U.S. patent application number 11/770299 was filed with the patent office on 2009-01-01 for chip to chip optical interconnect.
This patent application is currently assigned to INTERNATIONAL BUSINESS MACHINES CORPORATION. Invention is credited to Punit P. Chiniwalla, Philip Hobbs, Theodore G. van Kessel.
Application Number | 20090003762 11/770299 |
Document ID | / |
Family ID | 40160618 |
Filed Date | 2009-01-01 |
United States Patent
Application |
20090003762 |
Kind Code |
A1 |
Chiniwalla; Punit P. ; et
al. |
January 1, 2009 |
CHIP TO CHIP OPTICAL INTERCONNECT
Abstract
An apparatus for optical communication is provided. The
apparatus includes a first waveguide formed on a first surface and
a second waveguide formed on a second surface. The first and second
surfaces are bonded together to form an air gap between the first
and second surfaces and diffraction gratings of the first and
second waveguides are facing each other
Inventors: |
Chiniwalla; Punit P.; (Ann
Arbor, MI) ; Hobbs; Philip; (Briarcliff Manor,
NY) ; van Kessel; Theodore G.; (Millbrook,
NY) |
Correspondence
Address: |
PERMAN & GREEN, LLP
425 POST ROAD
FAIRFIELD
CT
06824
US
|
Assignee: |
INTERNATIONAL BUSINESS MACHINES
CORPORATION
Armonk
NY
|
Family ID: |
40160618 |
Appl. No.: |
11/770299 |
Filed: |
June 28, 2007 |
Current U.S.
Class: |
385/14 |
Current CPC
Class: |
G02B 6/1228 20130101;
G02B 6/124 20130101; G02B 6/43 20130101 |
Class at
Publication: |
385/14 |
International
Class: |
G02B 6/12 20060101
G02B006/12 |
Claims
1. A method for chip to chip communication, the method comprising:
forming a first waveguide on a first surface; forming a second
waveguide on a second surface; bonding the first surface to the
second surface so an air gap exists between the first and second
surface and diffraction gratings of the first and second wave
guides face each other; and passing a light beam through the first
waveguide, across the air gap and into the second waveguide;
forming a third waveguide on a third surface; and bonding the third
surface to the second surface so an air gap exists between the
third and second surface and diffraction gratings of the third and
second wave guides face each other; wherein the light beam passes
from the second wave guide across the air gap and into the third
waveguide.
2. (canceled)
3. The method of claim 1, wherein the first and third surfaces
comprise microchips.
4. The method of claim 1, where the first surface and second
surface are bonded by controlled collapse chip connection.
5. The method of claim 1 wherein forming the first and second
waveguides comprises: forming a rectangular waveguide and a
diffraction grating terminal connected by a tapered waveguide;
wherein the rectangular waveguide and tapered waveguide are formed
by lithography.
6. The method of claim, wherein the diffraction gratings are formed
by a dual damascene process.
7. The method of claim 1 wherein the diffraction gratings are metal
tuned.
8. The method of claim 1 further comprising the first waveguide and
second waveguide being formed on a top surface area of the first
surface and second surface, respectively.
9. An apparatus for optical communication comprising: a first
waveguide formed on a first surface; and a second waveguide formed
on a second surface; wherein the first and second surfaces are
bonded together to form an air gap between the first and second
surfaces and diffraction gratings of the first and second
waveguides are facing each other; and a third waveguide formed on a
third surface; wherein the third and second surfaces are bonded
together to form an air gap between the first and second surfaces
and diffraction gratings of the third and second waveguides are
facing each other.
10. (canceled)
11. The apparatus of claim 9, wherein the waveguides are configured
so the first waveguide and third waveguides are in communication
with each other via the second waveguide.
12. The apparatus of claim 9, wherein the first and third surfaces
comprise a microchip.
13. The apparatus of claim 9, wherein the first and second
waveguides comprise at least: a rectangular waveguide section; a
diffraction grating terminal; and a tapered waveguide section
longitudinally connecting the rectangular waveguide to the
diffraction grating terminal.
14. The apparatus of claim 13 wherein the rectangular waveguide
section comprises a 0.4 micron dimension, the tapered waveguide
section is tapered laterally 20 degrees and the diffraction grating
terminal is 10 microns wide and 10 microns long.
15. The apparatus of claim 9, wherein the first waveguide comprises
silicon or a polymer and second waveguide comprises silicon or a
polymer.
16. The apparatus of claim 9, wherein the diffraction gratings
comprise a two-dimensional diffraction grating made of a single
metal layer.
17. The apparatus of claim 9, wherein the diffraction gratings of
the first and second waveguides have non-uniform spacing.
18. The apparatus of claim 9 further comprising the first waveguide
and the second waveguide being formed on a top surface area of the
first surface and the second surface, respectively.
19. An apparatus for optical chip to chip communication comprising:
a first waveguide formed on an outer area of a first surface; a
second waveguide formed on an outer area of a second surface, the
second waveguide having at least two diffraction grating sections;
and a third waveguide formed on an outer area of a third surface;
wherein the outer areas of the first and third surfaces are bonded
to the outer area of the second surface so an air gap exists
between the outer areas of the first and second surfaces and the
third and second surfaces, a diffraction grating section of the
first waveguide is facing a first diffraction grating of the second
waveguide and a diffraction grating of the third surface is facing
a second diffraction grating of the second surface.
20. An apparatus for optical chip to chip communication comprising:
a first waveguide formed on a first surface; and a second waveguide
formed on a second surface; wherein the first and second surfaces
are bonded together to form an air gap between the first and second
surfaces and diffraction gratings of the first and second
waveguides are facing each other, the first waveguide having three
elements; and a third waveguide formed on a third surface; wherein
the third and second surfaces are bonded together to form an air
gap between the first and second surfaces and diffraction gratings
of the third and second waveguides are facing each other, the third
waveguide having three elements.
21. (canceled)
22. The apparatus of claim 20, wherein the first, second and third
waveguides comprise at least: a rectangular waveguide section; a
diffraction grating terminal; and a tapered waveguide section
longitudinally connecting the rectangular waveguide to the
diffraction grating terminal.
23. The apparatus of claim 19 wherein the waveguides are configured
so the first waveguide and third waveguides are in communication
with each other via the second waveguide.
24. The apparatus of claim 19, wherein the first and second
waveguides comprise at least: a rectangular waveguide section; a
diffraction grating terminal; and a tapered waveguide section
longitudinally connecting the rectangular waveguide to the
diffraction grating terminal.
25. The apparatus of claim 19 further comprising the first
waveguide and the second waveguide being formed on a top surface
area of the first surface and the second surface, respectively.
Description
BACKGROUND
[0001] 1. Field
[0002] The present embodiments relate to optical connections and,
more particularly, to optical connections between microchips.
[0003] 2. Brief Description of Related Developments
[0004] Large symmetric multiprocessor machines such as blade
servers are running out of bandwidth for blade-to-blade,
module-to-module and eventually chip-to-chip interconnections. The
bandwidth required for these interconnections is growing more
rapidly than Moore's law because this bandwidth is the product of
the increasing number of processors, the increasing complexity of
each processor and the increasing clock speed of those processors.
In order to build a complete optical interconnection hierarchy
several levels of connections should be addressed. The optical
signal travels from the chip to the module, from the module to the
board, from the board to the backplane, across the backplane and
then back through these levels in the reverse order.
[0005] In conventional mainframe and high-end applications, chips
interconnect via a substrate using controlled collapse chip
connection (C4 or flip chip) bonding. The substrate may be, for
example, a glass ceramic multichip module or silicon interposer.
Chip to chip electrical connections flow from a chip through the C4
interconnect to the multichip module and then back through the C4
interconnect to the next chip. These C4 electrical connections are
not sufficient to accommodate the necessary increase in
bandwidth.
[0006] Existing optical connections rely on multimode waveguides or
multimode fibers because the large cores of multimode waveguides do
not require tight mechanical tolerances during assembly, the losses
within the short range between chips is negligible and the sources
are often multimode. These existing optical connections use volume
holographic gratings that are not a feasible manufacturing approach
to the problem of the increasing bandwidth requirements because
these volume holographic gratings cannot be easily created using
microlithography.
[0007] It would be advantageous to replace some of the chip to chip
electrical connections with easily manufactured optical connections
to allow higher speeds over chip to chip distances.
SUMMARY
[0008] In one exemplary embodiment, a method for chip to chip
communication is provided. The method includes forming a first
waveguide on a first surface, forming a second waveguide on a
second surface, bonding the first surface to the second surface so
an air gap exists between the first and second surface and
diffraction gratings of the first and second wave guides face each
other and passing a light beam through the first waveguide, across
the air gap and into the second waveguide.
[0009] In another exemplary embodiment, an apparatus for optical
communication is provided. The apparatus includes a first waveguide
formed on a first surface and a second waveguide formed on a second
surface. The first and second surfaces are bonded together to form
an air gap between the first and second surfaces and diffraction
gratings of the first and second waveguides are facing each
other.
[0010] In one exemplary embodiment, an apparatus for optical chip
to chip communication is provided. The apparatus includes a first
waveguide formed on a first surface, a second waveguide formed on a
second surface, the second waveguide having at least two
diffraction grating sections and a third waveguide formed on a
third surface. The first and third surfaces are bonded to the
second surface so an air gap exists between the first and second
surfaces and the third and second surfaces, a diffraction grating
section of the first waveguide is facing a first diffraction
grating of the second waveguide and a diffraction grating of the
third surface is facing a second diffraction grating of the second
surface.
[0011] In still another exemplary embodiment, an apparatus for
optical chip to chip communication is provided. The apparatus
includes a first waveguide formed on a first surface and a second
waveguide formed on a second surface. The first and second surfaces
are bonded together to form an air gap between the first and second
surfaces and diffraction gratings of the first and second
waveguides are facing each other, the first waveguide having three
elements.
BRIEF DESCRIPTION OF THE DRAWINGS
[0012] The foregoing aspects and other features of the present
embodiments are explained in the following description, taken in
connection with the accompanying drawings, wherein:
[0013] FIG. 1 shows a chips in accordance with an exemplary
embodiment bonded to a substrate;
[0014] FIG. 2 illustrates an optical interconnect routing in
accordance with an exemplary embodiment;
[0015] FIG. 3 shows a side view of the optical interconnect of FIG.
2;
[0016] FIGS. 4A and 4B illustrate a top and side view of a
waveguide in accordance with an exemplary embodiment; and
[0017] FIG. 5 illustrates a flow diagram of a method in accordance
with an exemplary embodiment.
DETAILED DESCRIPTION OF THE EXEMPLARY EMBODIMENT(s)
[0018] FIG. 1 illustrates silicon chips bonded to a substrate.
Although the present embodiments will be described with reference
to the exemplary embodiments shown in the drawings and described
below, it should be understood that the present embodiments could
be embodied in many alternate forms of embodiments. In addition,
any suitable size, shape or type of elements or materials could be
used.
[0019] The chips 110A and 110B and the multichip module, carrier or
interposer 100 may contain the waveguide 300 of FIGS. 4A and 4B.
The waveguide 300 may be fabricated on, for example, the device
side or C4 mounting surface of a substrate 400 by any suitable
method such as lithographic thin film methods (FIG. 5, Blocks 500
and 510). The device side of the substrate 400 such as the chips
110A, 110B or the carrier 100 may be available for the disclosed
waveguides and is unobstructed by heatsinks, thermal hats or
grease. In alternate embodiments the waveguide 300 may be formed on
any suitable surface of the substrate 400. In one embodiment, the
waveguide 300 is not embedded in the corresponding structure on
three sides. The waveguide 300 is formed on a top or outer surface
of the respective structure, which can include for example a chip,
module, carrier, interposer or other suitable structure.
[0020] The waveguide 300 may be made of any suitable material
capable of transmitting light, such as for example silicone,
silicon/SiO2 or a polymer.
[0021] The waveguide 300 may be a multimode waveguide or any other
suitable waveguide. The waveguide 300 may include a rectangular
waveguide 410 and a grating terminal or section 430 that are
longitudinally connected by a tapered waveguide 420. The
rectangular waveguide 410 may have, for example, a width of about
0.4 microns. The tapered waveguide 420 may, for example, extend
longitudinally away from the rectangular waveguide 410 while
expanding the lateral dimension of the rectangular waveguide by,
for example, approximately twenty degrees to mate with the grating
terminal or section 430. The grating section 430 may be, for
example, approximately ten microns wide and approximately ten
microns long. In alternate embodiments the waveguide 300 and its
sections 410-430 may have any suitable shape, dimensions and/or
configuration. The grating terminal 430 may contain diffraction
gratings 440 having a directional selectivity to send the
propagated light in and out of the chip in the direction of 100A,
110B rather than into the body of chip 110A, 110B. For example, as
shown in FIG. 2, the gratings 440 of a waveguide 300 in chip 110B
may send the propagated light out of the chip 110B in the direction
130 and into a waveguide 300 in the carrier 100. The diffraction
gratings 440 may be, for example, two-dimensional gold gratings
formed by any suitable method such as, for example, lithography,
dual damascene, etching, chemical-mechanical planarization or any
combination thereof. The gratings 440 may be made of a single metal
layer or any suitable number of layers. The gratings 440 may be,
for example, blazed gratings (i.e. gratings with a nonsymmetrical
line profile) that provide directional selectivity formed with dual
Damascene technology in which two or more metal layers are directly
adjacent to each other and patterned separately. In alternate
embodiments, the gratings 440 may be tuned so that electrical or
magnetic fields induced within the gratings 440 exhibit phase
shifts (i.e. metal tuning). The interaction of these phase shifts
with asymmetric vertical placement of the metal in the waveguide
300 may provide directional selectivity in a way similar to the
copper ring (e.g. the shading-coil) of a shaded-pole AC motor.
[0022] The waveguide 300 may propagate any suitable wavelength of
light such as for example the infrared 1.5 micron or 3 micron
communication wavelengths. In alternate embodiments the waveguide
300 may be configured through suitable material choices to
propagate visible light.
[0023] The chips 110A and 110B containing the wave guide 300 may be
bonded to the carrier 100, which also contains a waveguide 300, in
any suitable manner such as by C4 connection 120 (FIG. 5, Block
520). The C4 connections 120 are solder balls placed on
corresponding contact pads of the chips 110A, 110B and the carrier
100. As heat is applied, the solder melts and flows causing the
chips 110A, 110B to be bonded to the carrier 100. These bonds
formed by the solder balls form the electrical connections between
the contact pads of the chips 110A, 110B and the contact pads of
the carrier 100. The C4 bonding process limits the lateral and
vertical alignment error of the chips 110A and 110B to
approximately .+-.1.0 micron. The C4 bonding process also produces
an air gap 340 between chips of approximately 50 microns.
[0024] Referring to FIG. 3, as the chip 110 is bonded to the
carrier 100, the inherent lateral and vertical alignment properties
of the C4 connection permit the grating terminals or regions 430 of
the waveguides 300A, 300B in the chip 110 and in the carrier 100
respectively to be in alignment. In essence, the optical connection
between the chip 110 and the carrier 100 is an "optical C4"
connection that is capable of being interspersed with ordinary
solder C4's without requiring any special alignment or preparation
beyond what is already used in, for example, fluxless C4 bonding.
Although, in this example, the C4 connection is used to describe
the bonding between the chip 110 and the carrier 100, any suitable
bonding method having sufficient alignment properties can be
used.
[0025] As noted above, for each optical connection two waveguides
300A, 300B are used in pairs and oriented to face each other across
the gap 340 as shown in FIG. 3. One waveguide 300A is, fabricated
on the device or bonding side of the chip 110 while the other is
similarly fabricated on a corresponding side of the carrier 100.
Because the two waveguides 300A, 300B communicate via the free
space or air gap 340, the waveguide technologies can be different
in the chip 100 and the carrier 100. For example, a silicon/SiO2
waveguide on the chip 110 could communicate with a polymer
waveguide on the carrier 100.
[0026] In the exemplary configuration shown in FIG. 2, the light
330 propagates laterally through the waveguides on the chips 110A,
110B and the carrier 100. To bridge the air gap 340 the light makes
a ninety-degree bend via the gratings 440. When the light 330
crosses the gap 340 it enters the gratings 440 of the corresponding
waveguide and makes another ninety-degree bend. Arrows 130 and 140
indicate the path of light travel in this exemplary embodiment. The
light 330 crossing the air gap 340 may have, for example, a waist
at a point between the two interacting surfaces (e.g. gratings 440)
of the wave guides 300A, 300B so that the coupling loss has a broad
minimum at the expected separation between the chip 100 and the
carrier 100.
[0027] By expanding the guided light beam 330 in the exit region up
to a sufficient diameter, such as for example, a 10 or 20 micron
diameter that is focused some distance away from the surface, the
alignment tolerances of the chip/carrier bonding process are
accommodated. In addition, the light beam may be configured so that
uncertainties in the distance between the chip 110 and the carrier
100 (i.e. the air gap 340) due to variations in the solder volume
in each C4 joint are within the depth of focus of the beam 330
leaving the exit region.
[0028] In operation, for example, light enters the waveguide
structure 300A in the chip 110 via the rectangular waveguide 410
section. The rectangular waveguide section mates with a grating
terminal or section 430 through the tapered waveguide 420 section.
As shown in FIG. 3, the rectangular waveguide 410 and tapered
waveguide 420 are shown as section 310. As the light enters the
grating section 430 it interacts with the two-dimensional
diffraction grating 440 that bends the light ninety-degrees and
sets the state of focus of the beam. The light beam 330 traverses
the air gap 340 and enters the grating section 320 of the
corresponding waveguide 300B where it is bent ninety-degrees and
coupled to the waveguide 300B for propagation through the carrier
100 for subsequent transmission to another chip (FIG. 5, Block
530). Both ends of the waveguide 300B in the in the carrier 100 may
have substantially the same diffraction gratings so that the light
beam 330 is passed from chip to chip through the waveguide 300B.
Light beam 330 transmission from the carrier 100 to the chip 110
occurs in substantially the opposite way as described above for
light beam transmission from the chip 110 to the carrier 100 (e.g.
light propagation is bidirectional). The two gratings 430, 320 may
have non-uniform spacing in general, because the field exiting the
tapered region may have curved phase fronts and because the
focusing of the light beam 330 may entail a quadratic phase change
across the exit region.
[0029] The disclosed embodiments may provide a cost effective,
easily manufactured and practical method of bringing optical
signals to and from a multichip module from outside the module as
well as from chip to chip.
[0030] It should be understood that the foregoing description is
only illustrative of the embodiments. Various alternatives and
modifications can be devised by those skilled in the art without
departing from the embodiments. Accordingly, the present
embodiments are intended to embrace all such alternatives,
modifications and variances that fall within the scope of the
appended claims.
* * * * *