PRXQuantum 5 010102

PRX QUANTUM 5, 010102 (2024)
Perspective
Scalable Fault-Tolerant Quantum Technologies with Silicon Color Centers

Stephanie Simmons*
Photonic, Inc.
(Received 19 October 2023; published 13 March 2024)
The scaling barriers currently faced by both quantum networking and quantum computing technologies
ultimately amount to the same core challenge of distributing high-quality entanglement at scale. In this
Perspective, a novel quantum information-processing architecture based on optically active spins in silicon
is proposed that offers a combined single technological platform for scalable fault-tolerant quantum com-
puting and networking. The architecture is optimized for overall entanglement distribution and leverages
color-center spins in silicon (T centers) for their manufacturability, photonic interface, and high-fidelity
information-processing properties. Silicon nanophotonic optical circuits allow for photonic links between
T centers, which are networked via telecom-band optical photons in a highly connected graph. This high
connectivity unlocks the use of low-overhead quantum error-correcting codes, significantly accelerating
the time line for modular scalable fault-tolerant quantum repeaters and quantum processors.
DOI: 10.1103/PRXQuantum.5.010102
I. CONTEXT [12]. Some post-quantum cryptography algorithms under

consideration have held up against attack so far and have
Quantum information processing unlocks novel tech-
the potential for strong long-lasting security in practice
nological capabilities that cannot be achieved classically.
but, similar to RSA, rest upon intrinsically unprovable
To this end, there is a long and growing list of quantum
computational security [13].
algorithms [1], to be executed on future quantum com-
In parallel to quantum computing, quantum networks
puters, which offer superpolynomial speed-ups over their
provide greater connectivity and security for quantum
known classical counterparts. Some of these algorithms
devices, and enable applications beyond the power of
are critical for useful tasks including quantum chemistry
stand-alone quantum devices. These include secure com-
and the design of novel materials, as they enable precise
munications [14] that even future quantum computers
simulations of chemical processes in important techno-
cannot break, “blind” computing of both classical and
logical applications ranging from catalysts [2] to batteries
quantum algorithms [15], and modular quantum comput-
[3] and pharmaceuticals [4]. Other quantum algorithms
ing [16], timing [17], and sensing [18]. Horizontal scaling
will impact our cybersecurity standards. For instance,
is a way of increasing the capacity of a quantum system by
once implemented on a large-scale fault-tolerant quantum
adding additional connected quantum chips, as opposed to
computer, Shor’s algorithm will allow efficient decryp-
increasing the capacity of existing chips by adding addi-
tion of any data protected by Rivest-Shamir-Adleman
tional qubits to each chip [19]. Analogously to today’s
(RSA) cryptography and its variants, which currently
high-performance modular classical supercomputers, mod-
underpin >90% of all financial and Internet traffic [5].
ular quantum supercomputers over quantum networks will
Interestingly, numerous so-called “quantum-safe” RSA-
enable astonishing computational capabilities by enabling
replacement encryption algorithms have already fallen to
true horizontal scaling of quantum resources, realizing
classical or quantum attacks, including Diffie-Hellman [6],
practical quantum advantage for the aforementioned appli-
Soliloquy [7], Elliptic Curve [8,9], Rainbow [10], SIKE
cations and giving rise to quantum applications yet to
[11], and (through a side channel) CRYSTALS-Kyber
be imagined. Importantly, modular quantum processors
connected by a quantum network directly offer all the capa-
bilities required of quantum repeaters, which are necessary
to scale quantum networks.
*
[email protected] However, today’s noisy intermediate-scale quantum
Published by the American Physical Society under the terms of (NISQ) devices cannot be trusted to execute practical-scale
the Creative Commons Attribution 4.0 International license. Fur- networking and computational instructions beyond small
ther distribution of this work must maintain attribution to the instances of a few specific problems [20–22]. For each of
author(s) and the published article’s title, journal citation, and today’s monolithic quantum architectures, it has been a
DOI. challenge to deliver high-fidelity operations at a scale of
2691-3399/24/5(1)/010102(18) 010102-1 Published by the American Physical Society

STEPHANIE SIMMONS PRX QUANTUM 5, 010102 (2024)
even a few hundred physical qubits and presently it seems deliver SFTQ technologies and the corresponding trans-
that more (and perhaps substantially more) high-fidelity formational real-world use cases far earlier than current
qubits will be necessary for commercial advantage. Simi- predictions.
larly, the scaling of quantum networks is currently blocked This technology is being built based upon networked
by the unavailability of reliable quantum repeaters. There- spins in silicon, specifically the T-center spin-photon inter-
fore, it remains an open question if any commercial value face [26]. The architecture itself is broadly applicable to
can be realized with these NISQ devices [23]. all long-lived spin-photon interfaces such as solid-state
Delivering the full potential of quantum information color centers [27] as well as some optically active atoms
processing requires the construction of scalable fault- in vacuum [28,29]. This architecture exploits the connec-
tolerant quantum (SFTQ) computing and networking tech- tivity and modularity of quantum networks to scale the
nologies. Although the global race to deliver SFTQ tech- power of fault-tolerant quantum processors and the error-
nology is already considerable and accelerating, and the corrected memory of distributed processors to scale the
performance of some small-scale quantum systems are reach of quantum networks. This paper is an overview of
approaching the levels needed to operate individual logical that vision.
qubits [24], scaling such systems seems to be a consider-
able challenge and hence the advent of large-scale SFTQ
II. CONNECTIVITY
networking and computing technologies is believed by
some to be a decade or more away [25] based upon the The stringent thresholds for fault-tolerant quantum error
quantum architectures known today. correction imply that qubits must operate in strictly
In this Perspective, the argument is made that quantum controlled environments, including, e.g., low tempera-
networks and quantum information processors will both tures, ultrahigh vacuum, electromagnetic shielding, with
achieve ultimate scale by combining them into the same high-purity materials, and more. Each of these physical
core entanglement-distribution technology. The network- constraints results in a certain system size, beyond which
ing of quantum computers will allow for true horizontal the marginal difficulty of adding the next qubit gets harder,
scaling of quantum resources through modularity; and the not easier. Constraints such as these imply that for each
introduction of quantum repeaters (small quantum com- quantum computing platform, there exists a natural system
puters) will scale the topologies and total distance of size past which it would be far easier to link multiple com-
telecom quantum networks. By way of illustration, a quan- puter modules instead of building ever-larger monolithic
tum architecture emphasizing entanglement distribution is quantum supercomputers. This shift into modular quan-
proposed (Fig. 1), which could be reasonably expected to tum processing has multiple advantages: it can directly
(a) (b) (c)

28
Si
13
C
Silicon 1
H
T Centers
e– Room-Temperature
Control Electronics Global-Scale
Satellite Links
Optical
al Cavities
Room-Temperature
Optical Switches
SNSPDs Optical Switches
1-K Cryostat
Metroscale
Quantum Processor Chip Fiber Links
FIG. 1. Photonic’s scalable quantum technology architecture. A quantum chip is cooled in a 1 K cryostat. This chip hosts inte-
grated silicon T centers within optical cavities, photonic switches, and single-photon detectors. Optical input-output (IO) ports connect
via telecom fiber to a room-temperature photonic switch network and control electronics. This naturally allows a highly connected
architecture with nonlocal connectivity even as the system scales in size. Telecom fiber also enables horizontal system scaling by
connecting multiple cryostats together through their optical IO. This enables both expansion of computing power and long-distance
quantum networks.
010102-2
SCALABLE FAULT-TOLERANT QUANTUM TECHNOLOGIES. . . PRX QUANTUM 5, 010102 (2024)
unlock quantum repeaters and scalable quantum networks (a) (b) (c)
if telecom photons are used to link the modules. For some
quantum computing architectures, the maximum number q1
of future qubits in any one module could, in principle, be q2
q1 q8 q1 q8 q3
quite large and the requirement to shift into horizontally
scalable modular quantum computing may be positioned q4
q2 q9 q2 q9
in the distant future. However, for competitive (rapid) scal- q5
ing of quantum resources, it is suggested in this Perspective q3 q10 q3 q10
q6
that modular quantum processing will be an ultimate long- q7
versus
term goal for all quantum architectures on a practical q4 q11 q4 q11 q8
basis. q9
If one assumes that modularity will unleash limit- q5 q12 q5 q12 q10
less horizontal scaling of fault-tolerant quantum net- q11
works and quantum supercomputers, as it has for classi- q6 q13 q6 q13 q12
cal networking and supercomputers, the role of connec- q13
tivity between modules deserves specific attention. For q7 q14 q7 q14
q14
reference, modern classical high-performance computing Module1 Module 2 Module1 Module2
hinges entirely upon parallelization across interconnected
computing modules [30,31]. In the quantum case, even (d)
higher degrees of intermodule connectivity will be crit- q1
ical. The architectural benefits of high connectivity have q2
been discussed in Refs. [32,33]. Even with quantum algo- q3
q4
rithms that minimize the number of logical operations
q5
across modules [34–36], entanglement will need to be dis- q6
tributed efficiently. This is deeply connected to how quan- q7
tum error correction is implemented in physical systems, q8
as is described next. q9
The scaling of connections between modules is a crit- q10
ical design parameter, as module size sets a lower bound q11
on the number of entangled pairs that must be shared q12
to apply an arbitrary operation on a joint two-module q13
system [37]. More importantly, connectivity remains sig- q14
nificant in the specific case where each module encodes
its own set of logical qubits. To implement logical two- FIG. 2. The temporal overheads associated with intermodule
qubit gates between separate modules, many physical connectivity. (a),(b) An illustrative example of two seven-qubit
qubits must be entangled between the individual modules. CSS code blocks with (a) limited (low) and (b) ideal (high)
intermodule connectivity. (c) Maximal intermodule connectiv-
By way of example, for a Calderbank-Shor-Steane (CSS)
ity, where each physical qubit can be directly entangled with
code (a quantum error correcting code where bit-flip and its partner qubit from the second module in parallel, results in
phase-flip errors are detected and corrected independently a single-time-step transversal CNOT gate, which implements the
by classical error-correcting codes), a transversal logical logical CNOT gate between these two modules. (d) The lim-
controlled-NOT (CNOT) gate between logical qubits from ited intermodule connectivity of this example results in a serial
two code blocks in separate modules requires transversal slowdown for the two-qubit logical CNOT operation with an
(pairwise) physical CNOT gates between every single one illustrative, but suboptimal, circuit implementation. Substantial
of their constituent physical qubits (Fig. 2). Without a high additional temporal overheads are incurred (not shown here) if
degree of intermodule connectivity, the entanglement dis- all-to-all connectivity within the module is not available, as is
assumed for this example.
tribution between the two modules becomes a bottleneck
in the performance of the distributed system [38,39]. As
systems scale, this bottleneck can only be avoided if the
connectivity of qubits between modules scales with the
error-detecting capacity of the code. In codes where this estimates of quantum networking and computing applica-
error-detection capacity scales proportionally with module tions [39]—and yet connectivity between and within mod-
size, this poses a stringent mandate: an interconnect for ules is almost never accounted for, and typically not even
each physical qubit in the ideal case [40]. In large-scale mentioned, in current quantum resource-estimation liter-
quantum systems, the interconnectivity of modules may in ature. Simply put, these one-to-one physical entangling
fact dominate the temporal and spatial quantum resource operations between the modules must be parallelizable
010102-3
for good overall system performance. This observation

(a)
implies certain physical qubit capabilities in the ideal case;
it is argued below that maximal parallel entanglement
distribution in a modular architecture preferably implies
direct telecom optical access to each physical qubit. 1 2 3 4 5 6 7 8 9 10
With the assumption of modular quantum technologies
and high connectivity between modules for the purpose
of high-performance parallel physical operations across
modules, the natural question becomes: how should all of
the qubits be connected to each other? The key to fast Swapped- Infidelity
10 2
logical gates across modules is efficient nonlocal entan-
0 1 2 3 4
glement distribution. If, indeed, each qubit is designed (b)
to be easily entangled with other physical qubits from
other modules, presumably that qubit supports a physical
process that allows for nonlocal, as opposed to proximity-
1
based, entanglement generation. Under such assumptions,
connectivities both within and across modules are not
necessarily constrained to proximity-based qubit topolo-
gies and can be connected in whatever way offers the best
total system performance.
Most of today’s quantum architectures are built on the
paradigm of planar proximity-based entanglement gener-
FIG. 3. A comparison between two-qubit operations on
ation and aim to leverage a variant of the well-studied devices with (a) planar nearest-neighbor connectivity graphs and
low-connectivity quantum error correction (QEC) code (b) nonlocal connectivity. A CNOT gate between two distant
known as the surface code [41,42], to achieve fault tol- qubits on the planar graph requires sequential swap operations.
erance. Even these near-optimal codes for planar architec- Each consecutive operation accumulates errors and the overall
tures require tremendous development and resources. For fidelity drops precipitously, even over the small distances shown.
example, under optimistic settings [43] for the factoring With nonlocal connectivity, two-qubit operations are equivalent
algorithm, the surface code would require upward of 3000 across the graph and the device can scale with greater fidelity.
physical qubits for every fault-tolerant logical qubit. More-
over, in many planar architectures, the vast majority of codes are unshackled from the constraint of local connec-
logical instruction cycles require swapping qubits in order tivity, they can achieve truly astounding performance. A
to implement proximity-based multiqubit operations [44]. recent flurry of work [51–56] has culminated in the con-
Rejection of the assumption of planar connectivity struction of QLDPC code families with asymptotically
(Fig. 3) for high-connectivity SFTQ architectures can optimal code parameters, closing this long-open question
give rise to a significant reduction of time- and physical- after decades of relatively stagnant progress. In conjunc-
resource requirements [45]. Avoiding these and similar tion with previous work [57], these code families have
overheads by allowing for higher connectivity can signif- established that physical overheads could be driven sig-
icantly reduce the number of operations required [9,46]. nificantly lower than those observed for planar code archi-
Even the distillation of magic states, which is largely the tectures. The inherent properties of QLDPC codes unlock
preferred method for providing the missing element of uni- additional performance perks including single-shot oper-
versality needed for surface-code architectures, would be 2 ation [58–61], which removes the requirement to repeat
times more resource efficient using high-connectivity dis- QEC cycles to separate measurement errors from qubit
tillation algorithms [47], and near-term codes could have errors [62,63]. The work to implement logical operations
10 times lower overhead [48]. Unfortunately, neither a has only just begun but early signs point to methods on
moderate amount of long-range connectivity [49] nor small par with those of planar codes [64,65]. Surprisingly, no-
modules of qubits with substantial internal connectivity go theorems implying the need for costly methods to
offer a significant improvement over simple planar con- achieve universality may not even apply to certain QLDPC
nectivity. The connectivity must keep pace with the system codes [66].
size to realize substantial improvements [50]. Even more encouraging than these theoretical results
Quantum low-density parity-check (QLDPC) codes, on high-connectivity QLDPC codes has been the assess-
quantum analogues of the classical LDPC codes underpin- ment of their practical performance, which appears to
ning 5G telecommunications networks, are a class of codes match or outperform the planar code in essentially
characterized by low-weight stabilizer checks. When these every way. They offer similar error thresholds [49,67],
010102-4
similar check weights [48], and proven efficient decoding given physical qubit type should be selected for further
algorithms for large finite codes [68]. QLDPC codes that development with this long-term requirement in mind.
encode one logical qubit per every three physical qubits
have been demonstrated to saturate the quantum hash- III. BUILDING BLOCKS: SILICON, TELECOM,
ing bound under depolarizing noise (one of the biggest MEMORY, ENTANGLEMENT
points of pride of the planar surface code) using decoders A. Silicon
already in practical use today [69]. Even with rather small
system sizes (where the asymptotic behavior of QLDPC Silicon is a pinnacle material for both quantum and
codes need not be representative of actual performance) classical applications. It is the industrial standard for high-
and under nuanced noise models, QLDPC codes have been performance integrated electronics, as well as for low-loss
demonstrated to lower resource costs by over a factor of 10 high-density photonic integrated circuits both active and
in comparison to planar codes [70], even when extending passive. Spin qubits within silicon have also proven to be
to physical-resource estimation [71]. Certain families of exceptional quantum memories, with coherence times of
QLDPC codes offer constant overheads as the code scales approximately 3 h [93] and high fidelity [94]. From an
in size. For instance, explicit codes of around 1000 physi- industrial perspective, no other semiconductor is remotely
cal qubits support well over 100 logical qubits [71]. With close to silicon in its global impact, process engineering
constant-overhead QLDPC codes, the encoding rate could capabilities, cost effectiveness, and purity. The industrial
remain the same as the system scales horizontally. Unfortu- dominance and extensive development of silicon offers
nately for purely planar architectures, there is no substitute such incomparable competitive advantages that, histor-
for high connectivity—it is a fundamental requirement for ically, if a solution is found using silicon, the silicon
any system hoping to capitalize on this tantalizing poten- solution usually wins.
tial [33,72]. While full space-time logical-circuit-overhead
estimates are currently scarce, existing methods [48,73] B. Telecom photons
already upper bound this overhead at O(n) and there is Telecommunications-band (telecom) photonic flying
optimism that QLDPC logical circuit compiling can match qubits will be the backbone of any highly connected global
or beat state-of-the-art methods for surface codes [74] in quantum network and the backbone of modular quan-
practice. tum computers. Telecom photons can be flexibly routed
Even in advance of SFTQ, connectivity is also prov- with arbitrary connectivity to connect matter qubits both
ing critical in NISQ applications. For example, to build locally and remotely, with low loss in cryogenic compat-
a quantum repeater, connectivity allows for substantially ible waveguides and at room temperature using modern
more efficient entanglement distribution across many users telecommunications infrastructure. Although a number of
[75,76] and higher entanglement throughput via multipath research efforts are now advancing the transduction of
routing [77]. Other examples where connectivity improves other qubits into telecom photons [95–99], the overhead of
near-term applications include mapping circuits to hard- transduction processes can be avoided entirely by working
ware [78] and quantum image processing [79]. with quantum systems that interact with telecom photons
Taken together, the rate and quality of nonlocal entan- directly—such as the silicon T center. In this Perspec-
glement generation should be seen as the key characteristic tive, the claim is made that telecom photons are essential
dictating the ultimate performance and resource require- for high-connectivity quantum technologies at scale but
ments of applications using large-scale modular fault- are likely not practically sufficient on their own without
tolerant quantum systems. This is true for both large-scale a quantum memory.
networks leveraging quantum repeaters as well as modular
and horizontally scalable quantum supercomputers. In this
vision, quantum supercomputers and quantum networks C. Quantum memory
are both ultimately large-scale entanglement-distribution The central challenge with photonic qubits of all wave-
systems. Many physical qubit types can, in principle, lengths, across quantum computing and quantum network-
support nonlocal optical entanglement generation, including, is that they suffer from unavoidable loss. Each pho-
ing trapped ions [32,80], neutral atoms [81,82], quan- tonic process over optical links succeeds only a fraction of
tum dots [83], and solid-state color centers [84–90], and the time, and this is true even for telecom photons, which
the newly rediscovered and hence relatively unexplored have the lowest-loss photonic components available by
silicon T center [26,91,92]. While the architecture pro- far (e.g., switches) [100]. In large-scale high-connectivity
posed here could be applied to each, practical details that settings, each switch layer introduces yet further loss.
impact the rate and quality of the nonlocal entanglement- Quantum memories offer a straightforward path to protect
generation capabilities of a qubit can have truly dramatic against photon loss for both high-fidelity computing and
consequences on the final system performance and any networking applications [101].
010102-5
Quantum memories are necessary for the construc- (a)

tion of quantum repeaters with the highest functionalities h
Projective
[102,103]. In repeaters, ultralong quantum memory life- Readout
times are essential for high-fidelity qubit storage and the
h
Energy
coordination of photon signals with long time-of-flight
(long distance) connections, as well as in high-loss envi- A B C D
ronments where many entanglement-generation attempts
are necessary on average before photons arrive success-
fully. Solid-state color-center qubits with access to a spin e
Microwave
degree of freedom [27,84], as well as some optically Controls
e
active atoms in vacuum [28,29,104], and rare-earth emit-
ters [105–109] offer all of the aforementioned ideal quan-
tum functionality. Each of these physical qubit types offer
a direct high-quality photon interface into at least one (b) B C
long-lived [110] high-fidelity [111] universally control- T-Center Optical-Cavity
Spectrum Response
lable spin qubit [112]. Every physical qubit in this category
Amplitude
has a direct and dedicated photonic interface allowing for
parallel entanglement generation.
For high-connectivity quantum computing, a similar
architectural advantage is proposed here, where the pho- A D
tons distribute entanglement but do not process quantum
information directly: the processing is done within the
spin qubits. This design is inherently tolerant to pho- Energy
ton loss as probabilistic entanglement-generation attempts
can be repeated until success, without losing the spin- FIG. 4. (a) The energy levels of a T center. Transition C is
qubit state [113]. Memory qubits within the spin register optically pumped to generate entanglement and for readout. h,
must be insulated from decoherence due to remote entan- excited-state hole spin; e, ground-state electron spin. (b) The
glement attempts between communication qubits by a optical spectrum of the electron spin of the T center, overlaid
with optical cavity to enhance emission of the C transition.
suitable memory-protection scheme [114,115]. Essentially,
the high connectivity between matter qubits—which could
be physically arranged on planar chips—is provided by of quantum interfaces, which will inevitably compromise
entanglement carried by telecom photons and the arbitrary entanglement-distribution rates and fidelities.
connectivity that these photons provide.
Maximally entangled Bell pairs (BPs—pairs of qubits in
one of the four Bell states) of long-lived spins can be pro- D. Computing and networking with T centers
duced via a variety of photonic methods (see Fig. 6 for a Even in a loss-tolerant design, achieving efficient distri-
specific implementation [116,117]). Once entanglement in bution of entanglement at scale is a critical performance
the form of BPs is delivered to the spins, this entanglement metric for high-connectivity SFTQ. This further motivates
can be used as a resource to construct cluster states to be the adoption of solid-state spin-photon interfaces, ideally
consumed for computation within the measurement-based telecom color centers, which can be directly placed into
computing paradigm [118,119] or teleported gates in the low-loss integrated photonic circuits in silicon. Direct inte-
traditional gate model of quantum computing [120,121]. gration not only maximizes photon collection efficiency
Both the cluster state and teleported gate models of quan- from every single qubit but also enhances the photon emis-
tum computation allow for blind-computing applications sion rate and quality using routine photonic engineering,
over a network [15,122]. Below, the focus is on teleported while the spins, particularly long-lived nuclear spins, can
gates as an example implementation. perform at the high levels necessary for low-overhead
For this kind of quantum technology, the repeaters QEC.
and quantum computing modules of quantum networks The above considerations foretell a future where scal-
will be almost identical in their core construction, which able fault-tolerant quantum networks and scalable fault
reflects the fact that the key to efficient high-performance tolerant quantum computers will be the same core technol-
large-scale quantum networking and quantum computing ogy, with long-lived spins embedded into silicon integrated
is high-bandwidth high-connectivity entanglement distri- photonics, directly interfacing with telecom photons. In
bution. In the long term, it is suggested that this unified particular, the silicon T center [26,91,123] is proposed as
technology will outperform architectures with disparate the exemplary foundational quantum unit: in addition to
networking and computing cores that require extra layers direct telecom access, it contains one unpaired electron
010102-6
spin and up to three spin-1/2 nuclear spins (one hydrogen high-fidelity operations according to the vision set out in
and two carbon atoms; see Fig. 1), each of which serves as Sec. III.
a high-fidelity and long-lived qubit with good performance
comfortably above 1 K [26], where substantial cooling
power is available for large-scale modular processors. A. State preparation and measurement
Solid-state spin performance is largely determined by For readout and initialization, the electron spin can be
the host-crystal environment. High-fidelity performance projectively measured and initialized to high fidelity using
and ultralong coherence times [93,124,125] are common spin-dependent optical excitation [Fig. 4(a)] [26,91,132].
to many silicon spin centers, including T centers [26]. In The initialization and readout of the full T center spin
particular, the capability to isotopically purify silicon to register could be realized via successive rounds of high-
the nuclear spin-free 28 Si isotope has shown nuclear spin- fidelity state swaps between the nuclear spins and the elec-
coherence times of 3 h [93], with fidelities above 99.9% tron spin. Additionally, quantum nondemolition (QND)
[94]. T centers in bulk 28 Si samples show spin-coherence readout of the nuclear spins can be achieved by executing
times (T2 Hahn echo) of 2.1 ms for the electron and 1.1 s two-qubit CNOT gates with the electron spin as the target,
for the hydrogen nuclear spin; both spins have T1 lifetimes optically reading out the electron spin, and then repeating
far longer than a second [26]. as necessary [133]. In either case, memory qubits must be
The T center is one of very few spin qubits with a protected from decoherence due to optical excitation by a
direct telecom interface in silicon [109,126,127]. It can suitable scheme. Readout schemes without excitation are
efficiently interact with a pump laser pulse and emit a spin- possible for strong-coupled emitter-cavity systems [134].
entangled O-band photon. A photonic cavity can change The four spin qubits in the local register of a T cen-
the photonic environment around the T and enhance the ter admit independent and high-fidelity single-qubit and
emission of the photon into desired optical modes through multiqubit gates through standard electron and nuclear
the Purcell effect [128,129]. These modes are coupled to magnetic resonance techniques [26]. Silicon donor spins
optical waveguides with well-defined modes and there- have demonstrated excellent spin-selective control, long
fore spin-entangled photons can be accurately (and with dephasing rates (T∗2 ), and frequency stability, as well as
low loss) routed to their destination either through inte- long spin lifetimes a mere 10 nm from interfaces [94,135].
grated photonic waveguides or by coupling into optical
fiber. Optical fibers can connect T centers across multiple
chips, enabling a naturally modular and horizontally scal- B. Bell-pair entanglement
able architecture. As monolithic quantum systems grow Two-qubit gates rely on proximity-based or nonlocal
in size, they typically face increasingly difficult IO and entangling operations between qubits. Remote T centers
environmental challenges, e.g., resonator wiring for super- interact through photon-mediated entanglement. Each T
conducting transmon qubits. Modular scaling can relieve center can be optically triggered to emit single telecom O-
IO density challenges and environmental constraints that band photons [91], optionally entangled with the electron
emerge as systems grow in size. spin. To generate entanglement between spins, these pho-
tons must be indistinguishable. This means matching every
degree of freedom between two photons: wavelength,
IV. QUANTUM OPERATIONS
linewidth, timing, polarization, and so on. Importantly, the
The silicon T center, only recently rediscovered for wavelength of these photons can be tuned with either strain
quantum applications in 2020 [26], has the potential to [136] or electric fields [137,138] to compensate for the
enable high-performance SFTQ in the near and long term. variations in the local environments of the T centers.
It merges the advantages of silicon fabrication and scal- The indistinguishability of spin-entangled photons dic-
ability, telecom emission, and long-lived spin memories tates the quality of the generated BPs and hence the max-
into a quantum system uniquely adapted to a modular high- imum fidelity of the multiqubit operations. Highly indis-
connectivity architecture. Compared to color centers such tinguishable photons can be produced on demand when
as the NV or SiV in diamond, which have decades of the optical-transition linewidth of the emitter approaches
development as qubits and photon sources, there is rel- its lifetime-limited value, rather than being broadened by
atively little literature concerning the performance of T environmental noise. This can be dramatically assisted by
centers as a spin-photon interface. From published results, photonic engineering, specifically by spin-selective Pur-
it is clear that the T center has progressed rapidly from cell enhancement of the optical transition when on res-
characterization to device integration [26,91,92,130,131]. onance [Fig. 4(b)] with a high-Q photonic cavity [139].
Although there is significantly more work to be done, This offers faster emission rates, leading to larger lifetime
recent results have established the striking spin [26] and contributions to the linewidth and therefore more indistin-
optical [131,132] properties of the T center. Below, it guishable photons, as well as higher cyclicity (defined as
is described how the qubits of the T center perform the conservation of the spin state through an optical cycle).
010102-7
(a) (a) (c)

dt e
Detector Clicks
TCenter A
n Z n n n
=
e H
TCenter B
Time
n X n n n
(b)
(b) Accepted Bell Pairs
1.0
Entangling Rate (Hz)

Bell-Pair Fidelity
0.9
104
Rejected
0.8 Bell Pairs
103
0.7
Initialize Interfere Invert states Interfere Entangled
0.6
102
0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 FIG. 6. The two-qubit-gate mechanism. The electron spin is
Time Difference Between Clicks (ns) initialized in a superposition state and optically pumped to emit
a spin-entangled photon. Repeating this process and heralding on
FIG. 5. (a) Two T centers emit synchronized photons. The photon detector clicks produces photon-mediated electron spin-
detection times will vary due to the finite time width of the pho- spin entanglement, which is consumed in remote T centers to
ton wave packets. (b) The detection time difference dt can be effect a CNOT gate between two nuclear spins.
used to herald only high-fidelity BPs. As the heralding thresh-
old grows more stringent, the BP-generation rate decreases. The
model curves have been calculated using reasonable projections
of system parameters from current performance.
The triggered emission of two indistinguishable high-

This architecture is quite tolerant to low yields for both purity spin-entangled photons can be used to project the
the cavities (frequency, Q factor) and the color centers participating electron spins into a maximally entangled
(frequency, coupling strength). With switchable optical Bell-pair spin state [145,146]. An illustrative example of
networks, it is possible to select high-performing qubits the BP-generation protocols that exist, the Barrett-Kok
from a larger pool. A quantum network of qubits is, for [116] protocol family, is notable in that it is which-way
this reason, far less sensitive to imperfect yields than symmetric, does not require interferometric stability of the
fixed nearest-neighbor processors. To achieve high indis- setup, and heralds success with minimal false positives.
tinguishability, the frequency of the optical cavities and The Barrett-Kok protocol proceeds as follows (Fig. 6).
the color centers must coincide within a relatively nar- The electron spins of two separated T centers are pre-
row range defined by the Purcell-enhanced linewidth. To pared in a superposition of spin-up and spin-down and
improve upon initially low yields resulting from fabri- triggered to emit a photon resonant with the spin-up transi-
cation variation, tunability, such as Stark shifting of the tion. The photonic modes are interfered on a beam splitter
optical transition or cavity tuning or trimming, can be and detected. Then, the spin states are inverted and the
employed. Fortunately, the precision of silicon nanofabri- emission and detection is repeated. Using exactly one pho-
cation is exceptional and many of these ideas have already ton detection in each of the first and second optical cycles
been explored [140–143]. as a herald signal, the spins are projected into a maximally
Integration into photonic cavities has already led to entangled BP.
demonstrations of > 20 times reductions in single T center Although photon loss can cause an individual attempt
excited-state lifetimes compared to the 940 ns unenhanced to fail, success is heralded and the entire process can be
value [130,131,138,144], enhancing indistinguishability repeated until success, with small degradation in the local
and increasing maximum photon rates. Instantaneous nuclear spin coherence for each attempt (see below). More-
linewidths for single T centers only 5 times larger than over, for this protocol, higher-fidelity BPs can be obtained
the lifetime-limited value have also been measured [130, at the expense of lower entanglement rates by shift-
131,138]. Additional improvements toward high indis- ing the heralding threshold, the time difference between
tinguishability include enhanced fabrication methods to detector clicks, e.g., using time-bin filtering for time-
reduce the optical spectral diffusion. frequency mismatch (Fig. 5) [147].
010102-8
C. Teleported gates detector module. Here, the silicon-on-insulator (SOI) plat-

The teleported-gate model has a number of advan- form can be leveraged to drive on-chip device integration
tages at scale. Teleportation delivers two-qubit gates on to a stage where photons never need to leave the pho-
demand, after coincident identical photons are heralded. tonic integrated circuit, maximizing the photon collection
This inherently eliminates accidental multiqubit operations efficiency and entanglement heralding rates. Furthermore,
and crosstalk that can occur when two-qubit interactions optical interconnects off-chip allow connections to outside
are “always on.” Furthermore, qubits coupled to com- networks and modular scaling of computing power.
mon local environments are susceptible to correlated errors A visual summary of this physical architecture is shown
that can dramatically degrade QEC performance [148– in Fig. 1. T centers are placed into photonic cavities that
150]. Since the teleported gate is a remote operation, each are directly coupled into low-loss photonic waveguides.
T center can occupy a unique local environment, thus Emitted photons are routed on-chip through switches and,
minimizing correlations in noise. optionally, through high-efficiency optical IO ports into
Once the BP entanglement is heralded and delivered one of the many optical fibers. An optical-switch fabric
to the two electron spins of the T centers, that BP can governed by run-time electronics guides T center emis-
be consumed in a teleported gate sequence to apply a sion to single-photon detectors and Bell-state measurement
nonlocal multiqubit gate between the two nuclear spins modules within the cryostat.
of the T centers using only local operations and classi- The ability to then link distinct photonic chips through
cal feed-forward operations (see Fig. 6) [121,151,152]. off-chip interconnects enables modularizing and distribut-
Alternatively, a series of successful BPs can be distilled ing IO across multiple chips to horizontally scale the
[153–155] through local measurements and gates onto a system size both within one and amongst many cryostats.
remote memory-qubit BP [156,157], with fidelity beyond Due to the low loss of telecom photons in optical fiber, the
the photon indistinguishability, before this distilled pair is cryostats can be meters to kilometers apart.
consumed to deliver higher-fidelity teleported gates. A reg-
ister of multiple-spin memory qubits such as the internal A. On-chip scalability
nuclear spins of the T center allows a tiered distillation pro-
tocol [158]. Distillation can prioritize the dominant error T centers are atomic in scale and can be localized
pathway of the teleported gate, yielding, e.g., a teleported within small photonic components without direct optical
CNOT process fidelity exceeding the heralded or distilled crosstalk, making it possible to pattern millions of indi-
BP fidelity [146]. vidually addressable qubits per chip [91]; however, the
The approach outlined here uses the electron as a spin- ultimate qubit count per chip will be governed by the larger
photon interface to generate and distribute entanglement footprint of active photonic components, control electron-
and nuclear spins for processing and memory. A configu- ics, and signal IO. For such an architecture, the on-chip
ration must be chosen that protects the memory qubits dur- qubit density will not be limited by the qubit geometry,
ing entanglement-distribution attempts. Repeated attempts as opposed, e.g., to qubits connected by superconducting
can, in general, degrade the information stored in memory traces.
by perturbing the nuclear spin states [84]. The low thermal conductivity of optical fiber enables
Protection techniques include working in a decoherence- incredibly high-count optical IO into cryogenic environ-
free subspace [108], aligning an external magnetic field ments (e.g., 37 000 optical fibers in Ref. [161]). High-
along one of the principal axes of the hyperfine ten- density optical connections to integrated photonic devices
sor [138], and reducing the optical excited-state lifetime have also been demonstrated in cryogenic environments
[159,160]. With sufficient protection, BP generation does [162]. Furthermore, the T center achieves operating per-
not materially affect data stored in the nuclear spins and formance (electron spin lifetime, optical linewidth) at only
so the teleported gate operations are made photon-loss tol- 1–2 K [26]. This relatively high operation temperature cir-
erant. Moreover, because of the long memory, this system cumvents the need for dilution-refrigeration technologies
works well for distillation and the creation of large entan- and permits increased cooling power to counteract the heat
gled states, even after minimizing the required number of load generated by the signal IO. As high-density photonic
attempts by minimizing optical losses. IO continues to develop, it is expected that each chip could
be able to support upward of 4000 fibers, with 1000-plus
qubits per system and beyond.
V. SCALABLE QUANTUM TECHNOLOGY
T center qubits can be controlled using electrical sig-
Scaling the number of interacting qubits in this archi- nals segmented into global fields and local fields. Global
tecture requires mediating entanglement across a grow- fields include a static magnetic field B0 to define the T cen-
ing network of qubits. Enabling multiqubit operations ter spin Hamiltonian and a spin control field B1 that can
between any two T centers amounts to routing their address qubits individually using global microwave con-
indistinguishable photons to the same beam-splitter and trol techniques [163–167]. Local control fields include the
010102-9
electric field E, a trimming magnetic field δB0 , and optical- All-to-all Connectivity
switch and detector signals. Multiplexing strategies reduce (networked)
the electrical line count for all signals, particularly the local
control [168] and detector [169] signals.
B. Local quantum networks

As mentioned in Sec. III, quantum computing and quan-
tum networking are essentially the same in the architecture Chip 1 Chip 3
(local) (remote)
presented here. The challenge that is overcome through
this approach to SFTQ technologies—high-bandwidth
high-connectivity entanglement distribution over lossy Chip 2
optical networks—is the same challenge faced in scaling (local)
up quantum communication networks.
For this high-connectivity architecture, horizontal scal- FIG. 7. The nonlocal connectivity of the system works in the
ing of quantum resources is achieved by racking multiple same way when entangling qubits across the chips in the same
units adjacent to each other into a modular local quan- cryostat 10 m away or 100 km away.
tum network of qubits (Fig. 1). Where additional network
topology, interchip links, or connections between distant T
communications over distances as large as 1000 km [172],
centers are needed, optical IO can route photons to switch
optical quantum repeaters remain the key to unlocking
networks. Because optical photons can noiselessly traverse
global quantum networking applications [102].
from cryogenics to room temperature without transduc-
Fiber-coupled T-center processor nodes contain telecom
tion, the switching network can be placed either within the
optical quantum memories that are capable of supporting
cryostat or at room temperature.
long-distance communication over low-loss fiber links and
This design allows for arbitrary and flexible connec-
are able to generate, store, and process optically entan-
tivity using switchable telecommunications hardware at
gled spin qubits. Hence, they can be naturally operated
room temperature with simple connections to cryogen-
as a telecom-wavelength quantum repeater to realize a
ics via optical fiber. Additionally, on-chip switching and
long-distance scalable multiuser quantum Internet, with
multiplexing can then be used to scale the qubit count
the potential to securely distribute entanglement or connect
within each chip beyond limitations imposed by the IO
quantum processors between any two users on the global
fiber count.
network (Fig. 1).
For the architecture described in this paper, T centers
As an initial implementation, the nested quantum
from distinct cryogenic units can be entangled in the same
repeater protocol [173] creates long-distance entangle-
way as adjacent T centers (Fig. 7). Put another way, the
ment by heralded entanglement generation over shorter
multiqubit operations between different cryostats can be
links and entanglement swapping between adjacent nodes
similar in performance to the multiqubit operations within
followed by entanglement distillation. The steps are
a chip. This design maximizes the performance of impor-
repeated between successive, now twice as long, entan-
tant distributed logical quantum algorithms [170]. The
gled links until the end nodes are entangled. The individual
physical and logical qubits will have effectively all-to-all
entanglement-generation steps taken during this procedure
connectivity by routing photons to common beam split-
are exactly the same as local distributed SFTQ computing
ters, by teleporting qubit states to new neighborhoods of
using the proposed architecture, allowing implementation
connectivity, or through entanglement-swapping protocols
by T center processors. This repeater protocol belongs to
[16].
the so-called first generation of quantum repeaters [174].
Here, the loss errors are addressed by heralded entangle-
C. Quantum repeaters ment generation and operational errors are addressed by
The broad adoption of a quantum Internet [171] is ham- entanglement distillation, which is a specific type of error
pered by issues that arise when quantum communication correction.
channels are extended beyond local (approximately 10 km) Looking to the second generation of quantum repeaters,
distances: the communication rate of point-to-point terres- one can leverage the ability of T center processors to per-
trial fiber links drops exponentially with distance due to form multiqubit encoding and logical operations between
the signal attenuation intrinsic in even the best available nodes. As before, loss errors on individual photons are
optical fibers. Similarly, beam divergence, atmospheric addressed by heralded entanglement generation but the
absorption, and turbulence impose fundamental losses and operational errors are addressed by logical encoding at
noise on free-space telescopic links. And even though each node, namely, with codes permitting transversal
telecom networks have achieved point-to-point quantum two-qubit operations [175], i.e., the teleported CNOT
010102-10
implemented in Fig. 6. Each node possesses one or more (a) (b) (d)
logical qubits, which become logically entangled to neigh- e
boring logical qubits through entanglement distribution.
n H
Subsequent logical entanglement swapping allows the end-
node logical qubits to be entangled, similar to the end-node LC1 LC2
LC1
physical qubits of the first generation of repeaters. This LC2
second generation uses the scalability and connectivity of
T centers in individual nodes to distribute error-corrected (c)
logical qubits over long distance.
For these reasons, the proposed T center-based archi-
tecture is not only best suited to tackle the challenges of
SFTQ computing but also offers a very attractive platform
for implementing large-scale quantum networks and their L E E L
various applications.
Encoded
Laser Client
VI. QUANTUM NETWORKING APPLICATIONS Initialize Interfere Invert Interfere Loaded
A truly scalable network, extendable both in dis-
tance and number of users, unlocks powerful applications FIG. 8. (a) A protocol for loading of a time-bin photonic qubit
[176,177], such as quantum key distribution (QKD), dis- state into the spin memory of a T center. (b) A quantum circuit
tributed quantum computing, blind quantum computing that utilizes the protocol in (a) to generate a cryoptographic key
for two users (laser clients, LCs) connected to the same hub (c).
[15], and enhanced sensing. Here, the focus is on two
Loading photons from two users and then performing a Bell-
applications, blind quantum computing and quantum key state measurement on them computes the parity of their states,
distribution, and we discuss how these could be naturally allowing secret key generation in MA-MDI QKD.
implemented in the proposed architecture.
A. Memory-assisted measurement-device independent than existing trusted-node networks [181]. At the same
quantum key distribution time, the memory assistance provides improved key-rate
Quantum key distribution is the quantum networking scaling [179] due to the ability to store the clients’ quantum
application nearest to widespread adoption. A first imple- information in the T center spins. This means that the client
mentation of a QKD network using T centers consists of photons do not have to arrive simultaneously; their loading
users possessing laser clients (LCs) that are optically con- to the spins is heralded and only once both are loaded is the
nected to a quantum processor containing T centers (see Bell-state measurement performed between spins.
Fig. 8), possibly in a hub-spoke configuration (a network Within this architecture, each hub will support thou-
with many outlying nodes and one central hub connected sands of users in a hub-spoke model, well exceeding the
to each outlying node). Users can encode quantum infor- usual point-to-point connections of currently commercially
mation onto weak coherent pulses (e.g., using time-bin
encoding) and send this to the processor hub. There, using
the spin-photon entangled state of the T center and quan- (a) (b)
tum teleportation, the encoded quantum state is loaded to e
a local spin state [178] for storage and further processing. n H
T Center A
Multiple hubs could be interconnected to extend range and
capacity (Fig. 9). e LC1
T Center B
The proposed networking configuration realizes n
memory-assisted measurement-device-independent (MA- LC1
LC2
MDI) QKD [179]. Within this protocol, a secret key
LC2
between two users is established by first loading each
user’s states onto the qubits of the hub, followed by a Bell-
FIG. 9. (a) A quantum circuit for loading qubit states from dis-
state measurement between these two qubits. The result of tant laser clients to two different quantum hubs and performing a
that measurement is announced to the users, who use this nonlocal Bell-state measurement between the two T center spins.
information to extract a shared secret key. The teleported CNOT (middle) is mediated by entanglement dis-
The MA-MDI-QKD protocol eliminates detector-side- tribution between the two hubs. (b) Because the CNOT between
channel attacks [180], allowing the hub to operate as an the two T centers is teleported, this circuit could be used to gener-
untrusted node. Entanglement-based linking of nodes can ate cryptographic keys between two users connected to different
also be untrusted, providing significantly more security cryostats that are linked by telecom fiber.
010102-11
available QKD systems. Performing entanglement distri- quantum systems, potentially distributed over global dis-
bution between hubs as in Fig. 9 allows implementing the tances, will be the ultimate version of horizontally scaled
same MA-MDI-QKD protocol in a distributed setting. The quantum information processing and networking. Under-
shared entanglement can be generated independently of the pinning this view is the observation that combining
qubit loading events and then consumed to implement the quantum computing and quantum networking technology
nonlocal Bell-state measurement. removes the fundamental obstacles to scale that each of
Whether connected to a single hub or a repeater chain, these technologies are facing in isolation, i.e., truly scal-
each user requires only a simple cost-effective source: a able architectures are horizontally scalable. Namely, to
room-temperature laser attenuated below the single-photon unlock quantum networks at scale, one needs to develop
level. It is equipped with a modulator able to produce time- quantum repeaters that, in the high-performance limit, are
bin-encoded photonic qubits, which are spectrally matched essentially fault-tolerant quantum computing modules, and
to T centers in the hub. Such a device is very similar to to unlock truly scalable quantum computing one needs
the telecom transceivers of a modern data center and the to leverage the entanglement-distribution capabilities of
photonic qubits produced by these sources can be simi- quantum networking to link quantum computing modules
larly routed through existing data-center or metropolitan into quantum supercomputers.
telecom fiber to the hub. Given that high-bandwidth high-quality entanglement
distribution ultimately sets the performance of both scal-
able (modular) fault-tolerant quantum computing and net-
B. Blind quantum computing working, quantum systems should be engineered to opti-
Although quantum repeaters allow for long-distance mize entanglement distribution. Under this model, quan-
quantum computation and telecom-wavelength photons tum computing and quantum networking are (in the ideal
allow for modular quantum design, quantum computers case) ultimately the same fundamental technology.
may remain constrained by the complex hardware required This Perspective proposes a scalable quantum (network-
to protect and manipulate quantum states. This restric- ing and computing) architecture with this end goal in
tion suggests that quantum supercomputers are expected mind. It suggests a specific implementation using telecom
to start, and perhaps remain, as network-accessible large- color centers in silicon—namely, the T center—but this
scale devices. Cloud access over the classical Internet may architectural model is broadly applicable to many qubit
not be able to provide the necessary privacy for all poten- systems. Because of the high connectivity offered by the
tial end-user applications. Additionally, the operators of spin-photon interface, this architecture can take advantage
the quantum computers would have access to both the data of fixed- and low-overhead QLDPC codes to deliver fault
and the algorithm run on it. Similar concerns for classi- tolerance. Due to the integrated silicon photonics plat-
cal cloud computing have galvanized research in the field form, thousands of qubits can be fabricated and addressed
of homomorphic encryption [182], where a computer per- on a single chip with optical and electronic control and
forms logic directly on encrypted data without ever gaining routing, and modules can be connected together across
the capability to decrypt and learn the underlying infor- existing global telecommunications infrastructure without
mation. Blind quantum computing [15], which may be any transduction losses. The use of a T center network
viewed as a quantum analogue to homomorphic encryp- to distribute verified quantum entanglement allows for
tion, allows users to perform arbitrary computations using device-independent networking protocols, providing the
remote quantum processing resources, while restricting the ultimate protection against eavesdropper attacks, as well
computer from having meaningful access to both the data as other high-value applications leveraging entanglement
and the algorithm. In this highly connected architecture, distribution, such as blind computing. Taken together, the
the same process of qubit loading employed for MA-MDI future for truly scalable quantum technology is bright.
QKD can also be employed to realize remote and blind Declaration of competing interest: Stephanie Simmons
quantum computation, without a change of user hardware. is the founder and Chief Quantum Officer at Photonic Inc.,
This is due to the intrinsic spin-photon interface of the a company for quantum technologies based on photoni-
T center, which allows the loading of arbitrary external- cally linked silicon spin qubits.
user qubits into the computation in a heralded way. Loaded
qubits are largely used to direct the computation remotely,
while some are reserved as checks to confirm the operation
of the computer [183]. [1] S. Jordan, Quantum Algorithm Zoo, https://quantumal-
gorithmzoo.org/.
VII. CONCLUSIONS [2] V. von Burg, G. H. Low, T. Häner, D. S. Steiger, M.
Reiher, M. Roetteler, and M. Troyer, Quantum comput-
The highest-value quantum applications known today ing enhanced computational catalysis, Phys. Rev. Res. 3,
require fault-tolerant capabilities at scale. Modular 033055 (2021).
010102-12
[3] I. H. Kim, Y.-H. Liu, S. Pallister, W. Pol, S. Roberts, [20] A. Peruzzo, J. McClean, P. Shadbolt, M.-H. Yung, X.-Q.
and E. Lee, Fault-tolerant resource estimate for quantum Zhou, P. J. Love, A. Aspuru-Guzik, and J. L. O’brien,
chemical simulations: Case study on Li-ion battery elec- A variational eigenvalue solver on a photonic quantum
trolyte molecules, Phys. Rev. Res. 4, 023019 (2022). processor, Nat. Commun. 5, 4213 (2014).
[4] J. J. Goings, A. White, J. Lee, C. S. Tautermann, M. [21] E. Farhi, J. Goldstone, and S. Gutmann, A quantum
Degroote, C. Gidney, T. Shiozaki, R. Babbush, and N. approximate optimization algorithm, arXiv:1411.4028.
C. Rubin, Reliably assessing the electronic structure of [22] E. Farhi, J. Goldstone, S. Gutmann, and M. Sipser, Quan-
cytochrome P450 on today’s classical computers and tum computation by adiabatic evolution, arXiv:quant-
tomorrow’s quantum computers, Proc. Natl. Acad. Sci. ph/0001106.
119, e2203533119 (2022). [23] J. W. Z. Lau, K. H. Lim, H. Shrotriya, and L. C. Kwek,
[5] L. Kee, RSA is dead—we just haven’t accepted it yet, NISQ computing: Where are we and where do we go?,
https://www.forbes.com/sites/forbestechcouncil/2021/ AAPPS Bulletin 32, 27 (2022).
05/06/rsa-is-dead—we-just-haventaccepted-ityet/. [24] Y. Wang, S. Simsek, T. M. Gatterman, J. A. Gerber, K.
[6] A. Raya and K. Mariyappn, in 2020 Fifth International Gilmore, D. Gresh, N. Hewitt, C. V. Horst, M. Matheny,
Conference on Research in Computational Intelligence T. Mengle, B. Neyenhuis, and B. Criger, Fault-tolerant
and Communication Networks (ICRCICN) (2020), p. 130. one-bit addition with the smallest interesting colour code,
[7] P. Campbell, M. G. Shepherd, and D. Shepherd, Soliloquy: arXiv:2309.09893.
A cautionary tale, Tech. Rep. [25] M. Mosca and M. Piani, 2022 Quantum Threat Timeline
[8] M. Webber, V. Elfving, S. Weidt, and W. K. Hensinger, Report, Tech. Rep.
The impact of hardware specifications on reaching quan- [26] L. Bergeron, C. Chartrand, A. T. K. Kurkjian, K. J. Morse,
tum advantage in the fault tolerant regime, AVS Quantum H. Riemann, N. V. Abrosimov, P. Becker, H. J. Pohl,
Sci. 4, 013801 (2022). M. L. W. Thewalt, and S. Simmons, Silicon-integrated
[9] E. Gouzien, D. Ruiz, F.-M. L. Regent, J. Guillaud, and N. telecommunications photon-spin interface, PRX Quantum
Sangouard, Computing 256-bit elliptic curve logarithm in 1, 020301 (2020).
9 hours with 126133 cat qubits, arXiv:2302.06639. [27] D. D. Awschalom, R. Hanson, J. Wrachtrup, and B. B.
[10] W. Beullens, Breaking rainbow takes a weekend on a lap- Zhou, Quantum technologies with optically interfaced
top, Cryptology ePrint Archive, Paper 2022/214 (2022), solid-state spins, Nat. Photonics 12, 516 (2018).
https://eprint.iacr.org/2022/214. [28] B. B. Blinov, D. L. Moehring, L.-M. Duan, and C. Mon-
[11] W. Castryck and T. Decru, An efficient key recovery roe, Observation of entanglement between a single trapped
attack on SIDH (preliminary version), Cryptology ePrint atom and a single photon, Nature 428, 153 (2004).
Archive, Paper 2022/975 (2022), https://eprint.iacr.org/ [29] T. Wilk, S. C. Webster, A. Kuhn, and G. Rempe, Single-
2022/975. atom single-photon quantum interface, Science 317, 488
[12] E. Dubrova, K. Ngo, and J. Gärtner, Breaking a fifth- (2007).
order masked implementation of CRYSTALS-Kyber by [30] J. L. Gustafson, Reevaluating Amdahl’s law, Commun.
copy-paste, Cryptology ePrint Archive, Paper 2022/1713 ACM 31, 532 (1988).
(2022), https://eprint.iacr.org/2022/1713. [31] K. Asanović, R. Bodik, B. C. Catanzaro, J. J. Gebis, P.
[13] R. Bavdekar, E. Jayant Chopde, A. Agrawal, A. Bhatia, Husbands, K. Keutzer, D. A. Patterson, W. L. Plishker, J.
and K. Tiwari, in 2023 International Conference on Infor- Shalf, S. W. Williams, and K. A. Yelick, The Landscape
mation Networking (ICOIN) (IEEE Computer Society, of Parallel Computing Research: A View from Berkeley,
Bangkok, 2023), p. 146. Tech. Rep. UCB/EECS-2006-183 (EECS Department,
[14] C. H. Bennett and G. Brassard, in Proceedings of the University of California, Berkeley, California, 2006).
International Conference on Computers, Systems & Sig- [32] C. Monroe, R. Raussendorf, A. Ruthven, K. R. Brown,
nal Processing (Steering Committee, Bangalore, 1984), P. Maunz, L. M. Duan, and J. Kim, Large-scale modular
p. 175. quantum-computer architecture with atomic memory and
[15] A. Broadbent, J. Fitzsimons, and E. Kashefi, in 2009 50th photonic interconnects, Phys. Rev. A 89, 022317 (2014).
Annual IEEE Symposium on Foundations of Computer [33] N. Baspin and A. Krishna, Connectivity constrains quan-
Science (IEEE, Atlanta, GA, USA, 2009), p. 517. tum codes, Quantum 6, 711 (2022).
[16] M. Caleffi, M. Amoretti, D. Ferrari, D. Cuomo, J. Illiano, [34] R. Van Meter and S. J. Devitt, The path to scalable
A. Manzalini, and A. S. Cacciapuoti, Distributed quantum distributed quantum computing, Computer 49, 31 (2016).
computing: A survey, arXiv:2212.10609. [35] D. Ferrari, A. S. Cacciapuoti, M. Amoretti, and M. Cal-
[17] P. Kómár, E. M. Kessler, M. Bishof, L. Jiang, A. S. effi, Compiler design for distributed quantum computing,
Sørensen, J. Ye, and M. D. Lukin, A quantum network IEEE Trans. Quantum Eng. 2, 1 (2021).
of clocks, Nat. Phys. 10, 582 (2014). [36] D. Cuomo, M. Caleffi, K. Krsulich, F. Tramonto, G.
[18] Z. Zhang and Q. Zhuang, Distributed quantum sensing, Agliardi, E. Prati, and A. S. Cacciapuoti, Optimized com-
Quantum Sci. Technol. 6, 043001 (2021). piler for distributed quantum computing, ACM Trans.
[19] L. Jiang, J. M. Taylor, A. S. Sørensen, and M. D. Lukin, Quantum Comput. 4, 1 (2023).
Distributed quantum computation based on small quantum [37] D. Stahlke and R. B. Griffiths, Entanglement requirements
registers, Phys. Rev. A—At., Mol., Opt. Phys. 76, 062323 for implementing bipartite unitary operations, Phys. Rev.
(2007). A 84, 032316 (2011).
010102-13
[38] R. Nigmatullin, C. J. Ballance, N. de Beaudrap, and S. [49] M. A. Tremblay, N. Delfosse, and M. E. Beverland,
C. Benjamin, Minimally complex ion traps as modules for Constant-overhead quantum error correction with thin
quantum communication and computing, New J. Phys. 18, planar connectivity, Phys. Rev. Lett. 129, 050504 (2022).
103028 (2016). [50] N. Baspin and A. Krishna, Quantifying nonlocality: How
[39] Y. Li and S. C. Benjamin, Hierarchical surface code for outperforming local quantum codes is expensive, Phys.
network quantum computing with modules of arbitrary Rev. Lett. 129, 050505 (2022).
size, Phys. Rev. A 94, 042303 (2016). [51] J.-P. Tillich and G. Zémor, Quantum LDPC codes with
[40] As a prototypical example, consider the case where each positive rate and minimum distance proportional to the
module accommodates a code block of a large n-qubit square root of the blocklength, IEEE Trans. Inf. Theory
error-correcting code with distance d. Here, it is argued 60, 1193 (2013).
that a fault-tolerant entangling logical operation between [52] M. B. Hastings, J. Haah, and R. O’Donnell, in Proceed-
these modules will always require O(d) simultaneous ings of the 53rd Annual ACM SIGACT Symposium on The-
physical entangling operations between the two modules. ory of Computing (Association for Computing Machinery,
Performing entangling logical operations between two Italy, 2021), p. 1276.
code blocks necessarily involves at least O(d) physical [53] N. P. Breuckmann and J. N. Eberhardt, Balanced product
entangling operations, as the support of a logical Pauli quantum codes, IEEE Trans. Inf. Theory 67, 6653 (2021).
in each module must grow by O(d) in the other mod- [54] P. Panteleev and G. Kalachev, Quantum LDPC codes with
ule, though these could be mostly within a code block in almost linear minimum distance, IEEE Trans. Inf. Theory
principle. If this operation is performed fault tolerantly, 68, 213 (2021).
any one single-qubit Pauli error cannot grow a support [55] A. Leverrier and G. Zémor, in 2022 IEEE 63rd Annual
larger than O(1) in either module. Since a logical Pauli Symposium on Foundations of Computer Science (FOCS)
operation in one module consists of the tensor product (IEEE, Denver, Colorado, USA, 2022), p. 872.
of O(d) single-qubit Pauli operations, O(d) single-qubit [56] P. Panteleev and G. Kalachev, in Proceedings of the 54th
Pauli operations in one module must each gain distinct Annual ACM SIGACT Symposium on Theory of Comput-
support from one another in the other module. This implies ing (Association for Computing Machinery, Rome, Italy,
that there needs to be O(d) entangling operations across 2022), p. 375.
the two modules. Since we would like the total circuit [57] D. Gottesman, Fault-tolerant quantum computation with
depth for the logical operation to be as close to O(1) as constant overhead, arXiv:1310.2984.
possible to maintain fault tolerance, we should be capable [58] N. P. Breuckmann and V. Londe, Single-shot decoding of
of performing these O(d) module-to-module entangling linear rate LDPC quantum codes with high performance,
operations simultaneously. For example, this is how lattice arXiv:2001.03568.
surgery will be accomplished in planar QEC architectures. [59] A. O. Quintavalle, M. Vasmer, J. Roffe, and E. T. Camp-
More importantly, if we consider an optimally scaling bell, Single-shot error correction of three-dimensional
code where d = O(n), this already implies that we need homological product codes, PRX Quantum 2, 020340
to be able to apply O(n) entangling operations in paral- (2021).
lel between two n-qubit modules. A classic example of a [60] O. Higgott and N. P. Breuckmann, Improved single-
logical operation with this requirement is the transversal shot decoding of higher-dimensional hypergraph-product
CNOT operation in CSS codes. codes, PRX Quantum 4, 020332 (2023).
[41] S. B. Bravyi and A. Y. Kitaev, Quantum codes on a lattice [61] S. Gu, E. Tang, L. Caha, S. H. Choe, Z. He, and A.
with boundary, arXiv:quant-ph/9811052 (1998). Kubica, Single-shot decoding of good quantum LDPC
[42] A. Kitaev, Fault-tolerant quantum computation by anyons, codes, arXiv:2306.12470.
Ann. Phys. 303, 2 (2003). [62] H. Bombín, Single-shot fault-tolerant quantum error cor-
[43] C. Gidney and M. Ekerå, How to factor 2048 bit RSA inte- rection, Phys. Rev. X 5, 031043 (2015).
gers in 8 hours using 20 million noisy qubits, Quantum 5, [63] E. T. Campbell, A theory of single-shot error correction
433 (2021). for adversarial noise, Quantum Sci. Technol. 4, 025006
[44] M. McEwen, D. Bacon, and C. Gidney, Relaxing hard- (2019).
ware requirements for surface code circuits using time- [64] A. O. Quintavalle, P. Webster, and M. Vasmer, Parti-
dynamics, arXiv:2302.02192. tioning qubits in hypergraph product codes to implement
[45] E. T. Campbell, B. M. Terhal, and C. Vuillot, Roads logical gates, arXiv:2204.10812.
towards fault-tolerant universal quantum computation, [65] A. Cowtan and S. Burton, CSS code surgery as a universal
Nature 549, 172 (2017). construction, arXiv:2301.13738.
[46] D. Litinski and N. Nickerson, Active volume: An archi- [66] P. Webster, M. Vasmer, T. R. Scruby, and S. D. Bartlett,
tecture for efficient fault-tolerant quantum computers with Universal fault-tolerant quantum computing with stabi-
limited non-local connections, arXiv:2211.15465 (2022). lizer codes, Phys. Rev. Res. 4, 013092 (2022).
[47] S. Bravyi and J. Haah, Magic-state distillation with low [67] J. Roffe, D. R. White, S. Burton, and E. Campbell, Decod-
overhead, Phys. Rev. A 86, 052329 (2012). ing across the quantum low-density parity-check code
[48] L. Z. Cohen, I. H. Kim, S. D. Bartlett, and B. J. landscape, Phys. Rev. Res. 2, 043423 (2020).
Brown, Low-overhead fault-tolerant quantum computing [68] P. Panteleev and G. Kalachev, Degenerate quantum LDPC
using long-range connectivity, Sci. Adv. 8, eabn1717 codes with good finite length performance, Quantum 5,
(2022). 585 (2021).
010102-14
[69] S. Yang and R. Calderbank, Spatially-coupled QDLPC [86] M. Raha, S. Chen, C. M. Phenicie, S. Ourari, A. M.
codes, arXiv:2305.00137. Dibos, and J. D. Thompson, Optical quantum nondemo-
[70] S. Bravyi, A. W. Cross, J. M. Gambetta, D. Maslov, P. lition measurement of a single rare earth ion qubit, Nat.
Rall, and T. J. Yoder, High-threshold and low-overhead Commun. 11, 1 (2020).
fault-tolerant quantum memory, arXiv:2308.07915. [87] G. Wolfowicz, F. J. Heremans, C. P. Anderson, S. Kanai,
[71] Q. Xu, J. Ataides, C. A. Pattison, N. Raveendran, D. Blu- H. Seo, A. Gali, G. Galli, and D. D. Awschalom, Quantum
vstein, J. Wurtz, B. Vasic, M. D. Lukin, L. Jiang, and H. guidelines for solid-state spin defects, Nat. Rev. Mater. 6,
Zhou, Constant-overhead fault-tolerant quantum compu- 906 (2021).
tation with reconfigurable atom arrays, arXiv:2308.08648. [88] J. A. Martínez, R. A. Parker, K. C. Chen, C. M. Purser,
[72] N. Delfosse, M. E. Beverland, and M. A. Tremblay, L. Li, C. P. Michaels, A. M. Stramma, R. Debroux, I. B.
Bounds on stabilizer measurement circuits and obstruc- Harris, M. H. Appel, E. C. Nichols, M. E. Trusheim, D.
tions to local implementations of quantum LDPC codes, A. Gangloff, D. Englund, and M. Atatüre, Photonic indis-
arXiv:2109.14599 (2021). tinguishability of the tin-vacancy center in nanostructured
[73] D. Gottesman, Fault-tolerant quantum computation with diamond, Phys. Rev. Lett. 129, 173603 (2022).
constant overhead, arXiv preprint arXiv:1310.2984 (2013). [89] C. Babin et al., Fabrication and nanophotonic waveguide
[74] M. Beverland, V. Kliuchnikov, and E. Schoute, Surface integration of silicon carbide colour centres with pre-
code compilation via edge-disjoint paths, PRX Quantum served spin-optical coherence, Nat. Mater. 21, 67 (2022).
3, 020342 (2022). [90] C. M. Knaut, A. Suleymanzade, Y.-C. Wei, D. R. Assump-
[75] Q. Zhuang and B. Zhang, Quantum communication capac- cao, P.-J. Stas, Y. Q. Huan, B. Machielse, E. N. Knall, M.
ity transition of complex quantum networks, Phys. Rev. A Sutula, G. Baranes, N. Sinclair, C. De-Eknamkul, D. S.
104, 022608 (2021). Levonian, M. K. Bhaskar, H. Park, M. Lončar, and M. D.
[76] Y. Lee, E. Bersin, A. Dahlberg, S. Wehner, and D. Lukin, Entanglement of nanophotonic quantum memory
Englund, A quantum router architecture for high-fidelity nodes in a telecommunication network, arXiv:2310.01316
entanglement flows in quantum networks, npj Quantum (2023).
Inf. 8, 75 (2022). [91] D. B. Higginbottom et al., Optical observation of single
[77] C. Harney and S. Pirandola, Analytical methods for high- spins in silicon, Nature 607, 266 (2022).
rate global quantum networks, PRX Quantum 3, 010349 [92] D. Dhaliah, Y. Xiong, A. Sipahigil, S. M. Griffin, and G.
(2022). Hautier, First-principles study of the T center in silicon,
[78] T. Peham, L. Burgholzer, and R. Wille, On opti- Phys. Rev. Mater. 6, L053201 (2022).
mal subarchitectures for quantum circuit mapping, [93] K. Saeedi, S. Simmons, J. Z. Salvail, P. Dluhy, H. Rie-
arXiv:2210.09321. mann, N. V. Abrosimov, P. Becker, H.-J. Pohl, J. J. L.
[79] A. Geng, A. Moghiseh, C. Redenbach, and K. Schladitz, Morton, and M. L. W. Thewalt, Room-temperature quan-
Improved FRQI on superconducting processors and its tum bit storage exceeding 39 minutes using ionized donors
restrictions in the NISQ era, Quantum Inf. Process. 22, in silicon-28, Science 342, 830 (2013).
104 (2023). [94] J. T. Muhonen, A. Laucht, S. Simmons, J. P. Dehollain, R.
[80] C. B. Young, A. Safari, P. Huft, J. Zhang, E. Oh, R. Kalra, F. E. Hudson, S. Freer, K. M. Itoh, D. N. Jamieson,
Chinnarasu, and M. Saffman, An architecture for quan- J. C. McCallum, A. S. Dzurak, and A. Morello, Quantify-
tum networking of neutral atom processors, Appl. Phys. ing the quantum gate fidelity of single-atom spin qubits in
B: Lasers Opt. 128, 151 (2022). silicon by randomized benchmarking, J. Phys.: Condens.
[81] T. van Leent, M. Bock, F. Fertig, R. Garthoff, S. Eppelt, Matter 27, 154205 (2015).
Y. Zhou, P. Malik, M. Seubert, T. Bauer, W. Rosenfeld, W. [95] M. Mirhosseini, A. Sipahigil, M. Kalaee, and O. Painter,
Zhang, C. Becher, and H. Weinfurter, Entangling single Superconducting qubit to optical photon transduction,
atoms over 33 km telecom fibre, Nature 607, 69 (2022). Nature 588, 599 (2020).
[82] Y. Li and J. Thompson, High-rate and high-fidelity mod- [96] G. Arnold, M. Wulf, S. Barzanjeh, E. S. Redchenko, A.
ular interconnects between neutral atom quantum proces- Rueda, W. J. Hease, F. Hassani, and J. M. Fink, Con-
sors, arXiv preprint arXiv:2401.04075 (2024). verting microwave and telecom photons with a silicon
[83] R. Uppu, L. Midolo, X. Zhou, J. Carolan, and P. Lodahl, photonic nanomechanical interface, Nat. Commun. 11,
Quantum-dot-based deterministic photon-emitter inter- 4460 (2020).
faces for scalable photonic quantum technology, Nat. [97] V. Krutyanskiy, M. Meraner, J. Schupp, V. Krcmarsky,
Nanotechnol. 2021 16:12 16, 1308 (2021). H. Hainzer, and B. P. Lanyon, Light-matter entanglement
[84] M. Pompili, S. L. N. Hermans, S. Baier, H. K. C. Beukers, over 50 km of optical fibre, npj Quantum Inf. 5, 72 (2019).
P. C. Humphreys, R. N. Schouten, R. F. L. Vermeulen, [98] A. Stolk, K. van der Enden, M.-C. Roehsner, A. Teepe, S.
M. J. Tiggelman, L. dos Santos Martins, B. Dirkse, S. Faes, C. Bradley, S. Cadot, J. van Rantwijk, I. te Raa, R.
Wehner, and R. Hanson, Realization of a multinode quan- Hagen, A. Verlaan, J. Biemond, A. Khorev, R. Vollmer,
tum network of remote solid-state qubits, Science 372, 259 M. Markham, A. Edmonds, J. Morits, T. Taminiau, E. van
(2021). Zwet, and R. Hanson, Telecom-band quantum interference
[85] G. Wolfowicz, C. P. Anderson, B. Diler, O. G. Poluek- of frequency-converted photons from remote detuned NV
tov, F. J. Heremans, and D. D. Awschalom, Vanadium spin centers, PRX Quantum 3, 020359 (2022).
qubits as telecom quantum emitters in silicon carbide, Sci. [99] R. D. Delaney, M. D. Urmey, S. Mittal, B. M. Brubaker,
Adv. 6, 18 (2020). J. M. Kindem, P. S. Burns, C. A. Regal, and K.
010102-15
W. Lehnert, Superconducting-qubit readout via low- tolerant under a broker-client [? ] model of entanglement
backaction electro-optic transduction, Nature 606, 489 distribution.
(2022). [114] S. L. N. Hermans, M. Pompili, H. K. C. Beukers, S. Baier,
[100] O. Alibart, V. D’Auria, M. De Micheli, F. Doutre, F. J. Borregaard, and R. Hanson, Qubit teleportation between
Kaiser, L. Labonté, T. Lunghi, É. Picholle, and S. Tanzilli, non-neighbouring nodes in a quantum network, Nature
Quantum photonics at telecom wavelengths based on 605, 663 (2022).
lithium niobate waveguides, J. Opt. 18, 104001 (2016). [115] P. Drmota, D. Main, D. P. Nadlinger, B. C. Nichol, M.
[101] A. I. Lvovsky, B. C. Sanders, and W. Tittel, Optical A. Weber, E. M. Ainley, A. Agrawal, R. Srinivas, G.
quantum memory, Nat. Photonics 3, 706 (2009). Araneda, C. J. Ballance, and D. M. Lucas, Robust quan-
[102] K. Azuma, S. E. Economou, D. Elkouss, P. Hilaire, tum memory in a trapped-ion quantum network node,
L. Jiang, H.-K. Lo, and I. Tzitrin, Quantum repeaters: Phys. Rev. Lett. 130, 090803 (2023).
From quantum networks to the quantum Internet, [116] S. D. Barrett and P. Kok, Efficient high-fidelity quantum
arXiv:2212.10820. computation using matter qubits and linear optics, Phys.
[103] In an alternative approach, all-optical quantum repeaters Rev. A 71, 060310 (2005).
also benefit from matter-based spin-photon interfaces in [117] H. K. Beukers, M. Pasini, H. Choi, D. Englund, R. Han-
the generation of the complex entangled states needed [90, son, and J. Borregaard, Tutorial: Remote entanglement
184,185]. protocols for stationary qubits with photonic interfaces,
[104] D. P. Nadlinger, P. Drmota, B. C. Nichol, G. Araneda, D. arXiv preprint arXiv:2310.19878 (2023).
Main, R. Srinivas, D. M. Lucas, C. J. Ballance, K. Ivanov, [118] R. Raussendorf, D. E. Browne, and H. J. Briegel,
E. Y. Tan, P. Sekatski, R. L. Urbanke, R. Renner, N. Measurement-based quantum computation on cluster
Sangouard, and J. D. Bancal, Experimental quantum key states, Phys. Rev. A 68, 022312 (2003).
distribution certified by Bell’s theorem, Nature 607, 682 [119] R. Jozsa, An introduction to measurement based quantum
(2022), original title: Device-Independent Quantum Key computation, arXiv:quant-ph/0508124.
Distribution Changed in response to referees. See referee [120] D. Gottesman and I. L. Chuang, Demonstrating the viabil-
reports on file. ity of universal quantum computation using teleportation
[105] M. Rančić, M. P. Hedges, R. L. Ahlefeldt, and M. J. and single-qubit operations, Nature 402, 390 (1999).
Sellars, Coherence time of over a second in a telecom- [121] J. Eisert, K. Jacobs, P. Papadopoulos, and M. B. Plenio,
compatible quantum memory storage material, Nat. Phys. Optimal local implementation of nonlocal quantum gates,
14, 50 (2018). Phys. Rev. A 62, 052317 (2000).
[106] E. Miyazono, I. Craiciu, A. Arbabi, T. Zhong, and A. [122] C. A. Pérez-Delgado and J. F. Fitzsimons, Iterated gate
Faraon, Coupling erbium dopants in yttrium orthosili- teleportation and blind quantum computation, Phys. Rev.
cate to silicon photonic resonators and waveguides, Opt. Lett. 114, 220502 (2015).
Express 25, 2863 (2017). [123] N. S. Minaev and A. V. Mudryi, Thermally-induced
[107] A. Dibos, M. Raha, C. Phenicie, and J. D. Thompson, defects in silicon containing oxygen and carbon, Physica
Atomic source of single photons in the telecom band, Status Solidi (a) 68, 561 (1981).
Phys. Rev. Lett. 120, 243601 (2018). [124] A. M. Tyryshkin, S. Tojo, J. J. Morton, H. Riemann, N. V.
[108] A. Reiserer, N. Kalb, M. S. Blok, K. J. van Bemme- Abrosimov, P. Becker, H.-J. Pohl, T. Schenkel, M. L. The-
len, T. H. Taminiau, R. Hanson, D. J. Twitchen, and walt, K. M. Itoh, and S. A. Lyon, Electron spin coherence
M. Markham, Robust quantum-network memory using exceeding seconds in high-purity silicon, Nat. Mater. 11,
decoherence-protected subspaces of nuclear spins, Phys. 143 (2012).
Rev. X 6, 021040 (2016). [125] T. Kobayashi, J. Salfi, C. Chua, J. Van Der Heijden, M. G.
[109] C. Yin, M. Rancic, G. G. de Boo, N. Stavrias, J. C. McCal- House, D. Culcer, W. D. Hutchison, B. C. Johnson, J. C.
lum, M. J. Sellars, and S. Rogge, Optical addressing of an McCallum, H. Riemann, N. V. Abrosimov, P. Becker, H.-
individual erbium ion in silicon, Nature 497, 91 (2013). J. Pohl, M. Y. Simmons, and S. Rogge, Engineering long
[110] T. H. Taminiau, J. Cramer, T. van der Sar, V. V. Dobrovit- spin coherence times of spin-orbit qubits in silicon, Nat.
ski, and R. Hanson, Universal control and error correction Mater. 20, 38 (2021).
in multi-qubit spin registers in diamond, Nat. Nanotech- [126] Y. Xiong, C. Bourgois, N. Sheremetyeva, W. Chen, D.
nol. 9, 171 (2014). Dahliah, H. Song, S. M. Griffin, A. Sipahigil, and G.
[111] H. H. Vallabhapurapu, I. Hansen, C. Adambukulam, Hautier, High-throughput identification of spin-photon
R. Stohr, A. Denisenko, C. H. Yang, and A. Laucht, interfaces in silicon (2023).
High fidelity control of a nitrogen-vacancy spin qubit [127] G. Zhang, Y. Cheng, J.-P. Chou, and A. Gali, Material
at room temperature using the SMART Protocol, platforms for defect qubits and single-photon emitters,
arXiv:2208.14671. Appl. Phys. Rev. 7, 031308 (2020).
[112] G. Waldherr, Y. Wang, S. Zaiser, M. Jamali, T. Schulte- [128] E. M. Purcell, H. C. Torrey, and R. V. Pound, Resonance
Herbrüggen, H. Abe, T. Ohshima, J. Isoya, J. F. Du, P. absorption by nuclear magnetic moments in a solid, Phys.
Neumann, and J. Wrachtrup, Quantum error correction in Rev. 69, 37 (1946).
a solid-state hybrid spin register, Nature 506, 204 (2014). [129] E. M. Purcell, in Confined Electrons and Photons: New
[113] Alternatively, the photons could themselves be considered Physics and Applications (Springer, 1995), p. 839.
as qubits to directly mediate nonlocal interactions through [130] F. Islam, C.-M. Lee, S. Harper, M. H. Rahaman, Y.
strong coupling schemes [? ], which could be made loss Zhao, N. K. Vij, and E. Waks, Cavity enhanced emission
010102-16
from a silicon T center, arXiv preprint arXiv:2310.13808 [144] J. L. Zhang, S. Sun, M. J. Burek, C. Dory, Y.-K. Tzeng,
(2023). K. A. Fischer, Y. Kelaita, K. G. Lagoudakis, M. Radu-
[131] A. Johnston, U. Felix-Rendon, Y.-E. Wong, and S. Chen, laski, Z.-X. Shen, N. A. Melosh, S. Chu, M. Lončar, and
Cavity-coupled telecom atomic source in silicon, arXiv J. Vučković, Strongly cavity-enhanced spontaneous emis-
preprint arXiv:2310.20014 (2023). sion from silicon-vacancy centers in diamond, Nano. Lett.
[132] A. DeAbreu, C. Bowness, A. Alizadeh, C. Chartrand, 18, 1360 (2018).
N. A. Brunelle, E. R. MacQuarrie, N. R. Lee-Hone, M. [145] D. L. Moehring, P. Maunz, S. Olmschenk, K. C.
Ruether, M. Kazemi, A. T. K. Kurkjian, S. Roorda, N. V. Younge, D. N. Matsukevich, L.-M. Duan, and C. Monroe,
Abrosimov, H. J. Pohl, M. L. W. Thewalt, D. B. Higgin- Entanglement of single-atom quantum bits at a distance,
bottom, and S. Simmons, Waveguide-integrated silicon T Nature 449, 68 (2007).
centres, Opt. Express 31, 15045 (2023). [146] H. Bernien, B. Hensen, W. Pfaff, G. Koolstra, M. S.
[133] P. Neumann, J. Beck, M. Steiner, F. Rempp, H. Fedder, Blok, L. Robledo, T. H. Taminiau, M. Markham, D. J.
P. R. Hemmer, J. Wrachtrup, and F. Jelezko, Single-shot Twitchen, L. Childress, and R. Hanson, Heralded entan-
readout of a single nuclear spin, Science 329, 542 (2010). glement between solid-state qubits separated by three
[134] R. Gehr, J. Volz, G. Dubois, T. Steinmetz, Y. Colombe, B. metres, Nature 497, 86 (2013).
L. Lev, R. Long, J. Estève, and J. Reichel, Cavity-based [147] J. Metz and S. D. Barrett, Effect of frequency-mismatched
single atom preparation and high-fidelity hyperfine state photons in quantum-information processing, Phys. Rev. A
readout, Phys. Rev. Lett. 104, 203602 (2010). 77, 042323 (2008).
[135] J. T. Muhonen, J. P. Dehollain, A. Laucht, F. E. Hudson, [148] C. D. Wilen, S. Abdullah, N. A. Kurinsky, C. Stanford, L.
R. Kalra, T. Sekiguchi, K. M. Itoh, D. N. Jamieson, J. C. Cardani, G. D’Imperio, C. Tomei, L. Faoro, L. B. Ioffe,
McCallum, A. S. Dzurak, and A. Morello, Storing quan- C. H. Liu, A. Opremcak, B. G. Christensen, J. L. DuBois,
tum information for 30 seconds in a nanoelectronic device, and R. McDermott, Correlated charge noise and relaxation
Nat. Nanotechnol. 2014 9:12 9, 986 (2014). errors in superconducting qubits, Nature 594, 369 (2021).
[136] A. N. Safanov, E. C. Lightowlers, and G. Davies, [149] J. Yoneda, J. S. Rojas-Arias, P. Stano, K. Takeda, A. Noiri,
Carbon-hydrogen deep level luminescence centre in sili- T. Nakajima, D. Loss, and S. Tarucha, Noise-correlation
con responsible for the T-line, Mater. Sci. Forum 196-201, spectrum for a pair of spin qubits in silicon (2022).
909 (1995). [150] J. S. Rojas-Arias, A. Noiri, P. Stano, T. Nakajima, J.
[137] T. Müller, I. Aharonovich, L. Lombez, Y. Alaverdyan, A. Yoneda, K. Takeda, T. Kobayashi, A. Sammak, G. Scap-
N. Vamivakas, S. Castelletto, F. Jelezko, J. Wrachtrup, S. pucci, D. Loss, and S. Tarucha, Spatial noise correlations
Prawer, and M. Atatüre, Wide-range electrical tunability beyond nearest-neighbor in 28 Si/SiGe spin qubits (2023),
of single-photon emission from chromium-based colour doi.org/10.48550/ARXIV.2302.11717.
centres in diamond, New J. Phys. 13, 075001 (2011). [151] D. I. Schuster, A. Wallraff, A. Blais, L. Frunzio, R.-S.
[138] Manuscript in preparation. Huang, J. Majer, S. M. Girvin, and R. J. Schoelkopf,
[139] A. L. Crook, C. P. Anderson, K. C. Miao, A. Bourassa, ac Stark shift and dephasing of a superconducting qubit
H. Lee, S. L. Bayliss, D. O. Bracher, X. Zhang, H. Abe, strongly coupled to a cavity field, Phys. Rev. Lett. 94,
T. Ohshima, E. L. Hu, and D. D. Awschalom, Purcell 123602 (2005).
enhancement of a single silicon carbide color center with [152] J. M. Pino, J. M. Dreiling, C. Figgatt, J. P. Gaebler, S. A.
coherent spin control, Nano Lett. 20, 3427 (2020). Moses, M. S. Allman, C. H. Baldwin, M. Foss-Feig, D.
[140] K. Ashida, M. Okano, M. Ohtsuka, M. Seki, N. Hayes, K. Mayer, C. Ryan-Anderson, and B. Neyenhuis,
Yokoyama, K. Koshino, M. Mori, T. Asano, S. Noda, and Demonstration of the trapped-ion quantum CCD computer
Y. Takahashi, Ultrahigh-Q photonic crystal nanocavities architecture, Nature 592, 209 (2021).
fabricated by CMOS process technologies, Opt. Express [153] F. Rozpedek, T. Schiet, D. Elkouss, A. C. Doherty, S.
25, 18165 (2017). Wehner et al., Optimizing practical entanglement distil-
[141] N. Quack, A. Y. Takabayashi, H. Sattari, P. Edinger, G. lation, Phys. Rev. A 97, 062333 (2018).
Jo, S. J. Bleiker, C. Errando-Herranz, K. B. Gylfason, F. [154] N. Kalb, A. A. Reiserer, P. C. Humphreys, J. J. W. Baker-
Niklaus, U. Khan, P. Verheyen, A. K. Mallik, J. S. Lee, mans, S. J. Kamerling, N. H. Nickerson, S. C. Benjamin,
M. Jezzini, P. Morrissey, C. Antony, P. O’Brien, and W. D. J. Twitchen, M. Markham, and R. Hanson, Entan-
Bogaerts, Integrated silicon photonic MEMS, Microsyst. glement distillation between solid-state quantum network
Nanoeng. 9, 1 (2023). nodes, Science 356, 928 (2017).
[142] Y.-L. Tang, Y.-L. Tang, L. Komza, L. Komza, P. Samut- [155] S. Bratzik, S. Abruzzo, H. Kampermann, and D. Bruß,
praphoot, P. Samutpraphoot, H. Song, M. Odeh, M. Odeh, Quantum repeaters and quantum key distribution: The
M. Mathew, M. Mathew, J. Chang, Z.-H. Zhang, A. impact of entanglement distillation on the secret key rate,
Sipahigil, A. Sipahigil, and A. Sipahigil, Tunable single Phys. Rev. A 87, 062335 (2013).
photons from an artificial atom in silicon photonics, CLEO [156] C. H. Bennett, G. Brassard, S. Popescu, B. Schumacher,
2023 (2023), paper FTu3C.3 , FTu3C.3 (2023). J. A. Smolin, and W. K. Wootters, Purification of noisy
[143] M. Prabhu, C. Errando-Herranz, L. D. Santis, I. Chris- entanglement and faithful teleportation via noisy channels,
ten, C. Chen, C. Gerlach, and D. Englund, Individu- Phys. Rev. Lett. 76, 722 (1996).
ally addressable and spectrally programmable artificial [157] N. Kalb, A. A. Reiserer, P. C. Humphreys, J. J. W. Bak-
atoms in silicon photonics, Nat. Commun. 14, 2380 ermans, S. J. Kamerling, N. H. Nickerson, S. C. Ben-
(2023). jamin, D. J. Twitchen, M. Markham, and R. Hanson,
010102-17
Entanglement distillation between solid-state quantum [170] B. R. de Supinski, M. Hall, T. Gamblin, T. Häner, D. S.
network nodes, Science 356, 928 (2017). Steiger, T. Hoefler, and M. Troyer, Distributed quantum
[158] Y. Li and S. C. Benjamin, High threshold distributed quan- computing with QMPI, SC21: International Conference
tum computing with three-qubit nodes, New J. Phys. 14, for High Performance Computing, Networking, Storage
093008 (2012). and Analysis 00, 1 (2021).
[159] L. Jiang, M. V. G. Dutt, E. Togan, L. Childress, P. [171] H. J. Kimble, The quantum Internet, Nature 453, 1023
Cappellaro, J. M. Taylor, and M. D. Lukin, Coherence of (2008).
an optically illuminated single nuclear spin qubit, Phys. [172] Y. Liu, W.-J. Zhang, C. Jiang, J.-P. Chen, C. Zhang, W.-
Rev. Lett. 100, 073001 (2008). X. Pan, D. Ma, H. Dong, J.-M. Xiong, C.-J. Zhang, H. Li,
[160] M. S. Blok, N. Kalb, A. Reiserer, T. H. Taminiau, and R.-C. Wang, J. Wu, T.-Y. Chen, L. You, X.-B. Wang, Q.
R. Hanson, Towards quantum networks of single spins: Zhang, and J.-W. Pan, Experimental twin-field quantum
analysis of a quantum memory with an optical interface in key distribution over 1000 km fiber distance, Phys. Rev.
diamond, Faraday Discuss. 184, 173 (2015). Lett. 130, 210801 (2023).
[161] A. J. R. MacDonald, G. G. Popowich, B. D. Hauer, P. H. [173] H.-J. Briegel, W. Dür, J. I. Cirac, and P. Zoller, Quan-
Kim, A. Fredrick, X. Rojas, P. Doolin, and J. P. Davis, tum repeaters: The role of imperfect local operations
Optical microscope and tapered fiber coupling apparatus in quantum communication, Phys. Rev. Lett. 81, 5932
for a dilution refrigerator, Rev. Sci. Instrum. 86, 013107 (1998).
(2015). [174] S. Muralidharan, L. Li, J. Kim, N. Lütkenhaus, M. D.
[162] M. Häußler, R. Terhaar, M. A. Wolff, H. Gehring, F. Beu- Lukin, and L. Jiang, Optimal architectures for long dis-
tel, W. Hartmann, N. Walter, M. Tillmann, M. Ahangar- tance quantum communication, Sci. Rep. 6, 20463 (2016).
ianabhari, M. Wahl, T. Röhlicke, H.-J. Rahn, W. H. P. [175] L. Jiang, J. M. Taylor, K. Nemoto, W. J. Munro, R. V.
Pernice, and C. Schuck, Scaling waveguide-integrated Meter, and M. D. Lukin, Quantum repeater with encoding,
superconducting nanowire single-photon detector solu- Phys. Rev. A 79, 032325 (2009), pauli frame.
tions to large numbers of independent optical channels, [176] S. Wehner, D. Elkouss, and R. Hanson, Quantum Inter-
arXiv:2207.12060. net: A vision for the road ahead, Science 362, eaam9288
[163] I. Hansen, A. E. Seedhouse, A. Saraiva, A. Laucht, A. S. (2018).
Dzurak, and C. H. Yang, Pulse engineering of a global [177] C. Simon, Towards a global quantum network, Nat. Pho-
field for robust and universal quantum computation, Phys. tonics 11, 678 (2017).
Rev. A 104, 062415 (2021). [178] W. Gao, P. Fallahi, E. Togan, A. Delteil, Y. Chin, J.
[164] E. Vahapoglu, J. P. Slack-Smith, R. C. C. Leon, W. H. Lim, Miguel-Sanchez, and A. Imamoğlu, Quantum teleporta-
F. E. Hudson, T. Day, J. D. Cifuentes, T. Tanttu, C. H. tion from a propagating photon to a solid-state spin qubit,
Yang, A. Saraiva, N. V. Abrosimov, H.-J. Pohl, M. L. W. Nat. Commun. 4, 2744 (2013).
Thewalt, A. Laucht, A. S. Dzurak, and J. J. Pla, Coherent [179] C. Panayi, M. Razavi, X. Ma, and N. Lütkenhaus,
control of electron spin qubits in silicon using a global Memory-assisted measurement-device-independent quan-
field, npj Quantum Inf. 8, 126 (2022). tum key distribution, New J. Phys. 16, 043005 (2014).
[165] B. E. Kane, A silicon-based nuclear spin quantum com- [180] H.-K. Lo, M. Curty, and B. Qi, Measurement-device-
puter, Nature 393, 133 (1998). independent quantum key distribution, Phys. Rev. Lett.
[166] E. I. Rosenthal, C. P. Anderson, H. C. Kleidermacher, 108, 130503 (2012).
A. J. Stein, H. Lee, J. Grzesik, G. Scuri, A. E. Rugar, [181] X. Liu, J. Liu, R. Xue, H. Wang, H. Li, X. Feng, F.
D. Riedel, S. Aghaeimeibodi, G. H. Ahn, K. Van Gasse, Liu, K. Cui, Z. Wang, L. You, Y. Huang, and W. Zhang,
and J. Vučković, Microwave spin control of a tin-vacancy 40-user fully connected entanglement-based quantum key
qubit in diamond, Phys. Rev. X 13, 031022 (2023). distribution network without trusted node, PhotoniX 3, 2
[167] J. J. Pla, K. Y. Tan, J. P. Dehollain, W. H. Lim, J. J. Mor- (2022).
ton, F. A. Zwanenburg, D. N. Jamieson, A. S. Dzurak, and [182] A. Acar, H. Aksu, A. S. Uluagac, and M. Conti, A sur-
A. Morello, High-fidelity readout and control of a nuclear vey on homomorphic encryption schemes: Theory and
spin qubit in silicon, Nature 496, 334 (2013). implementation, ACM Computing Surveys (Csur) 51, 1
[168] B. P. Wuetz, P. L. Bavdaz, L. A. Yeoh, R. Schouten, H. van (2018).
der Does, M. Tiggelman, D. Sabbagh, A. Sammak, C. G. [183] T. Kapourniotis, E. Kashefi, D. Leichtle, L. Music,
Almudever, F. Sebastiano, J. S. Clarke, M. Veldhorst, and and H. Ollivier, A framework for verifiable blind
G. Scappucci, Multiplexed quantum transport using com- quantum computation, arXiv 10.48550/arxiv.2206.00631
mercial off-the-shelf CMOS at sub-kelvin temperatures, (2022).
npj Quantum Inf. 6, 43 (2020). [184] D. Buterakos, E. Barnes, and S. E. Economou, Determin-
[169] S. Doerner, A. Kuzmin, S. Wuensch, I. Charaev, F. Boes, istic generation of all-photonic quantum repeaters from
T. Zwick, and M. Siegel, Frequency-multiplexed bias and solid-state emitters, Phys. Rev. X 7, 041023 (2017).
readout of a 16-pixel superconducting nanowire single- [185] S. C. Benjamin, D. E. Browne, J. Fitzsimons, and J. J. L.
photon detector array, Appl. Phys. Lett. 111, 032603 Morton, Brokered graph-state quantum computation, New
(2017). J. Phys. 8, 0 (2006).
010102-18

PRXQuantum 5 010102

Uploaded by

Copyright:

Available Formats

PRXQuantum 5 010102

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

PRXQuantum 5 010102

Uploaded by

Copyright:

Available Formats

PRX QUANTUM 5, 010102 (2024)

Scalable Fault-Tolerant Quantum Technologies with Silicon Color Centers

(Received 19 October 2023; published 13 March 2024)

I. CONTEXT [12]. Some post-quantum cryptography algorithms under

2691-3399/24/5(1)/010102(18) 010102-1 Published by the American Physical Society

(a) (b) (c)

SNSPDs Optical Switches

for good overall system performance. This observation

Quantum memories are necessary for the construc- (a)

(a) (a) (c)

Entangling Rate (Hz)

The triggered emission of two indistinguishable high-

C. Teleported gates detector module. Here, the silicon-on-insulator (SOI) plat-

B. Local quantum networks

You might also like