Abstract
The prospect of building quantum circuits^{1,2} using advanced semiconductor manufacturing makes quantum dots an attractive platform for quantum information processing^{3,4}. Extensive studies of various materials have led to demonstrations of twoqubit logic in gallium arsenide^{5}, silicon^{6,7,8,9,10,11,12} and germanium^{13}. However, interconnecting larger numbers of qubits in semiconductor devices has remained a challenge. Here we demonstrate a fourqubit quantum processor based on hole spins in germanium quantum dots. Furthermore, we define the quantum dots in a twobytwo array and obtain controllable coupling along both directions. Qubit logic is implemented allelectrically and the exchange interaction can be pulsed to freely program onequbit, twoqubit, threequbit and fourqubit operations, resulting in a compact and highly connected circuit. We execute a quantum logic circuit that generates a fourqubit Greenberger−Horne−Zeilinger state and we obtain coherent evolution by incorporating dynamical decoupling. These results are a step towards quantum error correction and quantum simulation using quantum dots.
This is a preview of subscription content, access via your institution
Relevant articles
Open Access articles citing this article.

Scalable algorithm simplification using quantum AND logic
Nature Physics Open Access 14 November 2022

Phase flip code with semiconductor spin qubits
npj Quantum Information Open Access 27 October 2022

Universal control of a sixqubit quantum processor in silicon
Nature Open Access 28 September 2022
Access options
Subscribe to Nature+
Get immediate online access to Nature and 55 other Nature journal
$29.99
monthly
Subscribe to Journal
Get full journal access for 1 year
$199.00
only $3.90 per issue
All prices are NET prices.
VAT will be added later in the checkout.
Tax calculation will be finalised during checkout.
Buy article
Get time limited or full article access on ReadCube.
$32.00
All prices are NET prices.
Data availability
All data underlying this study are available from the 4TU ResearchData repository at https://doi.org/10.4121/13663442.
References
Terhal, B. M. Quantum error correction for quantum memories. Rev. Mod. Phys. 87, 307–346 (2015).
Reiher, M., Wiebe, N., Svore, K. M., Wecker, D. & Troyer, M. Elucidating reaction mechanisms on quantum computers. Proc. Natl Acad. Sci. USA 114, 7555–7560 (2017).
Loss, D. & DiVincenzo, D. P. Quantum computation with quantum dots. Phys. Rev. A 57, 120–126 (1998).
Vandersypen, L. M. K. et al. Interfacing spin qubits in quantum dots and donors—hot, dense, and coherent. npj Quant. Inform. 3, 34 (2017).
Shulman, M. D. et al. Demonstration of entanglement of electrostatically coupled singlettriplet qubits. Science 336, 202–205 (2012).
Veldhorst, M. et al. A twoqubit logic gate in silicon. Nature 526, 410–414 (2015).
Zajac, D. M. et al. Resonantly driven CNOT gate for electron spins. Science 359, 439–442 (2018).
Watson, T. F. et al. A programmable twoqubit quantum processor in silicon. Nature 555, 633–637 (2018).
Huang, W. et al. Fidelity benchmarks for twoqubit gates in silicon. Nature 569, 532–536 (2019).
He, Y. et al. A twoqubit gate between phosphorus donor electrons in silicon. Nature 571, 371–375 (2019).
Ma¸dzik, M. T. et al. Conditional quantum operation of two exchangecoupled singledonor spin qubits in a MOScompatible silicon device. Nat. Commun. 12, 181 (2021).
Petit, L. et al. Universal quantum logic in hot silicon qubits. Nature 580, 355–359 (2020).
Hendrickx, N. W., Franke, D. P., Sammak, A., Scappucci, G. & Veldhorst, M. Fast twoqubit logic with holes in germanium. Nature 577, 487–491 (2020).
Arute, F. et al. Quantum supremacy using a programmable superconducting processor. Nature 574, 505–510 (2019).
Koppens, F. H. L. et al. Driven coherent oscillations of a single electron spin in a quantum dot. Nature 442, 766–771 (2006).
Petta, J. R. et al. Coherent manipulation of coupled electron spins in semiconductor quantum dots. Science 309, 2180–2184 (2005).
Itoh, K. M. & Watanabe, H. Isotope engineering of silicon and diamond for quantum computing and sensing applications. MRS Commun. 4, 143–157 (2014).
Muhonen, J. T. et al. Storing quantum information for 30 seconds in a nanoelectronic device. Nat. Nanotechnol. 9, 986–991 (2014).
Veldhorst, M. et al. An addressable quantum dot qubit with faulttolerant controlfidelity. Nat. Nanotechnol. 9, 981–985 (2014).
Yoneda, J. et al. A quantumdot spin qubit with coherence limited by charge noise and fidelity higher than 99.9%. Nat. Nanotechnol. 13, 102–106 (2018).
Yang, C. H. et al. Silicon qubit fidelities approaching incoherent noise limits via pulse engineering. Nat. Electron. 2, 151–158 (2019).
Zwanenburg, F. A. et al. Silicon quantum electronics. Rev. Mod. Phys. 85, 961–1019 (2013).
Scappucci, G. et al. The germanium quantum information route. Nat. Rev. Mater. https://doi.org/10.1038/s4157802000262z (2020).
Itoh, K. et al. High purity isotopically enriched 70Ge and 74Ge single crystals: isotope separation, growth, and properties. J. Mater. Res. 8, 1341–1347 (1993).
Bulaev, D. V. & Loss, D. Spin relaxation and decoherence of holes in quantum dots. Phys. Rev. Lett. 95, 076805 (2005).
Lodari, M. et al. Light effective hole mass in undoped Ge/SiGe quantum wells. Phys. Rev. B 100, 041304 (2019).
Lodari, M. et al. Low percolation density and charge noise with holes in germanium. Mater. Quantum. Technol. 1, 011002 (2020).
Hendrickx, N. W. et al. Gatecontrolled quantum dots and superconductivity in planar germanium. Nat. Commun. 9, 2835 (2018).
Lawrie, W. I. L. et al. Quantum dot arrays in silicon and germanium. Appl. Phys. Lett. 116, 080501 (2020).
Pillarisetty, R. Academic and industry research progress in germanium nanodevices. Nature 479, 324–328 (2011).
Bulaev, D. V. & Loss, D. Electric dipole spin resonance for heavy holes in quantum dots. Phys. Rev. Lett. 98, 097202 (2007).
Maurand, R. et al. A CMOS silicon spin qubit. Nat. Commun. 7, 13575 (2016).
Watzinger, H. et al. A germanium hole spin qubit. Nat. Commun. 9, 3902 (2018).
PioroLadrière, M. et al. Electrically driven singleelectron spin resonance in a slanting Zeeman field. Nat. Phys. 4, 776–779 (2008).
Tokura, Y., van der Wiel, W. G., Obata, T. & Tarucha, S. Coherent single electron spin control in a slanting Zeeman field. Phys. Rev. Lett. 96, 047202 (2006).
Sammak, A. et al. Shallow and undoped germanium quantum wells: a playground for spin and hybrid quantum technology. Adv. Funct. Mater. 29, 1807613 (2019).
van Riggelen, F. et al. A twodimensional array of singlehole quantum dots. Appl. Phys. Lett. 118, 044002 (2021).
Hendrickx, N. W. et al. A singlehole spin qubit. Nat. Commun. 11, 3478 (2020).
Lawrie, W. I. L. et al. Spin relaxation benchmarks and individual qubit addressability for holes in quantum dots. Nano Lett. 20, 7237–7242 (2020).
Danon, J. & Nazarov, Y. V. Pauli spin blockade in the presence of strong spin–orbit coupling. Phys. Rev. B 80, 041301 (2009).
Yang, C. H. et al. Charge state hysteresis in semiconductor quantum dots. Appl. Phys. Lett. 105, 183505 (2014).
HarveyCollard, P. et al. Highfidelity singleshot readout for a spin qubit via an enhanced latching mechanism. Phys. Rev. X 8, 021046 (2018).
Knill, E. et al. Randomized benchmarking of quantum gates. Phys. Rev. A 77, 012307 (2008).
Gullans, M. J. & Petta, J. R. Protocol for a resonantly driven threequbit Toffoli gate with silicon spin qubits. Phys. Rev. B 100, 085419 (2019).
Hetényi, B., Kloeffel, C. & Loss, D. Exchange interaction of holespin qubits in double quantum dots in highly anisotropic semiconductors. Phys. Rev. Res. 2, 033036 (2020).
Taylor, J. M. et al. Faulttolerant architecture for quantum computation using electrically controlled semiconductor spins. Nat. Phys. 1, 177–183 (2005).
Veldhorst, M., Eenink, H. G. J., Yang, C. H. & Dzurak, A. S. Silicon CMOS architecture for a spinbased quantum computer. Nat. Commun. 8, 1766 (2017).
Li, R. et al. A crossbar network for silicon quantum dot qubits. Sci. Adv. 4, eaar3960 (2018).
Hensgens, T. et al. Quantum simulation of a Fermi–Hubbard model using a semiconductor quantum dot array. Nature 548, 70–73 (2017).
Seedhouse, A. E. et al. Pauli blockade in silicon quantum dots with spinorbit control. PRX Quant. 2, 010303 (2021).
Russ, M. et al. Highfidelity quantum gates in Si/SiGe double quantum dots. Phys. Rev. B 97, 085421 (2018).
Chan, K. W. et al. Assessment of a silicon quantum dot spin qubit environment via noise spectroscopy. Phys. Rev. Appl. 10, 044017 (2018).
Wang, Z. et al. Suppressing chargenoise sensitivity in highspeed Ge hole spinorbit qubits. Preprint at https://arxiv.org/abs/1911.11143 (2019).
Barnes, E., Kestner, J. P., Nguyen, N. T. T. & Das Sarma, S. Screening of charged impurities with multielectron singlettriplet spin qubits in quantum dots. Phys. Rev. B 84, 235309 (2011).
Acknowledgements
We thank L. M. K. Vandersypen for discussions and S. G. J. Philips for his contributions to software development. M.V. acknowledges support through a Vidi grant, two projectruimte grants, and an NWA grant, all associated with the Netherlands Organization of Scientific Research (NWO).
Author information
Authors and Affiliations
Contributions
N.W.H. performed the experiments and analysed the data. W.I.L.L. fabricated the device. N.W.H. contributed to the design and development. M.R. contributed to theoretical analysis. F.v.R. contributed to the preparation of the experiment. A.S. and G.S. supplied the heterostructures. S.L.d.S. and R.N.S. contributed to software and hardware development. N.W.H. and M.V. wrote the manuscript with input from all other authors. M.V. supervised the project.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Peer review information Nature thanks Hendrik Bluhm, Michel PioroLadrière, David Reilly and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.
Extended data
is available for this paper at https://doi.org/10.1038/s41586021033326.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Extended data figures and tables
Extended Data Fig. 1 Schematic of the measurement setup.
The sample is bonded to a PCB mounted on the cold finger of a dilution refrigerator. Directcurrent (dc) voltages (orange) are applied to the gates using galvanically isolated digitaltoanalogue converters (DACs) through lines filtered at the mixing chamber stage of the refrigerator. The DACs are custom built and part of a custom serial peripheral interface (SPI) measurement rack. Voltage pulses (red) are applied to the same gates using Keysight M3202A AWGs. The Keysight AWGs and digitizer are part of a Keysight M9019A peripheral component interconnect express extensions for instrumentation (PXIe) chassis. The coaxial cables (coax) pass through a ferrite common mode choke and are attenuated at different stages of the fridge and combined with the dc signal using onPCB bias tees. The microwave excitation used for spin resonance is generated by three separate vector sources and is combined with the AWG signal on plungers P2, P3 and P4 through roomtemperature diplexers with pass bands dc400 MHz (LF) and 1.5–10 GHz (RF). The output of the vector sources is modulated by quadrature modulation using signals generated on the AWGs (dark blue). By applying sine waves with a phase difference of π/2 to the inphase (I) and quadrature (Q) inputs of the vector source, a singlesideband signal with controllable amplitude, frequency, phase and duration is acquired. Additionally, we apply a pulse modulation (PM) envelope around the microwave pulses to improve the microwave suppression to a total of −120 dB. Using an inhouse built reflectometry setup, we generate two microwave tones resonant with the tank circuit on the PCB (light blue). The radiofrequency sources (rf src) are pulse modulated by a trigger (trig) generated on the AWG and 50Ω split (split) across all sources, to mute them during qubit manipulation. The radiofrequency signals are combined (rf cmb), filtered and attenuated before reaching a directional coupler at the mixing chamber stage. Here, the signal propagates through two onPCB biastees to the two NbTiN resonators, which form the tank circuits. The reflected signal is then split off and amplified by 35 dB at the 4 K stage of the fridge using a CITLF3 cryogenic amplifier. The radiofrequency signal is amplified (rf amp) again at room temperature (RT) after which it is demodulated separately at both of the carrier frequencies (ref) to retrieve the charge sensor signal. After filtering the signal, it is recorded and temporally averaged \(\bar{V}(t)\) using the fieldprogrammable gate array (FPGA) in the digitizer (digi) to reduce the amount of data transferred to the computer.
Extended Data Fig. 2 Screening of qubit devices.
a, The qubit devices undergo a visual screening as well as a transport screening at a temperature of T = 4.2 K. Out of the full batch of 16 nominally identical devices, 15 passed visual inspection. Seven of these devices were tested at T = 4.2 K and two devices were found to pass all testing, of which one (the device presented in this work) was mounted in a dilution refrigerator. b, Transport current I at a temperature of 4.2 K for the device presented in this work. Three different channels are turned on by sweeping all gates down to V_{gate} = −1,500 mV (black), thereby accumulating charge in the undoped strained Ge quantum well. Next, the effect of the individual gates is tested by sweeping them up and down to V_{gate} = 0 V (coloured lines). Both the north and south single hole transistor (SHT) sensors are tested, as well as a transport channel through the full device. All channels turn on and the gates affect the transport current as expected from the device layout.
Extended Data Fig. 3 Readout characteristics.
a, b, We measure the difference in charge sensor signal between the blocked and nonblocked states as a function of the measurement time at the readout point. An exponential decay can be observed related to the tunnelling time T_{in} of Q2 (or Q4) to the reservoir for the Q1Q2 (a) and Q3Q4 (b) readout system, respectively. c–f, We vary the ramp time t_{ramp} between the manipulation phase and the readout phase and measure the blocked state probability P_{blocked} of the four different twoqubit basis states by applying preparation π pulses to the relevant qubits, both for the Q1Q2 readout system (a, b) and the Q3Q4 readout system (c, d). By increasing the interdot coupling during the readout and elongating the ramp between the manipulation and readout point, we can switch between a parity readout (a, c) and a singlestate readout (b, d). The dashed line corresponds to the optimized readout ramp time used for our measurements.
Extended Data Fig. 4 Randomized benchmarking of the Clifford group and spin relaxation times of the different qubits.
a, We quantify the quality of the singlequbit gates by performing randomized benchmarking of the singlequbit Clifford group^{43}. The decay curve of the qubit state is measured as a function of the number of Clifford gates applied. Each data point consists of 1,000 single shots for 30 different randomly selected Clifford sequences of length N_{Cliffords}. The decay is fitted to P_{up} = aexp(−(N_{Cliffords}m)) + y_{0}, with a the initial spinup probability, m the decay parameter, and y_{0} an offset. F = 1 – m/(2 × 1.875) is extracted based on the average single qubit gate length of 1/1.875 Clifford gates. Uncertainty margins (in parentheses) correspond to 1σ. b, The spin relaxation time T_{1} is measured at the manipulation point by applying a π_{X}pulse separated by a waiting time t_{wait} from the readout phase. By fitting the normalized spinup fraction to P = exp(−t_{wait}/T_{1}), we find spin relaxation times of \({T}_{1}^{{\rm{Q}}1}\) = 0.84(6) ms, \({T}_{1}^{{\rm{Q}}2}\) = 7.6(5) ms, \({T}_{1}^{{\rm{Q}}3}\) = 16.1(8) ms, and \({T}_{1}^{{\rm{Q}}4}\) = 11.5(5) ms. Uncertainty margins (in parentheses) correspond to 1σ.
Extended Data Fig. 5 Ramsey, Hahn echo and CPMG measurements on the different qubits.
a, The phase coherence time \({T}_{2}^{* }\) is measured using a Ramsey sequence consisting of two X(π/2)pulses separated by a waiting time τ as illustrated in the schematic at the top. By fitting the data to \(P=\,\cos (2{\rm{\pi }}\Delta f\tau +{\varphi }_{0})\exp ({(\tau /{T}_{2}^{* })}^{\alpha })\), with Δf the frequency detuning, ϕ_{0} a phase offset and α the power of the decay, we find spin dephasing times of \({T}_{2,{\rm{Q}}1}^{* }\) = 201 ns, \({T}_{2,{\rm{Q}}2}^{* }\) = 146 ns, \({T}_{2,{\rm{Q}}3}^{* }\) = 445 ns, and \({T}_{2,{\rm{Q}}4}^{* }\) = 150 ns for Q1 to Q4, respectively. b, Using an additional X(π)pulse, lowfrequency fluctuations of the qubit resonance frequency can be echoed out, allowing us to probe the Hahnecho decay time \({T}_{2}^{{\rm{Hahn}}}\). Fitting the data to \(P=\exp ({(\tau /{T}_{2}^{{\rm{Hahn}}})}^{\alpha })\), we find Hahn echo times of \({T}_{2,{\rm{Q}}1}^{{\rm{Hahn}}}\) = 4.3 μs, \({T}_{2,{\rm{Q}}2}^{{\rm{Hahn}}}\) = 5.5 μs, \({T}_{2,{\rm{Q}}3}^{{\rm{Hahn}}}\) = 3.8 μs, and \({T}_{2,{\rm{Q}}4}^{{\rm{Hahn}}}\) = 2.9 μs. c, Using a CPMG sequence of repeated Y(π) pulses, we can increase the echo bandwidth and extend the phase coherence to over \({T}_{2,{\rm{Q}}1}^{{\rm{CPMG}}}\) >100 μs. The phase coherence can be observed to increase with the amount of refocusing pulses N_{π} (left), with exemplary decay traces for Q1 plotted in the right panel.
Extended Data Fig. 6 Noise spectroscopy using Ramsey and CPMG measurements.
We measure the effective noise spectrum acting on the qubit, both tracing the resonance frequency using repeated Ramsey measurements^{20} (in blue), as well as by using the filter function of a dynamical decoupling measurement^{18,52} (in red). Dashed blue and red lines are fits to the Ramsey and CPMG data respectively. The black line is a fit to the combined dataset, where the weight of both sets is normalized to the amount of data points. The effective noise can be observed to increase towards low frequencies, consistent with the upwards trend of \({T}_{2}^{{\rm{CPMG}}}\) observed in Extended Data Fig. 3c. The effective charge noise measured in this heterostructure is S_{cn}(f) = 6 μV \(\sqrt{{\rm{Hz}}}\)^{–1} at 1 Hz (ref. ^{27}). Combining this with a typical resonance frequency slope of df/dV = 5 MHz mV^{–1} (ref. ^{39}) results in an effective resonancefrequency noise power of S(f) = 9 × 10^{8} Hz^{2} Hz^{–1}, comparable to what is observed experimentally, suggesting that coherence is limited by charge noise in our system. The effect of charge noise could be mitigated by careful optimization of the electric field environment^{53} or moving to a multihole charge occupancy, screening the influence of charge impurities^{54}, potentially enabling even higherfidelity operations. Alternatively, noise could originate in the nuclear spin bath present in natural germanium, which could be overcome by isotopically enriching the material.
Extended Data Fig. 7 Driving of all resonance lines of the coupled three and fourqubit system.
a, Both the coupling between Q2 and Q1 as well as Q2 and Q3 are enabled, using the respective virtual barrier gates. This splits the resonance line into four, as shown in Fig. 3. Driving each of the separate lines results in the conditional rotations of Q2 depending on the states of Q1 and Q3. We measure the spinup probability P_{up} after driving each of the four resonance lines for time t_{p}, for all four permutations of the Q1 and Q3 basis states as initial state, following the colour scheme of Fig. 3. The driving power is adjusted for each of the transitions to synchronize the πrotation times, with \({a}_{{f}_{1}}\) = 330 mV, \({a}_{{f}_{2}}\) =500 mV, \({a}_{{f}_{3}}\) = 280 mV, and \({a}_{{f}_{4}}\) = 400 mV, for f_{1}−f_{4} from low to high. b, Similarly, by additionally opening up the coupling between Q3 and Q4 as well, the resonance line splits into eight and we can drive all separate lines individually. The eight lines are driven using the same microwave power in this figure and a strong difference in rotation frequencies can be observed for the different transitions f_{1}−f_{8} from low to high. This also results in a small offresonant driving effect for some of the lines.
Extended Data Fig. 8 Tuning of the CZgates.
a, b, The CZ gates between all four qubit pairs are tuned using a Ramsey sequence (analogous to Fig. 4), where the spinup probability is measured as a function of the phase θ of the final π/2 pulse as well as the depth of the exchange pulse V_{vBmn}, with m and n the relevant qubits (a). We choose to tune the height of the voltage pulse rather than its length, owing to the limited temporal resolution of the exchange pulses (1 ns). The acquired phase θ_{0} is obtained by fitting each line to P = Acos(θ + θ_{0}) + y_{0}, with A the visibility and y_{0} an offset. A CZ gate is achieved when the difference in acquired phase is exactly Δθ = π, for the situation where the control qubit is ↓⟩ (blue) compared to ↑⟩ (orange). The barrier gate voltage at which this occurs is obtained from the intersection of two locally linear fits to the extracted acquired phase (b). c, The CZ gates between all four qubit pairs probed by a Ramsey sequence using the inverse qubits as target and control as compared to the data in Fig. 4b. The target qubit is marked in boldface.
Extended Data Fig. 9 Time evolution of the fourqubit GHZ state.
a, Circuit diagram of the experiments performed in b, c. We first apply a preparation pulse to Q3 and then generate a fourqubit GHZ state analogously to Fig. 5. Next we let the entangled system evolve for time t_{wait}, then apply an optional Y^{2} decoupling pulse and finally disentangle the GHZ state again. b, c, We vary both the waiting time t_{wait} and preparation time t_{prep} and plot the nonblocked state fraction for Q3Q4 in the case without (b) and with (c) a decoupling pulse. It can be clearly observed that without the echo pulse, the system has fully decohered at the end of the algorithm. However, by applying the decoupling pulse, the coherence of the entangled system can be maintained for a prolonged timescale, with a characteristic decay time of τ = 390 ns.
Extended Data Fig. 10 Dephasing of the fourqubit GHZ state.
We model the quantum circuit performed in Fig. 5 and account for qubit decoherence by applying a depolarizing channel Λ_{λ}(ρ) = λρ + [(1 − λ)/d]\({\mathbb{1}}\), with ρ the density matrix, λ the depolarization parameter, d the dimension of the Hilbert space, and \({\mathbb{1}}\) the identity operator. We plot the expected measurement outcomes for qubit pairs Q1Q2 (blue) and Q3Q4 (orange). The top row corresponds to the case of perfect coherence in each panel. In the centre row, we fit the depolarization parameter to the measurement data. The finite rotations visible in panels IV and V in Fig. 5b can be reproduced by including gate or readout errors in the model. Finally, the bottom row corresponds to a full depolarization of the state. If the qubit system is completely dephased at any point in time, no recovery of the signal can be observed in panel IX.
Rights and permissions
About this article
Cite this article
Hendrickx, N.W., Lawrie, W.I.L., Russ, M. et al. A fourqubit germanium quantum processor. Nature 591, 580–585 (2021). https://doi.org/10.1038/s41586021033326
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1038/s41586021033326
This article is cited by

Scalable algorithm simplification using quantum AND logic
Nature Physics (2023)

Phase flip code with semiconductor spin qubits
npj Quantum Information (2022)

A silicon singlet–triplet qubit driven by spinvalley coupling
Nature Communications (2022)

Universal control of a sixqubit quantum processor in silicon
Nature (2022)

A single hole spin with enhanced coherence in natural silicon
Nature Nanotechnology (2022)
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.