可编程光子芯片：Nature综述
需积分: 47 88 浏览量
2022-11-21
13:55:36
上传
评论
收藏 3.68MB PDF 举报
Nature | Vol 586 | 8 October 2020 | 207
Review
Programmable photonic circuits
Wim Bogaerts
1,2
✉
, Daniel Pérez
3,4
, José Capmany
3,4
, David A. B. Miller
5
, Joyce Poon
6,7
,
Dirk Englund
8
, Francesco Morichetti
9
& Andrea Melloni
9
The growing maturity of integrated photonic technology makes it possible to build
increasingly large and complex photonic circuits on the surface of a chip. Today, most
of these circuits are designed for a specic application, but the increase in complexity
has introduced a generation of photonic circuits that can be programmed using
software for a wide variety of functions through a mesh of on-chip waveguides,
tunable beam couplers and optical phase shifters. Here we discuss the state of this
emerging technology, including recent developments in photonic building blocks
and circuit architectures, as well as electronic control and programming strategies.
We cover possible applications in linear matrix operations, quantum information
processing and microwave photonics, and examine how these generic chips can
accelerate the development of future photonic circuits by providing a higher-level
platform for prototyping novel optical functionalities without the need for custom
chip fabrication.
Photonic integrated circuits (PICs) have recently become an estab-
lished and powerful technology that supports many applications
1,2
.
Like electronic integrated circuits, PICs are implemented on the surface
of a chip, but they manipulate light instead of electrical signals, using
on-chip optical waveguides, beam couplers, electro-optic modulators,
photodetectors and lasers. Whereas electronic circuits are good at
digital computations, photonics circuits are good at transporting and
processing analogue information. Therefore, today PICs are mostly
used in fibre-optic communications, but they are also useful in various
applications in which light has an important role, such as chemical, bio-
logical or spectroscopic sensors, metrology, and classical and quantum
information processing. Because most photonic circuits are tailored for
one of these applications, they are called application-specific photonic
integrated circuits (ASPICs) and, given that the flow of light is essentially
fixed, they can be very compact and power-efficient.
Programmable PICs are based on the idea that the flow of light on
the chip can be manipulated at run-time, for example, by electrically
controlling tunable beam couplers connected by optical waveguides
3
.
This way, light is distributed and spatially rerouted under software con-
trol. These chips can implement various linear functions by interfering
signals along different paths, and they can define programmable wave-
length filters
3
, which are essential building blocks for communication
or sensor applications and for the manipulation of microwave signals in
the optical domain
4,5
. When scaling up such meshes of connected wave-
guides, the interferences can perform linear optical computations, such
as real-time matrix–vector products
6–8
. These are essential operations
in quantum information processing
9–12
, neuromorphic computing and
artificial intelligence
6,7
, and we are already seeing rapid development of
programmable PIC technologies in these applications. As in electronics,
programmability makes it possible to (re)configure the functionality
at run-time, which lowers the economic and technological barriers to
using the circuit and provides a path to upgradability.
In conventional optics, a system with even a few interferometric
elements becomes difficult to line up, in terms of both space and
wavelength. However, we can now fabricate complex interferometric
systems on a chip, with architectures and algorithms for program-
ming, stabilization and control. Some of these systems even allow
self-configuration, adapting the circuit in real time to the optical prob
-
lem being solved, without high-level calculations
8,13,14
. This combina-
tion of complex circuits and control techniques is opening the field of
programmable photonics.
Here, we summarize recent developments in this emerging field. We
start by explaining the core concepts of waveguide meshes and how
they route light and perform analogue matrix and filtering operations.
We then look at the necessary technologies for such photonic circuits.
Because programmable PICs are more generic than ASPICs, they can
be deployed in various applications, but there are some fields in which
their unique capability to perform matrix and parallel operations is
especially valuable. Based on this, we look at the future potential of
programmable photonics.
Mesh architectures and algorithms
In programmable photonic integrated circuits, the flow of light is con-
trolled by waveguides connected in a mesh using 2×2 blocks, or ‘ana-
logue gates’, the on-chip equivalent of free-space optical beam splitters.
The mesh connectivity determines the possible functions of the pro-
grammable circuit, and how it can be configured. Some architectures
enable arbitrary matrix operations
4–6,8,12–31
, and can even automatically
adapt to changing problems
8,13,14,29–32
.
We can separate waveguide meshes into two broad classes:
(1) forward-only, where the light flows from one side of the mesh to the
other
7,8,12,14,32,33
and (2) recirculating, where light can also be routed in
loops and even back to the input ports
3,5,18,23
. Both architectures use the
https://doi.org/10.1038/s41586-020-2764-0
Received: 17 March 2020
Accepted: 10 July 2020
Published online: 7 October 2020
Check for updates
1
IMEC, Department of Information Technology, Ghent University, Ghent, Belgium.
2
Center of Nano- and Biophotonics, Ghent University, Ghent, Belgium.
3
Universitat Politècnica València, ITEAM
Research Institute, Valencia, Spain.
4
iPronics, Programmable Photonics, Valencia, Spain.
5
Ginzton Laboratory, Stanford University, Stanford, CA, USA.
6
Max Planck Institute of Microstructure
Physics, Halle, Germany.
7
Edward S. Rogers Department of Electrical and Computer Engineering, University of Toronto, Toronto, Ontario, Canada.
8
Research Laboratory of Electronics,
Massachusetts Institute of Technology, Cambridge, MA, USA.
9
Dipartimento di Elettronica, Informazione e Bioingegneria, Politecnico di Milano, Milan, Italy.
✉
e-mail: wim.bogaerts@ugent.be
资源评论
