Abstract
Measuring the state of a qubit is a key fundamental operation of a quantum computer. Highfidelity singleshot readout of superconducting qubits can be achieved using parametric amplifiers at millikelvin temperatures. However, scaling parametric amplifiers beyond hundreds of qubits is challenging due to practical size and power limitations. Nanobolometers can, in contrast, offer scalability, sensitivity and speed suitable for qubit readout. Here we show that a bolometer can provide singleshot qubit readout with a readout duration of 13.9 μs and a singleshot fidelity of 0.618. The fidelity is mainly limited by the energy relaxation time of the qubit (28 μs), and a fidelity of 0.927 is found after removing errors arising from this relaxation. In the future, higherfidelity singleshot readout may be achieved through improvements in chip design and experimental setup, as well as a change in the bolometer absorber material to reduce the readout time to the level of hundreds of nanoseconds and below.
Similar content being viewed by others
Main
Qubit readout in a quantum computer is required to determine the result at the end of a computation^{1,2} as well as for error correction, which is necessary for fault tolerance^{3,4,5}. Superconducting qubits are a promising platform for practical quantum computing^{6,7,8}, and readout is one of the main bottlenecks restricting the development of quantumerrorcorrected largescale superconducting quantum processors. In particular, recent experiments have shown that the readout phase is a key source of error in error correction cycles^{9,10,11}.
The most common method of measuring the state of superconducting qubits in the framework of circuit quantum electrodynamics is known as dispersive readout^{6,12,13}. Here a qubit is offresonantly coupled to a readout resonator, the frequency of which shifts depending on the qubit state. With this approach, singleshot readout fidelity above 99% with an averaging time below 100 ns has been achieved for single qubits^{14,15}, and 97–99% on average for the simultaneous multiplexed readout of several qubits^{10,11,16}.
To achieve a signaltonoise ratio (SNR) sufficient for highfidelity singleshot dispersive readout, the output signal from the readout resonator is typically amplified at the millikelvin stage by a parametric amplifier, such as a Josephson parametric amplifier^{17} or a travellingwave parametric amplifier (TWPA)^{18}. These amplifiers can be quantum limited, where the noise added by the amplification stems solely from the Heisenberg uncertainty principle for the inphase and quadrature components of the amplified signal^{19,20,21,22}. Parametric amplifiers can offer high gain and low noise, but introduce challenges in terms of scaling to large numbers of qubits. These challenges include narrow bandwidth, which is undesirable in multiplexed qubit readout^{16}. Although TWPAs provide a broad bandwidth, they typically incorporate more than 10^{3} Josephson junctions, making their onchip footprint large and highyield largescale fabrication challenging. Importantly, both types of amplifier currently require strong isolation between the amplifier and qubit–resonator system: Josephson parametric amplifiers work in the reflection mode via a circulator, and TWPAs utilize a relatively strong pumping tone potentially near the qubit or resonator frequencies. TWPAs also amplify vacuum and other noise, which may reflect back into the TWPA input and cause decoherence in the qubit. To mitigate these issues, permanentmagnetbased microwave isolators are typically placed between the readout transmission line and amplifier input. Such isolators introduce losses in the signal. They are also large, costly and require shielding to protect both qubit and amplifier from the magnetic fields they introduce.
These scalability issues have driven the development of alternative readout techniques. Microwave photon counters offer an alternative to voltage amplification at the millikelvin stage, but require a more involved pulse sequence and suffer from backaction arising from the creation of quasiparticles and their tunnelling. Qubit readout can also be achieved without a parametric amplifier by driving the qubit to the second excited state before readout^{23}. However, the required nonparametric microwave amplifier introduces hightemperature noise close to the qubit frequency, which necessitates bulky microwave isolators. Another approach to promote scalability is to deliver signals to and from the cryostat in the optical domain^{24,25}.
Nanobolometers^{26,27,28} have been shown to be fast and sensitive enough for the readout of superconducting qubits, reaching thermal time constants in the range of hundreds of nanoseconds and energy resolution of a few typical microwave photons^{29}. Such bolometers have several attractive properties for qubit readout. In contrast to parametric amplifiers, bolometers can be driven by a subgigahertz probe tone^{26}, a frequency well below the typical qubit and resonator frequencies of 4–8 GHz. In addition, bolometers can be probed with a low power of approximately −120 or −130 dBm at the chip^{27}, whereas TWPAs typically require pump powers of −75 dBm or higher^{18}. Furthermore, the absorbing port of the bolometer can be conveniently matched to 50 Ω (ref. ^{28}). The absorber presents a cold bath to the readout transmission line, allowing for the detection frequency to vary by orders of magnitude given a fixed probe frequency^{30}. These features may remove the need for isolators between the qubit–resonator system and the bolometer; together with the small size of the bolometers, this makes them promising in terms of scalability.
Since a bolometer measures power, or photon number, it is not bound to add quantum noise stemming from the Heisenberg uncertainty principle. The vacuum noise does not promote detection events in the bolometer since no energy can be extracted from a vacuum. Thus, bolometric readout is fundamentally different from the readout utilizing voltage amplification. In addition, bolometers are relatively simple to fabricate and operate^{28}. They do not require engineering of many small Josephson junctions, and a bolometer requires only a single continuous probe tone with two parameters—power and frequency—to optimize performance.
In this Article, we report the integration of a sensitive bolometer at millikelvin temperatures with the readout circuitry of a superconducting qubit. We demonstrate singleshot qubit readout at a fidelity of 0.618. Our experiment does not use the fastest compatible bolometers^{29}, but those that are orders of magnitude slower^{28}, and thus, bolometric singleshot qubit readout has great potential for the highfidelity scalable readout of superconducting qubits.
Qubittobolometer setup
A schematic of the experimental setup is shown in Fig. 1 (Extended Data Fig. 1 provides further details). The qubit–resonator system and bolometer reside on separate chips (Extended Data Fig. 2). The qubit chip consists of six typical fluxtunable Xmon qubits^{31}, each with a capacitively coupled coplanar waveguide readout resonator in a notch configuration. The readout resonators share a common readout feedline and the flux is adjusted using a shared external magnetic coil; otherwise, the qubits are uncoupled and individually addressable via separate qubit drive lines (Methods). The qubit–resonator pairs have varying qubit and resonator frequencies, as well as qubit–resonator coupling strengths.
We choose a qubit with a transition frequency of f_{q} = 7.655 GHz at the flux sweet spot, anharmonicity α/(2π) = −273 MHz, energy relaxation time of T_{1} = 28.0 μs and Ramsey dephasing time of T_{2} = 7.6 μs. This qubit is coupled to a readout resonator with a coupling strength of g/(2π) = 61 MHz. The resonator has a fundamental resonance frequency of f_{r,g} = 5.473 GHz if the qubit is in its ground state g〉 and linewidth of κ_{r}/(2π) = 1.0 MHz. These parameters produce a dispersive shift of χ/2π = −0.3 MHz, that is, the resonator frequency shifts to f_{r,e} = f_{r,g} + 2χ/(2π) if the qubit is in its excited state e〉.
To read the qubit state, a rectangular microwave pulse of length t_{RO} ≈ 10 μs, frequency f_{d} and power P_{d} is applied to the feedline of the readout resonator. In a typical dispersive readout, the drive frequency is chosen to be in the middle of the dressed resonator frequencies, that is, f_{d} = (f_{r,g} + f_{r,e})/2, and the photons reflected from the resonator accumulate a phase shift depending on the qubit state^{6}. In contrast, here we operate in the photodetection mode^{32} by driving close to one of the dressed frequencies, namely, f_{d} ≈ f_{r,g}. In this driving scheme, information about the qubit state is mostly carried by the power of the signal, that is, the number of photons emitted by the resonator into the feedline. A power difference is necessary for the bolometer, since it is a thermal detector, sensitive to the power but insensitive to the phase of its input signal. Note that for typical dispersive readout, the ratio ∣χ∣/κ_{r} = 1/2 yields the optimal SNR^{33}, whereas for photodetectionbased readout, the optimal ratio is higher^{32}.
The output of the feedline for the readout resonator is connected to the absorber port of the bolometer chip, which is similar to the device discussed elsewhere^{28}. The main component of the bolometer is a resistive Au_{x}Pd_{1−x} (x ≈ 0.6) nanowire. A segment of the nanowire works as the resistive absorber and the remaining wire is interrupted by a series of superconducting Al islands, forming a chain of superconductor–normal metal–superconductor (SNS) junctions. The impedance Z(T_{e}) of this junction chain depends on the electron temperature of the nanowire, T_{e}. The junction chain is embedded in an effective LC circuit formed by a shunt capacitor of C_{1} = 134 pF in parallel with Z(T_{e}), which can be modelled as a parallel resistance and inductance^{27}. The nanowire is grounded between the absorber and junctions, so that the essentially purely realvalued impedance of the absorber does not contribute to the LC circuit.
The bolometer is probed by reflecting a continuous tone at power P_{p} and frequency f_{p} from the gate capacitor C_{g} = 0.87 pF. With a low P_{p} value, the LC circuit resonates at a frequency of f_{b} = 585 MHz with a linewidth of 7.6 MHz. As radiation is absorbed by the absorber, T_{e} increases, which shifts f_{b} down, and thus, by observing changes in the reflection coefficient Γ at the gate capacitor, it is possible to detect the radiation incident on the bolometer input. The reflected probe signal is amplified by a lownoise highelectronmobilitytransistor (HEMT) amplifier at 4 K, and further amplified, demodulated and digitized at room temperature in a heterodyne configuration.
The shift in f_{b} due to the readout pulse coming from the readout resonator is observed as a change in the digitized voltage (Fig. 2a,b). For low P_{p}, the readout pulse causes T_{e}, and thus, the reflected signal to exponentially approach a steadystate value with a thermal time constant τ_{b}. The time constant depends on f_{p}, P_{p} and the power of the readout pulse P_{d} in a nontrivial way due to electrothermal feedback^{27,28}. For the relevant parameter regime considered here, τ_{b} varies between 10 μs and 1 ms (Extended Data Fig. 3c). In particular, we have τ_{b} ≳ T_{1}, which implies that for qubit readout, we must operate the bolometer in a calorimetric fashion, that is, with t_{RO} < τ_{b}. This is highlighted in Fig. 2a, where the time constant, extracted from a measurement with a long readout pulse of t_{RO} > 1 ms, is τ_{b} = 36.2 μs. With t_{RO} = 10 μs, which is more feasible for qubit readout than t_{RO} > τ_{b}, the steady state is far from the reached maximum signal level.
For highenough P_{p}, the electrothermal feedback results in bistability for the electron temperature of the SNS junction. This bistability can be exploited for highfidelity photodetection by operating the bolometer in the latching mode using a pulsed probe tone^{27}. However, this scheme introduces substantial dead time for the detection, and requires detailed calibration of the pulse shape. For simplicity, we, thus, focus on the continuous detection scheme with a sufficiently low P_{p} value.
Using the timedomain data digitized from the bolometer output, we define the detector signal as
where \({\bar{V}}_{0}\) is the time average of the digitized voltage before the readout pulse and \(\bar{V}\) is the time average of the voltage over some averaging window [t_{0}, t_{RO}]. Since t_{RO} < τ_{b}, increasing the time over which the average is taken can substantially decrease \(\bar{V}\), but may increase the SNR. In practice, we observe that choosing t_{0} = 0.75 × t_{RO} balances these two effects reasonably well.
Both qubit–resonator and bolometer chips are mounted on their individual sample holders, which are placed in separate magnetic shields and attached to the mixingchamber plate of a dilution refrigerator with a base temperature of 12 mK. Note that in the setup presented in Fig. 1, an additional microwave switch and directional coupler are placed between the qubit and bolometer. These allow measuring the qubit and driving the bolometer individually, and are only used for the initial separate characterization of the qubit and bolometer. The auxiliary readout channel called for the use of an isolator between the chips, but this can be removed in future experiments.
Qubit measurements
During the initial characterization, we apply no driving to the readout resonator of the qubit and find the bolometer resonance by measuring the reflection coefficient of the bolometer probe signal as a function of f_{p} and P_{p}. Next, we apply pulses to the readout resonator with f_{d} well detuned from f_{r,g}, and map the detector signal S as a function of f_{p} and P_{p}. These results (Fig. 2c and Extended Data Fig. 3) provide us with a feasible operation point (f_{p}, P_{p}) = (580.5 MHz, −116.7 dBm) where S is maximized with P_{p} below the region where the electrothermal feedback induces bistability, as discussed above. Table 1 summarizes the parameter values used during the initial characterization.
With the bolometer operation point fixed, we carry out common qubit characterization measurements^{34} and monitor the bolometer signal. First, we use singletone spectroscopy to find the resonator frequency (Extended Data Fig. 6a), and confirm that it matches the results obtained from the initial characterization carried out via the auxiliary readout port. We also measure S as a function of the qubit readout power P_{d}, both on resonance with f_{d} = f_{r,g} and off resonance with f_{d} = f_{r,g} − 70 MHz, with no excitation pulse applied to the qubit. These results are shown in Fig. 2d for different readout pulse durations t_{RO}. The difference between the onresonance and offresonance values of S corresponds to the maximum contrast attainable with this readout scheme if the dispersive shift χ of the qubit–resonator system was infinite. Importantly, we observe that the suitable values of P_{d} for qubit readout are near the minimum power the bolometer can detect in the given integration time. The highest feasible readout power is P_{d} ≈ −108 dBm, above which the onresonance signal exhibits a sudden jump since the resonator becomes nonlinear and the quantum nondemolition (QND) nature of the readout breaks down^{35,36}. With a very long t_{RO} of 2 ms, a good signal is achieved for P_{d} well below the onset of nonlinearity. However, reducing t_{RO} to 1 μs, which is still an order of magnitude longer than current stateoftheart qubit readout, a reasonable SNR is not achieved even with P_{d} ≈ −108 dBm. The low SNR with short t_{RO} further highlights that the long bolometer time constant limits the readout signal. Fortunately, a graphenebased bolometer^{29} seems suitable for operation at the 100 ns timescale.
After characterizing the resonator, we find an estimate for the qubit frequency using twotone spectroscopy (Extended Data Fig. 6b). Figure 2e shows the subsequent results of a Rabi oscillation measurement where we initialize the qubit to its ground state, drive it with a rectangular pulse of variable length and frequency, and measure the readout resonator using the bolometer. We observe a clear chevron pattern as desired, from which we determine a πpulse length of 100 ns. Here we employ a 10 μs readout pulse and ensemble average the result 512 times. Importantly, we do not observe any drift in the bolometer signal as a function of the pulse length and the contrast between the local minima and maxima in Fig. 2e does not noticeably deteriorate even with a 1 µs Rabi length, indicating that the qubit drive pulse is not leaking to the bolometer.
With the qubit frequency and πpulse length calibrated, we carry out singleshot qubit readout by alternating between preparing the qubit in g〉 and e〉, and recording S with no ensemble averaging. For each prepared state, we record 10^{4} data points. The data are binned to produce a histogram, with the bin width chosen using Scott’s rule^{37}. To maximize the readout fidelity, defined as F = 1 − P(g∣e) − P(e∣g), we optimize the threshold value for S to decide the measurement outcome of g〉 or e〉. Here P(a∣b) is the probability of measuring the qubit in state a〉 provided that it was prepared in state b〉, where the probabilities are obtained from the measured distributions. The highest fidelity is obtained by increasing the readout pulse length to t_{RO} = 20 μs, so that the abovediscussed choice of t_{0} = 0.75 × t_{RO} yields an averaging time of 5 μs for \(\bar{V}\), and setting the readout power to P_{d} = −108 dBm, which is just below the point of nonlinearity for the readout resonator. Figure 3a shows the measured probability distributions for the probe signal S with these parameters, from which we extract the fidelity as F = 0.49.
In an effort to further optimize the singleshot readout fidelity, we carry out an additional experiment, where we intentionally stretch the readout pulse to be unreasonably long (that is, 40 μs) and instead of just storing S derived from the timeaveraged quantities \({\bar{V}}_{0}\) and \(\bar{V}\), we record the full time traces of the bolometer output signal for 1,000 single shots. With these data, we may vary the effective readout pulse length and averaging time of \(\bar{V}\) during postprocessing. Figure 3c shows the resulting singleshot readout fidelity as a function of the digitally determined pulse length and averaging time. We observe a region of relatively high fidelity (F > 0.600; Fig. 3c (red boundary)), with the highest fidelity of 0.618 achieved with t_{RO} = 13.9 μs and t_{RO} − t_{0} = 10.6 μs. Figure 3b shows the probability distributions of the signal with these parameter values.
The fidelity is overall higher in Fig. 3b,c than in Fig. 3a. This is mostly because in the latter, \({\bar{V}}_{0}\) is individually calculated for each single shot. Since each value of \({\bar{V}}_{0}\) is averaged over a relatively short 1.1 µs window, it introduces considerable noise into the value of S calculated using equation (1). In contrast, a common value of \({\bar{V}}_{0}\) is used for the whole dataset in Fig. 3b,c, meaning that only the fluctuations in \(\bar{V}\) contribute to the noise in S. Rescaling the standard deviation of the data in Fig. 3b,c by a factor corresponding to the noise from this 1.1 µs window reproduces the increased noise level shown in Fig. 3a.
For the qubit prepared in e〉, the singleshot probability distribution is clearly bimodal (Fig. 3a,b). This is expected, since t_{RO} is comparable with the qubit energy decay time T_{1}, and thus, the qubit experiences considerable spontaneous relaxation during the readout. Even in the case of infinitely narrow distributions and ideal state preparation, the T_{1} decay produces an error of approximately 1 − exp(−t_{RO}/(2T_{1})) ≈ 22% for t_{RO} = 13.9 μs and T_{1} = 28.0 μs (ref. ^{38}). The long readout stems from the relatively long effective thermal time constant of the bolometer, namely, τ_{b} = 9.4 μs, extracted from the data in Fig. 3c for the qubit prepared in g〉. In Fig. 3c, the trend of relatively high and constant fidelity obtained for simultaneously increasing readout pulse length and averaging time indicates that the increase in fidelity owing to the increasing bolometer SNR approximately compensates the decrease in fidelity owing to the increasing qubit decay. With short readout times, the fidelity is low because the distributions corresponding to the different qubit states are not well separated.
The data shown in Fig. 3a, as well as the g〉state data shown in Fig. 3b, are well modelled by a sum of two Gaussian distributions. This simple model deviates from the e〉state data shown in Fig. 3b between the centres of the Gaussian distributions, since it does not take into account the T_{1} decay of the qubit during readout. Fortunately, we capture this effect using an advanced model (Methods). For the e〉state data shown in Fig. 3a, there is no substantial difference between the Gaussian and advanced models, since the effect of decay on the distribution is masked by noise. Using the parameters extracted from these fits, we determine the fidelity with t_{RO} = 13.9 μs and the T_{1} error removed to be F ≈ 0.927 (Methods). The e〉state data shown in Fig. 3b deviate from the fit above roughly 150 mV. We attribute this to qubit excitations outside the computational subspace owing to the high readout power.
Conclusions
We have shown that a thermal detector can achieve reasonablefidelity singleshot readout of a superconducting qubit, measuring the power of the qubit readout signal, not its voltage. We reported qubit characterization measurements with less ensemble averaging than in previous setups lacking millikelvin amplifiers. Using a 13.9 µs readout pulse, we achieve a singleshot readout fidelity of 0.618, and a fidelity of 0.927 after removing errors arising from a finite qubit lifetime.
Several factors in our experiment can be improved to increase the singleshot readout fidelity. The SNR in the singleshot readout is fairly low, which mainly arises from the long effective thermal time constant of the bolometer—on the order of tens of microseconds. This requires the use of a long readout pulse, which degrades the fidelity due to T_{1} decay of the qubit. To reduce the time constant, a material with lower heat capacity than AuPd—such as graphene—can be used as the absorber^{29,39}.
The long thermal time constant also introduces a long idle time between consecutive qubit measurements and prevents a demonstration of the QND nature of the bolometric readout. Fortunately, and regarding the qubit–resonator interactions, our scheme is essentially identical to the typical dispersive readout, the QND property of which is well known^{6,14,15,19,38}. Minor modifications to our experimental setup may improve the SNR. In this work, we have a number of additional microwave components between the qubit and bolometer, which may be removed to reduce the signal loss between the two chips. Ultimately, the bolometer may be directly connected to the readout feedline, either on chip or in a layered flipchip architecture^{40}. Having the qubit and bolometer on separate chips is an important factor in degrading the fidelity in our experiment. By placing both devices on the same chip and optimizing the impedance matching between the readout resonator and bolometer absorber, undesired reflections can be eliminated (Methods and Extended Data Fig. 5).
The qubit chip design may also be optimized for photodetectionbased readout. In our design, the readout resonator is placed in a notchtype configuration with symmetric coupling to the input and output feedlines. This is disadvantageous as half the signal reflected from the resonator is lost as it escapes through the input port. Changing to a transmission configuration with a weak coupling to the input port could substantially increase the power incident on the bolometer. The ratio ∣χ∣/κ_{r} = 0.3 in our sample is close to the value of 1/2, which maximizes the SNR in a typical dispersive readout, but it is suboptimal for photodetectionbased readout where a larger ratio is desirable. Furthermore, our readout resonator lies below the qubit frequency, which leads to the onset of nonadiabatic and chaotic resonator dynamics at a much lower photon number than for an elevated readout frequency^{13,35,36}. Thus, designing the readout circuit to resonate well above the qubit frequency may lead to a greatly increased signal power at the bolometer input.
The readout fidelity can be improved by advanced pulsing schemes and improved data analysis methods. Instead of a simple rectangular pulse, the resonator can be driven by a twostep pulse^{14}. This may enable the use of a power level that the bolometer can detect with a higher SNR, and still maintain the QND nature of the readout. Alternatively, the SNR of the signal reaching the bolometer may be increased by preparing the qubit in a highly excited state before readout^{23}, or by using a twotone drive that induces an effective longitudinal coupling between the qubit and bolometer^{41,42}. Data analysis can be improved using optimized signal processing. For example, the time averaging can be weighted by the separation between the average trajectories corresponding to g〉 and e〉 (ref. ^{14}), or the singleshot trajectories can be classified using a machine learning algorithm^{43}. Here we have neglected the falling edge of the voltage signal that can be used in the advanced analysis methods to increase the SNR. The averaging over the falling edge does not increase the T_{1}related infidelity since during this averaging time as the readout pulse is off, and hence, the possible qubit decay will not lead to a change in the bolometer signal. Using a graphene bolometer in the calorimetric mode, which accounts for the falling edge, could improve the readout fidelity (Methods).
The next step for bolometric readout is to demonstrate its scalability. Placing qubits and bolometers on the same chip to perform highfidelity readout without the need for isolators is yet to be experimentally demonstrated, to the best of our knowledge. Qubit readout should be multiplexed to minimize the number of necessary control lines. Multiple qubits can be read by coupling their readout resonators to individual bolometers and multiplexing the bolometers via a shared probe line. A chip containing three bolometers, which can probe in a frequencymultiplexed way, is being explored (Methods and Extended Data Fig. 7).
In largescale quantum processors, it may be desirable to produce a binary output from the qubit readout signal inside the cryostat, thereby evading the need for roomtemperature feedback. The bolometer can possibly be used for such a readout by operating in the bistable regime^{27}, or by coupling the output to a clicktype detector such as a Josephson bifurcation amplifier^{44}. By incorporating the SNR improvements discussed above, we estimate that it may be possible to carry out singleshot qubit readout at 99.9% level of fidelity using bolometers (Methods). Thus, bolometers are a promising component for the scalable highfidelity readout of superconducting qubits due to their low power consumption, large probetone frequency offset from qubits, resilience against quantum noise, small footprint, reduced need for microwave isolators and naturally introduced lowtemperature bath for qubits that can be impedance matched at a broad range of readout frequencies.
Methods
Bolometer fabrication
To fabricate the bolometer, we begin with a fourinch silicon wafer (100) of high resistivity (ρ > 10 kΩ cm), covered by a 300 nm thermal oxide. Then, we sputter 200 nm of pure Nb onto the wafer. We use an AZ5214E photoresist in the positive mode with a hard contact to define the waveguide in a Karl Suss MA6 mask aligner. After development, the sample undergoes etching using a Plasmalab 80Plus reactive ion etching system (Oxford Instruments). The plasma operates with a gas flow of SF_{6}/O_{2} at 40/20 s.c.c.m. with a radiofrequency power of 100 W.
We clean the resist residuals in acetone and isopropyl alcohol using ultrasonic excitation and then dry the chip with a nitrogen gun. Next, we use an atomic layer deposition method to grow a 45 nm dielectric layer of Al_{2}O_{3} in a Beneq TFS500 system. Subsequently, we protect the dielectric layer at the desired capacitor regions using the AZ5214E resist, and wet etch the rest of the atomiclayerdeposited oxide with an ammonium fluoride–hydrofluoric acid mixture. We then cleave the fourinch wafer into a 2 × 2 cm^{2} chip using Disco DADdy.
The nanowire is patterned using EPBG5000pES electronbeam lithography system employing a bilayer of methyl methacrylate/polymethyl methacrylate resist on a single chip. We deposit a 30nmthick AuPd layer in an electronbeam evaporator at a rate of 0.5 Å s^{−1}. After liftoff in acetone overnight, we galvanically pattern the superconducting leads connected to the nanowire by electronbeam lithography and deposit them with 100 nm of Al at a rate of 5 Å s^{−1}. Finally, we cleave each pixel (5 × 5 mm^{2}) using a laser micromachining system and package the chosen chip. We employ Al bonding wires to connect the chip to the printed circuit board of the sample holder. Extended Data Fig. 2c–e shows the micrographs of the bolometer chip.
Qubit fabrication
The qubit samples are fabricated using the following steps. First, a layer of 200 nm of Nb is sputtered onto a highresistivity silicon substrate. Next, we define the transmission line, readout resonator and transmon shunt capacitor using photolithography, followed by dry reactive ion etching.
The Al/Al_{2}O_{3}/Al Josephson junctions are subsequently fabricated using electronbeam lithography and the Dolan bridge method, where oxidation is used between the deposition of two Al layers to form the Josephson junctions. To ensure a galvanic contact between Al and Nb, the niobium oxide is removed by argon milling before the deposition of any Al.
The excess metal is lifted off in acetone. The roomtemperature resistance of the Al/Al_{2}O_{3}/Al junctions is measured to select a sample that will most probably yield the desired qubit frequency. Finally, the chip is diced into individual samples and the selected sample is wire bonded to a sample holder. Extended Data Fig. 2a,b shows the micrographs of the qubit chip.
Details of the experimental setup
Extended Data Fig. 1 shows a diagram of the full experimental setup. A PXI control computer initiates a measurement by sending a software trigger to a National Instruments NI5782 transceiver module connected to an NI PXIe7962R analoguetodigital converter running custom fieldprogrammable gate array code. The transceiver sends further digital trigger signals to initiate pulses to the readout resonator as well as excite the qubit. The clocks of all the devices are synchronized using a 10 MHz Rb reference.
The qubit excitation pulse is generated by an Active Technologies AT1212 digitaltoanalogue converter, upconverted to the gigahertz range and directed into the cryostat via a series of attenuators and filters. A microwave switch placed at the mixingchamber plate is used to select which of the six qubits on the chip is being driven. The qubit frequency is adjusted by applying a magnetic field to the qubit–resonator chip using a handwound coil outside the sample holder.
The readout pulse is generated by a microwave source using pulse modulation. The pulse is reflected off of the resonator on the qubit–resonator chip, after which it is directed through an isolator, a doublepole doublethrow switch, a directional coupler, a highpass filter with a 3 dB cutoff frequency at 3.1 GHz and a lowpass filter with a 3 dB cutoff frequency at 7.6 GHz, before reaching the bolometer absorber port. We have chosen the filters in an effort to place the passband around the readout frequency and filtering out possible leakage from the qubit drive tone. The qubit frequency, unfortunately, turned out to be close to the edge of the passband, but we did not observe leakage of the drive tone (Methods and Extended Data Fig. 6) and were able to perform the experiment. The other components between the qubit–resonator and bolometer chips are needed only to individually characterize the qubit–resonator chip (by toggling the doublepole doublethrow switch and measuring from the auxiliary readout port) and the bolometer (by applying a pulse from the auxiliary drive port through the directional coupler), and can be removed in future experiments.
The Thermocoax cable in the auxiliary drive line acts as a lowpass filter to prevent highfrequency radiation from leaking in; consequently, we needed to apply roughly 25 dBm higher power at f_{d} = 5.5 GHz during the initial bolometer characterization than when driving via the qubit chip.
To probe the bolometer, a continuous microwave tone is first attenuated by 30 dB for the available output power of the source to lie within a suitable range for our experiment. Subsequently, this tone is split into two paths, namely, a reference and a signal. The signal is directed into the cryostat, reflected off of the probe port of the bolometer, and amplified by an HEMT amplifier at 4 K and additional amplifiers at room temperature. Isolators are placed between the HEMT and bolometer to attenuate the thermal noise from the HEMT. Such noise may heat up the bolometer and thus degrade the SNR. However, it is possible to operate the bolometer without isolators on the probe output line, as demonstrated elsewhere^{26}, as well as in some of the measurements discussed below. Thermal noise from 4 K over the κ_{b} = 7.6 MHz linewidth of the bolometer has a power of k_{B}Tκ_{b} = −123.8 dBm, which is nearly an order of magnitude below the powers of the probe and qubit readout tones. Removing the isolators from the probe output line, thus, seems to be a feasible option.
At room temperature, filters and 3 dB attenuators are placed between the amplifiers to avoid standing waves and possible amplifier saturation. After amplification, the signal is demodulated to an intermediate frequency using a local oscillator, which is detuned from the probe tone by a fixed intermediate frequency of 70.3125 MHz. The reference is demodulated by the local oscillator without passing through the cryostat, and is used as an amplitude and phase reference for the signal. Both intermediatefrequency signals are amplified further and digitized by the NI5782 module at a sampling rate of 250 megasamples per second and digitally demodulated into inphase (I) and quadrature (Q) components. The NI5782 module also handles the ensemble averaging of the data, where applicable, as well as downsampling by boxcar averaging a variable number of adjacent points. From the timedomain data in the I–Q plane, we calculate the average signal before the pulse (\({\tilde{V}}_{0}\)) and the average signal during the pulse (\(\tilde{V}\)), as determined by t_{0} and t_{RO}. The data are then rotated in the I–Q plane such that \(\tilde{V}{\tilde{V}}_{0}\) lies on the I axis. We define \(\bar{V}\) and \({\bar{V}}_{0}\) (discussed in the main text) as the I components of \(\tilde{V}\) and \({\tilde{V}}_{0}\), respectively, after applying this rotation. For the singleshot experiments, we apply a common rotation across all the shots, with the rotation angle chosen such that the readout fidelity is maximized. Note that the phase of the signal in the I–Q plane is due to the reflection off of the effective LC circuit of the bolometer and is thus completely independent of the phase of the photons emitted by the readout resonator.
The data acquisition and storage is managed by the QCoDeS^{45} data acquisition framework. Data analysis and fitting are carried out using the NumPy^{46}, SciPy^{47}, xarray^{48} and lmfit^{49} libraries.
Sample characterization
Extended Data Figs. 3–5 show the characterization measurements of the bolometer. Extended Data Fig. 3a shows the signal coming out of the cryostat normalized by the reference, as discussed above. The data are averaged for 1 ms over 16 repetitions. For each pixel in Extended Data Fig. 3b, a 2 ms pulse is applied to the bolometer absorber port via the readout resonator with an offresonant tone of f_{d} = 5.4 GHz and P_{d} = −108 dBm. From these data, we extract the time constant τ_{b} (Extended Data Fig. 3c) by fitting the timedomain signal to an exponential model.
To characterize the bandwidth of the bolometer, we disconnect the qubit chip and directly connect the readout drive line to the bolometer absorber port without any filters (Extended Data Fig. 4). A circulator and a microwave switch are used in front of the absorber port to be able to measure the reflection coefficient of the absorber, as discussed below. We repeat the measurement shown in Fig. 2d with a readout pulse length of 2 ms and the probe frequency adjusted to 549 MHz. The bolometer resonance frequency shifts owing to the thermal cycling of the cryostat as well as the increased thermal noise from the HEMT since the isolators were removed from the probe output. We vary the readout power P_{d} and frequency f_{d}, and use these data to characterize the gain of the bolometer as follows: for a given readout drive power P_{d}, the detector signal is given by S = G(P_{d})P_{d}, where G is a gain factor characterizing the powertovoltage conversion of the bolometer and the total amplification of the amplification chain of the probe signal, including the HEMT and roomtemperature amplifiers. We describe the gain suppression of the bolometer using the following simple phenomenological model:
where P_{sat} is the saturation power and G_{0} is the linear gain factor at low readout powers, where it coincides with the quasistatic responsivity^{28}. Here \({\bar{V}}_{{{{\rm{noise}}}}}\) is the average noise level, calculated as the median of the data across f_{d} with P_{d} = 0, since with no power applied, the detector signal consists of noise only, independent of f_{d}. We calculate the noise floor as \({V}_{{{{\rm{NF}}}}}={\bar{V}}_{{{{\rm{noise}}}}}+3{\sigma }_{{f}_{{{{\rm{d}}}}}}(S)\), where \({\sigma }_{{f}_{{{{\rm{d}}}}}}(S)\) is the sample standard deviation of the data across f_{d} with P_{d} = 0. We define the 1 dB compression point as P_{1 dB} = (10^{1/10} − 1)P_{sat}, which is found by solving S(P_{1 dB}) = 10^{−1/10}G_{0}P_{1 dB} from equation (2) with \({\bar{V}}_{{{{\rm{noise}}}}}=0\). The minimumdetectable power P_{min} is given by V_{NF}/G_{0}, above which S from equation (2) is above the noise floor.
Extended Data Fig. 5a shows S measured with the qubit chip removed, along with the data for t_{RO} = 2 ms and f_{d} = 5.4 GHz from Fig. 2d for comparison. For the data shown in Fig. 2d, we calculate \({\bar{V}}_{{{{\rm{noise}}}}}\) and V_{NF} from the median and sample standard deviation, respectively, of S for the values where P_{d} ≤ −128 dBm, since no other data are available. We find excellent agreement between the data and the model of equation (2). Extended Data Fig. 5a also shows the extracted noise floor, P_{min} and P_{1 dB}.
The extracted G_{0}, P_{1 dB} and P_{min}, as well as the headroom (defined as P_{1 dB}/P_{min}), are shown in Extended Data Fig. 5b–d over a broad range of frequencies. Overall, there is around 10 dB of variation in P_{1 dB} and P_{min}, but the headroom is fairly flat and consistently above 11 dB. On the basis of this, the performance of the bolometer can be expected to be similar to that discussed in the main text across a broad range of frequencies, provided that the impedance of the absorber is matched (as discussed below). Note that the gain suppression of the bolometer does not arise from saturation, but rather because the resonance frequency of the bolometer shifts, which results in a smaller change in S away from resonance. This also means that P_{1 dB}, P_{min} and the headroom vary with f_{p}, with the headroom maximized near the resonance.
We further study the impedance matching of the absorber port using the setup shown in Extended Data Fig. 4. A microwave signal from a vector network analyser is sent to the absorber port and measured via a circulator. The obtained S_{21} transmission coefficient is normalized by a reference measurement with the microwave switch between the circulator and bolometer connected to a fully reflecting open line. The resulting absorber power reflection coefficient ∣Γ_{abs}(f_{d})∣^{2} is shown in Extended Data Fig. 5e. We find that ∣Γ_{abs}∣^{2} is independent of the drive power P_{d}, and ∣Γ_{abs}∣^{2} ≲ −3 dB ≈ 0.50 for f_{d} around 4.6–5.6 GHz and 7.2–8.6 GHz, and ∣Γ_{abs}∣^{2} ≲ −6 dB ≈ 0.25 for f_{d} around 4.7–5.4 GHz. We also observe 10MHzwide resonance peaks near 4.79 GHz with ∣Γ_{abs}∣^{2} < −22 dB ≈ 6 × 10^{−3} and at 5.28 GHz with ∣Γ_{abs}∣^{2} < −31 dB ≈ 8 × 10^{−4}. At f_{d} = f_{r} ≈ 5.47 GHz, we find ∣Γ_{abs}∣^{2} = −8.9 dB, which corresponds to a real impedance of Z_{abs} = Z_{0}(1 + Γ_{abs})/(1 − Γ_{abs}) = 106.2 Ω, where Z_{0} = 50.0 Ω is the impedance of the coaxial transmission line connected to the bolometer sample holder. We attribute the overall impedance mismatch to reflections from the sample holder, reflections from the bonding wires between the sample holder and chip, and the geometry of the absorber. This suggests that the SNR of the qubit readout can be improved at least by a factor of
by placing the qubit and bolometer on the same chip and optimizing the absorber geometry.
Extended Data Fig. 6 shows the characterization experiments of the qubit–resonator system using the bolometer. Extended Data Fig. 6a,b shows the singletone and twotone spectroscopy, respectively, from which we extract an initial estimate for the resonator and qubit frequencies. Note that in contrast to dispersive readout with voltage amplifiers where the tones may be continuous, we pulse the readout tone in all these experiments. Subsequently, we calibrate the qubit frequency more accurately, as well as the duration of the π pulse (Fig. 2e).
With the π pulse calibrated, we are able to observe the effect of the dispersive shift by measuring the resonator spectrum after preparing the qubit in g〉 and e〉 (Extended Data Fig. 6c). Note that the difference between the locations of the minima in Extended Data Fig. 6c is somewhat smaller than the calculated dispersive shift of χ/2π, since the spectrum corresponding to state e〉 is distorted owing to the qubit decaying during readout.
In Fig. 6d, a π pulse is applied to the qubit, and the readout pulse is applied after different delays to extract the T_{1} decay time. Extended Data Fig. 6e shows the result of a Ramsey experiment, where the qubit is driven by two π/2 pulses with a varying idle time between them, for different modulation frequencies of the pulse. These data are used to extract the T_{2} time, given by the exponentially decaying envelope of the oscillation. The signal level varies in Extended Data Fig. 6, since different values of P_{d} and t_{RO} were used during the characterization and the optimal values for these parameters were not yet known. In addition, the cryostat was thermally cycled between the acquisition of the data shown in Extended Data Fig. 6a–c and those shown in Extended Data Fig. 6d,e. This is why the qubit frequency has decreased in Extended Data Fig. 6e compared with those in Extended Data Fig. 6b and Fig. 2e.
Detector signal probability distribution with long averaging time
Following another work^{38}, we develop a model for the probability distribution of the detector signal for the qubit nominally prepared in e〉 (with some preparation error) and for the averaging time being comparable with the energy decay time T_{1}. Let v_{g/e}(t, t_{d}) be the timedomain signal of the bolometer output when the qubit is in state g〉 or e〉 at the beginning of the readout pulse, that is, t = 0. Here t_{d} denotes the time at which the qubit instantaneously decays to g〉 if it was initially in e〉 (v_{g} is independent of t_{d}). We assume that no thermal excitation events occur, and that the noise in the signal is Gaussian. Formally, the signal is given by
where u_{g/e}(t, t_{d}) is the evolution of the output in the absence of noise, P_{N} is the noise power spectral density and dW(t) is the Wiener increment^{50}.
To incorporate the bolometer dynamics, we model u_{g/e}(t, t_{d}) as exponentially approaching some steadystate values c_{g} and c_{e}, for the qubit in state g〉 or e〉, respectively^{28}. Namely, we assume that the time evolution is of the form
and
where θ(x) is the step function and \({c}_{{{{\rm{e}}}}}^{{\prime} }={c}_{{{{\rm{e}}}}}(1{{{{\rm{e}}}}}^{{t}_{{{{\rm{d}}}}}/{\tau }_{{{{\rm{b}}}}}})\).
Averaging the bolometer output from t_{0} to t_{RO} yields the averaged detector signal as
where \({\bar{U}}_{{{{\rm{g}}}}/{{{\rm{e}}}}}({t}_{{{{\rm{d}}}}})\equiv {({t}_{{{{\rm{RO}}}}}{t}_{0})}^{1}\int\nolimits_{{t}_{0}}^{{t}_{{{{\rm{RO}}}}}}{u}_{{{{\rm{g}}}}/{{{\rm{e}}}}}(t,{t}_{{{{\rm{d}}}}})\,{{{\rm{d}}}}t\) and X[0, σ^{2}] is a normally distributed random variable with mean 0 and standard deviation \(\sigma =\sqrt{{P}_{{{{\rm{N}}}}}/({t}_{{{{\rm{RO}}}}}{t}_{0})}\). Above, we have neglected the constant \({\bar{V}}_{0}\), which simply shifts \({\bar{V}}_{{{{\rm{g}}}}/{{{\rm{e}}}}}\). With a fixed t_{d}, the probability of obtaining a given detector voltage V, thus, obeys the following probability distribution:
which is a Gaussian centred around \({\bar{U}}_{{{{\rm{g}}}}/{{{\rm{e}}}}}({t}_{{{{\rm{d}}}}})\) having standard deviation σ as defined above.
If the qubit is in state g〉 at t = 0, the probability distribution P_{g}(V) is simply a Gaussian centred around the constant value \({\bar{U}}_{{{{\rm{g}}}}}\). When the qubit is initially in e〉, the probability distribution is obtained by calculating the average of P_{e}(V∣t_{d}) for all the possible realizations of the qubit decay time t_{d} weighted by the probability density P(t_{d}) of the decay occurring at t_{d}:
where P(t_{d}) is exponentially distributed: \(P({t}_{{{{\rm{d}}}}})={{{{\rm{e}}}}}^{{t}_{{{{\rm{d}}}}}/{T}_{1}}/{T}_{1}\). Thus, the total probability distribution for the qubit nominally prepared in e〉 is the following weighted sum of these distributions:
where P_{x} is the probability that the qubit was actually in state g〉 at t = 0. Here P_{x} includes state preparation errors, as well as the T_{1} decay of the qubit during the short delay between the application of the π pulse and the start of the readout pulse.
In the case where u_{g/e}(t, t_{d}) is like step functions, equations (9) and (10) have analytical expressions^{38}. However, we do not obtain an analytical expression for P_{e}(V) given the temporal evolution arising from equation (6), and hence, we numerically calculate the integral of equation (9). Scaling this by the bin width used in Fig. 3b produces the prediction for the distribution, with c_{g}, c_{e}, T_{1}, σ and P_{x} as the fitting parameters. For the data shown in Fig. 3b, using the separately determined value of τ_{b} = 9.4 μs, we find c_{g} = 24.7 mV, c_{e} = 182.0 mV, T_{1} = 25.8 μs, σ = 17.4 mV and P_{x} = 0.20.
The readout fidelity is given by
where V_{th} is the threshold value for assigning the measurement outcome to g〉 or e〉, and we again assume that no thermal excitations occur.
If we further assume that P_{x} = 0 and that the qubit never relaxes, \({P}_{{{{\rm{e}}}}}^{{{{\rm{tot}}}}}(V\,)\) reduces to a Gaussian distribution centred around \({\bar{U}}_{{{{\rm{e}}}}}^{\infty }={\bar{U}}_{{{{\rm{e}}}}}({t}_{{{{\rm{d}}}}}\to \infty )\), which can be calculated by dropping the second term in equation (6). In this case, the readout fidelity is maximized with \({V}_{{{{\rm{th}}}}}=({\bar{U}}_{{{{\rm{e}}}}}^{\infty }+{\bar{U}}_{{{{\rm{g}}}}})/2\), which yields
where erf(x) is the error function. Using the values for c_{g}, c_{e}, τ_{b} and σ extracted from the fitting, and fixing t_{RO} = 13.9 μs, we calculate \({\bar{U}}_{{{{\rm{g}}}}}\) and \({\bar{U}}_{{{{\rm{e}}}}}^{\infty }\) as functions of t_{0}. After inserting these into equation (12), we maximize F^{∞} with respect to t_{0} and find F^{∞} ≈ 0.927 with t_{0} = 3.19 μs.
Estimate of parameters needed for 99.9% fidelity
Here we quantitatively discuss how to achieve 99.9% fidelity with bolometric readout following improvements to the experiment proposed in the main text. To be able to make a fair comparison between the performance of different bolometers presented in the literature^{28,29}, we carry out this analysis by assuming that we operate in the calorimetric mode. Namely, we interpret the distributions shown in Fig. 3b to represent a measurement of the energy of the absorbed pulses. In such an experiment, the signal is directly proportional to the energy packet arriving at the bolometer, whereas the width of the distribution is independent of the energy.
Let us first assume that the length of the readout pulse is reduced to 200 ns without changing the readout power. This means that the bolometerabsorbed energy, and hence the SNR (Fig. 3b), is reduced by a factor of 70. By changing the metallic bolometer to a graphene bolometer, the energy resolution has been observed to increase by a factor of 13 (ref. ^{29}). Thus, to obtain 99.9% fidelity, one needs to increase the amount of energy absorbed by the graphene bolometer in 200 ns by a factor of A_{99.9%} ≈ 1.8 × 70/13 ≈ 10, where the factor of 1.8 arises from the fact that to reduce the overlap infidelity of the data shown in Fig. 3b from 7.0% to 0.1%, one needs a factor of 1.8 improvement in the SNR in equation (12).
Below, we aim to show that by making different improvements to the measurement scheme, we can arrive at several factors A_{α} of increment to the energy absorbed by the bolometer; taking all these factors into account, it is possible to exceed the required SNR, that is, A_{99.9%} = 10 < ∏_{α}A_{α}. Since we have assumed a 200 ns readout time, having a qubit T_{1} of the order of 100 μs is sufficient to reach 99.9% readout fidelity. Note that the bolometer is a power sensor, and thus, the SNR is directly proportional to the absorbed power for the lowenough powers considered here^{29} (Fig. 2d).
First, switching to a transmissiontype setup, where all the readout photons are directed into the bolometer instead of half of them escaping through the input, increases the power incident on the bolometer by a factor of A_{t} = 2. Second, removing additional components between the qubit and bolometer chips reduces the losses by approximately A_{c} ≈ 1 dB ≈ 1.250, and improving the impedance matching of the absorber increases the SNR by a factor of A_{Z} = 1.148 according to equation (3). Third, by optimizing the dispersive shift χ for photodetectiontype readout, we expect that the photon number occupying the resonator during readout, and thus the incident power, can be increased by a factor of A_{χ} = 2 (ref. ^{32}). Fourth, switching the qubit–resonator system to a configuration where the resonance frequency of the resonator lies above the qubit frequency (keeping the detuning ∣f_{r,g} − f_{q}∣ constant) allows driving with a larger number of photons^{36}, which we estimate to be at least A_{a} = 1.5 times greater than with our current parameters. Fifth, increasing the admissible drive power by doubling the resonator drive frequency to 10 GHz may further double the photon energy hf_{r,g}, introducing a factor of A_{2f} = 2. In total, the improvement in SNR is given by
and thus, a high fidelity is feasibly achievable.
Multiplexed qubit readout using bolometers
Let us discuss the applicability of bolometerbased qubit readout for quantum processors consisting of multiple qubits. Although a bolometer cannot natively distinguish the frequency of incident photons, frequencymultiplexed qubit readout is feasible by coupling individual bolometers to the readout resonators of each qubit (Extended Data Fig. 7a). Here it is possible to multiplex the bolometer probe signals in the frequency domain using a shared probe line. Furthermore, this way of multiplexing provides relief to the available headroom in the total probe power since probing at the resonance frequency of one bolometer does not saturate the other bolometers. In threedimensional integrated quantum processors^{40}, these individual bolometers may be placed on the wiring layer below the qubits owing to the small footprint of the bolometers.
To take the first steps in demonstrating the feasibility of using bolometers for multiplexed qubit readout, we have fabricated a chip with three bolometers (Extended Data Fig. 7b). Each bolometer has a separate readout input port, which may be connected to qubit readout resonators. The design of bolometers is similar to the sample discussed in the main text, apart from the addition of meanderline inductors in series with the nanowire to control the designed resonance frequencies of the bolometer LC circuits. Extended Data Fig. 7c–h shows the preliminary characterization results from this threebolometer device. We find three distinct resonances (Extended Data Fig. 7c–e), which shift down in frequency with increasing probe power, indicating that they correspond to bolometers that heat up due to electrothermal feedback. We are able to clearly observe microwave signals between 5.0 and 7.6 GHz sent to the absorbers of each bolometer through separate input lines without major crosstalk (Extended Data Fig. 7f–h). These preliminary results on multiplexed bolometeric readout pave the way for future experiments where multiple qubits are read by their dedicated bolometer.
Data availability
The data that support the findings of this study are available via Zenodo at https://doi.org/10.5281/zenodo.7773980 (ref. ^{51}).
References
DiVincenzo, D. P. The physical implementation of quantum computation. Fortschr. Phys. 48, 771–783 (2000).
Nielsen, M. A. & Chuang, I. L. Quantum Computation and Quantum Information: 10th Anniversary Edition (Cambridge Univ. Press, 2011).
Shor, P. W. Scheme for reducing decoherence in quantum computer memory. Phys. Rev. A 52, R2493–R2496 (1995).
Preskill, J. Faulttolerant quantum computation. Preprint at https://arxiv.org/abs/quantph/9712048 (1997).
Terhal, B. M. Quantum error correction for quantum memories. Rev. Mod. Phys. 87, 307–346 (2015).
Blais, A., Huang, R.S., Wallraff, A., Girvin, S. M. & Schoelkopf, R. J. Cavity quantum electrodynamics for superconducting electrical circuits: an architecture for quantum computation. Phys. Rev. A 69, 062320 (2004).
Barends, R. et al. Logic gates at the surface code threshold: superconducting qubits poised for faulttolerant quantum computing. Nature 508, 500–503 (2014).
Kjaergaard, M. et al. Superconducting qubits: current state of play. Annu. Rev. Condens. Matter Phys. 11, 369–395 (2020).
Zhao, Y. et al. Realization of an errorcorrecting surface code with superconducting qubits. Phys. Rev. Lett. 129, 030501 (2022).
Krinner, S. et al. Realizing repeated quantum error correction in a distancethree surface code. Nature 605, 669–674 (2022).
Acharya, R. et al. Suppressing quantum errors by scaling a surface code logical qubit. Nature 614, 676–681 (2023).
Koch, J. et al. Chargeinsensitive qubit design derived from the Cooper pair box. Phys. Rev. A 76, 042319 (2007).
Blais, A., Grimsmo, A. L., Girvin, S. M. & Wallraff, A. Circuit quantum electrodynamics. Rev. Mod. Phys. 93, 025005 (2021).
Walter, T. et al. Rapid highfidelity singleshot dispersive readout of superconducting qubits. Phys. Rev. Appl. 7, 054020 (2017).
Sunada, Y. et al. Fast readout and reset of a superconducting qubit coupled to a resonator with an intrinsic Purcell filter. Phys. Rev. Appl. 17, 044016 (2022).
Heinsoo, J. et al. Rapid highfidelity multiplexed readout of superconducting qubits. Phys. Rev. Appl. 10, 034040 (2018).
Aumentado, J. Superconducting parametric amplifiers: the state of the art in Josephson parametric amplifiers. IEEE Microw. Magazine 21, 45–59 (2020).
Esposito, M., Ranadive, A., Planat, L. & Roch, N. Perspective on traveling wave microwave parametric amplifiers. Appl. Phys. Lett. 119, 120501 (2021).
Clerk, A. A., Devoret, M. H., Girvin, S. M., Marquardt, F. & Schoelkopf, R. J. Introduction to quantum noise, measurement, and amplification. Rev. Mod. Phys. 82, 1155–1208 (2010).
Bergeal, N. et al. Phasepreserving amplification near the quantum limit with a Josephson ring modulator. Nature 465, 64–68 (2010).
Macklin, C. et al. A near–quantumlimited Josephson travelingwave parametric amplifier. Science 350, 307–310 (2015).
Opremcak, A. et al. Highfidelity measurement of a superconducting qubit using an onchip microwave photon counter. Phys. Rev. X 11, 011027 (2021).
Chen, L. et al. Transmon qubit readout fidelity at the threshold for quantum error correction without a quantumlimited amplifier. NPJ Quantum Inf. 9, 26 (2023).
Lecocq, F. et al. Control and readout of a superconducting qubit using a photonic link. Nature 591, 575–579 (2021).
Delaney, R. D. et al. Superconductingqubit readout via lowbackaction electrooptic transduction. Nature 606, 489–493 (2022).
Govenius, J. et al. Microwave nanobolometer based on proximity Josephson junctions. Phys. Rev. B 90, 064505 (2014).
Govenius, J., Lake, R. E., Tan, K. Y. & Möttönen, M. Detection of zeptojoule microwave pulses using electrothermal feedback in proximityinduced Josephson junctions. Phys. Rev. Lett. 117, 030802 (2016).
Kokkoniemi, R. et al. Nanobolometer with ultralow noise equivalent power. Commun. Phys. 2, 124 (2019).
Kokkoniemi, R. et al. Bolometer operating at the threshold for circuit quantum electrodynamics. Nature 586, 47–51 (2020).
Girard, J.P. et al. Cryogenic sensor enabling broadband and traceable power measurements. Rev. Sci. Instrum. 94, 054710 (2023).
Barends, R. et al. Coherent Josephson qubit suitable for scalable quantum integrated circuits. Phys. Rev. Lett. 111, 080502 (2013).
Nesterov, K. N., Pechenezhskiy, I. V. & Vavilov, M. G. Counting statistics of microwave photons in circuit QED. Phys. Rev. A 101, 052321 (2020).
Gambetta, J. et al. Quantum trajectory approach to circuit QED: quantum jumps and the Zeno effect. Phys. Rev. A 77, 012112 (2008).
Chen, Z. Metrology of Quantum Control and Measurement in Superconducting Qubits. PhD thesis, Univ. California Santa Barbara (2018).
Khezri, M. et al. Measurementinduced state transitions in a superconducting qubit: within the rotating wave approximation. Phys. Rev. Appl. 20, 054008 (2023).
Cohen, J., Petrescu, A., Shillito, R. & Blais, A. Reminiscence of classical chaos in driven transmons. PRX Quantum 4, 020312 (2023).
Scott, D. W. On optimal and databased histograms. Biometrika 66, 605–610 (1979).
Gambetta, J., Braff, W. A., Wallraff, A., Girvin, S. M. & Schoelkopf, R. J. Protocols for optimal readout of qubits using a continuous quantum nondemolition measurement. Phys. Rev. A 76, 012325 (2007).
McKitterick, C. B., Prober, D. E. & Karasik, B. S. Performance of graphene thermal photon detectors. J. Appl. Phys. 113, 044512 (2013).
Kosen, S. et al. Building blocks of a flipchip integrated superconducting quantum processor. Quantum Sci. Technol. 7, 035018 (2022).
Touzard, S. et al. Gated conditional displacement readout of superconducting qubits. Phys. Rev. Lett. 122, 080502 (2019).
Ikonen, J. et al. Qubit measurement by multichannel driving. Phys. Rev. Lett. 122, 080503 (2019).
Lienhard, B. et al. Deepneuralnetwork discrimination of multiplexed superconductingqubit states. Phys. Rev. Appl. 17, 014024 (2022).
Siddiqi, I. et al. RFdriven Josephson bifurcation amplifier for quantum measurement. Phys. Rev. Lett. 93, 207002 (2004).
Nielsen, J. H. et al. QCoDeS/Qcodes: 0.34.0  June 2022 (20220613). Zenodo https://doi.org/10.5281/zenodo.6637581 (2022).
Harris, C. R. et al. Array programming with NumPy. Nature 585, 357–362 (2020).
Virtanen, P. et al. SciPy 1.0: fundamental algorithms for scientific computing in Python. Nat. Methods 17, 261–272 (2020).
Hoyer, S. & Joseph, H. xarray: ND labeled arrays and datasets in Python. J. Open Res. Software 5, 10 (2017).
Newville, M. et al. lmfit/lmfitpy: 1.1.0. Zenodo https://doi.org/10.5281/zenodo.3588521 (2019).
Jacobs, K. Stochastic Processes for Physicists: Understanding Noisy Systems (Cambridge Univ. Press, 2010).
Gunyho, A. M. et al. Data for “Singleshot readout of a superconducting qubit using a thermal detector”. Zenodo https://doi.org/10.5281/zenodo.10000511 (2023).
Acknowledgements
We acknowledge funding from the Academy of Finland Centre of Excellence program project nos. 352925 (A.M.G. and S.K.), 336810 (A.M.G., V. Vadimov and S.K.), 336817 (M.M. and V. Vesterinen) and 336819 (M.M. and V. Vesterinen); European Research Council under Advanced Grant no. 101053801 (ConceptQ) (M.M. and V. Vadimov); Horizon Europe programme HORIZONCL42022QUANTUM01SGA via the project 101113946 (OpenSuperQPlus100) (M.M. and Q.C.); the Future Makers Program of the Jane and Aatos Erkko Foundation and the Technology Industries of Finland Centennial Foundation (A.M.G., Q.C. and P.S.); Business Finland under the Quantum Technologies Industrial (QuTI) project (decision no. 41419/31/2020) (S.N., J.M. and M.M.); and the Finnish Foundation for Technology Promotion project no. 8640 (G.C.). We thank A. Alizadeh, S. Simbierowicz and R. Lake for useful discussions, and L. Grönberg for niobium deposition in the fabrication.
Funding
Open Access funding provided by Aalto University.
Author information
Authors and Affiliations
Contributions
A.M.G. and S.N. conducted the experiments and analysed the data. W.L. designed and fabricated the bolometer chip. The qubit chip was designed by S.K. and fabricated by J.M. with help from V. Vesterinen. G.C., P.S. and Q.C. assisted with characterizing and operating the bolometer. A.M.G. and V. Vadimov developed the model for the signal probability distribution. P.S. designed, fabricated and characterized the multiplexed bolometer chip. The paper was written by A.M.G. and M.M., with comments from all authors. The work was conceived and supervised by M.M.
Corresponding author
Ethics declarations
Competing interests
M.M. declares that that he is a cofounder and shareholder of the quantum computer company IQM Finland Oy. M.M. also declares that he is an inventor in granted patents FI122887B (Finland), US9255839B2 (US), JP5973445B2 (Japan) and EP2619813B1 (EU) titled ‘Detector of single microwave photons propagating in a guide’, applied by Aaltokorkeakoulusäätiö, and invented by M.M. and J. Pekola. This patent family describes an ultrasensitive microwave detector concept. M.M. further declares that he is an inventor in pending patent applications WO2022248759A1 and TW202303466A titled ‘Quantumstate readout arrangement and method’, applied by IQM Finland Oy, and invented by M.M. and J. Hassel. This patent family describes a concept of measuring the states of qubits using bolometers. The other authors declare no competing interests.
Peer review
Peer review information
Nature Electronics thanks David Pahl and Adam Sirois for their contribution to the peer review of this work.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Extended data
Extended Data Fig. 1 Full experimental setup.
The labels for the filters, directional couplers, amplifiers, and the power splitter refer to minicircuits model numbers. The bandpass filters with two labels denote a lowpass filter and highpass filter in series.
Extended Data Fig. 2 Chip design.
(a) Falsecolor optical micrograph of the qubit chip composited from two images. The chip contains six qubits with individual drive lines and readout resonators, with a shared readout line (highlighted in green). (b) Enlarged view of one of the qubits, showing the readout resonator (blue), the drive line (yellow), and the crossshaped qubit capacitor (yellow). The Josephson junctions are near the bottom of the capacitor. (c) Falsecolor optical micrograph of the bolometer chip, to scale with (a). The readout line from the qubit chip is connected to the transmission line (green) leading to the absorber of the bolometer. (d) Area enclosed in the red rectangle in (c) showing the Al grounding pad, bolometer shunt capacitor C_{1}, and the probe gate capacitor C_{g}. The circular artifacts originate from dust particles on the microscope lens. (e) Scanningelectron micrograph of the AuPd nanowire and Al islands forming the superconductor–normalmetal–superconductor (SNS) junctions of a bolometer similar to that used in this paper.
Extended Data Fig. 3 Bolometer characterization.
(a) Total transmission power divided by the reference power, (b) detector signal S, and (c) extracted thermal time constant τ_{b} of the bolometer as functions of the probe frequency f_{p} and power P_{p}. The red background approximately indicates the power region where the bolometer exhibits bistability.
Extended Data Fig. 4 Experimental setup for gain compression measurements.
The roomtemperature components connected to the probe input and output are as in Extended Data Fig. 1. To characterize the reflection coefficient at the absorber port of the bolometer, we connect an Agilent N9928A vector network analyzer to the readout input and output ports, and for the gain compression measurement we use the same microwave source for the readout input as in Extended Data Fig. 1, without the roomtemperature filters. See Extended Data Fig. 1 for the definition of the symbols.
Extended Data Fig. 5 Gain compression and bandwidth.
(a) Bolometer detector signal S without (blue crosses) and with (red circles) the qubit chip present, as functions of the readout drive power P_{d}, with f_{d} = 5.4 GHz, a 2 ms readout pulse and 16 ensemble averages. The solid lines are fits to a phenomenological model. The horizontal dotted lines indicate the calculated noise floor. The diagonal dashed lines show the signal gain without compression, given by S = G_{0}P_{d}. The vertical dashed and dotted lines indicate the 1 dB compression point and minimum detectable power, respectively. The compression points for the two datasets are 0.04 dBm apart from each other at −113.8 dBm. The data with the qubit chip present is from the same data set as the data of Fig. 2(d) with t_{RO} = 2 ms and f_{d} = 5.4 GHz. (b) Lowpower gain factor G_{0}, (c) 1 dB gain compression point (blue) and minimum detectable power (red), (d) headroom defined as the ratio of the gain compression point and the minimum detectable power, and (e) power reflection coefficient ∣Γ_{abs}∣^{2} of the bolometer absorber port, as functions of the readout frequency f_{d}, with no qubit connected to the bolometer. The vertical dotted line indicates the frequency of the readout resonator at 5.47 GHz. The horizontal dashed line in (e) marks 3 dB reflection.
Extended Data Fig. 6 Qubit characterization using bolometer.
(a) Detector signal S as a function of the bias voltage applied to the flux bias coil and of the frequency of the resonator drive pulse. The resonator frequency, indicated by the minimum detector signal at a given flux value, clearly changes as a function of the flux bias. The data of panel (a) are averaged over 4 repetitions. (b) Detector signal S as a function of the flux bias voltage and the frequency of a continuous tone applied to the qubit. The data is averaged over 32 repetitions. The data of panels (a) and (b) are acquired using a readout power of P_{d} = − 120.8 dBm and a pulse duration t_{RO} = 1 ms. (c) Detector signal S as a function of the readout drive frequency f_{d}, with the qubit prepared in the \(\left\vert {{{\rm{g}}}}\right\rangle\) (blue) and \(\left\vert {{{\rm{e}}}}\right\rangle\) (red) states. The resonance shifts if the qubit state changes. The vertical dotted line indicates the frequency used for the qubit readout. Here, we have used P_{d} = − 109.8 dBm, t_{RO} = 10 μs, and averaged over 16384 repetitions. (d) Detector signal S as a function of the idling time between a π pulse applied to the qubit and the readout pulse, providing the T_{1} = 28 μs of the qubit. In these measurements, P_{d} = − 112.8 dBm and t_{RO} = 20 μs, and the data are averaged over 4096 repetitions. (e) Detector signal as a function of the idle time between the π/2 pulses and the qubit drive frequency in a Ramsey experiment which is used to extract the T_{2} of the qubit. These data are acquired with P_{d} = − 107.8 dBm and t_{RO} = 25 μs, averaged over 1024 repetitions.
Extended Data Fig. 7 Multiplexed qubit readout using bolometers.
(a) Illustration of the proposed scheme for multiplexed qubit readout using bolometers. Readout resonators with resonance frequencies f_{r,i} in the gigahertz range are driven via a shared feedline and coupled to individual bolometers with resonance frequencies f_{b,i} in the subgigahertz range. The bolometers are probed with separate probe frequencies f_{p,i} ≈ f_{b,i} via a shared probe line. (b) Falsecolor optical micrograph of a chip with three bolometers sharing a probe line (orange). The meandering inductors (green) are used to tune the frequency of the LC circuits. (c)–(e) Normalized transmission amplitudes near the resonance frequencies of the three bolometers as functions of the probe frequency and probe power. (f)–(h) Measured detector signals as functions of time with 1 ms pulses applied to the absorber ports of the different bolometers at 1 ms. Each bolometer is driven via input passband filters (not shown) at their absorber port by one pulse at a time with frequency 5.0 GHz (blue), 5.8 GHz (green), or 7.6 GHz (red). For each bolometer, the signal is large only for one of the drive frequencies, indicating that the crosstalk is small. The signals corresponding to the stopband drive frequencies are offset by − 1 and −2 mV for clarity. The red cross in each of the panels (b)–(d) mark the probe parameters that were used to acquire the data in each of the panels (e)–(g), respectively.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Gunyhó, A.M., Kundu, S., Ma, J. et al. Singleshot readout of a superconducting qubit using a thermal detector. Nat Electron 7, 288–298 (2024). https://doi.org/10.1038/s41928024011477
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1038/s41928024011477
This article is cited by

Measuring qubits with thermometers
Nature Electronics (2024)