PHYSICAL REVIEW A 79, 052115 共2009兲

Quantum nondemolition photon detection in circuit QED and the quantum Zeno effect

Ferdinand Helmer,1 Matteo Mariantoni,2,3 Enrique Solano,1,4 and Florian Marquardt1

1
Department of Physics, Center for NanoScience, and Arnold Sommerfeld Center, Ludwig-Maximilians-Universität,
Theresienstrasse 37, D-80333 Munich, Germany
2
Walther-Meissner-Institut, Bayerische Akademie der Wissenschaften, Walther-Meissner-Str. 8, D-85748 Garching, Germany
3
Physics Department, Technische Universität München, James-Franck-Str., D-85748 Garching, Germany
4
Departamento de Quimica Fisica, Universidad del Pais Vasco-Euskal Herriko Unibertsitatea, 48080 Bilbao, Spain
共Received 17 December 2007; revised manuscript received 15 January 2009; published 20 May 2009兲
We analyze the detection of itinerant photons using a quantum nondemolition measurement. An important
example is the dispersive detection of microwave photons in circuit quantum electrodynamics, which can be
realized via the nonlinear interaction between photons inside a superconducting transmission line resonator. We
show that the back action due to the continuous measurement imposes a limit on the detector efficiency in such
a scheme. We illustrate this using a setup where signal photons have to enter a cavity in order to be detected
dispersively. In this approach, the measurement signal is the phase shift imparted to an intense beam passing
through a second cavity mode. The restrictions on the fidelity are a consequence of the quantum Zeno effect,
and we discuss both analytical results and quantum trajectory simulations of the measurement process.

DOI: 10.1103/PhysRevA.79.052115 PACS number共s兲: 03.65.Ta, 03.65.Xp, 42.50.Lc

I. INTRODUCTION to be already prepared in a certain state. Although the quan-

tum Zeno effect also plays a role in the CPB dynamics in
Quantum nondemolition 共QND兲 measurements are ideal 关19兴, the conclusions and reasoning for the optimal measure-
projective measurements that reproduce their outcome when ment rate in our case are different.
repeated 关1,2兴. Using them, it is possible to measure the state Such dispersive photon detection schemes could have a
of a system with the minimal disruption required by quantum particularly important application in the field of quantum
mechanics. Recent successful experimental demonstrations electrodynamics 共QED兲 in superconducting circuits. Circuit
of QND detection for superconducting qubits and microwave QED setups 关12,20–22兴 offer the possibility to implement
photons 关3–6兴 are both of fundamental interest and crucial ideas of quantum optics using microwave photons propagat-
for the development of quantum communication and infor- ing in transmission lines and they also could form the basis
mation processing. When QND detection is applied continu- for novel architectures capable of scalable, fault-tolerant
ously to a system that would otherwise undergo some intrin- quantum computing 关23兴. Proposals for generating nonclas-
sic dynamics, quantum jumps are observed, tracing the sical photon states exist or have been implemented
quantum evolution in real time 关7–11兴. As a consequence, the 关4,21,24,25兴. However, the on-chip single-shot detection of
dynamics tends to be frozen, a result now known as the itinerant photons is still missing.
quantum Zeno effect 关12–17兴. Recent experiments have demonstrated dispersive qubit
In the present paper, we show that the interplay of these detection 关3兴 and measurements of photon statistics 关4兴.
phenomena may put interesting constraints on the detection Based on the dispersive coupling strengths demonstrated
of itinerant quanta. The specific minimal example we will there, one could employ a superconducting qubit inside a
discuss concerns the continuous dispersive QND detection of microwave transmission line resonator to induce a nonlinear
single photons passing through a cavity. The crucial distinc- coupling between two modes of the resonator 共or alterna-
tion to be recognized is the following. For localized quanta tively couple two cavities兲, thus creating a dispersive photon
共e.g., a photon already created inside a cavity 关8,18兴兲, the detector of the type discussed here. Even though we find a
quantum Zeno effect could presumably only enhance the de- limit for the photon detection fidelity of about 30% for the
tection by suppressing the decay. However, this no longer simplest scheme considered here, this would still be a con-
holds for the detection of itinerant quanta if we require that siderable improvement over the present state of the art.
our detector is always working and can detect the quantum
without knowing the arrival time in advance. As we will II. MODEL
show, in this case the unavoidable back action of the mea-
surement device produces a quantum Zeno effect, suppress- We investigate a QND scheme utilizing the nonlinear
ing the fidelity of measurements. Kerr-type coupling 关26–28兴 of two discrete localized modes
Similar measurement physics is involved when trying to of a bosonic field. The presence of a quantum inside the
read out the phonon state of a nanoresonator using a single signal mode gives rise to a frequency shift of the detection
electron transistor 共SET兲 and a Cooper pair box 共CPB兲 as a mode, which can be observed dispersively via the phase shift
coupler 关19兴. However, there are some distinct differences to of a beam transmitted through that mode 共see Fig. 1兲. In turn,
the scheme considered here. The most prominent difference the signal mode frequency fluctuates due to the detection
is that we analyze a scheme to detect itinerant quanta 共in our beam’s shot noise. As a consequence, the incoming signal
case photons兲 and therefore do not assume the signal mode photon will be reflected with a probability that rises with

1050-2947/2009/79共5兲/052115共10兲 052115-1 ©2009 The American Physical Society

HELMER et al. PHYSICAL REVIEW A 79, 052115 共2009兲

III. STOCHASTIC MASTER EQUATION

In this section we derive a stochastic master equation for
this measurement situation. Starting from a standard Lind-
blad master equation, involving both driving and decay for
the two modes, we can derive a stochastic master equation
that keeps track of the individual measurement signal ob-
tained in each individual run of the experiment. Before pre-
FIG. 1. 共Color online兲 Schematic sketch of the model. Two cav- senting that equation 关Eq. 共2兲 below兴, let us briefly describe
ity modes are coupled anharmonically. The detector mode is irradi- in qualitative terms the purpose of using a stochastic master
ated with a strong coherent field that suffers a phase shift whenever equation. Imagine a general measurement situation where the
a photon is present in the signal mode. system in question cannot be directly projectively measured,
but instead one has to couple a measurement apparatus 共most
coupling strength and detection beam intensity. often a meso- or macroscopic apparatus兲 to the system. Only
This incarnation of the quantum Zeno effect generates a this apparatus can then be projectively measured. Assuming
trade-off that yields the highest detection efficiency at inter- that apparatus and coupling to the system are indeed suitable
mediate coupling strengths. In that way, such dispersive to extract information about a certain system observable via
schemes for itinerant quanta turn out to be similar to weak a projective measurement on an apparatus observable, during
measurements using general linear detectors and amplifiers such a measurement the following happens: We let system
关2兴. and apparatus, initially assumed to be in a product state,
We proceed as follows. 共i兲 We numerically evaluate quan- interact according to the coupling Hamiltonian for some time
tum jump trajectories for the phase-shift signal in a minimal interval ⌬t. Afterwards, the state will have changed, gener-
model of a QND photon detector and analyze the fraction of ally into an entangled state between system and measure-
detected photons, observing the trade-off described above. ment apparatus. It is by this entanglement that information
共ii兲 We interpret these findings using an analytical approxi- about the system state is transferred into the measurement
mation. 共iii兲 Finally, we briefly comment on possible experi- apparatus. Then we imagine to projectively measure the
mental realizations. measurement apparatus. This has several consequences: 共i兲
We consider a system of two cavity modes with a Kerr- the measurement apparatus experiences a projection onto one
type coupling of strength g, of the 共usually smoothly distributed兲 eigenstates of the appa-
ratus observable; 共ii兲 the system and the measurement appa-
ratus are now again in a product state; 共iii兲 this evidently also

冉 冊
Ĥ = ប␻ n̂ +
1
2
冉
+ ប␻det n̂det +
1
2
冊
+ បgn̂n̂det + Ĥdrive+decay .
changed the system state 共this is called measurement back
action兲, however not necessarily into an eigenstate of the
system. It is important to realize that this change is also ran-
共1兲 dom, as it is conditional on the measurement result we ob-
tained when projectively measuring the apparatus.
Very often and especially in the situation we consider, the
These modes might represent two different electromagnetic coupling between system and measurement apparatus is
field modes inside an optical or microwave cavity, the modes weak, meaning that during one measurement as we just de-
of two adjacent cavities 关29兴, or even two anharmonically scribed, the system state only changes very little. This is then
coupled modes of a nanomechanical resonator. Photons in referred to as a “weak” measurement. When we now imagine
the signal mode 共frequency ␻, number operator n̂兲 and the that we take the continuum limit ⌬t → 0, it is intuitively clear
detector mode 共␻det, n̂det兲 decay by leaking out of the cavity. that the infinitesimally small kicks the system experiences
The anharmonic Kerr-type coupling arises generically when should lead to stochastic dynamics. We can therefore expect
introducing any nonlinear medium, such as an atom, a qubit, 共i兲 having a stochastic term enter the master equation which
or a quantum dot, into a cavity and has been studied for the leads to 共ii兲 diffusionlike behavior of the system state and
purpose of QND measurements in quantum optics 关26–28兴. 共iii兲 eventually the system will be driven into an eigenstate of
It induces a phase shift in the strong detection beam 共具n̂det典 the system observable we indirectly measured. The time
Ⰷ 1兲 upon presence of a signal photon. scale on which this indirect projection happens corresponds
We are interested in analyzing individual realizations of exactly to the time scale on which we could extract enough
the phase-shift signal as a function of time. The phase shift information out of the measurement signal 共the sequence of
can be observed by continuously measuring an appropriate results from projectively measuring the apparatus兲 to infer
field quadrature of the detection beam 共e.g., in a homodyne the value of the system observable with certainty. On this
setup兲. As the beam passes through the cavity, the beam be- time scale, we have indirectly performed a QND measure-
comes entangled weakly with the cavity’s state. Thus, the ment of the system. No matter how much longer we keep on
stochastic measurement outcomes reveal information about measuring, the result of subsequent measurements will al-
that state, feeding back into the time evolution of the cavity’s ways give the same result, as the system has been projected.
density matrix. This physics is described by a stochastic mas- In addition to the back action modifying the quantum dy-
ter equation 关2,18,30–32兴 for the density matrix ␳ˆ condi- namics, we also obtain the classical measurement signal. It is
tioned on the output signal 共see 关18兴兲. again intuitively clear that this signal should be noisy 共it is a

052115-2
QUANTUM NONDEMOLITION PHOTON DETECTION IN… PHYSICAL REVIEW A 79, 052115 共2009兲

sequence of random projections of the measurement appara-

tus兲 but should contain the same realization of the noise as
the internal back-action dynamics.
X共t兲 ⬅ 具n̂典共t兲 +
1
4
冑 1
⌫
␰共t兲. 共4兲

To summarize, this method of analyzing weak measure-

ments gives us access to concrete realizations of the classical
noisy signal while we also obtain the system internal mea- IV. ANALYSIS OF DETECTION EFFICIENCY
surement back-action modified quantum dynamics for each
We analyze a situation with a continuous weak coherent
of those realizations. This allows to design and validate ex-
perimentally realistic detection schemes in a powerful way beam of photons entering at a rate Ṅin into the signal mode,
and is the method of choice for the present analysis. whose intensity decay rate is ␬ 关first line of Eq. 共3兲兴. We have
In this derivation we follow 关18,33–35兴, adapted to the chosen to work in the limit of a large detector mode decay
case of zero temperature and a two-sided cavity. As a starting rate, ␬det Ⰷ ␬, which is favorable for the detection process
point we write down a master equation to model the quantum and makes it possible to adiabatically eliminate that mode
dynamics of the system introduced in Fig. 1 and Eq. 共1兲, 关18兴, keeping only the signal mode n̂ = â†â and drastically

␳˙ˆ = − i
2
冋
␬1
共␣â†1 + ␣ⴱâ1兲 + gâ†1â1â†0â0, ␳ˆ 册 reducing the numerical effort. After adiabatic elimination,
the coupling strength g and the detection beam intensity are
combined into the measurement rate 关18兴

冉 1 1
+ ␬ â0␳ˆ â†0 − â†0â0␳ˆ − ␳ˆ â†0â0
2 2
冊 ⌫ ⬅ g2具n̂det典/共4␬det兲, 共5兲

冉 冊
where 1 / ⌫ is the time scale needed to resolve different pho-
1 1 ton numbers. The last, stochastic term in Eq. 共3兲, describes
+ ␬1 â1␳ˆ â†1 − â†1â1␳ˆ − ␳ˆ â†1â1
2 2 the measurement back action.
冑
− i Ṅin/2␬关â0 + â†0, ␳ˆ 兴. 共2兲
It contains a systematic term depending on the average
number of signal photons, as well as a stochastic term rep-
The first Hamiltonian term describes driving of the detec- resenting the unavoidable vacuum noise, where 具␰典 = 0 and
tor mode 共annihilation operator â1, decay rate ␬1兲 as well as 具␰共t兲␰共t⬘兲典 = ␦共t − t⬘兲. In deriving Eqs. 共3兲 and 共4兲, we have
the Kerr-type nonlinear coupling between the modes. The assumed that the transmitted and reflected signals are super-
driving amplitude ␣ results in a coherent state in the signal imposed symmetrically to extract the maximum information.
mode with an average photon number 兩␣兩2. The signal mode As in any measurement of field quadratures, temporal fil-
annihilation operator is denoted as â0 and the its decay rate is tering is required to suppress the noise. We average over a
labeled ␬. The second and third terms are the Lindblad-decay time span ␶avg, which should be as large as possible while
terms for both modes reflecting the finite Q factor of the still remaining smaller than the expected temporal extent of
cavity. The fourth term is a weak driving of the signal mode the phase-shift signal due to a single photon, i.e., ␶avg not
to model the itinerant photons impinging onto the cavity. much bigger than ␬−1. We denote the averaged signal as X̄共t兲.
Here, Ṅin describes the rate of itinerant photons impinging
onto the cavity from the outside. Although the drive is co-
herent, after a suitable unraveling of the master equation it A. Numerical results
will become apparent that the Poissonian nature of the co- We numerically solve the master equation, using it to
herent drive will indeed model the situation we have in mind
compute the signal X̄共t兲 and the occupation of the signal
correctly—namely, that once in a while a photon comes
along the transmission line, hits the cavity, and can then be mode 具n̂典共t兲 as a function of time. We then implement the
detected or not. Note that we have chosen to work in an minimal model of a threshold detector. Time points when the
interaction picture and in a rotating frame to eliminate the quantum jump trajectory X̄共t兲 first exceeds the threshold Xthr
explicit time dependence of the drives. are counted as detection events, and the detector is then set
It is convenient to follow the steps of 关18,33–36兴 to 共i兲 insensitive for a dark time ␶dark, suitably chosen to avoid
derive the unraveled version of Eq. 共A1兲 and to 共ii兲 adiabati-
in Ⰷ ␶dark Ⰷ ␬ .
multiple detection, i.e., Ṅ−1 −1
cally eliminate the detector mode from this equation assum-
ing ␬det Ⰷ ␬. This is done in the Appendix. Finally, we obtain Our discussion will focus on small values of Ṅin, making
a stochastic master equation for the signal mode alone, the results independent of ␶dark, while we will analyze the
greatly facilitating the numerical study of the detection effi- dependence on Xthr in some details. In Fig. 2 we show two
ciency. Its stochastic term accounts for the back action of the example trajectories. Whereas the expected number of signal
measurements performed on the field quadrature leaking out photons is the same for both cases, the increase in the mea-
of the detector mode correctly and reads surement rate ⌫ / ␬ decreases the number of photons actually

冑
detected, while at the same time enhancing the signal-to-
␳ˆ˙ s = − i
Ṅin␬
2
冉 1 1
关â + â†, ␳ˆ 兴 + ␬ â␳ˆ ↠− n̂␳ˆ − ␳ˆ n̂
2 2
冊 noise ratio in the trajectory. This is an indication of the quan-
tum Zeno effect in the detection of itinerant quanta, which
we now want to study in more quantitative details.
− 2⌫†n̂,关n̂, ␳ˆ 兴‡ − 冑4⌫关n̂␳ˆ + ␳ˆ n̂ − 2␳ˆ 具n̂典共t兲兴␰共t兲, 共3兲 We plot the rate of photon detection events Ṅdet versus the
while the classical noisy measurement signal that corre- rate of incoming photons Ṅin 关Fig. 3共a兲兴. The detection effi-
sponds to this internal dynamics is given by ciency ␩ is naturally defined as the ratio of detected vs in-

052115-3
HELMER et al. PHYSICAL REVIEW A 79, 052115 共2009兲

FIG. 2. 共Color online兲 Quantum jump trajectories illustrating

dispersive photon detection. The observable homodyne signal X̄共t兲
关red 共light gray兲 lines兴 and the corresponding signal mode occupa-
tion 具n̂典共t兲 关blue 共dark gray兲 lines兴, for two different values of the
measurement rate ⌫ / ␬ at a fixed input rate Ṅin. Photon detection
events are indicated as filled circles. The relative noise strength
⬀共⌫␶avg兲−1/2 is suppressed with increasing ⌫ / ␬, but the number of
photons actually detected also decreases due to the quantum Zeno
effect 共see main text兲. The size of the noise floor, the detector
threshold Xthr 关共green兲 dashed line兴, and the dark time are indicated.
Here and in the following plots ␬␶avg = 2.
FIG. 3. 共Color online兲 共a兲 Detector profile: Rate of detected vs
incoming photons, at ⌫ / ␬ = 0.6, for two different thresholds Xthr.
coming photons, obtained at small input rates Ṅin, Observe the dark count rate 共offset at Ṅin = 0兲, the detector effi-

冏 冏
ciency ␩ defined from the slope at Ṅin = 0, and the saturation for
dṄdet large Ṅin ⬃ ␶dark
−1
. 共b兲 Suppression of the signal photon number n̄
␩⬅ . 共6兲 inside the cavity as a function of measurement rate ⌫ / ␬ compared
dṄin ˙ =0
N in to perfect transmission 共n̄ideal兲. 共c兲 Detector efficiency ␩, obtained
from quantum trajectory simulations, as a function of ⌫ and Xthr
Figure 3共c兲 displays the efficiency ␩ as a function of ⌫ / ␬ 共3D inset兲, and comparison to the analytical results 共main plot兲. The
and Xthr. The statistics for this figure were obtained from blue 共lower line of兲 data points display ␩ for fixed Xthr = 0.5 共Xthr
extensive numerical simulations by generating O共104兲 trajec- = 0.5 cut in 3D inset兲. When maximizing the ␩ over Xthr for any
tories of length 102 / ␬ for seven different rates Ṅin at each given ⌫ / ␬, the red 共upper line of兲 data points are obtained 关labeled
value of ⌫ / ␬. Apparently, the detector efficiency ␩ is “max共␩兲”兴, agreeing well with the analytical asymptote 关green
strongly suppressed both for ⌫ / ␬ Ⰶ 1 共low signal-to-noise 共solid兲 thin line兴 at higher values of ⌫ / ␬. Small inset: Lin-log en-
ratio兲 and ⌫ / ␬ Ⰷ 1. largement of the region of the maximum efficiency of the main plot.
共Here, the x axis is scaled logarithmically while the y axis is scaled
linearly to obtain the optimum resolution of the peak structure.兲
B. Analytical results
To interpret these results, we now calculate the total trans- 具␦␻共t兲␦␻共0兲典 − 具␦␻典2 = g2n̄dete−␬det兩t兩/2 . 共8兲
mission probability through the signal mode, whose fre-
quency fluctuates due to the shot noise in the detection mode, To obtain an expression for the transmission probability, we
which is treated as classical noise. We start from the semi- write down the formal solution for ␣共t兲,
classical equation of motion for the complex field amplitude
␣共t兲 in the signal mode,
␣共t兲
冑␬ L ␣ L in
= 冕
−⬁
t

冋
dt⬘ exp − i 冕
t⬘
t
␬
册
␦␻共t⬙兲dt⬙ − 共t − t⬘兲 . 共9兲
2

冉
␣˙ 共t兲 = − i␦␻共t兲 −
␬
2
冊 冑
␣共t兲 +
␬ in
␣ .
2 L
共7兲 Note that the fluctuations ␦␻共t兲 themselves are non-
Gaussian. Still, the integral in the exponent is approximately
Gaussian for time intervals that fulfill ␬det兩t − t⬘兩 Ⰷ 1 due to
Here ␣Lin is the amplitude of the signal photon field entering the central limit theorem. These times yield the main contri-
the cavity from the left side and ␦␻共t兲 ⬅ gndet共t兲 is the fluc- bution under our assumption of a “fast detector,” ␬det Ⰷ ␬.
tuating frequency shift 共ndet Ⰷ 1兲. The correlator of the noise Thus, we can evaluate 具兩␣兩2典 using the formula 具exp关−iY兴典
is given by = exp关−i具Y典 − 21 VarY兴 for a Gaussian random variable Y and

052115-4
QUANTUM NONDEMOLITION PHOTON DETECTION IN… PHYSICAL REVIEW A 79, 052115 共2009兲

inserting Eq. 共8兲. From this, we obtain the average transmit- mode, i.e., we consider the case of dispersive coupling. The
ted intensity unperturbed Hamiltonian Ĥ0 关redefined to absorb the vacuum
␬ energy of the harmonic oscillators and using rotating wave
具兩aRout兩2典 = 具兩␣兩2典 = 具T 典兩␣Lin兩2 共10兲 approximation 共RWA兲兴 thus reads
2
and the average transmission probability ␻q
Ĥ0 ⬅ ប ␴ˆ z + ប␻␣â␣† â␣ + ប␻␤â␤† â␤ + g0,␣共␴ˆ +â␣ + â␣† ␴ˆ −兲.

冉 冊
2
⌫ −1
具T 典 = 1 + 4 . 共11兲 This contains the resonant coupling of the qubit to the
␬
signal mode 共mode index ␣兲, while the coupling to the de-
Before we can correlate the suppression of the transmission tection mode 共index ␤兲 will be considered perturbatively.
with the reduction of the detector efficiency ␩ in the limit of Now imagine the incoming signal photon being on resonance
⌫ / ␬ Ⰷ 1, one more consideration is necessary. In this limit, with one of the states of the coupled system 共qubit/signal
any photon that has entered the cavity will almost certainly mode兲, e.g., 兩␣−典 ⬅ 1 / 冑2共兩0 , ↑典 + 兩1 , ↓典兲 丢 兩n␤典. This state is an
be detected. Once detected, the photon loses the coherence eigenstate of the qubit-signal mode system which is only
with the incoming beam, which is needed for perfect trans- weakly 共dispersively兲 coupled to the detector mode. Pro-
mission on resonance in the ideal coherent case. As a conse- vided we consider only situations with at most one signal
quence, it acquires an equal probability to leave the cavity photon inside the signal mode 共␣兲 and n␤ photons in the
through the left or the right port. This means that, on aver- detection 共␤兲 mode 共as was the case in the preceding discus-
age, the number of detected photons is twice the number of sion兲, we can restrict ourselves to the subspace 兵兩0 , ↓典
transmitted photons. The expected relation is thus ␩ = 2具T 典, 丢 兩n␤典 , 兩␣−典其. This set of states will effectively represent the
which is indeed observed nicely when comparing to the nu- two lowest states of the signal mode in our detection scheme
merical data 关Fig. 3共c兲兴. 共photon absent or present兲.
The reduction of detector efficiency at ⌫ / ␬ Ⰷ 1 thus has The state 兩␣−典 has energy ␻兩␣−典 = ␻␣ + g0,␣ + n␤␻␤. Reso-
found its explanation in the quantum Zeno effect: many pho- nantly irradiating with ␻兩␣−典 will populate the corresponding
tons remain undetected because they are reflected due to de- state 兩␣−典 only and, being an eigenstate, will not induce any
tector back action. As low values of ⌫ / ␬ are also unfavor- internal dynamics.
able, due to a bad signal-to-noise ratio, the maximum Now we consider populating the ␤ mode with n␤ photons
efficiency is found near the intermediate value ⌫ / ␬ = 1 / 4 to dispersively detect the difference between the states
共see Fig. 3兲. 兩vac, n␤典 ⬅ 兩n␣ = 0 , ↓ , n␤典 and 兩␣−典 and thus the fact if a
␣-mode photon has entered the cavity or not. So, we con-
sider the coupling to the ␤ mode via the perturbation Hamil-
C. Possible realization in superconducting circuit quantum tonian
electrodynamics setups
Cavity QED setups in superconducting circuits Ĥint = g0,␤共␴ˆ +â␤ + â␤† ␴ˆ −兲. 共12兲
关12,20–22兴 have been used to implement ideas of quantum
optics on the chip and are considered a promising candidate The idea is that the second-order energy shift of the tran-
for scalable, fault-tolerant quantum computing 关23兴. While sition frequency between states 兩vac, n␤典 and 兩␣−典 will con-
proposals for generating nonclassical photon states exist or tain a term ⬀n␤ which can be used to read off the effective
have been implemented 关4,21,24,25兴, the on-chip single-shot coupling g between the hybrid system of ␣ mode 共qubit兲 and
detection of itinerant photons is still missing. the ␤ mode. We find that the system can be described by the
Building on recent experiments that demonstrated disper- effective Hamiltonian analogous to Eq. 共1兲,
sive qubit detection 关3兴 and measurements of photon statis-
tics 关4兴, one could employ the superconducting qubit to in- Heff = ប␻sâs†âs + ប␻␤â␤† â␤ + gâs†âsâ␤† â␤ , 共13兲
duce a nonlinear coupling between two modes of the
microwave transmission line resonator 共or coupling two with a renormalized transition frequency of our hybrid qubit/
cavities 关37兴兲, thus creating a dispersive photon detector of cavity system given by
the type discussed here.
These experiments realize a Jaynes-Cummings coupling 兩g0,␤兩2
␻s ⬅ ␻␣ + g0,␣ −
between qubit and resonator of up to 2␲ ⫻ 100 MHz, reso- 2共␻␤ − ␻␣兲
nators with frequencies of about 2␲ ⫻ 5 GHz, and a large
spread of resonator decay rates ␬ between 1 and 100 MHz. and an effective cross-Kerr coupling of

冋 册
To make the example more concrete we suggest to employ
the qubit as a nonlinear coupler between two modes of a 兩g0,␤兩2 g0,␣ 兩g0,␤兩2
g⬅− 1+ ⬇− . 共14兲
transmission line cavity. 共 ␻ ␤ − ␻ ␣兲 共 ␻ ␤ − ␻ ␣兲 共 ␻ ␤ − ␻ ␣兲
Imagine the qubit being tuned into resonance with the
signal mode, such that their resonances hybridize, as de- Note that in the newly introduced system with subscript s
scribed by the Jaynes-Cummings model. The detector mode 共our hybrid system兲 only one excitation is possible such that
is assumed to be far detuned from the qubit and the signal 兩ns = 1典 ⬅ 兩␣−典.

052115-5
HELMER et al. PHYSICAL REVIEW A 79, 052115 共2009兲

In order to evaluate the result numerically for different ible light. This makes measurement much harder due to the
modes, it is appropriate to use Eq. 共5兲 and the following extremely short time spans involved compared to micro-
relations for the dependence of the parameters on the mode waves. However, if we assume the same coupling paradigm
index: as before, we have to plug in the numbers typically found in
present-day experiments 关39–42兴. Cavity resonance frequen-
g0,j = 冑 jg0 , cies are, e.g., ␻0 = 2␲ ⫻ 0.3⫻ 106 GHz, the fundamental
Jaynes-Cummings coupling of an exciton to the cavity mode
␻ j = j ␻0 , is g0 ⬇ 2␲ ⫻ 25 GHz, while present-day cavity decay rates
also range in the O共2␲ ⫻ 30 GHz兲 region. If the cavity qual-
␬ j = j ␬0 , ity factors could be improved by about a factor of 100 and
the coupling of the cavity to the pump could be improved by
which lead to the following expression for the coupling and integrating the pump source on the chip and connecting it
for the dimensionless measurement rate: with a waveguide to the cavity 共the technology for both ex-
兩g0兩2 ␤ ists兲, then we can reach optimal detection efficiencies ⌫ / ␬
g=− , 共15兲 ⬇ 1 / 4 with O共104兲 photons in the detection mode. We there-
␻0 ␤ − ␣ fore conclude that our scheme would also be applicable in

冉 冊
the field of photonic crystals and self-assembled quantum
⌫ 兩g兩2具n̂␤典 兩g0兩4 1 ␤ 2
dots once certain design improvements have been realized in
= = 具n̂␤典 2 2 . 共16兲
␬ ␣ 4 ␬ ␣␬ ␤ 4␻0␬0 ␣␤ ␤ − ␣ those systems. Another experiment in which essentially the
same physics could be observed is the detection of single
Assuming g0 = 2␲ ⫻ 50 MHz, ␬0 = 10 MHz, ␻0 = 2␲ photons in a microwave cavity by employing the dispersive
⫻ 6 GHz, and 共␣ , ␤兲 = 共3 , 5兲 we observe that the optimal interaction with a stream of Rydberg atoms 关8兴.
measurement rate of ⌫ / ␬ = 1 / 4 is reached at a measurement
mode photon number of 具n̂␤典 ⬇ 35, well below the critical V. CONCLUSIONS
photon number, where the dispersive approximation starts to
break down as discussed, e.g., in 关38兴. Note that the dropped In this paper we have analyzed a rather generic scheme
term in Eq. 共14兲 amounts to a correction of only about 2% in for the detection of itinerant photons in a QND measurement
the needed photon number. process, employing quantum trajectory simulations. One par-
Simply adjusting the decay rate of the cavity or the qubit ticularly important and feasible implementation would be in
resonator coupling will enable the experimentalist to observe superconducting circuit QED, where this scheme would en-
all the features discussed in this paper as the full range of able the on-chip detection of single microwave photons. We
measurement rates are available, starting from the ideal de- have shown how the quantum Zeno effect enters the detec-
tection limit ⌫ / ␬ = 1 / 4 to the quantum Zeno limit ⌫ / ␬ Ⰷ 1. tion efficiency, a result that will be relevant to many other
The detector efficiency, although limited by the quantum situations, such as the detection of electrons tunneling
Zeno effect as shown before, can then reach values of about through a quantum dot by current passing through a nearby
30% even without considering more elaborate detector and quantum point contact 关43兴, the detection of itinerant
signal analysis schemes. phonons entering a micromechanical cantilever or membrane
We note that in an alternative scheme, both the signal and 共e.g., in an optomechanical setup 关44兴兲, and other similar
the detection modes can be dispersively coupled to the qubit, settings in mesoscopic physics, quantum optics, and atomic
although generically this would lead to a weaker overall cou- physics.
pling g. Finally, we remark that having control over the in-
dividual mode frequencies involved 共e.g., in a setup with two ACKNOWLEDGMENTS
cavities instead of one兲 would be advantageous for two rea-
sons: first, it could avoid spurious higher-order processes in We thank S. M. Girvin, J. M. Gambetta, A. A. Houck, M.
which the signal mode is contaminated by the decay of pho- Blencowe, A. Blais, J.-M. Raimond, and Dirk Englund for
tons from the detection mode by making those processes discussions. Support from the SFB 631, NIM, and the
strongly off resonant; second, it would allow to tune the Emmy-Noether program 共F.M.兲 of the DFG and EuroSQIP
signal and detection modes relatively close to each other, are gratefully acknowledged. E. S. thanks the Ikerbasque
thereby enhancing the coupling. Foundation, the EU EuroSQIP project, and the UPV-EHU
Grant GIU07/40.
D. Possible realization using self-assembled quantum dots and
photonic crystals APPENDIX: DERIVATION OF THE ADIABATICALLY
ELIMINATED STOCHASTIC MASTER EQUATION
Photonic crystal cavities have been proven to couple to
excitons in self-assembled quantum dots inside the cavity. 1. Unraveling the master equation
The system is in principle analogous to the circuit QED sys-
a. Motivation
tem discussed above 共in the sense that the dot coupling to the
cavity is well described by a Jaynes-Cummings Hamil- The goal of this procedure is to obtain a stochastic master
tonian兲, however, the crucial difference is that photonic crys- equation which models the system including the measure-
tals and excitons operate at frequencies corresponding to vis- ment back action and the corresponding noisy measurement

052115-6
QUANTUM NONDEMOLITION PHOTON DETECTION IN… PHYSICAL REVIEW A 79, 052115 共2009兲

signal in a consistent way starting from Eq. 共2兲 which reads Ĥint = i 兺 gd共␻n兲共b̂d,n
†
â1 − b̂d,nâ†1兲. 共A3兲

冋
␳˙ˆ = − i
␬1
2
共␣â†1 + ␣ⴱâ1兲 + gâ†1â1â†0â0, ␳ˆ 册 n

We employ the usual continuum and Markovian limit and

冉 1 1
+ ␬ â0␳ˆ â†0 − â†0â0␳ˆ − ␳ˆ â†0â0
2 2
冊 define the detector mode decay rate ␬det ⬅ ␲D共␻1兲兩gd共␻1兲兩2,
where D is the density of states. 关Note that this assumption
cannot be easily relaxed for the purpose of deriving a sto-

冉 1 1
+ ␬1 â1␳ˆ â†1 − â†1â1␳ˆ − ␳ˆ â†1â1
2 2
冊 chastic master equation. It guarantees that the projective
measurement of the bath will always disentangle system and
bath which is not true for general non-Markovian baths兴.
冑
− i Ṅin/2␬关â0 + â†0, ␳ˆ 兴. 共A1兲 Details have been recently discussed in 关45兴. The operator of
the measured signal is then
This way, we will gain insights beyond the ensemble-average
X̂共t兲 = 兺 关b̂d,n 共t兲 + b̂d,n共t兲兴 = 冑␬det关â1共t兲 + â†1共t兲兴
description of the usual master equation which is crucial for †
the analysis the detection efficiency. n

b. Remark on stochastic master equations + 兺 关b̂d,n

†
共t0兲ei␻n共t−t0兲 + b̂d,n共t0兲e−i␻n共t−t0兲兴,
The master equation in general is an ensemble-average n
description of a quantum system. It can be thought of as
arising from averaging a stochastic unraveled master equa- where the last two lines represent the inhomogeneous and
tion. homogeneous parts of the solution of the equations of motion
Note that there is, without a physical concept of the pro- for the Heisenberg operators b̂d,n, respectively. Using
cesses involved, no unique way of obtaining a stochastic 具b̂n共t0兲典 = 具b̂†n共t0兲典 = 0, the average signal can be expressed as
master equation from the averaged master equation
关33,34,45兴, a process which is known as unraveling. How- 具X̂共t兲典 = 冑␬det具â共t兲 + â†1共t兲典.
ever, once we consider the physical details of the measure-
ment, quantum mechanics allows us to unravel the master Now the idea is to let this interaction Hamiltonian Eq. 共A3兲
equation Eq. 共A10兲 in such a way that we obtain 共i兲 a clas- act for a small time ⌬t before projecting the bath onto an
sical measurement signal as produced by homodyne detec- eigenstate. Every time the bath is projected, the previously
tion of the light field leaking out the detector mode and 共ii兲 generated entanglement between bath and system ensures
simultaneously a master equation conditioned on this mea- that this will also have a 共slight兲 effect on the system state.
surement signal. This enables us to self-consistently simulate By this mechanism it becomes clear that the random choice
the measured signal and the quantum dynamics of our sys- of an eigenvalue of the bath will in the continuum limit give
tem which gave rise to exactly this signal. rise to a stochastic term in the measured signal 共noise兲 as
c. Outline of calculation well as in the system dynamics 共back action兲. This random
choice of eigenvalue/eigenstate of X̂ happens according to
We will observe that pure number states of the signal
the probability distribution given by
mode are attractor solutions. As a result of the measurement,
the signal mode’s state is stochastically forced toward a pure
P共X兲 = 具X兩␳ˆ 共t + ⌬t兲兩X典
number state on a time scale that corresponds to the time that
we need to extract the information about this number state
and projects the density matrix onto the eigenstate such that
from the measured signal. The attractor nature of the number
states also becomes apparent from the fact that the stochastic
兩X典具X兩␳ˆ 共t + ⌬t兲兩X典具X兩
term in the master equation becomes strictly zero as can be ␳ˆ 哫 . 共A4兲
seen from Eq. 共3兲. 具X兩␳ˆ 共t + ⌬t兲兩X典
In the following derivation, we follow Refs. 关18,33–36兴.
Note that the choice of unraveling formally corresponds to In order to simplify the notation, we introduce global bath
projecting the field leaking out of the detector mode into the
1
operators defined as B̂共t兲 ⬅ 冑2␲D共 ␻兲
e+i␻1t兺nb̂d,n共t兲. We are in-
1
free space field modes 共which we also call the measurement terested in the evolution of the operator B̂共t兲 over the time
bath兲 onto the correct basis states. In this case we project scale ⌬t which is much shorter than the internal time scale of
onto the eigenstates of the field quadrature we are interested
in to extract the phase shift imposed by a signal photon in the the intrinsic dynamics of B̂共t兲. Thus we can define an opera-
cavity. We start by writing the measurement bath as an infi- tor dB̂共t兲 as
nite set of harmonic oscillators 共e.g., the modes of a trans-
mission line兲,
dB̂共t兲 ⬅
1
冑⌬t 冕 t+⌬t
B̂共␶兲d␶ .
Ĥbath = 兺
t
␻d,nb̂d,n
†
b̂d,n , 共A2兲
n
To leading order in ⌬t, the evolution of the density matrix
interacting with the detector mode by the Hamiltonian will then have the following form:

052115-7
HELMER et al. PHYSICAL REVIEW A 79, 052115 共2009兲

␳ˆ 共t + ⌬t兲 = ␳ˆ 共t兲 丢 ␳bath共t兲 + 冑␬det⌬t关dB̂â†1 + â1dB̂†, ␳ˆ 共t兲 be done by applying the displacement operator D̂共−␣兲
= exp关−␣â†1 + ␣쐓a1兴 to Eq. 共A1兲. As a result, we have elimi-
丢 ␳ˆ bath共t兲兴 + O共⌬t兲. nated the driving term for the detector mode and obtain as
Note that we have already worked out all O共⌬t兲 terms which the new master equation
constitute the deterministic evolution so we can focus on the
O共冑⌬t兲 terms. If one does so and keeps all the terms up to ␳˙ˆ = − ig关â†1â1â†0â0, ␳ˆ 兴 − ig关共␣â†1 + ␣ⴱâ1兲â†0â0, ␳ˆ 兴
order O共冑⌬t兲, one realizes that the state after measurement
and projection will have the form 冉 1
2
1
+ ␬ â0␳ˆ â†0 − â†0â0␳ˆ − ␳ˆ â†0â0
2
冊
⌬␳ˆ 共t + ⌬t兲 =
兩X典具X兩␳ˆ 共t + ⌬t兲兩X典具X兩
具X兩␳ˆ 共t + ⌬t兲兩X典
− ␳ˆ 共t兲 冉 1 1
+ ␬1 â1␳ˆ â†1 − â†1â1␳ˆ − ␳ˆ â†1â1
2 2
冊
= 冑⌬tX关â1␳ˆ 共t兲 + ␳ˆ 共t兲â†1 − 具â1 + â†1典共t兲␳ˆ 共t兲兴 + O共⌬t兲. 冑
− i Ṅin/2␬关â0 + â†0, ␳ˆ 兴. 共A8兲
共A5兲
Note that the new vacuum state that the detector mode re-
Similarly we find for the signal X, assuming zero tempera- laxes to is actually the coherent state 兩␣典. The first two terms
ture, that the bath state at time t is a Gaussian centered at capture the fluctuations in the ancilla excitation while we
X = 0 and with width 1. At time t + ⌬t, after interaction and have eliminated the frequency renormalization term
projection, the bath will still be in a Gaussian state, but cen- −i兩␣兩2g关â†0â0 , ␳ˆ 兴 by absorbing it into the interaction picture.
tered around 冑␬det具â1 + â†1典⌬t. Thus we can identify the func- The adiabatic elimination 关18,33–35兴 is a well-controlled
tional form of the noisy signal. It reads in differential form approximation and can be expressed as an expansion of the
density matrix in a small parameter. Introducing this small
X共t兲 = 冑␬det具a1 + a†1典共t兲 + ␰共t兲, 共A6兲 parameter ⑀ Ⰶ 1 such that ␬ / ␬det = O共⑀兲 and g兩␣兩 / ␬det = O共⑀兲
where ␰共t兲 is white noise, with 具␰共t兲典 = 0 and 具␰共t兲␰共t⬘兲典 are satisfied, we can formally expand the density matrix in
= ␦共t − t⬘兲. Writing Eq. 共A5兲 in differential form and inserting orders of ⑀ as follows:
Eq. 共A6兲 and again keeping terms up to O共⌬t兲, we find the ␳ˆ = ␳ˆ s共00兲 丢 ␴ˆ + ␳ˆ s共10兲 丢 â†1␴ˆ + ␳ˆ s共01兲 丢 ␴ˆ â1 + ␳ˆ s共11兲 丢 â†1␴ˆ â1
stochastic contribution ␳ˆ˙ st共t兲 of the master equation that de-
scribes the measurement back action given we have mea- + ␳ˆ s共20兲 丢 â†2
1 ␴ˆ + ␳ˆ s共02兲 丢 ␴ˆ â21 + O共⑀3兲. 共A9兲
sured a certain signal trajectory X共t兲 as Up to now, we have just rewritten the density matrix in a
form that makes the different orders of ⑀ as well as off-
␳˙ˆ st共t兲 = 冑␬det关â1␳ˆ 共t兲 + ␳ˆ 共t兲â†1 − 具â1 + â†1典共t兲␳ˆ 共t兲兴␰共t兲. diagonal terms and diagonal terms apparent. ␴ˆ ⬅ 兩vac典具vac兩
Together with the deterministic evolution of Eq. 共A1兲, the denotes the displaced coherent state of the detector mode
unraveled master equation reads which is a vacuum state. Using this decomposition, one can
straightforwardly realize that taking the trace over the detec-
␳ˆ˙ = − ig关â†1â1â†0â0, ␳ˆ 兴 − ig关共␣ⴱâ†1 + ␣â1兲â†0â0, ␳ˆ 兴 tor mode results in the following form for the signal mode

冉 1
2
1
+ ␬ â0␳ˆ â†0 − â†0â0␳ˆ − ␳ˆ â†0â0
2
冊 density matrix:
␳ˆ s = ␳ˆ s共00兲 + ␳ˆ s共11兲 .

冉 1
2
1
冊 冑
+ ␬1 â1␳ˆ â†1 − â†1â1␳ˆ − ␳ˆ â†1â1 − i Ṅin/2␬关â0 + â†0, ␳ˆ 兴
2
We now evaluate Eq. 共A8兲 term by term using Eq. 共A9兲,
关â1 , â†1兴 = 1, and the fact that the coherent state ␴ˆ is a vacuum
state. After lengthy but simple manipulation of the expres-
+ 冑␬关â1␳ˆ 共t兲 + ␳ˆ 共t兲â†1 − 具â1 + â†1典共t兲␳ˆ 共t兲兴␰共t兲. 共A7兲 sions we can eliminate all detector mode operators from the
master equation. Key steps in this calculation are 共a兲 replac-
The last steps of this procedure can be found in more detail
ing the off-diagonal terms in the signal mode density matrix
in 关33兴. Note that the stochastic term is not of Lindblad form
but has nonlinear superoperator form 关33兴. by their steady-state solution, i.e., setting ␳ˆ˙ s共10兲 = ␳ˆ˙ s共01兲 = ␳ˆ˙ s共02兲
= ␳˙ˆ s共20兲 = 0 and 共b兲 disregarding off-diagonal terms in the de-
2. Adiabatic elimination of the detector mode tector mode density matrix such as, e.g., â†1â†1â†1 coh␳ˆ det, which
are further away from the diagonal than the expansion range
The key assumption for the adiabatic elimination proce- of the ancilla state. The result 共see 关18兴兲 reads

冑
dure to be correct and appropriate is that the dynamics of the
ancilla system, in our case the detector mode, is much faster
than the dynamics of the system, in our case the signal mode.
In this limit ␬det / ␬ Ⰷ 1, the detector mode will relax to its
␳˙ˆ = − i
Ṅin␬
2
冉 1 1
关â + â†, ␳ˆ 兴 + ␬ â␳ˆ ↠− n̂␳ˆ − ␳ˆ n̂
2 2
冊
displaced driven state on a time scale that is much faster than g 2兩 ␣ 兩 2
the intrinsic time scales in the signal mode. −2 关n̂,关n̂, ␳ˆ 兴兴. 共A10兲
␬det
The first step to be taken is to transform the master equa-
tion such that the phase-space origin for the detector mode is From now on we will introduce the measurement rate ⌫
centered at its steady state under the coherent drive. This can ⬅ g2兩␣兩2 / ␬det which is the coefficient of the measurement

052115-8
QUANTUM NONDEMOLITION PHOTON DETECTION IN… PHYSICAL REVIEW A 79, 052115 共2009兲

induced diffusion term. This simplified master equation 关Eq. the adiabatic elimination calculation we also find the very
共A10兲兴 for the signal mode alone later enables us to perform useful identity
detailed numerical studies due to the more manageable size
of the density matrix. Furthermore, only in the adiabatically
eliminated version of the master equation we will be able to
explicitly see that the chosen limit of ␬det Ⰷ ␬ exactly corre- 冑␬det具â1 + â†1典共t兲 = 2冑2⌫具â†0â0典共t兲, 共A12兲
sponds to measuring the photon number of the signal mode
by looking at the phase shift on the detection mode.
The last step of the calculation follows the lines of the
elimination of the detector mode operators from the deter- which allows us to write the measurement signal 共rescaled兲
ministic terms in the master equation. However, to adiabati- Eq. 共4兲 as
cally eliminate the detector mode operators from the stochas-
tic term, one more consideration is necessary. Simply
replacing the off-diagonal terms by their stationary solution
will not be enough as the stochastic term averages to zero.
Instead we have to compute the variance of the stochastic
term and integrate it over the time scale of the diagonal terms
X共t兲 ⬅ 具n̂典共t兲 +
1
4
冑 1
⌫
␰共t兲. 共A13兲

⌬t = ␬det
−1
. Doing so, we find that the resulting master equation
Eq. 共3兲 reads

␳ˆ s = − i 冑 Ṅin␬
2
冉 1 1
关â + â†, ␳ˆ 兴 + ␬ â␳ˆ ↠− n̂␳ˆ − ␳ˆ n̂
2 2
冊 Having obtained the final form of the adiabatically elimi-
nated stochastic master equation, we can turn to the detector
logic and the evaluation of the photon detection efficiency.
− 2⌫关n̂,关n̂, ␳ˆ 兴兴 − 冑4⌫关n̂␳ˆ + ␳ˆ n̂ − 2␳ˆ 具n̂典共t兲兴␰共t兲. This will be done by numerically integrating Eqs. 共3兲 and
共A13兲 and applying a suitable nonlinear filter to the signal
共A11兲
followed by a statistical analysis of the dependence of the
For the rest of the paper 共in the main text兲 we will drop the photon count rate on the physical parameters in the system,
subscript s which denotes the signal mode. In the course of especially ⌫ / ␬, as explained in the main text.

关1兴 V. B. Braginsky, Y. I. Vorontsov, and K. S. Thorne, Science Phys. Rev. A 41, 2295 共1990兲.
209, 547 共1980兲. 关14兴 K. Jacobs, P. Lougovski, and M. Blencowe, Phys. Rev. Lett.
关2兴 V. B. Braginsky and F. Y. Khalili, Quantum Measurement 98, 147201 共2007兲.
共Cambridge University Press, Cambridge, England, 1992兲. 关15兴 R. Rossi, Jr., M. Nemes, and A. de Magalhaes, Phys. Rev. A.
关3兴 A. Wallraff, D. I. Schuster, A. Blais, L. Frunzio, J. Majer, M. 77, 012107 共2008兲.
H. Devoret, S. M. Girvin, and R. J. Schoelkopf, Phys. Rev. 关16兴 J. A. Damborenea, I. L. Egusquiza, G. C. Hegerfeldt, and J. G.
Lett. 95, 060501 共2005兲. Muga, Phys. Rev. A 66, 052104 共2002兲.
关4兴 D. I. Schuster, A. A. Houck, J. A. Schreier, A. Wallraff, J. M. 关17兴 X.-B. Wang, J. Q. You, and F. Nori, Phys. Rev. A 77, 062339
Gambetta, A. Blais, L. Frunzio, J. Majer, B. Johnson, M. H. 共2008兲.
Devoret, S. M. Girvin, and R. J. Schoelkopf, Nature 共London兲 关18兴 D. H. Santamore, A. C. Doherty, and M. C. Cross, Phys. Rev.
445, 515 共2007兲. B 70, 144301 共2004兲.
关5兴 J. Gambetta, W. A. Braff, A. Wallraff, S. M. Girvin, and R. J. 关19兴 K. Jacobs, A. Jordan, and E. Irish, EPL 82, 18003 共2008兲.
Schoelkopf, Phys. Rev. A 76, 012325 共2007兲. 关20兴 A. Wallraff, D. I. Schuster, A. Blais, L. Frunzio, R.-S. Huang,
关6兴 J. Gambetta, A. Blais, M. Boissonneault, A. A. Houck, D. I. J. Majer, S. Kumar, S. M. Girvin, and R. J. Schoelkopf, Nature
Schuster, and S. M. Girvin, Phys. Rev. A 77, 012112 共2008兲. 共London兲 431, 162 共2004兲.
关7兴 J. I. Cirac, R. Blatt, A. S. Parkins, and P. Zoller, Phys. Rev. 关21兴 A. A. Houck, D. I. Schuster, J. M. Gambetta, J. A. Schreier, B.
Lett. 70, 762 共1993兲. R. Johnson, J. M. Chow, L. Frunzio, J. Majer, M. H. Devoret,
关8兴 S. Gleyzes, S. Kuhr, C. Guerlin, J. Bernu, S. Deléglise, U. S. M. Girvin and R. J. Schoelkopf, Nature 共London兲 449, 328
Busk Hoff, M. Brune, J.-M. Raimond, and S. Haroche, Nature 共2007兲.
共London兲 446, 297 共2007兲. 关22兴 J. Johansson, S. Saito, T. Meno, H. Nakano, M. Ueda, K.
关9兴 W. Nagourney, J. Sandberg, and H. Dehmelt, Phys. Rev. Lett. Semba, and H. Takayanagi, Phys. Rev. Lett. 96, 127006
56, 2797 共1986兲. 共2006兲.
关10兴 M. Porrati and S. Putterman, Phys. Rev. A 36, 929 共1987兲. 关23兴 F. Helmer, M. Mariantoni, A. G. Fowler, J. von Delft, E. Sol-
关11兴 S. Peil and G. Gabrielse, Phys. Rev. Lett. 83, 1287 共1999兲. ano, and F. Marquardt, EPL 85, 50007 共2009兲.
关12兴 B. Misra and E. C. G. Sudarshan, J. Math. Phys. 18, 756 关24兴 M. Mariantoni, M. Storcz, F. Wilhelm, W. Oliver, A. Emmert,
共1977兲. A. Marx, R. Gross, H. Christ, and E. Solano, e-print
关13兴 W. M. Itano, D. J. Heinzen, J. J. Bollinger, and D. J. Wineland, arXiv:cond-mat/0509737.

052115-9
HELMER et al. PHYSICAL REVIEW A 79, 052115 共2009兲

关25兴 F. Marquardt, Phys. Rev. B 76, 205416 共2007兲. 关38兴 J. Gambetta, A. Blais, D. I. Schuster, A. Wallraff, L. Frunzio,
关26兴 N. Imoto, H. A. Haus, and Y. Yamamoto, Phys. Rev. A 32, J. Majer, M. H. Devoret, S. M. Girvin, and R. J. Schoelkopf,
2287 共1985兲. Phys. Rev. A 74, 042318 共2006兲; e-print arXiv:cond-mat/
关27兴 P. Yeh, J. Opt. Soc. Am. B 3, 747 共1986兲. 0602322.
关28兴 H. A. Bachor, M. D. Levenson, D. F. Walls, S. H. Perlmutter, 关39兴 D. Englund, A. Majumdar, A. Faraon, M. Toishi, N. Stoltz, P.
and R. M. Shelby, Phys. Rev. A 38, 180 共1988兲.
Petroff, and J. Vuckovic, e-print arXiv:0902.2428.
关29兴 M. Mariantoni, F. Deppe, A. Marx, R. Gross, F. K. Wilhelm,
关40兴 D. Englund, I. Fushman, A. Faraon, and J. Vučković, Photon-
and E. Solano, Phys. Rev. B 78, 104508 共2008兲.
ics Nanostruct. Fundam. Appl. 7, 56 共2009兲.
关30兴 A. Barchielli, Phys. Rev. A 34, 1642 共1986兲.
关31兴 H. Carmichael, An Open Systems Approach to Quantum Op- 关41兴 M. Kaniber, A. Neumann, A. Laucht, M. F. Huck, M. Bichler,
tics, Springer Lecture Notes Monographs Series No. 18 M.-C. Amann, and J. J. Finley, New J. Phys. 11, 013031
共Springer-Verlag, Berlin, 1993兲. 共2009兲.
关32兴 D. F. Walls and G. J. Milburn, Quantum Optics 共Springer, New 关42兴 A. Laucht, F. Hofbauer, N. Hauke, J. Angele, S. Stobbe, M.
York, 1995兲. Kaniber, G. Böhm, P. Lodahl, M. Amann, and J. Finley, New
关33兴 H. M. Wiseman, Ph.D. thesis, University of Queensland, 1994. J. Phys. 11, 023034 共2009兲; e-print arXiv:0810.3010.
关34兴 H. M. Wiseman and G. J. Milburn, Phys. Rev. A 49, 1350 关43兴 E. Buks, R. Schuster, M. Heiblum, D. Mahalu, and V. Uman-
共1994兲. sky, Nature 共London兲 391, 871 共1998兲.
关35兴 A. C. Doherty and K. Jacobs, Phys. Rev. A 60, 2700 共1999兲. 关44兴 J. D. Thompson, B. M. Zwickl, A. M. Jayich, F. Marquadt, S.
关36兴 D. H. Santamore, Ph.D. thesis, California Institute of Technol- M. Girvin, and J. G. E. Harris, Nature 共London兲 452, 72
ogy, 2003. 共2008兲.
关37兴 A. Blais, J. Gambetta, C. Cheung, R. Schoelkopf, and S. M. 关45兴 H. M. Wiseman and J. M. Gambetta, Phys. Rev. Lett. 101,
Girvin 共unpublished兲. 140401 共2008兲.

052115-10