Superconducting traveling-wave parametric amplifier

Abstract

A system and method are disclosed for a superconducting traveling-wave parametric amplifier (TWPA) with improved control and performance. In a preferred embodiment, the amplifier comprises an integrated array of symmetric rf-SQUIDs in a transmission line structure. A device was fabricated using niobium superconducting integrated circuits, and confirmed predicted performance, with a maximum gain up to 17 dB and a bandwidth of 4 GHz. A similar device can be applied as a low-noise, low-dissipation microwave amplifier for output from a superconducting quantum computer, or as a preamplifier, switch, or frequency converter for a sensitive microwave receiver, or as an output amplifier for a frequency-multiplexed superconducting detector array.

Claims

1. A traveling-wave parametric amplifier, comprising: an input port configured to receive a signal comprising a combination of an input microwave signal, and a microwave pump signal having a pump frequency which is outside a bandwidth of the input microwave signal; a plurality of successive repeated amplifier stages configured to form a microwave transmission line, each amplifier stage comprising a shunt capacitor and a serial inductor, an inductance of the serial inductor of each amplifier stage being a nonlinear function of a total signal in the serial inductor, wherein the serial inductor of each amplifier stage comprises at least one Radio Frequency superconducting quantum interference device (rf-SQUID) comprising a Josephson junction in parallel with an inductive loop; at least two control ports configured to accept at least two control signals, each of which modulates the inductance of the serial inductor of each amplifier stage; and an output port, configured to present an output microwave signal comprising an amplified replica of the microwave input signal.

2. The traveling-wave parametric amplifier according to claim 1, wherein each serial inductor comprises at least two Josephson junctions.

3. The traveling-wave parametric amplifier according to claim 2, wherein the Josephson junctions comprise the superconductors niobium or aluminum.

4. The traveling-wave parametric amplifier according to claim 2, wherein the plurality of successive repeated amplifier stages are configured to operate at a temperature of 4.2 K or less.

5. The traveling-wave parametric amplifier according to claim 2, wherein the plurality of successive repeated amplifier stages are configured to operate at a temperature less than 1 K.

6. The traveling-wave parametric amplifier according to claim 1, where the input port is configured to receive signals from superconducting photon detectors.

7. The traveling-wave parametric amplifier according to claim 1, wherein the serial inductor of each amplifier stage comprises two parallel rf-SQUIDs.

8. The traveling-wave parametric amplifier according to claim 7, wherein the two parallel rf-SQUIDs are mutually symmetrical, constituting a symmetric rf-SQUID.

9. The traveling-wave parametric amplifier according to claim 7, wherein the inductive loop of each rf-S QUID is configured to inductively couple magnetic flux from a respective control signal in proximity to the loop.

10. The traveling-wave parametric amplifier according to claim 1, wherein the serial inductor of each amplifier stage exhibits a second-order nonlinearity and a third-order nonlinearity, the two control signals being configured to separately control the second-order nonlinearity and the third-order nonlinearity of each amplifier stage.

11. The traveling-wave parametric amplifier according to claim 1, wherein the plurality of successive repeated amplifier stages are fabricated as an integrated circuit on a single chip.

12. The traveling-wave parametric amplifier according to claim 1, comprising at least 100 repeated stages of the plurality of successive repeated amplifier stages.

13. The traveling-wave parametric amplifier according to claim 1, wherein the input port is configured to receive an input microwave signal having frequency components in the range of 3-10 GHz and a power level of 120 dBm or less.

14. The traveling-wave parametric amplifier according to claim 1, wherein the microwave pump signal has a pump frequency of about 10 GHz and a pump power level of about 65 dBm.

15. The traveling-wave parametric amplifier according to claim 1, wherein the output port is configured to present the output microwave signal exhibiting a signal power gain greater than 12 dB with respect to the input microwave signal.

16. The traveling-wave parametric amplifier according to claim 1, wherein each control signal comprises an electrical current between about 1 A and 1 mA.

17. The traveling-wave parametric amplifier according to claim 1, wherein the input port and output port each match a microwave impedance of about 50 ohms.

18. The traveling-wave parametric amplifier according to claim 1, wherein the output port is further configured to present a frequency-shifted replica of the input signal at a frequency 2f.sub.pf.sub.s, where f.sub.p is the pump frequency and f.sub.s is a component of the input microwave signal.

19. The traveling-wave parametric amplifier according to claim 1, wherein the plurality of successive repeated amplifier stages are responsive to the microwave pump signal to increase a power of the output microwave signal by at least 20 dB with respect to an absence of the pump microwave signal.

20. The traveling-wave parametric amplifier according to claim 1, where the input port is configured to receive signals from superconducting qubits.

21. A method of amplifying a weak broadband microwave input signal emitted from a device operating at cryogenic temperatures, comprising: providing a series of nonlinear tunable superconducting inductors in parallel with an inductive loop, each having nonlinear properties tunable using external controls, in a series array, each nonlinear tunable superconducting inductor having an associated shunt capacitor, the series array forming a transmission line that comprises a distributed amplifier with input, output, and control lines; cooling the distributed amplifier to a cryogenic temperature below the critical temperature of the nonlinear tunable superconducting inductors; combining the input signal with a strong microwave pump signal that is offset in frequency from the input signal to form a combined signal; coupling the combined signal into the input port of the distributed amplifier; measuring the output of the distributed amplifier for gain, bandwidth, and distortion; and tuning nonlinearities of the nonlinear tunable superconducting inductors using the control lines to optimize the performance of the distributed amplifier, based on the gain, bandwidth, and distortion.

22. The method according to claim 21, wherein the combined signal represents a frequency-shifted replica of the weak broadband microwave input signal at a frequency 2f.sub.pf.sub.s, where f.sub.p is the frequency of the strong microwave pump signal and f.sub.s is a component of the weak broadband microwave input signal.

23. A tunable superconducting microwave inductor, comprising: an input port configured to receive a microwave current; two radio frequency superconducting quantum interference devices (rf-SQUIDs) in parallel and defining an inductance, being configured to together pass the microwave current, each rf-SQUID comprising a Josephson junction in parallel with an inductive loop, the inductance having a value dependent on an amplitude of the microwave current and a control current as a nonlinear function of the amplitude with terms that are at least one of quadratic and cubic in the amplitude; an output port, configured to present the microwave current from the two radio frequency superconducting quantum interference devices (rf-SQUIDs) in parallel; a shunt capacitor at the output port; and a control port configured to receive a control currents coupled as a magnetic flux into the inductive loop of each rf-SQUID, to tune the inductance value over a range of inductances and independently tune a gain and bandwidth of the tunable superconducting microwave inductor.

24. The tunable superconducting microwave inductor according to claim 23, further comprising a plurality of tunable superconducting microwave inductors, provided as respective repeated stages in a microwave transmission line configured to function as a traveling-wave parametric amplifier.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) FIG. 1 (prior art) shows a generic block diagram of a parametric amplifier.

(2) FIG. 2 (prior art) shows the circuit diagram of a nonlinear transmission line.

(3) FIG. 3A (prior art) shows the circuit diagram of a basic Josephson TWPA, where the inductance L has been replaced with a Josephson junction.

(4) FIG. 3B (prior art) shows the circuit diagram of an alternative Josephson TWPA, where the inductance L has been replaced with an rf-SQUID.

(5) FIG. 4 shows the circuit diagram of a symmetric rf-SQUID.

(6) FIG. 5A shows the flux-phase relation for the symmetric rf-SQUID.

(7) FIG. 5B shows the dependence of nonlinearity parameters on phase bias of the symmetric rf-SQUID.

(8) FIG. 6 shows the circuit diagram of a symmetric TWPA, or STWPA.

(9) FIG. 7 show a sample STWPA chip and the experimental setup for gain measurements.

(10) FIG. 8 shows the measured gain of the STWPA as a function of frequency.

(11) FIG. 9 shows the output power of the STWPA as a function of the two control currents.

(12) FIG. 10A shows the output power of the STWPA with the pump turned off.

(13) FIG. 10B shows the output power of the STWPA with the pump turned on.

(14) FIG. 11 presents a method for parametric amplification using tunable nonlinear superconducting inductors.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

(15) Superconducting quantum computing demands non-dissipative, quantum-limited cryogenic amplifiers. In order to improve scalability to many qubits, wide-bandwidth amplifiers are required for readout of multiple qubits which can be sensed with a single amplifier. Josephson Parametric Amplifiers (JPAs) were introduced to meet these needs, and proposed in many different designs. As a parametric amplifier, the JPA transfers power from a strong microwave tone (the pump) to a weak one (the signal), as well as to the idle frequency f.sub.i=f.sub.pf.sub.s (see FIG. 1). It has the advantage of operating at cryogenic temperatures with very low loss, as well as having the strong intrinsic nonlinearity of the Josephson inductance. While achieving good gain and noise performance, JPAs cannot guarantee a wide instantaneous bandwidth for many-qubit multiplexing, because of a lumped-circuit based design.

(16) An alternative approach is built around a superconducting transmission line, consisting of superconducting inductors and capacitors (see FIG. 2). If the superconducting inductor is nonlinear, that can create a Traveling-Wave Parametric Amplifier, or TWPA. Good gain-bandwidth results have been achieved by exploiting the weak nonlinear kinetic inductance of NbTiN films, but very long lengths on the order of meters are required, leading to difficult integration.

(17) Recently, the Josephson Traveling Wave Parametric Amplifier (JTWPA) was introduced as a nonlinear transmission line containing Josephson junctions as the nonlinear inductor. Two different approaches were demonstrated in the prior art, as depicted in FIG. 3: using a single Josephson junction as the nonlinear inductor (see FIG. 3A), and using an rf SQUID (a Josephson junction in parallel with a linear resistor, see FIG. 3B) as the nonlinear inductor. Both of these approaches can achieve high gain and good bandwidth, but both have problems. The basic Josephson TWPA is based on a third-order nonlinearity, and requires a resonant circuit at each stage to cancel phase modulation distortion.

(18) The rf-SQUID approach uses external flux bias to a tune the nonlinearity, and can take advantage of more efficient second-order nonlinearity. However, with a single DC flux bias, second and third order nonlinearities cannot be independently tuned, limiting the performance.

(19) To allow independent control of both second and third order Josephson nonlinearities, two independent control currents are needed.

(20) The Symmetric Traveling Wave Parametric Amplifier (STWPA) according to the present technology is based on a new symmetric rf-SQUID scheme. Experimental measurements of the first wafer demonstrated the correct TWPA operation at 4.2 K, showing a gain up to 17 dB and a 4 GHz 3 dB bandwidth.

(21) The key to understanding the performance of the STWPA lies in the nonlinear behavior of the symmetric rf-SQUID with two coupling loops. A symmetric rf-SQUID is obtained by shunting two identical rf-SQUIDs. Taking into account a small parasitic inductance L.sub.p connecting the two rf-SQUIDs, the schematic in FIG. 4 is obtained. The circuit equations can be expressed in terms of the symmetric and antisymmetric phase and flux parameters
.sub.+=(.sub.1+.sub.2)/2, .sub.=(.sub.1.sub.2)/2, .sub.+=(.sub.1+.sub.2)/2, and .sub.=(.sub.1.sub.2)/2, as:
.sub.++.sub.L(1+k)sin(.sub.+)cos(.sub.)=2.sub./.sub.0,
.sub.+=.sub.L[(1+k)/(1+)]sin(.sub.)cos(.sub.+)=[2/(1+)].sub.+/.sub.0,
where =L(1k)/L.sub.p>1 and k=M/L is the normalized mutual inductance.

(22) FIG. 5A shows a graph where .sub.L=0.8, =5, and k<<1. The relation between these parameters is single-valued if .sub.L<1, which defines the preferred non-hysteretic or diffusive mode.

(23) One can further show that for reasonably large values of , the second-order and third-order nonlinearity parameters are given respectively by =(.sub.L/2) cos(.sub.) sin(.sub.+) and =(.sub.L/6) cos(.sub.) cos(.sub.+), which are plotted in FIG. 5B. (Note that with on top in FIG. 5B is the same as .sub.L here.) For a conventional rf-SQUID, only the outer circle is accessible. In contrast, for the symmetric rf-SQUID, any set of parameters inside the circle is also accessible. This enables a much higher degree of control. More specifically, each point on a circle represents a radius r=|cos(.sub.)| and angle .sub.+.

(24) To integrate a symmetric rf-SQUID into a TWPA, it must be coupled to a capacitor C shunted to ground to form a unit cell, which is repeated to form a transmission line. The SQUID inductance is given by L.sub.symL/[2(1+.sub.L cos(.sub.+) cos (.sub.))]. This also means that the effective characteristic impedance of the TWPA, Z.sub.0=(L.sub.sym/C).sup.0.5 is somewhat dependent on these phase parameters. A TWPA that is designed to match 50 ohms for one set of parameters may be slightly mismatched for other parameters.

(25) It is also necessary to cut the small coupling inductance L.sub.p in half to attach the cells together, in order to maintain the symmetry to the device, as shown in FIG. 6. A capacitor may be attached for every SQUID to form a single cell. Alternatively, it may be more convenient to fabricate a double value of C attached to every second cell, for example. That should be functionally equivalent, provided that the effective wavelength at the given signal frequency on the transmission line is much longer than the cell size.

(26) All of the symmetric rf-SQUIDs in a given STWPA are typically essentially identical, and all are flux-biased by the same control currents I.sub.1 and I.sub.2 along the entire length of the lines. The measurements indicate that the devices as fabricated were in fact sufficiently similar along the full length. The flux bias was tested as a low-frequency, nominally dc control, but it is believed that switching of this flux bias up to at least MHz frequencies should be feasible; and the SQUID itself should be responsive up to much higher frequencies.

(27) The total number of cells is not arbitrary. Since this is a distributed amplifier, additional cells will tend to increase the gain, at least until the pump starts to be depleted. After this point, however, further cells can actually lead to power reduction due to harmonic generation at the signal frequency. Simulations and measurements have confirmed that of order 100-200 cells can provide useful gain, and can also fit on a 5-mm chip, with some meandering as shown in the chip photograph in FIG. 7.

(28) In the samples fabricated, the capacitor was based on two layers of superconducting niobium separated by a thin insulating layer of silicon dioxide (SiO.sub.2), with dielectric constant of .sub.r=3.9. For example, if the SiO.sub.2 layer is d=100 nm thick, the specific capacitance would be C=.sub.r.sub.0/d=0.35 pF/m.sup.2. So if L.sub.sym=5 pH and C=2 fF (corresponding to an area of 6 m.sup.2), this would correspond to Z.sub.0=(L.sub.sym/C).sup.0.5=50 ohms. A key advantage of using a superconductor for both L and C is that resistive loss is extremely small, corresponding to negligible attenuation due to loss.

(29) Regarding the measurements of amplifier gain using the spectrum analyzer, the gain in the signal channel was defined as G.sub.s=P.sub.on/P.sub.off, where P.sub.off and P.sub.on were the signal power (in dBm units) at the output port (connected to the spectrum analyzer) with the amplifier turned off (bias currents and pump off) and with the amplifier on (bias currents and pump on). The off-state values could have been affected by some frequency-dependent mismatch, because the STWPA was designed to match an impedance of 50 ohms with the current bias on. However, with the current bias turned off, the inductance and hence the impedance was shifted slightly. The maximum deviation of the OFF state signal power due to mismatch was then estimated, resulting in a maximum reflection loss of about 2 dB.

(30) FIG. 8 shows the gain for a particular STWPA as a function of input frequency, for a fixed pump frequency of 10.2 GHz, a fixed pump power of 65 dBm, and a fixed set of control currents. This demonstrated a gain >9.5 dB in a 4 GHz bandwidth centered at 5.9 GHz, with a nominal peak value of 12.5 dB. Moreover, by varying the pump power around this value, higher gain was achieved for some signal frequencies, such as a 17 dB gain at 7.9 GHz, 11 dB at 8.4 GHz and 9.4 GHz. The dip at 4.9 GHz was unexpected, and may reflect an unintended resonance in the transmission line or in its connections. This might also be responsible for the variation with pump power. However, we believe that this prototype clearly demonstrated the operating principles of the STWPA, even if this leaves room for future improvement.

(31) FIG. 9 shows the output power for a fixed input frequency of 4.4 GHz, where the output signal power with the pump OFF is plotted as a function of control currents. This would not be expected to exhibit gain, and in fact shows a strong loss for those same control parameters that correspond to gain with the pump ON. As expected, this loss is symmetric around the I.sub.1=I.sub.2 line (I.sub.=0), and also is periodic in I.sub.+the diagonal segments of greatest attenuation are equally spaced by a value of I.sub.+=0.625 mA, starting at I.sub.+=0.125 mA. These diagonal segments correspond to a very high second-order effect, which is maximized when .sub.+=/2. The attenuation is greatest here because 2.sup.nd harmonic generation is large here, and indeed, there is a corresponding peak (not shown) at twice the input frequency.

(32) Similar behavior is exhibited for a different STWPA sample in FIGS. 10A and 10B. This compares signal power vs. bias currents with pump off (FIG. 10A) and pump on (FIG. 10B), for a fixed frequency. For FIG. 10A, on the segment corresponding to .sub.+=/2, with pump off, the signal is mostly attenuated and converted into its second harmonic component. When .sub.+=/2, the second-order effect has maximum efficiency, giving a strong second harmonic generation. With the pump on (FIG. 10B), the segment corresponding to .sub.+=/2 is where the signal is most strongly amplified. In fact, in the ON state, the second-order effect generates a strong second-order parametric conversion process, corresponding to the three-wave mixing regime.

(33) More generally, for the biases corresponding to .sub.+=/2, across the entire 4 GHz bandwidth, the difference in output signal power between pump ON and OFF was at least 20 dB, at pump OFF, the This mechanism could then be used as a broadband, non-resonant Pump Driven On/Off RF-Switch. This could play a useful role in many applications, where very weak signals need to be accurately controlled.

(34) Moreover, we verified the correct second-order operation by checking the idler frequency f.sub.i=f.sub.pf.sub.s at the output port, in addition to the signal frequency. The idler was relatively high in power, reaching values between 135 dB and 123 dB. Nonlinear mixing is also generated at the sum of the component frequencies. This means that the STWPA could be potentially useful also as a coherent Up-Down converting rf-Mixer for weak signals, allowing the direct implementation of wave mixing at 4K or even at the mK stage of a dilution refrigerator. This STWPA approach could help the transition to fully superconducting control electronics, replacing room-temperature mixers. Together with the amplifier itself, these may make the STWPA a versatile system that could perform a wide range of operations with the same on-chip circuit.

(35) In addition to the specific systems described in this application, FIG. 11 presents a general method for implementing amplifiers based on tunable nonlinear superconducting inductors. The key is the tunability based on two or more controls, which enables independent optimization of multiple performance metrics. By successive iteration of tuning and measurement, one may obtain superior performance in terms of an appropriate combination of gain, bandwidth, efficiency, distortion, and noise.

(36) Other devices, apparatus, systems, methods, features and advantages of the invention will be or will become apparent to one with skill in the art upon examination of the following figures and detailed description. It is intended that all such additional systems, methods, features and advantages be included within this description, be within the scope of the invention, and be protected by the accompanying claims.