Observation of quantum jumps in a superconducting artificial atom

R. Vijay∗ , D. H. Slichter∗ , and I. Siddiqi
Quantum Nanoelectronics Laboratory, Department of Physics, University of California, Berkeley CA 94720 (Dated: September 16, 2010)

A continuously monitored quantum system prepared in an excited state will decay to its ground state with an abrupt jump. The jump occurs stochastically on a characteristic time scale T1 , the lifetime of the excited state. These quantum jumps, originally envisioned by Bohr, have been observed in trapped atoms and ions1–3 , single molecules4 , photons5 , and single electrons in cyclotrons6 . Here we report the first observation of quantum jumps in a macroscopic quantum system, in our case a superconducting “artificial atom” or quantum bit (qubit)7 coupled to a superconducting microwave cavity8 . We use a fast, ultralow-noise parametric amplifier9 to amplify the microwave photons used to probe the qubit state, enabling continuous high-fidelity monitoring of the qubit. This technique represents a major step forward for solid state quantum information processing, potentially enabling quantum error correction and feedback10 , which are essential for building a quantum computer. Our technology can also be readily integrated into hybrid circuits involving molecular magnets, nitrogen vacancies in diamond, or semiconductor quantum dots. Quantum error correction, an essential component of a practical quantum computer, requires real-time, highfidelity readout to accurately determine the quantum state of the system and to allow rapid feedback10 . The readout must also be quantum non-demolition (QND), that is, it must leave the system in an eigenstate of the measured observable11 , thus allowing repeated measurements. To fulfill these requirements, we use the circuit quantum electrodynamics (cQED) architecture, where the superconducting qubit is dispersively coupled to a superconducting cavity12 , in analogy to an atom in a Fabry-Perot cavity. The cavity decouples the qubit from its environment, improving coherence times8,13 while also enabling a continuous, high visibility, QND measurement of the qubit state by probing the resonant frequency of the cavity14 . These properties make the cQED architecture an ideal system for observing quantum jumps15 . Despite successfully demonstrating QND measurement with several kinds of superconducting qubits8,13,16 , cQED implementations with linear cavities have typically suffered from low single-shot fidelity, precluding the observation of quantum jumps. This is primarily due to inefficient amplification of the photons leaving the cavity. State of the art cryogenic semiconductor microwave amplifiers add about 30 photons of noise, necessitating many averages to see the few-photon readout signal14 . Using

more readout photons induces qubit state mixing17 , thus limiting the fidelity. Other high fidelity readout schemes implemented for superconducting qubits are either too slow18 or scramble the qubit state19 . Josephson parametric amplifiers20,21 with near quantum limited noise performance can potentially enable single shot readout in the cQED architecture, but most existing designs have an instantaneous bandwidth below 1 MHz, too small to enable real time monitoring of the qubit state. Since superconducting qubit lifetimes are typically around 1 µs, one would need a bandwidth of order 10 MHz to resolve quantum jumps between qubit states with high fidelity. We achieve this by using a low quality factor (Q) nonlinear resonator as a parametric amplifier9 . Our experimental setup, shown schematically in Fig. 1, is anchored to the mixing chamber of a dilution refrigerator at 30 mK. The superconducting readout cavity is implemented as a quasi-lumped element linear resonator consisting of a meander inductor (orange) in parallel with an interdigitated capacitor (blue). A transmon qubit (yellow)22 is capacitively coupled to the cavity. This arrangement is different than typical cQED setups which use transmission line resonators for the cavity. Our design has a smaller footprint and avoids the detrimental higher cavity modes found in transmission line resonators13 . Further details of the qubit and cavity are presented in the Methods section. Probe photons enter from the input port and reflect off the readout cavity, acquiring a phase shift that depends on the qubit state. These photons then travel through a series of circulators, which allow microwave signals to propagate in one direction as indicated by the arrows in the figure, and reflect off a low Q non-linear resonator. When biased near its critical point23 by a strong pump tone from the drive port, this resonator amplifies the readout signal and sends it to the output port. This amplifier design simultaneously achieves high gain, low noise and large bandwidth9 . We operate with the qubit frequency set at 4.753 GHz, corresponding to a detuning ∆ = 2π × 1.170 GHz from the bare readout cavity frequency of 5.923 GHz. With a coupling strength g = 2π × 109 ± 0.5 MHz, the dispersive shift of the cavity12 due to the qubit state χ = g 2 /∆ = 2π × 10.15 MHz is considerably larger than the cavity linewidth κ = 2π × 4.9 MHz. Therefore, measurement photons at 5.932 GHz exiting the readout cavity will have a relative phase shift of 0 or 180 degrees, depending on the state of the qubit. When these photons arrive at the amplifier resonator, they coherently add to or subtract from the pump (also at 5.932 GHz, and tuned to be in phase with the readout photons), causing

arXiv:1009.2969v1 [cond-mat.mes-hall] 15 Sep 2010

2
b

a
Output

Amplifier resonator

Input

Drive

c

Readout cavity + qubit

FIG. 1: Experimental setup. In (a), readout photons (black arrow) enter from the input port and are directed through a microwave circulator to a 180◦ hybrid, which converts the single-ended microwave signal into a differential one. The photons interact with the readout cavity and the reflected signal (purple arrows) carries information about the qubit state toward the non-linear resonator through three circulators, which isolate the readout and qubit from the strong pump of the amplifier resonator. A directional coupler combines this signal with pump photons (green arrow) from the drive port. The pumped nonlinear resonator amplifies the readout signal, and the amplified signal (red arrows) is reflected and sent through the third circulator to the output port. The signal is further amplified by cryogenic and room temperature amplifiers before being mixed down to zero frequency, digitized, and stored in a computer. Qubit manipulation pulses enter on the same line as the readout pulses. (b) false-color optical image of the readout resonator, formed by a meander inductor (orange) shunted by an interdigitated capacitor (blue). The transmon qubit (yellow) is capacitively coupled to the resonator. The detail view shows the qubit loop. Scale bars are 100 µm (main view) and 10 µm (detail view). (c) false-color optical image of the amplifier resonator showing the superconducting loop (pink) and junctions (black), shunted by large parallel-plate capacitors (cyan) on a SiNx -coated Nb ground plane (brown). Scale bars are 100 µm (main view) and 5 µm (detail view).

a phase shift of up to 180 degrees in the reflected pump photons which form the output signal. This can be seen in Fig. 2a, where the average phase of the output signal is plotted as a function of pump power. The two traces correspond to measurements taken with the qubit prepared in the ground (blue) and excited (red) states, with an average readout cavity population n of one photon ¯ (calibrated using the ac Stark effect24 ). By choosing an appropriate pump power bias point, the two qubit states are faithfully mapped to distinct phases of the reflected pump. Since there are many more pump photons than signal photons, this effectively amplifies the input signal (See Supplementary Information for more details). The improvement in signal to noise is evident in Figs. 2b and 2c, where we plot Rabi oscillations of the qubit both without and with parametric amplification. With this technique, we can perform single-shot measurements of the qubit state and observe quantum jumps. We prepare the qubit state with a 20 ns pulse of varying amplitude at the qubit frequency of 4.753 GHz and immediately probe the cavity with a readout power corresponding to n = 8 photons, a level which optimizes read¯ out fidelity while keeping the measurement QND. The amplified signal is then mixed down to zero frequency, effectively converting the phase shift signal of the readout into a single-quadrature voltage signal. This voltage

+90

o

a Digitizer voltage (mV)

14

b

Pump phase

0

o

0

0

Rabi pulse width (ns)

400 c

300 0

-90 -94.2

o

-94.0 -93.8 -93.6 Pump power (dBm)

-300

FIG. 2: Amplifier resonator. In (a), we plot the measured average phase of the reflected pump as a function of pump power with the qubit prepared in the ground (blue) or excited (red) states and an average readout cavity occupation of one photon. By choosing a suitable pump power, the two qubit states are mapped to two different phases of the output signal. Panels (b) and (c) show Rabi oscillations of the qubit with 104 averages for each time point, both without (b) and with (c) the amplifier resonator energized. Note both the improved signal-to-noise ratio and the increased magnitude of the signal (on the left axis) when using the amplifier.

is then digitized at 10 ns intervals. Fig. 3a shows the

3
a
qubit excitation

b 2π π
0 0 200

c
Digitizer voltage (V)

0.8 0.4 0.0 -0.4 0 200 400 600 Time (ns) 800

readout

-150 -100 -50 0 50 Time (ns)

100

-0.5

Digitizer voltage (V)

0

0.5 1.0

d
qubit excitation

e
-0.5 0 0.5 Digitizer voltage (V)

400 600 Time (ns)

800

1000

f
Digitizer voltage (V) 0.4 0.2 0.0 -0.2 -0.4 -0.6 -2 0 2

readout

-3

0

3 6 Time (μs)

9

12 -2 0 2 4 6 Time (μs) 8 10

4 6 Time (μs)

8

10

FIG. 3: Quantum jumps. (a) is a timing diagram for (b) and (c). The qubit is excited with a pulse of varying amplitude, and the readout (black) is immediately energized, causing the cavity population (purple) to rise and effect a measurement. (b) shows 20 representative single-shot traces from each of three different qubit pulse amplitudes corresponding to rotations of 0, π, and 2π. Abrupt quantum jumps from the excited state (white) to the ground state (blue) are clearly visible for the data corresponding to the π pulse, while the traces corresponding to 0 and 2π mostly do not show such features. Three individual single-shot time traces are shown in (c); the blue trace was taken following a 2π pulse, while the red and green traces were taken following a π pulse. (d) is a timing diagram for (e) and (f ). Here the readout is energized with the qubit in the ground state, and then a continuous qubit drive is applied after a 3 µs delay. (e) and (f ) show 60 traces and one trace, respectively, of the qubit jumping between the ground and excited state under the influence of both the qubit drive and measurement pinning.

timing diagram and Fig. 3b plots 20 individual traces for each of three pulse amplitudes corresponding to 0, π, and 2π qubit rotations. One can clearly see abrupt quantum jumps from the excited state (white) to the ground state (blue) for the data corresponding to a π pulse, while the traces corresponding to 0 and 2π show the qubit mostly in the ground state. A few traces after 0 and 2π pulses show jumps to the excited state, and a few traces after a π pulse are never measured to be in the excited state. We attribute the first effect to qubit state mixing due to high photon numbers in the readout cavity17 , and the second effect to the qubit spontaneously decaying before the cavity can ramp up14 . Three representative traces of the quantum jumps are shown in Fig. 3c, one where the qubit was prepared in the ground state (blue) and two where it was prepared in the excited state and subsequently decayed (red and green). We can also look at the effects of simultaneous qubit excitation and measurement. We energize the readout and then turn on a long qubit excitation pulse after a few µs, as shown in Fig. 3d. This qubit drive tries to coherently change the qubit state while the projective measurement forces the qubit to be in the ground or excited state, resulting in the random telegraph signal seen in Figs. 3e and 3f. This inhibition of qubit state evolution due to measurement is the essence of the quantum

Zeno effect15,25 and will be the subject of future work using samples with longer coherence times, allowing us to examine the dependence of the phenomenon on measurement strength and qubit excitation power. Finally, we look at the statistics of these quantum jumps. Fig. 4a plots a histogram of 2 × 104 individual measurements with the qubit prepared in the excited state, as a function of digitizer voltage and time (see Supplementary Video 1 which shows the evolution of the histogram as a function of qubit rabi drive strength). Most of the population is measured in the excited state (positive voltage) at t = 0 and then decays to the ground state (negative voltage) with a time constant of 290 ns (Fig. 4b, inset). Despite the large separation between the ground and excited state peaks, the maximum qubit readout fidelity is about 70%. This can be almost entirely attributed to the lifetime of the qubit (290 ns) being comparable to the cavity rise time (66 ns)14 , which means that around 30% of the excited state population decays to the ground state before the measurement is made. Since we can resolve individual decay events, we can also plot a histogram of the jump (excited to ground state) times as shown in Fig. 4b (see Supplementary Video 2 which shows the evolution of this histogram as it builds up). The histogram shows an exponential decay with a time constant of 270 ns. Both these time constants are con-

4 Methods The qubit and readout resonator were fabricated on a bare high-resistivity Si wafer using electron beam lithography and double-angle Al evaporation with an intervening oxidation step. The evaporated films were 35 and 80 nm thick respectively and were deposited with an e-beam evaporator. The amplifier resonator was fabricated on a high-resistivity Si wafer. The wafer was patterned with a rectangular 300-nm-thick sputtered Nb ground plane and coated with 125 nm of SiNx grown by PECVD. The amplifier resonator was then fabricated on top of the SiNx layer using electron beam lithography and double-angle Al evaporation with oxidation, in the same manner as the readout resonator9 . The experiment was carried out in a dilution refrigerator with a base temperature of 30 mK. The readout resonator had inductance L=1.25 pH and capacitance C=575 fF, with a bare resonant frequency of 5.923 GHz and a linewidth of 4.9 MHz. These numbers and the behavior of the resonator match well with finite element simulations performed using Microwave Office. The amplifier resonator was formed by a two junction superconducting quantum interference device (SQUID) with a maximum critical current of 4 µA shunted by two parallel plate capacitors in series with a total capacitance of 6.5 pF, with a resonant frequency tunable between 4 and 6.2 GHz. The transmon qubit had measured parameters EJ = 11.4 GHz and EC = 280 MHz at the operating frequency of 4.753 GHz, corresponding to a maximum EJ of 17.9 GHz at zero flux bias. The qubit transition frequency could be tuned between 4 and 5.8 GHz. Readout cavity photon occupation was calibrated using the ac Stark effect24 . Measurements of T1 as a function of qubit frequency agreed well with decay due to the singlemode Purcell effect in combination with an unknown relaxation channel giving an effective Q of 11,50013 . Ramsey fringes yielded T2 = 290 ns at the operating point. We observed a small leakage of pump photons from the amplifier resonator back to the readout resonator, consistent with the finite isolation of the circulators. Using the ac Stark effect, we determined the leakage corresponded to a readout cavity population of much less than one photon. The leakage did cause dephasing, which we verified by Ramsey fringes. We were able to cancel this leakage by applying a coherent microwave tone to the input port. Because the amplifier resonator has a very fast response, one could avoid the leakage problem by turning the paramp off during qubit manipulation and evolution. Acknowledgments We acknowledge useful discussions with M. H. Devoret, S. M. Girvin, R. J. Schoelkopf, D. M. Stamper-Kurn, J. Gambetta, M. Hatridge and O. Naaman. R.V. and I.S. acknowledge funding from AFOSR under Grant No. FA9550-08-1-0104. D.H.S. acknowledges support from a Hertz Foundation Fellowship endowed by Big George Ventures.

a

1.0 Digitizer voltage (V) 0.5 0.0 -0.5
0

-1.0 0 b 2000
Counts (x 103) 10 5 0 0

800 1,600 Counts/bin

200

400 Time (ns)

600

800

1500 Counts 1000 500 0

500 1000 Time (ns)

1500

0

500

1000 Time (ns)

1500

FIG. 4: Jump statistics. (a) shows a histogram of 2 × 104 individual traces with the qubit prepared with a π pulse and n = 8. The excited state signal, centered around 0.6 V, is ¯ clearly resolved from the ground state signal at -0.3 V. The ensemble population is predominately in the excited state at t = 0, and decays to the ground state with a time constant (b, inset) of 290 ns. (b) shows a histogram of jump times from the excited state to the ground state extracted from individual measurements . The solid line is an exponential fit with a decay time of 270 ns. We do not plot jumps that occur less than two cavity time constants after the readout is energized. The pulse protocol is shown in Fig. 3a.

sistent with each other and with the measured T1 of the qubit, as would be expected for a QND measurement17 . The observation of quantum jumps in a superconducting qubit is the first fruit of a powerful technique for quantum measurements in solid state systems. The real-time QND measurement technology demonstrated here is a major required step for implementing quantum error correction. Our techniques can be readily extended to a variety of other systems of interest, including nanomechanical devices, nitrogen vacancy centers in diamond, and lower-dimensional semiconductor systems. These techniques can be used to study the quantum Zeno effect15 and to shed further light on non-idealities in quantum measurement processes17 . They can also enable the development of precision time-resolved single photon sources and detectors operating at microwave frequencies15 , enabling a new class of quantum optics experiments in solid state systems.

5 Author Contributions D.H.S. and R.V. designed the samples. D.H.S. fabricated the samples. R.V. and D.H.S. performed the experiments, analyzed the data, and wrote the manuscript. All work was carried out under the supervision of I.S.

1

2

3

4

5

6

7

8

9

10

11

12

Nagourney, W., Sandberg, J. & Dehmelt, H. Shelved optical electron amplifier: Observation of quantum jumps. Phys. Rev. Lett. 56, 2797–2799 (1986). Bergquist, J. C., Hulet, R. G., Itano, W. M. & Wineland, D. J. Observation of quantum jumps in a single atom. Phys. Rev. Lett. 57, 1699–1702 (1986). Sauter, T., Neuhauser, W., Blatt, R. & Toschek, P. E. Observation of quantum jumps. Phys. Rev. Lett. 57, 1696– 1698 (1986). Basche, T., Kummer, S. & Brauchle, C. Direct spectroscopic observation of quantum jumps of a single molecule. Nature 373, 132–134 (1995). Gleyzes, S. et al. Quantum jumps of light recording the birth and death of a photon in a cavity. Nature 446, 297– 300 (2007). Peil, S. & Gabrielse, G. Observing the quantum limit of an electron cyclotron: QND measurements of quantum jumps between Fock states. Phys. Rev. Lett. 83, 1287– 1290 (1999). Clarke, J. & Wilhelm, F. K. Superconducting quantum bits. Nature 453, 1031–1042 (2008). Wallraff, A. et al. Strong coupling of a single photon to a superconducting qubit using circuit quantum electrodynamics. Nature 431, 162–167 (2004). Hatridge, M., Vijay, R., Slichter, D. H., Clarke, J. & Siddiqi, I. Dispersive magnetometry with a quantum limited SQUID parametric amplifier (2010). URL http: //arxiv.org/abs/1003.2466v1. Ahn, C., Doherty, A. C. & Landahl, A. J. Continuous quantum error correction via quantum feedback control. Phys. Rev. A 65, 042301 (2002). Braginsky, V. B. & Khalili, F. Y. Quantum nondemolition measurements: the route from toys to tools. Rev. Mod. Phys. 68, 1 (1996). 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).

13

14

15

16

17

18

19 20

21

22 23

24

25

Houck, A. A. et al. Controlling the spontaneous emission of a superconducting transmon qubit. Phys. Rev. Lett. 101, 080502 (2008). Wallraff, A. et al. Approaching unit visibility for control of a superconducting qubit with dispersive readout. Phys. Rev. Lett. 95, 060501 (2005). Gambetta, J. et al. Quantum trajectory approach to circuit QED: Quantum jumps and the Zeno effect. Phys. Rev. A 77, 012112 (2008). Manucharyan, V. E., Koch, J., Glazman, L. I. & Devoret, M. H. Fluxonium: Single Cooper-pair circuit free of charge offsets. Science 326, 113–116 (2009). Boissonneault, M., Gambetta, J. M. & Blais, A. Dispersive regime of circuit QED: Photon-dependent qubit dephasing and relaxation rates. Phys. Rev. A 79, 013819 (2009). Mallet, F. et al. Single-shot qubit readout in circuit quantum electrodynamics. Nature Phys. 5, 791–795 (2009). Martinis, J. M. Superconducting phase qubits. Quant. Info. Proc. 8, 81–103 (2009). Castellanos-Beltran, M. A. & Lehnert, K. W. Widely tunable parametric amplifier based on a superconducting quantum interference device array resonator. Appl. Phys. Lett. 91, 083509 (2007). Bergeal, N. et al. Phase-preserving amplification near the quantum limit with a Josephson ring modulator. Nature 465, 64–68 (2010). Koch, J. et al. Charge-insensitive qubit design derived from the Cooper pair box. Phys. Rev. A 76, 042319 (2007). Vijay, R., Devoret, M. H. & Siddiqi, I. Invited review article: The Josephson bifurcation amplifier. Rev. Sci. Instrum. 80, 111101 (2009). Schuster, D. I. et al. AC stark shift and dephasing of a superconducting qubit strongly coupled to a cavity field. Phys. Rev. Lett. 94, 123602 (2005). Itano, W. M., Heinzen, D. J., Bollinger, J. J. & Wineland, D. J. Quantum Zeno effect. Phys. Rev. A 41, 2295–2300 (1990).

Supplement to “Observation of quantum jumps in a superconducting artificial atom”
R. Vijay∗ , D. H. Slichter∗ , and I. Siddiqi
Quantum Nanoelectronics Laboratory, Department of Physics, University of California, Berkeley CA 94720 (Dated: September 16, 2010)

I.

DETAILS OF AMPLIFICATION MECHANISM

arXiv:1009.2969v1 [cond-mat.mes-hall] 15 Sep 2010

a
Pump power bistable (fc,Pc)

+90 Pump phase

o

b

paramp

-90 Pump frequency

o

Pump power

effective resonant frequency as a function of bias frequency and power. For drive power P > PC and drive frequency f < fC , the system is bistable within the boundary indicated by the dotted lines. This bistability has been employed for qubit readout in the past1,2 . In this work, we bias the resonator just outside the bistable regime. This bias region, marked as “paramp” in Fig. S1a, has been accessed to demonstrate parametric amplification3–5 . Fig. S1b explains the principle of small signal degenerate parametric amplification. The reflected pump phase is plotted as a function of pump power, giving the transfer function of the amplifier. A weak input signal is combined with the pump and modulates its amplitude, resulting in a large change in the phase of the reflected pump. Since there are many more pump photons than signal photons, this effectively amplifies the input signal. Two key points should be noted here. First, this technique implements a phase-sensitive amplifier6 , since only signals in phase with the pump will be amplified. Secondly, the amplifier is linear only for input signals which keep the bias conditions in the linear region of the transfer function shown in Fig. S1a. However, in this work we are only concerned with two possible cavity states corresponding to the two states of the qubit. Therefore, the linearity of the amplifier is inconsequential here as long as it faithfully maps the two cavity states to two distinct output states.

FIG. S1: The phase diagram of the amplifier resonator is shown schematically in (a). The resonance frequency (solid line) gradually decreases with increasing drive power. Above a critical drive power PC and below a critical frequency fC the resonator is bistable. At powers slightly below PC , the resonator functions as a parametric amplifier (paramp). (b) shows a linecut of (a) at constant pump frequency in the paramp regime. A weak signal modulating the drive power leads to large changes in the reflected phase of the pump.

The phase diagram of our non-linear resonator is shown schematically in Fig. S1a. The solid line indicates the

1

2 3

4

Vijay, R., Devoret, M. H. & Siddiqi, I. Invited review article: The josephson bifurcation amplifier. Rev. Sci. Instrum. 80, 111101 (2009). Mallet, F. et al. Single-shot qubit readout in circuit quantum electrodynamics. Nature Phys. 5, 791–795 (2009). Castellanos-Beltran, M. A. & Lehnert, K. W. Widely tunable parametric amplifier based on a superconducting quantum interference device array resonator. Appl. Phys. Lett. 91, 083509 (2007). Vijay, R. Josephson Bifurcation Amplifier: Amplifying

5

6

quantum signals using a dynamical bifurcation. Ph.D. thesis, Yale University (2008). Hatridge, M., Vijay, R., Slichter, D. H., Clarke, J. & Siddiqi, I. Dispersive magnetometry with a quantum limited squid parametric amplifier (2010). URL http://arxiv.org/abs/ 1003.2466v1. Caves, C. M. Quantum limits on noise in linear amplifiers. Phys. Rev. D 26, 1817–1839 (1982).

Sign up to vote on this title
UsefulNot useful