1: \documentclass[prl,twocolumn,showpacs,amsmath,amsfonts]{revtex4}
2: %\documentclass[aps,preprint,showpacs]{revtex4}
3: \usepackage{bm}
4: \usepackage{graphicx}
5:
6: \newcommand{\bra}[1]{\langle #1 | \,}
7: \newcommand{\ket}[1]{\, | #1 \rangle}
8: \newcommand{\braket}[2]{\langle #1 | #2 \rangle}
9: \newcommand{\la}{\lambda}
10: \newcommand{\om}{\omega}
11: \newcommand{\Om}{\Omega}
12: \newcommand{\ga}{\gamma}
13: \newcommand{\Ga}{\Gamma}
14: \newcommand{\de}{\delta}
15: \newcommand{\De}{\Delta}
16:
17:
18: \begin{document}
19:
20: \title{Scalable solid-state quantum processor using subradiant two-atom
21: states}
22:
23: \author{David Petrosyan}
24: \author{Gershon Kurizki}
25: %\homepage{http://www.weizmann.ac.il/chemphys/gershon/}
26: \affiliation{Department of Chemical Physics,
27: Weizmann Institute of Science, Rehovot 76100, Israel}
28:
29: \date{\today}
30:
31: \begin{abstract}
32: We propose a realization of a scalable, high-performance quantum processor
33: whose qubits are represented by the ground and subradiant states of effective
34: dimers formed by pairs of two-level systems coupled by resonant dipole-dipole
35: interaction. The dimers are implanted in low-temperature solid host material
36: at controllable nanoscale separations. The two-qubit entanglement either
37: relies on the coherent excitation exchange between the dimers or is mediated
38: by external laser fields.
39: \end{abstract}
40:
41: \pacs{03.67.Lx, 42.50.Fx}
42:
43: \maketitle
44:
45: The main stumbling blocks en route to the realization of useful quantum
46: computers, comprised of many qubits, are \cite{RevSteane}:
47: (i) {\em fidelity loss} due to decoherence, which grows with the amount
48: of single- and two-qubit operations and requires large redundancy for the
49: application of error-correction methods;
50: (ii) {\em scalability} of the quantum processor (QP), which restricts the
51: choice of candidate systems and gives preference to solid-state structures.
52: QP proposals and realizations have thus far predominantly involved optical
53: manipulations of atoms in ion traps \cite{CZ,MS,SKKLMMRTIWM}, high-Q cavities
54: \cite{PGCZ}, and optical lattices \cite{BCJD}. Yet, the decoherence caused by
55: radiative (spontaneous emission) and nonradiative processes, as well as
56: difficulties with the scalability, cast doubts on the suitability of these
57: schemes for truly large-scale quantum computation \cite{BKMPV}. Solid-state
58: QP realizations \cite{LDV,Kane,BHHFKWSF,LH} appear to be more promising,
59: both principally and technologically.
60:
61: Here we propose a combined optical/solid-state approach that can significantly
62: enhance the speed, fidelity and scalability of a QP. The crux of this approach
63: is the hitherto unexplored concept of a ``subradiant dimer'' (SD) qubit: two
64: similar two-level systems (atoms or quantum dots) that are separated by a few
65: nanometers and interact via the resonant dipole-dipole interaction (RDDI)
66: \cite{DDI}, thereby forming an effective ``dimer'', whose ground and
67: subradiant (``dark'') states serve as the qubit basis. {\em All the basic
68: ingredients of quantum computation} (state preparation, universal logic gates
69: and qubit readout) \cite{DiVincenzo} are shown to be realizable by high-speed
70: optical manipulations of these dimers with very small error probability,
71: due to strong inhibition of radiative decay. A scalable QP is envisioned in
72: a low-temperature solid host material doped with such dimers at controllable
73: nanoscale separations.
74:
75: \begin{figure}[t]
76: \centerline{\includegraphics[width=8.5cm]{dd_int.eps}}
77: \caption{(a) Two TLAs 1 and 2, separated by normalized distance $\zeta$,
78: interact via RDDI and exchange a single excitation.
79: (b) Energy level diagram of the resulting ``dimer'' states of the system.
80: \label{fig:dd_int}}
81: \end{figure}
82:
83: Let us recall the cooperative properties of two identical two-level atoms
84: (TLAs), 1 and 2, at fixed positions $\textbf{r}_1$ and $\textbf{r}_2$, whose
85: ground and excited states are labeled as $\ket{g_{1,2}}$ and $\ket{e_{1,2}}$,
86: respectively [Fig. \ref{fig:dd_int}(a)]. The effective (non-Hermitian)
87: Hamiltonian of the system can be cast in a form \cite{DDI}
88: \begin{equation}
89: H = H_{\text{A}}+V_{\text{RDDI}} \label{ham},
90: \end{equation}
91: where $H_{\text{A}} = \hbar (\om_{eg} - i \ga/2)
92: (\ket{e_1} \bra{e_1} + \ket{e_2} \bra{e_2})$ represents the atomic Hamiltonian,
93: with $\om_{eg}$ being the resonant frequency and $\ga$ the radiative decay rate
94: on the atomic transition $\ket{e}\to \ket{g}$, and
95: $V_{\text{RDDI}} = \hbar \left(\De - i \ga_{12}/2 \right)
96: (\ket{e_1g_2} \bra{g_1e_2} + \ket{g_1 e_2} \bra{e_1 g_2})$
97: describes the interatomic RDDI potential, whose real part $\De$ is equal to
98: the rate of coherent excitation exchange (hopping) between the atoms, and
99: the imaginary part $\ga_{12}$ is responsible for the cooperative radiative
100: decay of the system. Both $\De$ and $\ga_{12}$ are functions of the
101: normalized distance between the atoms $\zeta = q r_{12}$, with
102: $q = \om_{eg}/c$ and $r_{12} = |\textbf{r}_1 - \textbf{r}_2|$.
103: The diagonalization of Hamiltonian (\ref{ham}) yields the ``dimer''
104: eigenstates $\ket{G} = \ket{g_1 g_2}$,
105: $\ket{\pm} = \frac{1}{\sqrt{2}}(\ket{e_1 g_2} \pm \ket{g_1 e_2})$, and
106: $\ket{E} = \ket{e_1 e_2}$, with the energy eigenvalues
107: $\la_G = 0$, $\la_{\pm} = \om_{eg} \pm \De -i \Ga_{\pm}/2$, and
108: $\la_E = 2 \om_{eg} -i \Ga_E/2$, respectively [Fig. \ref{fig:dd_int}(b)].
109: At small separations $\zeta \ll 1$, the symmetric $\ket{+}$ and doubly-excited
110: $\ket{E}$ eigenstates are {\em superradiant}, having the corresponding decay
111: rates $\Ga_+ = \ga+\ga_{12} \approx \Ga_E = 2 \ga$, while the antisymmetric
112: eigenstate $\ket{-}$ is {\em subradiant}, with the decay rate
113: $\Ga_- = \ga - \ga_{12} \approx \ga \zeta^2/5 \ll \ga$
114: \cite{DDI}. The energy levels of states $\ket{\pm}$ are then
115: shifted from that of state $\ket{e}$ by $\pm \De$, with
116: $|\De| \approx 3 \ga/(4 \zeta^3) \gg \ga$.
117:
118: The coupling strength of a laser field $\mathcal{E}$, having frequency
119: $\om \sim \om_{eg}$ and wave vector $\textbf{k}$, with the dimer is expressed
120: by its Rabi frequencies, which are equal to $\pm \Om_{-}$ on the transitions
121: $\ket{G} \to \ket{-}$ and $\ket{-} \to \ket{E}$, respectively, and to
122: $\Om_{+}$ on the transitions $\ket{G} \to \ket{+}$ and $\ket{+} \to \ket{E}$,
123: where $\Om_{\pm} = 2^{-1/2} \Om [1 \pm e^{-i \textbf{k} \textbf{r}_{12}}]$ and
124: $\Om = \mu \mathcal{E}/\hbar$ is the Rabi frequency of the field for a single
125: isolated atom, with $\mu$ being the dipole matrix element for the atomic
126: transition $\ket{g} \to \ket{e}$. In the limit of small interatomic
127: separations, $\Om_+ \simeq 2^{1/2} \Om$ and
128: $\Om_- \simeq i 2^{-1/2} \Om \zeta \cos \phi$, where $\phi$ is the angle
129: between the vectors $\textbf{k}$ and $\textbf{r}_{12}$. Hence, $\Om_-$
130: identically vanishes if the propagation direction of the field is perpendicular
131: to the interatomic axis, $\textbf{k} \perp \textbf{r}_{12}$, while it is
132: maximized in the $\textbf{k} \parallel \textbf{r}_{12}$ configuration, for
133: $\zeta \ll 1$. In physical terms, the subradiant $\ket{G} \to \ket{-}$
134: transition exhibits a {\em quadrupolar} behaviour and dipole-moment
135: suppression, due to destructive interference of the two-atom interactions
136: with the field, as opposed to their constructive interference in the
137: superradiant $\ket{G} \to \ket{+}$ transition.
138:
139: Now we are in a position to introduce the concept of the ``subradiant dimer''
140: (SD) qubit. The two qubit states correspond to the ground $\ket{G}$ and
141: subradiant $\ket{-}$ states of the dimer. An arbitrary single-qubit operation
142: (rotation) can be performed by the laser field $\mathcal{E}_r$ with wave
143: vector $\textbf{k}_r \parallel \textbf{r}_{12}$ and frequency
144: $\om_r = \om_{eg} -\De$ that is resonant with the qubit transition
145: $\ket{G} \to \ket{-}$ [Fig. \ref{fig:dd_int}(b)]. During the qubit flip-time
146: $T_{\text{flip}} = \pi / (2 |\Om_-^{(r)}|)$, the probability of error
147: $P_-^{\text{sp}}$ due to spontaneous emission from the subradiant state
148: $\ket{-}$ has the upper bound
149: $P_-^{\text{sp}} \leq \Ga_- T_{\text{flip}}=\pi \ga \zeta/(5 \sqrt{2} \Om_r)$,
150: while the probability of error due to population transfer from the ground
151: state $\ket{G}$ to the superradiant state $\ket{+}$ satisfies
152: $P_+^{\text{tr}} \leq \Ga_+ |\Om_+^{(r)}|^2 T_{\text{flip}} /(2 \De)^2 =
153: 8\sqrt{2} \pi \Om_r \zeta^5/(9 \ga)$. As an example, for the parameters
154: $\zeta \simeq 0.02$ and $\Om_r/\ga \simeq 30$, the decay rate of the
155: antisymmetric state is $\Ga_- \approx 8 \times 10^{-5} \ga$ and the
156: error probabilities during the flip-time of a SD qubit are
157: $P_-^{\text{sp}} \simeq 3 \times 10^{-4}$ and
158: $P_+^{\text{tr}} \ll P_-^{\text{sp}}$, as compared to the corresponding
159: error probability for a single atom,
160: $P_{\text{atom}}^{\text{sp}} \leq \pi \ga/(2 |\Om_r|) \simeq 0.05$.
161: Such small errors of the SD qubit are amenable to error correction
162: \cite{RevSteane}.
163:
164: In order to {\em read-out} (measure) the state of the qubit, we may use a
165: modification of the electron-shelving technique \cite{QJumps}.
166: Let us apply for a time $T_{\text{rout}}$ a probe field
167: $\mathcal{E}_p$ at a frequency $\om_p = \om_{eg} + \De$ that is resonant with
168: the dimer transition $\ket{G} \to \ket{+}$. Since the Rabi frequency
169: on that transition is much larger than on the qubit transition
170: $\ket{G} \to \ket{-}$, from which the probe field is detuned by $2\De$,
171: the presence or absence of fluorescence from $\ket{+}$ would indicate
172: whether the qubit state is $\ket{G}$ or $\ket{-}$, respectively. However,
173: since the frequency $\om_p$ exactly matches that of the transition
174: $\ket{-} \to \ket{E}$ [Fig. \ref{fig:dd_int}(b)], the dimer in state
175: $\ket{-}$ can first be excited to $\ket{E}$ by absorbing a probe photon,
176: then decay to $\ket{+}$, subsequently producing the same fluorescence signal
177: as if it were initially in state $\ket{G}$. Therefore, for a reliable
178: measurement, the condition $\ga_{-+} T_{\text{rout}} < 1$ should be satisfied,
179: where $\ga_{-+} =|\Om_p|^2 \zeta^2/\ga$ is the rate of transition
180: $\ket{-} \to \ket{+}$. This leads to the condition
181: $\Om_p /\ga < \sqrt{2 \eta} /\zeta$, where $\eta <1$ is the detector
182: efficiency. With $\eta \simeq 0.3 $, $\Om_p /\ga \simeq 5$ and
183: $\zeta \simeq 0.02$, we obtain 98\% measurement reliability. If, however,
184: the probe laser is applied for a time $T_{\text{rout}} \geq \ga_{-+}^{-1}$,
185: it will {\em initialize} the state of the qubit to its ground state $\ket{G}$.
186:
187: \begin{figure}[t]
188: \centerline{\includegraphics[width=8.5cm]{cswp.eps}}
189: \caption{(a) Dimers $A$ and $B$ are separated by normalized distance
190: $\xi > \zeta$. An external ac Stark field can switch on and off the RDDI
191: between the dimers.
192: (b) When the qubit transitions of dimers $A$ and $B$ are brought to resonance,
193: they start swapping a single excitation.
194: \label{fig:cswp}}
195: \end{figure}
196:
197: We next consider the RDDI-induced entanglement between two neighboring
198: dimers of size $\zeta$, labeled as $A$ and $B$, whose normalized
199: separation $\xi = q r_{AB}$ satisfies the condition $\zeta < \xi \ll 1$
200: [Fig. \ref{fig:cswp}(a)]. The rate of coherent excitation exchange between
201: the dimers on the qubit transitions $\ket{G}_{A,B} \to \ket{-}_{A,B}$
202: is given by
203: $\De_{AB}^{(-)} \simeq 3 \Ga_- /(4 \xi^3) = 3 \ga \zeta^2/(20 \xi^3)$.
204: If the difference in the qubit transition frequencies of the two dimers
205: exceeds $\De_{AB}^{(-)}$ (as is usually the case in a solid host),
206: then their excitation exchange is effectively switched off. To switch
207: their interaction on, one can apply an off-resonant, intense,
208: standing-wave field, such that dimers $A$ and $B$ are exposed to different
209: field amplitudes and therefore undergo different ac Stark shifts
210: [Fig. \ref{fig:cswp}(a)]. The standing-wave pattern is then shifted along the
211: $A-B$ axis until the qubit
212: transitions of the two dimers become resonant. Then, during the time
213: $T_{\textsc{swap}} = \pi /(2 \De_{AB}^{(-)})$, the \textsc{swap}
214: transformation takes place,
215: $\ket{-}_{A(B)}\ket{G}_{B(A)} \to -i \ket{G}_{A(B)}\ket{-}_{B(A)}$, while
216: other initial states of the two qubits, $\ket{-}_{A}\ket{-}_{B}$ and
217: $\ket{G}_{A}\ket{G}_{B}$, remain unaffected [Fig. \ref{fig:cswp}(b)].
218: In the same way, one can realize the {\em square-root of swap}
219: ($\sqrt{\textsc{swap}}$) gate between two qubits. By switching on the
220: interaction for time $T_{\sqrt{\textsc{swap}}} = \pi /(4 \De_{AB}^{(-)})$,
221: one fully entangles the two qubits, attaining an equally-weighted
222: superposition of \textsc{swap} and no-\textsc{swap},
223: \begin{eqnarray}
224: & & \ket{-}_{A(B)}\ket{G}_{B(A)} \to \nonumber \\
225: & & \;\;\; \;\;\;\; \frac{1}{\sqrt{2}} [\ket{-}_{A(B)}\ket{G}_{B(A)} -
226: i \ket{G}_{A(B)}\ket{-}_{B(A)}]
227: \label{sqrt_swap} .
228: \end{eqnarray}
229: The main source of error in this scheme is the cooperative spontaneous decay
230: of the excited states of the qubits,
231: $P_{\text{swap}}^{\text{sp}} \leq 2 \Ga_- T_{\textsc{swap}} = 4 \pi \xi^3/3$.
232: With inter-dimer separation $\xi \simeq 0.1 \gg \zeta$, this leads to
233: $P_{\text{swap}}^{\text{sp}} \leq 4\times 10^{-3}$, which can be taken care of
234: by error correction schemes \cite{RevSteane}.
235:
236: \begin{figure}[t]
237: \centerline{\includegraphics[width=8.5cm]{cphs_qpr.eps}}
238: \caption{Inset: Schematic drawing of the proposed QP and the geometry for the
239: \textsc{cphase} gate: The interatomic axis of each dimer (SD qubit) is
240: perpendicular to the inter-dimer axis,
241: $\textbf{r}_{12}^{A,B} \perp \textbf{r}_{AB}$.
242: Each SD qubit can be separately addressed by a laser field with
243: $\textbf{k}_r \parallel \textbf{r}_{12}$.
244: Two-qubit interaction is mediated by a coupling field with
245: $\textbf{k}_c \parallel \textbf{r}_{AB}$.
246: (a) Internal level structure of two dimers.
247: (b) Eigenstates of the combined system of two dimers.
248: \label{fig:cphs}}
249: \end{figure}
250:
251: A fast {\em controlled-phase} (\textsc{cphase}) logic gate between two closely
252: spaced SD qubits can be realized by a laser field acting on the auxiliary
253: transition $\ket{G} \to \ket{+}$, and thereby populating the state $\ket{+}$.
254: This will induce the RDDI between the dimers, causing an excitation
255: exchange between state $\ket{+}_A$ of dimer $A$ and state
256: $\ket{G}_B$ of dimer $B$ and vice versa [Fig. \ref{fig:cphs}(a)]. From
257: the above analysis, the rate of this exchange is given by
258: $\De_{AB}^{(+)} \simeq 3 \Ga_+ /(4 \xi^3) = 3 \ga /(2 \xi^3)$, which is much
259: larger than $\De_{AB}^{(-)}$, since $\Ga_+/\Ga_- \simeq 10/\zeta^2 \gg 1$.
260: Therefore, during a time interval that is small compared to
261: $|\De_{AB}^{(-)}|^{-1}$, we can neglect the RDDI between the dimers on the
262: qubit transitions $\ket{G}_{A,B} \to \ket{-}_{A,B}$ in comparison to that on
263: the auxiliary transitions $\ket{G}_{A,B} \to \ket{+}_{A,B}$. To the same
264: accuracy, the eigenstates of the two-dimer system are $\ket{G_A G_B}$,
265: $\ket{M} = \frac{1}{\sqrt{2}} (\ket{+_A G_B} - \ket{G_A +_B})$,
266: $\ket{P} = \frac{1}{\sqrt{2}} (\ket{+_A G_B} + \ket{G_A +_B})$, and
267: $\ket{+_A +_B}$. The singly excited states $\ket{M}$ and $\ket{P}$, having
268: the decay rates $\Ga_M \simeq \Ga_+ \xi^2/5$ and $\Ga_P \simeq 2 \Ga_+$,
269: correspond, respectively, to the antisymmetric and symmetric combinations
270: of the superradiant states of the two dimers [Fig. \ref{fig:cphs}(b)].
271: To perform the \textsc{cphase} gate, we irradiate the system with
272: the coupling field $\mathcal{E}_c$ having the wave vector
273: $\textbf{k}_c \parallel \textbf{r}_{AB}$ and frequency
274: $\om_c = \om_{eg} + \De - \De_{AB}^{(+)}$ that is resonant with the transition
275: $\ket{G_A G_B} \to \ket{M}$ (Fig. \ref{fig:cphs}-inset). The Rabi frequencies
276: of this field on the transitions $\ket{G_A G_B} \to \ket{M}$ and
277: $\ket{G_A G_B} \to \ket{P}$ are equal, respectively, to
278: $\Om_M^{(c)} = \Om_c \xi$ and $\Om_P^{(c)} = 2 \Om_c$.
279: Since $\textbf{k}_c \perp \textbf{r}_{12}^{A,B}$, this field does not
280: couple to the qubit transitions of the dimers. During the time
281: $T_{\textsc{cphase}} = \pi/\Om_M^{(c)}$, the system of two dimers, being
282: initially in the state $\ket{G_A G_B}$, undergoes the Rabi cycle from
283: $\ket{G_A G_B}$ to $\ket{M}$ and back, resulting in the $\pi$-phase-shift
284: \begin{equation}
285: \ket{G_A G_B} \to - \ket{G_A G_B}. \label{cphase}
286: \end{equation}
287: This transformation corresponds to the \textsc{cphase} logic gate,
288: since all other initial states, such as $\ket{-}_{A}\ket{-}_{B}$ and
289: $\ket{-}_{A(B)}\ket{G}_{B(A)}$, remain unaffected, due to the
290: fact that the RDDI between the dimers is present only if their combined
291: state is either $\ket{G}_{A}\ket{+}_{B}$ or $\ket{+}_{A}\ket{G}_{B}$,
292: otherwise the coupling field is off-resonant with the system.
293: The error during this gate operation is due to the spontaneous emission from
294: the state $\ket{M}$, with the probability
295: $P_{\textsc{cphase}}^{\text{sp}} \leq \Ga_M T_{\textsc{cphase}}=
296: 2 \pi \ga \xi/5 \Om_c$, as well as due to population transfer from the
297: state $\ket{G_A G_B}$ to the state $\ket{P}$, with the probability
298: $P_{\textsc{cphase}}^{\text{tr}} \leq \Ga_P |\Om_P^{(c)}|^2
299: T_{\textsc{cphase}} /(2 \De_{AB}^{(+)})^2 =
300: 16 \pi \Om_c \xi^5/(9 \ga)$. With $\Om_c/\ga \simeq 30$
301: and $\xi \simeq 0.1 > \zeta$, we obtain
302: $P_{\textsc{cphase}}^{\text{sp}} \leq 4 \times 10^{-3}$ and
303: $P_{\textsc{cphase}}^{\text{tr}} \ll P_{\textsc{cphase}}^{\text{sp}}$.
304: This error probability is exactly the same as for a \textsc{swap} gate
305: with similar $\xi$, but the \textsc{cphase} gate is then 25 times faster
306: than the \textsc{swap} gate, since
307: $T_{\textsc{swap}}/T_{\textsc{cphase}} = 10 \Om_c \xi^4/(3 \ga \zeta^2)$.
308:
309: Having established all the basic principles of the proposed QP, we now
310: describe its possible realization. We envision a solid-state host
311: doped with active atoms having {\em non-degenerate} ground state, so as
312: to avoid mixing of various degenerate atomic states, which would invalidate
313: our simple two-level atomic model. Possible candidate systems include
314: sulphur-doped silicon, rare-earth (Yb or Nd) doped crystals \cite{rearth},
315: or semiconductor based nanostructures (quantum dots) \cite{QDots}. The
316: implantation of dopants and dots with controllable separations of few
317: nanometers is achievable with reasonable accuracy \cite{nano,BHHFKWSF}.
318:
319: With the arrangement of dopants shown in Fig. \ref{fig:cphs}-inset,
320: our scheme is capable of implementing arbitrary one-qubit rotations and
321: two-qubit logic gates, so as to obtain any desired unitary transformation
322: \cite{RevSteane}.
323: {\em (a) Individual SD qubits} would be rotated or read-out (and initialized)
324: by laser fields with frequency $\om_r$ or $\om_p$, respectively (see above),
325: and wave vector parallel to the interatomic axis, using the ``near-field''
326: technique (Fig. \ref{fig:cphs}-inset). The polarization of these fields can be
327: chosen such that they act only on the atomic transition from the nondegenerate
328: ground state to one of the magnetic sublevels of the excited state,
329: consistently with our two-level description of the atoms.
330: {\em (b) The \textsc{cphase} gate} between a chosen pair of qubits $A$ and $B$
331: is executed by a coupling field with frequency $\om_c$ and wave vector
332: $\textbf{k}_c \parallel \textbf{r}_{AB}$ that are specific for that pair.
333: {\em (c) The \textsc{swap} action} between neighboring qubits can be used to
334: convey the information in the QP, step-by-step, over large distances for which
335: the direct RDDI vanishes. To neutralize the \textsc{swap}, one can flip the
336: qubits at time intervals short compared to $[\De_{AB}^{(-)}]^{-1}$, which
337: is equivalent to the spin echo technique used in NMR \cite{NMR}.
338: Alternatively, the $\sqrt{\textsc{swap}}$ gate between two qubits $A$ and $B$
339: can be switched on and off via external ac Stark fields.
340:
341: Throughout this paper we have only dealt with the {\em radiative relaxation}
342: of the excited atomic state $\ket{e}$. This is adequate provided the competing
343: {\em nonradiative relaxation processes} are strongly {\em suppressed} by
344: working below the liquid helium temperature \cite{phonons} and/or using fast
345: ac Stark modulation of the vibrationally relaxing levels \cite{zeno}.
346: Another important consideration is the {\em inhomogeneous broadening} of the
347: atomic resonances. Consider two atoms having slightly different resonant
348: frequencies, $\om_{eg}^{(2)} - \om_{eg}^{(1)} = \de$, due to the host
349: inhomogeneity. This frequency mismatch results in an increase of the decay
350: rate $\Ga_-$ of the SD qubit in the amount $\ga \de^2/(8 \De^2)$. If we
351: require that this additional relaxation rate does not exceed $\Ga_-$ for
352: two resonant atoms, we obtain that the inhomogeneous width $\de$ must be
353: less than $\ga/\zeta^2$, which, for $\zeta \simeq 0.02$, yields
354: $\de\leq 2.5 \times 10^{3} \ga$.
355:
356: It is instructive to compare our scheme with previously considered
357: optically-controlled single- and two-qubit quantum gates:
358:
359: 1) In a commonly used optical scheme \cite{BCJD,PGCZ,LH},
360: a Raman qubit is represented by two metastable ground states $\ket{g_1}$
361: and $\ket{g_2}$ that are manipulated by two laser fields detuned by the amount
362: $\de_e \gg \ga_e$ from the intermediate excited state $\ket{e}$ having the
363: spontaneous decay rate $\ga_e$. The error probability during the qubit flip
364: $P_e^{\text{sp}} \leq \pi \ga_e /(2\de_e)$ is then an order of magnitude
365: larger than for the SD qubit, given {\em similar} values of the single-photon
366: $\Om_R$ and the effective two-photon $\Om_R^2/\de_e$ Rabi frequencies.
367:
368: 2) A \textsc{cphase} logic gate between two closely spaced Raman qubits
369: [see 1) above], $A$ and $B$, trapped in an optical lattice \cite{BCJD}, is
370: realized by an off-resonant ``catalysis'' field with Rabi frequency $\Om_C$,
371: which induces a RDDI-dependent ac Stark shift of the two-qubit states. One
372: then can show that during the gate operation, the probability of error due to
373: spontaneous decay of the excited states $\ket{e}_{A,B}$ is given by
374: $P_{\textsc{cphase}}^{(R)} \simeq 8 \pi \xi^3/3$,
375: where $\xi$ is the Lamb-Dicke parameter. With $\xi \simeq 0.1$, we obtain that
376: $P_{\textsc{cphase}}^{(R)} \simeq 8 \times 10^{-3}$, which is
377: twice worse than in our scheme with the same $\xi$. More dramatically, for
378: {\em similar} field strengths, e.g., $\Om_C/ \ga_e \simeq 30$ and
379: $\de_e \simeq 5 \De_{AB}^{(R)} \gg \Om_C$, where $\De_{AB}^{(R)}$ is
380: the RDDI coupling strength between the atoms on the transitions
381: $\ket{g_2}_{A,B} \to \ket{e}_{A,B}$, we find that the SD qubit
382: implementation of the \textsc{cphase} gate is $\sim 30$ times faster.
383:
384: 3) A \textsc{cphase} gate in an ion trap \cite{CZ} operates with speed
385: and error probability similar to our scheme. The error in the ion trap QP
386: is caused by the radiative decay of the auxiliary excited state, but one must
387: also reckon with error due to the phonon-mode decoherence
388: \cite{MS,SKKLMMRTIWM}. The main limitations of ion trap schemes are related
389: to difficulties with their scalability.
390:
391: To conclude, our proposal for an optically-manipulated, solid-state quantum
392: processor has no principal limitations on scalability. It allows us to
393: suppress radiative decoherence and enhance the speed of photon-mediated
394: quantum-logic gates, owing to the use of the ground and subradiant states
395: of effective dimers formed by resonant dipole-dipole interacting two-level
396: systems. These states constitute a physically realistic, simple and robust
397: ``decoherence-free subspace'' \cite{DFS}, whose implementation draws
398: efficiently upon the system resources (only two atoms per qubit). The highly
399: challenging experimental realization of such a quantum computer requires
400: nanofabrication techniques with nanometer precision of dopant or quantum
401: dot implantation \cite{nano,BHHFKWSF}.
402:
403: %\begin{acknowledgments}
404: %We acknowledge the support of the EC (QUACS Network) and the Feinberg
405: %School (D.P.).
406: %\end{acknowledgments}
407:
408: %\bibliography{dd,q_comp}
409: %\bibliographystyle{apsrev}
410:
411: \begin{thebibliography}{99}
412:
413: \bibitem{RevSteane}
414: A.~Steane, Rep. Prog. Phys. \textbf{61}, 117 (1998).
415:
416: \bibitem{CZ}
417: J.~I.~Cirac and P.~Zoller, Phys. Rev. Lett. \textbf{74}, 4091 (1995).
418:
419: \bibitem{MS}
420: K.~M{\o}lmer and A.~S{\o}rensen, Phys. Rev. Lett. \textbf{82}, 1835 (1999).
421:
422: \bibitem{SKKLMMRTIWM}
423: C.~A.~Sackett \textit{et al.}, Nature \textbf{404}, 256 (2000).
424:
425: \bibitem{PGCZ}
426: T.~Pellizzari, S.~A. Gardiner, J.~I. Cirac, and P.~Zoller,
427: Phys. Rev. Lett. \textbf{75}, 3788 (1995).
428:
429: \bibitem{BCJD}
430: G.~K. Brennen, C.~M. Caves, P.~S. Jessen, and I.~H. Deutsch,
431: Phys. Rev. Lett. \textbf{82}, 1060 (1999).
432:
433: \bibitem{BKMPV}
434: S.~Bose \textit{et al.}, Phil. Trans. Roy. Soc. Lond. A \textbf{356}, 1823
435: (1998).
436:
437: \bibitem{LDV}
438: D.~Loss and D.~P. DiVincenzo, Phys. Rev. A \textbf{57}, 120 (1998).
439:
440: \bibitem{Kane}
441: B.~E. Kane, Nature \textbf{393}, 133 (1998).
442:
443: \bibitem{BHHFKWSF}
444: M.~Bayer \textit{et al.}, Science \textbf{291}, 451 (2001).
445:
446: \bibitem{LH}
447: M.~D. Lukin and P.~R. Hemmer, Phys. Rev. Lett. \textbf{84}, 2818 (2000).
448:
449: \bibitem{DDI}
450: R.~H. Lehmberg, Phys. Rev. A \textbf{2}, 883 (1970); \textbf{2}, 889 (1970);
451: D.~P. Craig and T.~Thirunamachandran,
452: \emph{Molecular Quantum Electrodynamics}
453: (Academic Press, London, 1984), chap.~7.
454:
455: \bibitem{DiVincenzo}
456: D.~P. DiVincenzo, Fortschr. Phys. \textbf{48}, 771 (2000).
457:
458: \bibitem{QJumps}
459: W.~Nagourney, J.~Sandberg, and H.~Dehmelt,
460: Phys. Rev. Lett. \textbf{56}, 2797 (1986);
461: T. Sauter, W.~Neuhauser, R.~Blatt, and P.~E. Toschek,
462: \textit{ibid}. \textbf{57}, 1696 (1986);
463: J.~C. Bergquist, R.~G. Hulet, W.~M. Itano, and D.~J. Wineland,
464: \textit{ibid}. \textbf{57}, 1699 (1986).
465:
466: \bibitem{rearth}
467: R.~M. Macfarlane and R.~M. Shelby, \emph{Spectroscopy of Solids
468: Containing Rare Earth Ions} (North-Holland, Amsterdam, 1987).
469:
470: \bibitem{QDots}
471: L.~Jacak, P.~Hawrylak, and A.~Wijs, \emph{Quantum Dots} (Springer-Verlag,
472: Berlin, 1998).
473:
474: \bibitem{nano}
475: J.~W. Lyding, Proc. IEEE \textbf{85}, 589 (1997).
476:
477: \bibitem{NMR}
478: D.~Cory \textit{et al.}, Fortschr. Phys. \textbf{48}, 875 (2000).
479:
480: \bibitem{phonons}
481: T.~Takagahara, J Lumin. \textbf{70}, 129 (1996).
482:
483: \bibitem{zeno}
484: A.~G. Kofman and G.~Kurizki, Phys. Rev. Lett. \textbf{87}, 270405 (2001).
485:
486: \bibitem{DFS}
487: P.~Zanardi and M.~Rasetti, Phys. Rev. Lett. \textbf{79}, 3306 (1997);
488: D.~Bacon, J.~Kempe, D.~A. Lidar, and K.~B. Whaley,
489: \textit{ibid}. \textbf{85}, 1758 (2000).
490:
491: \end{thebibliography}
492:
493: \end{document}
494: