U.S. patent application number 16/385692 was filed with the patent office on 2019-08-15 for build sequences for mechanosynthesis.
The applicant listed for this patent is CBN Nano Technologies Inc.. Invention is credited to Damian G. Allis, Jeremy Barton, Michael Drew, Robert A. Freitas, Aru Hill, Matthew Robert Kennedy, Michael Shawn Marshall, Ralph C. Merkle, Tait Takatani.
Application Number | 20190250187 16/385692 |
Document ID | / |
Family ID | 62022243 |
Filed Date | 2019-08-15 |
View All Diagrams
United States Patent
Application |
20190250187 |
Kind Code |
A1 |
Allis; Damian G. ; et
al. |
August 15, 2019 |
Build Sequences for Mechanosynthesis
Abstract
Build sequences for fabricating an atomically-precise product
can be determined using computational chemistry algorithms to
simulate mechanosynthetic reactions, and which may use the
mechanosynthesis process conditions or equipment limitations in
these simulations, to determine a set of mechanosynthetic reactions
that will build an atomically-precise workpiece with a desired
degree of reliability. Methods for error correction of pathological
reactions or avoidance of pathological reactions are disclosed.
Libraries of reactions may be used to reduce simulation
requirements.
Inventors: |
Allis; Damian G.; (Syracuse,
NY) ; Barton; Jeremy; (Ottawa, CA) ; Drew;
Michael; (Union City, CA) ; Freitas; Robert A.;
(Pilot Hill, CA) ; Hill; Aru; (San Jose, CA)
; Kennedy; Matthew Robert; (Tucker, GA) ; Merkle;
Ralph C.; (Santa Clara, CA) ; Takatani; Tait;
(Plano, TX) ; Marshall; Michael Shawn; (Lilburn,
GA) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
CBN Nano Technologies Inc. |
Ottawa |
|
CA |
|
|
Family ID: |
62022243 |
Appl. No.: |
16/385692 |
Filed: |
April 16, 2019 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
15806201 |
Nov 7, 2017 |
10309985 |
|
|
16385692 |
|
|
|
|
15588494 |
May 5, 2017 |
10197597 |
|
|
15806201 |
|
|
|
|
14712506 |
May 14, 2015 |
9676677 |
|
|
15588494 |
|
|
|
|
PCT/US13/28419 |
Feb 28, 2013 |
|
|
|
14712506 |
|
|
|
|
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
B82B 3/0019 20130101;
G01Q 80/00 20130101; C01B 32/28 20170801; G01Q 20/02 20130101; C01B
32/26 20170801 |
International
Class: |
G01Q 80/00 20060101
G01Q080/00; C01B 32/28 20060101 C01B032/28; C01B 32/26 20060101
C01B032/26; B82B 3/00 20060101 B82B003/00 |
Claims
1. A method of employing a mechanosynthetic build sequence,
comprising the steps of: providing a computer access to a
mechanosynthetic build sequence, the build sequence comprising an
ordered sequence of mechanosynthetic reactions, the build sequence
having been determined capable of creating a specified workpiece
with a desired degree of reliability, providing a positional
device; and operating the positional device under control of the
computer so as to carry out a plurality of mechanosynthetic
reactions from the build sequence.
2. The method of claim 1 wherein said step of providing a computer
access comprises loading the build sequence into a computer memory
connected to the computer.
3. The method of claim 1 wherein at least one of the
mechanosynthetic reactions in the build sequence is used to
temporarily passivate or reduce the valence of at least one atom
while creating the specified product, and at least one subsequent
reaction depassivates or increases the valence of said at least one
atom.
4. The method of claim 1 wherein said workpiece is
three-dimensional.
5. The method of claim 1 wherein the workpiece has at least 100
atoms.
6. The method of claim 1 wherein said workpiece comprises
diamondoid.
7. The method of claim 1 wherein said workpiece comprises
diamond.
8. The method of claim 1 wherein the workpiece is aperiodic.
9. The method of claim 1 wherein the order in which the
mechanosynthetic reactions are performed is determined at least in
part by steric considerations.
10. The method of claim 1 wherein the order in which the
mechanosynthetic reactions are performed is determined at least in
part to avoid undesired rearrangements in intermediate workpiece
structures.
11. A method for implementing a build sequence to fabricate a
specified atomically-precise product, the method comprising the
steps of: providing at least one mechanosynthetic tip mounted to at
least one positional device having the capability of sub-angstrom
positional accuracy; and using the at least one positional device
to transfer feedstock onto a workpiece by moving the at least one
mechanosynthetic tip, wherein the at least one mechanosynthetic tip
is moved according to an ordered sequence of mechanosynthetic
reactions that have a calculated degree of reliability at a given
temperature and with the limitations of the at least one positional
device taken into account, and the entire ordered sequence having
been determined capable of creating the specified product with a
desired degree of reliability based upon the calculated
reliabilities of the individual reactions in the ordered
sequence.
12. The method of claim 11 wherein at least one of the
mechanosynthetic reactions in the build sequence is used to
temporarily passivate or reduce the valence of at least one atom
while creating the specified product, and at least one subsequent
reaction depassivates or increases the valence of said at least one
atom.
13. The method of claim 11 wherein the workpiece is
three-dimensional.
14. The method of claim 11 wherein the workpiece has at least 100
atoms.
15. The method of claim 11 wherein the workpiece comprises
diamondoid.
16. The method of claim 11 wherein the workpiece comprises
diamond.
17. The method of claim 11 wherein the workpiece is aperiodic.
18. The method of claim 11 wherein the order in which the
mechanosynthetic reactions are performed is determined at least in
part by steric considerations.
19. The method of claim 11 wherein the order in which the
mechanosynthetic reactions are performed is determined at least in
part to avoid undesired rearrangements in intermediate workpiece
structures.
Description
CROSS-REFERENCE TO RELATED APPLICATIONS
[0001] This application is a continuation of pending application
Ser. No. 15/806,201 (filed 2017 Nov. 7), which is a continuation of
application Ser. No. 15/588,494 (filed 2017 May 5--issued as U.S.
Pat. No. 10,197,597), which is a continuation-in-part of
application Ser. No. 14/712,506 (filed 2015 May 14--issued as U.S.
Pat. No. 9,676,677), which is a continuation-in-part of Application
No. PCT/US13/28419 (filed 2013 Feb. 28), all of these prior
applications being incorporated herein by reference.
FEDERALLY SPONSORED RESEARCH
[0002] Not applicable.
SEQUENCE LISTING OR PROGRAM
[0003] Not applicable.
TECHNICAL FIELD
[0004] The present application relates to mechanosynthesis, the
fabrication of atomically precise tools and materials using
individual atoms or small groups of atoms as the fundamental
building blocks, and more particularly, to devices, methods, and
systems for performing ordered sequences of site-specific
positionally controlled chemical reactions that are induced by use
of mechanical force.
BACKGROUND
Mechanosynthesis and Related Techniques
[0005] Scanning Probe Microscopy (SPM, in which we include all
related techniques such as AFM, STM and others) laboratories have
been manipulating individual atoms and molecules for decades.
(Eigler and Schweizer, "Positioning Single Atoms with a Scanning
Tunnelling Microscope," Nature. 1990. 344:524-526; Eigler, Lutz et
al., "An atomic switch realized with the scanning tunneling
microscope," Nature. 1991. 352:600-603; Stroscio and Eigler,
"Atomic and Molecular Manipulation with the Scanning Tunneling
Microscope," Science. 1991. 254:1319-1326; Meyer, Neu et al.,
"Controlled lateral manipulation of single molecules with the
scanning tunneling microscope," Applied Physics A. 1995.
60:343-345; MEYER, NEU et al., "Building Nanostructures by
Controlled Manipulation of Single Atoms and Molecules with the
Scanning Tunneling Microscope," phys Stat Sol (b). 1995.
192:313-324; Bartels, Meyer et al., "Basic Steps of Lateral
Manipulation of Single Atoms and Diatomic Clusters with a Scanning
Tunneling Microscope Tip," PHYSICAL REVIEW LETTERS. 1997.
79:697-700; Bartels, Meyer et al., "Controlled vertical
manipulation of single CO molecules with the scanning tunneling
microscope: A route to chemical contrast," Applied Physics Letters.
1997. 71:213; Huang and Yamamoto, "Physical mechanism of hydrogen
deposition from a scanning tunneling microscopy tip," Appl. Phys.
A. 1997. 64:R419-R422; Bartels, Meyer et al., "Dynamics of
Electron-Induced Manipulation of Individual CO Molecules on
Cu(111)," PHYSICAL REVIEW LETTERS. 1998. 80; Ho and Lee, "Single
bond formation and characterization with a scanning tunneling
microscope," Science 1999.1719-1722; Hersam, Guisinger et al.,
"Silicon-based molecular nanotechnology," Nanotechnology. 2000;
Hersam, Guisinger et al., "Silicon-based molecular nanotechnology,"
Nanotechnology. 2000. 11:70; Hla, Bartels et al., "Inducing All
Steps of a Chemical Reaction with the Scanning Tunneling Microscope
Tip--Towards Single Molecule Engineering," PHYSICAL REVIEW LETTERS.
2000. 85:2777-2780; Lauhon and Ho, "Control and Characterization of
a Multistep Unimolecular Reaction," PHYSICAL REVIEW LETTERS. 2000.
84:1527-1530; Oyabu, Custance et al., "Mechanical vertical
manipulation of selected single atoms by soft nanoindentation using
near contact atomic force microscopy," Phys. Rev. Lett. 2003. 90;
Basu, Guisinger et al., "Room temperature nanofabrication of
atomically registered heteromolecular organosilicon nanostructures
using multistep feedback controlled lithography," Applied Physics
Letters. 2004. 85:2619; Morita, Sugimoto et al., "Atom-selective
imaging and mechanical atom manipulation using the non-contact
atomic force microscope," J. Electron Microsc. 2004. 53:163-168;
Ruess, Oberbeck et al., "Toward Atomic-Scale Device Fabrication in
Silicon Using Scanning Probe Microscopy," Nano Letters. 2004. 4;
Stroscio and Celotta, "Controlling the Dynamics of a Single Atom in
Lateral Atom Manipulation," Science. 2004. 306:242-247; Duwez,
Cuenot et al., "Mechanochemistry: targeted delivery of single
molecules," Nature Nanotechnology. 2006. 1:122-125; Iancu and Hla,
"Realization of a four-step molecular switch in scanning tunneling
microscope manipulation of single chlorophyll-a molecules," Proc
Natl Acad Sci USA. 2006. 103:13718-21; Ruess, Pok et al.,
"Realization of atomically controlled dopant devices in silicon,"
Small. 2007. 3:563-7; Sugimoto, Pou et al., "Complex Patterning by
Vertical Interchange Atom Manipulation Using Atomic Force
Microscopy," Science. 2008. 322:413-417; Randall, Lyding et al.,
"Atomic precision lithography on Si," J. Vac. Sci. Technol. B.
2009; Owen, Ballard et al., "Patterned atomic layer epitaxy of
Si/Si(001):H," Journal of Vacuum Science & Technology B:
Microelectronics and Nanometer Structures. 2011. 29:06F201; Wang
and Hersam, "Nanofabrication of heteromolecular organic
nanostructures on epitaxial graphene via room temperature
feedback-controlled lithography," Nano Lett. 2011. 11:589-93;
Kawai, Foster et al., "Atom manipulation on an insulating surface
at room temperature," Nat Commun. 2014. 5:4403) These efforts have
generally been limited to simple one- or two-dimensional
structures, but the techniques are powerful enough to have already
demonstrated basic molecular-scale logic (Heinrich, Lutz et al.,
"Molecule Cascades," Science. 2002. 298:1381-1387) and to have
inspired commercial efforts to build atomically-precise structures,
including work towards quantum computers. (Ruess, Oberbeck et al.,
"Toward Atomic-Scale Device Fabrication in Silicon Using Scanning
Probe Microscopy," Nano Letters. 2004. 4; Ruess, Pok et al.,
"Realization of atomically controlled dopant devices in silicon,"
Small. 2007. 3:563-7; Randall, Lyding et al., "Atomic precision
lithography on Si," J. Vac. Sci. Technol. B. 2009.)
[0006] Previously, atom manipulation was performed using one of
three techniques: Feedback Controlled Lithography (FCL), horizontal
atom manipulation, or vertical atom manipulation. FCL uses a
scanning probe tip to remove atoms (e.g., passivating hydrogens)
from a surface, creating chemically-reactive radical patterns on
that surface, followed by bulk chemical reactions that take
advantage of the new radical sites to create a surface modified at
specific atomic locations. Horizontal atom manipulation relies upon
dragging atoms across flat surfaces to place them at specific
locations, in effect decorating a surface with atomically-precise
designs. Vertical atom manipulation, often referred to as
mechanosynthesis, includes the deposition of single atoms or
molecules, such as CO, as well as vertical atom interchange, which
allows a surface and tip atom to be swapped. (Oyabu, Custance et
al., "Mechanical vertical manipulation of selected single atoms by
soft nanoindentation using near contact atomic force microscopy,"
Phys. Rev. Lett. 2003. 90; Morita, Sugimoto et al., "Atom-selective
imaging and mechanical atom manipulation using the non-contact
atomic force microscope," J. Electron Microsc. 2004. 53:163-168;
Oyabu, Custance et al., "Mechanical Vertical Manipulation of Single
Atoms on the Ge(111)-c(2.times.8) Surface by Noncontact Atomic
Force Microscopy," Seventh International Conference on non-contact
Atomic Force Microscopy. Seattle, Wash. 2004.34; Sugimoto, Pou et
al., "Complex Patterning by Vertical Interchange Atom Manipulation
Using Atomic Force Microscopy," Science. 2008. 322:413-417;
Tarasov, Akberova et al., "Optimal Tooltip Trajectories in a
Hydrogen Abstraction Tool Recharge Reaction Sequence for
Positionally Controlled Diamond Mechanosynthesis," J. Comput.
Theor. Nanosci. 2010. 7:325-353; Herman, "Toward Mechanosynthesis
of Diamondoid Structures: IX Commercial Capped CNT Scanning Probe
Microscopy Tip as Nowadays Available Tool for Silylene Molecule and
Silicon Atom Transfer," Journal of Computational and Theoretical
Nanoscience. 2012. 9:2240-2244; Herman, "Toward Mechanosynthesis of
Diamondoid Structures: X. Commercial Capped CNT SPM Tip as Nowadays
Available C2 Dimer Placement Tool for Tip-Based Nanofabrication,"
Journal of Computational and Theoretical Nanoscience. 2013.
10:2113-2122; Kawai, Foster et al., "Atom manipulation on an
insulating surface at room temperature," Nat Commun. 2014.
5:4403)
[0007] As previously implemented, each of these atom manipulation
techniques modifies a single atomic layer on a surface, does so
using a very limited palette of reactions and reactants, and cannot
manufacture complex, three-dimensional products.
[0008] Previous work by the current inventors, including U.S. Pat.
Nos. 8,171,568, 8,276,211, 9,244,097, 9,676,677, US Patent
Publication 20160167970 and PCT Application WO/2014/133529 sought
to address some of the shortcomings of prior atom manipulation
techniques via improved implementations of mechanosynthesis. These
references describe various aspects of mechanosynthesis, including
a bootstrap process for preparing atomically-precise tips from
non-atomically-precise tips, reactions that can be used to build
three-dimensional workpieces, methods for ordering such reactions
into build sequences, provisioning of feedstock, and disposal of
waste atoms.
[0009] Nonetheless, room for improvement still exists. Accordingly,
it is an object of the invention to improve the manufacturing of
three-dimensional workpieces via mechanosynthesis.
SUMMARY
[0010] The present invention is directed to processes for creating
build sequences which are determined using computational chemistry
algorithms to simulate mechanosynthetic reactions, and which may
use the mechanosynthesis process conditions or equipment
limitations in these simulations, and which facilitate determining
a set of mechanosynthetic reactions that will build an
atomically-precise workpiece with a desired degree of reliability.
Included are methods for error correction of pathological reactions
or avoidance of pathological reactions. Libraries of reactions may
be used to reduce simulation requirements.
BRIEF DESCRIPTION OF DRAWINGS
[0011] For a more complete understanding of the present invention,
reference is now made to the following descriptions taken in
conjunction with the accompanying drawings, in which:
[0012] FIG. 1 depicts the modular parts of an exemplary tip.
[0013] FIG. 2 depicts the modular parts of another exemplary
tip.
[0014] FIG. 3 depicts the AbstractionO tip surface-mounted on
Silicon.
[0015] FIG. 4 depicts the HDonationO tip surface-mounted on
Silicon.
[0016] FIG. 5 depicts the C2DonationO tip surface-mounted on
Silicon.
[0017] FIG. 6 depicts the MeDonationO tip surface-mounted on
Silicon.
[0018] FIG. 7 depicts a tip surface-mounted on Silicon which can be
SiH3DonationO, GeH3DonationO, SiMe3DonationO or GeMe3DonationO.
[0019] FIG. 8 depicts the AbstractionNH tip surface-mounted on
Silicon.
[0020] FIG. 9 depicts the HDonationNH tip surface-mounted on
Silicon.
[0021] FIG. 10 depicts the C2DonationNH tip surface-mounted on
Silicon.
[0022] FIG. 11 depicts the MeDonationNH tip surface-mounted on
Silicon.
[0023] FIG. 12 depicts a tip surface-mounted on Silicon which can
be SiH3DonationNH, GeH3DonationNH, SiMe3DonationNH or
GeMe3DonationNH.
[0024] FIG. 13 depicts the AbstractionS tip surface-mounted on
Gold.
[0025] FIG. 14 depicts the HDonationS tip surface-mounted on
Gold.
[0026] FIG. 15 depicts the C2DonationS tip surface-mounted on
Gold.
[0027] FIG. 16 depicts the MeDonationS tip surface-mounted on
Gold.
[0028] FIG. 17 depicts a tip surface-mounted on Silicon which can
be SiH3DonationS, GeH3DonationS, SiMe3DonationS or
GeMe3DonationS.
[0029] FIG. 18 depicts a synthetic route for the AbstractionO
tip.
[0030] FIG. 19 depicts a synthetic route for the HDonationO
tip.
[0031] FIG. 20 depicts a synthetic route for the C2DonationO
tip.
[0032] FIG. 21 depicts a synthetic route for the MeDonationO
tip.
[0033] FIG. 22 depicts a synthetic route for the SiH3DonationO
tip.
[0034] FIG. 23 depicts a synthetic route for the GeH3DonationO
tip.
[0035] FIG. 24 depicts a synthetic route for the SiMe3DonationO
tip.
[0036] FIG. 25 depicts a synthetic route for the GeMe3DonationO
tip.
[0037] FIG. 26 depicts a synthetic route for the AbstractionNH
tip.
[0038] FIG. 27 depicts a synthetic route for the HDonationO
tip.
[0039] FIG. 28 depicts a synthetic route for the C2DonationO
tip.
[0040] FIG. 29 depicts a synthetic route for the MeDonationO
tip.
[0041] FIG. 30 depicts a synthetic route for the SiH3DonationO
tip.
[0042] FIG. 31 depicts a synthetic route for the GeH3DonationO
tip.
[0043] FIG. 32 depicts a synthetic route for the SiMe3DonationO
tip.
[0044] FIG. 33 depicts a synthetic route for the GeMe3DonationO
tip.
[0045] FIG. 34 depicts a synthetic route for the AbstractionS
tip.
[0046] FIG. 35 depicts a synthetic route for the HDonationS
tip.
[0047] FIG. 36 depicts a synthetic route for the C2DonationS
tip.
[0048] FIG. 37 depicts a synthetic route for the MeDonationS
tip.
[0049] FIG. 38 depicts a synthetic route for the SiH3DonationS
tip.
[0050] FIG. 39 depicts a synthetic route for the GeH3DonationS
tip.
[0051] FIG. 40 depicts a synthetic route for the SiMe3DonationS
tip.
[0052] FIG. 41 depicts a synthetic route for the GeMe3DonationS
tip.
[0053] FIG. 42 depicts a synthetic route for the FHD-104X
intermediate.
[0054] FIG. 43 depicts a synthetic route for the NHD-103X
intermediate.
[0055] FIG. 44 depicts photo-activation of a halogen-capped
tip.
[0056] FIG. 45 depicts photo-activation of a Barton ester-capped
tip.
[0057] FIG. 46 depicts an exemplary synthesis of a tip with Barton
ester cap.
[0058] FIG. 47 depicts the use of surface-mounted tips where the
workpiece moves.
[0059] FIG. 48 depicts the use of surface-mounted tips where the
surface moves.
[0060] FIG. 49 depicts a metrology setup for measuring six degrees
of freedom.
[0061] FIG. 50a-f depicts a way of implementing the sequential tip
method.
[0062] FIG. 51 depicts a conventional mode tip that can be used for
the sequential tip method.
[0063] FIG. 52a-o depicts a build sequence for building a
half-Si-Rad tip.
[0064] FIG. 53 depicts a synthetic pathway for synthesizing an
AdamRad-Br tip.
[0065] FIG. 54 depicts exemplary methods of using strain to alter
affinity.
[0066] FIG. 55 is a flowchart of an exemplary process for
specifying a workpiece.
[0067] FIG. 56 is a flowchart of an exemplary process for designing
reactions.
[0068] FIG. 57 is a flowchart of an exemplary process for
performing reactions.
[0069] FIG. 58 is a flowchart of an exemplary process for testing
reaction outcomes.
DETAILED DESCRIPTION
Definitions
[0070] The following definitions are used herein:
[0071] An "adamantane" molecule comprises a 3D cage structure of
ten carbon atoms, each terminated with one or two hydrogen atoms,
having the chemical formula C10H16 in its fully hydrogen-terminated
form. Adamantane is the smallest possible unit cage of crystalline
diamond.
[0072] An "adamantane-like" structure includes one or more
adamantanes, one or more adamantanes where one or more atoms have
been substituted with atoms or molecular fragments of like or
similar valence, including e.g., Nitrogen, Oxygen, and
Sulfur-substituted variations, and similar molecules comprising
polycyclic or cage-like structures. By way of example, and not of
limitation, adamantane-like structures would include adamantane,
heteroadamantanes, polymantanes, lonsdaleite, crystalline silicon
or germanium, and versions of each of the foregoing where, for
example, Fluorine or another halogen is used for termination
instead of Hydrogen, or where termination is incomplete.
[0073] An "aperiodic" workpiece is one where the overall shape or
atomic constituents do not result directly from the crystal
structure or lattice of the workpiece. For example, diamond
crystals tend to form an octahedral shape due to the bond angles of
the underlying atoms. An octahedral diamond crystal, or variations
thereof, could be said to be periodic because both the internal
structure and the external shape is determined by the periodic
structure of the crystal. On the other hand, a diamond shaped like
a car cannot be said to be periodic because, internal structure
aside, there is no way the lattice cell of diamond could have
specified the shape of a car. Another example of aperiodic diamond
would be a crystal composed largely of diamond, but with irregular
(with respect to the crystal matrix) substitutions made within its
matrix, such as the replacement of some carbon atoms with silicon
or germanium. Almost any complex shape or part is going to be
aperiodic because of its shape, its atomic constituents, or both.
Note that aperiodic does not necessarily mean irregular. Take, for
example, a conventional gear made of diamond. The round,
symmetrical shape of a gear and its teeth are radially symmetric
and have a kind of periodicity. However, this periodicity is not
derived from the underlying crystal structure. For a structure to
be periodic, it is not enough that it be regular; it must be
regular in a manner that is derived from its crystal structure.
While this definition may seem pedantic, it is useful when
discussing the differences between an engineered,
atomically-precise material versus a naturally-occurring or
bulk-synthesized crystal. Naturally-occurring or bulk-synthesized
crystals are generally, impurities or bonding errors
notwithstanding, periodic structures. There is no way know to the
authors to make them both atomically-precise and aperiodic since
their method of manufacture inherently relies upon the periodic
crystal structure given elements form under a particular set of
conditions, rather than controlling the structure atom by atom as
can be done with a positionally-controlled technology like
mechanosynthesis.
[0074] An "atom" includes the standard use of the term, including a
radical, which, for example, may be just a proton in the case of
H.sup.+.
[0075] "Atomically-precise" in the context of a reaction means
where the position and identity of each atom is known to a
precision adequate to enable the reaction to be directed to a
particular atomic site ("site-specific"). In the context of a
workpiece, atomically-precise refers to the actual molecular
structure being identical to the specified structure (e.g., as
specified by a molecular model or build sequence).
[0076] The "bridgehead" position in an adamantane-like molecular
structure refers to a structural atom that is bonded to three other
structural atoms and may be terminated by one or more nonstructural
atoms. This is contrasted with a "sidewall" position which refers
to a structural atom that is bonded to two other structural atoms
and is terminated by one or more nonstructural atoms.
[0077] A "build sequence" is one or more mechanosynthetic reactions
arranged in an ordered sequence that permits the assembly,
disassembly, or modification of a workpiece.
[0078] A "chemical bond" is an interatomic covalent bond, an
interatomic ionic bond, or interatomic coordination bond, as these
terms are commonly understood by practitioners skilled in the
art.
[0079] A "chemical reaction" is said to occur when chemical bonds
are formed, broken, or altered.
[0080] "Conventional mode" is where one or more tips are affixed to
a positional means/device (e.g., an SPM probe) to facilitate
mechanosynthetic reactions between the tips and a workpiece. This
contrasts with "inverted mode" where a workpiece is affixed to a
positional means and the workpiece moves to the tips. Although
uncommon in practice, since in theory both tips and workpiece could
be affixed to a positional means, another way to distinguish
between the modes would be to say that if the workpiece is
connected to apparatus which indicates that the workpiece is being
used as a probe (e.g., if STM is being done through the workpiece),
the system is operating in inverted mode. Otherwise, the system is
operating in conventional mode. Conventional mode tips are
generally affixed to a positional means singly or in small numbers,
while in inverted mode, a larger, generally stationary,
presentation surface allows the provisioning of large numbers of
surface-mounted tips. Note that although inverted mode and surface
mounted tips may be used together, inverted mode should not be
conflated with surface-mounted tips. As is described herein (the
sequential tip method), surface-mounted tips can be used in a
system which is operating in conventional mode.
[0081] A "conventional mode tip" is a tip affixed to a positional
means or otherwise being employed in conventional mode as described
in that definition, just as an "inverted mode tip" is a tip affixed
to a presentation surface or otherwise being employed in "inverted
mode" as described in that definition.
[0082] "Diamond" is a crystal of repeating adamantane cage units
arranged in various well-known crystallographic lattice
geometries.
[0083] "Diamondoid" materials include any stiff covalent solid that
is similar to diamond in strength, chemical inertness, or other
important material properties, and possesses a three-dimensional
network of bonds. Examples of such materials include but are not
limited to (1) diamond, including cubic and hexagonal lattices and
all primary and vicinal crystallographic surfaces thereof, (2)
carbon nanotubes, fullerenes, and other graphene structures, (3)
several strong covalent ceramics of which silicon carbide, silicon
nitride, and boron nitride are representative, (4) a few very stiff
ionic ceramics of which sapphire (monocrystalline aluminum oxide)
is representative, and (5) partially substituted variants of the
above that are well-known to those skilled in the art.
[0084] "Feedstock" is the supply of atoms used to perform
mechanosynthetic reactions. Feedstock may take the form of one or
more atoms, including radicals (e.g., .GeH2, .CH2). Feedstock
includes atoms removed from a workpiece. For example, a hydrogen
atom from a workpiece may be the feedstock for a hydrogen
abstraction tip. In such cases, since frequently nothing is
subsequently to be done with atoms removed from a workpiece, such
feedstock may be referred to as "waste atoms." Feedstock must be
atomically-precise.
[0085] A "handle structure" comprises a plurality of atoms whose
bonding pattern is not altered during a site-specific
mechanosynthetic chemical reaction and whose primary function is to
hold a tip(s) or workpiece(s) to facilitate a mechanosynthetic
chemical reaction when the handle structure is manipulated by a
positional device. Handle structure may include the null case
(e.g., a tip or workpiece bound directly to a positional
means).
[0086] An "inert environment" includes, but is not limited to,
ultra-high vacuum (UHV), argon, nitrogen, helium, neon, or other
gases or liquids, either individually or in combination, that do
not react with the tip(s), feedstock, or workpiece(s) during
mechanosynthetic operations.
[0087] "Inverted mode": see definition within "Conventional Mode"
definition.
[0088] "Mechanical force" may include applied mechanical forces
having positive, negative, or zero magnitude. Chemical reactions
driven by the application of mechanical force include reactions
that are (1) driven through its reaction barrier by mechanically
forcing reactants or products through the transition state, or (2)
driven away from an undesired reaction by mechanically restraining
potentially reactive sites from attaining closer physical
proximity, or (3) allowed to occur by bringing potentially reactive
sites into closer physical proximity when zero mechanical force is
required to do so, as for example when no reaction barrier exists,
or when thermal energy alone is sufficient to surmount the reaction
barrier.
[0089] "Mechanosynthesis" is the use of positional control and
mechanical force to facilitate site-specific chemical reactions
involved in the building, alteration, or disassembly of a
workpiece. Mechanosynthesis does not require voltage biases, but
neither does it exclude their use.
[0090] A "mechanosynthetic reaction" (sometimes referred to as a
"reaction" when context makes it clear that the reaction is
mechanosynthetic) is a chemical reaction carried out using
mechanosynthesis.
[0091] A "meta-tip" is a handle to which multiple tips are
attached. For example, a meta-tip could be prepared using a
conventional SPM probe with a flat surface on the end, which is
then functionalized with multiple tips.
[0092] A "modular tip" is a tip with a modular design. Modules
include an active site, a body, feedstock, legs, and linkers. Some
of these modules may be considered to be modular themselves. For
example, a body contains an active site, and the active site may be
said to include feedstock. Similarly, linkers can be thought of as
part of the leg module. A modular tip may be referred to as simply
a "tip" when context makes the type of tip clear.
[0093] A "positional device" is a device capable of exerting
atomically-precise positional control on a mechanosynthetic tip,
tool, or workpiece, and may include, but is not limited to,
scanning probe microscopes (SPM) and atomic force microscopes (AFM)
and related devices, a miniaturized or MEMS-scale SPM or AFM, a
robotic arm mechanism of any size scale, or other appropriate
manipulation system capable of atomically-precise positional
control and appropriate force application. Many types of such
positional devices are known to those skilled in the art, but for
example, actuators can be based upon piezo elements or
electrostatics. Metrology based upon piezo elements, or optical
(e.g., interferometry), capacitive, or inductive techniques, or
other technology, can be used for positional feedback if
required.
[0094] A "presentation surface" is a surface which can be used to
bind feedstock or tips for use in mechanosynthesis, and as a base
on which to build a workpiece. Although generally monolithic, a
presentation surface can be composed of more than one material
(e.g., gold and silicon could both be used where each has
advantageous aspects), or composed of multiple non-adjacent
surfaces. A presentation surface may be referred to simply as a
"surface" when context makes the meaning clear. Presentation
surfaces include the appropriate area(s) on handle structures and
meta-tips. Presentation surfaces are preferably as close as
possible to atomically-flat, but this is largely a convenience
having to do with standard equipment design, and to facilitate
higher speeds and reduced scanning (e.g., to create topological
maps of non-flat surfaces), rather than an absolute
requirement.
[0095] "Site-specific" refers to a mechanosynthetic reaction taking
place at a location precise enough that the reaction takes place
between specific atoms (e.g., as specified in a build sequence).
The positional accuracy required to facilitate site-specific
reactions with high reliability is generally sub-angstrom. With
some reactions that involve large atoms, or those with wide
trajectory margins, positional uncertainty of about 0.3 to 1
angstrom can suffice. More commonly, a positional uncertainty of no
more than about 0.2 angstroms is needed for high reliability. Some
reactions, for example, due to steric issues, can require higher
accuracy, such as 0.1 angstroms. These are not hard cutoffs;
rather, the greater the positional uncertainty, the less reliable a
reaction will be.
[0096] A "structural atom" in an adamantane-like molecular
structure refers to an atom comprising the cage framework, for
example a carbon atom in an adamantane molecule. More generally, a
structural atom is an atom that comprises part of the backbone or
overall structure in a highly-bonded molecule.
[0097] A "synthetic tip" is an atomically-precise tip manufactured
via a bulk method, such as gas or solution-phase chemistry, rather
than via mechanosynthesis. A synthetic tip be referred to as simply
a "tip" when context makes the type of tip clear.
[0098] A "terminating atom" refers to an atom that does not serve
as a structural atom but absorbs unused valences of a structural
atom. For example, a hydrogen atom in an adamantane molecule.
[0099] A "three-dimensional" workpiece means a workpiece including
a lattice of atoms whose covalent structure occupies more than a
single plane, discounting bond angles. Under this definition, for
example, most proteins (discounting e.g., disulfide inter- or
intra-molecular bonds) and other polymers would be two dimensional,
as would a plane of graphene. A covalent network solid or a carbon
nanotube would be three-dimensional.
[0100] A "tip" is a device for facilitating mechanosynthetic
reactions which includes one or more "active" atoms or sites whose
bonding pattern or electronic state is altered during a
mechanosynthetic operation, and one or more "support" atoms whose
bonding pattern or electronic state is not altered during a
mechanosynthetic operation. The support atoms hold the active atoms
in position, and may also modify the chemical behavior of the one
or more active atoms. Unless otherwise specified, a tip is
atomically-precise.
[0101] "Tip swapping" is the process of connecting a new tip and
handle structure to a positional means. In conventional SPM, this
may be done by, for example, manually changing the probe, or using
equipment with probe magazines which hold multiple probes and can
automate tip swapping.
[0102] A "tool" comprises a tip, potentially bonded to a handle,
controlled by a positional device or means.
[0103] A "workpiece" is an apparatus, article of manufacture, or
composition of matter, built via mechanosynthesis. A system may
have more than one workpiece. A workpiece may be connected to, but
does not include, non-atomically-precise structures such as a
support substrates or pre-existing structures onto which a
workpiece is built.
Chemical Structure and Scientific Notation
[0104] A dot (".") is may be used in chemical structures herein to
represent an electron, as in the radical group ".CH2". For ease of
typesetting, the notation herein generally omits subscript or
non-standard characters. Superscript may be written using the
"{circumflex over ( )}" character when required for clarity.
Synthetic Tips
[0105] Previous literature described (see, e.g., U.S. Pat. No.
9,244,097 or WO2014/133529) a bootstrap process to facilitate the
creation of atomically-precise tips from atomically-imprecise tips
using mechanosynthesis. This is a potentially complex process,
requiring the characterization of atomically-imprecise tips, to
then perform mechanosynthetic reactions with those tips, to build
atomically-precise tips. Being able to skip this step is therefore
quite useful. As an alternate method of preparing
atomically-precise tips, we describe the bulk synthetic chemical
preparation (and if appropriate, passivation, and depassivation or
activation) of various atomically-precise tips.
[0106] Bulk synthetic preparation of tips allow the avoidance of a
bootstrapping process. As will also been seen, bulk presentation of
such tips on a surface allows a fundamentally different way of
dealing with feedstock provisioning, waste atom disposal, and
access to multiple tips.
[0107] With respect to feedstock provisioning, previous work such
as WO2014/133529 describes the use of feedstock depots and trash
depots. Feedstock depots are presentation surfaces to which
feedstock has been directly bound. Trash depots are surfaces which
provide for waste disposal by allowing a tip to transfer unwanted
atoms from the tip to the surface. One drawback to this method is
the lack of chemical diversity available on the surface(s). On a
uniform surface, different feedstock will have different affinity,
potentially higher or lower than optimal. Herein we describe a way
to completely avoid needing to bind feedstock or waste atoms
directly to a surface by using "tips on a surface." In addition to
using presentation surfaces directly as feedstock and trash depots,
previous proposals describe rechargeable tips, employing strategies
that use a relatively small number of tips over and over again
during a build sequence. Herein we describe methods for partially
or completely avoiding tip reuse, and hence we are able to omit tip
recharge steps (e.g., as described in WO2014/133529), streamlining
the entire process.
[0108] Synthetic tips, because they can be made via bulk chemistry
techniques, are available in very large numbers after synthesis
(like the molecules in most bulk chemical reactions, "very large
numbers" can mean up to millions, billions, or even far more,
ranging into numbers that require scientific notation to easily
express). Therefore, a large number of synthetic tips could be
affixed to a presentation surface. The synthetic tips can be
pre-charged (meaning, the tips are already in the chemical state
desired to carry out the intended reactions, such as already being
bonded to feedstock), and they can include large numbers of every
type of tip required for a given build sequence. In this way, the
presentation surface can serve purposes including being a feedstock
depot (the synthetic tips already being charged with their
feedstock), a trash depot (e.g., radical tips could be used to bind
waste atoms), and a varied collection of tips that can carry out
all necessary reactions (for example, almost any number of tips,
including all the tips described herein, or in previous work such
as WO2014/133529, could be present on a presentation surface, and
all in large numbers). Using a large number of synthetic tips also
allows each tip to be disposable, rather than requiring recharge
for subsequent use, avoiding the need to design and perform
recharge operations.
[0109] The availability of large numbers of tips on a surface
raises the idea that a workpiece could be connected to a positional
means, allowing the workpiece to move to the tips ("inverted
mode"), rather than tips moving to the workpiece ("conventional
mode"). Conceptually, if the workpiece moves and the presentation
surface is stationary, one could think of a build sequence as a
workpiece moving around a presentation surface, aligning itself
with a desired tip, and then being brought into contact with that
tip with sufficient force to trigger a mechanosynthetic reaction.
The tip that was used is then spent, but the presentation surface
can easily provide large numbers of tips. The build sequence
proceeds by then aligning the workpiece with the next appropriate
unspent tip and bringing them together. This process repeats until
the entire workpiece is built.
[0110] Note that, as is discussed elsewhere herein, in some
embodiments, the process of mechanosynthesis may involve scanning
the presentation surface to establish a topological map and the
positions of the tips to be used. If the tips have been mapped,
software can be used to keep track of which locations have been
used and which have not. An alternative implementation would be to
simple scan for unused tips as they are needed, since a used tip
and an unused tip would have markedly different characteristics
when evaluated via, e.g., STM.
[0111] Other variations on this concept are also possible,
including a tool which holds multiple tips (a "meta-tip"). Such
designs may be more efficient than a tool holding a single tip
because multiple reactions could be performed without requiring tip
swapping or tip recharge. Whether the tips reside on a presentation
surface, or on a tool, and whether the presentation surface, the
tool, the workpiece, or some combination thereof are coupled to
positional means, the overarching point is a design which has at
least some of the following characteristics and advantages, among
others.
[0112] First, a plurality of tips can be made available. These tips
could be all the same, or could include many different types of
tips. If multiple tip types are present, they could be randomly
intermingled, segregated by sector or position, or the tips could
be laid out in an order which maximizes the efficiency of a build
sequence (for example, by arranging different tip sectors in a
manner that minimizes the movement required to perform the
mechanosynthetic operations to build a particular workpiece, or
considering a more general design, locating tips that are apt to be
used more frequently closer to the workpiece, or locating tip
sectors concentrically around a workpiece to minimize total tip to
workpiece distance regardless of the order of reactions).
[0113] Second, due to the large number of tips that are accessible
to the system, tip recharge may be reduced or eliminated during a
build sequence. Each tip can be used once, and then ignored once it
is spent. By eliminating recharge reactions, shorter, faster build
sequences are facilitated. If additional tips were still required,
e.g., for a workpiece requiring a number of tips beyond that which
are available, the strategy of mounting a large number of tips,
preferably in their ready-to-use state, on a surface, allows the
bulk replacement of tips by swapping in a new surface. In this
scenario, tip recharge is not completely eliminated, but it is
greatly reduced.
[0114] Third, tips do not have to be swapped for chemical diversity
because every type of tip needed for a given build sequence can be
present somewhere on the presentation surface. This reduces or
eliminates the need for multiple positional means or tip
swapping.
[0115] Fourth, large numbers of atomically-precise tips can be
prepared and affixed via bulk chemical reactions (and bulk
activated, if required). This eliminates the need for a bootstrap
process that uses non-atomically-precise tips to create
atomically-precise tips. It also reduces or eliminates the need to
build tips using mechanosynthesis, which can be useful where
mechanosynthetic operations are the rate limiting step of a
manufacturing process. Exemplary synthetic pathways for multiple
synthetic tips are described herein.
[0116] Fifth, system complexity is kept relatively low, and the
number of tips and feedstock moieties available can be relatively
high, as compared to other proposals for providing feedstock via,
for example, methods which require cartridges or conveyor belts
(Rabani, "Practical method and means for mechanosynthesis and
assembly of precise nanostructures and materials including diamond,
programmable systems for performing same; devices and systems
produced thereby, and applications thereof." United States. 2009.
Ser. No. 12/070,489.).
[0117] With respect to the number of tips that may be available
under some of these scenarios, this can vary greatly. For example,
on a very small surface, such as a small flat at the end of a probe
tip (which would traditionally hold one tip and could do so in some
embodiments of the present invention), small numbers of tips could
be provided for chemical diversity. For example, two to ten tips
could be placed on the end of a probe, requiring no more than a few
square nanometers of space. This would provide convenient access to
tips of varying chemical nature without needing to swap probes.
Assuming a build sequence requiring more reactions than a small
batch of tips like this can provide, such tips would still have to
be recharged, but the advantage is that this could be done
chemically (e.g., by touching the tip to appropriate surfaces to
abstract or donate feedstock) rather than requiring physical
swapping of an entire tip and handle.
[0118] On larger surfaces, much larger numbers of tips could be
presented. For example, a presentation surface on the order of
square nanometers could provide dozens, hundreds, or thousands of
tips. A presentation surface on the order of square microns could
provide room for millions or billions of tips. And, if even larger
numbers or greater space are desired, long-distance metrology can
allow presentation surfaces on the order of square millimeters or
centimeters while still maintaining the requisite positional
accuracy. (Lawall, "Fabry-Perot metrology for displacements up to
50 mm," J. Opt. Soc. Am. A. OSA. 2005. 22:2786-2798)
[0119] When using a plurality of tips, the tips could all be the
same (helping to reduce recharge reactions, as described herein),
but as chemical diversity is also useful, there could also be
almost any number of different types, from two different types, to
the at least eight main tip/feedstock combinations described in,
e.g., FIGS. 3-7 (or nine including the later-described AdamRad-Br
tip), or even substantially more given the different types of
linkers, feedstock, other tip designs that could be used, and the
potential desire for tips to facilitate new reactions or that would
work under different conditions.
Surface-Mounted Tips
[0120] Synthetic tips, if properly designed, can be chemically
bound to a presentation surface, or "surface-mounted." In addition
to being amenable to synthesis using traditional chemistry, and
carrying out one or more mechanosynthetic reactions,
surface-mounted tips are designed to allow efficient bonding to a
presentation surface (often in large quantity).
[0121] Surface-mounted tips differ from the tips normally used in
SPM work in that they are not simply integral to a handle structure
(e.g., commercially available tips often have a tip where the
crystal structure of the tip is contiguous with the handle
structure; essentially the tip is just the end of the handle
structure), nor are they a handle structure to which only a trivial
functionalization has been added (e.g., bonding a single CO
molecule to the end of an existing tip is a common technique to
increase resolution). Surface-mounted tips differ from
previously-proposed mechanosynthetically-created tips in that they
do not require mechanosynthesis to manufacture (which has not only
process implications, but structural and chemical implications
since this requires that surface-mounted tips be able to bind to
the desired surface without the aid of mechanosynthesis). Given
this, while surface-mounted tips may appear superficially similar
to other tips described in the literature, the requirements for the
design of tips which are to be surface-mounted are substantially
different.
[0122] Binding orientation is one issue that must be addressed when
designing surface-mounted tips. It would be preferable that the
tips only affix themselves to a surface in a manner that renders
them properly oriented for use in mechanosynthetic reactions
(although multiple possible orientations could be acceptable given
the number of redundant tips that could be present--the system
could scan to identify and use only tips in the desired
orientation, but this reduces efficiency).
[0123] Active sites and legs are discussed in more detail herein,
but are major factors in ensuring that correct binding orientation
is obtained. For example, tips with radical active sites will be
highly reactive in their active form. Due to this high reactivity,
the active site may bind to the presentation surface instead of the
legs. If this happens, the tip would end up bound to the
presentation surface upside down or at least improperly oriented.
Reactive sites may also form bonds to other parts of the same tip,
or may form bonds to other tips, such as two tips dimerizing. This
problem may be avoided in the case of reactive active sites by
binding the tip to the presentation surface while the active sites
are neutralized. The active sites can then be activated after leg
binding. A similar issue presents itself with respect to the legs.
The legs (or leg linkers) need to be reactive enough that they will
bind to the presentation surface, but they must resist pathological
reactions with themselves or other tips (e.g., forming a leg-leg
bond instead of a leg-surface bond, or undergoing any other
undesired reactions).
[0124] Of course, there are other design consideration for tips,
including that they perform the desired reactions reliably during a
build sequence, but the above concerns are unique to
bulk-synthesized, surface-mounted tips. Tips created using
mechanosynthesis can largely avoid the problems described above via
the positional specificity of the reactions used in their
synthesis.
Modular Tip Design
[0125] As will be seen in subsequent examples, surface-mounted tips
can be thought of as being modular. Each tip can be thought of as
having an active site (one or more atoms that bind a desired atom
or group of atoms, which could be, e.g., feedstock for a donation
reaction, or some moiety to be removed from a workpiece for an
abstraction reaction), a body (adamantane or an adamantane
derivative in our examples, but other structures could be used
given the teachings herein), and one or more legs that serve to
attach the tip to a surface. The feedstock of a tip could also be
considered a module, as could the surface, which, although not
technically part of the tip, can be important to tip design and
function.
[0126] To aid in understanding how tips function, and how they can
be rationally designed, considerations pertinent to each module are
described below. Note that the specific examples presented use
adamantane, or adamantane-like bodies. Many reactions for
functionalizing adamantanes are known, and their stiffness, small
size, computational tractability and other favorable
characteristics lead us to use these structures as exemplary tips,
although many different molecules, including other adamantane-like
structures, could serve the same purpose.
[0127] The active site's main characteristic is that it reliably
facilitate the desired reaction on a workpiece. However, how to
efficiently synthesize and deliver tips to a surface, and prepare
them for use, must be considered in their design. Particularly when
a tip's ready-to-use form includes a radical, a tip may incorporate
a protective cap (what in solution-phase chemistry is commonly
referred to as a "protecting group"). This cap reduces the active
site's reactivity prior to use to avoid, for example, tip-tip
dimerization, binding of the active site to the surface, or other
undesired reactions. However, the cap must be removable so that the
tip can be activated for use. One way to do this is to make the cap
photo-cleavable, but other methods are possible and well-known in
the field of chemistry.
[0128] The body may contain, or serve as a point of attachment for,
the active site. The body also serves as a point of attachment for
one or more legs. The body can also serve to tune the active site,
and to isolate it from other chemical influences. With respect to
tuning the active site, for example, substitutions which alter bond
lengths, angles, or electronegativity may be used to increase or
decrease the affinity of the active site for its feedstock. With
respect to isolation, the body provides chemical isolation from,
for example, the legs. Such isolation is one of the aspects of this
modular design paradigm that eases the design of new tips by
allowing modules to be put together combinatorially. For example,
if an active site and body combination that accomplish the desired
reaction are already known, but one desires to use a different
surface which necessitates different legs, it is likely that the
new legs can be swapped in without redesign of the body and active
site. If the legs were connected directly to the active site, their
chemical nature would tend to have more of an effect on the active
site, potentially requiring redesign of the body, or unnecessarily
constraining the choice of legs. Another characteristic of the body
is that it is preferably rigid. A rigid body will tend to be more
versatile because a rigid body will better resist deformation when
forces are applied to it during mechanosynthetic reactions.
[0129] The legs serve to attach the body to the surface. The legs
preferably have a geometry that permits them to bind the body to a
surface without excessive strain, including surfaces that are
functionalized prior to leg attachment. Functionalized surfaces,
such as chlorinated Si, may make longer legs preferable because
the, e.g., Cl atoms, can be directly under the tip body, making
some clearance between the body of the tip and Si surface
preferable. Legs are also preferably fairly rigid, and strong
enough so that reactions require the application of force proceed
reliably rather than the tip tilting, otherwise moving, or breaking
a leg bond. While legs that are too short may be unable to bond to
the surface reliably, legs that are too long may be too flexible,
adding to the positional uncertainty of the tip atoms during a
mechanosynthetic operation. Where issues such as surface
functionalization and lattice mismatches between the surface and
body are not issues, legs can be very short (e.g., a single oxygen
atom could serve as each leg).
[0130] With respect to the number of legs, the examples provided
depict tips with three legs. Three legs helps provide stability
against forces acting upon the active site or feedstock at varied
angles, and can reduce the force on any given leg by spreading it
amongst all legs. However, tips with one or two legs could be used,
as could tips with four, or more, legs. Note that tips with more
than one leg may be usable when not all of their legs have bound to
the presentation surface, as long as the required stability is
provided. On a tip with multiple legs, each leg does not need to be
identical.
[0131] Legs may incorporate linkers (if not, the leg may be
considered to also be the linker, or vice versa), which serve to
provide a bridge between the rest of the leg and the body or
surface. The advantage of linkers is in providing an appropriate
chemistry with which to bind a surface. For example, if the rest of
the leg does not have the necessary reactivity or bond strength
with a surface, a linker may address the issue. This is
demonstrated with the exemplary tips described herein, wherein each
tip may have, e.g., a trifluorobenzene leg, and to that leg may be
attached a linker which is, e.g., NH, O, or S. This modular
swapping of linkers allows otherwise-identical tips to be adapted
to various surfaces without compromising the characteristics of the
active site. Linkers may also be used to adjust the geometry of the
legs, for example, helping them to fit the surface lattice spacing
better, adjusting their length, or altering their rigidity.
[0132] Feedstock serves as a source of atoms which can be added to
a workpiece and is generally attached to the "top" (with respect to
the orientation depicted in, e.g., FIG. 1-17, although the
real-world orientation may differ) of the tip to provide access to
the feedstock without steric interference from other parts of the
tip or the surface. Feedstock is chosen not only by what atom or
atoms it contains, but by how it binds to a tip's active site and
the desired location on a workpiece. There are many ways, for
example, to donate carbon atoms to a workpiece, and examples using
C2, CH2, and CH3 are all presented herein. Context will determine
which is most appropriate, though often more than one could be used
to build a given workpiece, assuming appropriate alterations in the
build sequence.
[0133] The surface to which a tip is being attached has a variety
of important characteristics, including chemical reactivity,
surface smoothness, lattice spacing, linker-surface bond strength,
and internal bond strength. In terms of chemical reactivity, the
surface must bind to the linkers during the tip binding process,
but preferably not to other parts of the tip. The surface's lattice
spacing must allow linker binding without excessive strain. The
linker-surface bond strength must suffice so that the bonds do not
rupture if pulling forces are required. And, the internal
(surface-surface) bonds must be of sufficient strength that, if
pulling forces are required, the entire tip, along with one or more
surface atoms, is not ripped from the surface.
[0134] With surface-mounted tips being broken down into the
described modules, and the important functional characteristics of
each module described, and realizing that this modular design at
least to an extent isolates various modules from one another,
facilitating module re-use and combinatorial creational of new
tips, along with the examples presented herein, this provides a
design paradigm for the design and synthesis of new tips that can
be generalized well beyond the specific examples provided.
[0135] FIG. 1 depicts one version of an abstraction tip that may be
used to remove hydrogen, among other moieties, from a workpiece.
Radical 101 is used to bind the moiety to be abstracted, and serves
as the tip's active site. The active site is connected to body 102,
which in this example is adamantane. The body is connected to three
methyl group legs, exemplified by leg 103. Each leg contains a
sulfur linker, exemplified by linker 104. Each linker is bound to
surface 105. As an abstraction tip being depicted in its
ready-to-use state, no feedstock is present.
[0136] As a different example, with feedstock, FIG. 2 depicts one
version of a tip capable of donating hydrogen to many atom types.
Active site 201 is a Ge atom, which in this case is part of a
substituted adamantane body 202. Trifluorobenzene (which could be
viewed as trifluorophenol if considered together with the linkers)
legs are used, exemplified by leg 203, and each leg is connected to
an oxygen linker 204, which connects to surface 205. Feedstock 206
is connected to active site 201.
Exemplary Tips
[0137] Surface-mounted tips, along with their routes of synthesis,
have been devised which carry out mechanosynthetic reactions while
minimizing or eliminating issues such as tip dimerization and
improper tip orientation during surface mounting, and allow for
proper leg length, flexibility and linker chemistry to bind to the
exemplary surfaces. These synthetic routes allow for the bulk
manufacture of many diverse tip types, thereby facilitating many
different mechanosynthetic reactions while having the benefits
described for surface-mounted tips and the processes for using such
tips.
[0138] The set of tips described includes an abstraction tip with a
C2-based active site (capable of extracting many atoms from many
different types of workpieces, including, e.g., hydrogen from
diamond), a hydrogen donation tip, a C2 donation tip, a Methyl
donation tip, and a donation tip which can donate SiH3, GeH3,
Si(CH3)3, or Ge(CH3)3, depending on the feedstock attached to the
Ge active atom in its substituted adamantane body.
[0139] To demonstrate the modular design described herein, various
versions of each tip are depicted. Specially, each tip is shown
with three trifluorobenzene legs which can be linked to either a
chlorinated silicon surface, or a partially-hydrogenated
partially-chlorinated silicon surface, via an oxygen linker or an
NH linker. A version of each tip is also depicted where the legs
are methyl groups, using sulfur linkers to connect to an Au
surface. These various versions provide for a variety of surface
properties and surface attachment chemistries and demonstrate how a
body can be used to isolate an active site from other changes in
the tip, as the tips continue to function as desired after changing
the legs, linkers, and surface.
[0140] Note that a silicon surface has stronger intra-surface bonds
than a gold surface. When placing tips on a gold surface, reactions
that require substantial pulling forces (exceeding a few nN) may
pull the tip from the surface (taking one or more gold atoms with
it), or cause the tip to slide sideways across the surface.
Nonetheless, the thiol linker chemistry is very accessible, making
gold a useful surface (along with lead and other similar materials)
if reactions with substantial pulling forces are not required.
[0141] Each exemplary tip is shown in detail, bonded to an
appropriate surface for the linker chemistry depicted, in FIGS.
3-17. FIGS. 3-7 all depict tips that use trifluorobenezene legs and
oxygen linkers on a silicon surface. Specifically: FIG. 3 depicts
an abstraction tip having a C2-radical-based active site, an
adamantane body, trifluorobenzene legs, and oxygen linkers, on a
silicon surface (all Si surfaces include, e.g., chlorinated,
partially-chlorinated, and partially-hydrogenated,
partially-chlorinated Si). This tip will be referred to as
AbstractionO. FIG. 4 depicts a hydrogen donation tip with hydrogen
feedstock, a Ge-based active site incorporated into a substituted
adamantane body, trifluorobenzene legs, and oxygen linkers, on a
silicon surface. This tip will be referred to as HDonationO (or
"HDonation," omitting the specific linker group, to denote any of
the variants, a conventional which can apply to any of the tip
names). FIG. 5 depicts a C2 donation tip with .C2 feedstock, and
otherwise the same structure as FIG. 4. This tip will be referred
to as C2DonationO. FIG. 6 depicts a methyl donation tip with .CH2
feedstock, and otherwise the same structure as FIG. 4. This tip
will be referred to as MeDonationO. FIG. 7 depicts a donation tip
that can be used to donate a variety of feedstock moieties
depending on the identity of the M and R groups. M can be Si or Ge,
and R can be H or CH3, allowing the tip to donate SiH3, GeH3,
Si(CH3)3 or Ge(CH3)3. These tips will be referred to, respectively,
as SiH3DonationO, GeH3DonationO, SiMe3DonationO, and
GeMe3DonationO. FIG. 7 has otherwise the same structure as FIG.
4.
[0142] FIGS. 8-12 depict tips with the same feedstock (if present),
active site, bodies, and legs as FIGS. 3-7, respectively, but each
tip in FIGS. 8-12 uses NH linkers instead of oxygen linkers to
connect to a silicon surface. These tips will be referred to,
respectively, as AbstractionNH, HDonationNH, C2DonationNH,
MeDonationNH, and for the various versions of FIG. 12,
SiH3DonationNH, GeH3DonationNH, SiMe3DonationNH, and
GeMe3DonationNH.
[0143] FIGS. 13-17 depict tips with the same feedstock (if
present), active site, and bodies as FIGS. 3-7, respectively, but
each tip in FIGS. 13-17 uses methyl legs and a sulfur linker to
connect the tip to a gold surface. These tips will be referred to,
respectively, as AbstractionS, HDonationS, C2DonationS,
MeDonationS, and for the various versions of FIG. 17,
SiH3DonationS, GeH3DonationS, SiMe3DonationS, and
GeMe3DonationS.
[0144] In addition to the use of these tips in their charged state,
some tips could be used in their uncharged state. For example,
several of the tips, such as the hydrogen donation tip, have a Ge
radical active site in their discharged state. This can be a useful
form of these tips, for example, to break into a C.dbd.C bond, or
as a trash depot for unwanted atoms (assuming appropriate
affinity).
[0145] With respect to naming conventions, note that sometimes tips
are described in terms of what reaction they perform, and sometimes
in terms of their structure and payload. For example "MeDonation"
(regardless of whether the tip's legs are based on NH, O, S,
phenylpropargyl alcohol, or something else) stands for "methyl
donation" since that is what the tip does. With respect to naming
via structure and payload, for example, many of the donation tips
described herein have Ge-substituted adamantane bodies. With no
feedstock, the Ge atom would be a radical, and so may be referred
to as "GeRad." Similarly, "AdamRad" is an adamantane molecule
without the C to Ge substitution, but rather having a radical
carbon at the active site. An adamantane can also be substituted
with a silicon atom at its active site, which may be called SiRad.
Obviously, these are just examples used to describe naming
conventions, not a list of all possible structures or
substitutions, which are numerous. To convey what feedstock is
attached, the names may be written as, for example, GeRad-CH2
(which is one implementation of an MeDonation tip), GeRad-H (one
implementation of an HDonation tip). Understanding these
conventions, the tip name normally makes its structure and/or
function obvious.
Tip Synthesis
[0146] Exemplary synthetic pathways for each tip are depicted in
FIGS. 18-41. Note that multiple synthetic pathways for the tip
depicted in FIGS. 7, 12 and 17 due to the various possible
combinations of M and R. Tips with radicals in their active form
are synthesized with a protective cap. Procedures for cap removal
are described herein.
[0147] FIG. 18 depicts a synthetic pathway for AbstractionO. The
synthesis steps are as follows: Commercially available
1,3,5-trihydroxyadamantane reacts with 2,4,6-trifluorophenol while
heated between 50-80.degree. C. under acidic conditions to give
OFA-1. Treating OFA-1 with an excess dimethyldioxirane (DMDO) in
acetone at room temperature selectively oxidizes the tertiary C--H
bond to give alcohol OFA-2. Using Koch-Haaf conditions (Stetter,
H., Schwarz, M., Hirschhorn, A. Chem. Ber. 1959, 92, 1629-1635), CO
is formed from the dehydration of formic acid by concentrated
sulfuric acid between -5-0.degree. C. The CO forms a bond with the
tertiary carbocation formed from the dehydration of the bridgehead
alcohol at room temperature. Upon aqueous workup the carboxylic
acid OFA-3 is obtained. Esterification of the carboxylic acid OFA-3
with dry methanol and catalytic sulfuric acid between 40-60.degree.
C. yields the methyl ester OFA-4. The phenolic --OH groups in OFA-4
are protected with tert-butyldimethylsilyl chloride (TBSCl) in the
presence of imidazole at room temperature to give the TBS-silyl
ether OFA-5. Reduction of the methyl ester with LiAlH4 in
tetrahydrofuran (THF) between 0.degree. C. and room temperature
gives the methyl alcohol OFA-6. Oxidation of the methyl alcohol to
the aldehyde OFA-7 proceeds with catalytic tetrapropylammonium
perruthenate ((Pr4N)RuO4, TPAP) and stoichiometric
N-methylmorpholine-N-oxide (NMO). The presence of 4 .ANG. powdered
molecular sieves in the reaction mixture adsorbs any water present
and decreases the probability of over-oxidation to the carboxylic
acid (Ley, S. V., Norman, J., Griffith, W. P., Marsden, S. P.,
Synthesis, 1994, 639-666). Using a modified Corey-Fuchs procedure
(Michel, P., Rassat, A. Tetrahedron Lett. 1999, 40, 8570-8581), the
aldehyde in THF is added to a premixed solution of iodoform (CHI3),
triphenylphosphine, and potassium tert-butoxide at room temperature
in THF to undergo a carbon-carbon bond forming reaction to give the
1,1-diiodoalkene. Single elimination of the vinyl iodide with
excess potassium tert-butoxide and careful temperature control
(-78.degree. C.--50.degree. C.) yields the iodoalkyne OFA-8. It is
possible to form the terminal alkyne from this reaction if
temperature is not carefully controlled, however, the terminal
alkyne can be iodinated with N-iodosuccinimide/AgNO3 or,
alternatively, with 12 in basic methanol. The final global
deprotection of the TBS-silyl ether groups is performed with
tetra-n-butylammonium fluoride (TBAF). Upon aqueous workup, the
Abstraction.RTM. tip with free phenol linkers OFA-9 is
obtained.
[0148] FIG. 19 depicts a synthetic pathway for HDonationO. The
synthesis steps are as follows: FHD-104X is reduced by excess
lithium aluminum hydride in THF solvent at 0.degree. C., converting
the germanium halide to the germanium hydride FHD-105.
Tetra-n-butylammonium fluoride is used to deprotect the
tert-butyldimethylsilyl protecting groups from FHD-105 in THF to
yield the triphenol FHD-106, the HdonationOHtip.
[0149] FIG. 20 depicts a synthetic pathway for C2DonationO. The
synthesis steps are as follows: The Grignard reagent
ethynylmagnesium bromide in THF solution is added to FHD-104X
dissolved in dry THF and cooled to -78 C dropwise with rapid
stirring. The reaction is stirred for 1 hour, warmed to 0 C for 1
hour, and stirred for 1 hour at room temperature to form FC2D-101.
FC2D-101 is dissolved in dry THF and cooled to -78 C. A solution of
n-butyllithium in hexanes is added and the reaction is stirred for
1 hour at -78 C. A solution of iodine in dry THF is added and the
reaction is allowed to warm to room temperature to yield FC2D-102.
FC2D-102 is dissolved in THF and stirred rapidly at room
temperature. Tetra-n-butylammonium fluoride is added and the
reaction is stirred for 1 hour to yield FC2D-103, the C2DonationO
tip.
[0150] FIG. 21 depicts a synthetic pathway for MeDonationO. The
synthesis steps are as follows: The germanium halide FHD-104X in
THF solution is reduced with lithium metal to generate a lithiated
germanium species in situ. The solution is then slowly added
dropwise to a solution of 10-fold excess methylene iodide (CH2I2)
in THF cooled to 0 C. This method of addition favors the formation
iodomethyl germane FMeD-101 over methylene-bridged germanes.
Stoichiometric tetra-n-butylammonium fluoride is used to deprotect
the tert-butyldimethylsilyl protecting groups from FMeD-101 in THF
to yield the triphenol FMeD-102, the MeDonationO tip.
[0151] FIG. 22 depicts a synthetic pathway for SiH3DonationO. The
synthesis steps are as follows: The phenols of FHD-106 are acylated
with mesitoyl chloride in dichloromethane with pyridine base.
(Corey et al., JACS 1969, 91, 4398) The mesitoate protecting group
is utilized due to its stability to the lithiation conditions
necessary for FSiHD-102. FSiHD-101 in dry THF solution is
deprotonated with n-butyllithium in hexanes at -78 C and slowly
warmed to room temperature. The resulting lithiated anion is
silylated with chlorotriethoxysilane in THF solution to yield
FSiHD-102. FSiHD-102 in dry THF solution is cooled to 0 C and
lithium aluminum hydride in THF solution is added to cleave the
mesitoate esters and reduce the triethoxysilyl group, yielding
FSiHD-103, the SiH3DonationO tip.
[0152] FIG. 23 depicts a synthetic pathway for GeH3DonationO. The
synthesis steps are as follows: To form FGeHD-101, the germanium
halide FHD-104X in THF solution is reduced with lithium metal to
generate a lithiated germanium species in situ. The solution is
then removed by syringe to separate the lithiated germanium species
from the unreacted lithium metal and then slowly added dropwise to
a solution of chloro(phenyl)germane (Ohshita, J.; Toyoshima, Y.;
Iwata, A.; Tang, H.; Kunai, A. Chem. Lett. 2001, 886-887) in THF
cooled to 0 C. It is necessary to separate the lithiated germanium
species from excess lithium metal before addition to the
trimethylgermanium chloride because lithium is capable of exchange
reactions with germanium halides. FGeHD-101 is dephenylated with
trifluoromethanesufonic acid in dichloromethane at 0 C. The crude
reaction isolate after neutralization and workup is then dissolved
in dry THF. The reaction is cooled to 0 C and lithium aluminum
hydride is added dropwise to produce the germane FGeHD-102, the
GeH3DonationO tip.
[0153] FIG. 24 depicts a synthetic pathway for SiMe3DonationO. The
synthesis steps are as follows: To prepare FSiHD-101, the phenols
of FHD-106 are acylated with mesitoyl chloride in dichloromethane
with pyridine base. (Corey et al., JACS 1969, 91, 4398) The
mesitoate protecting group is utilized due to its stability to the
lithiation conditions necessary for FSiHD-102. FSiHD-101 in dry THF
solution is deprotonated with n-butyllithium in hexanes at -78 C
and slowly warmed to room temperature. The resulting lithiated
anion is silylated with trimethylsilyl chloride in THF solution to
yield FSiMeD-102. FSiMeD-102 in dry THF solution is cooled to 0 C
and lithium aluminum hydride in THF solution is added to cleave the
mesitoate esters, yielding FSiMeD-103, the SiMe3DonationO tip.
[0154] FIG. 25 depicts a synthetic pathway for GeMe3DonationO. The
synthesis steps are as follows: To prepare FGeMeD-101, the
germanium halide FHD-104X in THF solution is reduced with lithium
metal to generate a lithiated germanium species in situ. The
solution is then removed by syringe to separate the lithiated
germanium species from the unreacted lithium metal and then slowly
added dropwise to a solution of trimethylgermanium chloride in THF
cooled to 0 C. It is necessary to separate the lithiated germanium
species from excess lithium metal before addition to the
trimethylgermanium chloride because lithium is capable of exchange
reactions with germanium halides. Stoichiometric
tetra-n-butylammonium fluoride is used to deprotect the
tert-butyldimethylsilyl protecting groups from FMeD-101 in THF to
yield the triphenol FGeMeD-102, the GeMe3DonationO tip.
[0155] FIG. 26 depicts a synthetic pathway for AbstractionNH. The
synthesis steps are as follows: Commercially available
1,3,5-trihydroxyadamantane reacts with 2,4,6-trifluoroaniline while
heated to 50-80.degree. C. under acidic conditions in
1,2-dichloroethane to give NFA-1. Treating NFA-1 tetrafluoroboric
acid forms the tetrafluoroborate amine salt in situ to prevent
oxidation of the amines (Asencio, G., Gonzalez-Nunez, M. E.,
Bernardini, C. B., Mello, R., Adam, W. J. Am. Chem. Soc., 1993,
115, 7250-7253) Following the salt formation, an excess of
dimethyldioxirane (DMDO) in acetone at room temperature selectively
oxidizes the tertiary C--H bond to give alcohol NFA-2. Using
Koch-Haaf conditions (Stetter, H., Schwarz, M., Hirschhorn, A.
Chem. Ber. 1959, 92, 1629-1635), CO is formed from the dehydration
of formic acid by concentrated sulfuric acid. The CO forms a bond
with the tertiary carbocation formed from the dehydration of the
bridgehead alcohol. Upon aqueous workup the carboxylic acid NFA-3
is obtained. Esterification of NFA-3 with dry methanol and
catalytic sulfuric acid yields the ester NFA-4 that can be reduced
readily with diisobutylaluminum hydride. Di-tert-butyl-dicarbonate
(Boc2O) is used to protect the --NH2 groups and to be removable by
acid hydrolysis. Treating NFA-4 with Boc2O yields the protected
compound NFA-5. Reduction of the methyl ester with LiAlH4 in
tetrahydrofuran (THF) gives the methyl alcohol NFA-6. Oxidation of
the methyl alcohol to the aldehyde NFA-7 proceeds with catalytic
tetrapropylammonium perruthenate (TPAP) and stoichiometric
N-methylmorpholine-N-oxide (NMO). The presence of 4 .ANG. powdered
molecular sieves in the reaction mixture adsorbs any water present
and decreases the probability of over-oxidation to the carboxylic
acid. (Ley, S. V., Norman, J., Griffith, W. P., Marsden, S. P.,
Synthesis, 1994, 639-666) Using a modified Corey-Fuchs procedure
(Michel, P., Rassat, A. Tetrahedron Lett. 1999, 40, 8570-8581), the
aldehyde in THF is added to a premixed solution of iodoform (CHI3),
triphenylphosphine, and potassium tert-butoxide at room temperature
in THF to undergo a carbon-carbon bond forming reaction to give the
1,1-diiodoalkene. Single elimination of iodide with careful
temperature (-78.degree. to -50.degree. C.) and excess potassium
tert-butoxide control yields the iodoalkyne NFA-8. It is possible
to form the terminal alkyne from this reaction if temperature is
not carefully controlled, however, the terminal alkyne can be
iodinated with N-iodosuccinimide/AgNO3 or, alternatively, with 12
in basic methanol. The final global deprotection of the Boc- groups
is performed with trifluoroacetic acid (TFA) in dichloromethane at
RT. Upon aqueous workup, NFA-9, the AbstractionNH tip, is
obtained.
[0156] FIG. 27 depicts a synthetic pathway for HDonationNH. The
synthesis steps are as follows: NHD-103X in dry THF solution is
cooled to 0 C and lithium aluminum hydride in THF solution is added
to reduce the germanium halide, yielding NHD-104. NHD-104 is
dissolved in dry MeOH and added to a reaction vessel suitable for
pressurized hydrogenations. Palladium hydroxide catalyst is added
and the vessel pressurized with hydrogen gas. Agitation of the
reaction under the pressurized hydrogen atmosphere yields NHD-105,
the HDonationNH tip.
[0157] FIG. 28 depicts a synthetic pathway for C2DonationNH. The
synthesis steps are as follows: (Triisopropylsilyl)acetylene is
dissolved in dry THF and cooled to -78 C. n-Butyllithium solution
in hexanes is slowly added dropwise to deprotonate the acetylene
hydrogen. The solution is stirred for 1 hour, allowed to warm to
room temperature, and is added dropwise to NHD-103X in dry THF
solution cooled to -78 C. The reaction is stirred for 1 hour,
warmed to 0 C for 1 hour, and stirred for 1 hour at room
temperature to form NC2D-101. NC2D-101 is dissolved in dry MeOH and
added to a reaction vessel suitable for pressurized hydrogenations.
Palladium hydroxide catalyst is added and the vessel pressurized
with hydrogen gas. Agitation of the reaction under the pressurized
hydrogen atmosphere yields NC2D-102. The steric bulk of both the
triisopropylsilyl group and the germaadamantane core prevent
hydrogenation of the alkyne. NC2D-102 is dissolved in THF and
stirred rapidly at room temperature. Tetra-n-butylammonium fluoride
is added and the reaction is stirred for 1 hour at RT to yield
NC2D-103. NC2D-103 is dissolved in MeOH and rapidly stirred.
Potassium hydroxide is added and a solution of iodine in methanol
is added slowly dropwise at RT to yield NC2D-104, the C2DonationNH
tip.
[0158] FIG. 29 depicts a synthetic pathway for MeDonationNH. The
synthesis steps are as follows: The germanium halide NHD-103X in
THF solution is reduced with lithium metal to generate a lithiated
germanium species in situ. The solution is then slowly added
dropwise to a solution of 10-fold excess methylene iodine (CH2I2)
in THF cooled to 0 C. This method of addition favors the formation
iodomethyl germane NMeD-101 over methylene-bridged germanes.
NMeD-101 is dissolved in dry MeOH and added to a reaction vessel
suitable for pressurized hydrogenations. Palladium hydroxide
catalyst is added and the vessel pressurized with hydrogen gas.
Agitation of the reaction under the pressurized hydrogen atmosphere
yields NMeD-102, the MeDonationNH tip.
[0159] FIG. 30 depicts a synthetic pathway for SiH3DonationNH. The
synthesis steps are as follows: The germanium halide NHD-103X in
THF solution is reduced with lithium metal at -78 C to generate a
lithiated germanium species in situ. The solution is then removed
by syringe to separate the lithiated germanium species from the
unreacted lithium metal and then slowly added dropwise to a
solution of excess chlorotriethoxysilane in THF cooled to 0 C and
the reaction is allowed to warm to room temperature to produce
NSiHD-101. NSiHD-101 in THF solution cooled to 0 C is reduced with
lithium aluminum hydride to generate NSiHD-102. NSiHD-102 is
dissolved in cyclohexane and added to a reaction vessel suitable
for pressurized hydrogenations. Palladium hydroxide catalyst is
added and the vessel pressurized with hydrogen gas. Agitation of
the reaction under the pressurized hydrogen atmosphere yields
NSiHD-103, the SiH3DonationNH tip.
[0160] FIG. 31 depicts a synthetic pathway for GeH3DonationNH. The
synthesis steps are as follows: The germanium halide NHD-103X in
THF solution is reduced with lithium metal at -78 C to generate a
lithiated germanium species in situ. The solution is then removed
by syringe to separate the lithiated germanium species from the
unreacted lithium metal and then slowly added dropwise to a
solution of chloro(phenyl)germane in THF cooled to 0 C and the
reaction is allowed to warm to room temperature to produce
NGeHD-101. It is necessary to separate the lithiated germanium
species from excess lithium metal before addition to the
trimethylgermanium chloride to prevent lithium-halogen exchange
reactivity with the chloro(phenyl)germane. NGeHD-101 is
dephenylated with trifluoromethanesufonic acid at 0 C. The crude
reaction isolate after neutralization of acid and workup is then
dissolved in dry THF. The reaction is cooled to 0 C and lithium
aluminum hydride is added to produce the germane NGeHD-102.
NGeHD-102 is dissolved in cyclohexane and added to a reaction
vessel suitable for pressurized hydrogenations. Palladium hydroxide
catalyst is added and the vessel pressurized with hydrogen gas.
Agitation of the reaction under the pressurized hydrogen atmosphere
yields NGeHD-103, the GeH3DonationNH tip.
[0161] FIG. 32 depicts a synthetic pathway for SiMe3DonationNH. The
synthesis steps are as follows: The germanium halide NHD-103X in
THF solution is reduced with lithium metal at -78 C to generate a
lithiated germanium species in situ. The solution is then removed
by syringe to separate the lithiated germanium species from the
unreacted lithium metal and then slowly added dropwise to a
solution of excess chlorotrimethylsilane in THF cooled to 0 C and
the reaction is allowed to warm to room temperature to produce
NSiMeD-101. NSiMeD-101 is dissolved in cyclohexane and added to a
reaction vessel suitable for pressurized hydrogenations. Palladium
hydroxide catalyst is added and the vessel pressurized with
hydrogen gas. Agitation of the reaction under the pressurized
hydrogen atmosphere yields NSiMeD-102, the SiMe3DonationNH tip.
[0162] FIG. 33 depicts a synthetic pathway for GeMe3DonationNH. The
synthesis steps are as follows: The germanium halide NHD-103X in
THF solution is reduced with lithium metal at -78 C to generate a
lithiated germanium species in situ. The solution is then removed
by syringe to separate the lithiated germanium species from the
unreacted lithium metal and then slowly added dropwise to a
solution of trimethylgermanium chloride in THF cooled to 0 C and
the reaction is allowed to warm to room temperature to produce
NGeMeD-101. It is necessary to separate the lithiated germanium
species from excess lithium metal before addition to the
trimethylgermanium chloride to prevent lithium reduction of the
germanium chloride. NGeMeD-101 is dissolved in cyclohexane and
added to a reaction vessel suitable for pressurized hydrogenations.
Palladium hydroxide catalyst is added and the vessel pressurized
with hydrogen gas. Agitation of the reaction under the pressurized
hydrogen atmosphere yields NGeMeD-102, the GeMe3DonationNH tip.
[0163] FIG. 34 depicts a synthetic pathway for AbstractionS. The
synthesis steps are as follows: Commercially available
1-bromoadamantane undergoes a Friedel-Crafts alkylation with three
separate benzene molecules under Lewis acidic conditions with AlCl3
at 90 C to yield SHA-1. Careful control of the stoichiometry of the
tert-butyl bromide (2.0 equivalents) yields the 1,3,5-triphenyl
adamantane (Newman, H. Synthesis, 1972, 12, 692-693). Treatment of
SHA-1 in fluorobenzene and 50% aqueous NaOH solution with a phase
transfer catalyst gives SHA-2. This reaction is selective at
brominating the tertiary C--H bond in the adamantane (Schreiner, P.
R.; Lauenstein, O.; Butova, E. D.; Gunchenko, P. A.; Kolomitsin, I.
V.; Wittkopp, A.; Feder, G.; Fokin, A. A., Chem. Eur. J. 2001, 7,
4996-5003). Oxidative cleavage of the aromatic rings by RuCl3 in a
biphasic mixture gives the tricarboxylic acid SHA-3 (Carlsen, P. H.
J.; Katsuki, T.; Martin, V. S.; Sharpless, K. B., J. Org. Chem.
1981, 46, 3936-3938). Esterification of SHA-3 with dry methanol and
catalytic sulfuric acid between 50-60.degree. C. yields the
triester SHA-4 that can be reduced readily with LiAlH4 at 0 C. The
triol SHA-4 can react readily with triflic anhydride and pyridine
in dichloromethane at 0 C to give the compound SHA-5. Condensing
vinyl bromide at -20.degree. C. with catalytic AlBr3 in the
presence of the adamantyl bromide SHA-5 gives a
dibromoethyladamantane intermediate that is used with potassium
tert-butoxide to eliminate to give the alkyne SHA-6 (Malik, A. A.;
Archibald, T. G.; Baum, K.; Unroe, M. R., J. Polymer Sci. Part A:
Polymer Chem. 1992, 30, 1747-1754). Three equivalents of potassium
thioacetate displaces the triflate groups in refluxing acetonitrile
to give the compound SHA-7. The use of 18-crown-6 enhances the
nucleophilicity of the thioacetate and can be added to enhance the
rate of the reaction at room temperature (Kitagawa, T., Idomoto,
Y.; Matsubara, H.; Hobara, D.; Kakiuchi, T.; Okazaki, T.; Komatsu,
K., J. Org. Chem. 2006, 71, 1362-1369). Silver nitrate with
N-iodosuccinimide in THF creates the iodoalkyne at room temperature
and treatment with potassium hydroxide removes the acetate groups
to give compound SHA-8, the AbstractionS tip.
[0164] FIG. 35 depicts a synthetic pathway for HDonationS. The
synthesis steps are as follows: Allowing RHD-101 to react with
benzene and trifluoroacetic acid (TFA) at room temperature in
dichloromethane forms the triphenylgermaadamantane SHD-101.
Oxidative cleavage of the phenyl groups with catalytic RuCl3 in a
solvent mixture of CCl4, CH3CN, and H2O with periodic acid added as
stoichiometric oxidant cleaves the aromatic rings between 0.degree.
C. to room temperature gives the tricarboxylic acid SHD-102.
Esterification of SHD-102 with methanol with sulfuric acid between
40-60.degree. C. gives the triester that can subsequently be
reduced with LiAlH4 at 0.degree. C. to give the triol SHD-103.
Triol SHD-103 can be treated with triflic anhydride at 0.degree. C.
with pyridine in dichloromethane to give the triflate SHD-104.
Displacement of the triflate groups with potassium thioacetate in
the presence of 18-crown-6 ether in acetonitrile at room
temperature yields the acetate-protected thiols in SHD-105.
Treatment of SHD-105 with a Lewis acid source including to but not
limited to SnCl4, I2, or Br2 in dichloromethane at -78.degree. C.
to room temperature selectively cleaves the Ge-Me bond to give the
respective Ge--X (X=Cl, Br, I) bond in SHD-106X. Treating the
resulting Ge--X compound SHD-106X with LiAlH4 at 0.degree. C. to
room temperature reduces the Ge--X bond as well as simultaneously
removing the thioacetate groups from the thiols to yield the
trithiol SHD-107, the HDonationS tip, upon aqueous workup.
[0165] FIG. 36 depicts a synthetic pathway for C2DonationS. The
synthesis steps are as follows: The intermediate SHD-106X from the
HDonationS synthesis is allowed to react with an excess of
commercially available ethynylmagnesium bromide solution in diethyl
ether at 0.degree. C. to room temperature to form SC2D-101. The
excess of the ethynylmagnesium bromide ensures full deprotection of
the thioacetate protective groups upon aqueous workup. The thiols
in SC2D-101 are protected with acetate groups by treating it with
acetic anhydride (Ac2O). The protected compound is then treated
with silver nitrate and a slight excess of N-iodosuccinimide in THF
at room temperature to form the iodoalkyne in SC2D-102. Subsequent
treatment of the crude reaction mixture in basic methanol at room
temperature yields SC2D-102, the C2DonationS tip.
[0166] FIG. 37 depicts a synthetic pathway for MeDonationS. The
synthesis steps are as follows: The synthesis of the thiol methyl
donation tool begins from intermediate SHD-105. The acetate groups
must be exchanged with a thioether protective group, specifically
the tert-butyl group, to withstand the synthetic conditions. The
acetate groups are removed in basic methanol at room temperature
and then subsequently treated with an acidic solution of
tert-butanol at room temperature to form SMeD-101. The Ge-Me bond
is cleaved with a Lewis acid between -78.degree. C. and room
temperature with a reagent such as SnCl4, I2, or Br2 to yield the
Ge--Cl bond in SMeD-102X. Treating SMeD-102X with lithium metal and
excess CH2I2 at 0 C in THF at high dilution yields SMeD-103.
Removal of the tert-butyl groups is performed with
2-nitrobenzenesulfenyl chloride in acetic acid and yields a mixed
disulfide (Pastuszak, J. J., Chimiak, A., J. Org. Chem., 1981, 46,
1868. Quintela, J. M., Peinador, C., Tetrahedron, 1996, 52, 10497).
Treating this disulfide with NaBH4 at low temperature between
-20.degree. C. and 0.degree. C. allows the recovery of the free
thiol SMeDon-104, the MeDonationS tip, without reducing the C--I
bond.
[0167] FIG. 38 depicts a synthetic pathway for SiH3DonationS. The
synthesis steps are as follows: Intermediate SMeD-102X with t-butyl
protected thiols is treated with lithium metal in THF at 0.degree.
C. followed by the addition of triethoxychlorosilane to give
SSiHD-101 upon workup. This reaction forms the Ge--Si bond
necessary for the SiH3 donor. The removal of the t-butyl groups is
performed with the reagent 2-nitrobenzenesulfenyl chloride in
acetic acid at room temperature to give the mixed disulfide.
Treatment with LiAlH4 cleaves the S--S bonds to give the free
thiols in SSiHD-102, the SiH3DonationS tip, as well as
simultaneously reducing the triethoxysilyl group to --SiH3.
[0168] FIG. 39 depicts a synthetic pathway for GeH3DonationS. The
synthesis steps are as follows: Intermediate SMeD-102X with t-butyl
protected thiols is treated with lithium metal in THF at
-78.degree. C. The solution is then removed by syringe to separate
the lithiated germanium species from the unreacted lithium metal
and then slowly added dropwise to a solution of PhGeH2Cl at
0.degree. C. to give SGeHD-101 upon workup. This reaction forms the
Ge--Ge bond necessary for the --GeH3 donor. Treatment of SGeHD-101
with triflic acid cleaves the Ph-Ge bond to form a Ge--OSO2CF3
bond. Triflic acid also removes of the t-butyl thioether groups.
Treatment of the this intermediate with LiAlH4 in diethyl ether at
0.degree. C. cleaves any S--S bonds to give the free thiols in
SGeHD-102, the GeH3DonationS tip, as well as simultaneously
reducing the Ge triflate group to --GeH3.
[0169] FIG. 40 depicts a synthetic pathway for SiMe3DonationS. The
synthesis steps are as follows: Intermediate SMeD-102X with t-butyl
protected thiols is treated with lithium metal in THF at -78 C
followed by the addition of chlorotrimethylsilane upon warming to
0.degree. C. Upon workup the compound SSiMeD-101 with the Ge--Si
bond is obtained. The removal of the t-butyl groups is performed
with the reagent 2-nitrobenzenesulfenyl chloride in acetic acid at
room temperature to give the mixed disulfide. Treatment with NaBH4
in chloroform and methanol at room temperature cleaves the S--S
bonds to give the free thiols in SSiMeD-102, the SiMe3DonationS
tip.
[0170] FIG. 41 depicts a synthetic pathway for GeMe3DonationS. The
synthesis steps are as follows: Intermediate SMeD-102X with t-butyl
protected thiols is treated with lithium metal in THF at -78 C. The
solution is then removed by syringe to separate the lithiated
germanium species from the unreacted lithium metal and then slowly
added dropwise to a solution of chlorotrimethylgermane at 0 C. Upon
workup the compound SGeMeD-101 with the Ge--Ge bond is obtained.
The removal of the t-butyl groups is performed with the reagent
2-nitrobenzenesulfenyl chloride in acetic acid at room temperature
to give the mixed disulfide. Treatment with NaBH4 in chloroform and
methanol at room temperature cleaves the S--S bonds to give the
free thiols in SGeMeD-102, the GeMe3DonationS tip.
[0171] FIG. 42 depicts a synthetic pathway for intermediate
FHD-104X, from which some of the other syntheses begin. The
synthesis steps are as follows: Cis, cis-Tri-O-alkyl
1,3,5-Cyclohexanetricarboxylate is reduced with lithium aluminum
hydride in refluxing THF and vigorous mechanical stirring to yield
cis, cis-1,3,5-tris(hydroxymethyl)cyclohexane HD-1. The procedure
used resembles that found in Boudjouk et al., Organometallics 1983,
2, 336. Cis, cis-1,3,5-Tris(hydroxymethyl)cyclohexane, HD-1, is
brominated utilizing triphenylphosphine dibromide generated in
situ. This is accomplished by slow addition of bromine to a
solution of the triol and triphenylphosphine in DMF at room
temperature to yield cis, cis-1,3,5-tris(bromomethyl)cyclohexane,
HD-2. The procedure used resembles that found in Boudjouk et al.,
Organometallics 1983, 2, 336. The tri-Grignard is generated in situ
by adding cis, cis-1,3,5-Tris(bromomethyl)cyclohexane, HD-2, at
room temperature to magnesium turnings in THF and heating to
reflux. The tri-Grignard is then transferred to a second reaction
vessel to separate the reagent from the excess magnesium turnings
(Mg is capable of inserting into a Ge--Cl bond).
Trimethylchlorogermane, previously dried over calcium hydride and
degassed, is added slowly dropwise to the reaction at 0 C. After 2
hours, the reaction is warmed to room temperature for two hours,
and finally refluxed overnight. The reaction yields predominantly
cis, cis-1,3,5-Tris(trimethylgermylmethyl)cyclohexane, HD-3. Cis,
cis-1,3-dimethyl-5-(trimethylgermylmethyl)cyclohexane and cis,
cis-1-methyl-3,5-bis(trimethylgermylmethyl)cyclohexane are also
produced in small amounts. The procedure used is similar to that
found in Boudjouk and Kapfer, Journal of Organometallic Chemistry,
1983, 296, 339. HD-3 in benzene solution is subjected to
redistribution reaction conditions using high purity anhydrous
aluminum trichloride and heating to reflux to yield
1-methyl-1-germaadamantane. HD-3 side products cis,
cis-1,3-dimethyl-5-(trimethylgermylmethyl)cyclohexane and cis,
cis-1-methyl-3,5-bis(trimethylgermylmethyl)cyclohexane may also be
present in the reaction or isolated and reacted under these
conditions to yield HD-4 as well. HD-4 is reacted with excess
"ketone free" dimethyldioxirane (DMDO) (Crandall, J. K. 2005.
Dimethyldioxirane. e-EROS Encyclopedia of Reagents for Organic
Synthesis.) in methylene chloride solution at -20 C to yield
1-methyl-3,5,7-trihydroxy-1-germaadamantane RHD-101. The absence of
acetone in the reaction conditions allows for RHD-101 to
precipitate from the reaction mixture, preventing over-oxidation.
Upon completion of the reaction, isopropyl alcohol is used to
quench the excess DMDO, preventing over-oxidation by excess reagent
during reaction workup. RHD-101 is subjected to strongly acidic
conditions in the presence of 2,4,6-trifluorophenol at room
temperature to yield FHD-102. The use of Bronsted acidic conditions
favors carbocation formation at the 3,5,7 bridgehead positions of
the adamantane cage structure over redistribution reactivity at the
germanium center. The 1-methyl group of FHD-102 can be exchanged
with a halide (X=F, Cl, Br, I) with a variety of electrophilic
reagents at low temperatures ranging from -78 C up to room
temperature, depending on the halide desired. Reagents include, but
are not limited to: Lewis acids such as SnCl4 or GaCl3, elemental
halides Br2 and 12 with Lewis acid catalyst, alkyl halides such as
isopropyl chloride with Lewis acid catalyst, and interhalogen
compounds such as IBr and ICl. Furthermore, heavier FHD-103X
halides can be converted to lighter halides utilizing the
appropriate lighter silver halide (e.g. FHD-103Br and AgCl will
produce FHD-103Cl). The phenolic alcohols of FHD-103X (X=F, Cl, Br,
I) can be protected utilizing tert-butyl(chloro)diphenysilane and
imidazole in DMF at RT to yield FHD-104X (X=F, Cl, Br, I).
[0172] FIG. 43 depicts a synthetic pathway for intermediate
NHD-103X, from which some of the other syntheses begin. The
synthesis steps are as follows: RHD-101 is subjected to strongly
acidic conditions such as methanesulfonic acid in the presence of
2,4,6-trifluoroaniline at room temperature to yield NHD-102. The
use of Bronsted acidic conditions favors carbocation formation at
the 3,5,7 bridgehead positions of the adamantane cage structure
over redistribution reactivity at the germanium center. To form
NHD-103, NHD-102 is alkylated at room temperature with
4-methoxybenzyl bromide in DMF with potassium carbonate base in the
presence of potassium iodide. To form NHD-103X, the 1-methyl group
of NHD-103 can be exchanged with a halide (X=F, Cl, Br, I) with a
variety of electrophilic reagents at low temperatures ranging from
-78 C up to room temperature depending on the halide desired.
Reagents include, but are not limited to: Lewis acids such as SnCl4
or GaCl3, elemental halides Br2 and I2 with Lewis acid catalyst,
alkyl halides such as isopropyl chloride with Lewis acid catalyst,
and interhalogen compounds such as IBr and ICl. Furthermore,
heavier NHD-103X halides can be converted to lighter halides
utilizing the appropriate lighter silver halide (e.g. NHD-103Br and
AgCl will produce NHD-103Cl).
Surface Preparation
[0173] Various exemplary surfaces are described herein, including
diamond, silicon and gold. Preferably, these surfaces would more
specifically be depassivated diamond, partially-hydrogenated
partially-chlorinated Si(111), and Au(111). Of course, similar
surfaces could be used, including germanium, and lead, although
they may require leg or linker modifications.
[0174] With respect to diamond, methods for obtaining surfaces
appropriate for both presentation of tips and building of
workpieces are well known in the literature (for example, see
(Hayashi, Yamanaka et al., "Atomic force microscopy study of
atomically flat (001) diamond surfaces treated with hydrogen
plasma," Applied Surface Science. 1998. 125:120-124; Watanabe,
Takeuchi et al., "Homoepitaxial diamond film with an atomically
flat surface over a large area," Diamond and Related Materials.
1999. 8:1272-1276; Okushi, "High quality homoepitaxial CVD diamond
for electronic devices," Diamond and Related Materials. 2001.
10:281-288; Tokuda, Umezawa et al., "Atomically flat diamond (111)
surface formation by homoepitaxial lateral growth," Diamond and
Related Materials. 2008. 17:1051-1054; Yatsui, Nomura et al.,
"Realization of an atomically flat surface of diamond using dressed
photon-phonon etching," Journal of Physics D: Applied Physics.
2012. 45:475302)).
[0175] Partially-hydrogenated partially-chlorinated Si(111) is used
in preference to a fully-chlorinated Si surface because the partial
chlorination reduces the energy barrier to the tip molecules
binding as compared to just chlorinated Si(111) because the
hydrogen, being smaller in size than Cl, helps reduce steric
congestion as the tip approaches the surface. Hydrogenation is
preferably in the 33%-50% range, although wider ranges will work,
as will not using hydrogenation at all. Partially hydrogenated
partially-chlorinated Si(111) can be prepared in a number of ways.
One is the following.
[0176] Clean, atomically flat doped Si(111) surfaces are prepared
by direct current annealing the Si for several hours at .about.650
C followed by rapid heating to .about.1200 C for 1-20 sec while
keeping the chamber pressure <1.times.10-9 Torr. This procedure
gives the 7.times.7 reconstructed Si(111) surface, as in J Phys
Cond Matt 26, 394001 (2014).
[0177] The Si(111) surface can be chlorinated by depositing C12
from an electrochemical cell similar to the one in J Vac Sci and
Tech A 1, 1554 (1983), while the Si(111) is heated to .about.400 C.
Atomically flat halogenated Si(111) surfaces have been prepared
this way, as in Phys Rev Lett 78, 98 (1997).
[0178] Si(111)-Cl surfaces can then be partially hydrogenated by
exposing the surface to 600 L of atomic hydrogen from a H2 cracker,
as in Surf Sci 402-404, 170-173 (1998), with the Si(111)-Cl at room
temperature.
[0179] Clean, atomically flat Au(111) surfaces are prepared by
repeated cycles of sputtering and annealing a single crystal
Au(111) surface, as in Phys Rev Lett 80, 1469 (1998).
Tip Bonding
[0180] Once synthesized, a tip can be bound to a presentation
surface, including large surfaces, and smaller surfaces such as
meta-tips or a single-tip tool surface. Many ways of binding tips
to surfaces are possible, and these may vary with the exact nature
of the tip and the surface.
[0181] One method of depositing isolated tips on a surface is via
thermal evaporation in vacuum. In this technique, purified
molecules in the form of a solid or liquid are heated up in a
vacuum chamber until they evaporate as a gas of isolated molecules.
By placing the presentation surface within this gas, individual
tips will adhere to the surface. (See tetramantane deposition as
described in "Spatially resolved electronic and vibronic properties
of single diamondoid molecules," Nature Materials 7, 38-42 (2008)).
This method has the advantage of depositing molecules without
surface contamination from a solvent and can be used with masks.
The use of masks allows creating sectors which could each contain
different tips, or different mixtures of tips, allowing for logical
and efficient layout of tips.
[0182] The tips having sulfur or thiol-based linkers will bond to
gold spontaneously at room temperature. The tips with O or NH
linkers designed to bond to chlorinated silicon surfaces require
heating of the surface to overcome reaction barriers. This is the
reason partial hydrogenation/chlorination is favored: The reduction
in steric interference keeps the reaction barrier to tip binding as
far below the tip decomposition temperature as possible.
[0183] A simple way to evaporate molecules is to place the
molecules in a glass or alumina crucible with a tungsten wire
wrapped around the crucible. Passing a current through the wire
heats the crucible and molecules, generating a molecular gas that
exits the front of the crucible. A thermocouple on the crucible
measures its temperature. A quartz crystal microbalance can be used
to determine how much is evaporating as a function of time and
temperature.
[0184] This is just one example of how tips could be bonded to a
surface. Such techniques, including how to create sectors of
specific molecules, are well-known in the respective arts. (Zahl,
Bammerlin et al., "All-in-one static and dynamic nanostencil atomic
force microscopy/scanning tunneling microscopy system," Review of
Scientific Instruments. 2005. 76:023707; Sidler, Cvetkovic et al.,
"Organic thin film transistors on flexible polyimide substrates
fabricated by full-wafer stencil lithography," Sensors and
Actuators A: Physical. 2010. 162:155-159; Vazquez-Mena, Gross et
al., "Resistless nanofabrication by stencil lithography: A review,"
Microelectronic Engineering. 2015. 132:236-254; Yesilkoy, Flauraud
et al., "3D nanostructures fabricated by advanced stencil
lithography," Nanoscale. 2016. 8:4945-50)
Tip Activation
[0185] Tips, particularly those with exposed radicals at their
active site, may be bonded to a surface in an inactive form. One
method of activating such tips is through photo-cleavage of the
structure. For example, the halogen-capped tip examples herein can
be activated through exposure to 254 nm light. FIG. 44 depicts an
activating reaction for halogen-capped tips. Other wavelengths and
chemistries can also be used. For example, if different synthetic
steps were used, a tip could be protected with a Barton ester,
which can then be cleaved, activating the tip, with 365 nm light.
FIG. 45 provides an example of the activation reaction that could
be used with a Barton ester.
[0186] While not the only way to remove a tip cap, photo-activation
is convenient in that different areas of a surface can be masked.
Different wavelengths can also be used, choosing wavelengths which
affect some tips but not others. This makes photo-activation a
versatile technique even when multiple types of tips are present,
or when potentially-complex layout patterns are desired.
Barton Ester Caps
[0187] Other examples are provided herein of synthetic routes to
halogen-capped tips, and how to activate them. To demonstrate
another chemistry for synthesizing tips with protective caps, the
Barton ester is an alternative that fragments upon being irradiated
with, for example, 355-365 nm wavelength light to give the carbon
centered radical, CO2, and the pyrithiyl radical. (Barton, D. H.
R., Crich, D., Potier, P. Tetrahedron Lett., 1985, 26, 5943-5946.
For a review of thiohydroxamic acids chemistry see: Crich, D.,
Quintero, L. Chem. Rev. 1989, 89, 1413-1432) These types of
activated molecules can be made from the described compounds and
one such synthetic route is described below, resulting in the
Barton ester version of the Abstraction.RTM. tip.
[0188] FIG. 46 depicts the synthesis of the Barton ester
Abstraction.RTM. tip, which is as follows: To synthesize the Barton
ester for photoactivation, propynoic acid OFAB-1 is made from OFA-7
using the traditional Corey-Fuchs procedure and quenching by
bubbling gaseous CO2 through the reaction mixture. (Corey, E. J.,
Fuchs, P. L. Tetrahedron Lett. 1972, 36, 3769-3772) The first step
forms the 1,1-dibromoalkene in solution at -78 C. The addition of 2
more equivalents of butyllithium forms the lithium acetylide in the
reaction mixture. By bubbling with the carbon dioxide the desired
carboxylic acid OFAB-1 is obtained after an aqueous workup. To make
the Barton ester, carboxylic acid derivative OFAB-1 is activated to
the acid halide by oxalic acid and catalytic N,N-dimethylformamide
(DMF) in dichloromethane at room temperature. To this reaction
mixture the sodium pyrithione salt is added to the mixture to form
the desired ester bond in compound OFAB-2. The Barton ester is
unstable to aqueous acidic and basic media, so careful control of
reaction conditions must be taken when removing the protective
groups. Multiple techniques are possible for removal of silyl
ethers such as OFAB-2 that are pH sensitive. One is to use more
labile silyl ethers such as trimethylsilyl- (TMS-) or
triethylsilyl- (TES-) ethers in place of the more stable TBS silyl
ethers. Another method is to use OFAB-2 and catalytic solid
tetra-n-butylammonium fluoride (TBAF) or cesium fluoride in 100:1
THF-buffer solution to produce OFAB-3. A solution of K2HPO4
buffered at pH=7.1 could be used in the TBAF deprotection.
(DiLauro, A. M.; Seo, W.; Phillips, S. T., J. Org. Chem. 2011, 76,
7352-7358) This decreases the risk of hydrolyzing the Barton ester
bond and increases the likelihood of obtaining the free phenols in
OFAB-3, the Barton ester AbstractionO tip.
Methods of Tip Use
[0189] One of the ways in which surface mounted tips can be used is
depicted in FIG. 47. This figure is diagrammatic and not to scale.
In FIG. 47, handle 4701 is connected to surface 4702. Surface 4702
is optional, serving to provide the desired materials and chemistry
to bind workpiece 4703 in the case where the material of the handle
is unsuitable for doing this directly. It may be possible to bind
workpiece 4703 directly to handle 4701. Handle 4701 would be
connected to a positional means (not shown) for the purposes of
moving handle 4701, and thereby workpiece 4703 with respect to tips
(of which tip 4704 is representative) mounted on surface 4705.
[0190] In the depicted position, workpiece 4703 could be descending
upon a tip, or it could be rising from just having been acted upon
by a tip. Regardless, the point is that surface 4705 can contain
many tips, of many different types, including non-functional tips
(which either failed to synthesize correctly or have already been
used). Knowledge of tip position, for example, because sectoring
was used to place certain tip types in certain locations, or via
scanning the surface (before or during a build sequence), allows
the workpiece to be moved to a desired tip, at which time a
mechanosynthetic reaction occurs, and the workpiece then moves to
the next desired tip. This process is repeated until the workpiece
is complete.
[0191] Another way to use surface-mounted tips is to create a
meta-tip, which is a handle upon which a plurality of tips may be
mounted, directly, or via a surface. FIG. 48 depicts this mode of
using surface-mounted tips, where handle 4801 is connected to
(optional) surface 4802. Handle 4801 is also connected to a
positional means (not shown). Tips, of which tip 4804 is
representative, are shown mounted on surface 4802, but could be
mounted directly to handle 4801. In this scenario, the tips move to
act upon workpiece 4803, which resides upon surface 4805.
[0192] The main difference between the scenarios of FIG. 47 and
FIG. 48 is whether the workpiece moves or the tips move. In
actuality, it is possible that both move (e.g., one for course
adjustments, one for fine), and the distinction is mainly one of
equipment design.
[0193] FIG. 48 perhaps provides the clearest illustration of the
advantages surface-mounted tips have over previous mechanosynthesis
techniques. If surface 4802 only had one tip affixed to it, it
would be analogous to the tips commonly used for mechanosynthesis.
In this scenario, to create complex workpieces, the affixed tip
would have to a) be capable of multiple reactions and b) be
regenerated frequently, or, frequent tip swapping employed. Using
either the scenario of FIG. 47 or FIG. 48 (and modifications
thereof which would be possible given the teachings herein), many
tips are available to provide mechanosynthetic reactions,
potentially (depending on the number of tips initially available
and the number of reactions required to build the workpiece)
without tip recharge and without tip swapping. Any reduction in tip
recharge or tip swapping can help decrease the average time it
takes to perform a reaction.
Number of Available Tips
[0194] The total number of available tips could span a very wide
range, depending on factors such as the total number of reactions
needed to make a workpiece, the number of different types of
reactions needed to make a workpiece, the available size of the
presentation surface, and the exact methods being used. Also, it is
conceptually important to distinguish between the total number of
available tips, and the number of different types of tips.
[0195] For example, if tip recharge is acceptable, then the number
of tips might be limited to only providing one tip for each type of
reaction needed by a build sequence. For example, as described
herein, one way of building diamond requires four different tips
(and row initiation and termination each take only three tips,
while row extension requires four). Ignoring feedstock and
differences only in legs or linkers, about 7 different types of
tips are described herein. Counting feedstock, given the structures
in Table 1, in addition to those in, e.g., FIGS. 1-17 and FIG. 51,
this number becomes about 20 or more since some tips can use a
variety of feedstocks. Given these examples, it will be obvious
that the number of types of tips present in a system can include
less than 4, 4 to 7, 8 to 20, or more. Note that this says nothing
about the number of positional means in a system, since multiple
types of tips can be affixed to a single positional means.
[0196] Having a single tip of any required type present is useful
for avoiding tip swapping, but not as useful for avoiding tip
recharge. To avoid tip recharge, ideally each type of tip would be
present at least as many times as that tip is used in a build
sequence. Given that build sequences can essentially be arbitrarily
long, this is one example where it becomes useful to have the total
number of tips present be, e.g., 10 to 100 for even quite small
workpieces, and between one hundred and a thousand, or between a
thousand and a million, or between a million and a billion, or
more, for larger workpieces. It can easily be seen by determining
the surface area available to an appropriate system, and the size
of the average tip, that even while allowing for some wasted space
given, for example, imperfect tiling of tips on a presentation
surface and the possible presence of some percentage of defective
tips, the presentation surface can hold a very large number of
tips.
Mechanosynthesis-Adapted Equipment
[0197] Typical commercial atomic microscopy systems combine course
and fine motion controllers to provide both long range of motion,
and atomic resolution. Multi-tip systems are also available (or can
be constructed, for example (Eder, Kotakoski et al., "Probing from
both sides: reshaping the graphene landscape via face-to-face
dual-probe microscopy," Nano Letters. 2013. 13:1934-40)), whereby
more than one tip can be employed simultaneously. For example,
Omicron's (Scienta Omicron GmbH, Germany) LT Nanoprobe provides a
pre-integrated SPM, having 4 probe tips, a course motion controller
with a range of 5 mm.times.5 mm.times.3 mm, a fine motion
controller with a range of 1 um.times.1 um.times.0.3 um, and atomic
resolution in STM mode. Such equipment suffices for
mechanosynthesis work, and given that mechanosynthesis work has
been carried out for decades, even what would currently be
considered outdated equipment can suffice. However, typical SPM
equipment is not optimized for carrying out high-volume
mechanosynthetic reactions. Typical SPM work involves analysis
rather than manufacture, the point generally being to scan
specimens to create an image or collect other data. Scan speed is
frequently the limiting factor, and increasing scan speed is an
active area of research (Dai, Zhu et al., "High-speed metrological
large range AFM," Measurement Science and Technology. 2015.
26:095402).
[0198] Scan speed is less important to systems for mechanosynthesis
as long as the system can obtain the necessary accuracy without
scanning, which is well within the state-of-the-art. Ideally,
systems adapted for mechanosynthesis would not need to scan, at
least for position determination or refinement. Realistically, some
scanning will probably be necessary, including an initial surface
scan to map surface topology and tip location and identity, and, if
desired, small areas around a reaction site could be scanned after
a reaction to verify that the reaction occurred correctly (it
should be noted that this may not be necessary given the extremely
high reliability of many of the exemplary reactions). Note that
such scanning and tip or workpiece characterization capabilities
are clearly present in the state-of-the-art; see for example
(Giessibl, "Forces and frequency shifts in atomic-resolution
dynamic-force microscopy," Physical Review B. American Physical
Society. 1997. 56:16010-16015; Perez, Stich et al., "Surface-tip
interactions in noncontact atomic-force microscopy on reactive
surfaces: Si(111)," PHYSICAL REVIEW B. 1998. 58:10835-10849; Pou,
Ghasemi et al., "Structure and stability of semiconductor tip
apexes for atomic force microscopy," Nanotechnology. 2009.
20:264015; Yurtsever, Sugimoto et al., "Force mapping on a
partially H-covered Si(111)-(7.times.7) surface: Influence of tip
and surface reactivity," Physical Review B. 2013. 87; Hofmann,
Pielmeier et al., "Chemical and crystallographic characterization
of the tip apex in scanning probe microscopy," Phys Rev Lett. 2014.
112:066101; Hapala, Ondra ek et al., "Simultaneous nc-AFM/STM
Measurements with Atomic Resolution," Noncontact Atomic Force
Microscopy: Volume 3. Cham, Springer International Publishing.
2015.29-49).
[0199] Regardless of the fact that some scanning will likely be
used at various points in the mechanosynthetic process, doing away
with frequent scanning for position refinement, and instead using
metrology that allows the requisite point-to-point accuracy
(meaning, moving directly from one tip or workpiece location to
another, without using scanning in between to refine position),
would considerably speed up the process of mechanosynthesis.
[0200] Note that while the ideal attributes for analytical or
metrological SPM are different than those for systems for
mechanosynthesis, even previous work on mechanosynthesis did not
provide systems well-adapted for such work, presumably due to the
simple and low-volume nature of the work being performed, for which
conventional equipment suffices. For example, many commercial
atomic microscopes are open-loop, meaning, they do not use
metrology to refine tip position. However, closed-loop systems are
also available, can be built, or metrology can be added to an
existing open-loop system (e.g., see (Silver, Zou et al.,
"Atomic-resolution measurements with a new tunable diode
laser-based interferometer," Optical Engineering. 2004. 43:79-86)).
Closed-loop systems are generally more accurate due to metrology
feedback and positional means capable of very high accuracy over
large distances are available. For example, piezo elements are
often used to position tips very precisely, and using
interferometry, angstrom or even picometer-level accuracy has been
shown to be possible, even at distances up to 50 mm (Lawall,
"Fabry-Perot metrology for displacements up to 50 mm," J. Opt. Soc.
Am. A. OSA. 2005. 22:2786-2798; Durand, Lawall et al., "Fabry-Perot
Displacement Interferometry for Next-Generation Calculable
Capacitor," Instrumentation and Measurement, IEEE Transactions on.
2011. 60:2673-2677; Durand, Lawall et al., "High-accuracy
Fabry-Perot displacement interferometry using fiber lasers," Meas.
Sci. Technol. 2011. 22:1-6; Chen, Xu et al., "Laser straightness
interferometer system with rotational error compensation and
simultaneous measurement of six degrees of freedom error
parameters," Optics Express. 2015. 23:22) Further, although this
could be unnecessary with high-accuracy closed loop systems,
software capable of compensating for positional errors due to
hysteresis, creep, and other phenomenon is available; for example
see (Mokaberi and Requicha, "Compensation of Scanner Creep and
Hysteresis for AFM Nanomanipulation," IEEE Transactions on
Automation Science and Engineering. 2008. 5:197-206; Randall,
Lyding et al., "Atomic precision lithography on Si," Journal of
Vacuum Science & Technology B: Microelectronics and Nanometer
Structures. 2009. 27:2764; Follin, Taylor et al., "Three-axis
correction of distortion due to positional drift in scanning probe
microscopy," Rev Sci Instrum. 2012. 83:083711). Software also
exists that essentially uses image recognition for positional
refinement; for example see (Lapshin, "Feature-oriented scanning
methodology for probe microscopy and nanotechnology,"
Nanotechnology. 2004. 15:1135-1151; Lapshin, "Automatic drift
elimination in probe microscope images based on techniques of
counter-scanning and topography feature recognition," Measurement
Science and Technology. 2007. 18:907-927; Lapshin,
"Feature-Oriented Scanning Probe Microscopy," Encyclopedia of
Nanoscience and Nanotechnology. 2011. 14:105-115). Ideally, this
would not be necessary since the required scanning would slow down
the overall process, but it is available if desired.
[0201] Note that 50 mm is far longer than the working distance
needed to accommodate a very large number of tips (billions,
trillions, or more) and complex workpieces. Distances on the order
of microns (or even smaller for small workpieces), thousands of
times smaller than the technology has been proven capable of, would
suffice for many types of workpieces.
[0202] In a metrological system, the tip is generally not exactly
at the point being measured (which may be, e.g., a reflective flat
when using laser interferometry), such metrology has to be
carefully implemented to avoid, e.g., Abbe error which can be
induced by slightly non-linear movement of the tip or workpiece
with respect to, e.g., the reflective flat. One way to address this
issue it to measure not only the X, Y and Z coordinates of the
reflective flat, but also to measure (and so be able to account
for) any rotation that might be occurring around these axis as
well.
[0203] One way to measure both linear and angular position is to
use 6 interferometers (e.g., Michelson or Fabry-Perot optical
interferometers). FIG. 49 illustrates one way interferometers can
be used to measure six degrees of freedom (X, Y, and Z, and
rotation about each of those axes).
[0204] In FIG. 49, Reflective mirrors 4901-4906 and, and their
respective beams, BeamZ1 4907, BeamZ2 4908, BeamZ3 4909, BeamX1
4910, BeamY1 4911 and Beam Y2 4912 can be used together to
determine position in all six degrees or freedom. The spacing
between various pairs of beams must be known to compute rotations.
In this scenario, BeamX1 provides the X position. BeamY1 or BeamY2
provide the Y position. BeamZ1, or BeamZ2, or BeamZ3 provides the Z
position. BeamZ1 and BeamZ2, together with the distance between the
two beams allows the rotation about the X axis to be calculated.
BeamZ2 and BeamZ3, together with the distance between the two beams
allows the rotation about the Y axis to be calculated. And, BeamY1
and BeamY2, together with the distance between the two beams allows
the rotation about the Z axis to be calculated.
[0205] Coupling the ability to provide, ideally, sub-Angstrom
linear distance measurement over distances up to the millimeter
scale, while also measuring and accounting for angular errors,
with, for example, a microscope that operates at 4K (room
temperature is feasible but more technically challenging) in
ultra-high vacuum, while using, e.g., a qPlus sensor, provides for
a system that can access precise locations on large presentation
surfaces with a greatly-reduced need to use scanning and image
recognition to refine the relative position of tips and workpieces.
These adaptations themselves are valuable for mechanosynthesis.
Using such equipment with surface-mounted tips and the processes
described herein provides systems adapted for mechanosynthesis that
can provide much greater reaction throughput than conventional
systems.
[0206] Other useful adaptations that are somewhat unique to the
requirements of mechanosynthesis include reducing tip recharge and
reducing tip swapping (which does occur in more conventional uses
of SPM equipment, but normally because a tip has been damaged, not
because many tips of different chemical natures are required).
Surface mounted tips have been discussed herein as one way to
reduce the need for tip recharge and tip swapping.
Sequential Tip Method
[0207] Surface-mounted tips and inverted mode offer important
improvements over conventional mode. However, inverted mode,
because the workpiece is being built on the handle (e.g., an SPM
probe), does have some drawbacks. For example, if the workpiece is
not conductive, some modes such as STM may not be possible. Also,
the geometry of the workpiece can pose a problem. For example, if a
workpiece has a sizeable flat surface adjacent to the site of the
next reaction, as the reaction site on the workpiece approaches the
surface-mounted tips, other portions of the workpiece will also be
approaching other surface-mounted tips, potentially causing
undesired reactions. Ideally, one would like to combine the
benefits of both inverted mode and conventional mode, keeping the
high aspect ratio, versatile mode capabilities and other desirable
characteristics of conventional mode, without sacrificing the
important improvements that inverted mode with surface mounted tips
offers, such as the reduction or elimination of tip swapping due to
the availability of large numbers of any type of tips required for
a given build sequence, and the elimination of feedstock
provisioning and trash depots as separate entities from
surface-mounted tips.
[0208] Obtaining the benefits of both inverted mode with
surface-mounted tips and conventional mode is possible if the tip
thermodynamics are engineered to allow an additional tip-to-tip
feedstock transfer, resulting in what we refer to as a
"thermodynamic cascade." Rather than a surface-mounted tip
interacting directly with the workpiece, the sequential tip method
consists of a surface-mounted tip interacting with a conventional
mode tip. The conventional mode tip interacts with the workpiece.
The surface mounted tips thus serve as what can be conceptualized
as a surface with tunable affinity. Since the surface mounted tips
can be engineered to have any desired affinity for their feedstock,
they can present or accept a much wider range of feedstocks to the
conventional tip than would be possible if the feedstock was
attached directly to the presentation surface. Note that the
workpiece is preferably located on the presentation surface along
with the surface mounted tips, although this is not always true, as
is explained herein.
[0209] FIG. 50a-f shows one way of implementing the sequential tip
method, with sub-FIGS. 50a-e depicting sequential states of the
same system and FIG. 50f showing an overhead view.
[0210] FIG. 50a, which we arbitrarily use as a starting state,
shows handle 5001 (which would be connected to positional control
means, not shown) with a tip 5003 (a conventional mode tip) bound
to its apex. Tip 5003 has an active site 5002, which in this case,
is empty and awaiting feedstock. A presentation surface 5007 holds
tips, of which tip 5004 (an inverted mode tip) is exemplary, and a
workpiece 5006. The tip 5004 includes feedstock 5005.
[0211] In FIG. 50b, handle 5001 and tip 5003 have been positioned
so that active site 5002 binds to feedstock 5005. In other words, a
mechanosynthetic reaction occurs between tip 5003 and feedstock
5005. At this point, feedstock 5005 is bound to both tip 5003 and
tip 5004.
[0212] In FIG. 50c, handle 5001, and thus tip 5003, have been
pulled away from tip 5004, and feedstock 5005 has transferred to
tip 5003. This transfer occurs upon pulling the two tips away from
each other because tip 5003 has been engineered to have greater
affinity for feedstock 5005 than tip 5004.
[0213] In FIG. 50d, handle 5001 brings tip 5003 and its feedstock
5005 to a specific location on workpiece 5006, facilitating a
mechanosynthetic reaction between feedstock 5005 and workpiece
5006. At this point feedstock 5005 is bound to both tip 5003 and
workpiece 5006.
[0214] In FIG. 50e, handle 5001 and tip 5003 have been pulled away
from workpiece 5006, leaving feedstock 5005 bound to workpiece
5006. Like the previous tip-to-tip transfer between tip 5004 and
tip 5003, feedstock 5005 remains bound to workpiece 5006, instead
of pulling away with tip 5003, because tip 5003 has been engineered
to have lower affinity for feedstock 5005 than does the chosen
specific location on workpiece 5006.
[0215] FIG. 50f depicts a top view of the system shown in side
views in FIG. 50a-e. Workpiece 5006 is shown partially under handle
5001 (dotted lines representing the hidden borders of the
workpiece) and tip 5003 (denoted with dotted lines as it is under
handle 5001). Tip 5004 is representative of many surface-mounted
tips arrayed in sectors set off by a grid of dotted lines, such as
exemplary sector 5008. Of course, this is not to scale, nor
necessarily the actual arrangement that would be used. The
workpiece could be next to the surface-mounted tips, in the middle
of the surface-mounted tips, or at any other convenient location,
even on a different presentation surface. The sectors could be
rectangular, concentric, shaped like pie wedges, or any other
convenient shape, or sectors could not exist at all, with tips of
different types being intermingled.
[0216] The addition of the tip-to-tip transfer step may complicate
the system design from a chemical perspective, but overall creates
a more efficient and versatile system. The increased chemical
complexity stems from the fact that to carry out the sequential tip
method, assuming a donation reaction, the affinity of the
surface-mounted tip for the feedstock must be less than the
affinity of the conventional tip for the feedstock (a requirement
that does not exist in conventional or inverted mode, since no
tip-to-tip transfer takes place), and the affinity of the
conventional tip for the feedstock must be less than the affinity
of the workpiece for the feedstock.
[0217] The chemistry is further complicated by the desire to have a
single conventional tip be able to receive many different
feedstocks from surface-mounted tips, and be able to donate those
feedstocks to various specific locations on a workpiece, which may
vary in their chemical nature, and therefore in their affinity for
feedstock. Note that while these reactions are generally described
in terms of a tip donating feedstock to a workpiece, the same
principles apply to abstraction reactions, although the
thermodynamics and sequence of events need to be changed as
appropriate.
[0218] Subsequently, we describe how to design and build tips, both
surface mounted and conventional, that meet the necessary
thermodynamic requirements. We also provide a work-around for
situations where it is not possible or desirable for one
conventional tip to carry out all the reactions of a given build
sequence.
[0219] Note that, while the sequential tip method is generally
described as involving two tips and therefore a single tip-to-tip
transfer for a given reaction on a workpiece, if desired, there is
no reason the sequential tip method could not be performed with
more than two tips as long as the tip affinities are appropriately
designed.
Tip Design for the Sequential Tip Method
[0220] Two types of tips are used in the sequential tip method:
surface-mounted tips and conventional tips. Herein we describe a
set of tips that can be used as surface-mounted tips and allow the
transfer of a wide variety of feedstock (including atoms abstracted
from a workpiece, such as with the AbstractionO, AbstractionNH and
AbstractionS tips). Using these surface-mounted tips as examples,
we now turn to the design of a conventional tip which has an
affinity for many of the various feedstocks which is between that
of the surface-mounted tips and that of an exemplary diamond
workpiece.
[0221] Note that in mechanosynthetic reactions it is not
necessarily the energy levels of the products and reactants that
specify their relative affinities. Bond stiffness is also a factor.
Consider the hypothetical reaction
Tip-F+Workpiece.fwdarw.Tip-+F-Workpiece. It is possible that the
reactants have lower energy than the products. However, the
mechanosynthetic reaction can still be successful if the
F-Workpiece bond is stiffer than the Tip-F bond. In such a case, as
the tip is retracted from the workpiece, the Tip-F bond gradually
stretches and then breaks, unable to overcome the stiffness of the
F-Workpiece bond, even though the overall energy of the Tip-F bond
may be greater. This is not merely hypothetical; some of the
reactions of which the exemplary tips are capable work in this
manner. Given this, affinity is not defined by bond energy. Rather,
we use the practical definition that when two structures (e.g., two
tips, or a tip and a surface, or a tip and a workpiece, or a
workpiece and a surface) are brought together to potentially
transfer feedstock in a mechanosynthetic reaction, the structure to
which the feedstock is bound after the two structures are separated
has the higher affinity for that feedstock.
[0222] FIG. 51 depicts one possible structure of a conventional tip
for use in the sequential tip method. The tip is built on surface
5101 (which would be connected to a positional means, not shown)
and comprises support atoms 5102, 5103 and 5104, and active atom
5105. In this state, active atom 5105 is a radical, ready to e.g.,
bind feedstock from surface-mounted tips, or abstract one or more
atoms from a workpiece. Passivating atom 5106 is used to satisfy
unused valences, and is representative of many such atoms bonded to
the tip and surface.
[0223] In one embodiment, surface 5101 is silicon, support atoms
5102, 5103 and 5104 are carbon, and active atom 5105 is silicon.
For building diamond-based structures, this embodiment has an
affinity which is conveniently between that of the described
surface-mounted tips and the workpiece for multiple different
feedstocks and reactions. In one embodiment passivating atom 5106
and other passivating atoms could be any atom of appropriate
chemical nature such as hydrogen or fluorine.
[0224] We refer to the embodiment where the active atom is silicon,
connected to three support atoms which are carbon, as half-Si-Rad
(because it is a partial or "half" adamantane structure with an
apical silicon radical in its basic form). With various feedstock
attached, the tip can take forms which include half-Si-Rad-CC (a
carbon dimer bound to the active atom, and a radical itself, which
for some reactions actually makes the apical carbon of the carbon
dimer the active atom as it can be used to abstract other atoms
from tips or workpieces), half-Si-Rad-H (a hydrogen bound to the
active atom), and half-Si-Rad-CH2 (CH2 bound to the active atom),
among others.
[0225] Exemplary reactions that various versions of the half-Si-Rad
tip can carry out include: H Abstraction from C(111) to
half-Si-Rad-CC, H Donation to from half-Si-Rad-H to C(111)-Radical,
H Abstraction from C(111)-CH3 to half-Si-Rad-CC, H Donation from
half-Si-Rad-H to C(111)-CH2, CH2 Donation from half-Si-Rad-CH2 to
C(111)-Radical, CH2 Donation from half-Si-Rad to C(111)-CH2 and C2
Dimer Donation from half-Si-Rad-CC to C(111)-Radical.
[0226] While half-Si-Rad can carry out many useful reactions, it is
not capable of carrying out all reactions, particularly when
different classes of workpieces are considered. For example,
silicon bonds tend to be weaker than carbon bonds, and germanium
bonds tend to be weaker still. Given this, for Si- or Ge-based
workpieces, the half-Si-Rad tip will often have an affinity for
feedstock that is higher than the affinity of the workpiece for the
feedstock. This means that it could not donate the feedstock to the
workpiece. A systematic method of adjusting tip affinity would be
useful to assist in the rational design of tips with different
feedstock affinities. There are two main ways of adjusting tip
affinity without departing from the basic bonding structure of the
tip depicted in FIG. 51.
[0227] First, active atom 5105 can be substituted with an atom of
different affinity. For example, to increase the affinity of the
active atom for feedstock, carbon could be substituted for silicon,
and to reduce the affinity of the active atom for feedstock, in
order of descending affinity, germanium, tin, or lead could be used
(although it should be recognized that this is a rule of thumb and
will not be accurate for all tip-feedstock combinations; those
familiar with the relevant arts will understand more nuanced ways
of predicting affinity).
[0228] Second, one or more of the support atoms 5102, 5103 and 5104
can be substituted with different atoms which can affect the
affinity of active atom 5105. For example, the embodiment described
above where the support atoms are each carbon is, for most
diamond-based reactions, superior to an all-silicon tip because the
affinity of the all-silicon tip is lower than desired. The carbon
atoms strengthen the bond between the active atom and the
feedstock. Our computational studies indicate that active atom
affinity for feedstock, in general, is affected by the support
atoms in the following manner: O>N>C>S>P>Si.
Meaning, using oxygen as a support atom results in the highest
affinity of the active atom for the feedstock, while using silicon
results in the lowest affinity of the active atom for the
feedstock, although like the affinity comments above, this is a
rule of thumb. Regardless, this hierarchy provides a useful
starting point for the design of new tips. Obviously, tips with
different basic structures, but with the desired feedstock
affinity, could also be designed given the examples and teachings
herein.
[0229] The ability to rationally design new conventional tips
raises the issue of how these tips can be synthesized and bound to
the positional means. While we could design and affix conventional
tips in a manner like that described for surface-mounted tips, this
would likely mean that multiple handles, each with a different tip,
would be needed. Assuming a single positional means, this implies
that tip swapping would be required. Tip swapping is, as described
herein, preferably avoided. Using equipment with multiple
positional means is one way to overcome this problem. For example,
systems with two to four positional means exist, and if each
positional means was affixed to a tip of different affinity, the
overall set of tips would allow a greater diversity of reactions
than a single tip. However, multiple positional means complicates
equipment design and increases cost. A method to avoid tip swapping
even with only a single positional means may be preferable.
In Situ Tip Synthesis
[0230] Tip swapping can be avoided if conventional tips are
disassembled and reassembled (in modified form, as appropriate) on
the same surface (e.g., a presentation surface connected to a
handle) as needed. For example, if the half-Si-Rad tip described
above was the initial tip bound to a handle, a build sequence could
be carried out up until the point when a tip of different affinity
was needed. At that point, the conventional tip (half-Si-Rad in
this example) essentially becomes a workpiece, with the system
temporarily operating in inverted mode rather than sequential
mode.
[0231] By this, it is meant that the surface-mounted tips act upon
the conventional tip, modifying it as desired. The surface mounted
tip can be used to remove any (or all, creating a completely new
structure) of the atoms in the conventional tip. The
surface-mounted tips then provide the new atoms to manufacture a
tip that can complete the next part of the build sequence. This
process can be repeated as many times as necessary to complete a
build sequence, although preferably the need to change the
conventional tip would be minimized to streamline the manufacturing
process. This suggests a refinement to the process of creating a
build sequence where build sequences are ordered, at least in part,
in a manner that minimizes the need to rebuild the conventional
tips.
[0232] As an example of in situ tip synthesis, FIGS. 52a-o depict a
build sequence which creates the half-Si-Rad tip starting from a
depassivated silicon surface. Depassivated silicon surfaces are
well-known in the relevant fields, and can be created via bulk
chemical methods or heating. Also, a patch of depassivated silicon
atoms could be created using mechanosynthesis. For example,
starting with a conventional passivated silicon probe, three
hydrogens could be removed from a small flat area on the apical end
via the abstraction tips described herein.
[0233] In FIG. 52a, an exemplary silicon structure is depicted as a
stand-alone structure terminated with passivating hydrogens, of
which hydrogen atom 5201 is representative, except on its lower
face, which is depassivated. In reality, the structure depicted
would be part of a larger structure (which may itself be connected
to larger structures such as a handle and positioning means), but
only the small area needed for a presentation surface is shown for
clarity. Three depassivated silicon atoms are present, of which
silicon atom 5202 is representative. This silicon structure, with
its small patch of depassivated silicon atoms, serves as the
starting point for building the half-Si-rad tip.
[0234] In FIG. 52b, a bromine atom is donated to one of the
depassivated silicon atoms. This can be accomplished with a tip
comprising an adamantane body with a carbon radical active site, to
which a bromine atom has been bound. We will refer to this tip as
AdamRad-Br.
[0235] In FIG. 52c, another bromine atom has been added to one of
the other depassivated silicon atoms, also using an AdamRad-Br.
[0236] In FIG. 52d, the third and final bromine is added to the
last unpassivated silicon atom, again using AdamRad-Br.
[0237] Note that the three bromine atoms which were added in the
first three steps of this sequence will end up being removed. This
raises the question of why the bromine atoms were added in the
first place. The reason is that it is preferable to satisfy the
valences of the depassivated silicon atoms at certain points in the
sequence to prevent unwanted rearrangements (a useful technique in
many build sequences). The question might also be raised as to why
the sequence does not just start from a hydrogenated silicon
surface, since on that surface there are no unused valences to lead
to potential reactivity problems. The issue is one of chemical
convenience. Hydrogen, and in general, passivating atoms other than
bromine, could be made to work. However, using the particular tips
we have chosen for this sequence, bromine is found to more reliably
facilitate the desired reactions than other atoms that were
investigated.
[0238] In FIG. 52e, the structure shows that one of the bromine
atoms has been removed. This is accomplished using a GeRad tip.
[0239] In FIG. 52f, a CH2 group has been added to the radical
silicon that was created by the bromine removal in the previous
step. This CH2 donation reaction is accomplished using a tip like
MeDonationO or its variants, described herein.
[0240] In FIG. 52g, a hydrogen atom is added to the CH2 radical
that was added in the previous step. This is accomplished using
HDonation (whether it is HDonationNH, HDonationO, or HDonationS not
being relevant to the reaction).
[0241] In FIG. 52h, one of the remaining bromine atoms is removed,
using GeRad.
[0242] In FIG. 52i, a methyl group is donated to the silicon
radical that was created by the bromine abstraction in the previous
step. The methyl donation reaction is accomplished using MeDonation
(again, the specific variant not being relevant).
[0243] In FIG. 52j, the methyl group donated by the MeDonation tool
in the previous step is given a hydrogen atom, using an HDonation
tip.
[0244] In FIG. 52k, the sole remaining bromine is removed from the
structure, using GeRad.
[0245] In FIG. 52l, a methyl group is donated to the silicon
radical that was created by the bromine abstraction in the previous
step. The methyl donation reaction is accomplished using an
MeDonation tip. Note that unlike the previous methyl groups, this
methyl group does not have its open valence satisfied via a
hydrogen donation reaction.
[0246] In FIG. 52m, one of the previously-created CH3 groups has a
hydrogen abstracted from it, via an Abstraction tip, resulting in a
surface that has two CH2 groups and one CH3 group.
[0247] In FIG. 52n, the remaining previously-created CH3 group has
a hydrogen abstracted from it, via an Abstraction tip, resulting in
three CH2 groups on the surface of the structure.
[0248] In FIG. 52o, a silicon atom is bound to all three CH2
groups. The silicon atom is donated from an already-described tip
loaded with a different payload. Specifically, the Abstraction tip
can have a silicon atom bound to its radical active site, and will
then donate that silicon atom to the structure. The Abstraction tip
can be charged with a silicon feedstock atom by abstracting a Si
atom from anywhere else on the conventional tip which is not
crucial to the build sequence. The resulting structure is the
half-Si-rad tip, which will be obvious when realizing that the
structure shown in FIG. 52o, although differing in how termination
is depicted at the top of the diagram, is essentially the structure
from FIG. 51.
[0249] The build sequence for the half-Si-Rad as described requires
the AdamRad-Br tip. This is an adamantane radical with a bromine
feedstock. The synthesis for this tip is depicted in FIG. 53. The
synthesis starts with chemical SHA-2, previously described in FIG.
34 and the respective synthesis. SHA-2 can be iodinated at the
4-position of the aromatic rings using 12 and
[bis(trifluoroacetoxy)iodo]benzene in CHCl3 to yield AdBr-1.
Sonogashira coupling conditions of AdBr-1 with
triisopropylsilylacetylene (TIPS acetylene) produces the protected
alkyne AdBr-2. Deprotection of the TIPS group proceeds with TBAF in
THF to make the terminal acetylene AdBr-3. The terminal acetylene
is deprotonated with n-butyllithium at low temperature and
paraformaldehyde is added to produce the tripropargylic alcohol
AdBr-4, also called AdamRad-Br. Note that this version of
AdamRad-Br depicts a new leg structure, phenylpropargyl alcohol,
which has been found to be useful in conjunction with
adamantane-based bodies and silicon surfaces and could be coupled
with any of the other tips described herein.
[0250] Note that it is possible to perform a modified version of
the half-Si-Rad build sequence without using AdamRad-Br at all. The
only purpose AdamRad-Br serves in the build sequence is to
brominate a depassivated silicon surface. If the silicon surface is
bulk passivated with bromine, rather than the more common hydrogen,
the build sequence can start from a structure equivalent to that of
FIG. 52d, eliminating all the bromine donation reactions.
Techniques for bulk bromination (and halogenation in general) of
silicon are known in the literature, e.g., see (He, Patitsas et
al., "Covalent bonding of thiophenes to Si(111) by a
halogenation/thienylation route," Chemical Physics Letters. 1998.
286:508-514; Eves and Lopinski, "Formation and reactivity of high
quality halogen terminated Si (111) surfaces," Surface Science.
2005. 579:89-96).
[0251] While the example given describes building a conventional
tip using surface-mounted tips, this need not be the only such
process. For example, conventional tips could build surface-mounted
tips, using either feedstock from other surface-mounted tips, or
feedstock provisioned directly off presentation surfaces. This
could be useful if, for example, there were one or more
surface-mounted tips that were only needed in small quantity and so
it is more efficient to build them mechanosynthetically rather than
via bulk chemistry.
Additional Tip Design Guidelines and Examples
[0252] Herein we have described many different tips, and how a
modular tip design can facilitate the creation of new tips. Some
other comments on tip structure and design criteria may further
facilitate new tip and reaction design.
[0253] First, the use of a rigid tip geometry can be helpful so
that the bonds between the apical atom and the other tip atoms do
not deform excessively or break as a feedstock atom is transferred.
However, where there is a small or non-existent reaction barrier,
this requirement may be relaxed. For various reasons (e.g., ease of
synthesis, tip size, tip aspect ratio) a rigid tip may not be
desired, and relaxing the requirement expands the possible design
space. For example, if a given feedstock-workpiece reaction
requires no physical force (meaning, the reaction will simply occur
if the feedstock is brought into proximity of the desired site on
the workpiece) to surmount a reaction barrier, there may be no need
for a design with three or more legs. One or two legs may work
fine.
[0254] The tip shape preferably allows the tip to approach a
workpiece and perform the desired reaction without steric
hindrance, leading to the observation that higher aspect ratios can
be advantageous (although steric considerations can also be
addressed through reaction order, for example, by avoiding the need
for high aspect ratios). Further, tip geometry could also be
exploited to hold feedstock at a particular angle. For example,
equipment limitations may dictate that, e.g., an SPM probe, must be
kept perpendicular to the work surface. But, there may be reactions
where a perpendicular alignment of the feedstock with the workpiece
is not a desirable trajectory. In that case, it is possible to
design a tip that holds the feedstock at e.g., 45 degrees (or any
other angle desired) to the rest of the tip or handle.
[0255] With regards to a rigid tip geometry, a tetrahedral
structure with respect to the apical atom can be useful as, with a
feedstock atom bound to one leg of the tetrahedron, the other three
bonds serve to stabilize the apical atom when force is applied
during a reaction. However, other geometries are possible. For
example, in addition to VSEPR AX4 (tetrahedral, or other variations
of AX4), AX5 and higher hybridizations can also provide the
necessary free electrons to bond a feedstock atom while having the
ability to form at least three other bonds to create a rigid tip
structure. However, the primary concern is simply whether or not a
given tip will reliably perform the intended reaction, and
certainly working tips can deviate from these suggestions.
[0256] To facilitate the design of new tips and reactions by
example, and to provide a library of known tips and reactions
(which may in themselves constitute a set of tips and reactions
sufficient for some build sequences, and the availability of a set
of pre-vetted reactions certainly speeds up the build sequence
design process), below we provide a table of various donating
structures (e.g., tips), receiving structures (e.g., workpieces,
although in the examples the receiving structures are also
tip-sized to facilitate computational analysis) and reactions which
can be facilitated between the two. These structures and reactions
have been vetted using multiple computational chemistry algorithms
and approaches, including B3LYP/6-311G(d,p) using the Gaussian09
software package with default DFT grid size and convergence
criteria. Other computational chemistry algorithms and basis sets
can also be employed, as can multi-scale methods such as ONIOM.
[0257] Computational means (hardware) were generally individual or
clustered Intel CPU-based, multi-core servers, having between 8 and
256 GB RAM, solid state hard drives, and network interconnects as
appropriate. This hardware is only exemplary. Other computational
means can be used (as could other software). For example,
processors may be CPU-, GPU-, ASIC-based, or other. Memory means or
data storage means, both volatile and non-volatile, could also take
many forms including various flavors of RAM, hard drives, or Flash,
among others.
[0258] The data provided include net energy changes and reaction
barriers, and the feedstock transferred includes Al, B, Be, Br, C,
Cl, F, Ge, H, Ir, Li, Mg, N, Na, O, P, S, and Si. While many
examples are provided, they are indeed only examples. These are
certainly not the only structures and reactions that would be
possible given the teachings presented herein.
[0259] With respect to the reactions in Table 1, the tip always
approached the workpiece coaxially. "Coaxial" means that the bond
that is being broken (e.g., the tip-feedstock bond) and the bond
being formed (e.g., the feedstock-workpiece bond) lie on the same
line. The coaxial trajectory has been found to be reliably
facilitate most reactions we have studied. This fact, along with
the extensive data provided, should enable the facile design of a
vast number of related reactions. Also, (Tarasov, Akberova et al.,
"Optimal Tooltip Trajectories in a Hydrogen Abstraction Tool
Recharge Reaction Sequence for Positionally Controlled Diamond
Mechanosynthesis," J. Comput. Theor. Nanosci., 2, 2010) teaches a
process that may be used to determine other trajectories, and we
incorporate by reference this material.
[0260] In the table below, "Tip" is the donating structure, "FS"
(feedstock) is the atom being transferred, "Workpiece" is the
structure to which the feedstock is transferred, "Delta (eV)"
indicates the change in energy for the reaction, and "Barrier (eV)"
indicates the reaction barrier.
[0261] "300K" is the probability of reaction failure at 300 Kelvin
(room temperature), while "77K" is the probability at 77 Kelvin
(liquid nitrogen temperature). Scientific notation is used due to
the very small numbers. These calculations were performed using the
formulas disclosed in Code Listing 1. 300K and 77K are
representative temperatures only. Any temperature at which the
reactions are reliable enough for a given purpose could be used,
and another common temperature, 4K, is easily-accessible with
liquid helium and would show much higher reliability figures. Also,
it is noteworthy that most of the reactions listed would have over
99.99% reliability even at room temperature.
[0262] With respect to the structures, C9H14[Al,B,N,P] have the
apical atom, to which the feedstock atom is attached, at the
sidewall position of an adamantane frame. C9H15[C,Si,Ge] have the
apical atom, to which the feedstock atom is attached, at the
bridgehead position of an adamantane frame. The notation for the
workpieces are the same, except that the apical atoms are listed
first. For example, the reaction where a C914Al tip using a Be
feedstock atom donates the feedstock atom to CC9H15 could be
expressed as:
AdamantaneSidewall-Al--Be.+.C-AdamantaneBridgeHead->AdamantaneSidewal-
l-Al.+.Be--C-AdamantaneBridgeHead
TABLE-US-00001 TABLE 1 Element Transfers with Energy Calculations
and Reliabilities at Various Temperatures Delta Barrier Tip FS
Workpiece (eV) (eV) 77K 300K C9H14Al Al CC9H15 -0.64 0.02 1.15E-42
1.72E-11 C9H14Al B NC9H14 -3.40 0.00 1.18E-222 1.09E-57 C9H14Al Be
CC9H15 -1.46 0.00 2.39E-96 2.87E-25 C9H14Al Be NC9H14 -2.71 0.00
1.14E-177 3.84E-46 C9H14Al H BC9H14 -1.05 0.15 4.94E-69 2.94E-18
C9H14Al H CC9H15 -0.90 0.22 1.77E-59 8.32E-16 C9H14Al H SiC9H15
-0.49 0.23 1.06E-32 6.21E-09 C9H14Al Li NC9H14 -0.76 0.00 1.30E-50
1.57E-13 C9H14Al Mg BC9H14 -0.22 0.00 2.48E-15 1.78E-04 C9H14Al Mg
NC9H14 -0.61 0.00 1.53E-40 6.04E-11 C9H14Al N BC9H14 -1.73 0.04
6.14E-114 8.75E-30 C9H14Al P BC9H14 -0.75 0.14 1.47E-49 2.93E-13
C9H14Al P NC9H14 -0.42 0.00 4.85E-28 9.76E-08 C9H14Al P SiC9H15
-0.21 0.00 3.30E-14 3.47E-04 C9H14Al S BC9H14 -0.90 0.00 2.69E-59
9.27E-16 C9H14B Al CC9H15 -0.13 0.00 3.72E-09 6.86E-03 C9H14B Be
NC9H14 -1.26 0.00 4.21E-83 7.19E-22 C9H14B Li NC9H14 -0.78 0.00
5.61E-52 7.01E-14 C9H14B Na NC9H14 -0.13 0.00 3.15E-09 6.58E-03
C9H14N Br AlC9H14 -2.48 0.00 7.75E-163 2.46E-42 C9H14N S AlC9H14
-0.65 0.02 1.95E-43 1.09E-11 C9H14N S BC9H14 -1.55 0.00 5.25E-102
1.01E-26 C9H14N S SiC9H15 -0.41 0.11 2.18E-27 1.44E-07 C9H14P Al
NC9H14 -1.67 0.07 6.91E-110 9.60E-29 C9H14P Mg AlC9H14 -0.05 0.00
6.87E-04 1.54E-01 C9H14P Mg BC9H14 -0.27 0.02 1.71E-18 2.75E-05
C9H14P P BC9H14 -0.87 0.07 1.31E-57 2.51E-15 C9H15C Br AlC9H14
-1.23 0.01 3.73E-81 2.27E-21 C9H15C Br BC9H14 -1.50 0.00 1.44E-98
7.71E-26 C9H15C Br GeC9H15 -0.60 0.06 5.25E-40 8.28E-11 C9H15C Br
SiC9H15 -1.01 0.04 1.27E-66 1.22E-17 C9H15C Cl AlC9H14 -1.22 0.17
9.07E-81 2.86E-21 C9H15C Cl BC9H14 -1.62 0.18 8.02E-107 5.87E-28
C9H15C Cl GeC9H15 -0.52 0.32 1.27E-34 2.00E-09 C9H15C Cl SiC9H15
-1.02 0.21 1.29E-67 6.79E-18 C9H15C Li NC9H14 -1.06 0.00 6.19E-70
1.72E-18 C9H15C Mg NC9H14 -0.61 0.00 8.90E-41 5.25E-11 C9H15C O
BC9H14 -2.68 0.00 1.58E-175 1.36E-45 C9H15C S AlC9H14 -0.88 0.00
2.90E-58 1.71E-15 C9H15C S BC9H14 -1.78 0.00 7.93E-117 1.59E-30
C9H15C S GeC9H15 -0.24 0.00 2.11E-16 9.47E-05 C9H15C S NC9H14 -0.23
0.00 1.49E-15 1.56E-04 C9H15C S SiC9H15 -0.63 0.00 3.25E-42
2.25E-11 C9H15Ge Br AlC9H14 -0.63 0.11 7.10E-42 2.75E-11 C9H15Ge Br
BC9H14 -0.90 0.14 2.73E-59 9.31E-16 C9H15Ge Br SiC9H15 -0.41 0.21
2.39E-27 1.47E-07 C9H15Ge C CC9H15 -1.15 0.00 9.46E-76 5.54E-20
C9H15Ge C SiC9H15 -0.46 0.00 7.39E-31 1.85E-08 C9H15Ge Cl AlC9H14
-0.71 0.31 7.12E-47 1.43E-12 C9H15Ge Cl SiC9H15 -0.51 0.47 1.00E-33
3.39E-09 C9H15Ge F AlC9H14 -1.08 0.01 2.00E-71 7.15E-19 C9H15Ge F
BC9H14 -1.79 0.18 1.19E-117 9.76E-31 C9H15Ge Ge CC9H15 0.02 0.00
6.18E-02 4.89E-01 C9H15Ge H SiC9H15 -0.35 0.23 1.12E-23 1.29E-06
C9H15Ge Li NC9H14 -0.46 0.00 1.62E-30 2.26E-08 C9H15Ge O BC9H14
-2.96 0.00 3.94E-194 2.29E-50 C9H15Ge O SiC9H15 -0.96 0.00 9.41E-64
6.66E-17 C9H15Ge P BC9H14 -0.79 0.03 5.05E-52 6.82E-14 C9H15Ge S
BC9H14 -1.54 0.15 3.71E-101 1.67E-26 C9H15Ge Si CC9H15 -0.21 0.00
3.21E-14 3.44E-04 C9H15Si Al CC9H15 -0.25 0.02 4.97E-17 6.54E-05
C9H15Si B CC9H15 -1.12 0.14 4.39E-74 1.48E-19 C9H15Si Br BC9H14
-0.49 0.43 1.13E-32 6.31E-09 C9H15Si H BC9H14 -0.56 0.27 4.65E-37
4.73E-10 C9H15Si Li NC9H14 -0.57 0.00 5.33E-38 2.71E-10 C9H15Si P
BC9H14 -0.54 0.16 4.44E-36 8.44E-10 C9H15Si S BC9H14 -1.14 0.00
2.44E-75 7.07E-20 C9H15Si Si CC9H15 -0.11 0.00 6.11E-08 1.41E-02
C9H15Si Ge CC9H15 -0.08 0.00 5.83E-06 4.53E-02 C9H15Ge Ir CC9H15
-0.04 0.00 1.97E-03 2.02E-01 C9H15Ge Ir SiC9H15 -0.33 0.00 1.82E-22
2.63E-06 C9H15C Ir SiC9H15 -0.29 0.00 9.36E-20 1.31E-05 C9H15C Ir
BC9H14 -1.07 0.00 6.78E-71 9.77E-19
[0263] Note that it is possible for the change in energy (eV) to be
positive. This is due to the fact that energy and force are not
equivalent. A mechanosynthetic tip may exert force over a distance
that results in a net change in energy which is positive, even if
the reaction product resides in a local energy minima. This is
discussed in more detail herein with respect to bond stiffness and
affinity.
[0264] As the Table 1 data indicates, high reliability transfers of
atoms including Al, B, Be, Br, C, Cl, F, Ge, H, Ir, Li, Mg, N, Na,
O, P, S, and Si have been shown to be possible, using tips which
employ active atoms Al, B, C, Ge, N, P, and Si. Obviously, these
are examples only, and an even wider range of tips and reactions
can be designed given the teachings herein.
Bond Strain in Tip, Reaction and Workpiece Design
[0265] A number of strain types exist, such as Van der Waals,
stretch, torsion, and angle (or "bend," including ring) strain. In
aggregate the various types of strain are often referred to as
"steric energies," and these steric energies, or strain, are known
to influence molecular stability and chemical reaction
energetics.
[0266] For example, cyclobutane, with 7.5% kcal/mol/bond strain, is
more reactive than the larger cycloalkanes in which the ring strain
is relaxed. Fullerenes are similarly affected by bond strain. Since
the lowest energy configuration for individual fullerene units is
planar, higher curvatures generally lead to more reactive molecules
due at least in part, to angle strain. In terms of individual bond
energy, less than about 2% strain tends to have little effect on
reactivity. 3-5% strain tends to cause at least some increase in
reactivity, while at 5-10% strain, major increases in reactivity
are generally apparent. Of course, this trend cannot continue
indefinitely; if strain is too high, a bond can spontaneously
rupture, leading to rearrangement of the molecule.
[0267] Note that overall, a molecule could have very little strain,
but one or more strained bonds can still cause it to be highly
reactive, so the distribution of strain is also important.
Conversely, a molecule could have many bonds which are only
slightly strained (perhaps less than the 5% figure), yet when
accumulated across multiple bonds, the overall strain energy is
substantial. In such cases, modest amounts of strain on per-bond
basis can lead to substantial effects on molecule conformation and
various other properties. These observations lead to the conclusion
that using strain to alter bond strength, and therefore reactivity,
can be a useful technique in the design of tips and workpieces.
[0268] One scenario is that of feedstock held to a tip by a single
bond. Strain within the tip may be used to change the bond angles,
and thereby energies, of the apical tip atom to the feedstock. For
example, consider an adamantane structure where a bridgehead carbon
is bonded to the feedstock. This bridgehead carbon would normally
be bonded to three other carbons, and the uniform length of the
carbon-carbon bonds throughout the adamantane structure allows the
bridgehead carbon to achieve a perfect tetrahedral configuration
where each bond to the bridgehead carbon is about 109.5 degrees.
However, if a Ge atom is substituted for each of the three carbons
to which the bridgehead carbon is attached, the Ge--C-feedstock
angle becomes about 112.9 degrees, causing angle strain.
[0269] In addition to angle strain, other type of strain can also
be employed. For example, Van der Waals strain can be created by
replacing, e.g., H atoms with larger diameter atoms of the same
valence, adjacent to the feedstock. In this case, the larger
diameter atom need not be bonded to the feedstock or to the apical
tip atom. It need only impinge upon the feedstock's Van der Waals
radius to cause steric strain.
[0270] While a tip designed in this manner can cause Van der Waals
strain by having two or more parts of the same tip interfere (where
one part is the feedstock site and the other part is a portion of
the tip designed to at least partially impinge upon the feedstock
location), a second tip could also be used to apply mechanical
force to feedstock. For example, consider a first tip with
feedstock bound to it. Using a second tip to apply force to the
feedstock perpendicularly (or at any useful angle) to its point of
attachment could weaken the bond between the first tip and the
feedstock. This is conceptually similar to building such strain
into a single tip, but more versatile as the timing, amount of
force, and angle of force application can all be varied.
[0271] Another scenario where strain could be employed is when
feedstock is held by more than one bond to a tip. To reduce tip
bond strength to the feedstock, the bonding points can be pulled
apart until the bonds are strained by the desired amount. This is
more easily illustrated in a slightly larger structure than a
single adamantane, so that rigidity of the tip backbone can be used
to create strain without excessive deformation. For example, the
native distance between two methyl groups connected by an oxygen
(3HC--O--CH3) is about 2.36 .ANG., and the angle is about 110.7
degrees. However, due to the lattice spacing, this configuration
cannot be obtained on (111) diamond. If two adjacent carbons on the
(111) face of diamond each have a hydrogen removed, and an oxygen
atom is then bound to those carbons, with a very small structure
composed of 3 interlocked adamantanes (larger structures would
likely allow less deformation of the tip backbone), the oxygen
becomes bound to the two carbons at an angle of about 87.8 degrees
with the carbons being spaced about 2.02 A apart. Clearly, this is
a substantial distortion of the minimal energy configuration and so
if the oxygen is the feedstock, it will require less energy to
remove from the tip structure than if it were bound in a
configuration closer to its energy minima. Substitutions could be
used to alter the diamond lattice spacing to increase or decrease
the amount of strain created. An analogous technique could be used
by a single feedstock moiety held by more than one tip. The tip
spacing could be used to adjust tip-feedstock bond strength, and
this could be changed on-the-fly if desired.
[0272] Note that with one single bond, as they are free to rotate,
torsion is generally irrelevant. But, if a feedstock moiety was
multiply-bonded, or one or more, e.g., double bonds (or any bond
type not free to rotate), were used to bind the feedstock to one or
more tips, or one or more points on a single tip, torsion could
also be used to create strain, as could any other well-known
strain-inducing modifications.
[0273] Many of the same techniques could be employed on the
workpiece. In some cases, modulating bond strength on the workpiece
instead of, or in addition to, the tip may be convenient. And,
build sequence order can be chosen to create intermediate
structures with strain if this alters the reactivity favorably.
[0274] It should be noted that creating strain and releasing strain
are two sides of the same effect. If one considers a strained
structure the default structure, releasing strain could be used to,
for example, strengthen, instead of weaken, bonds. Further, strain
levels need not be static. Levels of strain could be changed curing
the course of a reaction. For example, to increase tip affinity
when picking up feedstock, and then decreasing tip affinity when
releasing feedstock.
[0275] FIG. 54 depicts various one way of creating adjustable
strain, and hence affinity, for feedstock. In FIG. 54a, a first tip
(5401) is connected to feedstock (5405) via bond (5403). A second
tip (5402) is also connected to feedstock (5405) via bond (5404).
We assume this to be the minimum energy configuration. Various
movements of the two tips would change the bond angles and lengths,
causing strain and thereby reducing the affinity of the feedstock
for the tips. For example, in FIG. 54b, the two tips have been
moved part, stretching and changing the angle of the bonds to the
feedstock. In FIG. 54c, the two tips have been move closer
together, potentially compressing and changing the angle of the
bonds to the feedstock. And, in FIG. 54d, one tip has been moved
vertically with respect to the other, potentially resulting in
stretching of bond (5403) and compression of bond (5404), plus
angle changes. In a complete system, the tips would be attached to
positional means (not shown). It is possible that each tip has its
own position means. It is also possible that both tips reside on a
single positional means (and actually may be considered two halves
of the same tip) in which case relative movement can still be
caused in various ways. For example, the surface onto which the
tips are affixed could be a piezo element which can expand and
contract. Or, changing temperature, charge, or other parameters
could result in a conformation change in either the tips, or the
surface to which they are affixed.
Workpiece Specification and Build Sequences
[0276] Many structures and reactions have been discussed herein,
along with teachings which enable the creation of additional
structures and reactions. However, to apply this information to the
building of a workpiece, it helps to define the workpiece in an
atomically-precise manner, and to define a build sequence which
will create the workpiece.
[0277] A workpiece for mechanosynthesis can be defined by
specifying each atom in the workpiece and its atomic coordinates,
directly or indirectly (for example, via an algorithm which
generates the desired structure). Many computational chemistry
programs allow the creation of models based on atomic coordinates,
or algorithms to generate such coordinates.
[0278] Once the atomic coordinates have been specified, a build
sequence can be created that specifies the order in which each atom
is to be added to, or removed from, the workpiece. Reactions that
do not add or remove atoms are also possible, such as those that
change the bonding structure of the workpiece, or if necessary,
charge or otherwise alter tips. The reactions must be ordered so
that they result in the desired workpiece, while avoiding, for
example, intermediate states prone to pathological reactions, or
unstable structures that undesirably rearrange. These topics are
addressed in more detail below.
Process Flowcharts and Descriptions
[0279] To aid in the understanding of the general process of
creating a workpiece, FIGS. 55 through 58 illustrate embodiments of
the invention using exemplary flowcharts. Note that many variations
on these processes are possible, and even without changing the
steps involved, one might change the decision logic or loop through
some processes more than once. For example, to optimally design a
workpiece for manufacturability (55-2) may require an iterative
process where the workpiece design is revised based on the outcome
of subsequent steps or processes, such as the reaction design
process described in FIG. 56.
[0280] The process can be started in FIG. 55, which provides an
overview of how a workpiece definition can be created, at step
(55-1), "Create Workpiece Functional Specifications." This step is
similar to that for any traditionally-manufactured product in that
product requirements must be defined before the product can be
designed from an engineering perspective.
[0281] Step (55-2), "Design Workpiece for Manufacturability" also
has an analog in traditional manufacturing. The product must be
designed with the limitations of the manufacturing process in mind.
In the case of mechanosynthesis, this means that a device is
preferably designed with elements and bonding patterns whose
properties are understood, for which tips and build sequences have
been, or can be, designed and are compatible with equipment
capabilities, using geometries accessible to the relevant tips,
among other limitations which will be obvious to those skilled in
the art given the teachings herein.
[0282] Once the device has been designed, step (55-3) is to
"Specify Atomic Coordinates of Workpiece." That is, define each
atom type and its position within the structure. This step may also
include determining bonding structure, as this step can be
informative although technically redundant since the bonding
structure may be fully specified via the atomic coordinates. This
may be done in any molecular modeling or computational chemistry
software with the appropriate capabilities, such as HyperChem,
Gaussian, GROMACS or NAMD.
[0283] Step (55-4) "Determine Reaction Reliability Requirements"
involves performing an impact analysis of potential defects and the
resultant establishment of reaction reliability requirements.
Although the goal of mechanosynthesis is the production of
atomically-precise products, unintended reactions can occur at
frequencies which depend on factors including the chemical
reactions being used, the tip design, the reaction trajectory,
equipment capabilities and temperature. For each reaction one could
analyze the most likely pathological side reactions that might
occur and their impact upon the finished workpiece. For example,
one could determine the impact of a feedstock atom failing to
transfer, a feedstock atom bonding to a workpiece atom adjacent to
the intended position, or the workpiece undergoing an unintended
rearrangement. The workpiece could be simulated with each potential
defect, or more general heuristics or functional testing could be
used to determine the likely impact of possible errors in the
workpiece.
[0284] As an example of how a defect could be insignificant in one
context but not in another, consider a simple part such as a
structural beam: A small number of mistakes may not substantially
affect the properties of the finished part, and may not affect the
overall product, particularly if the part has been over-engineered
to allow for some defects. In such a scenario, one might decide
that some number of defects were tolerable and therefore require
relatively low reaction reliability. On the other hand, if the
workpiece being constructed were, for example, a single-molecule
transistor that would not function correctly, or at all, if crucial
atoms were misplaced, one might require a very low number
(including 0) of defects.
[0285] One alternative to defect impact analysis is to require that
each reaction be reliable enough that it is statistically unlikely
that the final workpiece contains any errors. This is quite
feasible, as seen from the reaction reliability calculations
presented herein. Also, the ability to correct errors may have an
impact on reaction reliability requirements. If errors can be
fixed, one might decide to reduce reliability requirements and
simply fix errors as they occur.
[0286] FIG. 56, which describes how a build sequence can be
designed, begins with step (56-1) "Determine Order of Reactions,
Reaction Conditions and Trajectories." Each atom, as specified in
the atomic coordinates of the workpiece, generally (but not
necessarily since, for example, one could use dimers or larger
molecules as feedstock) requires that a particular reaction be
performed on the workpiece to deposit that atom. Abstraction
reactions may also be required, as may be reactions which alter the
bonding structure of the workpiece without adding or subtracting
any atoms.
[0287] There may be many different build sequences that would
permit the construction of a particular workpiece. Steric
constraints will be one determinant of the order in which atoms are
added, as a three-dimensional workpiece requires adding atoms in an
order which permits access by the necessary tools for later
reactions. The stability of the intermediate structures should also
be considered. For example, certain atoms, when left as radicals,
might rearrange, forming undesired bonds with adjacent atoms. In
addition to a logical order to the addition of atoms, other
techniques can be employed to prevent undesired rearrangement. For
example, terminating atoms can be added to radical sites to
temporarily satisfy empty valances, or temperature can be
reduced.
[0288] When a presumptive build order has been established, the
build sequence may be simulated to determine if it works correctly
(56-2). The same simulations can test reaction parameters including
which tip to use, what temperature is required, and what trajectory
a tip will follow. As has been previously noted, lower temperatures
will favor accuracy, and frequently the coaxial trajectory will
enable successful reactions.
[0289] Note that, given that rearrangement and abstraction
reactions may be required in a build sequence, workpieces may
require more reactions than the number of atoms in the finished
workpiece. And, even if this were not the case, workpieces with
many atoms will generally require many reactions. If the reactions
are being implemented manually, this leads to a substantial
requirement for labor. Automating the reaction steps may therefore
be desirable. CAD programs can be used to specify AFM trajectories
(Chen, "CAD-guided automated nanoassembly using atomic force
microscopy-based nonrobotics," IEEE Transactions on Automation
Science and Engineering, 3, 2006; Johannes, "Automated
CAD/CAM-based nanolithography using a custom atomic force
microscope," IEEE Transactions on Automation Science and
Engineering, 3, 2006), atomic force microscopes that are
programmable are commercially available, and programming languages
or environments (e.g., LabVIEW) to control scientific equipment are
well known (Berger et al., "A versatile LabVIEW and
field-programmable gate array-based scanning probe microscope for
in operando electronic device characterization," Review of
Scientific Instruments 85, 123702 (2014)).
[0290] Based on the outcome of the simulations, a decision is
reached as to whether the reactions as specified are correct
(56-3). If not, the sequence is revised. If so, the process
proceeds to (56-4) where a decision is made as to whether any of
the calculated reactions may pose reliability concerns, for
example, based on rearrangements or incorrect reactions that were
seen during simulation in (56-2).
[0291] In (56-5) the reaction reliabilities can be calculated (for
example, by energy barrier calculations or Monte Carlo
simulations). (56-6) is a determination as to whether the proposed
reaction reliabilities meet production quality needs, and, if the
answer to (56-6) is no, the process proceeds to (56-7) where
requirements are reviewed to see if the build sequence restrictions
can be relaxed since they were not met. From (56-7) if the answer
is yes, a new iteration is started at (55-4) to determine revised
reaction reliability requirements. If the answer to (56-7) is no,
alternate reactions, reaction order, reaction trajectories, or
reaction conditions can be simulated (56-1) to find a revised build
sequence that meets the reaction reliability requirements. If the
answer to (56-6) is yes, the process continues in FIG. 57, step
(57-1).
[0292] FIG. 57 describes a process for carrying out
mechanosynthetic reactions per a build sequence. Starting at (57-1)
"Perform Mechanosynthetic Reactions," the reactions determined in
the build sequence are carried out using SPM/AFM-like equipment, or
other suitable equipment. This step involves, whether manually or
in a computer-controlled manner, using a positionally-controlled
tip to perform each mechanosynthetic reaction in the build
sequence. This means picking up a feedstock atom from a
presentation surface (or potentially a gaseous or liquid source of
feedstock) and bonding it to the workpiece, or removing an atom
from the workpiece, or changing the bonding structure of the
workpiece without adding or removing an atom. This step would also
encompass other reactions, including reactions not involving the
workpiece, such as tip refresh or pre-reaction feedstock
manipulation as may be necessary.
[0293] Step (57-2) is a decision point. If the answer is "no,"
testing is not required (for example, such as when the reactions
being used are reliable enough that testing is superfluous), the
process proceeds to (57-3). The action taken from (57-3) depends on
whether all reactions in the build sequence have been completed. If
no, reactions are repeated until the answer is yes, at which point
the workpiece is complete. Back at (57-2), if the answer were
"yes," testing is required, the process continues in FIG. 58,
starting with step (58-1).
[0294] In FIG. 58, testing may done by, for example, scanning the
surface of a workpiece using AFM or SPM-like techniques and
checking to see that the expected structure is present. If no
errors are found in (58-2), the process continues at (57-3). If an
error is present at (58-2), a decision must be made in (58-3) as to
whether the error is ignorable (e.g., not an error that would
prevent the workpiece from functioning). If it is ignorable, the
process again continues with (57-3), although the build sequence
may require adjustment if key atoms were moved as a result of the
error (not depicted). If the error is not ignorable, it must be
determined if the error can be fixed (58-4). This is largely a
question of whether the tools and processes exist to fix the
error.
[0295] Note that errors could be fixed in various ways, such as
directly reversing the last reaction if possible, or using
abstraction tips to remove the local area of the workpiece
entirely, paring the workpiece back to a point where the edges can
be left in a correct and stable configuration. The build sequence
would then be altered to fill back in the removed area, before
proceeding on with the rest of the sequence.
[0296] If the error can be fixed, this is done in (58-6) and the
process continues with (57-3). If the error cannot be fixed, given
that it was previously determined to be a crucial error, the build
sequence must be started over (58-5).
[0297] The embodiment of the process shown in FIG. 58 assumes the
ability to detect and fix errors (58-6). This is not necessarily
the case, and this flow chart represents only one possible process
of implementing mechanosynthesis. For example, it is possible to
desire testing without the ability to fix errors, or at least not
all errors, if only to know that the workpiece must be discarded
and the process started anew, as in (58-5). It is also possible to
forgo error checking completely, and this is a reasonable solution
especially for high-reliability reactions. Product requirements and
process capabilities, among other considerations, will determine
which steps are actually used, and in what order.
[0298] A variation on the error correction described above is the
use of "conditional" build sequences. Some reactions might be known
to be error-prone, having one or more pathological outcomes which
cannot be prevented reliably. In this case, the creation of a
branching, or conditional, build sequence can be useful. A
conditional build sequence has multiple paths, or sub-sequences,
within it. The exact path chosen will be determined by which
structures result from the previous reactions. For example, assume
that a build sequence reaches a reaction which is likely to have
one of three outcomes. The exact outcome cannot be controlled, but
it can be accommodated by a build sequence which contains logic
such as "If the product of reaction X is A, then do this. If the
product of reaction X is B, then do something else. If the product
of reaction X is C, then follow yet another path."
Exemplary Build Sequences
[0299] Now that the process of designing a build sequence has been
described, several exemplary build sequences, in addition to the
half-Si-Rad build sequence already described, are presented. The
following sequences can be used to create diamond (or with
modification, diamondoid) structures. Reactions are logically
grouped into sets of sequences which provide the ability to
initiate, extend, and terminate, rows in a diamond structure. In
these particular sequences, the assumed starting surface is the 110
face of diamond, although this is exemplary only, and other faces
can be built upon, and other surfaces used (e.g., diamond can also
be built on Si, given the minimal lattice spacing mismatch).
[0300] These build sequences were computed using the representative
density functional method with the B3LYP/6-311G** basis set, which
typically provides a good tradeoff between accuracy and
computational expense. Higher reaction accuracies could be obtained
using more computationally-demanding techniques such as coupled
clusters. (Lee, Scuseria et al., "Achieving Chemical Accuracy with
Coupled-Cluster Theory," Quantum Mechanical Electronic Structure
Calculations with Chemical Accuracy, Kluwer Academic Publisher,
1995) 4 degrees Kelvin was assumed for this sequence (readily
accessible with liquid helium) although the reactions would likely
prove reliable at higher temperatures.
Reactions
[0301] The reactions in Table 2 are grouped into one of three
functions: Row Initiation, Row Extension, or Row Termination. For
example, to start a new row of diamond on a build surface, one
would use the Row Initiation reactions, #1 to #11. To then extend
that row, Row Extension reactions #12 to #17 would be used (as many
times as necessary to achieve the desired length). To terminate the
row, Row Termination reactions #18 to #22 would be used.
[0302] Each set of reactions can be repeated as many times as
necessary, at different locations as appropriate, to build
workpieces with varied geometry. This is conceptually similar to
how a 3D printer lays down lines or blobs of material which, in
aggregate, build a desired shape. This analogy only goes so far
however, because "3D printing" using mechanosynthesis must take
into account the varying chemical nature of different sites on a
workpiece. For example, as the different sub-sequences for building
diamond show, placing the first carbon in a row is not the same as
placing a middle carbon, or the carbon at the far end.
[0303] The tips used in these build sequences are described in
detail elsewhere herein. They are: the Abstraction tip, the
HDonation tip, the Germanium Radical tip (GeRad), and the
MeDonation tip. Additionally, while the descriptions should make
obvious the sequence of events, molecular models depicting the
products and reactants of the reactions described below can be
found in US Patent Application 20160167970. Similar reactions and
build sequences, along with a pyramidal exemplary workpiece, can be
found in PCT Patent Application WO2014/133529.
TABLE-US-00002 TABLE 2 Exemplary Build Sequence Reactions #
Description Tip Row Initiation Reaction Sequence 1 First step for
row initiation, via abstracting Abstraction the hydrogen from the
outer edge carbon. 2 Second step for row initiation, via donating
MeDonation the radical methyl group to the radical outer edge
carbon. 3 Third step for row initiation, via donating a HDonation
hydrogen to the radical outer edge methyl group. 4 Fourth step for
row initiation, via abstracting Abstraction the hydrogen from the
surface carbon adjacent to the outer edge methyl group. 5 Fifth
step for row initiation, via donating a MeDonation radical methyl
group to the radical surface carbon adjacent to the outer edge
methyl group. 6 Sixth step for row initiation, via abstracting
Abstraction a hydrogen from the outer edge methyl group, allowing
radical-radical coupling between the carbon site of the outer edge
methyl group and the adjacent radical methyl group to form a
6-member ring. 7 Seventh step for row initiation, via abstracting
Abstraction a hydrogen from a secondary carbon within a 6-member
ring. 8 Eighth step for row initiation, via abstracting Abstraction
a hydrogen from a secondary carbon adjacent to a radical carbon
both within a 6-member ring, allowing radical-radical coupling
between the two adjacent secondary radical carbons forming a
C.dbd.C double bond. 9 Ninth step for row extension, via
abstracting a Abstraction hydrogen from the surface carbon adjacent
to the 6-member ring. 10 Tenth step for row extension, via donating
a MeDonation radical methyl group. On approach of the tip to the
surface, the radical methyl group breaks into the C.dbd.C double
bond of the 6-member ring, allowing for subsequent radical-radical
coupling of the radical surface carbon with the radical methyl
carbon on retraction of the tool from the surface. 11 Final step
for row extension, via donating a HDonation hydrogen to the radical
secondary carbon. Row Extension Reaction Sequence 12 First step for
row extension, via abstracting a Abstraction hydrogen from the
surface carbon adjacent to the cage. 13 Second step for row
extension, via abstracting Abstraction a hydrogen from the
secondary carbon within the cage adjacent to the surface radical
carbon, allowing for radical-radical coupling creating a strained
tertiary carbon site. 14 Third step for the row extension, via
Abstraction abstracting a hydrogen from the strained tertiary
carbon. 15 Fourth step for row extension, via donating a MeDonation
radical methyl group to the strained radical tertiary carbon. On
retraction of the tip from the surface, the bond between the
strained tertiary carbon and the surface carbon breaks with
preference to form an unstrained C.dbd.C double bond. 16 Fifth step
for row extension, via approaching GeRad the secondary carbon of
the C.dbd.C double bond with tip, allowing the radical surface
carbon to break into the C.dbd.C double bond thereby forming a C--C
single bond between the primary carbon and the surface carbon. 17
Final step for the row extension, via HDonation saturating the
radical tertiary carbon. Row Termination Reaction Sequence 18 First
step for the row termination, via Abstraction abstracting a
hydrogen from a tertiary carbon. 19 Second step for row
termination, via Abstraction abstracting a hydrogen from the
secondary carbon adjacent to the radical tertiary carbon, allowing
radical-radical coupling to form a strained C.dbd.C double bond. 20
Third step for row termination, via donating a MeDonation radical
methyl group to the secondary carbon of the strained C.dbd.C double
bond. On retraction of the tip from the surface, the position of
the radical methyl group facilitates the migration of a hydrogen
from the outer edge carbon thereby saturating the methyl group and
leaving a radical outer edge carbon. 21 Fourth step for row
termination, via donating a HDonation hydrogen to the radical
tertiary carbon. 22 Final step for row termination, via abstracting
Abstraction a hydrogen from the methyl group, allowing
radical-radical coupling to occur between the radical methyl group
and the radical outer edge carbon, closing the row.
Differentiating Mechanosynthesis Products
[0304] It should be noted that, while a pyramidal workpiece is
mentioned here, the reaction sequences provided can make many other
shapes. In general, workpieces can be virtually any shape permitted
by the chemistry of the workpiece, though some shapes and
substitutions may require the design of additional reactions. While
shapes such as pyramids, cuboids, cylinders, spheres, ellipsoids
and other simple geometric shapes can obviously be made, they are
perhaps not the most interesting or most useful examples of what
can be built with mechanosynthesis. This is for a variety of
reasons, including the fact that their simplicity limits their
functionality (although different parts can be combined to address
this issue), and because at least some of these shapes can be
approximated, even if not in an atomically-precise manner, by other
technologies. For example, it may be possible to grow some simple,
approximate shapes using chemical vapor deposition.
[0305] What are perhaps more interesting cases are where the
workpiece is not a simple shape, or any periodic shape derived
directly from its crystal structure (which might permit its
manufacture by CVD, self-assembly, or some other known process). We
will refer to such workpieces as being "aperiodic", and aperiodic
workpieces are interesting because as far as we know,
mechanosynthesis is the only way to produce such workpieces. For
example, consider an arbitrary shape such as the outline of a car
(to use a familiar shape, if not a relevant scale). Even if CVD
could be used to grow atomically-precise crystals, there is no way
it could be used to achieve such an irregular shape. Also included
in aperiodic workpieces would be workpieces that may largely be
periodic, but which have aperiodic substitutions. For example,
consider a diamond cube, perfect and regular in all respects except
that nitrogen vacancies have been placed in specific locations.
Again, this would be impossible to create via CVD, or any other
technology of which we are aware besides mechanosynthesis, yet this
could be a very useful workpiece for realizing a quantum computer.
The vast majority of parts, whether mechanical or electronic, used
in devices today, are aperiodic. Being aperiodic is the rule rather
than the exception, and while such parts are easily manufactured at
the macro-scale using subtractive manufacturing (e.g., machining)
and other techniques, it is very difficult to manufacture such
parts with atomic precision. In most cases we would say that it is
impossible without mechanosynthesis.
[0306] Another way to view the difference between mechanosynthesis
products and other natural or synthetic products is to compare some
other aspects of their makeup aside from periodic versus a
periodic. Specifically, it is informative to consider stiffness,
bonding structure, size, and complexity (which can be related to,
but is not the same as periodicity, or lack thereof).
[0307] Large numbers of natural and synthetic chemical structures,
and synthesis pathways, are known outside of mechanosynthesis. And,
given these known structures and synthesis pathways, the
manufacture of many more structures would be possible. Some of
these structures are large (as molecules go), some are stiff and
highly-bonded, some have strained bonds, some are
atomically-precise, and some, by various measures, could be
considered complex. However, no natural or synthetic structure
prepared without the aid of mechanosynthesis, possesses all of
these characteristics.
[0308] For example, DNA of essentially arbitrary length and
sequence can be prepared using conventional techniques. And, given
that DNA need not be simply a repetition of the same monomer, by
some measures DNA sequences could have high complexity. However,
DNA is essentially a floppy, one-dimensional polymer. Although DNA
can fold into 3D structures, even then, DNA is not stiff or
highly-bonded.
[0309] Large, three-dimensional polymers can be synthesized. For
example, a dendritic polymer of 2.times.108 Daltons has been
synthesized (Zhang, Wepf et al., "The Largest Synthetic Structure
with Molecular Precision: Towards a Molecular Object," Angewandte
Chemie International Edition, 3, WILEY-VCH Verlag, 2011). However,
the ability to precisely control the composition of such polymers
is lacking, and they tend to be relatively simple polymeric
sequences which have been joined in a manner that allows them to
assume a three-dimensional shape. The dendritic polymer synthesized
by (Zhang, Wepf et al., "The Largest Synthetic Structure with
Molecular Precision: Towards a Molecular Object," Angewandte Chemie
International Edition, 3, WILEY-VCH Verlag, 2011) is not stiff,
highly-bonded, or complex, and subsequent work on error rates at
various points in the molecule indicate that it is not
atomically-precise.
[0310] Structures consisting of multiple adamantane units in random
configurations have been purified from petroleum. These structures
are stiff and highly-bonded. Additionally, various chemical
processes are known to make modified or functionalized adamantane
(Szinai, "ADAMANTANE COMPOUNDS," U.S. Pat. No. 3,859,352, United
States, Eli Lilly and Company (Indianapolis, Ind.), 1975; Baxter,
"Adamantane derivatives," U.S. Pat. No. 6,242,470, United States,
AstraZeneca AB (Sodertalje, S E), 2001). However, the adamantane
aggregates obtained from natural sources are connected randomly,
and so the chances of finding any particular arrangement of
adamantanes as the size of the molecule grows becomes vanishingly
small. In practicality, these molecules are neither large nor
atomically-precise. The functionalized adamantanes used in the
pharmaceutical industry are atomically-precise, but they are not
large or highly-bonded (since such molecules tend to be, for
example, a single adamantane connected to a long, flexible side
chain).
[0311] Diamond, whether natural or synthetic (e.g., grown via
chemical vapor deposition) is neither complex, being (with the
exception of errors) a uniformly repeated three-dimensional polymer
of adamantane, nor atomically-precise, as even the most perfect
such diamond has flaws at the atomic level.
[0312] With respect to strained bonds, the creation of individual
strained bonds is routine in chemistry, and molecules like
cyclopropane and cubane exemplify the structures that can be
created with strained bonds. Larger structures containing many
strained bonds also exist, e.g., Fullerenes of various
configurations. While the specific mechanisms of formation are very
different, there is a commonality between the synthesis of
cyclopropane, cubane, Fullerenes, and other strained molecules in
that there are energetically-feasible sequential reaction pathways
leading from the initial reactants to the final product.
[0313] However, there are classes of strained structures for which
this is not true; there is no practical pathway from the component
atoms or molecules to the final product using only conventional
chemistry. To conceptually illustrate this principle, consider a
stiff, rod-shaped molecule. Now, bend the rod into a circle and
connect the ends. A hoop-shaped molecule is formed. While
hoop-shaped molecules abound, including all the cycloalkanes, and
the many other cyclo-polymers, the formation of such structures
rely upon some fairly restrictive requirements. The main
requirement for the formation of these strained structures is that
the two ends can be brought close enough together so that they can
be bonded together, changing the molecule from a linear structure
into a circular structure. The two ends of the linear molecule can
be closely approximated in a variety of ways. For example, the
molecule can be very small to begin with, so that even if the
molecule is straight, the two ends are both within reach of a
single reaction. Or, the molecule can be flexible enough that it
can bend into the necessary configuration. Or, the linear molecule
could have an inherent curve to it, making it already a partial
hoop and thereby leaving only a small gap to bridge.
[0314] But, consider a class of molecules that do not meet these
requirements. A long rod, if stiff enough, even if somewhat curved,
with a substantial gap between its ends, cannot be made into a hoop
through conventional chemistry techniques. Similarly, a stiff
two-dimensional molecule (e.g., a plane of diamond just one or two
adamantane layers thick) will be unable to curl into a tube
structure, both because of its stiffness, and possibly because
multiple bonds would have to simultaneously form to hold the new
tubular structure in place--a statistically unlikely event.
[0315] A stiff, long, potentially wide, structure with two sides
which are, atomically speaking, far apart, but which need to be
brought together to then undergo a bonding reaction to form a
stable hoop or cylinder may sound like a very contrived class of
structures. It is not. For example, it is exemplary of many of the
bearing designs which have been proposed for nano-devices, where an
axle revolves inside a stiff cylindrical ring or tube.
Mechanosynthesis can form such structures in a variety of ways,
such as by using force to approximate the necessary ends, or by
building a temporary jig around the structure that forces
intermediate structures into the necessary shape (and which can
then be removed once the desired structure is complete).
[0316] These are only examples. Comments similar to those about DNA
and dendritic polymers apply to other polymers as well, comments
similar to those about adamantane apply to the existence or
synthesis of other structures, comments similar to those made about
diamond apply to other crystals, and certainly rod or plane-shaped
structures that need to be folded into hoops or cylinders are not
the only example of how positional control allows the formation of
structures which could not be made via conventional chemistry due
to geometric issues.
[0317] Another problem with traditional chemical synthesis methods,
geometry issues like those described above aside, is that there is
no way to differentiate multiple sites which have similar or
identical chemical properties, and yet the end product requires
that they be treated differently. Linear polymer synthesis (e.g.,
DNA synthesis) is an exception, since it is possible to work only
at one or a few specific locations (e.g., the ends) of a growing
one-dimensional polymer, but these polymers are not stiff, or
amenable to the formation of precise, highly-bonded
three-dimensional structures.
[0318] Once molecules become two or three dimensional, the problem
of chemically-equivalent sites at different locations appears. For
example, consider a perfectly flat plane of diamond, onto which a
structure is to be built. Reactions are known which can add
additional carbon (or other) atoms to diamond; this is the basis
for CVD-based growth of diamond. However, with the exception of the
edges and corners of the plane, which have different bonding
structures by virtue of not having the same number of neighboring
carbon atoms as the atoms away from the edge, all the sites on the
surface of the plane are essentially chemically equivalent. There
is no way that CVD, or any non-positional technique can, for
example, start adding new atoms to the plane at arbitrary,
atomically-precise coordinates.
[0319] This concept of multiple chemically-similar or
chemically-identical sites is the reason that three-dimensional
dendritic polymers have a simple, repetitious structure: Whatever
reaction happens to one branch tends to happen to the equivalent
sites on all branches. Beyond dendritic polymers, this general
concept is one of the main reasons that synthetic chemistry cannot
create arbitrarily large and complex structures.
[0320] Certainly mechanosynthesis could be used to make products
including DNA and other polymers, small molecules, or repetitious
structures of low complexity. In fact, such products would be
superior in some ways. For example, products of 100% purity could
be created, potentially improving the properties of the product, as
well as eliminating waste, and the need for purification steps.
[0321] However, when speaking of the possible products of
mechanosynthesis, these are not the most important cases since such
products, even if inefficiently or imperfectly, can already be
created. The more important cases are those structures which cannot
reasonably be created or obtained by other means. For the
aforementioned reasons, these tend to be structures that are
atomically precise, large, highly bonded, and complex. Such
structures may occur with or without strain; the presence of at
least some kinds of strain makes it even more unlikely that any
method other than positionally-controlled chemistry can create such
a structure.
Reliability
[0322] Reliability is an important consideration in the design of
build sequences for multi-atom workpieces. Reaction reliability can
be achieved in a variety of ways, including use of reactions with
energy barriers sufficient to prevent spontaneous reactions at a
given temperature, reactions designed to avoid pathological side
reactions (for example, by approaching a workpiece using a
trajectory that favors only the desired reaction, or by ordering a
build sequence to avoid leaving unsatisfied valences in
self-reactive positions), or the introduction of a testing step
during mechanosynthesis. These topics are discussed in more detail
below.
[0323] In some cases, primarily with respect to hydrogen due to its
low atomic mass, tunneling can contribute to reaction error. These
errors can be reduced with slight modifications in build sequences
to avoid problematic situations. Also, deuterium could be used in
place of standard hydrogen. Deuterium's different mass and Van der
Waal's radius also has effects on reaction rates (the kinetic
isotope effect), vibrational frequencies, torsional coupling and
other properties. All of these effects may be exploited by choosing
to use hydrogen or deuterium on a case by case basis. Note that in
general, any isotope of an element could be used where its
properties are advantageous, and the ability to positionally
control isotopes of an element may useful, just as the positional
control of different elements is useful.
Reaction Barriers and Temperature
[0324] Note that equipment capabilities could have an effect on
reaction reliability. For example, the error in a positional means
is unlikely to be zero. However, it is well within the limits of
conventional atomic microscopy technology to attain high enough
positional accuracy that it essentially becomes irrelevant. With
equipment that can position a tip with a precision of, e.g., <20
pm, temperature becomes the dominating variable in reaction
reliability. As the positional means become less accurate, reaction
reliability suffers regardless of temperature, and for example,
positional errors of 50 pm or more will substantially reduce the
reliability of many mechanosynthetic reactions. Those skilled in
the art will understand how to incorporate such equipment
limitations into reaction reliability calculations, if necessary.
For exemplary purposes, only temperature is considered in the
following example of calculating reaction reliability.
[0325] One of the advantages of mechanosynthesis is that it
facilitates specific, desired reactions by using directed
mechanical force to overcome reaction barriers. In conventional
chemistry, reaction barriers or energy deltas are often overcome by
thermal energy. However, thermal energy is nonspecific and
facilitates desired and undesired reactions alike. Reducing
temperature decreases the thermal energy available to cause
non-specific reactions. This reduces the likelihood of pathological
side reactions while directed mechanical force, even at low
temperatures, still facilitates desired reactions.
[0326] The Arrhenius equation and other principles of
thermodynamics and computational chemistry may be used in
conjunction with data on net energy differences and energy barriers
to determine the reliability of a given reaction at a given
temperature. For example, the following Mathematica v8 code may be
used to determine reaction reliability at a given temperature when
considering the net energy difference between two structures (e.g.,
the before and after workpiece structures):
TABLE-US-00003 Code Listing 1: (** calculate reliability of a
reaction at a given temperature **) (** Define Constants and Unit
Conversions **) (** Boltzmann constant = 1.38*10{circumflex over (
)}-23 J/K **) boltzmann = 1.381*10{circumflex over ( )}-23; (**
convert eV to Joules **) jouleBarrier = barrier*1.6*10{circumflex
over ( )}-19; (** inputs for specific reaction **) (** reaction
barrier in eV **) barrier = Abs[-0.6418]; (** temp in Kelvin **)
temperature = 300; (** Calculate Probability of Failure **)
probability = NumberForm[Exp[-jouleBarrier/
(boltzmann*temperature)], 4]
Reliability in Build Sequences
[0327] The reliability of reactions across a build sequence can
provide one way of assessing the statistical error rate. And,
depending on which, or how many, errors are considered significant
enough to compromise workpiece function, these data can then be
used to assess workpiece yield (or performance, in a scenario where
workpieces do not simply pass/fail a quality check and the effect
of certain errors on workpiece function are known) in a
manufacturing setting. This is most easily explained by
example.
[0328] Consider a workpiece which requires 10{circumflex over ( )}6
reactions to create. For the sake of simplicity, assume that each
of these reactions are identical in their energy barrier to a
pathological reaction (an error), and that the barrier is 0.2 eV.
Another assumption is that simulations, practical experience, or
other information provide guidelines as to how many errors, on
average, may be present before a workpiece is deemed defective.
Arbitrarily, since this would vary with the workpiece design, a
limit of 10 errors is used for this example. Which is to say, a
workpiece having between 0 and 10 errors is acceptable, while a
workpiece having over 10 errors will be rejected as defective.
Finally, (again, arbitrarily to demonstrate the logic, since this
number will vary depending on the business and technical
requirements) a yield of at least 90% is required.
[0329] Since an error is presumed to be a rare event, error
occurrence is modeled as a Poisson distribution. The problem then
becomes one of determining 1, the number of expected events, where
the Cumulative Distribution Function is equal to or greater than
0.90 (a 90% yield) when the number of events is 10 (the maximum
number of tolerable errors). In this case, 1 is 7. Meaning, if one
expects, on average, that 7 errors will occur during the build
sequence, then 90% of the time, no more than 10 events will occur.
So, the expected number of errors must be <=7. Since the
hypothetical workpiece requires 10{circumflex over ( )}6 reactions
to build, the threshold for mistakes is 7/10{circumflex over ( )}6.
Using the equations herein to solve for the maximum allowable
temperature to attain this accuracy given a 0.2 eV barrier, the
answer is about 195 degrees Kelvin. Obviously this number can
change depending on actual reaction barriers, manufacturing
requirements, equipment capabilities, and other factors.
[0330] Note that these calculations assume that temperature is the
sole factor limiting reliability. As previously noted, there may be
other sources of error, caused by factors such as positional
uncertainty in the equipment, or Hydrogen tunneling, and these
could be factored in when assessing an actual manufacturing
process. Also, note the assumption that errors are statistically
independent. Error independence is unlikely in some scenarios,
since a missing or mis-bonded atom may cause subsequent problems
when placing neighboring atoms. However, this is not necessarily
the case, and regardless, the issue can be made irrelevant by
requiring an error rate approaching 0%.
[0331] Temperature and reaction barriers aside, considering the
statistics of the case where zero errors is the requirement
provides a way to compare the literature processes to the
reliability requirements needed for the creation of more complex
workpieces. The literature often describes experiments involving
between one and about twelve reactions. The literature does not
report error rates, but theoretically, how reliable must the
reactions be to perform, for example, twelve reactions with no
errors? A simple calculation (Reliability #Reactions=Yield) shows
that 90% reliability for each reaction would give a 28% yield. That
may be an acceptable, or even excellent, yield for a laboratory
process, but a fairly poor yield for an industrial manufacturing
process, and that is with only 12 reactions.
[0332] If the workpiece requires 20 reactions, a 90% reliability
for each reaction gives a yield of 12%. At 50 reactions, 90%
reliability provides a yield of only 0.5%. By 100 reactions, 90%
reliability is no longer reasonable as an error-free workpiece
would almost never be created. For 100 reactions, the reliability
needs to be more in the 95-99% range. And, for 1,000 reactions or
more, assuming that a yield of more than a few percent is desired,
the reliability needs to approach 100%.
[0333] Note that some reactions will be abstraction or
rearrangement reactions, while some will be addition reactions
which may add more than one atom at a time. On average, the number
of reactions probably exceeds the number of atoms in a given
workpiece, but the order of magnitude will be the same, so for ease
of discussion we will assume that a workpiece containing 20 atoms
takes about 20 reactions, a workpiece containing 50 atoms takes
about 50 reactions, a workpiece containing 100 atoms takes about
100 reactions, and a workpiece containing 1000 atoms takes about
1000 reactions, and so on.
[0334] Clearly, error rates that are acceptable for workpieces
requiring trivial numbers of reactions are unsatisfactory for
building more complex workpieces. Of course, this statement comes
with a number of assumptions, such as no error correction
processes, and little tolerance for errors in the finished
workpiece. But, in general, this illustrates the need for
rationally-designed build sequences, based on reactions of known
reliability, that permit reliability far in excess of that
evidenced in the literature (but well within the capabilities of
the reactions reported herein).
[0335] Of course, some useful build sequences are quite short. For
example, depending on whether the starting point is a
dehydrogenated Si surface or a brominated Si surface, the
half-Si-Rad build sequence described herein is only 11 to 15 steps
long. Similarly, initiating a new row on a diamond surface takes 11
reactions, extending the row takes 5 steps, and terminating a row
takes 6 steps (ignoring that such steps will often need to be
repeated--while this would frequently be the case, it cannot be
said to always be the case). Clearly, some build sequences may be
between 5 and 10, or between 10 and 20, steps long and still
accomplish something of value. In such circumstances, reliability
requirements for the individual reactions might be lower and still
result in success some reasonable percentage of the time, as
opposed to build sequences which have, e.g., 20 to 50, 50 to 100,
100 to 1,000, or more, steps.
* * * * *