U.S. patent application number 14/569372 was filed with the patent office on 2015-06-11 for integrated sensor arrays for biological and chemical analysis.
The applicant listed for this patent is LIFE TECHNOLOGIES CORPORATION. Invention is credited to James Bustillo, Kim L. Johnson, David Marran, Mark James Milgrew, Todd Rearick, Jonathan M. ROTHBERG, Jonathan Schultz.
Application Number | 20150160154 14/569372 |
Document ID | / |
Family ID | 42223357 |
Filed Date | 2015-06-11 |
United States Patent
Application |
20150160154 |
Kind Code |
A1 |
ROTHBERG; Jonathan M. ; et
al. |
June 11, 2015 |
INTEGRATED SENSOR ARRAYS FOR BIOLOGICAL AND CHEMICAL ANALYSIS
Abstract
The invention is directed to apparatus and chips comprising a
large scale chemical field effect transistor arrays that include an
array of sample-retaining regions capable of retaining a chemical
or biological sample from a sample fluid for analysis. In one
aspect such transistor arrays have a pitch of 10 .mu.m or less and
each sample-retaining region is positioned on at least one chemical
field effect transistor which is configured to generate at least
one output signal related to a characteristic of a chemical or
biological sample in such sample-retaining region. In one
embodiment, the characteristic of said chemical or biological
sample is a concentration of a charged species and wherein each of
said chemical field effect transistors is an ion-sensitive field
effect transistor having a floating gate with a dielectric layer on
a surface thereof, the dielectric layer contacting said sample
fluid and being capable of accumulating charge in proportion to a
concentration of the charged species in said sample fluid. In one
embodiment such charged species is a hydrogen ion such that the
sensors measure changes in pH of the sample fluid in or adjacent to
the sample-retaining region thereof. Apparatus and chips of the
invention may be adapted for large scale pH-based DNA sequencing
and other bioscience and biomedical applications.
Inventors: |
ROTHBERG; Jonathan M.;
(Guilford, CT) ; Bustillo; James; (Castro Valley,
CA) ; Milgrew; Mark James; (Brandford, CT) ;
Schultz; Jonathan; (Guilford, CT) ; Marran;
David; (Durham, CT) ; Rearick; Todd;
(Cheshire, CT) ; Johnson; Kim L.; (Carlsbad,
CA) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
LIFE TECHNOLOGIES CORPORATION |
Carlsbad |
CA |
US |
|
|
Family ID: |
42223357 |
Appl. No.: |
14/569372 |
Filed: |
January 21, 2015 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
13125133 |
Jul 1, 2011 |
8936763 |
|
|
PCT/US09/05745 |
Oct 22, 2009 |
|
|
|
14569372 |
|
|
|
|
61205626 |
Jan 22, 2009 |
|
|
|
61198222 |
Nov 4, 2008 |
|
|
|
61196953 |
Oct 22, 2008 |
|
|
|
Current U.S.
Class: |
506/38 ;
506/39 |
Current CPC
Class: |
C12Q 1/6874 20130101;
G01N 27/4145 20130101; C12Q 1/6874 20130101; H01L 21/82 20130101;
C12Q 2565/607 20130101; G01N 27/4148 20130101; C12Q 1/6869
20130101 |
International
Class: |
G01N 27/414 20060101
G01N027/414; C12Q 1/68 20060101 C12Q001/68 |
Claims
1. An apparatus comprising a chemical field effect transistor array
in a circuit-supporting substrate, such transistor array having
disposed on its surface an array of sample-retaining regions
capable of retaining a chemical or biological sample from a sample
fluid, wherein such transistor array has a pitch of 10 .mu.m or
less and each sample-retaining region is positioned on at least one
chemical field effect transistor which is configured to generate at
least one output signal related to a characteristic of a chemical
or biological sample in such sample-retaining region.
Description
RELATED APPLICATIONS
[0001] This application is a continuation of U.S. patent
application Ser. No. 13/125,133 filed Jul. 1, 2011, which is a
national stage filing under 35 U.S.C. .sctn.371 of PCT
International application No. PCT/US2009/005745 filed Oct. 22,
2009, which claims priority under 35 U.S.C. .sctn.119(e) to U.S.
Provisional Applications 61/196,953, 61/198,222, 61/205,626, filed
Oct. 22, 2008, Nov. 4, 2008 and Jan. 22, 2009, respectively, and
under 35 U.S.C. .sctn.120 to U.S. Non-Provisional application Ser.
Nos. 12/474,897 and 12/475,311, both filed May 29, 2009, the entire
contents of all of which are incorporated by reference herein.
FIELD OF THE DISCLOSURE
[0002] The present disclosure is directed generally to
semiconductor chips for making chemical measurements, and more
particularly, to single chip ISFET arrays (and arrays of single
chip ISFET arrays) for monitoring one or more analytes.
BACKGROUND
[0003] Rapid and accurate measurement of biological and chemical
analytes is important in many fields, ranging from diagnostics, to
industrial process control, to environmental monitoring, to
scientific research. Chemically sensitive, and in particular,
ion-sensitive field effect transistors ("chemFETs" and "ISFETs"
respectively) have been used for such measurements for many years,
e.g. Bergveld, Sensors and Actuators, 88: 1-20 (2003); Yuqing et
al., Biotechnology Advances, 21: 527-534 (2003); and the like. More
recently, attempts have been made to fabricate arrays of such
sensors using integrated circuit technologies to obtain spatially
distributed and multi-analyte measurements using a single device,
e.g., Yeow et al., Sensors and Actuators B 44: 434-440 (1997);
Martinoia et al., Biosensors & Bioelectronics, 16: 1043-1050
(2001); Milgrew et al., Sensors and Actuators B 103: 37-42 (2004);
Milgrew et al., Sensors and Actuators B, 111-112: 347-353 (2005);
Hizawa et al., Sensors and Actuators B, 117: 509-515 (2006); Heer
et al., Biosensors and Bioelectronics, 22: 2546-2553 (2007); and
the like. Such efforts face several difficult technical challenges,
particularly when ISFET sensor arrays have scales in excess of
thousands of sensor elements and densities in excess of many
hundreds of sensor elements per mm.sup.2. Such challenges include
making large-scale arrays with sensor elements having uniform
performance characteristics from sensor to sensor within the array,
and making sensor elements with footprints on the order of microns
which are capable of generating signals detectable against a
background of many noise sources from both the sensor array itself
and a fluidics system that conveys reactants or analyte-containing
samples to the array, For ISFET arrays comprising sensor elements
with charge-sensitive components, such as floating gates, the
former challenge is exacerbated by the accumulation of trapped
charge in or adjacent to such components, which is a common side
product of semiconductor fabrication technologies. The latter
challenge is exacerbated by the requirement that analytes of
interest directly or indirectly generate a charged species that
accumulates at or on a charge-sensitive component of the ISFET
sensor. In very dense arrays, diffusion, reactivity of the analyte
or its surrogate, cross-contamination from adjacent sensors, as
well as electrical noise in the sample fluid can all adversely
affect measurements. The availability of large-scale ISFET arrays
that overcome these challenges would be highly useful in the above
fields, particularly where ever highly parallel multiplex chemical
measurements are required, such as in the large scale genetic
analysis of genomes.
SUMMARY
[0004] Aspects of the invention relate in part to large arrays of
chemFETs or more specifically ISFETs for monitoring reactions,
including for example nucleic acid (e.g., DNA) sequencing
reactions, based on monitoring analytes present, generated or used
during a reaction. More generally, arrays including large arrays of
chemFETs may be employed to detect and measure static and/or
dynamic amounts or concentrations of a variety of analytes (e.g.,
hydrogen ions, other ions, non-ionic molecules or compounds, etc.)
in a variety of chemical and/or biological processes (e.g.,
biological or chemical reactions, cell or tissue cultures or
monitoring, neural activity, nucleic acid sequencing, etc.) in
which valuable information may be obtained based on such analyte
measurements. Such arrays may be employed in methods that detect
analytes and/or methods that monitor biological or chemical
processes via changes in charge at the surfaces of sensors in the
arrays, either by direct accumulation of charged products or by
indirect generation or capture of charged species related to the
concentration or presence of an analyte of interest. The present
invention is exemplified in a number of implementations and
applications, some of which are summarized below.
[0005] In one aspect the invention is directed to an apparatus
comprising a chemical field effect transistor array in a
circuit-supporting substrate, such transistor array having disposed
on its surface an array of sample-retaining regions capable of
retaining a chemical or biological sample from a sample fluid,
wherein such transistor array has a pitch of 10 .mu.m or less and
each sample-retaining region is positioned on at least one chemical
field effect transistor which is configured to generate at least
one output signal related to a characteristic of a chemical or
biological sample in such sample-retaining region. In one
embodiment, the characteristic of said chemical or biological
sample is a concentration or an amount of a charged species and
wherein each of said chemical field effect transistors is an
ion-sensitive field effect transistor having a floating gate with a
dielectric layer on a surface thereof, the dielectric layer
contacting said sample fluid and being capable of accumulating
charge in proportion to a concentration of the charged species in
said sample fluid. In one embodiment, such charged species is a
hydrogen ion such that the sensors measure changes in pH of the
sample fluid in or adjacent to the sample-retaining region thereof.
In an aspect of such embodiment, the dielectric layer has a
thickness selected to maximize capacitance thereacross, consistent
with other requirements. Such thickness may be selected in the
range of from 1 to 1000 nanometers (nm), or in a range of from 10
to 500 nm, or in a range of from 20 to 100 nm.
[0006] In another aspect, the invention is directed to an
integrated sensor array that comprises a plurality of sensors
formed in a circuit-supporting substrate, each sensor comprising a
chemical field effect transistor, the sensors being in a planar
array of greater than 256 sensors at a density greater than 100
sensors per mm.sup.2, each sensor of the array being configured to
provide at least one output signal related to a concentration or
presence of a chemical or biological sample proximate thereto, such
output signal being substantially the same for each sensor of the
array in response to the same concentration or presence of the same
chemical or biological sample. In one embodiment of this aspect,
the integrated sensor array further includes a plurality of
sample-retaining regions in said circuit-supporting substrate, each
sample-retaining region on (or alternatively below or beside) and
operationally associated with at least one of said sensors. In
another embodiment, such sample-retaining regions are each
microwells configured to hold a sample within a volume of sample
fluid. In another embodiment, sensors of the integrated sensor
array detect or measure a concentration of a charged species and
each of the sensors is an ion-sensitive field effect transistor
having a floating gate with a dielectric layer on a surface
thereof, the dielectric layer contacting a sample fluid containing
said chemical or biological sample and being capable of
accumulating charge in proportion to a concentration of the charged
species in the sample fluid adjacent thereto. The dielectric layer
has a thickness selected to maximize capacitance thereacross,
consistent with other requirements.
[0007] In a further aspect, the invention is directed to a single
chip chemical assay device that comprises: (a) a sensor array
formed in a circuit supporting substrate, each sensor of the array
comprising a chemical field-effect transistor and being configured
to provide at least one output signal related to a concentration or
presence of a chemical or biological sample proximate thereto, such
output signal being substantially the same for each sensor of the
array in response to the same concentration or presence of the same
chemical or biological sample; (b) a plurality of sample-retaining
regions in the circuit supporting substrate, each sample-retaining
region disposed on at least one sensors; and (c) control circuitry
in the circuit supporting substrate coupled to the sensor array to
receive samples of the output signals from said chemical field
effect transistors at a rate of at least one frame per second.
[0008] In still another aspect the invention is directed to a
single chip nucleic acid assay device that comprises: (a) a sensor
array formed in a circuit supporting substrate, each sensor of the
array comprising a chemical field-effect transistor and being
configured to provide at least one output signal related to a
concentration or presence of a chemical or biological sample
proximate thereto, such output signal being substantially the same
for each sensor of the array in response to the same concentration
or presence of the same chemical or biological sample; (b) a
plurality of sample-retaining regions in the circuit supporting
substrate, each sample-retaining region disposed on at least one
sensors; (c) supports, including solid supports such as particulate
solid supports, disposed on the sample-retaining regions, each
support having a concatemerized template attached thereto; and (d)
control circuitry in the circuit supporting substrate coupled to
the sensor array to receive samples of the output signals from said
chemical field effect transistors at a rate of at least one frame
per second. Supports may include beads, including beads having
solid or porous cores and/or solid or porous surfaces,
microspheres, microparticles, gel microdroplets and other separable
particulate supports for attaching DNA templates, particularly as
clonal populations. Such supports may be on the order of microns or
nanometers, depending on the application. For example, the beads
may be microbeads or they may be nanobeads.
[0009] These above-characterized aspects, as well as other aspects,
of the present invention are exemplified in a number of illustrated
implementations and applications, some of which are shown in the
figures and characterized in the claims section that follows.
However, the above summary is not intended to describe each
illustrated embodiment or every implementation of the present
invention.
BRIEF DESCRIPTION OF THE DRAWINGS
[0010] In the drawings, like reference characters generally refer
to the same parts throughout the different views. Also, the
drawings are not necessarily to scale, emphasis instead being
placed upon generally illustrating the various concepts discussed
herein.
[0011] FIG. 1 generally illustrates a nucleic acid processing
system comprising a large scale chemFET array, according to one
inventive embodiment of the present disclosure.
[0012] FIG. 2 illustrates one column of an chemFET array similar to
that shown in FIG. 1, according to one inventive embodiment of the
present disclosure.
[0013] FIG. 3 shows a composite cross-sectional view of multiple
neighboring pixels illustrating a layer-by-layer view of pixel
fabrication according to another inventive embodiment of the
present disclosure.
[0014] FIGS. 4A-4L are top views of patterns corresponding layers
of material laid down in the fabrication of sensors shown in FIG.
3.
[0015] FIG. 5A-5B illustrates process steps for fabricating an
array with thin dielectric layers on sensor floating gates, and a
dielectric layer comprising a charge-sensitive layer and an
adhesion layer.
[0016] FIG. 6 is a block diagram of an embodiment of the electronic
components of one embodiment of the invention.
[0017] FIG. 7 is an exemplary timing diagram for components shown
in FIG. 6.
[0018] FIG. 8A is a high-level, partially block, partially circuit
diagram showing a basic passive sensor pixel in which the voltage
changes on the ISFET source and drain inject noise into the
analyte, causing errors in the sensed values.
[0019] FIG. 8B is a high-level partially block, partially circuit
diagram showing a basic passive sensor pixel in which the voltage
changes on the ISFET drain are eliminated by tying it to ground,
the pixel output is obtained via a column buffer, and CDS is
employed on the output of the column buffer to reduce correlated
noise.
[0020] FIG. 8C is a high-level partially block, partially circuit
diagram showing a two-transistor passive sensor pixel in which the
voltage changes on the ISFET drain and source are substantially
eliminated, the pixel output is obtained via a buffer, and CDS is
employed on the output of the column buffer to reduce correlated
noise.
[0021] FIG. 9 is a cross-sectional view of a flow cell.
[0022] FIGS. 10A and B are graphs showing a trace from an ISFET
device (A) and a nucleotide readout (B) from a sequencing reaction
of a 23-mer synthetic oligonucleotide.
[0023] FIGS. 11A and B are graphs showing a trace from an ISFET
device (A) and a nucleotide readout (B) from a sequencing reaction
of a 25-mer PCR product.
DETAILED DESCRIPTION
[0024] While the invention is amenable to various modifications and
alternative forms, specifics thereof have been shown by way of
example in the drawings and will be described in detail. It should
be understood, however, that the intention is not to limit the
invention to the particular embodiments described. On the contrary,
the intention is to cover all modifications, equivalents, and
alternatives falling within the spirit and scope of the invention.
For example, the microelectronics portion of the apparatus and
array is implemented in CMOS technology for purposes of
illustration. It should be appreciated, however, that the
disclosure is not intended to be limiting in this respect, as other
semiconductor-based technologies may be utilized to implement
various aspects of the microelectronics portion of the systems
discussed herein. Guidance for making arrays of the invention is
found in many available references and treatises on integrated
circuit design and manufacturing and micromachining, including, but
not limited to, Allen et al., CMOS Analog Circuit Design (Oxford
University Press, 2.sup.nd Edition, 2002); Levinson, Principles of
Lithography, Second Edition (SPIE Press, 2005); Doering and Nishi,
Editors, Handbook of Semiconductor Manufacturing Technology, Second
Edition (CRC Press, 2007); Baker, CMOS Circuit Design, Layout, and
Simulation (IEEE Press, Wiley-Interscience, 2008); Veendrick,
Deep-Submicron CMOS ICs (Kluwer-Deventer, 1998); Cao,
Nanostructures & Nanomaterials (Imperial College Press, 2004);
and the like, which relevant parts are hereby incorporated by
reference.
[0025] In one aspect the invention is directed to integrated sensor
arrays (e.g., a two-dimensional array) of chemically-sensitive
field effect transistors (chemFETs), where the individual chemFET
sensor elements or "pixels" of the array are configured to detect
analyte presence (or absence), analyte levels (or amounts), and/or
analyte concentration in a sample fluid. "Sample fluid" means a
fluid which is used to deliver sample or reagents to a
sample-retaining region or to remove products or reactants from
sample-retaining regions, such as a wash fluid of a multi-step
reaction. Sample fluids may remain the same or may change in the
course of an analytical process, or processes, taking place on an
array. Analytical processes detected or monitored by arrays of the
invention include chemical and/or biological processes (e.g.,
chemical reactions, cell cultures, neural activity, nucleic acid
sequencing reactions, etc.) occurring in proximity to the array.
Examples of chemFETs contemplated by various embodiments discussed
in greater detail below include, but are not limited to,
ion-sensitive field effect transistors (ISFETs) and
enzyme-sensitive field effect transistors (EnFETs). In one
exemplary implementation, one or more microfluidic structures
is/are fabricated above the chemFET sensor array to provide for
retention, containment and/or confinement of a biological or
chemical reaction in which an analyte of interest may be captured,
produced, or consumed, as the case may be. Such structures define
regions on an array referred to herein as "sample-retaining
regions." For example, in one implementation, the microfluidic
structure(s) may be configured as one or more wells (or microwells,
or reaction chambers, or reaction wells as the terms are used
interchangeably herein) disposed above one or more sensors of the
array, such that the one or more sensors over which a given well is
disposed detect and measure analyte presence, level, and/or
concentration in the given well. Preferably, there is a 1:1 ratio
of wells and sensors.
[0026] A chemFET array according to various inventive embodiments
of the present disclosure may be configured for sensitivity to any
one or more of a variety of analytes. In one embodiment, one or
more chemFETs of an array may be particularly configured for
sensitivity to one or more analytes, and in other embodiments
different chemFETs of a given array may be configured for
sensitivity to different analytes. For example, in one embodiment,
one or more sensors (pixels) of the array may include a first type
of chemFET configured to be sensitive to a first analyte, and one
or more other sensors of the array may include a second type of
chemFET configured to be sensitive to a second analyte different
from the first analyte. In one embodiment, the first and second
analytes may be related to each other. As an example, the first and
second analytes may be byproducts of the same biological or
chemical reaction/process and therefore they may be detected
concurrently to confirm the occurrence of a reaction (or lack
thereof). Such redundancy is preferable in some analyte detection
methods. Of course, it should be appreciated that more than two
different types of chemFETs may be employed in any given array to
detect and/or measure different types of analytes, and optionally
to monitor biological or chemical processes such as binding events.
In general, it should be appreciated in any of the embodiments of
sensor arrays discussed herein that a given sensor array may be
"homogeneous" and thereby consist of chemFETs of substantially
similar or identical type that detect and/or measure the same
analyte (e.g., pH or other ion concentration), or a sensor array
may be "heterogeneous" and include chemFETs of different types to
detect and/or measure different analytes. In another embodiment,
the sensors in an array may be configured to detect and/or measure
a single type (or class) of analyte even though the species of that
type (or class) detected and/or measured may be different between
sensors. As an example, all the sensors in an array may be
configured to detect and/or measure nucleic acids, but each sensor
detects and/or measures a different nucleic acid.
[0027] As used herein, an array is a planar arrangement of elements
such as sensors or wells. The array may be one or two dimensional.
A one dimensional array is an array having one column (or row) of
elements in the first dimension and a plurality of columns (or
rows) in the second dimension. An example of a one dimensional
array is a 1.times.5 array. A two dimensional array is an array
having a plurality of columns (or rows) in both the first and the
second dimensions. The number of columns (or rows) in the first and
second dimensions may or may not be the same. An example of a two
dimensional array is a 5.times.10 array.
[0028] Accordingly, one embodiment is directed to an apparatus,
comprising an array of CMOS-fabricated sensors, each sensor
comprising one chemically-sensitive field effect transistor
(chemFET) and occupying an area on a surface of the array of 10
.mu.m.sup.2 or less, 9 .mu.m.sup.2 or less, 8 .mu.m.sup.2 or less,
7 .mu.m.sup.2 or less, 6 .mu.m.sup.2 or less, 5 .mu.m.sup.2 or
less, 4 .mu.m.sup.2 or less 3 .mu.m.sup.2 or less, or 2 .mu.m.sup.2
or less. Another embodiment of the invention is directed to an
apparatus comprising a planar array of CMOS-fabricated sensors in a
density of at least 100 sensors/mm.sup.2, or at least 250
sensors/mm.sup.2, or at least 1000 sensors/mm.sup.2, or at least
5000 sensors/mm.sup.2, or at least 10000 sensors/mm.sup.2.
[0029] Another embodiment is directed to an apparatus, comprising
an array of CMOS-fabricated sensors, each sensor comprising one
chemically-sensitive field effect transistor (chemFET). The array
of CMOS-fabricated sensors includes more than 256 sensors, and a
collection of chemFET output signals from all chemFETs of the array
constitutes a frame of data. The apparatus further comprises
control circuitry coupled to the array and configured to generate
at least one array output signal to provide multiple frames of data
from the array at a frame rate of at least 1 frame per second. In
one aspect, the frame rate may be at least 10 frames per second. In
another aspect, the frame rate may be at least 20 frames per
second. In yet other aspects, the frame rate may be at least 30,
40, 50, 70 or up to 100 frames per second. Chemical or biological
phenomena measured by the sensors may result in electrical signals,
usually current or voltage level changes, having a wide variety of
durations and amplitudes. In some embodiments, such signals may
have a duration in the range of a few milliseconds, e.g. 10 msec,
to many seconds, e.g. 10-20 sec. For cyclical or sequential
processes monitored by arrays, such signals may be substantially
repeated over an interval of minute, hours, or days. Moreover, the
signal may have superimposed various types of noise, including
flicker noise from the array itself, and thermal noise,
particularly from the sample fluid. The control circuitry of the
apparatus is configured to sample each signal from each sensor in
the array for a complete frame of data. Such sampling may be for a
short duration, e.g. 1-100 .mu.sec, or the like, so that complete
frames of data may be readout in the frame per second values listed
above.
[0030] As explained more fully below, arrays of the invention have
been fabricated in a manner that reduces the effects of trapped
charge, which includes lack of stability and uniformity of
sensor-to-sensor responses to the same or similar sensing
conditions by sensor in the same array. In one embodiment, such
responses may be measured by exposing sensors of an array to
predefined changes in pH. In one embodiment, responses of at least
95 percent of the sensors in an array are substantially linear over
a range of from 7 to 9 pH and has a voltage output signal with a
sensitivity of at least 40 mV/pH unit. In another embodiment, at
least 98 percent of sensor of an array have such performance.
"Substantially linear" means that measured pH values over such pH
range has a coefficient of variation of deviation from linearity of
at least 10 percent, or at least 5 percent, or at least 2 percent.
In another embodiment, at least 95 percent of sensors of such
arrays each have a response time to a change of pH of less than 200
msec.
[0031] Another embodiment is directed to a sensor array, comprising
a plurality of electronic sensors arranged in a plurality of rows
and a plurality of columns. Each sensor comprises one
chemically-sensitive field effect transistor (chemFET) configured
to provide at least one and in some instances at least two output
signals representing a presence and/or a concentration of an
analyte proximate to a surface of the array. For each column of the
plurality of columns, the array further comprises column circuitry
configured to provide a constant drain current and a constant
drain-to-source voltage to respective chemFETs in the column, the
column circuitry including two operational amplifiers and a
diode-connected FET arranged in a Kelvin bridge configuration with
the respective chemFETs to provide the constant drain-to-source
voltage.
[0032] Another embodiment is directed to a sensor array, comprising
a plurality of electronic sensors arranged in a plurality of rows
and a plurality of columns. Each sensor comprises one
chemically-sensitive field effect transistor (chemFET) configured
to provide at least one output signal and in some instances at
least two output signals representing a concentration of ions in a
solution proximate to a surface of the array. The array further
comprises at least one row select shift register to enable
respective rows of the plurality of rows, and at least one column
select shift register to acquire chemFET output signals from
respective columns of the plurality of columns.
[0033] The apparatus and devices of the invention can be used to
detect and/or monitor interactions between various entities. These
interactions include biological and chemical reactions and may
involve enzymatic reactions and/or non-enzymatic interactions such
as but not limited to binding events. As an example, the invention
contemplates monitoring enzymatic reactions in which substrates
and/or reagents are consumed and/or reaction intermediates,
byproducts and/or products are generated. An example of a reaction
that can be monitored according to the invention is a nucleic acid
synthesis method such as one that provides information regarding
nucleic acid sequence. This reaction will be discussed in greater
detail herein.
Apparatus
[0034] Apparatus of the invention can vary widely depending on the
analyte being detected, whether assay reactions with auxiliary
reagents are required, and whether sequential or cyclical reactions
are required. In one aspect, apparatus of the invention comprises a
sensor array and an array of sample-retaining regions on a surface
thereof for retaining biological or chemical analytes delivered to
the surface by a sample fluid. In one embodiment, sample-retaining
regions are integral with the sensor array and may have a wide
variety of formats. Such regions may be defined by chemically
reactive group on the surface of the sensor array, by binding
compounds attached to the surface of the sensor array which are
specific for predetermined analytes, by regions of hydrophobicity
or hydrophilicity, or by spatial features such as microwells,
cavities, weirs, dams, reservoirs, or the like. In additional
embodiments, apparatus of the invention may comprise sample
carriers, such as beads, particles, gel microdroplets, or other
supports, structures or substances which hold analytes of interest
and which may be delivered to sample-retaining regions by a sample
fluid. Such sample carriers may include binding moieties or
reactive groups to permit attachment to sample-retaining regions.
Such attachment may be specific wherein such binding moieties or
reactive groups form linkages with only complementary binding
compounds or functionalities, or the attachment may be random where
a sample carrier has a substantially equal likelihood of being
retained in any sample-retaining region of an array. As described
more fully below, in one embodiment, sample-retaining regions are
arrays of microwells that each have walls and an interior for
physically retaining sample, analyte and/or one or more sample
carriers.
[0035] As is exemplified below, sample fluid and samples or
analytes may be delivered to retaining regions of a sensor array in
several ways. Where sensor arrays are employed as one-use sample
characterization devices, such as with process, environmental or
cellular monitoring, sample may be delivered by emersion,
pipetting, pouring, or by other direct methods. Where sensor arrays
are employed in sequential or cyclical reactions, such as in DNA
sequencing, sample fluidic, including nucleic acid templates,
reagents, wash solutions, and the like, may be delivered by a
fluidic system under computer control. For such latter
applications, embodiments of the invention may further include a
flow cell integrated with the sample-retaining regions and sensor
array. As described more fully below, in one embodiment, such flow
cell delivers sample fluids (including assay reactants, buffers,
and the like) to sample-retaining regions under controlled
conditions, which may include laminar flow, constant flow rate at
each sample-retaining region, controlled temperature, minimization
of bubbles or other flow disruptions, and the like. In one aspect,
a flow cell of an apparatus of the invention comprises an inlet, an
outlet, and an interior space, which when the flow cell is in
communication with, for example sealingly bonded to, the arrays of
sample-retaining regions and sensors forms a chamber that is closed
except for the inlet and outlet. In some embodiments, the device is
manufactured such that the flow cell and one or both the arrays are
integral to each other. In other embodiments, the flow cell is
sealingly bonded to the arrays. Either embodiment will prevent
fluid leakage, which, among other possible hazards, would introduce
electrical noise into the sample fluid. In one aspect, the
apparatus of the invention includes a reference electrode in fluid
contact with the sample fluid so that during operation an
electrical potential difference is established between the
reference electrode and the sensors of the array.
[0036] An exemplary apparatus is shown in FIG. 1 which is adapted
for nucleic acid sequencing. In the discussion that follows, the
chemFET sensors of the array are described for purposes of
illustration as ISFETs configured for sensitivity to static and/or
dynamic ion concentration, including but not limited to hydrogen
ion concentration. However, it should be appreciated that the
present disclosure is not limited in this respect, and that in any
of the embodiments discussed herein in which ISFETs are employed as
an illustrative example, other types of chemFETs may be similarly
employed in alternative embodiments, as discussed in further detail
below. Similarly it should be appreciated that various aspects and
embodiments of the invention may employ ISFETs as sensors yet
detect one or more ionic species that are not hydrogen ions.
[0037] The system 1000 includes a semiconductor/microfluidics
hybrid structure 300 comprising an ISFET sensor array 100 and a
microfluidics flow cell 200. In one aspect, the flow cell 200 may
comprise a number of wells (not shown in FIG. 1) disposed above
corresponding sensors of the ISFET array 100. In another aspect,
the flow cell 200 is configured to facilitate the sequencing of one
or more identical template nucleic acids disposed in the flow cell
via the controlled and ordered introduction to the flow cell of a
number of sequencing reagents 272 (e.g., dATP, dCTP, dGTP, dTTP
(generically referred to herein as dNTP), divalent cations such as
but not limited to Mg.sup.2+, wash solutions, and the like).
[0038] As illustrated in FIG. 1, the introduction of the sequencing
reagents to the flow cell 200 may be accomplished via one or more
valves 270 and one or more pumps 274 that are controlled by a
computer 260. A number of techniques may be used to admit (i.e.,
introduce) the various processing materials (i.e., solutions,
samples, reaction reagents, wash solutions, and the like) into the
wells of such a flow cell. As illustrated in FIG. 1, reagents
including dNTP may be admitted to the flow cell (e.g., via the
computer controlled valve 270 and pumps 274) from which they
diffuse into the wells, or reagents may be added to the flow cell
by other means such as an ink jet. In yet another example, the flow
cell 200 may not contain any wells, and diffusion properties of the
reagents may be exploited to limit cross-talk between respective
sensors of the ISFET array 100, or nucleic acids may be immobilized
on the surfaces of sensors of the ISFET array 100.
[0039] The flow cell 200 in the system of FIG. 1 may be configured
in a variety of manners to provide one or more analytes (or one or
more reaction solutions) in proximity to the ISFET array 100. For
example, a template nucleic acid may be directly attached or
applied in suitable proximity to one or more pixels of the sensor
array 100, or in or on a support material (e.g., one or more
"beads") located above the sensor array but within the reaction
chambers, or on the sensor surface itself. Processing reagents
(e.g., enzymes such as polymerases) can also be placed on the
sensors directly, or on one or more solid supports (e.g., they may
be bound to the capture beads or to other beads) in proximity to
the sensors, or they may be in solution and free-flowing. It is to
be understood that the device may be used without wells or
beads.
[0040] In the system 1000 of FIG. 1, according to one embodiment
the ISFET sensor array 100 monitors ionic species, and in
particular, changes in the levels/amounts and/or concentration of
ionic species, including hydrogen ions. In important embodiments,
the species are those that result from a nucleic acid synthesis or
sequencing reaction.
[0041] Via an array controller 250 (also under operation of the
computer 260), the ISFET array may be controlled so as to acquire
data (e.g., output signals of respective ISFETs of the array)
relating to analyte detection and/or measurements, and collected
data may be processed by the computer 260 to yield meaningful
information associated with the processing (including sequencing)
of the template nucleic acid.
[0042] With respect to the ISFET array 100 of the system 1000 shown
in FIG. 1, in one embodiment the array 100 is implemented as an
integrated circuit designed and fabricated using standard CMOS
processes (e.g., 0.35 micrometer process, 0.18 micrometer process),
comprising all the sensors and electronics needed to
monitor/measure one or more analytes and/or reactions. With
reference again to FIG. 1, one or more reference electrodes 76 to
be employed in connection with the ISFET array 100 may be placed in
the flow cell 200 (e.g., disposed in "unused" wells of the flow
cell) or otherwise exposed to a reference (e.g., one or more of the
sequencing reagents 172) to establish a baseline against which
changes in analyte concentration proximate to respective ISFETs of
the array 100 are compared. The reference electrode(s) 76 may be
electrically coupled to the array 100, the array controller 250 or
directly to the computer 260 to facilitate analyte measurements
based on voltage signals obtained from the array 100; in some
implementations, the reference electrode(s) may be coupled to an
electric ground or other predetermined potential, or the reference
electrode voltage may be measured with respect to ground, to
establish an electric reference for ISFET output signal
measurements, as discussed further below.
[0043] More generally, a chemFET array according to various
embodiments of the present disclosure may be configured for
sensitivity to any one or more of a variety of analytes. In one
embodiment, one or more chemFETs of an array may be particularly
configured for sensitivity to one or more analytes and/or one or
more binding events, and in other embodiments different chemFETs of
a given array may be configured for sensitivity to different
analytes. For example, in one embodiment, one or more sensors
(pixels) of the array may include a first type of chemFET
configured to be sensitive to a first analyte, and one or more
other sensors of the array may include a second type of chemFET
configured to be sensitive to a second analyte different from the
first analyte. In one exemplary implementation, both a first and a
second analyte may indicate a particular reaction such as for
example nucleotide incorporation in a sequencing-by-synthesis
method. Of course, it should be appreciated that more than two
different types of chemFETs may be employed in any given array to
detect and/or measure different types of analytes and/or other
reactions. In general, it should be appreciated in any of the
embodiments of sensor arrays discussed herein that a given sensor
array may be "homogeneous" and include chemFETs of substantially
similar or identical types to detect and/or measure a same type of
analyte (e.g., hydrogen ions), or a sensor array may be
"heterogeneous" and include chemFETs of different types to detect
and/or measure different analytes.
[0044] In other aspects of the system shown in FIG. 1, one or more
array controllers 250 may be employed to operate the ISFET array
100 (e.g., selecting/enabling respective pixels of the array to
obtain output signals representing analyte measurements). In
various implementations, one or more components constituting one or
more array controllers may be implemented together with pixel
elements of the arrays themselves, on the same integrated circuit
(IC) chip as the array but in a different portion of the IC chip,
or off-chip. In connection with array control, analog-to-digital
conversion of ISFET output signals may be performed by circuitry
implemented on the same integrated circuit chip as the ISFET array,
but located outside of the sensor array region (locating the analog
to digital conversion circuitry outside of the sensor array region
allows for smaller pitch and hence a larger number of sensors, as
well as reduced noise). In various exemplary implementations
discussed further below, analog-to-digital conversion can be 4-bit,
8-bit, 12-bit, 16-bit or other bit resolutions depending on the
signal dynamic range required.
[0045] In general, data may be removed from the array in serial or
parallel or some combination thereof. On-chip controllers (or sense
amplifiers) can control the entire chip or some portion of the
chip. Thus, the chip controllers or signal amplifiers may be
replicated as necessary according to the demands of the
application. The array may, but need not be, uniform. For instance,
if signal processing or some other constraint requires instead of
one large array multiple smaller arrays, each with its own sense
amplifiers or controller logic, that is quite feasible.
[0046] Having provided a general overview of the role of a chemFET
(e.g., ISFET) array 100 in an exemplary system 1000 for measuring
one or more analytes, following below are more detailed
descriptions of exemplary chemFET arrays according to various
inventive embodiments of the present disclosure that may be
employed in a variety of applications. Again, for purposes of
illustration, chemFET arrays according to the present disclosure
are discussed below using the particular example of an ISFET array,
but other types of chemFETs may be employed in alternative
embodiments. Also, again, for purposes of illustration, chemFET
arrays are discussed in the context of nucleic acid sequencing
applications, however, the invention is not so limited and rather
contemplates a variety of applications for the chemFET arrays
described herein.
Sensor Layout and Array Fabrication
[0047] Methods of sensor layout design and array fabrication are
described in Rothberg et al., U.S. patent publications 2009/0026082
and 2009/0127589. In particular, techniques are disclosed for
reducing or eliminating trapped charged during; accordingly, these
references are incorporated by reference. In one aspect, the sensor
design and signal readout circuitry may be employed in the present
invention. For example, in one embodiment shown in FIG. 2, each
chemFET of an array comprises a floating gate structure, and a
source and a drain having a first semiconductor type and fabricated
in a region having a second semiconductor type, wherein there is no
electrical conductor that electrically connects the region having
the second semiconductor type to either the source or the drain.
Each sensor consists of three field effect transistors (FETs)
including the chemFET, and each sensor includes a plurality of
electrical conductors electrically connected to the three FETs. The
three FETs are arranged such that the plurality of electrical
conductors includes no more than four conductors traversing an area
occupied by each sensor and interconnecting multiple sensors of the
array. All of the FETs in each sensor are of a same channel type
and implemented in a single semiconductor region of an array
substrate. A collection of chemFET output signals from all chemFETs
of the array constitutes a frame of data. The apparatus further
comprises control circuitry coupled to the array and configured to
generate at least one array output signal to provide multiple
frames of data from the array at a frame rate of at least 20 frames
per second.
[0048] As an example of the integration of microwell and sensor
arrays, FIG. 3 shows a composite cross-sectional view of
neighboring pixels illustrating a layer-by-layer view of the pixel
fabrication and relative positions of floating gates and
microwells. Three adjacent pixels are shown in cross-section. All
of the FET components of the pixel 105.sub.1 are fabricated as
p-channel FETs in the single n-type well 154. Additionally, in the
composite cross-sectional view of FIG. 3 the highly doped p-type
region 159 is also visible corresponding to the shared drain (D) of
the MOSFETs Q2 and Q3. For purposes of illustration, the
polysilicon gate 166 of the MOSFET Q3 also is visible in FIG. 3.
However, for simplicity, the respective sources of the MOSFETs Q2
and Q3 shown in FIG. 2, as well as the gate of Q2, are not visible
in FIG. 3, as they lie along the same axis (i.e., perpendicular to
the plane of the figure) as the shared drain. The topmost metal
layer 304 corresponds to the ISFETs sensitive area 178, above which
is disposed an analyte-sensitive passivation layer 172. The topmost
metal layer 304, together with the ISFET polysilicon gate 164 and
the intervening conductors 306, 308, 312, 316, 320, 326 and 338,
form the ISFETs floating gate structure 170. However, an electrical
connection to the ISFETs drain is provided by the conductors 340,
328, and 318, coupled to the line 116.sub.1 which is formed in the
Metal2 layer rather than the Metal3 layer. Additionally, the lines
112.sub.1 and 114.sub.1 also are shown in FIG. 3 as formed in the
Metal2 layer rather than the Metal3 layer. The configuration of
these lines, as well as the line 118.sub.1, may be further
appreciated from the respective images of FIGS. 4A through 4L; in
particular, it may be observed in FIG. 4F that the line 118.sub.1,
together with the metal conductor 322, is formed in the Metal1
layer, and it may be observed that the lines 112.sub.1, 114.sub.1
and 116.sub.1 are formed in the Metal2 layer, leaving only the
jumper 308 of the floating gate structure 170 in the Metal3 layer
shown in FIG. 4J.
[0049] Accordingly, by consolidating the signal lines 112.sub.1,
114.sub.1, 116.sub.1 and 118.sub.1 to the Metal1 and Metal2 layers
and thereby increasing the distance between these signal lines and
the topmost layer 304 of the floating gate structure 170 in the
Metal4 layer, parasitic capacitances in the ISFET may be at least
partially mitigated. It should be appreciated that this general
concept (e.g., including one or more intervening metal layers
between signal lines and topmost layer of the floating gate
structure) may be implemented in other fabrication processes
involving greater numbers of metal layers. For example, distance
between pixel signal lines and the topmost metal layer may be
increased by adding additional metal layers (more than four total
metal layers) in which only jumpers to the topmost metal layer are
formed in the additional metal layers. In particular, a
six-metal-layer fabrication process may be employed, in which the
signal lines are fabricated using the Metal1 and Metal2 layers, the
topmost metal layer of the floating gate structure is formed in the
Metal6 layer, and jumpers to the topmost metal layer are formed in
the Metal3, Metal4 and Metal5 layers, respectively (with associated
vias between the metal layers).
[0050] In yet another aspect relating to reduced capacitance, a
dimension "f" of the topmost metal layer 304 (and thus the ISFET
sensitive area 178) may be reduced so as to reduce
cross-capacitance between neighboring pixels. As may be observed in
FIG. 4 (and as discussed further below in connection with other
embodiments directed to well fabrication above an ISFET array), the
well 725 may be fabricated so as to have a tapered shape, such that
a dimension "g" at the top of the well is smaller than the pixel
pitch "e" but yet larger than a dimension "f" at the bottom of the
well. Based on such tapering, the topmost metal layer 304 also may
be designed with the dimension "f" rather than the dimension "g" so
as to provide for additional space between the top metal layers of
neighboring pixels. In some illustrative non-limiting
implementations, for pixels having a dimension "e" on the order of
9 micrometers the dimension "f" may be on the order of 6
micrometers (as opposed to 7 micrometers, as discussed above), and
for pixels having a dimension "e" on the order of 5 micrometers the
dimension "f" may be on the order of 3.5 micrometers.
[0051] Detection of hydrogen ions, and other analytes as determined
by the invention, can be carried out using a passivation layer made
of silicon nitride (Si.sub.3N.sub.4), silicon oxynitride
(Si.sub.2N.sub.2O), silicon oxide (SiO.sub.2), aluminum oxide
(Al.sub.2O.sub.3), tantalum pentoxide (Ta.sub.2O.sub.5), tin oxide
or stannic oxide (SnO.sub.2), and the like.
[0052] When a dielectric layer is added over the floating gate
structure of the ISFET sensor arrangement, the path from the
analyte to the ISFET gate may be modeled as a series connection of
three capacitances: (1) the capacitance attributable to the
above-described charge double layer at the analyte-dielectric layer
interface (labeled C.sub.DL), (2) the capacitance due to the
floating gate dielectric layer (C.sub.FGD), and (3) the gate oxide
capacitance (C.sub.OX). (Note that in the text above, the floating
gate dielectric layer is sometimes referred to as a "passivation"
layer. Here, we refer more specifically to the layer as a floating
gate dielectric layer in order to avoid any suggestion that the
material composition of the layer is necessarily related to the
so-called passivation material(s) often used in CMOS processing
(e.g., PECVD silicon nitride) to coat and protect circuit
elements.) The series capacitance string extends between the liquid
analyte in the wells and the ISFET gate.
[0053] It is well known that capacitances in series form a
capacitive voltage divider. Consequently, only a fraction of the
signal voltage, V.sub.S, generated by or in the analyte, is applied
to the gate oxide as the voltage V.sub.G that drives the ISFET. If
we define the gate gain as V.sub.G/V.sub.S, one would ideally like
to have unity gain--i.e., no signal loss across any of the three
capacitances. The value of C.sub.DL is a function of material
properties and is typically on the order of about 10-40
.mu.F/cm.sup.2. The gate oxide capacitance is typically a very
small value by comparison. Thus, by making C.sub.FGD much greater
than the series combination of C.sub.OX and C.sub.DL (for short,
C.sub.FGD>>C.sub.OX), the gate gain can be made to approach
unity as closely as is practical.
[0054] To achieve the relationship C.sub.FGD>>C.sub.OX, one
can minimize C.sub.OX, maximize C.sub.FGD, or both. Maximization of
C.sub.FGD can be achieved by using a thin layer of high dielectric
constant material, or by increasing the area of the floating gate
metallization. The capacitance C.sub.FGD is essentially formed by a
parallel plate capacitor having the floating gate dielectric layer
as its dielectric. Consequently, for a given plate (i.e., floating
gate metallization) area, the parameters principally available for
increasing the value of C.sub.FGD are (1) the thickness of the
dielectric layer and (2) the selection of the dielectric material
and, hence, its dielectric constant. The capacitance of the
floating gate dielectric layer varies directly with its dielectric
constant and inversely with its thickness. Thus, a thin,
high-dielectric-constant layer would be preferred, to satisfy the
objective of obtaining maximum gate gain.
[0055] One candidate for the floating gate dielectric layer
material is the passivation material used by standard CMOS foundry
processes. The standard (typically, PECVD nitride or, to be more
precise, silicon nitride over silicon oxynitride) passivation layer
is relatively thick when formed (e.g., about 1.3 .mu.m), and
typical passivation materials have a limited dielectric constant. A
first improvement can be achieved by thinning the passivation layer
after formation. This can be accomplished by etching back the CMOS
passivation layer, such as by using an over-etch step during
microwell formation, to etch into and consume much of the nitride
passivation layer, leaving a thinner layer, such as a layer only
about 200-600 Angstroms thick.
[0056] Two approaches have been used for etching a standard CMOS
passivation layer of silicon nitride deposited over silicon
oxynitride. A first approach is referred to as the "partial etch"
technique; it involves etching away the silicon nitride layer plus
approximately half of the silicon oxynitride layer before
depositing the thin-film metal oxide sensing layer. The second
approach referred to as the "etch-to-metal" technique, involves
etching away all of the silicon nitride and silicon oxynitride
layers before depositing the thin-film metal oxide sensing layer.
With an ALD Ta.sub.2O.sub.5 thin-film sensing layer deposited over
a "partial etch," ISFET gains from about 0.37 to about 0.43 have
been obtained empirically, with sensor sensitivities of about
15.02-17.08 mV/pH.
[0057] An alternative is to simply deposit a thinner layer of
dielectric (passivation) material in the first place, such as the
indicated 200-600 Angstroms instead of the 1.3 .mu.m of the
conventional CMOS passivation process. Materials useful for the
floating gate dielectric layer are metal oxides such as tantalum
oxide, tungsten oxide, aluminum oxide, and hafnium oxide, though
other materials of dielectric constant greater than that of the
usual silicon nitride passivation material may be substituted,
provided that such material is, or can be rendered, sensitive to
the ion of interest. The etch-to-metal approach is preferred, with
the CMOS process' passivation oxide on the floating gate being
etched completely away prior to depositing the floating gate
dielectric material layer. That dielectric layer may be applied
directly on the metal extended ISFET floating gate electrode. This
will help maximize the value of the capacitance C.sub.FGD. FIGS. 5A
and 5B show steps using readily available fabrication techniques
for generating a dielectric layer for high capacitive coupling to
array sensor plates. Process steps additionally provide electrical
access to bondpad structures for off-chip communication. Initially
wafer (500) from a semiconductor manufacturer is treated to apply
material layer (502) from which microwells are formed (in this
example, TEOS), after which microwells are formed by etching to the
metal of the sensor plate (504). Dielectric layer (506) is added,
for example, by atomic layer deposition. In one embodiment, as
illustrated in FIG. 5B, dielectric layer (506) comprises a
charge-sensitive layer (512) and an adhesion layer (514). Using
alternative techniques and materials either single or
multiple-component dielectric layers may be formed. Tantalum oxide
and aluminum are exemplary charge-sensitive and adhesion layers of
dielectric layer (506). As mentioned above, other materials from
which dielectric layer (506) may be formed include Ta.sub.2O.sub.5,
Al.sub.2O.sub.3, HfO.sub.3 or WO.sub.3. In particular, such
materials result in a larger signal in response to pH changes in a
sample fluid. Iridium oxide may also be used, e.g. as described in
D. O. Wipf et al., "Microscopic Measurement of pH with Iridium
Oxide Microelectrodes," Anal. Chem. 2000, 72, 4921-4927, and Y. J.
Kim et al., "Configuration for Micro pH Sensor," Electronics
Letters, Vol. 39, No. 21 (Oct. 16, 2003).
Chip Control and Readout Circuitry
[0058] A wide variety of on-chip architectures and circuit designs
may be used for acquiring and processing output signals generated
by sensors in arrays of the invention. Several approaches are
disclosed in Rothberg et al., U.S. patent publications 2009/0026082
and 2009/0127589, which may be used with arrays of the present
invention. For example, FIG. 6 illustrates a block diagram of the
sensor array 100 coupled to an array controller 250, according to
one inventive embodiment of the present disclosure. In various
exemplary implementations, the array controller 250 may be
fabricated as a "stand alone" controller, or as one or more
computer compatible "cards" forming part of a computer 260 In one
aspect, the functions of the array controller 250 may be controlled
by the computer 260 through an interface block 252 (e.g., serial
interface, via USB port or PCI bus, Ethernet connection, etc.). In
one embodiment, all or a portion of the array controller 250 is
fabricated as one or more printed circuit boards, and the array 100
is configured to plug into one of the printed circuit boards,
similar to a conventional IC chip (e.g., the array 100 is
configured as an ASIC that plugs into a chip socket, such as a
zero-insertion-force or "ZIF" socket, of a printed circuit board).
In one aspect of such an embodiment, an array 100 configured as an
ASIC may include one or more pins/terminal connections dedicated to
providing an identification code that may be accessed/read by the
array controller 250 and/or passed on to the computer 260. Such an
identification code may represent various attributes of the array
100 (e.g., size, number of pixels, number of output signals,
various operating parameters such as supply and/or bias voltages,
etc.) and may be processed to determine corresponding operating
modes, parameters and or signals provided by the array controller
250 to ensure appropriate operation with any of a number of
different types of arrays 100. In one exemplary implementation, an
array 100 configured as an ASIC may be provided with three pins
dedicated to an identification code, and during the manufacturing
process the ASIC may be encoded to provide one of three possible
voltage states at each of these three pins (i.e., a tri-state pin
coding scheme) to be read by the array controller 250, thereby
providing for 27 unique array identification codes. In another
aspect of this embodiment, all or portions of the array controller
250 may be implemented as a field programmable gate array (FPGA)
configured to perform various array controller functions described
in further detail below.
[0059] Generally, the array controller 250 provides various supply
voltages and bias voltages to the array 100, as well as various
signals relating to row and column selection, sampling of pixel
outputs and data acquisition. In particular, the array controller
250 reads one or more analog output signals (e.g., Vout1 and Vout2)
including multiplexed respective pixel voltage signals from the
array 100 and then digitizes these respective pixel signals to
provide measurement data to the computer 260, which in turn may
store and/or process the data. In some implementations, the array
controller 250 also may be configured to perform or facilitate
various array calibration and diagnostic functions. Array
controller 250 generally provides to the array 100 the analog
supply voltage and ground (VDDA, VSSA), the digital supply voltage
and ground (VDDD, VSSD), and the buffer output supply voltage and
ground (VDDO, VSSO). In one exemplary implementation, each of the
supply voltages VDDA, VDDD and VDDO is approximately 3.3 Volts. In
another implementation, the supply voltages VDDA, VDDD and VDDO may
be as low as approximately 1.8 Volts. As discussed above, in one
aspect each of these power supply voltages is provided to the array
100 via separate conducting paths to facilitate noise isolation. In
another aspect, these supply voltages may originate from respective
power supplies/regulators, or one or more of these supply voltages
may originate from a common source in a power supply 258 of the
array controller 250. The power supply 258 also may provide the
various bias voltages required for array operation (e.g., VB1, VB2,
VB3, VB4, VBO0, V.sub.BODY) and the reference voltage VREF used for
array diagnostics and calibration.
[0060] In another aspect, the power supply 258 includes one or more
digital-to-analog converters (DACs) that may be controlled by the
computer 260 to allow any or all of the bias voltages, reference
voltage, and supply voltages to be changed under software control
(i.e., programmable bias settings). For example, a power supply 258
responsive to computer control (e.g., via software execution) may
facilitate adjustment of one or more of the supply voltages (e.g.,
switching between 3.3 Volts and 1.8 Volts depending on chip type as
represented by an identification code), and or adjustment of one or
more of the bias voltages VB1 and VB2 for pixel drain current, VB3
for column bus drive, VB4 for column amplifier bandwidth, and VBO0
for column output buffer current drive. In some aspects, one or
more bias voltages may be adjusted to optimize settling times of
signals from enabled pixels. Additionally, the common body voltage
V.sub.BODY for all ISFETs of the array may be grounded during an
optional post-fabrication UV irradiation treatment to reduce
trapped charge, and then coupled to a higher voltage (e.g., VDDA)
during diagnostic analysis, calibration, and normal operation of
the array for measurement/data acquisition. Likewise, the reference
voltage VREF may be varied to facilitate a variety of diagnostic
and calibration functions.
[0061] As shown in FIG. 6, the reference electrode 76 which is
typically employed in connection with an analyte solution to be
measured by the array 100 may be coupled to the power supply 258 to
provide a reference potential for the pixel output voltages. For
example, in one implementation the reference electrode 76 may be
coupled to a supply ground (e.g., the analog ground VSSA) to
provide a reference for the pixel output voltages. In other
exemplary implementations, the reference electrode voltage may be
set by placing a solution/sample of interest having a known pH
level in proximity to the sensor array 100 and adjusting the
reference electrode voltage until the array output signals Vout1
and Vout2 provide pixel voltages at a desired reference level, from
which subsequent changes in pixel voltages reflect local changes in
pH with respect to the known reference pH level. In general, it
should be appreciated that a voltage associated with the reference
electrode 76 need not necessarily be identical to the reference
voltage VREF discussed above (which may be employed for a variety
of array diagnostic and calibration functions), although in some
implementations the reference voltage VREF provided by the power
supply 258 may be used to set the voltage of the reference
electrode 76.
[0062] Regarding data acquisition from the array 100, in one
embodiment the array controller 250 of FIG. 6 may include one or
more preamplifiers 253 to further buffer one or more output signals
(e.g., Vout1 and Vout2) from the sensor array and provide
selectable gain. In one aspect, the array controller 250 may
include one preamplifier for each output signal (e.g., two
preamplifiers for two analog output signals). In other aspects, the
preamplifiers may be configured to accept input voltages from 0.0
to 1.8 Volts or 0.0 to 3.3 Volts, may have programmable/computer
selectable gains (e.g., 1, 2, 5, 10 and 20) and low noise outputs
(e.g., <10 nV/sqrtHz), and may provide low pass filtering (e.g.,
bandwidths of 5 MHz and 25 MHz). With respect to noise reduction
and increasing signal-to-noise ratio, in one implementation in
which the array 100 is configured as an ASIC placed in a chip
socket of a printed circuit board containing all or a portion of
the array controller 250, filtering capacitors may be employed in
proximity to the chip socket (e.g., the underside of a ZIF socket)
to facilitate noise reduction. In yet another aspect, the
preamplifiers may have a programmable/computer selectable offset
for input and/or output voltage signals to set a nominal level for
either to a desired range.
[0063] The array controller 250 of FIG. 6 also comprises one or
more analog-to-digital converters 254 (ADCs) to convert the sensor
array output signals Vout1 and Vout2 to digital outputs (e.g.,
10-bit or 12-bit) so as to provide data to the computer 260. In one
aspect, one ADC may be employed for each analog output of the
sensor array, and each ADC may be coupled to the output of a
corresponding preamplifier (if preamplifiers are employed in a
given implementation). In another aspect, the ADC(s) may have a
computer-selectable input range (e.g., 50 mV, 200 mV, 500 mV, 1V)
to facilitate compatibility with different ranges of array output
signals and/or preamplifier parameters. In yet other aspects, the
bandwidth of the ADC(s) may be greater than 60 MHz, and the data
acquisition/conversion rate greater than 25 MHz (e.g., as high as
100 MHz or greater).
[0064] In the embodiment of FIG. 6, ADC acquisition timing and
array row and column selection may be controlled by a timing
generator 256. In particular, the timing generator provides the
digital vertical data and clock signals (DV, CV) to control row
selection, the digital horizontal data and clock signals (DH, CH)
to control column selection, and the column sample and hold signal
COL SH to sample respective pixel voltages for an enabled row. The
timing generator 256 also provides a sampling clock signal CS to
the ADC(s) 254 so as to appropriately sample and digitize
consecutive pixel values in the data stream of a given array analog
output signal (e.g., Vout1 and Vout2), as discussed further below
in connection with FIG. 7. In some implementations, the timing
generator 256 may be implemented by a microprocessor executing code
and configured as a multi-channel digital pattern generator to
provide appropriately timed control signals. In one exemplary
implementation, the timing generator 256 may be implemented as a
field-programmable gate array (FPGA).
[0065] FIG. 7 illustrates an exemplary timing diagram for various
array control signals, as provided by the timing generator 256, to
acquire pixel data from the sensor array 100. For purposes of the
following discussion, a "frame" is defined as a data set that
includes a value (e.g., pixel output signal or voltage V.sub.S) for
each pixel in the array, and a "frame rate" is defined as the rate
at which successive frames may be acquired from the array. Thus,
the frame rate corresponds essentially to a "pixel sampling rate"
for each pixel of the array, as data from any given pixel is
obtained at the frame rate.
[0066] In the example of FIG. 7, an exemplary frame rate of 20
frames/sec is chosen to illustrate operation of the array (i.e.,
row and column selection and signal acquisition); however, it
should be appreciated that arrays and array controllers according
to the present disclosure are not limited in this respect, as
different frame rates, including lower frame rates (e.g., 1 to 10
frames/second) or higher frame rates (e.g., 25, 30, 40, 50, 60, 70
to 100 frames/sec., etc.), with arrays having the same or higher
numbers of pixels, are possible. In some exemplary applications, a
data set may be acquired that includes many frames over several
seconds to conduct an experiment on a given analyte or analytes.
Several such experiments may be performed in succession, in some
cases with pauses in between to allow for data transfer/processing
and/or washing of the sensor array ASIC and reagent preparation for
a subsequent experiment.
[0067] For example, with respect to the method for detecting
nucleotide incorporation, appropriate frame rates may be chosen to
sufficiently sample the ISFET's output signal. In some exemplary
implementations, a hydrogen ion signal may have a full-width at
half-maximum (FWHM) on the order of approximately 1 second to
approximately 2.5 seconds, depending on the number of nucleotide
incorporation events. Given these exemplary values, a frame rate
(or pixel sampling rate) of 20 Hz is sufficient to reliably resolve
the signals in a given pixel's output signal. Again, the frame
rates given in this example are provided primarily for purposes of
illustration, and different frame rates may be involved in other
implementations.
[0068] In one implementation, the array controller 250 controls the
array 100 to enable rows successively, one at a time. For example,
a first row of pixels is enabled via the row select signal
RowSel.sub.1. The enabled pixels are allowed to settle for some
time period, after which the COL SH signal is asserted briefly to
close the sample/hold switch in each column and store on the
column's sample/hold capacitor C.sub.sh the voltage value output by
the first pixel in the column. This voltage is then available as
the column output voltage V.sub.COLj applied to one of the two (odd
and even column) array output drivers 198.sub.1 and 198.sub.2
(e.g., see FIG. 16). The COL SH signal is then de-asserted, thereby
opening the sample/hold switches in each column and decoupling the
column output buffer 111j from the column amplifiers 107A and 107B.
Shortly thereafter, the second row of pixels is enabled via the row
select signal RowSel.sub.2. During the time period in which the
second row of pixels is allowed to settle, the column select
signals are generated two at a time (one odd and one even; odd
column select signals are applied in succession to the odd output
driver, even column select signals are applied in succession to the
even output driver) to read the column output voltages associated
with the first row. Thus, while a given row in the array is enabled
and settling, the previous row is being read out, two columns at a
time. By staggering row selection and sampling/readout (e.g., via
different vertical and horizontal clock signals and column
sample/hold), and by reading multiple columns at a time for a given
row, a frame of data may be acquired from the array in a
significantly streamlined manner.
[0069] FIG. 7 illustrates the timing details of the foregoing
process for an exemplary frame rate of 20 frames/sec. In a
512.times.512 array, each row must be read out in approximately 98
microseconds, as indicated by the vertical delineations in FIG. 7.
Accordingly, the vertical clock signal CV has a period of 98
microseconds (i.e., a clock frequency of over 10 kHz), with a new
row being enabled on a trailing edge (negative transition) of the
CV signal. The left side of FIG. 7 reflects the beginning of a new
frame cycle, at which point the vertical data signal DV is asserted
before a first trailing edge of the CV signal and de-asserted
before the next trailing edge of the CV signal. Also, immediately
before each trailing edge of the CV signal (i.e., new row enabled),
the COL SH signal is asserted for 2 microseconds, leaving
approximately 50 nanoseconds before the trailing edge of the CV
signal.
[0070] In FIG. 7, the first occurrence of the COL SH signal is
actually sampling the pixel values of row 512 of the 512.times.512
array. Thus, upon the first trailing edge of the CV signal, the
first row is enabled and allowed to settle (for approximately 96
microseconds) until the second occurrence of the COL SH signal.
During this settling time for the first row, the pixel values of
row 512 are read out via the column select signals. Because two
column select signals are generated simultaneously to read 512
columns, the horizontal clock signal CH must generate 256 cycles
within this period, each trailing edge of the CH signal generating
one odd and one even column select signal. As shown in FIG. 7, the
first trailing edge of the CH signal in a given row is timed to
occur two microseconds after the selection of the row (after
deactivation of the COL SH signal) to allow for settling of the
voltage values stored on the sample/hold capacitors C.sub.sh and
provided by the column output buffers. It should be appreciated
however that, in other implementations, the time period between the
first trailing edge of the CH signal and a trailing edge (i.e.,
deactivation) of the COL SH signal may be significantly less than
two microseconds, and in some cases as small as just over 50
nanoseconds. Also for each row, the horizontal data signal DH is
asserted before the first trailing edge of the CH signal and
de-asserted before the next trailing edge of the CH signal. The
last two columns are selected before the occurrence of the COL SH
signal which, as discussed above, occurs approximately two
microseconds before the next row is enabled. Thus, in the above
example, columns are read, two at a time, within a time period of
approximately 94 microseconds (i.e., 98 microseconds per row, minus
two microseconds at the beginning and end of each row). This
results in a data rate for each of the array output signals Vout1
and Vout2 of approximately 2.7 MHz.
[0071] Noise coupled into the sample fluid by the sensors in every
column of the array may be present in the output signals of a
sensor. When a row is selected in the array, the drain terminal
voltage shared between all of the ISFETs in a column moves up or
down (as a necessary requirement of the source-and-drain follower).
This changes the gate-to-drain capacitances of all of the
unselected ISFETs in the column. In turn, this change in
capacitance couples from the gate of every unselected ISFET into
the fluid, ultimately manifesting itself as noise in the fluid
(i.e., an incorrect charge, one not due to the chemical reaction
being monitored). That is, any change in the shared drain terminal
voltage can be regarded as injecting noise into the fluid by each
and every unselected ISFET in the column. Hence, if the shared
drain terminal voltage of the unselected ISFETs can be kept
constant when selecting a row in the array, this mechanism of
coupling noise into the fluid can be reduced or even effectively
eliminated. When a row is selected in the array, the source
terminal voltage of all of the unselected ISFETs in the column also
changes. In turn, that changes the gate-to-source capacitance of
all of these ISFETs in the column. This change in capacitance
couples from the gate of every unselected ISFET into the fluid,
again ultimately manifesting itself as noise in the fluid. That is,
any change in the source terminal voltage of an unselected ISFET in
the column can be regarded as an injection of noise into the fluid.
Hence, if the source terminal voltage of the unselected ISFETs can
be kept when selecting a row in the array, this mechanism of
coupling noise into the fluid via can be reduced or even
effectively eliminated.
[0072] A column buffer may be used with some passive pixel designs
to alleviate the ISFET drain problem but not the ISFET source
problem. Thus, a column buffer most likely is preferable to the
above-illustrated source-and-drain follower. With the illustrated
three-transistor passive pixels employing a source-and-drain
follower arrangement, there are essentially two sense nodes, the
ISFET source and drain terminals, By connecting the pixel to a
column buffer and grounding the drain terminal of the ISFET, there
will be only one sense node: the ISFET source terminal. So the
drain problem is eliminated.
[0073] The above-described readout circuit, which comprises both
sample-and-hold and multiplexer blocks, also has a gain that is
less than the ideal value of unity. Furthermore, the
sample-and-hold block contributes a significant percentage of the
overall chip noise, perhaps more than 25%. From switched-capacitor
theory, the sample-and-hold "kT/C" noise is inversely proportional
to capacitance. Hence, by choosing a larger capacitor, the
sample-and-hold noise can be reduced. Another approach to reducing
noise is to employ Correlated Double Sampling (CDS), where a second
sample-and-hold and difference circuit is used to cancel out
correlated noise. This approach is discussed at greater length,
below.
[0074] Correlated Double Sampling (CDS) is a known technique for
measuring electrical values such as voltages or currents that
allows for removal of an undesired offset. The output of the sensor
is measured twice: once in a known condition and once in an unknown
condition. The value measured from the known condition is then
subtracted from the unknown condition to generate a value with a
known relation to the physical quantity being measured. The
challenge here is how to be efficient in implementing CDS and how
to address both correlated noise and the minimization of noise
injection into the analyte fluid.
[0075] A starting point is the sensor pixel and its readout
configuration as expressed in earlier parts of this application.
Referring to FIG. 8A, the basic passive sensor pixel 77A1 is a
three-transistor arrangement of an ISFET 77A2 and a pair of row
select transistors, 77A3 and 77A4 connected to the ISFET source.
Transistor 77A3 is connected in turn to a current source or sink
77A5. A readout is obtained via transistor 77A4 which is connected
to the input of sense amplifier 77A6. A diode-connected transistor
77A7 in series with another amplifier, 77A8, connects in a feedback
loop from the output of the sense amplifier to the drain of the
ISFET. The sense amplifier output is captured by a sample-and-hold
circuit 77A9, which feeds an output amplifier 77A10.
[0076] As discussed above, the voltage changes on the ISFET source
and drain inject noise into the analyte, causing errors in the
sensed values. Two constructive modifications can reduce the noise
level appreciably, as shown in FIG. 8B.
[0077] The first change is to alter the signals on the ISFET. The
feedback loop to the drain of the ISFET is eliminated and the drain
is connected to a stable voltage, such as ground. A column buffer
77B is connected to the emitter of transistor.
[0078] The second change is to include a circuit to perform CDS on
the output of the column buffer. As mentioned above, CDS requires a
first, reference value. This is obtained by connecting the input of
column buffer 77B1 to a reference voltage via switch 7782, during a
first, or reference phase of a clock, indicated as the "SH" phase.
A combined CDS and sample-and-hold circuit then double samples the
output of the column buffer, obtaining a reference sample and a
sensed value, performs a subtraction, and supplies a resulting
noise-reduced output value, since the same correlated noise appears
in the reference sample and in the sensor output.
[0079] The operation of the CDS and sample-and-hold circuit is
straightforward. The circuit operates on a two-phase clock, the
first phase being the SH phase and the second phase being the SHb
phase. Typically, the phases will be symmetrical and thus inverted
values of each other. The reference sample is obtained in the SH
phase and places a charge (and thus a voltage) on capacitor Cin,
which is subtracted from the output of the column buffer when the
clock phase changes.
[0080] An alternative embodiment, still with a passive sensor
pixel, is shown in FIG. 8C. The sensor pixel in this embodiment is
a two-transistor circuit comprising ISFET whose drain is connected
to a fixed supply voltage, VSSA. There is no transistor comparable
to 77A4, and the pixel output is taken from the emitter of
transistor 77A3, instead. The CDS and sample-and-hold circuit has
been simplified slightly, by the elimination of a feedback loop,
but it serves the same function, in conjunction with the charge
(voltage) stored on capacitor Cbl, of subtracting a reference value
on capacitor Cin from the signal supplied by the sensor pixel.
Microwell Arrays
[0081] As discussed elsewhere, for many uses, such as in DNA
sequencing, it is desirable to provide over the array of
semiconductor sensors a corresponding array of microwells, each
microwell being small enough preferably to receive only one
DNA-loaded bead, in connection with which an underlying pixel in
the array will provide a corresponding output signal.
[0082] The use of such a microwell array involves three stages of
fabrication and preparation, each of which is discussed separately:
(1) creating the array of microwells to result in a chip having a
coat comprising a microwell array layer; (2) mounting of the coated
chip to a fluidic interface; and in the case of DNA sequencing, (3)
loading DNA-loaded bead or beads into the wells. It will be
understood, of course, that in other applications, beads may be
unnecessary or beads having different characteristics may be
employed.
[0083] The systems described herein can include an array of
microfluidic reaction chambers integrated with a semiconductor
comprising an array of chemFETs. In some embodiments, the invention
encompasses such an array. The reaction chambers may, for example,
be formed in a glass, dielectric, photodefineable or etchable
material. The glass material may be silicon dioxide.
[0084] Various aspects or embodiments of the invention involve an
apparatus comprising an array of chemFET sensors overlayed with an
array of reaction chambers wherein the bottom of a reaction chamber
is in contact with (or capacitively coupled to) a chemFET sensor.
In some embodiments, each reaction chamber bottom is in contact
with a chemFET sensor, and preferably with a separate chemFET
sensor. In some embodiments, less than all reaction chamber bottoms
are in contact with a chemFET sensor. In some embodiments, each
sensor in the array is in contact with a reaction chamber. In other
embodiments, less than all sensors are in contact with a reaction
chamber. The sensor (and/or reaction chamber) array may be
comprised of 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 30, 40, 50, 60,
60, 80, 90, 100, 200, 300, 400, 500, 1000, 10.sup.4, 10.sup.5,
10.sup.6, 10.sup.7, 10.sup.8, or more chemFET sensors (and/or
reaction chambers). As used herein, it is intended that an array
that comprises, as an example, 256 sensors or reaction chambers
will contain 256 or more (i.e., at least 256) sensors or reaction
chambers. It is further intended that aspects and embodiments
described herein that "comprise" elements and/or steps also fully
support and embrace aspects and embodiments that "consist of" or
"consist essentially of" such elements and/or steps.
[0085] Various aspects and embodiments of the invention involve
sensors (and/or reaction chambers) within an array that are spaced
apart from each other at a center-to-center distance or spacing (or
"pitch", as the terms are used interchangeably herein) that is in
the range of 1-50 microns, 1-40 microns, 1-30 microns, 1-20
microns, 1-10 microns, or 5-10 microns, including equal to or less
than about 9 microns, or equal to or less than about 5.1 microns,
or 1-5 microns including equal to or less than about 2.8 microns.
The center-to-center distance between adjacent reaction chambers in
a reaction chamber array may be about 1-9 microns, or about 2-9
microns, or about 1 microns, about 2 microns, about 3 microns,
about 4 microns, about 5 microns, about 6 microns, about 7 microns,
about 8 microns, or about 9 microns.
[0086] In some embodiments, the reaction chamber has a volume of
equal to or less than about 1 picoliter (pL), including less than
0.5 pL, less than 0.1 pL, less than 0.05 pL, less than 0.01 pL,
less than 0.005 pL.
[0087] The reaction chambers may have a square cross section, for
example, at their base or bottom. Examples include an 8 .mu.m by 8
.mu.m cross section, a 4 .mu.m by 4 .mu.m cross section, or a 1.5
.mu.m by 1.5 .mu.m cross section. Alternatively, they may have a
rectangular cross section, for example, at their base or bottom.
Examples include an 8 .mu.m by 12 .mu.m cross section, a 4 .mu.m by
6 .mu.m cross section, or a 1.5 .mu.m by 2.25 .mu.m cross
section.
[0088] In another exemplary implementation, the invention
encompasses a system comprising at least one two-dimensional array
of reaction chambers, wherein each reaction chamber is coupled to a
chemically-sensitive field effect transistor ("chemFET") and each
reaction chamber is no greater than 10 .mu.m.sup.3 (i.e., 1 pL) in
volume. Preferably, each reaction chamber is no greater than 0.34
pL, and more preferably no greater than 0.096 pL or even 0.012 pL
in volume. A reaction chamber can optionally be 2.sup.2, 3.sup.2,
4.sup.2, 5.sup.2, 6.sup.2, 7.sup.2, 8.sup.2, 9.sup.2, or 10.sup.2
square microns in cross-sectional area at the top. Preferably, the
array has at least 10.sup.2, 10.sup.3, 10.sup.4, 10.sup.5,
10.sup.6, 10.sup.7, 10.sup.8, 10.sup.9, or more reaction chambers.
The reaction chambers may be capacitively coupled to the chemFETs,
and preferably are capacitively coupled to the chemFETs. Such
systems may be used for high-throughput sequencing of nucleic
acids.
[0089] In some embodiments, the reaction chamber array (or
equivalently, microwell array) comprises 10.sup.2, 10.sup.3,
10.sup.4, 10.sup.5, 10.sup.6 or 10.sup.7 microwells or reaction
chambers. In some embodiments, individual reaction chambers in the
reaction chamber array are in contact with or capacitively coupled
to at least one chemFET. In one embodiment, a reaction chamber of
an array is in contact with or capacitively coupled to one chemFET
or one ISFET. In some embodiments, the chemFET array may optionally
comprise 10.sup.2, 10.sup.3, 10.sup.4, 10.sup.5, 10.sup.6 or
10.sup.7 chemFETs.
[0090] In these and in other aspects and embodiments, the chemFET
or ISFET arrays may comprise 256 or more chemFETs or ISFETs. The
chemFETs or ISFETs of such arrays may have a center-to-center
spacing (between adjacent chemFETs or ISFETs) of 1-10 microns. In
some embodiments, the center-to-center spacing is about 9 microns,
about 8 microns, about 7 microns, about 6 microns, about 5 microns,
about 4 microns, about 3 microns, about 2 microns or about 1
micron. In particular embodiments, the center-to-center spacing is
about 5.1 microns or about 2.8 microns.
[0091] In some embodiments, the bead is in a reaction chamber, and
optionally the only bead in the reaction chamber. In some
embodiments, the reaction chamber is in contact with or
capacitively coupled to an ISFET. In some embodiments, the ISFET is
in an ISFET array In some embodiments, the bead has a diameter of
less than 6 microns, less than 3 microns, or about 1 micron. The
bead may have a diameter of about 1 micron up to about 7 microns,
or about 1 micron up to about 3 microns.
[0092] In some embodiments, the reaction chambers have a
center-to-center distance of about 1 micron to about 10 microns. In
some embodiments, the reaction chamber array comprises 10.sup.2,
10.sup.3, 10.sup.4, 10.sup.5, 10.sup.6 or 10.sup.7 reaction
chambers.
[0093] In accordance with the invention, a dielectric layer on a
gate of an ISFET is part of the ISFET. It is recognized that the
charge in the reaction chamber builds up on one side of the
dielectric and forms one plate of a capacitor and which has as its
second plate the floating gate metal layer; thus, a reaction
chamber is referred to as being capacitively coupled to the
ISFET.
[0094] In some embodiments, the ISFET is in an ISFET array. The
ISFET array may comprise 10.sup.2, 10.sup.3, 10.sup.4, 10.sup.5,
10.sup.6 or 10.sup.7 ISFETs.
[0095] In some embodiments, the template nucleic acid is in a
reaction chamber in contact with or capacitively coupled to the
ISFET. In some embodiments, the reaction chamber is in a reaction
chamber array. In some embodiments, the reaction chamber array
comprises 10.sup.2, 10.sup.3, 10.sup.4, 10.sup.5, 10.sup.6 or
10.sup.7 reaction chambers.
[0096] The microwells may vary in size between arrays. The size of
these microwells may be described in terms of a width (or diameter)
to height ratio. In some embodiments, this ratio is 1:1 to 1:1.5.
The bead to well size (e.g., the bead diameter to well width,
diameter, or height) is preferably in the range of 0.6-0.8.
[0097] The microwell size may be described in terms of cross
section. The cross section may refer to a "slice" parallel to the
depth (or height) of the well, or it may be a slice perpendicular
to the depth (or height) of the well. The microwells may be square
in cross-section, but they are not so limited. The dimensions at
the bottom of a microwell (i.e., in a cross section that is
perpendicular to the depth of the well) may be 1.5 .mu.m by 1.5
.mu.m, or it may be 1.5 .mu.m by 2 .mu.m. Suitable diameters
include but are not limited to at or about 100 .mu.m, 95 .mu.m, 90
.mu.m, 85 .mu.m, 80 .mu.m, 75 .mu.m, 70 .mu.m, 65 .mu.m, 60 .mu.m,
55 .mu.m, 50 .mu.m, 45 .mu.m, 40 .mu.m, 35 .mu.m, 30 .mu.m, 25
.mu.m, 20 .mu.m, 15 .mu.m, 10 .mu.m, 9 .mu.m, 8 .mu.m, 7 .mu.m, 6
.mu.m, 5 .mu.m, 4 .mu.m, 3 .mu.m, 2 .mu.m, 1 .mu.m or less. In some
particular embodiments, the diameters may be at or about 44 .mu.m,
32 .mu.m, 8 .mu.m, 4 .mu.m, or 1.5 .mu.m. Suitable heights include
but are not limited to at or about 100 .mu.m, 95 .mu.m, 90 .mu.m,
85 .mu.m, 80 .mu.m, 75 .mu.m, 70 .mu.m, 65 .mu.m, 60 .mu.m, 55
.mu.m, 50 .mu.m, 45 .mu.m, 40 .mu.m, 35 .mu.m, 30 .mu.m, 25 .mu.m,
20 .mu.m, 15 .mu.m, 10 .mu.m, 9 .mu.m, 8 .mu.m, 7 .mu.m, 6 .mu.m, 5
.mu.m, 4 .mu.m, 3 .mu.m, 2 .mu.m, 1 .mu.m or less. In some
particular embodiments, the heights may be at or about 55 .mu.m, 48
.mu.m, 32 .mu.m, 12 .mu.m, 8 .mu.m, 6 .mu.m, 4 .mu.m, 2.25 .mu.m,
1.5 .mu.m, or less. Various embodiments of the invention
contemplate the combination of any of these diameters with any of
these heights. In still other embodiments, the reaction well
dimensions may be (diameter in .mu.m by height in .mu.m) 44 by 55,
32 by 32, 32 by 48, 8 by 8, 8 by 12, 4 by 4, 4 by 6, 1.5 by 1.5, or
1.5 by 2.25.
[0098] The reaction well volume may range (between arrays, and
preferably not within a single array) based on the well dimensions.
This volume may be at or about 100 picoliter (pL), 90, 80, 70, 60,
50, 40, 30, 20, 10, or fewer pL. In important embodiments, the well
volume is less than 1 pL, including equal to or less than 0.5 pL,
equal to or less than 0.1 pL, equal to or less than 0.05 pL, equal
to or less than 0.01 pL, equal to or less than 0.005 pL, or equal
to or less than 0.001 pL. The volume may be 0.001 to 0.9 pL, 0.001
to 0.5 pL, 0.001 to 0.1 pL, 0.001 to 0.05 pL, or 0.005 to 0.05 pL.
In particular embodiments, the well volume is 75 pL, 34 pL, 23 pL,
0.54 pL, 0.36 pL, 0.07 pL, 0.045 pL, 0.0024 pL, or 0.004 pL. In
some embodiments, each reaction chamber is no greater than about
0.39 pL in volume and about 49 .mu.m.sup.2 surface aperture, and
more preferably has an aperture no greater than about 16
.mu.m.sup.2 and volume no greater than about 0.064 pL.
[0099] Thus, it is to be understood that various aspects and
embodiments of the invention relate generally to large scale FET
arrays for measuring one or more analytes or for measuring charge
bound to the chemFET surface. It will be appreciated that chemFETs
and more particularly ISFETs may be used to detect analytes and/or
charge. An ISFET, as discussed above, is a particular type of
chemFET that is configured for ion detection such as hydrogen ion
(or proton) detection. Other types of chemFETs contemplated by the
present disclosure include enzyme FETs (EnFETs) which employ
enzymes to detect analytes. It should be appreciated, however, that
the present disclosure is not limited to ISFETs and EnFETs, but
more generally relates to any FET that is configured for some type
of chemical sensitivity. As used herein, chemical sensitivity
broadly encompasses sensitivity to any molecule of interest,
including without limitation organic, inorganic, naturally
occurring, non-naturally occurring, chemical and biological
compounds, such as ions, small molecules, polymers such as nucleic
acids, proteins, peptides, polysaccharides, and the like.
[0100] In some embodiments, the invention encompasses a sequencing
apparatus comprising a dielectric layer overlying a chemFET, the
dielectric layer having a recess laterally centered atop the
chemFET. Preferably, the dielectric layer is formed of silicon
dioxide.
[0101] After the semiconductor structures, as shown, are formed,
the microwell structure is applied to the die. That is, the
microwell structure can be formed right on the die or it may be
formed separately and then mounted onto the die, either approach
being acceptable. To form the microwell structure on the die,
various processes may be used. For example, the entire die may be
spin-coated with, for example, a negative photoresist such as
Microchem's SU-8 2015 or a positive resist/polyimide such as HD
Microsystems HD8820, to the desired height of the microwells. The
desired height of the wells (e.g., about 4-12 .mu.m in the example
of one pixel per well, though not so limited as a general matter)
in the photoresist layer(s) can be achieved by spinning the
appropriate resist at predetermined rates (which can be found by
reference to the literature and manufacturer specifications, or
empirically), in one or more layers. (Well height typically may be
selected in correspondence with the lateral dimension of the sensor
pixel, preferably for a nominal 1:1-1.5:1 aspect ratio,
height:width or diameter. Based on signal-to-noise considerations,
there is a relationship between dimensions and the required data
sampling rates to achieve a desired level of performance. Thus
there are a number of factors that will go into selecting optimum
parameters for a given application.) Alternatively, multiple layers
of different photoresists may be applied or another form of
dielectric material may be deposited. Various types of chemical
vapor deposition may also be used to build up a layer of materials
suitable for microwell formation therein.
[0102] Once the photoresist layer (the singular form "layer" is
used to encompass multiple layers in the aggregate, as well) is in
place, the individual wells (typically mapped to have either one or
four ISFET sensors per well) may be generated by placing a mask
(e.g., of chromium) over the resist-coated die and exposing the
resist to cross-linking (typically UV) radiation. All resist
exposed to the radiation (i.e., where the mask does not block the
radiation) becomes cross-linked and as a result will form a
permanent plastic layer bonded to the surface of the chip (die).
Unreacted resist (i.e., resist in areas which are not exposed, due
to the mask blocking the light from reaching the resist and
preventing cross-linking) is removed by washing the chip in a
suitable solvent (i.e., developer) such as
propyleneglycolmethylethylacetate (PGMEA) or other appropriate
solvent. The resultant structure defines the walls of the microwell
array.
[0103] For example, contact lithography of various resolutions and
with various etchants and developers may be employed. Both organic
and inorganic materials may be used for the layer(s) in which the
microwells are formed. The layer(s) may be etched on a chip having
a dielectric layer over the pixel structures in the sensor array,
such as a passivation layer, or the layer(s) may be formed
separately and then applied over the sensor array. The specific
choice or processes will depend on factors such as array size, well
size, the fabrication facility that is available, acceptable costs,
and the like.
[0104] Among the various organic materials which may be used in
some embodiments to form the microwell layer(s) are the
above-mentioned SU-8 type of negative-acting photoresist, a
conventional positive-acting photoresist and a positive-acting
photodefineable polyimide. Each has its virtues and its drawbacks,
well known to those familiar with the photolithographic art.
[0105] Naturally, in a production environment, modifications will
be appropriate.
[0106] Contact lithography has its limitations and it may not be
the production method of choice to produce the highest densities of
wells--i.e., it may impose a higher than desired minimum pitch
limit in the lateral directions. Other techniques, such as a deep
UV step-and-repeat process, are capable of providing higher
resolution lithography and can be used to produce small pitches and
possibly smaller well diameters. Of course, for different desired
specifications (e.g., numbers of sensors and wells per chip),
different techniques may prove optimal. And pragmatic factors, such
as the fabrication processes available to a manufacturer, may
motivate the use of a specific fabrication method. While novel
methods are discussed, various aspects of the invention are limited
to use of these novel methods.
[0107] Preferably the CMOS wafer with the ISFET array will be
planarized after the final metallization process. A chemical
mechanical dielectric planarization prior to the silicon nitride
passivation is suitable. This will allow subsequent lithographic
steps to be done on very flat surfaces which are free of back-end
CMOS topography.
[0108] By utilizing deep-UV step-and-repeat lithography systems, it
is possible to resolve small features with superior resolution,
registration, and repeatability. However, the high resolution and
large numerical aperture (NA) of these systems precludes their
having a large depth of focus. As such, it may be necessary, when
using such a fabrication system, to use thinner photodefinable
spin-on layers (i.e., resists on the order of 1-2 .mu.m rather than
the thicker layers used in contact lithography) to pattern transfer
and then etch microwell features to underlying layer or layers.
High resolution lithography can then be used to pattern the
microwell features and conventional SiO.sub.2 etch chemistries can
be used--one each for the bondpad areas and then the microwell
areas--having selective etch stops; the etch stops then can be on
aluminum bondpads and silicon nitride passivation (or the like),
respectively. Alternatively, other suitable substitute pattern
transfer and etch processes can be employed to render microwells of
inorganic materials.
[0109] Another approach is to form the microwell structure in an
organic material. For example, a dual-resist "soft-mask" process
may be employed, whereby a thin high-resolution deep-UV resist is
used on top of a thicker organic material (e.g., cured polyimide or
opposite-acting resist). The top resist layer is patterned. The
pattern can be transferred using an oxygen plasma reactive ion etch
process. This process sequence is sometimes referred to as the
"portable conformable mask" (PCM) technique. See B. J. Lin et al.,
"Practicing the Novolac deep-UV portable conformable masking
technique", Journal of Vacuum Science and Technology 19, No. 4,
1313-1319 (1981); and A. Cooper et al., "Optimization of a
photosensitive spin-on dielectric process for copper inductor coil
and interconnect protection in RF SoC devices."
[0110] Alternatively a "drill-focusing" technique may be employed,
whereby several sequential step-and-repeat exposures are done at
different focal depths to compensate for the limited depth of focus
(DOF) of high-resolution steppers when patterning thick resist
layers. This technique depends on the stepper NA and DOF as well as
the contrast properties of the resist material.
[0111] Thus, microwells can be fabricated by any high aspect ratio
photo-definable or etchable thin-film process, that can provide
requisite thickness (e.g., about 4-10 .mu.m). Among the materials
believed to be suitable are photosensitive polymers, deposited
silicon dioxide, non-photosensitive polymer which can be etched
using, for example, plasma etching processes, etc. In the silicon
dioxide family, TEOS and silane nitrous oxide (SILOX) appear
suitable. The final structures are similar but the various
materials present differing surface compositions that may cause the
target biology or chemistry to react differently.
[0112] When the microwell layer is formed, it may be necessary to
provide an etch stop layer so that the etching process does not
proceed further than desired. For example, there may be an
underlying layer to be preserved, such as a low-K dielectric. The
etch stop material should be selected according to the application.
SiC and SiN materials may be suitable, but that is not meant to
indicate that other materials may not be employed, instead. These
etch-stop materials can also serve to enhance the surface chemistry
which drives the ISFET sensor sensitivity, by choosing the
etch-stop material to have an appropriate point of zero charge
(PZC). Various metal oxides may be suitable addition to silicon
dioxide and silicon nitride.
[0113] The PZCs for various metal oxides may be found in various
texts, such as "Metal Oxides--Chemistry and Applications" by J.
Fierro. We have learned that Ta.sub.2O.sub.5 may be preferred as an
etch stop over Al.sub.2O.sub.3 because the PZC of Al.sub.2O.sub.3
is right at the pH being used (i.e., about 8.8) and, hence, right
at the point of zero charge. In addition Ta.sub.2O.sub.5 has a
higher sensitivity to pH (i.e., mV/pH), another important factor in
the sensor performance. Optimizing these parameters may require
judicious selection of passivation surface materials.
[0114] Using thin metal oxide materials for this purpose (i.e., as
an etch stop layer) is difficult due to the fact of their being so
thinly deposited (typically 200-500 A). A post-microwell
fabrication metal oxide deposition technique may allow placement of
appropriate PZC metal oxide films at the bottom of the high aspect
ratio microwells.
[0115] Electron-beam depositions of (a) reactively sputtered
tantalum oxide, (b) non-reactive stoichiometric tantalum oxide, (c)
tungsten oxide, or (d) Vanadium oxide may prove to have superior
"down-in-well" coverage due to the superior directionality of the
deposition process.
[0116] The array typically comprises at least 100 microfluidic
wells, each of which is coupled to one or more chemFET sensors.
Preferably, the wells are formed in at least one of a glass (e.g.,
SiO.sub.2), a polymeric material, a photodefinable material or a
reactively ion etchable thin film material. Preferably, the wells
have a width to height ratio less than about 1:1. Preferably the
sensor is a field effect transistor, and more preferably a chemFET.
The chemFET may optionally be coupled to a PPi receptor.
Preferably, each of the chemFETs occupies an area of the array that
is 10.sup.2 microns or less.
[0117] In some embodiments, the invention encompasses a sequencing
device comprising a semiconductor wafer device coupled to a
dielectric layer such as a glass (e.g., SiO.sub.2), polymeric,
photodefinable or reactive ion etchable material in which reaction
chambers are formed. Typically, the glass, dielectric, polymeric,
photodefinable or reactive ion etchable material is integrated with
the semiconductor wafer layer. In some instances, the glass,
polymeric, photodefinable or reactive ion etchable layer is
non-crystalline. In some instances, the glass may be SiO.sub.2. The
device can optionally further comprise a fluid delivery module of a
suitable material such as a polymeric material, preferably an
injection moldable material. More preferably, the polymeric layer
is polycarbonate.
[0118] In some embodiments, the invention encompasses a method for
manufacturing a sequencing device comprising: using
photolithography, generating wells in a glass, dielectric,
photodefinable or reactively ion etchable material on top of an
array of transistors.
[0119] Yet another alternative when a CMOS or similar fabrication
process is used for array fabrication is to form the microwells
directly using the CMOS materials. That is, the CMOS top
metallization layer forming the floating gates of the ISFET array
usually is coated with a passivation layer that is about 1.3 .mu.m
thick. Microwells 1.3 .mu.m deep can be formed by etching away the
passivation material. For example, microwells having a 1:1 aspect
ratio may be formed, 1.3 .mu.m deep and 1.3 .mu.m across at their
tops. Modeling indicates that as the well size is reduced, in fact,
the DNA concentration, and hence the SNR, increases. So, other
factors being equal, such small wells may prove desirable.
Flow Cells and Fluidics System
[0120] A complete system for using the sensor array will include
suitable fluid sources, valving and a controller for operating the
valving to low reagents and washes over the microarray or sensor
array, depending on the application. These elements are readily
assembled from off-the-shelf components, with and the controller
may readily be programmed to perform a desired experiment.
[0121] It should be understood that the readout at the chemFET may
be current or voltage (and change thereof) and that any particular
reference to either readout is intended for simplicity and not to
the exclusion of the other readout. Therefore any reference in the
following text to either current or voltage detection at the
chemFET should be understood to contemplate and apply equally to
the other readout as well. In important embodiments, the readout
reflects a rapid, transient change in concentration of an analyte.
The concentration of more than one analyte may be detected at
different times. In some instances, such measurements are to be
contrasted with methods that focus on steady state concentration
measurements.
[0122] The process of using the assembly of an array of sensors on
a chip combined with an array of microwells to sequence the DNA in
a sample is referred to as an "experiment." Executing an experiment
requires loading the wells with the DNA-bound beads and the flowing
of several different fluid solutions (i.e., reagents and washes)
across the wells. A fluid delivery system (e.g., valves, conduits,
pressure source(s), etc.) coupled with a fluidic interface is
needed which flows the various solutions across the wells in a
controlled even flow with acceptably small dead volumes and small
cross contamination between sequential solutions. Ideally, the
fluidic interface to the chip (sometimes referred to as a "flow
cell") would cause the fluid to reach all microwells at the same
time. To maximize array speed, it is necessary that the array
outputs be available at as close to the same time as possible. The
ideal clearly is not possible, but it is desirable to minimize the
differentials, or skews, of the arrival times of an introduced
fluid, at the various wells, in order to maximize the overall speed
of acquisition of all the signals from the array.
[0123] Flow cell designs of many configurations are possible; thus
the system and methods presented herein are not dependent on use of
a specific flow cell configuration. It is desirable, though, that a
suitable flow cell substantially conform to the following set of
objectives: [0124] have connections suitable for interconnecting
with a fluidics delivery system--e.g., via appropriately-sized
tubing; [0125] have appropriate head space above wells; [0126]
minimize dead volumes encountered by fluids; [0127] minimize small
spaces in contact with liquid but not quickly swept clean by flow
of a wash fluid through the flow cell (to minimize cross
contamination); [0128] be configured to achieve uniform transit
time of the flow over the array; [0129] generate or propagate
minimal bubbles in the flow over the wells; [0130] be adaptable to
placement of a removable reference electrode inside or as close to
the flow chamber as possible; [0131] facilitate easy loading of
beads; [0132] be manufacturable at acceptable cost; and [0133] be
easily assembled and attached to the chip package.
[0134] Satisfaction of these criteria so far as possible will
contribute to system performance positively. For example,
minimization of bubbles is important so that signals from the array
truly indicate the reaction in a well rather than being spurious
noise.
[0135] Each of several example designs will be discussed, meeting
these criteria in differing ways and degrees. In each instance, one
typically may choose to implement the design in one of two ways:
either by attaching the flow cell to a frame and gluing the frame
(or otherwise attaching it) to the chip or by integrating the frame
into the flow cell structure and attaching this unified assembly to
the chip. Further, designs may be categorized by the way the
reference electrode is integrated into the arrangement. Depending
on the design, the reference electrode may be integrated into the
flow cell (e.g., form part of the ceiling of the flow chamber) or
be in the flow path (typically to the outlet or downstream side of
the flow path, after the sensor array).
[0136] An example of a suitable experiment apparatus 3410
incorporating such a fluidic interface is shown in FIG. 9, the
manufacture and construction of which will be discussed in greater
detail below. The apparatus comprises a semiconductor chip 3412
(indicated generally, though hidden) on or in which the arrays of
wells and sensors are formed, and a fluidics assembly 3414 on top
of the chip and delivering the sample to the chip for reading. The
fluidics assembly includes a portion 3416 for introducing fluid
containing the sample, a portion 3418 for allowing the fluid to be
piped out, and a flow chamber portion 3420 for allowing the fluid
to flow from inlet to outlet and along the way interact with the
material in the wells. Those three portions are unified by an
interface comprising a glass slide 3422 (e.g., Erie Microarray Cat
#C22-5128-M20 from Erie Scientific Company, Portsmouth, N.H., cut
in thirds, each to be of size about 25 mm.times.25 mm).
[0137] Mounted on the top face of the glass slide are two fittings,
3424 and 3426, such as nanoport fittings Part #N-333 from Upchurch
Scientific of Oak Harbor, Wash. One port (e.g., 3424) serves as an
inlet delivering liquids from the pumping/valving system described
below but not shown here. The second port (e.g., 3426) is the
outlet which pipes the liquids to waste. Each port connects to a
conduit 3428, 3432 such as flexible tubing of appropriate inner
diameter. The nanoports are mounted such that the tubing can
penetrate corresponding holes in the glass slide. The tube
apertures should be flush with the bottom surface of the slide.
[0138] On the bottom of the glass slide, flow chamber 3420 may
comprise various structures for promoting a substantially laminar
flow across the microwell array. For example, a series of
microfluidic channels fanning out from the inlet pipe to the edge
of the flow chamber may be patterned by contact lithography using
positive photoresists such as SU-8 photoresist from MicroChem Corp.
of Newton, Mass. Other structures will be discussed below.
[0139] The chip 3412 will in turn be mounted to a carrier 3430, for
packaging and connection to connector pins 3432.
[0140] Achieving a uniform flow front and eliminating problematic
flow path areas is desirable for a number of reasons. One reason is
that very fast transition of fluid interfaces within the system's
flow cell is desired for many applications, particularly gene
sequencing. In other words, an incoming fluid must completely
displace the previous fluid in a short period of time. Uneven fluid
velocities and diffusion within the flow cell, as well as
problematic flow paths, can compete with this requirement. Simple
flow through a conduit of rectangular cross section can exhibit
considerable disparity of fluid velocity from regions near the
center of the flow volume to those adjacent the sidewalls, one
sidewall being the top surface of the microwell layer and the fluid
in the wells. Such disparity leads to spatially and temporally
large concentration gradients between the two traveling fluids.
Further, bubbles are likely to be trapped or created in stagnant
areas like sharp corners interior the flow cell. (The surface
energy (hydrophilic vs. hydrophobic) can significantly affect
bubble retention. Avoidance of surface contamination during
processing and use of a surface treatment to create a more
hydrophilic surface should be considered if the as-molded surface
is too hydrophobic.) Of course, the physical arrangement of the
flow chamber is probably the factor which most influences the
degree of uniformity achievable for the flow front.
[0141] In all cases, attention should be given to assuring a
thorough washing of the entire flow chamber, along with the
microwells, between reagent cycles. Flow disturbances may
exacerbate the challenge of fully cleaning out the flow
chamber.
[0142] Flow disturbances may also induce or multiply bubbles in the
fluid. A bubble may prevent the fluid from reaching a microwell, or
delay its introduction to the microwell, introducing error into the
microwell reading or making the output from that microwell useless
in the processing of outputs from the array. Thus, care should be
taken in selecting configurations and dimensions for the flow
disruptor elements to manage these potential adverse factors. For
example, a tradeoff may be made between the heights of the
disruptor elements and the velocity profile change that is
desired.
[0143] The flow cell, as stated elsewhere, may be fabricated of
many different materials. Injection molded polycarbonate appears to
be quite suitable. A conductive metal (e.g., gold) may be deposited
using an adhesion layer (e.g., chrome) to the underside of the flow
cell roof (the ceiling of the flow chamber). Appropriate
low-temperature thin-film deposition techniques preferably are
employed in the deposition of the metal reference electrode due to
the materials (e.g., polycarbonate) and large step coverage
topography at the bottom-side of the fluidic cell (i.e., the frame
surround of ISFET array). One possible approach would be to use
electron-beam evaporation in a planetary system.
[0144] Once assembly is complete--conductive epoxy (e.g., Epo-Tek
H20E or similar) may be dispensed on the seal ring with the flow
cell aligned, placed, pressed and cured--the ISFET flow cell is
ready for operation with the reference potential being applied to
the assigned pin of the package.
[0145] In some embodiments, the invention encompasses an apparatus
for detection of pH comprising a laminar fluid flow system.
Preferably, the apparatus is used for sequencing a plurality of
nucleic acid templates present in an array.
[0146] The apparatus typically includes a fluidics assembly
comprising a member comprising one or more apertures for
non-mechanically directing a fluid to flow to an array of at least
100K (100 thousand), 500K (500 thousand), or 1M (1 million)
microfluidic reaction chambers such that the fluid reaches all of
the microfluidic reaction chambers at the same time or
substantially the same time. Typically, the fluid flow is parallel
to the sensor surface. Typically, the assembly has a Reynolds
number of less than 1000, 500, 200, 100, 50, 20, or 10. Preferably,
the member further comprises a first aperture for directing fluid
towards the sensor array and a second aperture for directing fluid
away from the sensor array.
[0147] In some embodiments, the invention encompasses a method for
directing a fluid to a sensor array comprising: providing a
fluidics assembly comprising an aperture fluidly coupling a fluid
source to the sensor array; and non-mechanically directing a fluid
to the sensor array. By "non-mechanically" it is meant that the
fluid is moved under pressure from a gaseous pressure source, as
opposed to a mechanical pump.
[0148] In some embodiments, the invention encompasses an array of
wells, each of which is coupled to a lid having an inlet port and
an outlet port and a fluid delivery system for delivering and
removing fluid from said inlet and outlet ports
non-mechanically.
[0149] In some embodiments, the invention encompasses a method for
sequencing a biological polymer such as a nucleic acid utilizing
the above-described apparatus, comprising: directing a fluid
comprising a monomer to an array of reaction chambers wherein the
fluid has a fluid flow Reynolds number of at most 2000, 1000, 200,
100, 50, or 20. The method may optionally further comprise
detecting a pH or a change in pH from each said reaction chamber.
This is typically detected by ion diffusion to the sensor surface.
There are various other ways of providing a fluidics assembly for
delivering an appropriate fluid flow across the microwell and
sensor array assembly, and the forgoing examples are thus not
intended to be exhaustive.
pH-Based Nucleic Acid Sequencing
[0150] Apparatus of the invention may be adapted to detecting
hydrogen ions released by nucleotide incorporation, which detection
process is disclosed as a DNA sequencing method in Rothberg et al.,
U.S. patent publications 2009/0026082 and 2009/0127589. It is
important in these and various other aspects to detect as many
released hydrogen ions as possible in order to achieve as high a
signal (and/or a signal to noise ratio) as possible. Strategies for
increasing the number of released protons that are ultimately
detected by the chemFET surface include without limitation limiting
interaction of released protons with reactive groups in the well,
choosing a material from which to manufacture the well in the first
instance that is relatively inert to protons, preventing released
protons from exiting the well prior to detection at the chemFET,
and increasing the copy number of templates per well (in order to
amplify the signal from each nucleotide incorporation), among
others.
[0151] In one aspect, the invention provides arrays and devices
with reduced buffering capacity for monitoring and/or measuring
hydrogen ion changes (or pH changes) more accurately. As an
example, the invention provides apparatus and devices for
monitoring pH changes in polymerase extension reactions in an
environment with no or limited buffering capacity. Examples of a
reduced buffering environment include environments that lack pH
buffering components in the sample fluid and/or reaction mixtures;
environments in which surfaces of array components in contact with
sample fluid and/or reaction mixtures have little or no buffering
capacity; and environments in which pH changes on the order of
0.01, 0.05, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6. 0.7, 0.8, 0.9. or 1.0 pH
units are detectable for example via a chemFET and more
particularly an ISFET as described herein.
[0152] The buffering inhibitor may also be a phospholipid. The
phospholipids may be naturally occurring or non-naturally occurring
phospholipids. Examples of phospholipids to be used as buffering
inhibitors include but are not limited to phosphatidylcholine,
phosphatidylethanolamine, phosphatidylglycerol, and
phosphatidylserine. In some embodiments, phospholipids may be
coated on the chemFET surface (or reaction chamber surface). Such
coating may be covalent or non-covalent. In other embodiments, the
phospholipids exist in solution.
[0153] Some instances of the invention employ an environment,
including a reaction solution, that is minimally buffered, if at
all. Buffering can be contributed by the components of the solution
or by the solid supports in contact with such solution. A solution
having no or low buffering capacity (or activity) is one in which
changes in hydrogen ion concentration on the order of at least
about +/-0.005 pH units, at least about +/-0.01, at least about
+/-0.015, at least about +/-0.02, at least about +/-0.03, at least
about +/-0.04, at least about +/-0.05, at least about +/-0.10, at
least about +/-0.15, at least about +/-0.20, at least about
+/-0.25, at least about +/-0.30, at least about +/-0.35, at least
about +/-0.45, at least about +/-0.50, or more are detectable
(e.g., using the chemFET sensors described herein). In some
embodiments, the pH change per nucleotide incorporation is on the
order of about 0.005. In some embodiments, the pH change per
nucleotide incorporation is a decrease in pH. Reaction solutions
that have no or low buffering capacity may contain no or very low
concentrations of buffer, or may use weak buffers.
Concatemerized Templates
[0154] Increasing the number of templates or primers (i.e., copy
number) results in a greater number of nucleotide incorporations
per sensor and/or per reaction chamber, thereby leading to a higher
signal and thus signal to noise ratio. Copy number can be increased
for example by using templates that are concatemers (i.e., nucleic
acids comprising multiple, tandemly arranged, copies of the nucleic
acid to be sequenced), by increasing the number of nucleic acids on
or in beads up to and including saturating such beads, and by
attaching templates or primers to beads or to the sensor surface in
ways that reduce steric hindrance and/or ensure template attachment
(e.g., by covalently attaching templates), among other things.
Concatemer templates may be immobilized on or in beads or on other
solid supports such as the sensor surface, although in some
embodiments concatemers templates may be present in a reaction
chamber without immobilization. For example, the templates (or
complexes comprising templates and primers) may be covalently or
non-covalently attached to the chemFET surface and their sequencing
may involve detection of released hydrogen ions and/or addition of
negative charge to the chemFET surface upon a nucleotide
incorporation event. The latter detection scheme may be performed
in a buffered environment or solution (i.e., any changes in pH will
not be detected by the chemFET and thus such changes will not
interfere with detection of negative charge addition to the chemFET
surface).
[0155] RCA or CCR amplification methods generate concatemers of
template nucleic acids that comprise tens, hundreds, thousands or
more tandemly arranged copies of the template. Such concatemers may
still be referred to herein as template nucleic acids, although
they may contain multiple copies of starting template nucleic
acids. In some embodiments, they may also be referred to as
amplified template nucleic acids. Alternatively, they may be
referred to herein as comprising multiple copies of target nucleic
acid fragment. Concatemers may contain 2, 3, 4, 5, 6, 7, 8, 9, 10,
20, 30, 40, 50, 60, 70, 80, 90, 100, 500, 1000, or more copies of
the starting nucleic acid. They may contain 10-10.sup.2,
10.sup.2-10.sup.3, 10.sup.3-10.sup.4, 10.sup.3-10.sup.5, or more
copies of the starting nucleic acid. Concatemers generated using
these or other methods (such as for example DNA nanoballs) can be
used in the sequencing-by-synthesis methods described herein. The
concatemers may be generated in vitro apart from the array and then
placed into reaction chambers of the array or they may be generated
in the reaction chambers. One or more inside walls of the reaction
chamber may be treated to enhance attachment and retention of the
concatemers, although this is not required. In some embodiments of
the invention, if the concatemers are attached to an inside wall of
the reaction chamber, such as the chemFET surface, then nucleotide
incorporation at least in the context of a sequencing-by-synthesis
reaction may be detected by a change in charge at the chemFET
surface, as an alternative to or in addition to the detection of
released hydrogen ions as discussed herein. If the concatemers are
deposited onto a chemFET surface and/or into a reaction chamber,
sequencing-by-synthesis can occur through detection of released
hydrogen ions as discussed herein. The invention embraces the use
of other approaches for generating concatemerized templates. One
such approach is a PCR described by Stemmer et al. in U.S. Pat. No.
5,834,252, and the description of this approach is incorporated by
reference herein.
[0156] Important aspects of the invention contemplate sequencing a
plurality of different template nucleic acids simultaneously. This
may be accomplished using the sensor arrays described herein. In
one embodiment, the sensor arrays are overlayed (and/or integral
with) an array of microwells (or reaction chambers or wells, as
those terms are used interchangeably herein), with the proviso that
there be at least one sensor per microwell. Present in a plurality
of microwells is a population of identical copies of a template
nucleic acid. There is no requirement that any two microwells carry
identical template nucleic acids, although in some instances such
templates may share overlapping sequence. Thus, each microwell
comprises a plurality of identical copies of a template nucleic
acid, and the templates between microwells may be different.
[0157] It is to be understood therefore that the invention
contemplates a sequencing apparatus for sequencing unlabeled
nucleic acid acids, optionally using unlabeled nucleotides, without
optical detection and comprising an array of at least 100 reaction
chambers. In some embodiments, the array comprises 10.sup.3,
10.sup.4, 10.sup.5, 10.sup.6, 10.sup.7 or more reaction chambers.
The pitch (or center-to-center distance between adjacent reaction
chambers) is on the order of about 1-10 microns, including 1-9
microns, 1-8 microns, 1-7 microns, 1-6 microns, 1-5 microns, 1-4
microns, 1-3 microns, or 1-2 microns.
[0158] In various aspects and embodiments of the invention, the
nucleic acid loaded beads, of which there may be tens, hundreds,
thousands, or more, first enter the flow cell and then individual
beads enter individual wells. The beads may enter the wells
passively or otherwise. For example, the beads may enter the wells
through gravity without any applied external force. The beads may
enter the wells through an applied external force including but not
limited to a magnetic force or a centrifugal force. In some
embodiments, if an external force is applied, it is applied in a
direction that is parallel to the well height/depth rather than
transverse to the well height/depth, with the aim being to
"capture" as many beads as possible. Preferably, the wells (or well
arrays) are not agitated, as for example may occur through an
applied external force that is perpendicular to the well
height/depth. Moreover, once the wells are so loaded, they are not
subjected to any other force that could dislodge the beads from the
wells.
[0159] The Examples provide a brief description of an exemplary
bead loading protocol in the context of magnetic beads. It is to be
understood that a similar approach could be used to load other bead
types. The protocol has been demonstrated to reduce the likelihood
and incidence of trapped air in the wells of the flow chamber,
uniformly distribute nucleic acid loaded beads in the totality of
wells of the flow chamber, and avoid the presence and/or
accumulation of excess beads in the flow chamber.
[0160] In various instances, the invention contemplates that each
well in the flow chamber contain only one nucleic acid loaded bead.
This is because the presence of two beads per well will yield
unusable sequencing information derived from two different template
nucleic acids.
[0161] As part of the sequencing reaction, a dNTP will be ligated
to (or "incorporated into" as used herein) the 3' of the newly
synthesized strand (or the 3' end of the sequencing primer in the
case of the first incorporated dNTP) if its complementary
nucleotide is present at that same location on the template nucleic
acid. Incorporation of the introduced dNTP (and concomitant release
of PPi) therefore indicates the identity of the corresponding
nucleotide in the template nucleic acid. If no dNTP has been
incorporated, no hydrogens are released and no signal is detected
at the chemFET surface. One can therefore conclude that the
complementary nucleotide was not present in the template at that
location. If the introduced dNTP has been incorporated into the
newly synthesized strand, then the chemFET will detect a signal.
The signal intensity and/or area under the curve is a function of
the number of nucleotides incorporated (for example, as may occur
in a homopolymer stretch in the template. The result is that no
sequence information is lost through the sequencing of a
homopolymer stretch (e.g., poly A, poly T, poly C, or poly G) in
the template.
Example 1
On-Chip Polymerase Extension Detected by pH Shift on an ISFET
Array
[0162] Streptavidin-coated 2.8 micron beads carrying biotinylated
synthetic template to which sequencing primers and T4 DNA
polymerase are bound were subjected to three sequential flows of
each of the four nucleotides. Each nucleotide cycle consisted of
flows of dATP, dCTP, dGTP and dTTP, each interspersed with a wash
flow of buffer only. Flows from the first cycle are shown in blue,
flows from the second cycle in red, and the third cycle in yellow.
As shown in FIG. 10A, signal generated for both of the two dATP
flows were very similar. FIG. 10B shows that the first (blue) trace
of dCTP is higher than the dCTP flows from subsequent cycles,
corresponding to the flow in which the polymerase should
incorporate a single nucleotide per template molecule. FIG. 10C
shows that the first (blue) trace of dGTP is approximately 6 counts
higher (peak-to-peak) than the dGTP flows from subsequent cycles,
corresponding to the flow in which the polymerase should
incorporate a string of 10 nucleotides per template molecule. FIG.
10D shows that the first (blue) trace of dTTP is also approximately
6 counts higher (peak-to-peak) than the dTTP flows from subsequent
cycles, corresponding to the flow in which the polymerase should
incorporate 10 nucleotides per template molecule.
Example 2
Sequencing in a Closed System and Data Manipulation
[0163] Sequence has been obtained from a 23-mer synthetic
oligonucleotide and a 25-mer PCR product oligonucleotide. The
oligonucleotides were attached to beads which were then loaded into
individual wells on a chip having 1.55 million sensors in a
1348.times.1152 array having a 5.1 micron pitch (38400 sensors per
mm.sup.2). About 1 million copies of the synthetic oligonucleotide
were loaded per bead, and about 300000 to 600000 copies of the PCR
product were loaded per bead. A cycle of 4 nucleotides through and
over the array was 2 minutes long. Nucleotides were used at a
concentration of 50 micromolar each. Polymerase was the only enzyme
used in the process. Data were collected at 32 frames per
second.
[0164] FIG. 11A depicts the raw data measured directly from an
ISFET for the synthetic oligonucleotide. One millivolt is
equivalent to 68 counts. The data are sampled at each sensor on the
chip (1550200 sensors on a 314 chip) many times per second. The
Figure is color-coded for each nucleotide flow. With each
nucleotide flow, several seconds of imaging occur. The graph
depicts the concatenation of those individual measurements taken
during each flow. The Y axis is in raw counts, and the X axis is in
seconds. Superimposed just above the X axis are the expected
incorporations at each flow.
[0165] FIG. 11B depicts the integrated value for each nucleotide
flow, normalized to the template being sequenced. The integrated
value is taken from the raw trace measurements shown in FIG. 11A,
and the integral bounds have been chosen to maximize signal to
noise ratio. The results have been normalized to the signal per
base incorporation, and graphed per nucleotide flow. The Y axis is
incorporation count, and the X axis is nucleotide flow number,
alternating through TACG.
* * * * *