U.S. patent application number 17/502151 was filed with the patent office on 2022-02-10 for methods and systems for optical beam steering.
This patent application is currently assigned to Massachusetts Institute of Technology. The applicant listed for this patent is Massachusetts Institute of Technology. Invention is credited to DIRK ENGLUND, Jeffrey S. HERD, Paul William JUODAWLKIS, Mihika PRABHU, Scott A. SKIRLO, Marin SOLJACIC, Cheryl Marie SORACE-AGASKAR, Simon VERGHESE, Yi YANG.
Application Number | 20220043323 17/502151 |
Document ID | / |
Family ID | |
Filed Date | 2022-02-10 |
United States Patent
Application |
20220043323 |
Kind Code |
A1 |
SKIRLO; Scott A. ; et
al. |
February 10, 2022 |
METHODS AND SYSTEMS FOR OPTICAL BEAM STEERING
Abstract
An integrated optical beam steering device includes a planar
dielectric lens that collimates beams from different inputs in
different directions within the lens plane. It also includes an
output coupler, such as a grating or photonic crystal, that guides
the collimated beams in different directions out of the lens plane.
A switch matrix controls which input port is illuminated and hence
the in-plane propagation direction of the collimated beam. And a
tunable light source changes the wavelength to control the angle at
which the collimated beam leaves the plane of the substrate. The
device is very efficient, in part because the input port (and thus
in-plane propagation direction) can be changed by actuating only
log.sub.2 N of the N switches in the switch matrix. It can also be
much simpler, smaller, and cheaper because it needs fewer control
lines than a conventional optical phased array with the same
resolution.
Inventors: |
SKIRLO; Scott A.; (Boston,
MA) ; SORACE-AGASKAR; Cheryl Marie; (Bedford, MA)
; SOLJACIC; Marin; (Belmont, MA) ; VERGHESE;
Simon; (Arlington, MA) ; HERD; Jeffrey S.;
(Rowley, MA) ; JUODAWLKIS; Paul William;
(Arlington, MA) ; YANG; Yi; (Cambridge, MA)
; ENGLUND; DIRK; (Brookline, MA) ; PRABHU;
Mihika; (Cambridge, MA) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Massachusetts Institute of Technology |
Cambridge |
MA |
US |
|
|
Assignee: |
Massachusetts Institute of
Technology
Cambridge
MA
|
Appl. No.: |
17/502151 |
Filed: |
October 15, 2021 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
16842048 |
Apr 7, 2020 |
11175562 |
|
|
17502151 |
|
|
|
|
16284161 |
Feb 25, 2019 |
10649306 |
|
|
16842048 |
|
|
|
|
15630235 |
Jun 22, 2017 |
10261389 |
|
|
16284161 |
|
|
|
|
62353136 |
Jun 22, 2016 |
|
|
|
International
Class: |
G02F 1/313 20060101
G02F001/313; G02F 1/295 20060101 G02F001/295; G01S 7/481 20060101
G01S007/481 |
Goverment Interests
GOVERNMENT SUPPORT
[0002] This invention was made with Government support under
Contract No. FA8721-05-C-0002 awarded by the U.S. Air Force. The
Government has certain rights in the invention.
Claims
1. An optical beam steering apparatus comprising: a substrate; a
slab waveguide formed on the substrate; a planar lens, having a
focal surface patterned in the slab waveguide, to direct light
guided by the slab waveguide; and an output coupler, formed on the
substrate in optical communication with the slab waveguide, to
couple at least a portion of the light out of a plane of the slab
waveguide.
2. The optical beam steering apparatus of claim 1, wherein the
planar lens comprises a patterned layer of polysilicon disposed on
a layer of silicon nitride.
3. The optical beam steering apparatus of claim 1, wherein the
planar lens is one of a Luneburg lens or a Rotman lens.
4. The optical beam steering apparatus of claim 1, wherein the
planar lens has a height that is adiabatically tapered to reduce
loss at an interface between the planar lens and the slab
waveguide.
5. The optical beam steering apparatus of claim 1, wherein the
output coupler comprises a curved grating.
6. The optical beam steering apparatus of claim 5, wherein the
curved grating has a curvature selected to collimate the light
directed by the planar lens.
7. The optical beam steering apparatus of claim 5, wherein the
planar lens is configured to collimate the light guided by the slab
waveguide.
8. The optical beam steering apparatus of claim 1, wherein the
output coupler comprises a two-dimensional photonic crystal
coupler.
9. The optical beam steering apparatus of claim 1, wherein the
output coupler has a broken top-down mirror symmetry selected to
couple the light out of the plane of the slab waveguide
asymmetrically.
10. The optical beam steering apparatus of claim 1, wherein the
output coupler is configured to couple incident light into the
plane of the slab waveguide and the planar lens is configured to
couple the incident light into a mode guided by the slab
waveguide.
11. The optical beam steering apparatus of claim 10, further
comprising: a detector, in optical communication with the slab
waveguide, to detect the incident light in the mode guided by the
slab waveguide.
12. The optical beam steering apparatus of claim 1, further
comprising: at least one optical amplifier, in optical
communication with the slab waveguide, to amplify the light guided
by the slab waveguide.
13. The optical beam steering apparatus of claim 12, wherein the at
least one optical amplifier comprises a slab-coupled optical
waveguide amplifier integrated with the slab waveguide.
14. The optical beam steering apparatus of claim 12, wherein the at
least one optical amplifier comprises semiconductor optical
amplifier switches.
15. The optical beam steering apparatus of claim 1, further
comprising: a passive splitter network, formed on the substrate in
optical communication with the slab waveguide, to couple light into
and/or out of the slab waveguide.
16. The optical beam steering apparatus of claim 1, further
comprising: a network of switches, formed on the substrate in
optical communication with the slab waveguide, to couple light into
and/or out of the slab waveguide.
17. A system comprising: a seed laser to generate the light; and an
array of the optical beam steering apparatuses of claim 1 tiled
together and operably coupled to the seed laser.
18. A method of optical beam steering, the method comprising:
exciting an input to a planar lens with light guided by a slab
waveguide; directing the light within a plane of the slab waveguide
with the planar lens; and coupling the light out of the plane of
the slab waveguide.
19. The method of claim 18, wherein directing the light within
plane of the slab waveguide with the planar lens comprises
collimating the light.
20. The method of claim 18, wherein coupling the light out of the
plane of the slab waveguide comprises diffracting the light with a
curved grating in optical communication with the planar lens.
Description
CROSS-REFERENCE TO RELATED APPLICATIONS
[0001] This application is a continuation of U.S. application Ser.
No. 16/842,048, filed Apr. 7, 2020, and entitled "Methods and
Systems for Optical Beam Steering," which is a continuation of U.S.
application Ser. No. 16/284,161, now U.S. Pat. No. 10,649,306,
filed Feb. 25, 2019, and entitled "Methods and Systems for Optical
Beam Steering," which is a continuation of U.S. application Ser.
No. 15/630,235, now U.S. Pat. No. 10,261,389, filed Jun. 22, 2017,
and entitled "Methods and Systems for Optical Beam Steering," which
in turn claims the priority benefit, under 35 U.S.C. .sctn. 119(e),
of U.S. Application No. 62/353,136, filed Jun. 22, 2016, and
entitled "Integrated Lens-Enabled LIDAR System." Each of these
applications is incorporated by reference herein.
BACKGROUND
[0003] The meteoritic rise of autonomous navigation in real-world
settings for self-driving cars and drones has propelled rapidly
growing academic and commercial interest in LIDAR. One of the key
application spaces that has yet to be filled, but is of great
interest, is a non-mechanically steered LIDAR sensor which has
substantial range (e.g., 100-300 m), low power (e.g., 1-10 W), low
cost (e.g., hundreds of dollars), high resolution (e.g., 10.sup.4
to 10.sup.6 pixels) and small size (e.g., 10 cm.sup.3). There are
several candidate technologies including micro-mechanical mirrors,
liquid-crystal based devices, and integrated photonics that are
currently being explored academically and commercially to fill this
niche.
[0004] Current state-of-the-art chip-scale integrated-photonic
LIDARs are based on 1D or 2D phased array antennas. In this type of
architecture, a 1D or 2D array of dielectric grating antennas is
connected to electrically-controlled thermo-optic (TO) or
electro-optic phase shifters. These phase shifters are fed by
waveguides splitting off from one main dielectric waveguide which
brings power from an off-chip or on-chip source. By applying a
gradient to the phases tuning each antenna, in-plane or
out-of-plane beam-steering can be enabled.
[0005] The direct predecessor of this architecture are radio
frequency (RF) phased arrays developed for military and commercial
RADARs. Although the detailed implementation is different because
RF primarily relies on metallic waveguides and structures whereas
integrated photonics uses dielectrics, optical phased arrays are
essentially based on directly replacing RF components with their
optical equivalents. This direct translation brings a significant
disadvantage: whereas metallic waveguides can be spaced at
sub-wavelength pitches, eliminating parasitic grating lobes,
dielectric waveguides have to be separated by several wavelengths
to prevent excessive coupling, resulting in significant grating
lobes.
[0006] RF phased array radars are routinely produced with closely
spaced antennas (<.lamda./2 apart) in subarrays that can be
tiled to create very large apertures. This provides wide-angle
steering and scaling to large power-aperture designs.
Fundamentally, the radiating elements can be closely spaced with
independent control circuitry because the amplifiers, phase
shifters and switches in the RF are implemented as subwavelength
lumped elements.
[0007] Current chip-scale optical phased arrays often reproduce RF
phased array architectures, with RF elements replaced with their
optical analogs. Fundamentally, the optical analogs to RF
components are traveling-wave designs that are multiple wavelengths
long and spaced apart by more than .lamda./2. This design allows
beam-steering over very small angles. In an end-fed geometry, for
example, the grating antenna elements can be closely spaced for
wide-angle azimuthal steering and use wavelength tuning to change
the out-coupling angle of the gratings for elevation steering. But
this end-fed geometry cannot be tiled without introducing
significant grating lobes due to its sparsity.
SUMMARY
[0008] Embodiments of the present technology include an optical
beam steering apparatus comprising a substrate, a plurality of
waveguides formed on the substrate, a planar dielectric lens formed
on the substrate in optical communication with the waveguides, and
an output coupler formed on the substrate in optical communication
with the planar dielectric lens. In operation, the waveguides
comprise a first waveguide and a second waveguide. The planar
dielectric lens collimates light emitted by the first waveguide as
a first collimated beam propagating in a first direction in a plane
of the substrate and collimates light emitted by the second
waveguide as a second collimated beam propagating in a second
direction in the plane of the substrate different than the first
direction. And the output coupler guides the first collimated beam
in the first direction and the second collimated beam in the second
direction and couples at least a portion of the first collimated
beam and the second collimated beam out of the plane of the
substrate.
[0009] The optical beam steering apparatus may include at least 32,
100, or 1000 waveguides in optical communication with the planar
dielectric lens. The light emitted by these waveguides may not be
phase coherent (e.g., the first waveguide may have an arbitrary
phase relative to the light emitted by the second waveguide).
[0010] The planar dielectric lens may have a shape selected to
satisfy the Abbe sine condition. It can have a single focal point
or multiple (i.e., two or more) focal points.
[0011] The output coupler can comprise a one-dimensional grating
configured to diffract the first collimated beam and the second
collimated beam out of the plane of the substrate. It could also
include a two-dimensional photonic crystal that couples the first
and second collimated beams out of the substrate.
[0012] Examples of such an optical beam steering apparatus may also
include a tunable light source in optical communication with the
waveguides. This tunable light source tunes a wavelength of the
light emitted by the first waveguide and the light emitted by the
second waveguide. For instance, the tunable light source may tune
the wavelength to steer the first collimated beam to one of at
least 15, 50, 100, or 1000 resolvable angles with respect to a
surface normal of the plane of the substrate.
[0013] These examples may also include a network of optical
switches formed on the substrate in optical communication with the
tunable light source and the waveguides. This network guides the
light emitted by the first waveguide from the tunable light source
to the first waveguide when in a first state and guides the light
emitted by the second waveguide from the tunable light source to
the second waveguide when in a second state. In cases where there
are N waveguides, switching from the first state to the second
state involves actuating up to log.sub.2 N optical switches in the
network of optical switches. The optical beam steering apparatus
may also include a plurality of optical amplifiers formed on the
substrate in optical communication with the network of optical
switches and the waveguides. These amplifiers amplify the light
emitted by the first waveguide and the light emitted by the second
waveguide.
[0014] Other examples of the present technology include a lidar
with a light source, a network of optical switches, a planar
dielectric lens, and a periodic structure. In operation, the light
source emits a beam of light. The network of optical switches,
which are in optical communication with the tunable light source,
guides the beam of light to a first waveguide in a plurality of
waveguides. The planar dielectric lens, which is in optical
communication with the waveguides and has a shape selected to
satisfy the Abbe sine condition, collimates the beam of light
emitted by the first waveguide as a first collimated beam
propagating in a first direction. And the periodic structure, which
is in optical communication with the planar dielectric lens,
diffracts at least a portion of the first collimated beam at an
angle with respect to the first direction.
[0015] All combinations of the foregoing concepts and additional
concepts discussed in greater detail below (provided such concepts
are not mutually inconsistent) are part of the inventive subject
matter disclosed herein. In particular, all combinations of claimed
subject matter appearing at the end of this disclosure are part of
the inventive subject matter disclosed herein. The terminology used
herein that also may appear in any disclosure incorporated by
reference should be accorded a meaning most consistent with the
particular concepts disclosed herein.
BRIEF DESCRIPTIONS OF THE DRAWINGS
[0016] The skilled artisan will understand that the drawings
primarily are for illustrative purposes and are not intended to
limit the scope of the inventive subject matter described herein.
The drawings are not necessarily to scale; in some instances,
various aspects of the inventive subject matter disclosed herein
may be shown exaggerated or enlarged in the drawings to facilitate
an understanding of different features. In the drawings, like
reference characters generally refer to like features (e.g.,
functionally similar and/or structurally similar elements).
[0017] FIG. 1A shows an optical beam-forming device with a planar
dielectric lenses.
[0018] FIG. 1B shows a bifocal lens suitable for use in an optical
beam-forming device like the one shown in FIG. 1A.
[0019] FIG. 1C shows a bootlace lens suitable for use in an optical
beam-forming device like the one shown in FIG. 1A.
[0020] FIGS. 2A-2C show illustrate steering an optical beam with
the optical beam-forming device of FIG. 1A.
[0021] FIGS. 3A and 3B illustrate transmitting and receiving,
respectively, with the optical beam-forming device of FIG. 1A.
[0022] FIG. 4A shows a simulated aperture pattern of an optical
beam-forming device with a planar dielectric lens.
[0023] FIG. 4B shows the simulated far-field directivity and
far-field beam angles for the ideal aperture given in FIG. 4A
[0024] FIG. 5A shows ray-tracing simulations used to determine the
optimal port position and relative angle.
[0025] FIG. 5B is a heat plot showing far-field beam spots in
u.sub.x and u.sub.y space.
[0026] FIG. 5C shows three-dimensional (3D) beam patterns
corresponding to those in FIG. 5B.
[0027] FIG. 6A illustrates beam steering with a conventional
optical phased array using phase shifters.
[0028] FIG. 6B illustrates beam steering with a planar dielectric
lens and switch matrix.
[0029] FIG. 7 is a block diagram of an integrated LIDAR chip with a
planar dielectric lens for optical beam steering.
[0030] FIG. 8 shows a LIDAR with a tunable on-chip source that is
modulated by a microwave chirp.
[0031] FIG. 9 shows an integrated optical beamforming system that
scales the basic design of FIG. 8 to add functionality for N
independently controllable beams.
[0032] FIG. 10A shows how the unit cells shown in FIGS. 8 and 9 can
be modified for tiling an M by N array.
[0033] FIG. 10B shows the tiling of the unit cells to form larger
apertures.
[0034] FIG. 11 shows 3 dB overlapped far-field beam patterns for a
lens-enabled beam-forming system.
[0035] FIG. 12A shows a simulation of a far-field beam pattern to
extract phase center.
[0036] FIG. 12B shows a simulation of a far-field beam pattern to
extract gaussian beamwidth.
[0037] FIG. 12C shows a 2D simulation of on-axis port excitation of
a lens feed.
[0038] FIG. 12D shows a 2D simulation of off-axis port excitation
of the lens feed.
DETAILED DESCRIPTION
[0039] Although the optical analogy to RF phased arrays has been
well explored, there is an entire class of planar-lens based
devices developed in the RADAR literature that performs the same
function as an RF phased array with integrated-photonics analogs.
Instead of relying upon many continuously tuned thermal phase
shifters to steer beams, an integrated-photonics device excites the
focal plane of a specially designed planar lenses to generate a
discrete far-field beam.
[0040] This approach to making an on-chip beam-steering device
(e.g., for LIDAR applications) starts with widely-spaced
transmit/receive waveguides (>10.lamda. apart) that include
SOAs, phase shifters, directional couplers, and RF photodiodes. The
beam-steering device uses a wide-angle planar dielectric lens--an
optical equivalent of a Rotman lens for RF beamforming--to convert
the sparse array of waveguides into a dense array of output
waveguides (.about..lamda./2 apart) to enable wide-angle steering.
Exciting a given input port to the planar dielectric lens steers
the beam in the plane of the lens, and changing the beam's
wavelength steers the beam out of the plane.
[0041] This device can be tiled with optical overlapped subarrays
to suppress sidelobes. These passive beamforming structures can be
realized using silicon technology and may be butt-coupled to active
photonic chips that provide the active transmit/receive functions.
Fixed phase shifts in the beamformer chip could create a sinc-like
pattern in the near-field in the vertical direction. This
transforms into a rectangular beam pattern in the far-field,
suppressing the sidelobes. Advantages of using passive beamformers
with overlapped subarrays include: (1) a dramatic reduction in the
number of control lines needed; and (2) much reduced electrical
power dissipation per chip.
Optical Beam-Steering Device Architectures
[0042] FIG. 1A shows an example lens-enabled integrated photonic
LIDAR system 100 with light propagating through each component of
the system 100, which is formed on a substrate 104. Light is
coupled into the system from a fiber 90 via an on-chip waveguide
102. The waveguide 102 is formed by a 200 nm thick and 1 .mu.m wide
Silicon Nitride (SiN) section encapsulated in SiO.sub.2 (an upper
SiO.sub.2 layer is omitted from the diagram). This fiber 90
connects to a tunable IR source 80 that emits light centered at a
wavelength of about 1550 nm. The light source 80 can be mated or
bonded to the substrate 102 or integrated onto the substrate 102.
There may also be a preamplifier coupled between the light source
80 and the fiber 90 or disposed on the substrate 104 in optical
communication with the waveguide 102 to amplify the input beam from
the light source 80.
[0043] The waveguide 102 guides the light from the IR source 80 to
a switch matrix 110 composed of Mach-Zehnder interferometers (MZIs)
112. The MZIs 112 can switch the input into any one of 2.sup.D
output ports 114, where D is the depth of the tree in the switch
matrix 110. The optical path length of at least one arm of each MZI
112 is controlled by an integrated thermo-optic (TO) phase shifter
(not shown; external electronic control lines are also omitted),
which allows the optical beam to be electronically switched between
two output ports 114.
[0044] The system 100 may also include semiconductor optical
amplifiers (SOAs; not shown) integrated on the substrate 104 before
or after the switch matrix 110. These SOAs can be turned on and
off, depending on whether or not light is propagating through them,
to reduce power consumption.
[0045] Each output from the switch matrix 110 feeds into a slab
waveguide 122 that is patterned to form the focal surface of a
wide-angle planar dielectric lens 120, also called an aplanatic
lens. This lens may obey the Abbe Sine condition and has a lens
shape designed for wide-angle steering. There are many different
shapes and structures that provide wide-angle steering, including a
basic parabolic shape for the lens. Other lens shapes are also
possible, including those for a bifocal lens (shown in FIG. 1B),
bootlace lens (shown in FIG. 1C), integrated Luneburg lens, Rotman
lens, or standard compound lens, such as an achromatic doublet.
[0046] The light 121 rapidly diffracts upon entering the slab
waveguide 122 until being collimated by the aplanatic lens 120. The
lens 120 is formed by a patterned PolySi layer 20 nm or 40 nm
thick. The collimated beam 121 propagates into a output coupler,
implemented here as a one-dimensional (1D) grating 130 that guides
the collimated beam 121 within the plane of the substrate and
scatters the collimated beam out of the plane of the substrate 104
and into free-space. The grating 130 is formed out of a PolySi
layer the same height as the aplanatic lens 120.
[0047] In other examples, the lens redirects the beam without
collimating it to a curved grating (instead of the straight grating
130 shown in FIG. 1A). The curved grating's curvature is selected
to collimate the beam redirected by the lens. In other words, the
lens and curved grating work together to produce a collimated
beam.
[0048] The MZI switch matrix 120 can be replaced by a 3 dB
splitting tree (not shown) that illuminates all of the input ports
114 in parallel. In this implementation, beam-steering of this
"fan-beam" is accomplished through frequency tuning. A separate
aperture with an array of detectors processes the LIDAR return. At
100 m, a raster scan for a megapixel scale sensor would take 1
second, based on time-of-flight--far too slow for most
applications--whereas parallelization can allow an acquisition on
the scale of milliseconds. The tradeoff is that the power
requirements are increased by a factor of N and more detector
hardware is required.
[0049] In other embodiments, the output coupler is implemented as a
two-dimensional (2D) photonic crystal instead of a 1D grating. For
instance, photonic crystal resonances based on bound states in the
continuum (BIC) can implement the out-of-plane steering. BICs are
infinite quality factor resonances arising due to interference
effects. For a broad range of neighboring wave vector points, the
quality factor is also very high, potentially enabling large scale
photonic structures with efficient and focused emission across a
wide range of directions.
[0050] One factor enabling improved performance of a 2D photonic
crystal relative to other output couplers is the rigorously optimal
radiation quality factor of BICs. BIC photonic crystal gratings can
have quality factors as high as 10.sup.5, enabling propagation over
the order of 10.sup.5 periods, or 10 cm, before significantly
attenuating. These can easily be used to not just make 50%
illuminated, 1 mm long gratings, but even 70% and 95% illuminated,
1 cm long gratings. Slight tuning of the structure can change the
quality factor and allow the output to be tapered for a more even
aperture power distribution and a smaller beamwidth in the far
field.
[0051] The BICs close to k=0 points, protected by symmetry mismatch
between the electromagnetic resonance and radiation continuum,
feature many additional desirable properties. The relatively flat
and homogeneous bands here allow rapid tuning rates of the emission
direction by slightly tuning the frequency of excitation. The
angular tuning rate can be estimated as a function of frequency
using the following simple relation
d.theta.=d.omega./.omega.*n.sub.group, where n.sub.group is the
group index of the band. For n.sub.group=3, this yields a standard
tuning rate of 0.1 degrees/nm at 1500 nm. Using a flat band with
n.sub.group as high as 10, the expected tuning rates are about 0.4
degrees/nm, allowing a potential reduction in the required
bandwidth and tuning range of on-chip devices by a factor of 2 or
more.
[0052] In addition, the radiation field can also be designed to be
highly asymmetric by breaking the top-down mirror symmetry of the
structure, so that light can be more efficiently collected into the
desired direction. This enables lower losses and fewer interference
effects from radiation into the substrate.
[0053] Finally, a common problem with arrays of dielectric grating
antennas in conventional optical beam-steering devices is their
sensitivity to coupling at wavelength scale pitches. 2D photonic
crystals do not exhibit this problem because they are designed for
the strong "coupling" regime: the entire structure is wavelength
scale.
[0054] This reduced sensitivity comes at a price. The in-plane
steering and out-of-plane steering are no longer separately
controlled by on-chip beamforming or frequency-tuning: there is a
mixture. Despite this, it is possible to map a given set of
frequency and port settings to a given set of beam angles with a
lookup table.
Non-Mechanical Optical Beam Steering with a Planar Dielectric
Lens
[0055] FIGS. 2A-2C illustrate non-mechanical beam steering for the
chip 100 shown in FIG. 1A. Non-mechanical beam steering is
implemented by two mechanisms. The first is port switching, which
changes the in-plane propagation direction of the beam as shown in
FIGS. 2A and 2B. The desired input port, and hence the desired
in-plane propagation direction, can be selected by setting the MZIs
112 in the optical switching matrix 110 to route the input beam.
Thus, there may be up to one resolvable in-plane angle for each
input port to the planar dielectric lens. Systems with 32, 100,
1000, or 10,000 input ports would have up to 32, 100, 1000, or
10,000 resolvable in-plane steering angles, respectively.
[0056] As depicted in FIGS. 2B and 2C, the wavelength roughly
controls the out-of-plane angle, that is, the angle between the
beam center and the z-axis. Thus, there may be up to one resolvable
out-of-plane angle for each wavelength resolvable by the output
coupler. For example, systems with gratings that can resolve 15,
50, 100, or 1000 angles would have up to 15, 50, 100, or 1000
resolvable out-of-plane steering angles, respectively.
[0057] These 2D beam steering mechanisms are similar to those of RF
Rotman lenses feeding arrays of patch antennas. The 3D directivity
patterns of the generated beams are depicted in each subfigure. The
precise mathematical relationship between the emission angles and
the analytic form of the directivity pattern are detailed
below.
[0058] FIGS. 3A and 3B show how the system can be used to transmit
and receive, respectively. Transmission works as described above:
exciting an input to the planar dielectric lens yields a plane wave
that propagates in a given direction within the plane of the lens,
and tuning the wavelength changes the propagation angle with
respect to the surface normal of the plane of the lens. Receiving
works in reverse: the grating collects incident light, and the lens
focuses the incident light on the input port associated with the
corresponding in-plane angle-of-arrival. The out-of-plane angle of
arrival corresponds to the angle with the strongest transmission,
which represents a direct "reflection" from the object being
interrogated. The coupler may be illuminated by light from other
angles, e.g., caused by scattering or indirect "reflections," but
this light generally is not properly phased-matched to the grating
and therefore does not efficiently couple into the grating.
[0059] To better understand the system's operation, consider an
ideally preforming aperture. In operation, an ideal implementation
of the planar dielectric lens generates a plane wave propagating at
a finite angle. The scattered light from the plane wave propagation
through the 1D grating forms the near-field of the radiation
pattern. Assuming that the plane wave emitted from the lens is
uniform, that the lens introduces negligible aberrations, and that
the lens and grating parameters are wavelength and angle
independent yields a simplified aperture pattern of the following
form:
A .function. ( x , y ) = { exp .function. ( - q .times. x ) .times.
exp .function. ( i .times. k 0 .times. u x .times. 0 .times. x )
.times. exp .function. ( i .times. k 0 .times. u y .times. 0
.times. y ) 0 .ltoreq. x .ltoreq. L , - W 2 + x .times. .times. tan
.times. ( .PHI. ' ) .ltoreq. y .ltoreq. W 2 + x .times. .times. tan
.function. ( .PHI. ' ) .times. 0 else ##EQU00001##
[0060] FIG. 4A shows this ideal near-field aperture pattern. The
pattern in FIG. 4A can be thought of as a parallelogram with
uniform amplitude in the y-direction, and exponentially decaying
amplitude in the x-direction, determined by the grating decay
parameter q where L is the length of the grating and W is the
width. The inclination of the parallelogram is determined by the
grating propagation angle .PHI.', which is derived below and is
close in magnitude to the propagation angle of the beam output from
the lens .PHI..sub.in
[0061] u.sub.x,0=sin(.PHI..sub.0)cos(.theta..sub.0) and
u.sub.y,0=sin (.PHI..sub.0)sin(.theta..sub.0) characterize the
direction of the emitted mode and can be calculated by tracking the
phase accumulated by the collimated rays emitted from the lens and
discretely sampling them at the grating teeth. As shown below,
these can write these as:
u y , 0 = n 1 .times. sin .function. ( .PHI. in ) .times. .times. u
x , 0 = 0 .times. k x , avg .function. ( .PHI. in ) k 0 - 2 .times.
.pi. .times. .times. m .LAMBDA. k 0 ( 1 ) ##EQU00002##
where k.sub.x,avg is the average k component in the grating,
n.sub.1 is the effective index of the TE slab mode in the lens, m
is the grating order, and .LAMBDA. is the grating period. The
function of the grating can be understood from this equation: it
allows phase matching to radiating modes through the addition of
the crystal momentum 2.pi.m/.LAMBDA..
[0062] Making the approximation
k.sub.x,avg.apprxeq.n.sub.effk.sub.0 cos(.PHI..sub.in), where
n.sub.eff is the average effective index of the gratings, makes it
possible to show that u.sub.x0 and u.sub.y0 satisfy an elliptical
equation:
[ u x , 0 + m .times. .lamda. .LAMBDA. n e .times. f .times. f ] 2
+ [ u y , 0 n 1 ] 2 = 1 ( 2 ) ##EQU00003##
This elliptical equation has a simple physical interpretation.
Switching ports in-plane takes us along an elliptical arc in
x.sub.x,0 and u.sub.y,0 space, while tuning the wavelength 2 tunes
this arc forward and backwards as depicted in FIG. 6B (described
below).
[0063] Analytically calculating the directivity, which
characterizes the far-field distribution of radiation, yields:
D .function. ( .DELTA. .times. u x , .DELTA. .times. u y ) = W
.times. k 0 2 .times. cos .function. ( .theta. 0 ) .pi. .times.
.times. q .function. ( 1 - exp .function. ( - 2 .times. q .times. L
) ) .times. sin .times. .times. c 2 .times. ( W 2 .times. k 0
.times. .DELTA. .times. u y ) 1 + k 0 2 q 2 .times. ( .DELTA.
.times. u x + tan .function. ( .PHI. ' ) .times. .DELTA. .times. u
y ) 2 .times. ( 1 - 2 .times. cos .function. ( k 0 .times. L
.function. ( .DELTA. .times. u x + tan .function. ( .PHI. ' )
.times. .DELTA. .times. u y ) ) .times. exp .function. ( - q
.times. L ) + exp .function. ( - 2 .times. q .times. L ) ) ( 3 )
##EQU00004##
where .DELTA.u.sub.x=u.sub.x-u.sub.x,0 and
.DELTA.u.sub.y=u.sub.y-u.sub.y,0.
[0064] FIG. 4B is a plot of the computed far-field directivity and
far-field beam angles given in Eq. (3) for the ideal aperture given
in FIG. 4A. The peaks to the left and right of the main beam are
known as sidelobes and originate from the sinc factor in the
directivity function. There are several recognizable features to
this function, such as the sinc from the rectangular aperture and
the 1/(k.sub.0.sup.2+q.sup.2) from the exponential decay. The tan
.PHI.' components introduce a "shear" into the beam spots and come
directly from the tilted aperture pattern.
[0065] This result can be used to calculate estimates for the
number of resolvable points for port switching and wavelength
steering. Specifically, for wavelength tuning, the result is:
N w .times. a .times. velength .apprxeq. Q .times. n e .times. f
.times. f .times. .DELTA. .times. .lamda. .lamda. 0 .times. v g c +
1 ( 4 ) ##EQU00005##
where Q is the quality factor of the grating, v.sub.g is the group
velocity of propagation in the grating, and .DELTA..lamda. is the
bandwidth. This expression exactly resembles what would be
extracted from other phase-shifter based architectures which rely
on frequency tuning for beam-steering in one direction. The number
of resolvable points for steering in plane is approximately:
N in - plane .apprxeq. 2 .times. D p .times. e .times. a .times. k
3 = 2 .times. .pi. .times. W 3 .times. .lamda. ( 5 )
##EQU00006##
Planar Dielectric Lens Design
[0066] The wide-angle planar dielectric lens has a shape selected
to satisfy the Abbe Sine condition, which eliminates the Coma
aberration. Lenses designed this way tend to have good off-angle
performance to .+-.20.degree. or 30.degree.. In practice, this
quantity can translate to a field of view of 80.degree. or more in
.PHI..sub.0. The lens design depends on the focal length, lens
thickness, and lens index (the ratio of effective indices of a
transverse electric (TE) mode in a SiN slab (n.sub.2) to a SiN slab
with a layer of PolySi (n.sub.1)). After creating the lens, the
focal plane can be identified by conducting ray-tracing through the
lens and optimizing the feed position and angle based on maximizing
the 2D directivity from the 1D aperture pattern computed from
ray-tracing.
[0067] Following this, ray tracing is done through the grating to
compute the full 3D directivity for several optimized port
locations and angles. The aperture pattern can be extracted from
ray-tracing through the grating in the following way:
A .function. ( x , y ) = [ n = 0 N r .times. a .times. y .times. s
.times. m = 0 N grat .times. P n , m .times. .delta. .function. ( x
- m .times. .LAMBDA. ) .times. .delta. .function. ( y - y n , m )
.times. exp .function. ( - q .times. x ) .times. exp .function. ( i
.times. .PHI. n , m ) ( 6 ) ##EQU00007##
where the ray amplitudes P.sub.n,m and accumulated ray phases
.PHI..sub.n,m are discretely sampled for all N.sub.ray by
N.sub.grat, ray-grating intersections at [x.sub.n,m, y.sub.n,m].
The physical interpretation of this is that each ray-grating
intersection acts as a point radiation source driven by the
traveling wave (see below). An artificially "added" amplitude decay
of exp(-qx) accounts for the grating radiation as the rays
propagate. The power associated with a given ray-grating
intersection is calculated from the feed power based on
conservation arguments:
P.sub.feed(.PHI.)d.PHI.=P.sub.n,m(y)d.sub.n,m
[0068] FIGS. 5A-5C illustrate a full ray-tracing calculation and a
2D aperture pattern extracted by this method for lens-enabled
chip-scale LIDAR generated with .lamda..sub.0=1.55 .mu.m, q=0.025
.mu.m.sup.-1, .LAMBDA.=700 nm, duty cycle=0.1, feed beamwidth of
15.degree., and effective indices n.sub.1=1.39 and n.sub.2=1.96. In
FIG. 5A, ray-tracing simulations are used to determine the optimal
port position and relative angle. These rays are traced through the
grating and form an aperture pattern. The Fourier transform of the
aperture gives the Far-field pattern. Numerical details of
effective index calculations, port phase center, and feed patterns
are detailed below.
[0069] FIG. 5B is a heat plot showing far-field beam spots in
u.sub.x and u.sub.y space. The location of these ports is governed
by the equations above, where the beams along the elliptical curve
are generated from port switching, while the points formed from
translating the ellipse to the right and left correspond to
frequency tuning over .+-.50 nm around .lamda..sub.0=1.55 .mu.m.
FIG. 5C shows three-dimensional (3D) beam patterns corresponding to
those in FIG. 5B. The different shadings indicate the different
wavelengths used to generate the beam. The drooping effect of the
beams as they turn off-axis is caused by increasing in-plane
momentum.
[0070] Ray-tracing through the grating is valid in the regime where
the grating teeth individually cause low radiation loss and small
incoherent reflections (i.e., the excitation frequency is far from
the Bragg bandgap). When correct, this method is useful because it
can be used to compute the aperture pattern quickly for a large
many-wavelength structure while including the effects of lens
aberrations and a nonuniform power distribution, two features which
would be difficult to model analytically, and very costly to
simulate through 2D or 3D finite-difference time-domain (FDTD)
techniques.
[0071] Following standard RADAR design procedures, once a set of
far-field 3D directivity patterns are calculated, new ports are
placed to overlap the gain at each port by 3 dB to provide suitable
coverage in the field of view. To confirm the successful operation
of this design, the aperture patterns for multiple wavelengths
between 1500 and 1600 nm are calculated for all ports. The
wavelength dependence of the effective indices and the grating
decay factor q are included. Directivity patterns are plotted in
u.sub.x, u.sub.y space for a range of wavelengths in FIG. 5B, where
the 3 dB overlapped ports lie along an elliptical arc, and where
the arc is translated forward and backward by tuning the
wavelength. These same beams are plotted in 3D in FIG. 5C. Note
that beams towards the edge of the field of view tend to
"fall-into" the device plane because of increasing in-plane
momentum (see below).
[0072] Ray-tracing is one method that can be used to design an
optical beam-steering chip with a planar dielectric lens. The
parameters used for this method, such as the port phase centers,
feed beam width, grating decay length, and the effective indices,
can be extracted from other calculations. In addition, many other
simulations may be undertaken to validate the assumptions of our
ray-tracing computations to account for second order effects.
Finally, the outcome of the ray-tracing calculations may be
compared to the analytically predicted directivity functions and
beam directions to assess the performance and validity. Once a
design is validated, cadence layouts of the necessary components
can be generated automatically and verified to ensure they satisfy
design rule checks based on fabrication limitations and other
physics-based constraints.
Performance of Optical Beam Steering with a Planar Dielectric
Lens
[0073] The optical beam steering architecture shown in FIG. 1A has
several advantages over phase-shifter based solutions. RF lenses
were developed in part to reduce or minimize the use of phase
shifters, which are expensive, lossy, complicated, power hungry and
bulky. Some of the same considerations apply here: thermo-optic
phase shifters are power hungry components, typically using 10 mW
or more to achieve a 7 phase shift. To steer a beam with a
one-thousand pixel device, it would be necessary to actuate on the
order of 1000 phase shifters spread out in a 1 mm aperture as shown
in FIG. 6A. This is because, in this architecture, power is
uniformly fed to all output antenna elements through a 3 dB
splitting tree and the thermal phase shifters are actively cohered
to implement in-plane beam steering over 1000 resolvable points.
Actuating this many phase shifters would dissipate about 10
Watts.
[0074] Now consider the system of FIG. 1A with thermo-optic phase
shifters to operate the MZI switching matrix with N input ports to
the planar dielectric lens. The lens-based approach achieves N
resolvable points in-plane by switching with an MZI tree switching
matrix between N ports of a dielectric lens feed. The power
requirements for the MZI tree switching matrix in FIG. 6B scale
like log.sub.2 N as compared to N for the architecture in FIG. 6A
because only MZIs associated with the desired optical signal path
need to be activated (i.e., one MZI for each level of the switching
matrix hierarchy); the rest can be "off" and draw no power.
Consequently a lens-based device with 1000 resolvable points in
plane dissipates 100 times less power for in-plane steering
compared to the conventional phase-shifter based approach shown in
FIG. 6A. Thus, for lens-enabled LIDAR, the power budget is
dominated by the optical signal generation, whereas for the
phase-shifter architectures, it scales primarily with the feed
size.
[0075] Most practical phase-shifter approaches require active
feedback to maintain beam coherence because thermal cross-talk
causes changes in the path length of neighboring waveguides. This
means either making a measurement of the relative phases on chip
through lenses and detectors or measuring the beam in free space
through an IR camera to provide feedback. But a lens-based device
does not to actively cohere thousands of elements: it can use
"binary"-like switching to route the light to the appropriate port,
which is a simpler control problem. This means that the beams
emitted by the input ports to the lens can have arbitrary relative
phases. Lower power consumption additionally makes thermal
fluctuations less severe.
[0076] Using a solid 1D grating reduces or eliminates grating lobes
or high sidelobes that plague conventional optical phased arrays.
This is at the cost of not being able to "constrain" the ray path
to be in the forward direction and may result in having to use more
material for the grating coupler, hence the triangle shape of the
grating feed.
[0077] There is an alternate realization of this system, outlined
below, which does not use TO phase shifters and parallelizes the
in-plane ports. This architecture parallelizes one scanning
direction, as is commonly used in most commercial LIDARs to
increase scanning speed. This modification is not possible with the
conventional phase-shifter based approach.
[0078] Another advantage of a lens-based architecture is the
ability to use alternate material systems. One reason for using Si
for phased-array designs is its large TO coefficient, which makes
for lower power phase shifters. However, the maximum IR power a
single Si waveguide can carry is 50 mW, which significantly limits
the LIDAR range. SiN has much better properties in the IR and can
take the order of 10 W through a single waveguide. However, noting
that the power required to operate the phase shifter goes like
d .times. T d .times. n .times. .sigma. , where .times. .times. d
.times. n d .times. T ##EQU00008##
is the thermo-optic coefficient and u is the conductivity, phase
shifters on the SiN platform may use at least three times more
power than their Si equivalents. This would exacerbate the power
budget and control problems described above for any phase-shifter
approach based on SiN. The lens-enabled design can still benefit
from using SiN, and greatly improve the potential range, because
the feed power is practically negligible.
[0079] No architecture is perfect, and there are several
non-idealities which can alter the above story for our lens-based
solution. The first is the nonuniform field of view of the device,
which may cause problems for some applications. Another concern is
scaling the number of resolvable points to thousands of pixels in
each scanning direction. Although it is simple to ray-trace a lens
which would support up to a thousand resolvable points for in-plane
scanning, implementing such a lens in practice becomes more and
more difficult because the required fabrication tolerances scale as
1/N. An additional concern is the impact of lens aberrations on the
directivity degradation for the full aperture. Although it was
captured by ray-tracing, it was not rigorously modeled to determine
the required tolerances and behavior for high Q gratings.
LIDARs with Lens-Enable Optical Beamformers
[0080] FIG. 7 shows a lidar system 700 that includes a
lens-enabled, nonmechanical beam-forming system. The lidar system
100 includes a tunable IR light source 780 that emits a tunable IR
beam. An optical preamplifier 782 optically coupled to the tunable
IR light source 780 amplifies the tunable IR beam, which is coupled
to a 1-to-128 MZI switch matrix 710 via a 3 dB coupler 784 or
coupler. The 3 dB coupler picks off a portion of the amplified beam
for heterodyne detection of the received beam with a detector 790.
Signal processing electronics 792 coupled to the detector 790
process the received signal.
[0081] The switch matrix 710 is fabricated on a SiN platform 708
that is integrated with an InP platform 706 that supports a
slab-coupled optical waveguide amplifier (SCOWA) array 712. This
InP platrom is also integrated with another SiN platform 704 that
includes a passive beamforming chip 720 with both a planar
dielectric lens and an output coupler. The lens may be a 20 nm or
40 nm thick PolySi lenses, and output coupling gratings may be 10
nm, 20 nm, or 40 nm thick. The gratings support up to 300
resolvable points from wavelength tuning for the 10 nm variants
over a 100 nm bandwidth. Because of fabrication constraints, the
grating PolySi height may be the same as the lens height. This can
result in tradeoffs because thicker gratings had lower quality
factors, but thicker lenses have a better index contrast and can
support more resolvable points. The switching matrix 710 is
actuated by an off-chip digital controller 770.
[0082] In other examples, the system may be completely integrated.
For instance, the tunable source, detectors, and electronics may be
integrated on the chip as the switch matrix, lens, and output
coupler. Bringing all of these technologies together compactly,
cheaply, and robustly yields a new sensor capable of supporting the
next generation of autonomous machines.
[0083] FIG. 8 shows a lidar 800 with a tunable on-chip source 880,
which can be modulated with a microwave chirp. The chirped carrier
travels through a preamplifier 882 and a directional coupler 884
and is split with a 1-to-100 passive power splitter 810 to feed an
array of 100 20 dB amplifiers 812 on a 10 .mu.m pitch. Waveguides
from these amplifiers 812 are tapered to the edge of the
beamforming aplanatic lens 820 implemented with a SiN slab. These
amplifiers 812 are turned on and off with a digital controller 870
to the steer the output beam as explained below. The lens 820 ends
on a flat surface in front of a 1 mm.sup.2, 2D photonic crystal
(PhC) 830 that serves as the aperture and grating coupler. The
reflected return from a target comes back through the same aperture
and is beat against a local oscillator (provided by the tunable
source 880) and undergoes balanced detection with a heterodyne
detector 890 coupled to signal processing electronics 892.
[0084] FIG. 9 shows an integrated optical beamforming system 900
that scales the basic design to add functionality for N
independently controllable beams. A seed from a tunable source 980
is split into 16 waveguides with a 1 by 16 power splitter 910a.
Each of these 16 power splitter outputs feeds into its own
preamplifier 982 and heterodyne detection unit 990, which are
coupled to signal processing electronics 992. In turn, these feed
16 separate sections of the array with a power splitter 910b
coupled to an array of 30 dB amplifiers 912 (here, 128 amplifiers)
actuated by digital control electronics 970. A planar dielectric
lens 920 collimates the outputs of the amplifier array 912 for
diffraction by an output coupler 930, such as a 1D grating or 2D
photonic crystal. Here, 128 independent beams with scanning ranges
limited to 128 non-overlapping subsectors of the far field can be
realized by turning on and off amplifiers connected to a given
heterodyne detector.
[0085] The unit cells 800 and 900 shown in FIGS. 8 and 9 can be
modified for tiling an M by N array as in FIG. 10A, which shows a
unit cell 900 modified to receive a local oscillator 1080
distributed among the tiles with waveguides. This seed 1080 is
amplified with a preamplifier 1082 and serves as the source for the
tile. Before seeding the amplifiers 982, the source phase is
changed with a thermal phase shifter 1084. After the output lens
920, the neighboring subarrays are overlapped to suppress the
sidelobes of the tiled system. At the end of the grating 930, the
output light is sampled and undergoes balanced detection using a
balanced detector 1032 with the signal from a neighboring tile to
provide feedback to cohere the tiles.
[0086] FIG. 10B shows the tiling of the unit cells to form larger
apertures. The cells are flipped on one side to expand the
effective length of the PhC gratings.
[0087] Tunable Light Source and Preamplifier
[0088] As explained in greater detail below, the wavelength of the
light source controls the out-of plane angle of the optical beam.
Typical grating antennas show steering at the rate of 0.1-0.2
degrees/nm. For instance, a light source with about 100 nm of
tuning range provides a 12.degree. to 16.degree. field of regard.
2D photonic crystal gratings, discussed below, may have enhanced
steering rates. In addition, the laser provides seed power for
driving one or more optical amplifiers.
[0089] The power requirements for the laser source and optional
preamplifier can be determined by working through the signal chain
for the complete system. Consider a desired output of 500
mW/cm.sup.2 for a system with 100 input ports to the lens. This
corresponds to 5 mW from a 1 mm.sup.2 aperture. If there are 6 dB
losses in the grating and lens, the input to the lens should be
about 20 mW. If the system includes an amplifier that provides 20
dB gain (for this input port), the input power to the channel
should be 0.2 mW. To obtain 0.2 mW from a 1-to-128 splitter
requires 20 mW ignoring losses. This translates to 80 mW from the
light source and preamplifier, taking into account 6 dB losses from
the splitter and coupler. Assuming a nominal 10% efficiency, this
preamplifier would need 800 mW of electrical power for operation
and 12 dB of gain given a 5 mW source.
[0090] These specifications for an on-chip source are reasonable. A
recent work demonstrated a Vernier ring laser with 5.5 mW output
power and a 41 nm tuning range. A thermal phase shifter allows
tuning which can be adjusted on roughly 1 .mu.s timescales, giving
a sufficiently fast point-to-point sweep time for all realizations.
This source may also be directly modulated with an RF chirp with a
bandwidth of up to 9 GHz through plasma dispersion. RF modulation
can also be implemented with an integrated single sideband
modulator.
[0091] 1-to-N Optical Splitter
[0092] The system 100 in FIG. 1A includes a 1-to-N optical matrix
110. In the embodiments shown in FIGS. 8, 9 and 10A, the switch
matrix is replaced by one or two passive 1-to-N splitter trees
coupled to an array of N semiconductor amplifier switches
(discussed below). The splitter tree is created using a binary tree
of 50:50 splitters fabricated in SiN. Each 50:50 splitter is an
adiabatic 3 dB-coupler that is about 100-200 .mu.m long and about
10 .mu.m wide. Thus a 1-to-128 splitter tree has 7 levels of
splitting and is about 1 mm by 1 mm. A 1-to-1024 splitter tree has
10 levels of splitting and is about 1.5 mm by 2 mm. Each splitter
in the splitter tree has a very low excess loss of about 0.1 dB.
Thus, the entire splitter tree has a total excess loss of about 1
dB is expected for the entire tree. Because the splitting tree is
before the amplifier bank and each amplifier is either off or on
and saturated, the system is not sensitive to minor variations in
splitting ratio. This is in contrast to schemes in which splitters
directly feed the grating coupler antennas. Adiabatic splitters are
chosen to reduce or minimize back reflections and scattered light
(which can be problematic in multimode interference (MMI) 3 dB
splitters and Y-couplers) and to allow for uniform spitting over a
wide optical bandwidth (>40 nm, which is not achieved using
standard directional couplers).
[0093] For multiple beam and tiled realizations, a single laser
source may feed into an array of preamplifiers. Consider a tiled
realization with 16 tiles and a desired output of 500 mW from a 1
cm.sup.2 aperture. If there are 16 simultaneous beams, the
preamplifier array should provide a 125 mW output keeping the same
losses as above. For preamplifier with a 30 dB gain, the input
power should be at least 0.125 mW and the output show be at least
30 mW from one of the 20 dB preamplifier units before splitting.
The electrical power consumption for the preamplifiers for multiple
beam and tiled realizations will be 4.8 W for a single chip
assuming 10% efficiency.
[0094] The source and preamplifier devices may be created in InP
and picked and placed onto a passive SiN chip containing the
splitters, lens, and grating. For the single tile realization,
there may be one combined source, preamplifier, and detection InP
chip. For other realizations, one chip may have an array of
preamplifiers and detectors and another chip may have the
source.
[0095] Semiconductor Optical Amplifier (SOA) Switches
[0096] A passive splitter tree can be coupled to an array of SOA
switches as shown in FIGS. 8, 9, and 10A, with each SOA switch
amplifying or attenuating the input signal as a function of
externally applied power (control signal). SOAs have fast
(.about.10 ns) switching speeds compared to thermal phase shifters
(.about.1 .mu.s). SOAs having small-signal gain of 20 dB (e.g., in
the single-tile realization in FIG. 8) and 30 dB (FIGS. 9 and 10A)
boost the seed laser output to provide the desired output power.
For instance, the SOA output power may be 20 mW (FIG. 9), 125 mW
(FIGS. 9 and 10A), or any other suitable power level. SOAs can be
integrated with the SiN beamformer tiles using hybrid flip-chip
integration. The SOAs may be of any suitable type, including
conventional or low-confinement. Sample SOA specifications are
shown in Table 1.
TABLE-US-00001 TABLE 1 Specifications for SOAs Length Realization
Gain (dB) Psat (mW) Efficiency (cm) Approach Single Tile 20 20 10%
<0.1 cm Conventional SOA ST, MB 30 125 20% 0.32 cm High-
confinement SCOWA Arrayed Tiles 30 125 30% 0.32 cm High-
confinement SCOWA
[0097] For example, each SOA may be implemented as the slab-coupled
optical waveguide amplifier (SCOWA) developed by Lincoln Laboratory
(LL). At 1550-nm wavelength, a 1 cm long InP-based SCOWA having
small-signal gain of 30 dB and saturation output power of 400 mW
has been demonstrated. By increasing the SCOWA confinement factor
appropriately, a SOA having 30 dB gain and 125 mW output power
should be realizable with a 0.32 cm length. In addition to
providing enough gain and output power for this application, SCOWAs
also have a very large transverse optical mode (e.g., about
5.times.5 .mu.m), which increases the alignment tolerance when
using flip-chip integration to couple SOA and SiN chips. The
flip-chip coupling loss between a SCOWA and a SiP waveguide with
the appropriate mode-size converter is about 0.5 dB to 1 dB.
[0098] For an array of conventional SOAs or SCOWAs, the minimum
pitch is about 10 .mu.m to avoid optical coupling between
neighboring devices. This small pitch can be thermally managed as
only one SCOWA is on at a time during operation. Therefore, arrays
of 100 SCOWAs (single tile) and of 1000 SCOWAs (other realizations)
have footprints of 0.1.times.0.1 cm and 1.times.0.32 cm.
respectively.
[0099] Since these SOAs amplify to 20 mW for the single-tile
realization in FIG. 8, the design uses 200 mW electrical operating
power assuming 10% efficiency. Given that the on-chip source uses
800 mW, the total power requirements for a feasible realization can
be limited to 1 W. This gives a dissipated power density of
approximately 17 W/cm.sup.2. For the single-tile realization with
multiple beams shown in FIG. 9, given that 16 of the 125 mW
amplifiers are on simultaneously and assuming 20% efficiency, the
amplifier consumes 10 W. With 4.8 W for the preamplifier/source
unit, the dissipated power density comes to 6 W/cm.sup.2 for the
single-tile with multiple beam realization of FIG. 9 and less than
8 W/cm.sup.2 for the arrayed tile realization in FIG. 10A.
[0100] Given an operating power of 1 W, and an output power of 5
mW, a conservative estimate for the wall plug efficiency for the
realization of FIG. 8 is about 0.5%.
[0101] For the realization of FIG. 9, the operating power may be
about 15 W, including performance increases for the amplifier bank.
With a 500 mW optical output power, this puts the wall-plug
efficiency at 3%. Improving to a 30% efficient amplifier bank
yields 5% wall-plug efficiency for the realization of FIG. 10A.
[0102] The preamplifier architecture employed in the realizations
of FIGS. 9 and 10A makes it easier to increase the wall-plug
efficiency. The benefit of using an array of preamplifiers mid-way
through the splitting tree can be understood in the following way:
it is most efficient to turn on and off individual sources at each
input port of the beamforming lens. This means good solutions will
reduce or minimize the power of the preamplifier stage, increase or
maximize the gain of the final amplifier bank, and turn off any
unused amplifiers. In this case, placing an array of preamplifiers
after the 1-to-16 splitter 910a of FIGS. 9 and 10A reduces the
power requirements for the preamplifier stage by a factor of 16.
Consequently, whereas the preamplifier and source uses four times
more power than the final amplifier array for the realization of
FIG. 8, for the realizations of FIGS. 9 and 10A it uses four times
less, assuming equal efficiency of the components.
[0103] To create a 1 mm aperture, a conventional optical phased
array design needs on the order of 1000 thermal phase shifters.
With an operating power of 20 mW/phase shifter, such a system would
consume 20 W. This is an order of magnitude more power than the
single-tile realization shown in FIG. 8. The favorable power
scaling of the inventive approach extends to other beamforming
architectures.
[0104] Planar Dielectric Lens
[0105] The original microwave literature going back to 1946
explored the use of lenses for beam-steering applications. That
literature was chiefly concerned with mechanical displacement of
the feed to obtain wide-angle and diffraction-limited beams. This
specific approach was even implemented with MEMS and microlenses
for small steering angles. Over time, many mathematical techniques
were developed to numerically calculate the best lens shape to
minimize aberrations which would otherwise quickly degrade the beam
quality with increased steering angle. Specifically lenses with
wide-angle steering of -40 to +40 degrees can be developed by
numerically calculating a lens that satisfies a form of the Abbe
sine condition.
[0106] Additional approaches to create lenses with similar wide
angle ranges include bifocal and multi-focal lenses, which use
additional degrees of freedom to create structures which have
multiple perfect focal points in the imaging surface. The Rotman
lens is one such lens which utilizes delay lines to create three
focal points, one on-axis and two off-axis, for wide-angle
steering. Graded index lenses such as the Luneburg lens allow for
theoretically the widest angle steering possible by being
spherically symmetric. Beyond developing such a rich variety of
lenses, the microwave literature also explored many techniques for
optimally feeding the lenses, minimizing reflections, shaping the
feed end aperture field patterns, and dealing with a myriad of
other technical problems which may be relevant to our effort.
[0107] The planar dielectric lens can be implemented using any one
of a variety of designs. For example, it may be a dielectric slab
lens with a single perfect focal point in the imaging surface and
numerically designed to satisfy the Abbe sine condition. Fulfilling
the Abbe sine condition gives near diffraction limited performance
to up to .+-.40 degrees. For tiled realizations, alternate lenses,
such as the Rotman lens or the Luneburg lens, may be employed to
obtain up to 110 degrees or 180 degrees of in-plane beam steering,
respectively.
[0108] There are several approaches for implementing the lens in
integrated photonics, such as changing the height of the slab,
patterning an additional layer, doping, and varying the density of
subwavelength holes. These approaches have been used to implement
GRIN lenses, such as the Luneburg lens, on chip. For one
implementation, a thin layer of polysilicon can be patterned on a
silicon nitride slab. The high index of polysilicon compared to
silicon nitride creates a high effective index contrast thereby
increasing the focusing power of the lens. Adiabatically tapering
the height of the lens can reduce the radiation losses at the
interface of the slab with the lens.
[0109] Coupler
[0110] To use the same antenna for transmit and receive, the
antenna should capture backward-propagating return power. Off-chip,
an optical circulator would direct the backward propagating signal
to the receiver while providing good transmit/receive (T/R)
isolation. However, the magneto-optical materials used in such
reciprocity breaking devices are difficult to integrate on-chip.
Instead, a simple adiabatic 3 dB splitter sends half the backward
propagating power to the receiver. The low loss and large optical
bandwidths of such splitters should limit the performance penalty
to the 3 dB loss due to the splitting. This effect is especially
small in the transmit direction as the splitter is located before
the amplifier bank. Furthermore, since both the transmitter and
receiver ports are located on the same side of the device, decent
isolation of the receiver from the transmitter is provided. More
sophisticated possibilities for transmitter-receiver isolation are
also possible, including T/R switching or non-reciprocity from
modulation.
[0111] Heterodyne Detection
[0112] The heterodyne detection shown in FIGS. 8, 9, and 10A can be
performed with two balanced InP detectors. In operation, these
detectors record the heterodyne beat signal between the chirped
source and the chirped return. The use of balanced detectors allows
for changes in the amplitude of the source and return to be
decoupled from changes in the offset frequency, giving a more
robust measurement of round-trip time.
[0113] The outputs of the balanced detectors provide in-phase and
quadrature (I/Q) signals with an intermediate frequency (IF)
bandwidth determined by the time-bandwidth product required from
the transmitted linear-FM waveform. The outputs of the I/Q
detectors can be processed to create LIDAR imaging products.
[0114] Consider the simple example of a LIDAR on an autonomous
vehicle. This LIDAR uses a 1 GHz linearly frequency-modulated (LFM)
chirp over a 10 .mu.s period (1500 m range gate) in a repeating
sawtooth waveform. Stretched-pulse processing reduces the speed and
power consumption requirements for the IF analog-to-digital
converter (ADC) in the signal processing electronics on each
receive channel. At zero time lag between a target and the local
oscillator (LO), the IF frequency is direct current (dc; 0 Hz). For
a 0.2 .mu.s (30 m) target-range displacement, the IF frequency is
20 MHz. An IF ADC with 50 Msamples/sec and a few bits of dynamic
range can easily detect the displaced target and accurately
determine its range with an uncertainty of .about.15 cm (0.5*c/1
GHz). Such a compact circuit can be implemented in a 65 nm
complementary metal-oxide-semiconductor (CMOS) process.
[0115] Overlapped Subarrays
[0116] Tiling creates a larger effective aperture as depicted in
FIGS. 10A and 10B. Even if neighboring tiles are properly cohered,
there may be many narrowly spaced sidelobes in the far-field
because of the large distance between the center of each aperture.
At the same time, this comb of far-field sidelobes may be modulated
in magnitude by the far-field of the subarray pattern itself.
Engineering the subarray pattern to resemble a sinc function in the
near-field yields a box-like pattern in the far-field that
suppresses the sidelobes. It is possible to scan N beamwidths
within this box, where N is the number of subarray elements. For
example, in a tiled implementation, exciting 1 of 1,000 ports for
each tile and coherently phasing 10 neighboring tiles yields 10,000
resolvable points.
[0117] To produce a sinc pattern in the near-field, there are
several approaches to overlap and delay parts of the beam. One
strategy is to use multiple waveguide layers to route light from
the output grating of one tile to another to form the larger
subarray pattern. This arrangement can also work with a single
layer of waveguides by utilizing low-cross talk direct waveguide
crossings. Another approach uses an array of wedge-shaped
microlenses or photonic crystals. One implementation includes
super-collimating photonic crystals to keep the main part of the
beam going straight and defect waveguides to delay and route light
to neighboring tiles.
Analysis of Optical Beam Steering with a Planar Dielectric Lens
[0118] The following analysis is intended to elucidate operation of
an optical beam steering device with a planar dielectric lens. It
is not intended to limit the scope of the claims, nor is intended
to wed such a device to particular mode or theory of operation.
[0119] Far-Field Angles
[0120] The aperture phase is determined by the initial ray
directions, the grating parameters, and the wavelength. Since we
are propagating through a straight grating, the plane wave k.sub.y
generated by the lens feed system will be conserved, so
k.sub.y,avg=k.sub.y. k.sub.x on the other hand, will be more
complicated because it changes at each step of the grating.
Assuming an initial in-plane angle of .PHI..sub.in, an index of the
starting medium n.sub.1, an index of the steps n.sub.2 and a step
duty cycle d, we find that k.sub.y and k.sub.x,avg are given by the
following:
k y = n 1 .times. k 0 .times. sin .function. ( .PHI. in ) .times.
.times. k x , avg = d .times. n 1 .times. k 0 .times. cos
.function. ( .PHI. in ) + ( 1 - d ) .times. n 2 .times. k 0 .times.
( 1 - n 1 2 n 2 2 .times. sin 2 .function. ( .PHI. in ) ) ( 7 )
##EQU00009##
The effective indices for the grating n.sub.1 and n.sub.2 are also
functions of the wavelength. To compute the emission angle of this
aperture, we perform phase matching between these wavevectors and
those of a free-space plane wave with {right arrow over
(k)}=k.sub.0[sin(.theta..sub.0)cos(.PHI..sub.0),sin(.theta..sub.0)sin(.PH-
I..sub.0),cos(.theta..sub.0)].
k 0 .times. sin .function. ( .theta. 0 ) .times. sin .function. (
.PHI. 0 ) = n 1 .times. k 0 .times. sin .function. ( .PHI. in )
.times. .times. k 0 .times. sin .function. ( .theta. 0 ) .times.
cos .function. ( .PHI. 0 ) = k x , avg .function. ( .PHI. in ) - 2
.times. .pi. .times. .times. m .LAMBDA. ( 8 ) ##EQU00010##
[0121] Here we have subtracted a crystal momentum 2.pi.m/.LAMBDA.,
which originates from the discrete and periodic sampling
implemented by the scattering from each grating step. We can
rearrange this to derive the following expressions for the
far-field angles:
.theta. 0 = ( ( k avg , x .function. ( .PHI. in ) - 2 .times. .pi.
.times. .times. m .LAMBDA. ) 2 + ( n 1 .times. k 0 .times. sin
.function. ( .PHI. in ) ) 2 k 0 ) .times. .times. .PHI. 0 = ( n 1
.times. k 0 .times. sin .function. ( .PHI. in ) k avg , x
.function. ( .PHI. in ) - 2 .times. .pi. .times. .times. m .LAMBDA.
) ( 9 ) ##EQU00011##
We want to understand how the far-field angles depend on the
in-plane angle .PHI..sub.in and the wavelength .lamda.. We can
identify that .PHI..sub.0 will be significantly greater than
.PHI..sub.in. This results from the grating momentum
2.pi.m/.LAMBDA. being subtracted from k.sub.avg,x in the
denominator. This means that relatively small variations in the
input angle will greatly change the output in-plane angle
.PHI..sub.0, sweeping it across the field-of-view. This feature
ultimately allows us to use the lens in a small angle, aplanatic
regime.
[0122] As we sweep .PHI..sub.in we also expect variations in
.theta..sub.0. By examining (9), we see that the argument of the
arcsine term seems to increase with .PHI..sub.in, and ultimately
exceed 1, confining the beam in-plane. We can derive this cutoff
condition (where .theta..sub.0=0) more precisely by taking
k.sub.avg,x(.PHI..sub.in).apprxeq.n.sub.effk.sub.0
cos(.PHI..sub.in), where n.sub.eff=dn.sub.1+(1-d)n.sub.2. Even
though n.sub.eff is not truly constant, and its variations
significantly effect the far-field angles, qualitatively this
description holds. We find that the cutoff angle .PHI..sub.cut
satisfies:
1 = [ n e .times. f .times. f .times. cos .function. ( .PHI. cut )
- .lamda. .LAMBDA. ] 2 + n 1 2 .function. [ 1 - cos .function. (
.PHI. cut ) 2 ] ( 10 ) ##EQU00012##
[0123] This can be easily rearranged into a quadratic equation and
solved for .PHI..sub.cut. To get more intuition into the behavior
of this angle, we examine the case of normal (or broadside)
emission at .PHI..sub.in=0 and approximate
n.sub.1.apprxeq.n.sub.eff. Working this out we find:
cos .function. ( .PHI. in ) .apprxeq. 1 - 1 2 .times. n e .times. f
.times. f 2 ( 11 ) ##EQU00013##
where tan(.PHI..sub.0).apprxeq.-n.sub.eff/2 gives the corresponding
.PHI..sub.0 at this point. It makes sense that the magnitude of the
index will control the .PHI..sub.in because it determines how
rapidly k.sub.y increases as we move off axis. Overall, we can
envision how {right arrow over (k)} evolves as a function of
.PHI..sub.in: starting from emission normal to the surface, as we
adjust .PHI..sub.in away from 0, k turns rapidly to one side and
falls into the plane.
[0124] We can further visualize this trajectory by rearranging (8).
Taking u.sub.x,0=sin(.theta..sub.0)cos(.PHI..sub.0) and
u.sub.y,0=sin(.theta..sub.0)sin(.PHI..sub.0), we can manipulate (8)
to find:
[ u x , 0 + m .times. .times. .lamda. .LAMBDA. n e .times. f
.times. f ] 2 + [ u y , 0 n 1 ] 2 = 1 ( 12 ) ##EQU00014##
This is an ellipse centered at
[ - m .times. .lamda. .LAMBDA. , 0 ] . ##EQU00015##
As .PHI..sub.in is varied, the emission direction will traverse an
arc of this ellipse in u.sub.x,u.sub.y space. Tuning the wavelength
.lamda. will translate this ellipse forward and backward in the
u.sub.x direction. The total field-of-view in u.sub.x,u.sub.y space
will have the form of a curved band whose thickness will be
controlled by the total wavelength tuning range. We discuss the
number of 3 dB overlapped beams we can fit inside this
field-of-view below.
[0125] Far-Field Directivity
[0126] We can begin our derivation of the far-field pattern by
noting that we can completely specify the near-field amplitude to
have the following form:
A(x,y)=exp(-qx)exp(ik.sub.0u.sub.x0 exp(ik.sub.0u.sub.y0y) (13)
where u.sub.x0=sin(.theta..sub.0)cos(.PHI..sub.0) and
u.sub.y0=sin(.theta..sub.0)sin(.PHI..sub.0).
[0127] We have implicitly assumed a rectangular beam profile along
the y-direction to simplify our calculations. In general we expect
an additional function f (y, x) to modulate the amplitude of the
pattern according to the feed pattern, illumination position, and
lens geometry. This derivation captures the most critical features
of the far-field pattern and establishes an upper bound on the
gain. In addition, the performance of the aperture is largely
determined by its phase behavior, so smearing the amplitude
distribution relative to the ideal tends to lead to small
changes.
[0128] The physical aperture we are integrating over is a
parallelogram bounded by the following conditions:
0 .ltoreq. x .ltoreq. L ( 14 ) - W 2 + x .times. .times. tan
.function. ( .PHI. ' ) .ltoreq. y .ltoreq. W 2 + x .times. .times.
tan .function. ( .PHI. ' ) ( 15 ) ##EQU00016##
Here .PHI.' is equal to
( n 1 .times. sin .function. ( .PHI. in ) .times. k 0 k x , avg )
.apprxeq. ( n 1 .times. sin .function. ( .PHI. in ) n eff .times.
cos .function. ( .PHI. in ) ) ##EQU00017##
and is close in magnitude to .PHI..sub.in from the previous
section, but not identical because of the refraction at the grating
steps. To find the far-field pattern we can compute the Fourier
transform of this amplitude pattern over the domain:
F .function. ( u x - u x .times. .times. 0 , u y - u y .times.
.times. 0 ) = .intg. 0 L .times. dx .times. .times. .intg. - W 2 +
xtan .times. .times. .PHI. ' W 2 + xtan .times. .times. .PHI. '
.times. dy .times. .times. exp .function. ( - q .times. x ) .times.
exp .function. ( - i .times. k 0 .function. ( u x - u x .times. 0 )
.times. x ) .times. exp .function. ( - i .times. k 0 .function. ( u
y - u y .times. 0 ) .times. y ) ( 16 ) ##EQU00018##
Where u.sub.x=sin(.theta.)cos(.PHI.) and
u.sub.y=sin(.theta.)sin(.PHI.), which are the direction angles. For
convenience, from here we denote u.sub.x-u.sub.x0 with
.DELTA.u.sub.x and u.sub.y-u.sub.y0 with .DELTA.u.sub.y. We can
evaluate these integrals easily to find:
F .function. ( .DELTA. .times. u x , .DELTA. .times. u y ) = sin
.function. ( k 0 .times. W 2 .times. .DELTA. .times. u y ) k 0
.times. .DELTA. .times. u y .times. 1 - exp .function. ( - q
.times. L ) .times. exp .function. ( i .times. k 0 .times. L
.function. ( u x + tan .function. ( .PHI. ' ) .times. .DELTA.
.times. u y ) ) q + i .times. k 0 .function. ( u x + tan .function.
( .PHI. ' ) .times. .DELTA. .times. u y ) .times. A ( 17 )
##EQU00019##
The power of the far-field pattern is the magnitude of the field
pattern squared, that is P=|F|.sup.2. We use the power P below to
compute the directivity of the far-field pattern with the following
expression:
D .function. ( .theta. , .PHI. ) = P .function. ( .theta. , .PHI. )
1 2 .times. .pi. .times. .intg. 0 .pi. .times. d .times. .times.
.theta.sin.theta. .times. .intg. 0 2 .times. .pi. .times. d .times.
.times. .PHI. .times. .times. P .function. ( .theta. , .PHI. ) ( 18
) ##EQU00020##
[0129] The directivity gives the factor of the power emitted in a
given direction relative to an isotropic radiator. A well-designed
directional antenna tends to increase or maximize the peak gain,
the directivity of the main lobe, and reduce or minimize the power
into sidelobes, because these waste power and contribute to false
detections. We will discuss the directivity more below concerning
the range of the system and the number of resolvable points.
[0130] We can create a simpler expression by expanding the
direction angles about the far-field peak at .theta.=.theta..sub.0
and .PHI.=.PHI..sub.0, we can also take the limits of the integral
to infinity. This creates negligible error in the case of high-gain
beams and ultimately allows many of these gain integrals to be
evaluated analytically:
D .function. ( .DELTA. .times. .theta. , .DELTA..PHI. ) .apprxeq. P
.function. ( .DELTA. .times. .theta. , .DELTA..PHI. ) sin .times.
.theta. 0 2 .times. .pi. .times. .intg. - .infin. .infin. .times. d
.times. .times. .DELTA..theta. .times. .intg. - .infin. .infin.
.times. d .times. .times. .DELTA..PHI. .times. .times. P .function.
( .DELTA. .times. .theta. , .DELTA..PHI. ) ( 19 ) ##EQU00021##
[0131] It's illustrative to change coordinates of this expression
from 9 and 0 to .DELTA.u.sub.x and .DELTA.u.sub.y. We can find:
.DELTA.u.sub.x=cos(.theta..sub.0)sin(.PHI..sub.0).DELTA..theta.+sin(.the-
ta..sub.0)cos(.PHI..sub.0).DELTA..PHI.=u.sub.x-u.sub.x,0 (20)
.DELTA.u.sub.y=cos(.theta..sub.0)cos(.PHI..sub.0).DELTA..theta.-sin(.the-
ta..sub.0)sin(.PHI..sub.0).DELTA..PHI.=u.sub.y-u.sub.y,0 (21)
where we have taken .PHI.=.PHI..sub.0+.DELTA..PHI. and
.theta.=.theta..sub.0+.DELTA..theta.. We can use these expressions
to calculate the following Jacobian, where we have changed
variables from .PHI. and .theta. to .DELTA..PHI. and
.DELTA..theta.:
d .times. .times. .DELTA..theta. .times. .times. d .times. .times.
.DELTA..PHI.sin .function. ( .theta. 0 ) = d .times. .times.
.DELTA. .times. .times. u x .times. d .times. .times. .DELTA.
.times. .times. u y .times. sin .function. ( .theta. 0 ) .times.
sin .function. ( .PHI. 0 ) cos .function. ( .theta. 0 ) cos
.function. ( .PHI. 0 ) cos .function. ( .theta. 0 ) cos .function.
( .PHI. 0 ) sin .function. ( .theta. 0 ) - sin .function. ( .PHI. 0
) sin .function. ( .theta. 0 ) = d .times. .times. .DELTA. .times.
.times. u x .times. d .times. .times. .DELTA. .times. .times. u y
cos .function. ( .theta. 0 ) ( 22 ) ##EQU00022##
Taken together, we can use these results to rewrite our expression
for the peak gain as a function of .DELTA.u.sub.x and
.DELTA.u.sub.y:
D .function. ( .DELTA. .times. u x , .DELTA. .times. u y ) = 2
.times. .pi. .times. cos .function. ( .theta. 0 ) .times. P
.function. ( .DELTA. .times. u x , .DELTA. .times. u y ) .intg. -
.infin. .infin. .times. d .times. .DELTA. .times. u x .times.
.intg. - .infin. .infin. .times. d .times. .times. .DELTA. .times.
.times. u y .times. P .function. ( .DELTA. .times. u x , .DELTA.
.times. u y ) ( 23 ) ##EQU00023##
[0132] This unsimplified expression can already tell us something
very useful--that the peak gain of a given pattern is directly
proportional to cos(.theta..sub.0). This result emerges because the
far-field gain in general is proportional to the projected area. To
first order, neglecting additional aberrations and changes in the
grating parameters, effective indices, reflections, and feed
illumination, the peak gain fall-off as a function of angle is just
determined by the angle between the emission vector and the z-axis.
Another feature of this equation is that the peak shape is
essentially independent of the center of the main lobe: to leading
order the pattern just changes by the cos(.theta..sub.0) scale
factor.
[0133] We can directly evaluate these integrals for our far-field
pattern:
.intg. - .infin. .infin. .times. d .times. .DELTA. .times. u x
.times. .intg. - .infin. .infin. .times. d .times. .DELTA. .times.
u y .times. P .function. ( .DELTA. .times. u x , .DELTA. .times.
.times. u y ) = .intg. - .infin. .infin. .times. d .times. .times.
.DELTA. .times. .times. u x .times. .intg. - .infin. .infin.
.times. d .times. .times. .DELTA. .times. .times. u y .times. sin
.times. .times. c 2 .function. ( W 2 .times. k 0 .times. .DELTA.
.times. u y ) 1 + k 0 2 q 2 .times. ( .DELTA. .times. u x + tan
.function. ( .PHI. ) .times. .DELTA. .times. .times. u y ) 2
.times. ( 1 - 2 .times. cos .function. ( k 0 .times. L .function. (
.DELTA. .times. u x + tan .function. ( .PHI. ) .times. .DELTA.
.times. u y ) ) .times. exp .function. ( - q .times. L ) + exp
.function. ( - 2 .times. q .times. L ) ) ( 24 ) ##EQU00024##
We first start by performing a shear transformation on the
integrating variables given by:
.DELTA.u.sub.x,s=.DELTA.u.sub.x+tan(.PHI.).DELTA.u.sub.y and
.DELTA.u.sub.y,s=.DELTA.u.sub.y. With this transformation the
integral now becomes separable:
.intg. - .infin. .infin. .times. d .times. .DELTA. .times. u x
.times. .intg. - .infin. .infin. .times. d .times. .DELTA. .times.
u x .times. P .function. ( .DELTA. .times. u x , .DELTA. .times.
.times. u y ) = .intg. - .infin. .infin. .times. d .times. .DELTA.
.times. u x , s .times. ( 1 - 2 .times. cos .function. ( k 0
.times. L .times. .DELTA. .times. u x , s ) .times. exp .function.
( - q .times. L ) + exp .function. ( - 2 .times. q .times. L ) ) 1
+ k 0 2 q 2 .times. .DELTA. .times. u x , s 2 .times. .intg. -
.infin. .infin. .times. d .times. .DELTA. .times. u y , s .times.
sin .times. .times. c 2 .function. ( W 2 .times. k 0 .times.
.DELTA. .times. u y , s ) ( 25 ) ##EQU00025##
Note that the angle .theta..sub.p does not change the projected
area of the aperture, since it just shears the emitting surface.
Consequently we expect it to completely drop out of the integral,
which is indeed the case. Next we remove the dimensions and break
the integrals into parts and evaluate:
2 .times. q W .times. k 0 2 .times. .intg. - .infin. .infin.
.times. d .times. x .function. ( ( 1 + exp .function. ( - 2 .times.
q .times. L ) ) 1 + x 2 - 2 .times. cos .function. ( q .times. L
.times. x ) .times. exp .function. ( - q .times. L ) 1 + x 2 )
.times. .intg. - .infin. .infin. .times. d .times. y .times. ( 2
.times. ( y ) ) = 2 .times. q .times. .times. .pi. 2 W .times. k 0
2 .times. ( ( 1 + exp .function. ( - 2 .times. q .times. L ) ) - 2
.times. exp .function. ( - 2 .times. q .times. L ) ) ( 26 )
##EQU00026##
Using these results, finally we can write the directivity as:
D .function. ( .DELTA. .times. u x , .DELTA. .times. u y ) = W
.times. k 0 2 .times. cos .function. ( .theta. 0 ) .pi. .times.
.times. q .function. ( 1 - exp .function. ( - 2 .times. qL ) )
.times. sin .times. .times. c 2 .function. ( W 2 .times. k 0
.times. .DELTA. .times. u y ) 1 + k 0 2 q 2 .times. ( .DELTA.
.times. u x + tan .function. ( .PHI. ' ) .times. .DELTA. .times. u
y ) 2 .times. ( 1 - 2 .times. cos .function. ( k 0 .times. L
.function. ( .DELTA. .times. u x + tan .function. ( .PHI. ' )
.times. .DELTA. .times. u y ) ) .times. exp .function. ( - q
.times. L ) + exp .function. ( - 2 .times. q .times. L ) ) ( 27 )
##EQU00027##
We can also expression the peak directivity as:
D max = cos .function. ( .theta. 0 ) .times. W .times. k 0 2 .pi.
.times. .times. q .times. 1 1 + k 0 2 q 2 .times. ( 1 - exp
.function. ( - q .times. L ) ) 2 ( 1 - exp .function. ( - 2 .times.
q .times. L ) ) ( 28 ) ##EQU00028##
[0134] To gain a little insight into how this function behaves, we
can simplify it for large and small L. For L<<1/q, we
find:
lim q .times. L .fwdarw. 0 .times. D .function. ( .DELTA. .times. u
x , .DELTA. .times. .times. u y ) = W .times. L .times. k 0 2
.times. cos .function. ( .theta. 0 ) 2 .times. .pi. .times. sin
.times. .times. c 2 .function. ( W 2 .times. k 0 .times. .DELTA.
.times. u y ) .times. sin .times. .times. c 2 .function. ( L 2
.times. k 0 .function. ( .DELTA. .times. u x + tan .function. (
.PHI. ' ) .times. .DELTA. .times. u y ) ) ( 29 ) ##EQU00029##
This is just the directivity from a sheared rectangular aperture of
length L and width W, note that the peak gain is
W .times. L .times. k 0 2 .times. cos .function. ( .theta. 0 ) 2
.times. .pi. , ##EQU00030##
which is directly proportional to the projected area WL
cos(.theta..sub.0). Taking the opposite limit, we can find another
useful simplification:
lim q .times. L .fwdarw. .infin. .times. D .function. ( .DELTA.
.times. u x , .DELTA. .times. .times. u y ) = W .times. k 0 2
.times. cos .function. ( .theta. 0 ) .pi. .times. .times. q .times.
sin .times. .times. c 2 .function. ( W 2 .times. k 0 .times.
.DELTA. .times. u y ) 1 + k 0 2 q 2 .times. ( .DELTA. .times. u x +
tan .function. ( .PHI. ' ) .times. .DELTA. .times. u y ) 2 ( 30 )
##EQU00031##
[0135] Here the peak directivity scales as
W q .times. .times. cos .function. ( .theta. 0 ) , where .times.
.times. 1 q ##EQU00032##
becomes the effective length of the aperture. Even though these
expressions are much simpler than the general one we derived, even
if the aperture is several decay lengths long, the effect of the
finite length of the directivity is significant and properly
modeling it requires the full expression. An example of this is in
computing the number of far-field resolvable points.
[0136] Number of Resolvable Points with Wavelength Tuning
[0137] Another property of our system is the number of far-field
resolvable points. There are some relatively simple expressions we
can derive which will tightly bound the number of resolvable points
we can achieve in a particular system as a function of the aperture
parameters. We will first start with the number of resolvable
points we can achieve through wavelength tuning. Assuming normal
incidence from the feed, the far-field condition for the unit
vector in the x-direction is just:
n eff .times. k 0 - 2 .times. .pi. .LAMBDA. .times. m = k 0 .times.
u x , ##EQU00033##
where u.sub.x is the unit vector of the wavevector in the
x-direction, n.sub.eff is the effective index of the grating at
normal incidence, .LAMBDA. is the grating period, and m is the
grating order.
[0138] We want to count the number of full-width half-maximums
.DELTA.u.sub.FWHM, we can fit inside a total tuning range of
.DELTA.u.sub.range. .DELTA.u.sub.range in this case is just given
by
m .times. .DELTA. .times. .lamda. .LAMBDA. , ##EQU00034##
where .DELTA..lamda. is the tuning wavelength, and is typically
50-100 nm for integrated tunable sources. With this we can write a
simple expression for the number of resolvable points with
wavelength tuning N.sub.wavelength:
N wavelength .apprxeq. m .times. .DELTA. .times. .lamda. .LAMBDA. 0
.times. .DELTA. .times. u FWHM + 1 ( 31 ) ##EQU00035##
[0139] In the case of a finite length grating, .DELTA.u.sub.FWHM is
computed numerically from the full directivity formula to give a
precise calculation of the number of resolvable points. However, in
the case of a long grating, we can determine exactly that
.DELTA. .times. u FWHM = 2 .times. q k 0 . ##EQU00036##
Plugging this in gives the following relationship:
N wavelength .apprxeq. .pi. .times. .DELTA. .times. .lamda. .lamda.
0 .times. m q .times. .LAMBDA. 0 + 1 ( 32 ) ##EQU00037##
Assuming that at .lamda..sub.0, that the grating is emitting at
normal incidence, and using our expression relating the decay
length q to the grating quality factor Q, we find that:
N wavelength .apprxeq. Qn eff .times. .DELTA..lamda. .lamda. 0
.times. v g c + 1 ( 33 ) ##EQU00038##
[0140] Number of Resolvable Points from In-Plane Steering
[0141] We assume that we can determine the number of resolvable
points from the field pattern at the lens aperture, as opposed to
the pattern after being emitted from the grating. In 1D, the
directivity of a far-field pattern A(.theta.) is defined by
A 1 .pi. .times. .intg. A .function. ( .theta. ) .times. d .times.
.theta. . ##EQU00039##
Assuming that the power is confined to a single lobe of angular
width .DELTA..theta., we can approximate D.sub.peak as
.pi./.DELTA..theta.. Neglecting lens aberrations, the directivity
can be written:
D .function. ( .theta. ) = .pi. .DELTA. .times. .theta. .times. cos
.function. ( .theta. ) ( 34 ) ##EQU00040##
[0142] The steering range in this situation is limited by the
minimum acceptable gain usable by the system. Typically RADARs are
designed to have a directivity fall-off of 3 dB or 0.5 at the edge
of their usable FOV. This gives an effective steering range of
2.pi./3 radians. Conveniently approximating the beam-width to be
constant, we find that:
N In .times. - .times. plane .apprxeq. 2 .times. D p .times. e
.times. a .times. k 3 = 2 .times. .pi. .times. W 3 .times. .lamda.
( 35 ) ##EQU00041##
where we have substituted in the peak directivity of a rectangular
aperture of size W. This equation accurately reflects the scaling
of the number of resolvable points when a/.lamda. is between 10 and
40 or so. Beyond this, the path error for off-axis scanning angles
begins to become an appreciable fraction of the wavelength (since
the error is directly proportional to the lens size). The 3 dB
scanning limit will be squeezed inwards as a/.lamda. increases.
[0143] Abbe Sine Condition
[0144] If desired, we can shape a lens to satisfy the Abbe sine
condition. Satisfying the Abbe sine condition eliminates Coma
aberration on-axis and reduces it off axis in the regime where
sin(.PHI.)=.PHI.. We briefly outline the procedure for generating a
shaped lens given input parameters thickness T, focal length F,
effective focal length F.sub.e, and index n. The inner surface of
the lens is defined by r,.theta., while the outer surface is
defined by x, y. In this coordinate system, we satisfy the Abbe
sine condition when y=F.sub.e sin(.theta.). We can further relate r
and .theta. to x and y from the following expression calculated
from ray-propagation:
r+n {square root over ((y-r sin(.theta.)).sup.2+(x-r
cos(.theta.)).sup.2)}-x=(n-1)T (36)
This can be written as a quadratic equation for x and solved. Once
x is solved, r can be advanced by computing:
d .times. r d .times. .theta. = nr .times. .times. sin .function. (
.theta. - .theta. ' ) n .times. .times. cos .function. ( .theta. -
.theta. ' ) - 1 .times. .times. Where .times. : ( 37 ) .theta. ' =
tan - 1 [ ( F e - r ) .times. .times. sin .times. .times. ( .theta.
) x - r .times. .times. cos .function. ( .theta. ) ] ( 38 )
##EQU00042##
These equations can be solved iteratively to generate the entire
lens surface, beginning with .theta.=0 and r=F. Other methods can
be used to generate shaped lens surfaces, such as designing the
aperture power pattern based on the feed power pattern or forcing
the lens to have two off-axis focal points.
[0145] LIDAR Range
[0146] Generally, the minimum detectable received power P.sub.r,min
from a LIDAR return determines the maximum range of the device.
P.sub.r,min is determined by the integration time and sensor
architecture, which can be based on frequency modulated continuous
wave (FMCW) or pulsed direct detection type schemes. If a target
has a cross section .sigma., the maximum range we can observe that
target is given by the standard RADAR equation:
P r , min = D .function. ( .theta. , .PHI. ) 2 .times. .eta. 2 ( 4
.times. .pi. ) 3 .times. .lamda. 2 R max 4 .times. .sigma. .times.
P t ( 39 ) ##EQU00043##
where D(.theta.,.PHI.) is the directivity, .eta. is the device
efficiency, R.sub.max is the maximum range, and P.sub.t is the
transmitter power. We see here that the primary determinant of the
LIDAR preformance beyond the detection backend are the antenna
characteristics given by D(.theta.,.PHI.) and .eta..
[0147] In the case that the beam spot from the LIDAR is contained
completely within the target, which is a common application mode
for LIDARs, we can derive an alternate constraint, which is more
forgiving than the standard RADAR range equation in terms of
distance falloff:
P r , min = D .function. ( .theta. , .PHI. ) .times. .eta. ( 4
.times. .pi. ) 2 .times. .lamda. 2 R max 2 .times. P t ( 40 )
##EQU00044##
[0148] Overview of Numerical Methods and Verification
[0149] Because of structures are large and lack periodicity, full
3D FDTD simulations were not possible. However, we were able to do
smaller 2D and 3D FDTD simulations of individual components to help
verify the system performance.
[0150] First, we conducted simulations of the waveguides generated
from the routing algorithms to verify that they were defined with
enough points, were not too close, and satisfied minimum bend
radius requirements. Unfortunately, having 3 dB spaced far-field
spots results in wavelength-spaced ports in the focal surface.
Although the waveguides can be wavelength-spaced for short lengths
without significant coupling, generally the feed geometry results
in excessively high coupling between waveguides. We fixed this
problem by decimating the ports by a factor of two.
[0151] FIG. 11 shows 32 simulated 3 dB overlapped far-field beam
patterns. Each main peak represents a single port excitation. Peaks
to each side of main peak are sidelobes. These represent power
radiated in unintended directions and may result in false
detections. The dotted line indicates -3 dB. Port positions are
designed to overlap far-field resolvable spots by 3 dB. For a 2D
aperture, this is done along parameterized curve between each spot
in u.sub.x,u.sub.y space.
[0152] FIGS. 12A-12D show simulations for design of chip-scale
LIDAR. FIG. 12A shows a simulation of a far-field beam pattern to
extract phase center. FIG. 12B shows a simulation of a far-field
beam pattern to extract gaussian beamwidth. FIG. 12C shows a 2D
simulation of on-axis port excitation of a lens feed. FIG. 12D
shows a 2D simulation of off-axis port excitation of the lens
feed.
[0153] The simulations of FIGS. 12A and 12B involve the interface
between the waveguide and the SiN slab. With these simulations, we
visualized the creation of an effective point dipole source when
the waveguide terminated at the slab, and determined the far-field
radiation pattern and effective phase center. By fitting circles to
the phase front, we were able to extract a location for the phase
center, which we show in FIG. 12A. The phase front was
approximately 1 .mu.m behind the interface between the waveguide
slab and the waveguide. The far-field power as a function of angle
was well described by a Gaussian with a beam-width of 13.5.degree.,
as shown in FIG. 12B. The phase center and beam-width did not
change if the waveguide was incident at an angle on the interface:
the phase center and beam-profile remained the same relative to the
orientation of the waveguide. This feature was useful because it
allowed us to angle the waveguides to reduce or minimize the
spillover loss (the radiation that misses the lens) without having
to be concerned about the beam-width or phase center changing.
[0154] Another set of simulations we performed concerned the
interface of the lens and the slab. One assumption, which is also a
feature of other works on integrated planar lenses, is that we can
describe the in-plane propagation in terms of the effective mode
indices. We did several calculations of TE slab modes impacting 20
nm and 40 nm Si slab "steps" to verify this assertion and to
quantify the radiation loss at these interfaces. We found that for
a wide range of angles, the radiation loss was less than 5% in line
with previous experimental results for incident angles less than
40.degree..
[0155] We also performed effective 2D FDTD simulations of the
waveguides, the lens feed, and the lens itself to verify that
beam-steering worked properly. We see this in FIGS. 12C and 12D. We
confirmed the directivity derived from these simulations closely
matched those produced by ray-tracing. We also verified that the
expected lens roughness for fabrication would not result in
excessive gain degradation. 2D grating simulations of a 1D grating
were used to extract the grating Q as a function of wavelength. The
emission angle was compared to that predicted from the average
grating index and good agreement was obtained. Additionally we
modeled the photonic bandstructure using meep to confirm that our
excitation was far from the Bragg band edge.
[0156] Finally, we extracted the grating Q as a function of angle
and wavelength from meep calculations. We confirmed the on-axis
performance matched that predicted from the FDTD simulations.
Additionally we confirmed that the Q did not change too much for
off-axis propagation. Generally the behavior within .+-.20.degree.
was well-behaved, but beyond that there were large fluctuations.
For the ray-tracing simulations, in the regime of interest, the
grating Q could be considered constant, but in general it was a
complicated, rapidly varying function. Rigorously modeling Q as a
function of angle accounts for the unexpectedly strong dependence
in certain regimes.
Index Error
[0157] The effective index ratio n.sub.2/n.sub.1 of an experimental
system is different than that used in ray-tracing simulations,
because of finite fabrication tolerance, wavelength dispersion,
temperature variation, etc. In general, an error in the index may
cause the focal plane to shift by some amount. For a parabolic
lens, we find that the change is:
.DELTA. .times. f = f 2 R .times. .DELTA. .times. n ( 41 )
##EQU00045##
[0158] Since the depth of focus scales as .lamda., R: f, and f:
.lamda.N, where N is the number of resolvable points, we have that
our effective index tolerance scales inversely with the number of
resolvable points that the imaging system supports:
.DELTA. .times. n .apprxeq. 1 N ( 42 ) ##EQU00046##
Without any kind of external tunablity, meeting this constraint for
large N becomes increasingly difficult. For more than 100 ports,
wavelength dispersion over a 100 nm bandwidth already exceeds this
constraint for a 40 nm thick lens. For proper operation of a device
with 100 ports at a single wavelength, there should be better than
.+-.1 nm of precision in the layer heights, and better than 0.01
tolerance in the material index. Addressing these index tolerance
issues enables scaling the system to 1000s of resolvable
points.
CONCLUSION
[0159] While various inventive embodiments have been described and
illustrated herein, those of ordinary skill in the art will readily
envision a variety of other means and/or structures for performing
the function and/or obtaining the results and/or one or more of the
advantages described herein, and each of such variations and/or
modifications is deemed to be within the scope of the inventive
embodiments described herein. More generally, those skilled in the
art will readily appreciate that all parameters, dimensions,
materials, and configurations described herein are meant to be
exemplary and that the actual parameters, dimensions, materials,
and/or configurations will depend upon the specific application or
applications for which the inventive teachings is/are used. Those
skilled in the art will recognize, or be able to ascertain using no
more than routine experimentation, many equivalents to the specific
inventive embodiments described herein. It is, therefore, to be
understood that the foregoing embodiments are presented by way of
example only and that, within the scope of the appended claims and
equivalents thereto, inventive embodiments may be practiced
otherwise than as specifically described and claimed. Inventive
embodiments of the present disclosure are directed to each
individual feature, system, article, material, kit, and/or method
described herein. In addition, any combination of two or more such
features, systems, articles, materials, kits, and/or methods, if
such features, systems, articles, materials, kits, and/or methods
are not mutually inconsistent, is included within the inventive
scope of the present disclosure.
[0160] Also, various inventive concepts may be embodied as one or
more methods, of which an example has been provided. The acts
performed as part of the method may be ordered in any suitable way.
Accordingly, embodiments may be constructed in which acts are
performed in an order different than illustrated, which may include
performing some acts simultaneously, even though shown as
sequential acts in illustrative embodiments.
[0161] All definitions, as defined and used herein, should be
understood to control over dictionary definitions, definitions in
documents incorporated by reference, and/or ordinary meanings of
the defined terms.
[0162] The indefinite articles "a" and "an," as used herein in the
specification and in the claims, unless clearly indicated to the
contrary, should be understood to mean "at least one."
[0163] The phrase "and/or," as used herein in the specification and
in the claims, should be understood to mean "either or both" of the
elements so conjoined, i.e., elements that are conjunctively
present in some cases and disjunctively present in other cases.
Multiple elements listed with "and/or" should be construed in the
same fashion, i.e., "one or more" of the elements so conjoined.
Other elements may optionally be present other than the elements
specifically identified by the "and/or" clause, whether related or
unrelated to those elements specifically identified. Thus, as a
non-limiting example, a reference to "A and/or B", when used in
conjunction with open-ended language such as "comprising" can
refer, in one embodiment, to A only (optionally including elements
other than B); in another embodiment, to B only (optionally
including elements other than A); in yet another embodiment, to
both A and B (optionally including other elements); etc.
[0164] As used herein in the specification and in the claims, "or"
should be understood to have the same meaning as "and/or" as
defined above. For example, when separating items in a list, "or"
or "and/or" shall be interpreted as being inclusive, i.e., the
inclusion of at least one, but also including more than one, of a
number or list of elements, and, optionally, additional unlisted
items. Only terms clearly indicated to the contrary, such as "only
one of" or "exactly one of," or, when used in the claims,
"consisting of," will refer to the inclusion of exactly one element
of a number or list of elements. In general, the term "or" as used
herein shall only be interpreted as indicating exclusive
alternatives (i.e. "one or the other but not both") when preceded
by terms of exclusivity, such as "either," "one of," "only one of,"
or "exactly one of" "Consisting essentially of," when used in the
claims, shall have its ordinary meaning as used in the field of
patent law.
[0165] As used herein in the specification and in the claims, the
phrase "at least one," in reference to a list of one or more
elements, should be understood to mean at least one element
selected from any one or more of the elements in the list of
elements, but not necessarily including at least one of each and
every element specifically listed within the list of elements and
not excluding any combinations of elements in the list of elements.
This definition also allows that elements may optionally be present
other than the elements specifically identified within the list of
elements to which the phrase "at least one" refers, whether related
or unrelated to those elements specifically identified. Thus, as a
non-limiting example, "at least one of A and B" (or, equivalently,
"at least one of A or B," or, equivalently "at least one of A
and/or B") can refer, in one embodiment, to at least one,
optionally including more than one, A, with no B present (and
optionally including elements other than B); in another embodiment,
to at least one, optionally including more than one, B, with no A
present (and optionally including elements other than A); in yet
another embodiment, to at least one, optionally including more than
one, A, and at least one, optionally including more than one, B
(and optionally including other elements); etc.
[0166] In the claims, as well as in the specification above, all
transitional phrases such as "comprising," "including," "carrying,"
"having," "containing," "involving," "holding," "composed of," and
the like are to be understood to be open-ended, i.e., to mean
including but not limited to. Only the transitional phrases
"consisting of" and "consisting essentially of" shall be closed or
semi-closed transitional phrases, respectively, as set forth in the
United States Patent Office Manual of Patent Examining Procedures,
Section 2111.03.
* * * * *