nature nanotechnology

Article https://doi.org/10.1038/s41565-023-01332-3

Strong coupling between a photon and a hole

spin in silicon

Received: 13 July 2022 Cécile X. Yu1,4, Simon Zihlmann 1,4 , José C. Abadillo-Uriel 2,
Vincent P. Michal2, Nils Rambal3, Heimanu Niebojewski3, Thomas Bedecarrats3,
Accepted: 24 January 2023
Maud Vinet 3, Étienne Dumur 1, Michele Filippone 2, Benoit Bertrand 3,
Published online: xx xx xxxx Silvano De Franceschi 1, Yann-Michel Niquet 2 & Romain Maurand1

Check for updates

Spins in semiconductor quantum dots constitute a promising platform for
scalable quantum information processing. Coupling them strongly to the
photonic modes of superconducting microwave resonators would enable
fast non-demolition readout and long-range, on-chip connectivity, well
beyond nearest-neighbour quantum interactions. Here we demonstrate
strong coupling between a microwave photon in a superconducting
resonator and a hole spin in a silicon-based double quantum dot issued from
a foundry-compatible metal–oxide–semiconductor fabrication process.
By leveraging the strong spin–orbit interaction intrinsically present in the
valence band of silicon, we achieve a spin–photon coupling rate as high as
330 MHz, largely exceeding the combined spin–photon decoherence rate.
This result, together with the recently demonstrated long coherence of hole
spins in silicon, opens a new realistic pathway to the development of circuit
quantum electrodynamics with spins in semiconductor quantum dots.

Cavity quantum electrodynamics (QED) deals with the interaction As the spin does not directly couple to the cavity electric field, a
between the quantum degrees of freedom of an atom and the electro- spin–charge hybridization mechanism is needed to achieve coherent
magnetic modes of a cavity1. The extension of this concept to supercon- spin–photon interfaces. For electrons in Si/SiGe double quantum dots
ducting quantum circuits has led to the development of circuit QED, (DQDs), spin–photon coupling rates of a few tens of megahertz have
opening new opportunities for the study of light–matter interaction been demonstrated with the help of a synthetic spin–orbit (SO) interac-
and fostering the progress of solid-state quantum processors based on tion created by nearby micromagnets16–19. The reported coupling rates
superconducting qubits2–4. In the same footsteps, a variety of alterna- are several times larger than the spin dephasing rate, thereby enabling
tive realizations have been explored using different types of quantum coherent spin–photon coupling 16,17 and cavity-mediated interaction
system as artificial atoms5. Hybrid systems made of quantum dots cou- between spins in distant DQDs18,19. Yet, to fully profit from circuit QED
pled to superconducting microwave resonators are a prominent exam- tools, including long-range, high-fidelity two-qubit operations and
ple6–12. Of particular interest are silicon-based quantum dots owing to quantum non-demolition readout, a much stronger coupling strength
their ability to host long-coherence qubits encoded in a spin degree of is required, which necessitates more efficient coupling schemes.
freedom. Silicon-based spin qubits have made remarkable progress, In this work, we turn to a hole spin in a silicon nanowire metal–
reaching high fidelities in both one- and two-qubit gate operations, oxide–semiconductor (MOS) DQD to exploit the strong intrinsic SO
the latter being enabled by tunnelling-mediated exchange interaction interaction of the valence band states20,21, whose potential for cir-
between neighbouring qubits13. The co-integration with supercon- cuit QED22–26 has remained unexplored. In our device geometry, the
ducting cavities acting as quantum buses would allow for long-range quasi-one-dimensional hole confinement enhances this SO interac-
connectivity, largely facilitating the scalability of silicon spin qubits14,15. tion27 such that the SO length ℓso, that is, the distance over which a

1
Univ. Grenoble Alpes, CEA, Grenoble INP, IRIG-Pheliqs, Grenoble, France. 2Univ. Grenoble Alpes, CEA, IRIG-MEM-L_Sim, Grenoble, France. 3Univ.
Grenoble Alpes, CEA, LETI, Grenoble, France. 4These authors contributed equally: Cécile X. Yu, Simon Zihlmann. e-mail: [email protected];
[email protected]

Nature Nanotechnology
Article https://doi.org/10.1038/s41565-023-01332-3

a c d
A2 (dB) A2 (dB)
–33 –33 –31 –28 –26 –24

5.46
–823
5.44
W via ε
G4

VG1 (mV)

fp (GHz)
G3 –824 5.42
D χc/2π
G2
G1 5.40
z
–825
5.38
BSO x
S Φ B y –826 5.36
–829 –828 –827 –100 0 100
VG2 (mV) ε/h (GHz)

b e f
G1 RF in
2 µm
800 nm

hωc
hωr
|+

Energy (a.u.)

Energy (a.u.)
G3
0 hωc

|–

|L
D

|R


0
0 0
100 µm ε (a.u.) ε (a.u.)
G2 RF out

Fig. 1 | Silicon MOS device, superconducting circuitry and DQD charge consisting of an interdigitated capacitor and nanowire inductor. c, Transmission
properties. a, Schematic of the cross section of the DQD device. The gates probed at the bare resonator resonance frequency (5.43 GHz) as a function of
(G1–G4), source (S) and drain (D) of the nanowire transistor are connected VG1 and VG2 at B = 0. The energy detuning ε between the two dots is swept along
with vertical vias to the NbN circuitry located at the surface of the device. We the black arrow. d, Transmission as a function of fp and ε. At large ∣ε∣, the bare
refer to the dot formed below G1 (G2) as the left (right) dot. The effect of the SO resonator is probed, whereas near ε = 0, the DQD charge qubit dispersively
interaction is illustrated by a hole whose spin rotates coherently while tunnelling interacts with the resonator, leading to a frequency shift of χc/2π. e, Qualitative
from one dot to the other in the presence of an effective SO field Bso (blue arrow). energy diagram of the DQD as a function of energy detuning ε. The charge can be
The orange wavepacket on the via connecting G2 pictures the photon in the either in the right dot |R〉 (blue) or in the left dot |L〉 (red). Due to the finite tunnel
microwave cavity. The external magnetic field B is applied in-plane with angle coupling tc between the dots, |L〉 and |R〉 hybridize near ε = 0 into a bonding (|–〉)
ϕ to the nanowire axis. b, False-colour top-view scanning electron micrograph and antibonding (|+〉) state, forming the basis of a charge qubit with energy ℏωc.
of a representative NbN circuitry. The areas where NbN is etched away appear f, Qualitative energy diagram of the charge qubit energy and cavity energy as a
in dark grey. The 50 Ω feed line, resonator and its d.c. bias line are highlighted in function of ε. The coloured solid lines correspond to the non-interacting case,
green, red and yellow, respectively. The inset shows a close-up of the d.c. tap at whereas the dashed lines sketch the dispersive repulsion experienced by the
the centre of the resonator. Each d.c. line is equipped with an LC low-pass filter cavity and charge qubit in the presence of a finite charge–photon coupling.

spin rotates by π due to SO interaction, is reduced to a few tens of A high-impedance (2.5 kΩ) microwave cavity is then patterned in the NbN
nanometres, comparable with the DQD spatial extension d. The pres- film, along with a 50 Ω microwave feed line, ground planes and fanout
ence of such a strong SO interaction dramatically modifies the DQD lines (Fig. 1b). Besides being well suited for future large-scale integration,
energy levels, resulting in the formation of a flopping-mode SO qubit24 the used foundry-compatible MOS technology comes with large gate
whose energy is well separated from the other excitations of the DQD capacitances, resulting in tight electrostatic control and hence strong
system. Here we demonstrate that this spin qubit strongly interacts coupling to the electric-field component of the cavity mode.
with the quantized field of a high-impedance superconducting micro- Negative voltages applied to G1 and G2 accumulate holes in a DQD
wave resonator. We observe a spin–photon coupling rate as large as potential (Fig. 1a). The so-confined holes in the valence band of silicon
gs/2π = 330 MHz, exceeding the combined spin and photon decoher- experience a strong SO interaction20,21 that—along with an applied
ence rate by a factor 27 and hence leading to a cooperativity of 1,600. in-plane magnetic field B—controls the spin–charge mixing in the DQD
Furthermore, we explore the impact of the SO interaction in the limit (as discussed later).
of a spin confined in a single quantum dot. We measure a spin–photon We probe the microwave response of this hybrid system in trans-
coupling rate of ~1 MHz, in line with recent predictions25,26. mission at a temperature of 8 mK and at powers corresponding to less
than one photon on average in the cavity (navg ≈ 0.1), which is assumed
Hole-spin-based circuit QED architecture to be in its ground state. We first characterize the bare-cavity response
The spin circuit QED architecture designed here features a hole con- by sweeping the probe frequency fp = ωp/2π across the resonance fre-
fined in a silicon DQD device interacting with a single microwave photon quency and keeping the charges in the DQD fixed. This way, we extract
trapped in a superconducting cavity. The DQD is hosted in a natural the bare-cavity resonance frequency fr = ωr/2π = 5.428 GHz and cavity
silicon nanowire MOS transistor whose channel is controlled by four decay rate κ/2π = 14 MHz. To characterize the charge–photon coupling
Ω-shape gates crossing the nanowire (Fig. 1a). The back-end-of-line strength gc, we monitor the transmission at frequency fr as VG1 and VG2
fabrication of the silicon chip is interrupted to replace the first metal- are varied (Fig. 1c). In the dark-blue region (Fig. 1c), the levels of the two
lic interconnect layer with a 10-nm-thick niobium nitride (NbN) layer dots are aligned so that a hole oscillates between the dots in response
with large kinetic inductance28,29 and magnetic-field resilience29. to the cavity electric field. We next probe the transmission as a function

Nature Nanotechnology
Article https://doi.org/10.1038/s41565-023-01332-3

(A/A0)2 (Fig. 3a,b). This large modulation in the spin–photon coupling strength
a b results from the interplay between the Zeeman effect and SO interac-
0.6 0.7 0.8 0.9 1.0
Φ = 45°
tion, leading to gs ∝ gc∣(gB) × (gBso)∣, where g is the average gyromag-
netic g-matrix of the two dots and Bso is the effective SO field (note
|+↑
that ∣Bso∣ is related to the SO length (Supplementary Section IX)). The
5.6
|+↓ spin–photon coupling is, thus, expected to vanish when B is parallel
14 MHz to Bso, and is the maximum at ϕmax where the spin Larmor vector gB is
hωs |–↑
5.5
approximately perpendicular to gBso. The g-matrix of holes embodies
|–↓
an anisotropic Zeeman splitting EZ = μB∣gB∣, in contrast to electrons
fp (GHz)

5.4
ωr/2π (EZ ≈ 2μBB) (ref. 31). In the present case, both dots show similar g-matrix
184 MHz anisotropies, with EZ ≃ 1.3μBB when B is along the x axis and EZ ≃ 2μBB
when B is along the y axis. As shown in Fig. 3b, gs/2π gets almost entirely
5.3
suppressed around ϕ = 75°. Since gs does not vanish completely, Bso
must have a small out-of-plane component. Overall, however, we can
5.2
conclude that the orientation of Bso is rather close to the y axis. As dis-
S 2π
ω/

cussed in Supplementary Section IX, the orientation of Bso is primarily

5.1
determined by the device geometry. It is expected to be perpendicular
350 400 450 0.8 1.0
to the interdot tunnelling direction (~x axis) as well as to the average
B (mT) (A/A0)2
electric-field direction in the interdot barrier region (~z axis). Due to the
Fig. 2 | Strong spin–photon coupling. a, Normalized transmission as a function anisotropy of the g-matrix, the magnetic-field orientation maximizing
of probe frequency fp and amplitude of the magnetic field B oriented at ϕ = 45° gs is not exactly orthogonal to Bso.
with respect to the nanowire axis. The inset shows the qualitative energy diagram The quality of the spin–photon interface can be further quantified
of the DQD at finite magnetic field. The magnetic field splits the spin degenerate by the ratio between the coupling strength and decoherence rate, that
|–〉 and |+〉 states (Fig. 1e) into four states, namely, |–↓〉, |–↑〉, |+↓〉 and |+↑〉. An
is, 2gs/(γs + κ/2) (Fig. 3c). This ratio reaches up to 27 for ϕ = 0°, which—
avoided crossing—the signature of strong spin–photon coupling—is observed
along with a cooperativity C = 4g2s /(γs κ) = 1,600 (ref. 5)—highlights an
when the spin transition frequency ωs/2π (dashed line) matches the bare
extremely strong light–matter interaction. To evaluate the relative
resonator frequency ωr/2π (dash–dot line). b, Frequency linecut at resonance,
highlighting a vacuum Rabi-mode splitting 2gs/2π = 184 MHz. The solid line is a
impact of spin dephasing and cavity decay rates, we can extract the
fit to two superimposed Lorentzian functions, whose width (14 MHz) is twice the angular dependence of γs/2π (Supplementary Section VIII). We find
decoherence rate of the hybridized spin–photon state. values ranging from 2.5 to 17.0 MHz. Small (large) spin dephasing
generally coincides with a small (large) spin–charge mixing. Over a
large angular range centred around ϕ = 75°, γs remains rather small and
the quality of spin–photon interface is mostly limited by the relatively
of fp and energy detuning ε between the two dots (Fig. 1d). This reveals
large cavity decay rate κ/2π = 14 MHz.
a dispersive downshift in cavity resonance near ε = 0, due to the
A model for the spin–photon coupling of a hole DQD with SO
electric-dipole interaction with the DQD hole charge qubit3,30 with
interaction is presented in Supplementary Sections IX and X. With
energy ℏωc = √ε2 + 4t2c > ℏωr, where ℏ = h/2π is the reduced Planck distinct anisotropic Zeeman response for the two quantum dots and
constant (Fig. 1e,f). From the temperature dependence of this disper- spin-dependent tunnel couplings, this model captures gs at all the
sive shift, we extract a charge–photon coupling strength magnetic-field orientations (Fig. 3b). The effect of spin–charge mixing
gc /2π = 513 MHz together with an interdot tunnel coupling on the DQD energy diagram is illustrated in Fig. 3d,e for two different
tc/h = 9.57 GHz (Supplementary Section IV). magnetic-field orientations. In particular, Fig. 3e highlights the large
renormalization of energy levels in a DQD due to a strong SO interac-
Strong hole spin–photon coupling tion. Our model also catches the decrease in coherence by spin–charge
An in-plane magnetic field B lifts the spin degeneracy of the DQD charge mixing as the magnetic-field orientation approaches ϕmax (Fig. 3c and
states (Fig. 2a, inset). The two lowest spin-polarized states define an SO Supplementary Section XI). In Fig. 3c, the black solid line is calculated
flopping-mode qubit24. We probe the spin–photon interaction at ε ≈ 0, assuming a charge qubit decoherence rate γc/2π = 9.9 MHz, due to
where the electric dipole of the hole in the DQD is the maximum. When the dominant detuning noise, on top of an isotropic bare spin deco-
the Zeeman spin splitting ℏωs matches the resonance frequency of the herence rate γ0/2π = 3.4 MHz, collecting other mechanisms such as
cavity (fr ≃ fs = ωs/2π), spin–photon hybridization results in an avoided nuclear-spin noise32, electrically induced g-factor fluctuations25,33 or
crossing that splits the cavity response into two branches separated by phonon-mediated spin relaxation34.
the vacuum Rabi-mode splitting2,3. A representative measurement of
this avoided crossing is shown in Fig. 2a, where the normalized transmis- Spin–photon coupling in the single-dot limit
sion is plotted as a function of fp and B = ∣B∣ at ϕ = 45° with respect to the The spin–photon couplings reported so far benefit from the large
nanowire axis. The linecut at resonance shows two distinct dips sepa- electric-dipole moment of the DQD at ε ≈ 0. In the following, we explore
rated by a vacuum Rabi-mode splitting 2gs/2π = 184 MHz, where gs is the the interaction between the spin and cavity when the hole is localized
spin–photon coupling. The linewidth of these dips yields the decoher- in a single quantum dot. As previously reported for electrons17, this
1
ence rate of the hybridized spin–photon states (γs + κ/2)/2𝜋𝜋 = 7 MHz, interaction quickly vanishes for increasing detuning since charge
2
where γs is the spin decoherence rate. From κ/2π = 14 MHz, we extract localization in a single dot quenches the electric-dipole moment. This
γs/2π = 7 MHz. The fact that gs ≫ κ, γs demonstrates a strong coupling is evidenced by the detuning dependence of the charge–photon cou-
between the hole spin and photon in the cavity.
pling geff 2
c = 2gc tc /√4tc + ε . Figure 4a shows the transmission when
2

Spin–photon coupling versus magnetic-field the microwave cavity is resonant with the spin splitting of the hole
orientation confined in the right dot (ε ≪ −tc). Despite the considerable reduction
Varying the orientation of the in-plane magnetic field reveals pro- in the hole-dipole moment, two dips are still visible in the transmission,
nounced anisotropy in the vacuum Rabi-mode splitting with a measured implying γs < gs. Their separation reveals spin–photon coupling
maximum (minimum) gs/2π of 330 MHz (10 MHz) at ϕ = 3° (ϕ = 79°) gs/2π ≈ 1 MHz. Since gs < κ (the so-called bad-cavity limit), a clear vacuum

Nature Nanotechnology
Article https://doi.org/10.1038/s41565-023-01332-3

a 8
b d
300 Theory Φ = 45°
20

gS/2π (MHz)
7

E/h (GHz)
200
0
6
100
–20
5
0
0 25 50 75 100 125 150 175 –25 0 25
(A/A0)2

Φmax Φ (°) ε/h (GHz)

4
c e
30
3 45 Theory Φ = 0°
Φ (°) 135 20

2gS/(γS+κ/2)
3 112 20

E/h (GHz)
2 28 101 0
0 90
10
1 175 56
–20
34 79
0
–400 –200 0 200 400 0 25 50 75 100 125 150 175 –25 0 25
fp – fr (MHz) Φ (°) ε/h (GHz)

Fig. 3 | Spin–photon coupling versus magnetic-field orientation. a, The vertical dashed line indicates the magnetic field angle ϕmax = 17° at which the
Normalized transmission as a function of probe frequency fp for various spin–photon coupling is the maximum. The error bars in b and c represent the
magnetic-field orientations ϕ. All the curves are measured at resonance (that is, standard deviation from the fitting. d,e, Energy diagram of the DQD at resonance
for fs = fr) and all of them show a clear vacuum Rabi-mode splitting. The curves for ϕ = 45° (d) and ϕ = 0° (e). The dashed lines represent the spin states in the
are vertically offset and centred around the cavity resonance frequency fr for absence of SO interaction (Fig. 2a, inset), where the spin-splitting energies are—in
clarity. The dashed lines are fits to a superposition of two Lorentzians. For the first approximation—independent of DQD detuning. The SO interaction
each vacuum Rabi-mode cut shown here, a full map of the avoided crossing primarily couples the dashed |–↑〉 and |+↓〉 states, as highlighted by the colour of
(similar to Fig. 2) is shown in Supplementary Fig. 8. b,c, Angular dependence of the solid lines (increased mixing from black to red). Spin–charge mixing leads to
spin–photon coupling gs (b) and 2gs/(γs + κ/2) (c). The experimental data are in pronounced detuning dependence of the spin-splitting energies. Notice that for
excellent agreement with the theory (solid line). The grey-shaded area outlines ϕ = 0° (e), the spin–charge mixing is stronger than for ϕ = 45° (d), which results in
the magnetic-field orientations where the spin–photon resonance is achieved for a larger spin–photon coupling gs as shown in b.
magnetic fields larger than 1 T, which are inaccessible in our experimental setup.

Rabi-mode splitting cannot be resolved. To further support the exist- of SO interaction. This implies that the spin–photon resonance condi-
ence of a single-dot spin–photon interaction, we measure gs as a func- tion occurs for 2tc > ℏωr. In this regime, the flopping-mode qubit is well
tion of ε (Fig. 4b). We find that gs drops by more than two orders of isolated from the higher-energy levels and strongly coupled to the
magnitude when increasing ε , but tends to saturate once the hole is cavity mode. Moreover, operating the DQD at larger tunnel coupling
fully localized in the right dot. This limit presents potential interest (2tc ≈ 4ℏωr; Fig. 3e) reduces the impact of charge noise on the detuning
since the use of single dots25,26 would simplify the device architecture, energy, thereby enabling large cooperativity.
reduce the number of control parameters, allow for longer hole-spin
coherence, and enable alternative and possibly more efficient spin– Conclusions
photon architectures15,23. In addition, significant progress can be Looking further ahead, we foresee ample room to improve the spin–
expected from the implementation of spin–photon coupling schemes photon interface. On an engineering level, largely reduced resonator
relying on operational sweets spots22,23,25,26, where decoherence is losses with κ/2π < 1 MHz should be readily feasible29,36,37, and further
reduced and efficient electrical control is preserved. improvements harnessing the advanced MOS fabrication platform to
integrate a multilayer superconducting back-end-of-line would allow
Impact of strong SO interaction for a well-controlled microwave environment. We would also like to
The large spin–photon cooperativity observed in our experiment can emphasize that in our hole system, the spin–photon coupling strength
be seen as a combined effect of an efficient charge–photon coupling, is ruled by the geometry and electrostatic design of the DQD (Supple-
favoured by the MOS device layout, as well as the intrinsic SO interaction mentary Section IX) and controlled by the amplitude and direction of
of holes, much stronger than the synthetic SO interaction of electrons the externally applied magnetic field. We expect this should limit the
in silicon. We note that the maximum gs is generally achieved when impact of device-to-device variability and facilitate the development
2tc ≈ ℏωr ≈ EZ. Under this condition, the excited states |–↑〉 and |+↓〉 of large-scale quantum networks.
completely mix and gs approaches gc, regardless of the SO interac- Our work promotes holes in Si MOS devices as a powerful play-
tion strength. In the case of electrons in silicon, however, the weak SO ground for the development of spin circuit QED. As opposed to electrons,
interaction (ℓso ≫ d) cannot keep these two states sufficiently apart holes benefit from an intrinsically strong and versatile SO interaction
from each other to prevent unwanted excitations to the second excited (ℓso ≈ d), which is available without the need of micromagnets. This not
state |+↓〉 (ref. 35). This makes the limit of strong spin–charge mixing only simplifies the device architecture but also allows to engineer an SO
impractical, thereby preventing the achievement of maximum spin– two-level system, which remains well isolated from other energy levels
photon coupling. The hole system studied here does not suffer from of the DQD. Our demonstration of such a hole flopping-mode spin qubit,
this limitation. The level repulsion induced by the stronger SO interac- with an unprecedented spin–photon cooperativity of 1,600, opens the
tion (ℓso ≈ d) keeps the |–↑〉 and |+↓〉 states apart (Fig. 3e). Noticeably, door to a wide range of circuit QED implementations of fundamental and
the zero-detuning excitation energy of the flopping-mode qubit gets practical interest (for example, two-qubit operations between distant
significantly reduced compared with Zeeman splitting in the absence hole spins with fidelities as high as 90% already seem within reach38).

Nature Nanotechnology
Article https://doi.org/10.1038/s41565-023-01332-3

(A/A0)2 5. Clerk, A. A., Lehnert, K. W., Bertet, P., Petta, J. R. & Nakamura, Y.
0.6 0.7 0.8 0.9 1.0 1.1
Hybrid quantum systems with circuit quantum electrodynamics.
ε/h = –116 GHz
Nat. Phys. 16, 257–267 (2020).
a 6. Childress, L., Sørensen, A. S. & Lukin, M. D. Mesoscopic cavity
5.44 quantum electrodynamics with quantum dots. Phys. Rev. A 69,
2 MHz
042302 (2004).
5.43
7. Burkard, G. & Imamoglu, A. Ultra-long-distance interaction
fp (GHz)

between spin qubits. Phys. Rev. B 74, 041307(R) (2006).

8. Hu, X., xi Liu, Y. & Nori, F. Strong coupling of a spin qubit to a
5.42
superconducting stripline cavity. Phys. Rev. B 86, 035314 (2012).
9. Jin, P. Q., Marthaler, M., Shnirman, A. & Schön, G. Strong coupling
5.41 of spin qubits to a transmission line resonator. Phys. Rev. Lett. 108,
190506 (2012).
397 399 0.6 0.8 1.0
10. Frey, T. et al. Dipole coupling of a double quantum dot to a
B (mT) (A/A0)2
microwave resonator. Phys. Rev. Lett. 108, 046807 (2012).
b 11. Petersson, K. D. et al. Circuit quantum electrodynamics with a
Data spin qubit. Nature 490, 380–383 (2012).
2
10
Flopping mode 12. Viennot, J. J., Dartiailh, M. C., Cottet, A. & Kontos, T. Coherent
Flopping mode + SQD coupling of a single spin to microwave cavity photons. Science
1
gS/2π (MHz)

10 349, 408–411 (2015).

13. Burkard, G., Ladd, T. D., Nichol, J. M., Pan, A., and Petta, J.
0
10 R. Semiconductor spin qubits. Preprint at https://arxiv.org/
–1
abs/2112.08863 (2021).
10
14. Imamoglu, A. et al. Quantum information processing using
quantum dot spins and cavity QED. Phys. Rev. Lett. 83, 4204 (1999).
–2
10
–150 –100 –50 0 50 15. Vandersypen, L. M. K. et al. Interfacing spin qubits in quantum
ε/h (GHz) dots and donors—hot, dense, and coherent. npj Quantum Inf. 3,
34 (2017).
Fig. 4 | Spin–photon coupling in the single-dot limit. a, Normalized
16. Samkharadze, N. et al. Strong spin-photon coupling in silicon.
transmission as a function of fp and B at ε/h = −116 GHz and ϕ = 11.25° (left).
Science 359, 1123–1127 (2018).
Frequency linecut at the position of the arrows in the left panel (right). The spin
transition is clearly visible and a double-dip structure split by ~2 MHz is observed
17. Mi, X. et al. A coherent spin–photon interface in silicon. Nature
at resonance. b, gs/2π as a function of energy detuning ε. The DQD flopping-mode 555, 599–603 (2018).
model (black solid line) effectively captures the large spin–photon coupling 18. Borjans, F., Croot, X. G., Mi, X., Gullans, M. J. & Petta, J. R. Resonant
around ε ≈ 0, but underestimates gs by more than one order of magnitude at large microwave-mediated interactions between distant electron spins.
ε, where the hole is confined in a single quantum dot (SQD) on the left (ε > 0) or on Nature 577, 195–198 (2019).
the right (ε < 0). The measured gs is very well reproduced by adding an asymptotic 19. Harvey-Collard, P. et al. Coherent spin-spin coupling mediated by
(R,L) (R)
single-dot spin–photon coupling gs in each quantum dot. We find gs virtual microwave photons. Phys. Rev. X 12, 021026 (2022).
(L)
/2π= 1.16 MHz and gs /2π= 0.66 MHz, as shown by the red solid line. The 20. Luttinger, J. M. & Kohn, W. Motion of electrons and holes in
grey-shaded area outlines the energy detuning range where the spin–photon perturbed periodic fields. Phys. Rev. 97, 869 (1955).
resonance is achieved for magnetic fields larger than 1 T, which are inaccessible in 21. Winkler, R. Spin-Orbit Coupling Effects in Two-Dimensional
our experimental setup. The error bars represent the standard deviation from the
Electron and Hole Systems (Springer, 2003).
fitting.
22. Kloeffel, C., Trif, M., Stano, P. & Loss, D. Circuit QED with hole-spin
qubits in Ge/Si nanowire quantum dots. Phys. Rev. B 88, 241405
(2013).
23. Nigg, S. E., Fuhrer, A. & Loss, D. Superconducting grid-bus surface
Online content code architecture for hole-spin qubits. Phys. Rev. Lett. 118, 147701
Any methods, additional references, Nature Portfolio reporting sum- (2017).
maries, source data, extended data, supplementary information, 24. Mutter, P. M. & Burkard, G. Natural heavy-hole flopping mode
acknowledgements, peer review information; details of author con- qubit in germanium. Phys. Rev. Research 3, 013194 (2021).
tributions and competing interests; and statements of data and code 25. Michal, V. P. et al. Tunable hole spin-photon interaction based on
availability are available at https://doi.org/10.1038/s41565-023-01332-3. g-matrix modulation. Phys. Rev. B 107, L041303 (2023).
26. Bosco, S., Scarlino, P., Klinovaja, J. & Loss, D. Fully tunable
References longitudinal spin-photon interactions in Si and Ge quantum dots.
1. Haroche, S. and Raimond, J.-M. Exploring the Quantum: Atoms, Phys. Rev. Lett. 129, 066801 (2022).
Cavities, and Photons (Oxford Univ. Press, 2006). 27. Kloeffel, C., Rančić, M. J. & Loss, D. Direct Rashba spin-orbit
2. Wallraff, A. et al. Strong coupling of a single photon to a interaction in Si and Ge nanowires with different growth
superconducting qubit using circuit quantum electrodynamics. directions. Phys. Rev. B 97, 235422 (2018).
Nature 431, 162–167 (2004). 28. Niepce, D., Burnett, J. & Bylander, J. High kinetic inductance NbN
3. Blais, A., Huang, R.-S., Wallraff, A., Girvin, S. M. & Schoelkopf, R. J. nanowire superinductors. Phys. Rev. Appl. 11, 044014 (2019).
Cavity quantum electrodynamics for superconducting electrical 29. Yu, C. X. et al. Magnetic field resilient high kinetic inductance
circuits: an architecture for quantum computation. Phys. Rev. A superconducting niobium nitride coplanar waveguide resonators.
69, 062320 (2004). Appl. Phys. Lett. 118, 054001 (2021).
4. Blais, A., Grimsmo, A. L., Girvin, S. & Wallraff, A. Circuit 30. Gorman, J., Hasko, D. G. & Williams, D. A. Charge-qubit operation
quantum electrodynamics. Rev. Mod. Phys. 93, 025005 of an isolated double quantum dot. Phys. Rev. Lett. 95, 090502
(2021). (2005).

Nature Nanotechnology
Article https://doi.org/10.1038/s41565-023-01332-3

31. Crippa, A. et al. Electrical spin driving by g-matrix 37. Harvey-Collard, P. et al. On-chip microwave filters for
modulation in spin-orbit qubits. Phys. Rev. Lett. 120, 137702 high-impedance resonators with gate-defined quantum dots.
(2018). Phys. Rev. Appl. 14, 034025 (2020).
32. Fischer, J., Coish, W., Bulaev, D. & Loss, D. Spin decoherence of a 38. Benito, M., Petta, J. R. & Burkard, G. Optimized cavity-mediated
heavy hole coupled to nuclear spins in a quantum dot. Phys. Rev. dispersive two-qubit gates between spin qubits. Phys. Rev. B 100,
B 78, 155329 (2008). 081412 (2019).
33. Piot, N. et al. A single hole spin with enhanced coherence in
natural silicon. Nat. Nanotechnol. 17, 1072 (2022). Publisher’s note Springer Nature remains neutral with regard to
34. Li, J., Venitucci, B. & Niquet, Y.-M. Hole-phonon interactions jurisdictional claims in published maps and institutional affiliations.
in quantum dots: effects of phonon confinement and
encapsulation materials on spin-orbit qubits. Phys. Rev. B 102, Springer Nature or its licensor (e.g. a society or other partner) holds
075415 (2020). exclusive rights to this article under a publishing agreement with
35. Benito, M., Mi, X., Taylor, J. M., Petta, J. R. & Burkard, G. the author(s) or other rightsholder(s); author self-archiving of the
Input-output theory for spin-photon coupling in Si double accepted manuscript version of this article is solely governed by the
quantum dots. Phys. Rev. B 96, 235434 (2017). terms of such publishing agreement and applicable law.
36. Mi, X. et al. Circuit quantum electrodynamics architecture for
gate-defined quantum dots in silicon. Appl. Phys. Lett. 110, © The Author(s), under exclusive licence to Springer Nature Limited
043502 (2017). 2023

Nature Nanotechnology
Article https://doi.org/10.1038/s41565-023-01332-3

Data availability from C.X.Y., J.C.A.-U., É.D. and R.M. J.C.A.-U. developed the theoretical
The datasets generated during the current study are available via model with help from V.P.M., M.F. and Y.-M.N. S.Z., R.M., J.C.A.-U.,
Zenodo at https://doi.org/10.5281/zenodo.7533669. S.D.F. and Y.-M.N. co-wrote the manuscript with inputs from all the
authors. N.R., H.N, T.B., M.V. and B.B. were responsible for the front-end
Code availability fabrication of the device. R.M. initiated the project.
The code used to analyse the datasets are available via Zenodo at
https://doi.org/10.5281/zenodo.7533669. Competing interests
M.V. is co-founder and CEO of siquance.
Acknowledgements
We thank J.-L. Thomassin and F. Gustavo for help in the fabrication of Additional information
the NbN circuitry and M. Boujard and I. Matei for technical support Supplementary information The online version contains
in the lab. V. Renard is acknowledged for careful proofreading of supplementary material available at
the manuscript. This research has been supported by the European https://doi.org/10.1038/s41565-023-01332-3.
Union’s Horizon 2020 research and innovation programme under
grant agreement nos. 951852 (QLSI project), 810504 (ERC project Correspondence and requests for materials should be addressed to
QuCube) and 759388 (ERC project LONGSPIN), as well as by the Simon Zihlmann or Romain Maurand.
French National Research Agency (ANR) through the project MAQSi.
S.Z. acknowledges support by an Early Postdoc Mobility fellowship Peer review information Nature Nanotechnology thanks Malcolm
(P2BSP2_184387) from the Swiss National Science Foundation. Carroll, Hai-Ou Li and the other, anonymous, reviewer(s) for their
contribution to the peer review of this work.
Author contributions
C.X.Y. fabricated the NbN circuitry with help from S.Z. C.X.Y. and S.Z. Reprints and permissions information is available at
performed the measurements. S.Z. analysed the data with inputs www.nature.com/reprints.

Nature Nanotechnology