1: \documentclass[]{spie}
2: \usepackage[]{graphicx}
3: \usepackage{amsmath}
4: \usepackage{amsfonts}
5: \usepackage{bm}
6: \usepackage{color}
7:
8: \voffset 0.80 true cm
9: \hoffset 0.40 true cm
10: \setlength{\textwidth}{162mm}
11: \setlength{\textheight}{208mm}
12:
13: \title{
14: Coherence and Entanglement in Two-Qubit Dynamics: Interplay of the
15: Induced Exchange Interaction and Quantum Noise due to Thermal
16: Bosonic Environment
17: }
18:
19: \author{\textbf{Vladimir Privman} and \textbf{Dmitry Solenov}
20: \skiplinehalf %
21: Department of Physics, Clarkson
22: University, Potsdam, NY 13699--5721, USA}
23:
24: \authorinfo{Electronic addresses: Privman@clarkson.edu, Solenov@clarkson.edu}
25:
26: %/////////////////////////////////////////////////////////////////
27: %/////////////////////////////////////////////////////////////////
28: %/////////////////////////////////////////////////////////////////
29:
30: \newcommand{\ind}[1]{\textrm{#1}}
31: \newcommand{\com}[1]{}
32:
33: \newcommand{\Fig}[2]
34: {
35: \begin{figure}
36: \begin{center}
37: \begin{tabular}{c}
38: \includegraphics[height=5.5cm]{#1}
39: \end{tabular}
40: \end{center}
41: \caption[example]
42: { \label{#1} #2}
43: \vphantom{.}
44: \hrule
45: \end{figure}
46:
47: }
48:
49: %/////////////////////////////////////////////////////////////////
50: %/////////////////////////////////////////////////////////////////
51: %/////////////////////////////////////////////////////////////////
52:
53: \newcommand{\red}[1]{%
54: %\textcolor{red}% COMMENT ONLY THIS LINE AWAY (PREPEND %) TO REMOVE COLOR
55: {#1}}
56:
57: \begin{document}
58:
59: \maketitle
60:
61: \begin{abstract}
62:
63: We present a review of our recent results for the comparative
64: evaluation of the induced exchange interaction and quantum noise
65: mediated by the bosonic environment in two-qubit systems. We
66: report new calculations for P-donor-electron spins in Si-Ge type
67: materials. Challenges and open problems are discussed.
68:
69: {\noindent}\red{{\small{}{\bf Keywords:}\ decoherence, entanglement, qubit, exchange interaction, quantum noise, quantum computing.}}
70:
71: \vphantom{.}
72: \hrule
73: \end{abstract}
74: \
75:
76: \section{INTRODUCTION}\label{sec:intro}
77:
78: In this review we present results of our recent
79: investigations\cite{STPs,STPb} addressing several problems in the
80: field of open quantum systems; some have a rather long history.
81: With the actual experimental probes now being carried out at the
82: nanoscale, these problems have become more pressing, and some have
83: actually been suggested by the experimental
84: developments.\cite{Jiang,Jiang2,Craig,Elzerman,Koppens,Petta,MMJ,PF,Nakamura,Vion,Chiorescu,Yamamoto}
85: Perhaps the most fundamental (and difficult) of these problems is
86: the matter of accounting within a calculationally tractable
87: approach for relaxation vs.\ coherent dynamics in open quantum
88: systems. In particular the coherent and quantum-noise features
89: induced by the bosonic environment (bath of noninteracting bosonic
90: modes) are especially interesting, because this model is widely
91: applicable for quantum computing systems. Here we consider
92: bath-induced RKKY-type exchange interactions in two-qubit systems.
93:
94: This paper is organized as follows. In the next section, Sec.~\ref{sec:M}, we
95: formulate the model and discuss the issue of the initial
96: conditions. An example of the semiconductor (Si-Ge type) based qubit
97: model is given and the nature of the interactions is discussed.
98:
99: Section~\ref{sec:onset} is devoted to the initial dynamics of the
100: induced interaction and quantum noise. In particular, in
101: Sec.~\ref{ssec:OS} we present an exactly solvable model to obtain
102: the exchange interaction as well as the time-dependent correction
103: due to the initial conditions. The model also demonstrates the delay in
104: the response of the qubits caused by the finite propagation
105: velocities of the mediating \red{virtual} bosons. In Sec.~\ref{ssec:DM}, we study
106: the interplay of the coherent dynamics and quantum noise by calculating the
107: concurrence.\cite{Wootters1,Wootters2} Evolution of the reduced
108: density matrix elements is also investigated.
109:
110: At large times, as the system forgets the initial state, the
111: exchange interaction becomes stationary. This regime is discussed
112: in Sec.~\ref{sec:LT}. We utilize the master-equation formulation
113: of Markovian dynamics, Sec.~\ref{ssec:ME}, and go over an
114: illustrative one-dimensional (1D) example of the emerging
115: interaction and quantum noise in Sec.~\ref{ssec:1D}.
116: Generalization to higher dimensionality is given in
117: Sec.~\ref{ssec:GD}. A solid-state based three-dimensional (3D)
118: example of a semiconductor qubit model is presented in
119: Sec.~\ref{ssec:3D}. In the latter discussion we compare
120: bath-induced coupling and noise to other interactions, such as an
121: electromagnetic coupling of P-impurity spins (qubits).
122: Finally, brief summary and outline of some open
123: challenges are offered in Sec.~\ref{sec:discuss}.
124:
125: \hphantom{A}
126:
127: \section{THE MODEL AND INITIAL CONDITIONS}\label{sec:M}
128:
129: We consider two qubits, i.e., two two-state quantum systems,
130: modeled by localized electron spins ($1/2$) immersed in a
131: common quantum environment. The qubits are located at the distance
132: $\mathbf{d}$ from each other, far enough so that the direct
133: overlap of \red{the electron wave functions} is negligible. The environment is
134: modeled by a bath of bosonic modes, which are maintained at
135: temperature $T$. It is usually assumed that external
136: influences, as well as possibly
137: internal bath-mode interactions, set a fast time scale $t_B$ for the bath correlations to
138: equilibrate. The bath modes are then regarded as otherwise noninteracting. At least for the low-frequency bath modes, it is
139: usually argued that such a thermalization time for a generic case should be of
140: order $\hbar/k_{\textrm{B}}T$. The thermal-state density matrix of the bath modes, taken
141: here as noninteracting bosonic fields, is
142: %///////////////////////////////////////////////////////
143: \begin{equation}\label{Eq:M:rhoB}
144: \rho_B=\mathrm{Z}^{-1} \exp \bigg[{-{\sum\limits_{\mathbf{k},\mathbf{\xi}}
145: {\omega_{\mathbf{k},\mathbf{\xi}} a_{\mathbf{k},\mathbf{\xi}}^\dag
146: a_{\mathbf{k},\mathbf{\xi}}} }/{k_{\textrm{B}}T}}\bigg] \, ,
147: \end{equation}
148: %///////////////////////////////////////////////////////
149: where we set $\hbar = 1$, and the partition function is $\mathrm{Z}=1/\prod_k
150: (1-e^{-\omega_k/k_{\textrm{B}} T})$. The bath is linearly coupled to each
151: qubit,
152: %/////////////////////////////////////////////////////////////
153: \begin{equation}\label{Eq:M:H_SB}
154: H_{SB}=\sum\limits_{j=1,2}\sum\limits_{m=x,y,z}\sigma^j_m X^j_m \, ,
155: \end{equation}
156: %/////////////////////////////////////////////////////////////
157: where the superscripts ($j=1,2$) label the two spins, and the bath operators are given by
158: %/////////////////////////////////////////////////////////////
159: \begin{equation}\label{Eq:M:X_jm}
160: X^j_m=\sum_{\mathbf{k},\mathbf{\xi}}g^m_{\mathbf{k},\mathbf{\xi}}e^{i\mathbf{k}\cdot\mathbf{r}_j}
161: \left(a_{\mathbf{k},\mathbf{\xi}}+a_{-\mathbf{k},\mathbf{\xi}}^{\dagger}\right)\, .
162: \end{equation}
163: %//////////////////////////////////////////////////////////
164:
165: The overall Hamiltonian of the qubit-bath system is
166: %/////////////////////////////////////////////////////////////
167: \begin{equation}\label{Eq:M:H}
168: H = H_S + H_B + H_{SB}\, ,
169: \end{equation}
170: %//////////////////////////////////////////////////////////
171: where
172: %/////////////////////////////////////////////////////////////
173: \begin{equation}\label{Eq:M:H_B}
174: H_B=\sum_{\mathbf{k},\mathbf{\xi}}\omega
175: _{\mathbf{k},\mathbf{\xi}}a_{\mathbf{
176: k},\mathbf{\xi}}^{\dagger}a_{\mathbf{k},\mathbf{\xi}} \, ,
177: \end{equation}
178: %//////////////////////////////////////////////////////////
179: and the two spin-$1/2$ (qubits) will be assumed split by external
180: magnetic field,
181: %/////////////////////////////////////////////////////////////
182: \begin{equation}\label{Eq:M:H_S}
183: H_S=\Delta (\sigma_z^1 + \sigma_z^2 )/2 \, .
184: \end{equation}
185: %/////////////////////////////////////////////////////////////
186: Here $\Delta$ is the energy gap between the up and down states for
187: spins 1 and 2. A natural example of such a system are spins
188: of two localized electrons interacting via lattice vibrations
189: (phonons) by means of the spin-orbit
190: interaction.\cite{Mahan,Hasegawa,Roth,SO-Winkler} Another example
191: is provided by atoms or ions in a cavity, used as two-state
192: systems interacting with photons.\cite{PGCZ}
193:
194: Our emphasis here will be on calculating and comparing the
195: relative importance of the coherent (induced interaction) vs.\
196: quantum-noise effects of a given bosonic bath in the two-qubit
197: dynamics. We do not include other possible two-qubit interactions
198: in such comparative calculation of dynamical quantities. \red{In
199: Sec.~\ref{ssec:3D} we also include the direct electromagnetic (EM)
200: coupling for a comparison of the bath-induced and EM dipole-dipole
201: interaction strengths.}
202:
203: An interesting realization of the model formulated above can be
204: found in the impurity electron spin dynamics in semiconductors.
205: Let us consider P donor impurities embedded at controlled
206: positions in an otherwise very clean Si (or Ge) crystal
207: matrix. The system is maintained at very low temperature,
208: as appropriate for quantum computing. Therefore, the outer
209: donor-impurity electron remains bound. The spins of such localized
210: electrons can be utilized as
211: qubits. They are subject to external magnetic field
212: $\mathbf{H}$,
213: which produces the Zeeman splitting (\ref{Eq:M:H_S}). The
214: spin-orbit interaction couples the spin to the deformation
215: potential fluctuations of the host semiconductor, producing the
216: energy change
217: %/////////////////////////////////////////////////////////////
218: \begin{equation}\label{Eq:M:H-SO}
219: H_\textrm{SO}=\mu_{\textrm{B}} \! \! \sum\limits_{m,l\,=\,x,y,z}
220: \! \! {\sigma_m g_{ml}H_l}\,,
221: \end{equation}
222: %/////////////////////////////////////////////////////////////
223: where $\mu_\textrm{B}$ is the Bohr magneton, and
224: $\mathbf{H}=\{H_x,H_y,H_z\}$. Here the tensor $g_{ml}$ is
225: sensitive to lattice deformations. It was shown\cite{Roth} that
226: for the donor state which has tetrahedral symmetry (which is the
227: case for P in Si or Ge), the Hamiltonian (\ref{Eq:M:H-SO}) yields
228: the spin-deformation interaction of the form
229: %/////////////////////////////////////////////////////////////
230: \begin{equation}
231: H_i = {\rm A}\mu _{\textrm{B}} \left[ { \bar \varepsilon _{xx}
232: \sigma _x H_x + \bar \varepsilon _{yy} \sigma _y H_y + \bar
233: \varepsilon _{zz} \sigma _z H_z + {\bar\Delta }
234: (\mathbf{\sigma}\cdot\mathbf{H}})/3\right]
235: \\ \label{Eq:M:H-SD-Hxyz}
236: +\,{\rm B}\mu _{\textrm{B}} \left[ {\bar \varepsilon _{xy} \left({
237: \sigma_x H_y+\sigma_y
238: H_x}\right)+{\text{c}}{\text{.p}}{\text{.}}}\right]\,.
239: \end{equation}
240: %/////////////////////////////////////////////////////////////
241: Here c.p.\ denotes cyclic permutations and $\bar\Delta$ is the
242: effective dilatation. The tensor $\bar\varepsilon_{ij}$ already
243: includes averaging of the strain with the gradient of the
244: potential over the donor ground state wave function.
245:
246: As before, let us assume that we have two impurities separated by
247: distance \textbf{d}. For definiteness, one can direct the magnetic
248: field along the $z$-axis. Then the spin-deformation interaction
249: Hamiltonian simplifies to
250: %/////////////////////////////////////////////////////////////
251: \begin{equation}\label{Eq:M:H-SD-Hz}
252: H_i = {\rm A}\mu _{\textrm{B}} \bar \varepsilon _{zz} \sigma _z^i
253: H_z + {\rm B}\mu _{\textrm{B}} \left( {\bar \varepsilon _{yz}
254: \sigma _y^i H_z + \bar \varepsilon _{zx} \sigma _x^i H_z }
255: \right)\,.
256: \end{equation}
257: %/////////////////////////////////////////////////////////////
258: In terms of the quantized phonon field, we have\cite{Mahan,MKGB}
259: %/////////////////////////////////////////////////////////////
260: \begin{equation}\label{Eq:M:StrainTensor}
261: \bar \varepsilon _{ij} = \sum\limits_{{\mathbf{k}},{\mathbf{\xi
262: }}} {f({\mathbf{k}})\sqrt {\frac{\hbar } {{8\rho V\omega
263: _{{\mathbf{k}},{\mathbf{\xi }}} }}} \left( {\xi _{{\mathbf{k}},i}
264: k_j + \xi _{{\mathbf{k}},j} k_i } \right)\left(
265: {a_{{\mathbf{k}},{\mathbf{\xi }}}^\dag +
266: a_{{\mathbf{k}},{\mathbf{\xi }}} } \right)}\,,
267: \end{equation}
268: %/////////////////////////////////////////////////////////////
269: where in the spherical donor ground state
270: approximation\cite{Hasegawa,MKGB}
271: %/////////////////////////////////////////////////////////////
272: \begin{equation}\label{Eq:M:f(k)}
273: f({\mathbf{k}}) = \frac{1} {{\left( {1 + a_\textrm{B}^2 k^2 }
274: \right)^2}}\,.
275: \end{equation}
276: %/////////////////////////////////////////////////////////////
277: Here $a_\textrm{B}$ is {\it half\/} the effective Bohr radius of
278: the donor ground state wave function. In an actual Si or Ge
279: crystal, donor states are more complicated and include corrections
280: due to the symmetry of the crystal matrix including the fast
281: Bloch-function oscillations. However, the wave function of the
282: donor electrons in our case is spread over several atomic
283: dimensions (see below). Therefore, it suffices to consider
284: ``envelope'' quantities. Thus, in the spin-phonon Hamiltonian
285: (\ref{Eq:M:H_SB})-(\ref{Eq:M:X_jm}) coupling constants will be
286: taken in the form
287: %/////////////////////////////////////////////////////////////
288: \begin{equation}\label{Eq:M:gk-SD}
289: g_{{\mathbf{k}},{\mathbf{\xi }}}^m = \frac{D_m}{{\left( {1 +
290: a_\textrm{B}^2 k^2 } \right)^2}}\sqrt {\frac{\hbar } {{8\rho
291: V\omega _{{\mathbf{k}},{\mathbf{\xi }}} }}} \left( {\xi
292: _{{\mathbf{k}},z} k_m + \xi _{{\mathbf{k}},m} k_z } \right)\,,
293: \end{equation}
294: %/////////////////////////////////////////////////////////////
295: where $D_x=D_y={\rm B}\mu_\textrm{B} H_z$ and $D_z={\rm
296: A}\mu_\textrm{B} H_z$.
297:
298: It will be instructive to consider a one-dimensional calculation,
299: which simplifies the notation. \red{General results\cite{STPs,STPb} are presented later
300: in this article.} One can think of a 1D channel geometry along the
301: $z$ direction. This will give an example of an Ohmic bath model
302: discussed later. In a 1D channel the boundaries\cite{PBS} can
303: approximately quantize the spectrum of the phonons along $x$ and
304: $y$, depleting the density of states except at certain resonant
305: values. Therefore, the low-frequency effects, including the
306: induced coupling and quantum noise, will become effectively
307: one-dimensional, especially if the effective gap due to the
308: confinement is of the order of $\omega_c$. The frequency cutoff
309: comes from (\ref{Eq:M:f(k)}), namely, it is due to the bound
310: electron wave function localization. A channel of width comparable
311: to $\sim a_\textrm{B}$ will be required. This, however, may be
312: difficult to achieve in Si or Ge with the present-day technology.
313: Other systems may offer more immediately available 1D geometries
314: for testing similar theories, for instance, carbon nanotubes,
315: chains of ionized atoms suspended in ion
316: traps,\cite{Leibfried,Marquet,Porras} etc. In the 1D case, the
317: longitudinal acoustic (LA, $||$) phonons will account for the $g_{\mathbf{k},\parallel}^z$
318: component of the coupling, whereas the transverse acoustic (TA,
319: $\perp$) phonons will affect only the $x$ and $y$ spin
320: projections.
321:
322: It will be shown later that the contributions of the
323: cross-products of coupling constants,
324: $g_{\mathbf{k},\mathbf{\xi}}^m(g_{\mathbf{k},\mathbf{\xi}}^{m'})^*$
325: with $m\neq m'$, to quantities of interest vanish. The
326: combinations that enter the dynamics are
327: %/////////////////////////////////////////////////////////////
328: \begin{equation}\label{Eq:M:gk-1D}
329: |g_{k_z }^z|^2 = \frac{\textrm{A}^2 \mu_\textrm{B}^2 H_z^2}
330: {{4\rho V\omega _{k_z ,\parallel } }}\frac{{k_z^2 }} {{\left( {1 +
331: a_\textrm{B}^2 k_z^2 } \right)^4 }}\,,
332: \qquad
333: |g_{k_z }^x |^2 = |g_{k_z }^y |^2 = \frac{\textrm{B}^2
334: \mu_\textrm{B}^2 H_z^2} {{4\rho V\omega _{k_z , \bot }
335: }}\frac{{k_z^2 }} {{\left( {1 + a_{\rm B}^2 k_z^2 } \right)^4 }}\,.
336: \end{equation}
337: %/////////////////////////////////////////////////////////////
338: With the usual assumption for the low-frequency dispersion
339: relations $\omega _{k_z ,\parallel}\approx c_\parallel k_z$ and
340: $\omega _{k_z ,\perp}\approx c_\perp k_z$, the expressions
341: (\ref{Eq:M:gk-1D}) lead to the Ohmic bath model. The shape of the
342: frequency cutoff resulting from (\ref{Eq:M:gk-1D}) is not
343: exponential. However, to estimate the magnitude of the interaction
344: one can equivalently consider the exponential
345: cutoff,\cite{Leggett} with the summation over bosonic modes
346: carried out as
347: %/////////////////////////////////////////////////////////////
348: \begin{equation}\label{Eq:M:DOS}
349: \sum_{\mathbf{k},\xi}|g^m_{\mathbf{k},\xi}|^2 \to
350: \int\limits_0^\infty d\omega |g^m (\omega )|^2\Upsilon (\omega )=
351: \int\limits_0^\infty d\omega \ \alpha^m_n\omega^n
352: \exp(-\omega/\omega_c)\,,
353: \end{equation}
354: %/////////////////////////////////////////////////////////////
355: where $\Upsilon(\omega )$ is the density of states. The coupling
356: constants should then be taken as
357: %/////////////////////////////////////////////////////////////
358: \begin{equation}\label{Eq:M:alfas-1D}
359: \alpha_1^z = \frac{\textrm{A}^2 \mu_\textrm{B}^2 H_z^2}{8\pi
360: \rho S c^3_\parallel}\,,
361: \qquad\qquad
362: \alpha_1^x = \alpha_1^y = \frac{\textrm{B}^2 \mu_\textrm{B}^2
363: H_z^2}{8\pi \rho S c^3_\bot}\,,
364: \end{equation}
365: %/////////////////////////////////////////////////////////////
366: where $S$ is the cross section of the channel, and the cutoff is
367: $\omega_c \rightarrow c_{\parallel}/a_\textrm{B}$ for the $z$
368: component, and $\omega_c \rightarrow c_{\perp}/a_\textrm{B}$ for
369: the $x$ and $y$ components.
370:
371: For numerical estimates, we note that a typical
372: value\cite{Hasegawa,Roth,MKGB} of the effective Bohr radius in Si
373: for the P-donor-electron ground state wave function is
374: $2a_\textrm{B}=2.0\,$nm. The crystal lattice density is
375: $\rho=2.3\times10^3\,$kg/m$^3$, and the g-factor is $g^*=1.98$. For
376: an order-of-magnitude estimate, we take a typical value of the
377: phonon group velocity, $c_s=0.93 \times 10^4\,$m/s. The spin-orbit
378: coupling constants in Si are\cite{Hasegawa,Roth,MKGB}
379: $\textrm{A}^2\approx 10^2$ and $\textrm{B}^2\approx10^{-1}$. In
380: the Ge lattice, the spin-orbit coupling is dominated by the
381: non-diagonal terms,\cite{Hasegawa,Roth,MKGB} $\textrm{A}^2\approx
382: 0$ and $\textrm{B}^2\approx10^6$. The other parameters are
383: $2a_\textrm{B}=4.0\,$nm, $\rho=5.3\times10^3\,$kg/m$^3$,
384: $c_s=5.37\times10^3\,$m/s, and $g^* = 1.56$. This results in a
385: much stronger transverse component interaction. In both cases the
386: magnetic field will be taken of order $H_z=3\times 10^4\,$G. One could
387: use other experimentally suggested
388: values for the parameters, such as, for instance, $a_{\rm B}$. This will
389: not affect the results significantly.
390:
391: As mentioned earlier, a realistic phonon environment includes the
392: time scale at which it is reset to the thermal state. One can
393: think of the short- and long-time dynamics in reference to this
394: time scale. In the long-time regime, it is customary to introduce
395: Markovian-type assumptions, which include resetting the bath to
396: the thermal state instantaneously. We will use this approach in
397: the master-equation formulation of the large time dynamics,
398: Sec.~\ref{ssec:ME}.
399:
400: For the short-time dynamics, one has to address the matter of the
401: initial conditions at $t=0$. We will use the initially factorized
402: density matrix, with the thermal state assumed for the bath. In
403: quantum computing applications, such an initial condition has been
404: widely used for the qubit-bath
405: system,\cite{Privman,Tolkunov2,Solenov}
406: %/////////////////////////////////////////////////////////////
407: \begin{equation}\label{Eq:M:IC}
408: \rho(0) = \rho_S(0)\rho_B\,.
409: \end{equation}
410: %/////////////////////////////////////////////////////////////
411: This choice allows comparison with the Markovian results, and is
412: usually needed in order to make the short-time approximation
413: schemes tractable;\cite{Solenov} specifically, it is necessary for
414: the exact solvability of the model considered in the next section.
415:
416: A somewhat more ``physical'' excuse for the factorized initial
417: conditions is formulated as follows. Quantum computation is carried
418: out over a sequence of time intervals during which various
419: operations are performed on individual qubits and on pairs of
420: qubits. These operations include control gates, and measurements for
421: error correction. It is usually assumed that these ``control''
422: functions, involving rather strong interactions with external
423: objects, as compared to interactions with sources of quantum
424: noise, erase the fragile entanglement with the bath modes that
425: qubits can develop before those time intervals when they are
426: ``left alone'' to evolve under their internal (and bath induced)
427: interactions. Thus, for evaluating relative importance of the
428: quantum noise effects on the internal (and bath induced) qubit
429: dynamics, which is our goal here, we can assume that the state of the
430: qubit-bath system is ``reset'' to uncorrelated at $t=0$.
431:
432: \hphantom{A}
433:
434: \section{ONSET OF CORRELATIONS BETWEEN QUBITS}\label{sec:onset}
435:
436: Let us consider relatively short time scales and analyze how the
437: exchange interaction, accompanied by the quantum noise, sets in.
438: One can argue that such an investigation is only feasible provided
439: one knows the initial condition for the qubits-bath density matrix.
440: Indeed, the short-time interaction (and quantum noise) should depend on the
441: initial condition. The factorized initial condition, with the thermal-state bath, assumed
442: here, is expected in most quantum computing applications, as
443: discussed above.
444:
445: \subsection{Development of Induced Exchange Interaction}\label{ssec:OS}
446:
447: When the evolution of isolated qubits is slow with respect to the
448: other time scales such as that of decoherence, so that one can
449: assume vanishing qubit splitting energy, $\Delta =0$, and
450: if only one system operator enters the
451: qubit-bath interaction, then one obtains an exactly solvable model. A more general ``adiabatic'' system that allows exact solvability is obtained when $H_{S}$ commutes with $H_{SB}$. For
452: example, for electron spins of P-impurities in Si one has the spin-phonon
453: interaction dominated by such an adiabatic term, and only one such system operator in the
454: interaction matters for the onset dynamics. Though this situation
455: is rarely the case in quantum computing systems, the present model provides
456: a convenient tool for evaluating the initial dynamics of the exchange
457: interaction build-up, as well as for analyzing the response delay
458: due to the finite speed of the mediating \red{virtual} bosonic particles. These
459: features are difficult to capture analytically in other models.
460:
461: Therefore, let us presently take $\alpha _n^y=\alpha _n^z=0$,
462: while $\alpha _n^x \neq 0$. With the above assumptions, one can
463: utilize the bosonic operator techniques\cite{Louisell} to obtain
464: the reduced density matrix for the system (\ref{Eq:M:H_SB})-(\ref{Eq:M:X_jm}),
465: %/////////////////////////////////////////////////////////////
466: \begin{equation}\label{Eq:OS:AdiabaticSolution}
467: \rho _S (t) = \sum\limits_{\lambda ,\lambda '} {P_\lambda \rho _S
468: (0)P_{\lambda '} e^{\frak{L}_{\lambda \lambda '} (t)} }\,,
469: \end{equation}
470: %/////////////////////////////////////////////////////////////
471: where the projection operator is defined as
472: $P_\lambda=\left|{\lambda_1\lambda_2}\right\rangle\left\langle
473: {\lambda _1\lambda _2}\right|$, and
474: $\left|{\lambda_j}\right\rangle$ are the eigenvectors of
475: $\sigma_x^j$. The exponent in (\ref{Eq:OS:AdiabaticSolution})
476: consists of the real part, which leads to decay of off-diagonal
477: density-matrix elements resulting in decoherence,
478: %/////////////////////////////////////////////////////////////
479: \begin{equation}
480: \textrm{Re} \, \frak{L}_{\lambda \lambda '}(t)= -\sum\limits_k {G_k (t,T)
481: \bigg[ {\left( {\lambda '_1 - \lambda _1 } \right)^2 + \left({
482: \lambda '_2 - \lambda _2 } \right)^2 } }
483: \label{Eq:OS:DecoherenceFunction-ReL}
484: + \,{2\cos\! \left( {\frac{{\omega _k \left| {\mathbf{d}} \right|}}
485: {{c_s }}} \right)\!\left( {\lambda '_1 - \lambda _1 }
486: \right)\left( {\lambda '_2 - \lambda _2 } \right)} \bigg]\, ,
487: \end{equation}
488: %/////////////////////////////////////////////////////////////
489: and the imaginary part, which describes the coherent evolution,
490: %/////////////////////////////////////////////////////////////
491: \begin{equation}\label{Eq:OS:DecoherenceFunction-ImL}
492: \textrm{Im} \,\frak{L}_{\lambda \lambda '} (t) = \sum\limits_k {C_k
493: (t)\cos \left( {\frac{{\omega _k \left| {\mathbf{d}} \right|}}
494: {{c_s }}} \right)\left( {\lambda _1 \lambda _2 - \lambda '_1
495: \lambda '_2 } \right)}\, .
496: \end{equation}
497: %/////////////////////////////////////////////////////////////
498: Here we defined the standard spectral
499: functions\cite{Leggett,Privman}
500: %/////////////////////////////////////////////////////////////
501: \begin{equation}\label{Eq:OS:DecoherenceFunction-Gk}
502: G_k (t,T) = 2\frac{{\left| {g_k } \right|^2 }} {{\omega _k^2
503: }}\sin ^2 \left( \frac{{\omega _k t}} {{2}}\right) \coth \left(
504: {\omega _k \over 2k_{\textrm{B}} T} \right)
505: \end{equation}
506: %/////////////////////////////////////////////////////////////
507: and
508: %/////////////////////////////////////////////////////////////
509: \begin{equation}\label{Eq:OS:DecoherenceFunction-Ck}
510: C_k (t) = 2\frac{{\left| {g_k } \right|^2 }} {{\omega _k^2
511: }}\left( {\omega _k t - \sin \omega _k t} \right)\,.
512: \end{equation}
513: %/////////////////////////////////////////////////////////////
514: Calculating the sums by converting them to integrals over the
515: bath-mode frequencies $\omega$ in
516: (\ref{Eq:OS:DecoherenceFunction-ReL}) and
517: (\ref{Eq:OS:DecoherenceFunction-ImL}), assuming the Ohmic bath
518: with $n=1$, for $T>0$ one obtains a linear in time, $t$,
519: large-time behavior for both the temperature-dependent real part
520: and for the imaginary part. The coefficient for the former is
521: $\sim\! kT$, whereas for the latter it is $\sim\! \omega_c$. For
522: the super-Ohmic models, $n>1$, the real part grows slower, as was
523: also noted in the literature.\cite{Privman,PALMA,VKampen,Hanggi}
524:
525: \Fig{fig4ad.eps}{The magnitude of the time-dependent Hamiltonian
526: corresponding to the initial correction as a function of time and
527: distance. The Ohmic ($n=1$) case is shown. The inset demonstrates
528: the onset of the cross-qubit interaction on the same time scale.}
529:
530: First, let us analyze the coherent evolution part in
531: (\ref{Eq:OS:AdiabaticSolution}), namely, the effect that the
532: imaginary part of $\frak{L}_{\lambda \lambda '}(t)$ has on the
533: evolution of the reduced density matrix, since this contribution
534: leads to the induced interaction. If we omitted
535: (\ref{Eq:OS:DecoherenceFunction-ReL}) from
536: (\ref{Eq:OS:AdiabaticSolution}),
537: (\ref{Eq:OS:DecoherenceFunction-ImL}), and
538: (\ref{Eq:OS:DecoherenceFunction-Ck}), we would
539: obtain the (coherent) evolution operator in the form
540: %/////////////////////////////////////////////////////////////
541: \begin{equation}
542: e^{-i\left(H_\textrm{int}+F(t)\right)t}\;.
543: \end{equation}
544: %/////////////////////////////////////////////////////////////
545: The interaction $H_\textrm{int}$ comes from the first term in
546: (\ref{Eq:OS:DecoherenceFunction-Ck}),
547: %/////////////////////////////////////////////////////////////
548: \begin{equation}\label{Eq:OS:H-int}
549: H_\textrm{int}\! = \! - \frac{2\alpha _n^x \Gamma (n)c_s^n\omega
550: _c^n } {\left(c_s^2 + \omega _c^2 \left| {\mathbf{d}} \right|^2
551: \right)^{n/2} } \cos \left[{
552: n\arctan \left( {\frac{{\omega _c \left| {\mathbf{d}} \right|}}
553: {{c_s }}} \right)} \right]\!\sigma _x^1 \sigma _x^2\,.
554: \end{equation}
555: This expression gives the constant interaction that is important
556: in large time dynamics. We will obtain this induced exchange
557: interaction latter using different techniques for more general
558: cases, see Sec.~\ref{sec:LT}. The operator $F(t)$ is given by
559: %/////////////////////////////////////////////////////////////
560: \begin{equation}\label{Eq:OS:F(t)}
561: F(t) = 2\sigma _x^1 \sigma _x^2 \int\limits_0^\infty {d\omega
562: \frac{{D(\omega )\left| {g(\omega )} \right|^2 }} {\omega
563: }\frac{{\sin \omega t}} {{\omega t}}\cos \left( \frac{\omega
564: |\mathbf{d}|}{c_s} \right)} \, .
565: \end{equation}
566: %/////////////////////////////////////////////////////////////
567: It commutes with $H_\textrm{int}$ (and with itself at different
568: times), and therefore $d\left(tF(t)\right)/dt$ can be viewed as
569: the initial time-dependent correction to the interaction. In fact,
570: this term controls the onset of the induced coherent interaction;
571: note that $F(0)=-H_\textrm{int}$, but for large times $F(t) \sim
572: \alpha^x_n \omega_c^n/(\omega_c t)^n$.
573:
574: Let us consider in detail the time dependent correction
575: $H_F(t)=d\left(tF(t)\right)/dt$ to the interaction Hamiltonian during the initial
576: evolution,
577: %/////////////////////////////////////////////////////////////
578: \begin{equation}\label{Eq:OS:H_F}
579: H_F(t)= \sigma _x^1 \sigma _x^2 \alpha_n\Gamma(n)
580: \Big[
581: u(\omega_c|\mathbf{d}|/c_s - \omega_c t)%
582: +u(\omega_c|\mathbf{d}|/c_s + \omega_c t)%
583: \Big]\,,
584: \end{equation}
585: %/////////////////////////////////////////////////////////////
586: where $u(\xi)=\cos[n\arctan(\xi)]/(1+\xi^2)^{n/2}$. The above
587: expression is a superposition of two waves propagating in opposite
588: directions. In the Ohmic case, $n=1$, the shape of the wave is
589: simply $u(\xi)=1/(1+\xi^2)$.
590:
591: In Figure~\ref{fig4ad.eps}, we present the amplitude of $H_F(t)$,
592: defined via $H_F(t)={\cal H}_F\sigma _x^1 \sigma _x^2$, as well as
593: the sum of $H_\ind{int}$ and $H_F(t)$, for $n=1$. One can observe
594: that the ``onset wave'' of considerable amplitude and of shape
595: $u(\xi)$ propagates once between the qubits, ``switching on'' the
596: interaction. It does not affect the qubits once the interaction
597: has set in. One can also see in Figure~\ref{fig4ad.eps}, as well
598: as from (\ref{Eq:OS:H_F}), that the interaction between the qubits
599: is delayed by the time $\sim |\mathbf{d}|/c_s$, which is required
600: for the mediating \red{(virtual)} bosons to carry the response to the other qubit.
601: At the same time, relatively slow decay of the initial correction
602: $F(t)$ may necessitate a discussion regarding the meaning of the
603: ``coherent'' large-time induced interaction (when the noise
604: effects have also fully set in). We will offer additional comments
605: in the concluding section.
606:
607:
608: \subsection{Initial Dynamics of the Density Matrix Elements and Entanglement}\label{ssec:DM}
609:
610: Let us now take the entire solution
611: (\ref{Eq:OS:AdiabaticSolution}) which includes both the induced
612: interaction discussed above and the noise
613: (\ref{Eq:OS:DecoherenceFunction-ReL}). In the exact solution of
614: the short-time model, the bath is assumed to be thermalized
615: only initially. At
616: large enough times the externally induced equilibration of the bath should be considered.
617: We will account for this later when discussing the
618: perturbative Markovian approach. Here we continue our investigation of
619: the short-time model which assumes that the two qubits coupled with
620: the (noninteracting with each other) bath modes constitute a closed quantum system.
621:
622: To quantify the interplay of the effects of the induced
623: interaction vs.\ noise, we evaluate the
624: concurrence,\cite{Wootters1,Wootters2} which measures the
625: entanglement of the spin system and is monotonically related to
626: the entanglement of formation.\cite{Bennett,Vedral} For a mixed
627: state of two qubits, $\rho_S$, we first define the spin-flipped
628: state, $ \tilde \rho _S = \sigma^1_y \sigma^2_y \, \rho^*_S \,
629: \sigma^1_y \sigma^2_y $, and then the Hermitian operator
630: $R=\sqrt{\sqrt{\rho_S}\tilde\rho_S\sqrt{\rho_S}}$, with
631: eigenvalues $\lambda_{i=1,2,3,4}$. The concurrence is then
632: given\cite{Wootters2} by
633: %/////////////////////////////////////////////////////////////
634: \begin{equation}\label{Eq:DM:concurence}
635: C\left( {\rho _S } \right) = \max \Big\{ {0,2\mathop {\max
636: }\limits_i \lambda _i - \sum\limits_{j = 1}^4 {\lambda _j } }
637: \Big\}\,.
638: \end{equation}
639: %/////////////////////////////////////////////////////////////
640:
641: %
642: \Fig{fig_C1D_Si.eps}{ Development of the concurrence as a function
643: of time, calculated with $\alpha_1^z=0.5\cdot 10^{-7}$ and
644: $k_\textrm{B}T/\omega_c=0.5\cdot 10^{-2}$, which corresponds to
645: the magnetic field $H_z = 0.53\,$T and temperatures $T=0.34\,$K.
646: The left inset shows the distribution of the concurrence in the
647: $|\mathbf{d}|$-$t$ plane. The right inset presents the dynamics of
648: the diagonal density matrix elements $P_{++} \equiv \langle ++ |
649: \rho_S(t) | ++ \rangle $, etc., on the same time scale.}
650: %
651: Considering the Si channel geometry for the interaction, introduced
652: above as an illustrative example, we arrive at an approximately
653: adiabatic Hamiltonian ($\alpha_1^z \gg \alpha_1^{x,y}$) with
654: Ohmic-type coupling. The dynamics of the concurrence is presented
655: in Figure~\ref{fig_C1D_Si.eps}, and we note that the peak entanglement
656: can reach a sizable fraction of 1. The coupling constant $\alpha_1^z$ is quite small due to
657: the weakness of the spin-orbit coupling of P-impurity electrons in
658: Si, which results in low magnitude of the induced interaction (and
659: the noise due to the same environment). Nevertheless, one still
660: observes decaying periodic oscillations of entanglement, which indicate that
661: an approximately coherent dynamics can develop, due to the
662: bath-induced interaction, for several dynamical cycles.
663:
664: To understand the dynamics of the qubit system and its
665: entanglement, let us continue with the analysis of the coherent
666: part in (\ref{Eq:OS:AdiabaticSolution}). For each spin, we define the two states
667: $\left|\pm\right\rangle = \left[\,\left|\uparrow\right\rangle \pm
668: \left|\downarrow\right\rangle \right]/{\sqrt 2}$.
669: After the interaction,
670: $H_\ind{int}$, sets in (note the time scales in
671: Figures~\ref{fig4ad.eps} and \ref{fig_C1D_Si.eps}), it will split
672: the system energies into two degenerate pairs $E_0 = E_1=-{\cal
673: H}_\ind{int}$ and $E_2 = E_3={\cal H}_\ind{int}$. The wave
674: function is then
675: $|\psi(t)\rangle=\exp(-iH_\ind{int}t)|\psi(0)\rangle$. For the
676: initial ``++'' state, $|\psi(0)\rangle=\left|++\right\rangle $, it develops as
677: $\left|\psi(t)\right\rangle= \left|++\right\rangle \cos {\cal
678: H}_\ind{int}t + \left|--\right\rangle i\sin {\cal H}_\ind{int}t$,
679: where $H_\ind{int} = {\cal H}_\ind{int}\sigma^1_m\sigma^2_m$. One
680: can easily check that at times $t_E=\pi/4{\cal
681: H}_\ind{int},3\pi/4{\cal H}_\ind{int},\ldots$, maximally entangled
682: (Bell) states are obtained, while at times $t_0=0,\pi/2{\cal
683: H}_\ind{int},\pi/{\cal H}_\ind{int},\ldots$, the entanglement
684: vanishes; these special times can also be seen in Figure~\ref{fig_C1D_Si.eps}.
685:
686: The coherent dynamics just described is only approximate, because
687: the bath also induces decoherence that enters via
688: (\ref{Eq:OS:DecoherenceFunction-ReL}). The result for the
689: entanglement is that the decaying envelope function is
690: superimposed on the coherent oscillations described above. The
691: magnitudes of the first and subsequent peaks of the concurrence
692: are determined only by this function. As temperature increases,
693: the envelope decays faster resulting in lower values of the
694: concurrence.
695:
696: Note also that the non-monotonic behavior of the entanglement is
697: possible only provided that the initial state is not trivial with
698: respect to the induced interaction, otherwise only the phase
699: factor is developed. For example, taking the initial state
700: $\left[\,\left|-+\right\rangle + \left|+-\right\rangle \right]/\sqrt{2}$, in our
701: case would only lead to the destruction of entanglement, i.e., to
702: a monotonically decreasing concurrence, similar to results of
703: other studies.\cite{Eberly1,Eberly2}
704:
705: \red{For the model that allows the exact solution, i.e., for $H_S=0$,
706: one can notice that there is no relaxation by energy transfer
707: between the system and bath. The exponentials in
708: (\ref{Eq:OS:AdiabaticSolution}), with
709: (\ref{Eq:OS:DecoherenceFunction-ReL}), suppress only the
710: off-diagonal matrix elements, i.e., those with
711: $\lambda\neq\lambda'$. It happens, however, that at large times
712: the $\mathbf{d}$-dependence (the cosine term in
713: $\mathrm{Re}\,\frak{L}_{\lambda\lambda'}$) is not important in
714: (\ref{Eq:OS:DecoherenceFunction-ReL}). Therefore, one can show that
715: $\textrm{Re}\,\frak{L}_{\lambda\lambda'}(t\rightarrow\infty)$
716: vanishes for certain combinations of $\lambda\neq\lambda'$. Specifically, the limiting
717: $t\to \infty$ density matrix for our initial state
718: ($\left|++\right\rangle$) retains some non-diagonal elements,
719: %/////////////////////////////////////////////////////////////
720: \begin{equation}\label{Eq:DM:roLimit}
721: \rho(t\rightarrow\infty)\rightarrow\frac{1} {8}\left( {
722: {\begin{array}{*{20}c}
723: 3 & 0 & 0 & { - 1} \\
724: 0 & 1 & 1 & 0 \\
725: 0 & 1 & 1 & 0 \\
726: { - 1} & 0 & 0 & 3 \\
727: \end{array} }} \right)\, .
728: \end{equation}
729: %/////////////////////////////////////////////////////////////
730: The basis states here are $\left|++\right\rangle$,
731: $\left|+-\right\rangle$, $\left|-+\right\rangle$, and
732: $\left|--\right\rangle$. The significance of this and
733: similar\cite{Braun} results is in the fact that in the model with
734: $H_S=0$ and non-rethermalizing bath not all the off-diagonal
735: matrix elements need be suppressed by decoherence, even though the
736: concurrence of (\ref{Eq:DM:roLimit}) is zero.}
737:
738:
739: \section{INTERPLAY BETWEEN THE INDUCED INTERACTION AND QUANTUM NOISE AT LARGER TIMES}\label{sec:LT}
740:
741: \subsection{Master-Equation Approach}\label{ssec:ME}
742:
743: In this section we present the expressions for the induced
744: interaction and also for the noise effects due to the bosonic
745: environment, calculated perturbatively to the second order in the
746: spin-boson interaction, and with the assumption that the
747: environment is constantly reset to thermal.\cite{STPb}
748:
749: The dynamics of the system can be described by the equation for
750: the density matrix,
751: %/////////////////////////////////////////////////////////////
752: \begin{equation}\label{Eq:ME:Liuvolle}
753: i\dot \rho (t) = [H,\rho (t)] \, .
754: \end{equation}
755: %/////////////////////////////////////////////////////////////
756: In order to trace over the bath variables, we carry out the
757: second-order perturbative expansion. This dynamical description is
758: supplemented by the set of Markovian
759: assumptions,\cite{Leggett,VKampen,Louisell,Blum} one of which
760: invokes resetting the bath to thermal equilibrium, at temperature
761: $T$, after each infinitesimal time step, as well as at time $t=0$,
762: see (\ref{Eq:M:IC}), thereby decoupling the qubit system from the
763: environment.\cite{Leggett,VKampen} This is a physical assumption
764: appropriate for all but the shortest time scales of the system
765: dynamics.\cite{Privman,Tolkunov2,Solenov} It can also be viewed as
766: a means to phenomenologically account in part for the
767: randomization of the bath modes due to their interactions with
768: each other (anharmonicity) in real systems. This leads to the
769: master equation for the reduced density matrix of the qubits, $\rho
770: _S (t) = Tr_B \rho (t) $,
771: %/////////////////////////////////////////////////////////////
772: \begin{equation}\label{Eq:ME:MME}
773: i\dot \rho _S (t) = [H_S ,\rho _S (t)]
774: - i \!\int\limits_0^\infty {dt'Tr_B [H_{SB} ,[H_{SB} (t' - t),\rho
775: _B \rho _S (t)]]} \, ,
776: \end{equation}
777: where $H_{SB} (\tau)=e^{i(H_B+H_S)\tau}H_{SB}e^{-i(H_B+H_S)\tau}$.
778: Analyzing the structure of the integrand in (\ref{Eq:ME:MME}),
779: after lengthy calculations\cite{STPb} one can obtain the equation
780: with explicitly separated coherent and noise contributions,
781: %/////////////////////////////////////////////////////////////
782: \begin{equation}\label{Eq:ME:MRateEq}
783: i\dot \rho _S (t) = [H_\textrm{eff} ,\rho _S (t)] + i{\hat M}\rho
784: _S (t)\, .
785: \end{equation}
786: %/////////////////////////////////////////////////////////////
787: The effective coherent Hamiltonian $H_\textrm{eff}$ is
788: %/////////////////////////////////////////////////////////////
789: \begin{equation}
790: H_\textrm{eff}\!\!\!= H_S +\!\!\!\! \sum\limits_{m=x,y,z}
791: \!\!\! {2\chi _c^m ({\mathbf{d}})\sigma _m^1 \sigma _m^2 } - \chi
792: _s^x ({\mathbf{d}})\left( {\sigma _x^1 \sigma _y^2 + \sigma _x^2
793: \sigma _y^1 } \right)
794: + \chi _s^y ({\mathbf{d}})\left( {\sigma _y^1 \sigma _x^2 +
795: \sigma _y^2 \sigma _x^1 } \right) -
796: \left[\eta_s^x(0)+\eta_s^y(0)\right]
797: \left({\sigma_z^1+\sigma_z^2}\right)\, . \label{Eq:ME:H-eff}
798: \end{equation}
799: %/////////////////////////////////////////////////////////////
800: The expressions for the amplitudes $\chi_c^m(\mathbf{d})$,
801: $\chi_s^m(\mathbf{d})$, $\eta_c^m(\mathbf{d})$, and
802: $\eta_s^m(\mathbf{d})$ will be given bellow. The first three terms
803: following $H_S$ constitute the interaction between the two spins.
804: We will argue below that the leading induced exchange interaction
805: is given by the first added term, proportional to
806: $\chi_c^m(\mathbf{d})$. The last term gives the qubit Lamb shifts.
807:
808: The otherwise cumbersome expression for the noise terms can be
809: represented concisely by introducing the noise superoperator $\hat
810: M$, which involves single-qubit contributions, which are usually
811: dominant, as well as two-qubits terms,
812: %/////////////////////////////////////////////////////////////
813: \begin{equation}\label{Eq:ME:M-sums}
814: {\hat M} = \sum\limits_{m,i} \Big[ {\hat M}_m^i (0) +\sum_{j \ne
815: i} {\hat M}_m^{ij} (\mathbf{d} ) \Big] \,,
816: \end{equation}
817: %/////////////////////////////////////////////////////////////
818: where the summations are over the components, $m=x,y,z$, and the
819: qubits, $i,j=1,2$. One can define the superoperators ${\hat L}_a
820: (O_1)O_2=\{O_1,O_2\}$, ${\hat L}(O_1,O_2)O_3=O_1O_3O_2$, and
821: ${\hat L}_\pm(O_1,O_2)={\hat L}(O_1,O_2)\pm {\hat L}(O_2,O_1)$,
822: to write
823: %/////////////////////////////////////////////////////////////
824: \begin{equation}\label{Eq:ME:M-cross}
825: {\hat M}_m^{ij}(\mathbf{d})= \eta_c^m(\mathbf{d}) \left[{2{\hat
826: L}(\sigma^i_m,\sigma^j_m) - {\hat L}_a(\sigma^i_m\sigma^j_m)}
827: \right]
828: + \eta_s^m(\mathbf{d})\left[{{\hat
829: L}_+(\sigma^i_m,\varsigma^j_{m})-{\hat L}_a(\sigma^i_m
830: \varsigma^j_{m} )}\right]
831: - i\chi_s^m(\mathbf{d}){\hat L}_-(\sigma^i_m ,\varsigma^j_{m})\, ,
832: \end{equation}
833: %/////////////////////////////////////////////////////////////
834: where we denote $\varsigma^j_{m} = \frac{i}{2}[\sigma_z^j ,\sigma
835: _m^j]$, and
836: %/////////////////////////////////////////////////////////////
837: \begin{equation}\label{Eq:ME:M-local}
838: {\hat M}_m^j(0)= \eta _c^m (0)\left[{2{\hat L}(\sigma _m^j,\sigma
839: _m^j) -{\hat L}_a(\sigma _m^j \sigma _m^j )}\right]
840: + \eta _s^m(0){\hat L}_+(\sigma_m^j ,\varsigma^j_{m})
841: - i\chi _s^m (0)\left[{{\hat
842: L}_-(\sigma_m^j,\varsigma^j_{m})+{\hat L}_a(\sigma_m^j
843: \varsigma^j_{m})}\right]\,.
844: \end{equation}
845: %//////////////////////////////////////////////////////////
846:
847: The amplitudes in (\ref{Eq:ME:H-eff}), (\ref{Eq:ME:M-cross}), and
848: (\ref{Eq:ME:M-local}), calculated for the interaction defined in
849: (\ref{Eq:M:H_SB})-(\ref{Eq:M:X_jm}), are
850: %/////////////////////////////////////////////////////////////
851: \begin{equation}\label{Eq:ME:he-c}
852: \chi _c^m ({\mathbf{d}}) = - \sum\limits_\mathbf{\xi}
853: {\int\limits_{ - \infty }^\infty {\frac{{Vd{\mathbf{k}}}}
854: {{\left( {2\pi } \right)^3 }}} } \left|
855: {g_{{\mathbf{k}},\mathbf{\xi}}^m } \right|^2 \frac{{\omega
856: _{{\mathbf{k}},\mathbf{\xi}} \cos \left( {{\mathbf{k}} \cdot
857: {\mathbf{d}}} \right)}} {{\omega _{{\mathbf{k}},\mathbf{\xi}}^2 -
858: \Delta ^2 (1 - \delta _{m,z} )}}\,,
859: \end{equation}
860: %/////////////////////////////////////////////////////////////
861: %/////////////////////////////////////////////////////////////
862: \begin{equation}\label{Eq:ME:eta-c}
863: \eta _c^m ({\mathbf{d}}) = \frac{\pi}{2}\sum\limits_\mathbf{\xi}
864: {\int\limits_{ - \infty }^\infty {\frac{{Vd{\mathbf{k}}}}
865: {{\left( {2\pi } \right)^3 }}} } \left|
866: {g_{{\mathbf{k}},\mathbf{\xi}}^m } \right|^2 \coth \frac{{\omega
867: _{{\mathbf{k}},\mathbf{\xi}} }} {{2k_\textrm{B}T}}\cos \left(
868: {{\mathbf{k}} \cdot {\mathbf{d}}} \right)
869: \sum\limits_{q = \pm 1} {\delta \left( {\omega
870: _{{\mathbf{k}},\mathbf{\xi}} + (1 - \delta _{m,z} )q\Delta }
871: \right)}\,,
872: \end{equation}
873: %/////////////////////////////////////////////////////////////
874: %/////////////////////////////////////////////////////////////
875: \begin{equation}\label{Eq:ME:he-s}
876: \chi _s^m ({\mathbf{d}}) = - \left( {1 - \delta _{m,z} }
877: \right)\sum\limits_\mathbf{\xi} {\int\limits_{ - \infty }^\infty
878: {\frac{{Vd{\mathbf{k}}}} {{\left( {2\pi } \right)^3 }}} } \left|
879: {g_{{\mathbf{k}},\mathbf{\xi}}^m } \right|^2 \cos \left(
880: {{\mathbf{k}} \cdot {\mathbf{d}}} \right)
881: \sum\limits_{q = \pm 1} {\frac{{\pi s}} {2}\delta \left(
882: {\omega _{{\mathbf{k}},\mathbf{\xi}} + q\Delta } \right)}\, ,
883: \end{equation}
884: %/////////////////////////////////////////////////////////////
885: %/////////////////////////////////////////////////////////////
886: \begin{equation}\label{Eq:ME:eta-s}
887: \eta _s^m ({\mathbf{d}}) = \left( {1 - \delta _{m,z} } \right)
888: \sum\limits_\mathbf{\xi} {\int\limits_{ - \infty }^\infty
889: {\frac{{Vd{\mathbf{k}}}} {{\left( {2\pi } \right)^3 }}} } \left|
890: {g_{{\mathbf{k}},\mathbf{\xi}}^m } \right|^2 \coth \frac{{\omega
891: _{{\mathbf{k}},\mathbf{\xi}} }} {{2k_\textrm{B}T}}\frac{{\Delta
892: \cos \left( {{\mathbf{k}} \cdot {\mathbf{d}}} \right)}} {{\omega
893: _{{\mathbf{k}},\mathbf{\xi}}^2 - \Delta ^2 }}\,.
894: \end{equation}
895: %/////////////////////////////////////////////////////////////
896: Here the principal values of integrals are assumed.
897:
898: Note that $\chi _c^m ({\mathbf{d}})$ appears only in the induced
899: interaction Hamiltonian in (\ref{Eq:ME:H-eff}), whereas $\eta
900: _c^m ({\mathbf{d}})$, $\chi _s^m ({\mathbf{d}})$, and $\eta _s^m
901: ({\mathbf{d}})$ enter both the interaction and noise terms.
902: Therefore, in order to establish that the induced interaction can
903: be significant for some time scales, we have to demonstrate that
904: $\chi _c^m ({\mathbf{d}})$ can have a much larger magnitude than
905: the maximum of the magnitudes of $\eta _c^m ({\mathbf{d}})$, $\chi
906: _s^m ({\mathbf{d}})$, and $\eta _s^m ({\mathbf{d}})$. The third
907: and fourth terms in the expression for the interaction
908: (\ref{Eq:ME:H-eff}) are comparable to the noise and therefore have
909: no significant contribution to the coherent dynamics.
910:
911:
912: \subsection{Qubit Coupling in a 1D Channel}\label{ssec:1D}
913:
914: It is instructive to start with the simple 1D example, leaving the
915: derivations for higher dimensions to the next subsection. The 1D
916: geometry can actually be natural for certain ion-trap
917: quantum-computing schemes, in which ions in a chain are subject to
918: Coulomb interaction, developing a variety of phonon-mode lattice
919: vibrations.\cite{Leibfried,Marquet,Porras}
920:
921: In our case we allow the phonons to propagate in a single
922: direction, along $\mathbf{d}$, so that ${\mathbf{k}} \cdot
923: {\mathbf{d}} = k\left| {\mathbf{d}} \right|$. Here, for
924: definiteness, we also assume the linear dispersion, $\omega _k =
925: c_s k$. Furthermore, we will ignore the polarization index in the
926: coupling constants: $g_{{\mathbf{k}},\mathbf{\xi}}^m \to g^m
927: (\omega )$.
928:
929: The induced interaction and noise terms depend on the amplitudes
930: $\chi _c^m ({\mathbf{d}})$, $\eta _c^m ({\mathbf{d}})$, $\chi _s^m
931: ({\mathbf{d}})$, and $\eta _s^m ({\mathbf{d}})$, two of which can
932: be evaluated explicitly for the 1D case, because of the
933: $\delta$-functions \red{in (\ref{Eq:ME:eta-c}), (\ref{Eq:ME:he-s}).}
934: However, to derive an explicit expression for $\chi _c^m
935: ({\mathbf{d}})$ and $\eta _s^m ({\mathbf{d}})$, one needs to
936: specify the $\omega$-dependence in $\left| {g^m (\omega )}
937: \right|^2$. For the sake of simplicity, in this section we
938: approximate $|g^m (\omega )|^2$ by a linear function with
939: superimposed exponential cutoff. For a constant 1D density of
940: states $V/2\pi c_s$, this is a variant of the
941: Ohmic-dissipation\cite{Leggett} model, i.e., (\ref{Eq:M:DOS}) with
942: $n=1$ (the case when $n>1$ is considered in the next subsection).
943:
944: In most practical applications, we expect that $\Delta \ll
945: \omega_c$. With this assumption, we obtain
946: %/////////////////////////////////////////////////////////////
947: \begin{equation}\label{Eq:1D:he-c-explicit-1D}
948: \chi _c^m ({\mathbf{d}}) = \frac{{\alpha _1^m \omega _c }} {{1 +
949: \left( {{{\omega _c \left| {\mathbf{d}} \right|} \mathord{\left/
950: {\vphantom {{\omega _c \left| {\mathbf{d}} \right|} {c_s }}} \right.
951: \kern-\nulldelimiterspace} {c_s }}} \right)^2 }}\,,
952: \end{equation}
953: %/////////////////////////////////////////////////////////////
954: %/////////////////////////////////////////////////////////////
955: \begin{equation}\label{Eq:1D:he-s-explicit-1D}
956: \chi _s^m ({\mathbf{d}}) = \alpha _1^m \omega _c \frac{\pi }
957: {2}\left( {1 - \delta _{m,z} } \right)\frac{\Delta } {{\omega _c
958: }}\cos \frac{{\Delta \left| {\mathbf{d}} \right|}} {{c_s }}\,,
959: \end{equation}
960: %/////////////////////////////////////////////////////////////
961: %/////////////////////////////////////////////////////////////
962: \begin{equation}\label{Eq:1D:eta-c-explicit-1D}
963: \eta _c^m ({\mathbf{d}}) = \alpha _1^m \omega _c \frac{\pi }
964: {2}\left( {1 - \delta _{m,z} } \right)\frac{\Delta } {{\omega _c
965: }}\coth \frac{\Delta } {{2k_\textrm{B}T}}\cos \frac{{\Delta \left|
966: {\mathbf{d}} \right|}} {{c_s }}\,.
967: \end{equation}
968: %/////////////////////////////////////////////////////////////
969: The expression for $\eta _s({\mathbf{d}})$ could not be obtained
970: in closed form. However, numerical estimates suggest that $\eta
971: _s({\mathbf{d}})$ is comparable to $\eta _c^m ({\mathbf{d}})$. At
972: short spin separations $\mathbf{d}$, $\eta _s({\mathbf{d}})$ is
973: approximately bounded by $- \alpha_1^m \Delta
974: \ln\frac{\Delta}{\omega_c} \exp(-\frac{{\Delta \left| {\mathbf{d}}
975: \right|}} {{c_s }})$, while at lager distances it may be
976: approximated by $\alpha_1^m \Delta\frac{\pi}{2}\sin \frac{{\Delta
977: \left| {\mathbf{d}} \right|}} {{c_s }}
978: \coth\frac{\Delta}{2k_\textrm{B}T}$. The level of noise may be
979: estimated by considering the quantity
980: %/////////////////////////////////////////////////////////////
981: \begin{equation}\label{Eq:1D:M-explicit-1D}
982: {\cal M} = \max_{|\mathbf{d}|} \left|{\eta _c^m(\mathbf{d}), \chi
983: _s^m(\mathbf{d}), \eta _s^m(\mathbf{d})} \right|\,.
984: \end{equation}
985: %/////////////////////////////////////////////////////////////
986:
987: The interaction Hamiltonian takes the form
988: %/////////////////////////////////////////////////////////////
989: \begin{equation}\label{Eq:1D:H-int-1D}
990: H_{\operatorname{int} } = - \frac{2} {{1 + \left( {{{\omega _c
991: \left| {\mathbf{d}} \right|}/{c_s }}} \right)^2 }}
992: \sum\limits_{m = x,y,z} {\alpha _1^m \omega _c \sigma _m^1 \sigma
993: _m^2}\,.
994: \end{equation}
995: %/////////////////////////////////////////////////////////////
996: This induced interaction is temperature independent. It is
997: long-range and decays as a power law for large $\mathbf{d}$. Note
998: that the expression (\ref{Eq:1D:H-int-1D}) is the same as the
999: induced interaction obtained from the short-time model
1000: (\ref{Eq:OS:H-int}), provided the proper spin-phonon interaction
1001: component is considered.
1002:
1003: If the noise term were not present, the spin system would be
1004: governed by the Hamiltonian $H_S + H_{\textrm{int}}$. To be
1005: specific, let us analyze the spectrum, for instance, for $\alpha
1006: _1^x = \alpha _1^y$, $\alpha _1^z = 0$. The two-qubit
1007: states would consist of the singlet
1008: $(\left|{\uparrow\downarrow}\right\rangle-
1009: \left|{\downarrow\uparrow}\right\rangle)/\sqrt{2}$ and the split
1010: triplet $\left|{\uparrow\uparrow}\right\rangle$,
1011: $(\left|{\uparrow\downarrow}\right\rangle+
1012: \left|{\downarrow\uparrow}\right\rangle)/\sqrt{2}$, and
1013: $\left|{\downarrow\downarrow}\right\rangle$, with energies
1014: $E_2=-4\chi _c^x$, $E_0=-\Delta$, $E_1=4\chi _c^x$, $E_3=\Delta$,
1015: respectively. The energy gap $|E_1-E_2|$ between the two entangled
1016: states is defined by $4\chi _c^x$ ($=4\chi _c^y$). In the presence
1017: of noise, the oscillatory, approximately coherent evolution of the
1018: spins can be observed over several oscillation cycles provided
1019: that $2\alpha_1^m \omega_c/[1 +(\omega_c |\mathbf{d}|/c_s)^2]>
1020: {\cal M}$. The energy levels will acquire effective width due to
1021: quantum noise, of order $\eta_c^m(\mathbf{d})$.
1022:
1023: \subsection{Boson-Mediated Induced Qubit-Qubit Interaction in General Dimensions}\label{ssec:GD}
1024:
1025: Let us generalize the results of the previous subsections, where
1026: we considered the 1D case with Ohmic dissipation within the
1027: Markovian approach. In the general case, let us consider the
1028: Markovian model and, again, assume that $\Delta/\omega_c$ is
1029: small. We will also assume that the absolute square of the $m$th
1030: component of the spin-boson coupling, when multiplied by the
1031: density of states, can be modeled by
1032: $\alpha_{n_m}^m{}\omega^{n_m}\exp(-\omega/\omega_c)$; see
1033: (\ref{Eq:M:DOS}). The integration in (\ref{Eq:ME:he-c}) can then
1034: be carried out in closed form for any $n_m=1,2,\ldots\,\,$. The
1035: induced interaction (\ref{Eq:1D:H-int-1D}) is thus generalized to
1036: %/////////////////////////////////////////////////////////////
1037: \begin{equation}\label{Eq:GD:H-int}
1038: H_\textrm{int}=-\!\!\!\sum\limits_{m =
1039: x,y,z}{\alpha_{n_m}^m\omega_c^{n_m }\sigma_m^1\sigma _m^2}
1040: \frac{{2\Gamma (n_m
1041: )\textrm{Re}\left({1+i\omega_c\left|{\mathbf{d}}\right|/c_s}\right)^{n_m}}}
1042: {{\left[{1+\left({\omega_c\left|{\mathbf{d}}\right|/c_s}\right)^2}\right]^{n_m}}}\,.
1043: \end{equation}
1044: %/////////////////////////////////////////////////////////////
1045:
1046: With the appropriate choice of parameters, the result for the
1047: induced interaction, but not for the noise, coincides with the
1048: expression for $H_\textrm{int}$ obtained within the short-time
1049: model. The effective interaction has the large-distance asymptotic
1050: behavior $\left|{\mathbf{d}}\right|^{-n_m}$, for even $n_m$, and
1051: $\left|{\mathbf{d}}\right|^{-n_m-1}$, for odd $n_m$.
1052:
1053: In higher dimensions the structure of
1054: $g_{{\mathbf{k}},\mathbf{\xi}}^m$ in the $\textbf{k}$-space
1055: becomes important. Provided $\omega _{{\mathbf{k}},\mathbf{\xi}}$
1056: is nearly isotropic, the integrals in
1057: (\ref{Eq:ME:he-c})-(\ref{Eq:ME:eta-s}) will include (in 3D) a
1058: factor
1059: $\int_0^{2\pi}{d\varphi}\int_0^\pi{d\theta\sin\theta}\left|{
1060: g_{k\theta\varphi,\mathbf{\xi}}^m}\right|^2\cos\left({k\left|{\mathbf{d}}
1061: \right|\cos\theta}\right)$, which can be written as
1062: $[f^m_1(\omega,k|\mathbf{d}|)-f^m_2(\omega,
1063: k|\mathbf{d}|)\partial/\partial|\mathbf{d}|]\cos (k|\mathbf{d}|)$,
1064: \red{see Eqs.~(B1)-(B4) of Ref.~2 %\cite{STPb}
1065: for details.} When the
1066: dependence of $f^m_1,f^m_2$ on $k|\mathbf{d}|$ is negligible, the
1067: interaction is simply $H_\textrm{int}\rightarrow
1068: H_\textrm{int}|_{\{n_m\}\rightarrow
1069: a}-(\partial/\partial|\mathbf{d}|)H_\textrm{int}|_{\{n_m\}\rightarrow
1070: b}$, where $a$ and $b$ are sets of three integers representing the
1071: $\omega$-dependence of $f_1^m(\omega,k|\mathbf{d}|)$ and
1072: $f_2^m(\omega,k|\mathbf{d}|)$. Otherwise, a more complicated
1073: dependence on $|\mathbf{d}|$ is expected. The noise superoperator
1074: can be treated similarly.
1075:
1076:
1077: \subsection{Induced Interaction vs.\ Noise in the Semiconductor Impurity Qubit Model}\label{ssec:3D}
1078:
1079: Let us now proceed to an example of the semiconductor qubit model
1080: in the bulk material, which gives usual 3D geometry for phonon
1081: propagation. We consider spins of P-impurity donor electrons as
1082: qubits. Dilute P-impurities are embedded in the matrix of a Si or
1083: Ge crystal kept at a sufficiently low temperature such that the
1084: outer electrons remain bound. The two impurities placed next to
1085: each other at distances of about 10-30$\,$nm would constitute a
1086: two qubit register. The spins would then be interacting indirectly
1087: via the spin-orbit coupling and lattice deformation (phonons) as
1088: well as directly via the dipole-dipole electromagnetic coupling. In
1089: this section we will estimate and compare both. We will also
1090: present the dynamics due to the coherent vs.\ noise induced
1091: features and their interplay.
1092:
1093: Let us consider for simplicity only the LA phonon branch, ${\mathbf{\xi }} \to
1094: {\mathbf{k}}/\left| {\mathbf{k}} \right|$, and assume an isotropic
1095: dispersion
1096: $\omega_{\mathbf{k},\mathbf{\xi}}=c_s\left|\mathbf{k}\right|$. The
1097: expression for the coupling constants is then
1098: %/////////////////////////////////////////////////////////////
1099: \begin{equation}\label{Eq:3D:gk-3D}
1100: g_{{\mathbf{k}},{\mathbf{\xi }}}^m = D_m \frac{{k_z k_m }}
1101: {{\left( {1 + a_\textrm{B}^2 k^2 } \right)^2 }}\sqrt {\frac{\hbar
1102: } {{2\rho Vc_s k^3 }}}\,.
1103: \end{equation}
1104: %/////////////////////////////////////////////////////////////
1105: One can show\cite{STPb} that the cross terms, with $m \neq n$, of
1106: the correlation functions
1107: $Tr_B \left[ X_m^j X_n^i (t)\rho_B \right]$, depend
1108: on the combination $g_{\mathbf{k},
1109: \mathbf{\xi}}^m(g_{\mathbf{k}, \mathbf{\xi}}^n)^*$, which is
1110: always an odd function of one of the projections of the wave
1111: vector. Thus, all the non-diagonal terms vanish.
1112: % VPVPVPVPVP
1113:
1114: Integrating (\ref{Eq:ME:he-c}) and (\ref{Eq:ME:eta-c}) with
1115: (\ref{Eq:3D:gk-3D}), one can demonstrate\cite{STPb} that
1116: decoherence is dominated by the individual noise terms for each
1117: spin, with the typical amplitude
1118: %/////////////////////////////////////////////////////////////
1119: \begin{equation}\label{Eq:3D:eta-c-xy}
1120: \eta _c^{x,y} (0) = C_\textrm{B} \frac{{2\pi ^2 }}
1121: {{15}}\frac{b^3}{\left(
1122: {1+b^2}\right)^4}\coth\frac{\Delta}{2k_\textrm{B}T}\,,
1123: \end{equation}
1124: %/////////////////////////////////////////////////////////////
1125: where $b = \Delta a_{\rm B} /c_s$ and $C_\textrm{B}={\rm B}^2\mu_{\textrm{B}}^2
1126: H_z^2/\left({16\pi^3\rho \hbar a_{\rm B}^3 c_s^2}\right)$. The
1127: interaction amplitude $\chi _c^m ({\mathbf{d}})$ and, therefore,
1128: the induced spin-spin interaction, have inverse-square power-law
1129: asymptotic form for the $x$ and $y$ spin components, with a
1130: superimposed oscillation, and inverse-fifth-power-law decay for
1131: the $z$ spin components,
1132: %/////////////////////////////////////////////////////////////
1133: \begin{eqnarray}\label{Eq:3D:H-int-asymptotic}
1134: H_\textrm{int}&=&\sum\nolimits_m {2\chi _c^m ({\mathbf{d}})\sigma
1135: _m^1 \sigma _m^2 }
1136: \\ \nonumber
1137: &\xrightarrow{{r\gg1}}& - 4\pi ^2 C_\textrm{B} \frac{{2b}}
1138: {{\left( {1 + b^2 } \right)^4 }}\frac{{\sin br}} {{r^2 }}\left(
1139: {\sigma _x^1 \sigma _x^2 + \sigma _y^1 \sigma _y^2 } \right) +
1140: 384\pi ^2 C_\textrm{A} \frac{2} {{r^5 }}\sigma _z^1 \sigma _z^2
1141: \\ \nonumber
1142: &\xrightarrow{{r\ll 1}}& \left[ - 4 \pi^2 C_\mathrm{B}
1143: \frac{1+9b^2-9b^4+b^6}{120 (1+b^2)^4} + {\cal O}(r^2)
1144: \right]\left( {\sigma _x^1 \sigma _x^2 + \sigma _y^1 \sigma _y^2
1145: } \right) + \left[ - C_\mathrm{A}\frac{\pi^2}{10} +{\cal O}(r^2)
1146: \right] \sigma _z^1 \sigma _z^2 \,.
1147: \end{eqnarray}
1148: %/////////////////////////////////////////////////////////////
1149: Here $r=|\mathbf{d}|/a_\textrm{B}$ and
1150: $C_\textrm{A}=\textrm{A}^2C_\textrm{B}/\textrm{B}^2$. At small
1151: distances the interaction is regular and the amplitudes converge
1152: to constant values, see Figure~\ref{figHM.eps}. The complete
1153: expressions for $\chi_c^m(\mathbf{d})$ and $\eta_c^m(\mathbf{d})$
1154: are also available.\cite{STPb}
1155:
1156: In Figure~\ref{figHM.eps}, we plot the amplitudes of the induced
1157: spin-spin interaction, which has the asymptotic behavior
1158: (\ref{Eq:3D:H-int-asymptotic}), and noise for different values of
1159: the spin-spin separation and $b$, for electron impurity spins in
1160: 3D Si-Ge type structures. The value of $b$ can be controlled via
1161: the applied magnetic field, $b=\mu_\textrm{B}H_z g^*
1162: a_\textrm{B}/c_s$. The temperature dependence of the noise is
1163: insignificant provided $2k_{\textrm{B}}T/\Delta\ll 1$.
1164:
1165: As mentioned earlier, the obtained interaction
1166: (\ref{Eq:3D:H-int-asymptotic}) is always accompanied by noise
1167: coming from the same source, as well as by the direct
1168: interactions of the spins. When the electron wave
1169: functions overlap is negligible, the dominant direct interaction
1170: will be the electromagnetic dipole-dipole one,
1171: %/////////////////////////////////////////////////////////////
1172: \begin{equation}\label{Eq:3D:EM-spin-spin-interaction}
1173: H_\textrm{EM}(\mathbf{d})=\frac{\mu_0 \mu _\textrm{B}^2}{4\pi}
1174: \frac{\sigma _x^1 \sigma _x^2 + \sigma _y^1 \sigma _y^2 -
1175: 2\sigma _z^1 \sigma _z^2}{|\mathbf{d}|^3}\,.
1176: \end{equation}
1177: %/////////////////////////////////////////////////////////////
1178: The comparison of the two interactions and noise is shown in
1179: Figure~\ref{figGeHEM.eps}. We plot the magnitude of the effective
1180: induced interaction (\ref{Eq:3D:H-int-asymptotic}), the
1181: electromagnetic interaction
1182: (\ref{Eq:3D:EM-spin-spin-interaction}), measured by ${\cal
1183: H}_\textrm{EM}\equiv{\mu_0 \mu _\textrm{B}^2}/{4\pi
1184: |\mathbf{d}|^3}$, and a measure of the level of noise, for P-donor
1185: electron spins in Ge. In the plot the coupling constants for
1186: P-donors in Ge were taken as
1187: $C_\textrm{B}=1.3\times{}10^7\,$s$^{-1}$ and
1188: $C_\textrm{A}\approx0$, see Sec.~\ref{sec:M}. It transpires that
1189: the induced interaction can be considerable as compared to the
1190: electromagnetic spin-spin coupling. However the overall
1191: coherence-to-noise ratio is quite poor for Ge. In Si, the level of
1192: noise is lower as compared to the induced interaction. It is
1193: dominated primarily by the adiabatic ($\sigma_z^{1,2}$) term.
1194: However, the overall amplitude of the induced terms compares less
1195: favorably with the electromagnetic coupling.
1196: %
1197: \Fig{figHM.eps}{The magnitude of the induced spin-spin interaction
1198: for a 3D Si-Ge type structure: the dominant interaction amplitude,
1199: which is the same for the $x$ and $y$ spin components, is shown.
1200: The arrow indicates increasing $b$ values for the curves shown,
1201: with $b=\Delta a_{\rm B} /c_s=0.03$, $0.09$, $0.15$, $0.21$, $0.27$, $0.33$. The inset
1202: estimates the level of the noise (for $2k_{\textrm{B}}Ta_{\rm B}/c_s=0.01$): the
1203: bottom curve is $\eta_c^{m}(0)$, with $m=x,y$. For $b \leq 0.4$,
1204: the amplitude $\eta_s^{m}(\mathbf{d})$ can be comparable, and its
1205: values, calculated numerically for $0\leq r\leq 50$, are shown as
1206: long as they exceed $\eta_c^{m}(0)$, with the top curve
1207: corresponding to the maximum value, at $r=0$.}
1208: %
1209: \Fig{figGeHEM.eps}{The magnitudes, measured in units of s$^{-1}$,
1210: of the induced spin-spin interaction, the dipole-dipole coupling strength,
1211: and the level of noise for P-impurity electron spins in Ge. Here
1212: $H_z=3 \times 10^4\,$G, and low temperature,
1213: $2k_\textrm{B}T/\Delta \ll 1$, was assumed.}
1214:
1215: Let us estimate the development of entanglement for the P-in-Si
1216: case, considering only the bath-induced effects.
1217: Taking $\left|++\right\rangle$ as the initial state, one expects
1218: entanglement to develop due to the ${\cal H}_\ind{int} \sigma_z^1
1219: \sigma_z^2$ interaction term. The corresponding coupling constant
1220: is $C_\textrm{B}=7800\,$s$^{-1}$. At distances of about
1221: $12a_\ind{B}$ this part of the interaction can be already well
1222: approximated by ${\cal H}_\ind{int} \sim 384\pi ^2 C_\textrm{A}
1223: \frac{2 a_\ind{B}^5} {{|\mathbf{d}|^5 }}\sim 100\,$s$^{-1}$.
1224: The $x$ and $y$ components of the
1225: interaction are smaller by a factor $10^{-3}$ and will not be
1226: considered. The pure state would then be developing as
1227: $\left|\psi(t)\right\rangle= \left|++\right\rangle \cos {\cal
1228: H}_\ind{int}t + \left|--\right\rangle i\sin {\cal H}_\ind{int}t$.
1229: At times $t_E=\pi/4{\cal H}_\ind{int},3\pi/4{\cal
1230: H}_\ind{int},\ldots$, the maximally entangled (Bell) states are
1231: obtained, with the first occurring at $\simeq 0.8 \cdot 10^{-2}\,$s, and
1232: then at intervals of $\simeq 1.6 \cdot 10^{-2}\,$s. The actual
1233: entanglement is lowered by (i) the short-time (adiabatic) decoherence
1234: term, (ii) the large-time relaxation amplitudes (rates). The
1235: latter are due to the $x,y$ components of the qubit-bath coupling
1236: and are of order $\sim 1\,$s$^{-1}$. As a result the first peak of the
1237: concurrence is lowered by a factor of $\sim \exp(-0.8 \cdot 10^{-2})$,
1238: and the subsequent peaks decrease by $\sim \exp(-1.6 \cdot 10^{-2})$ per
1239: each cycle.
1240:
1241: \hphantom{A}
1242:
1243: \section{DISCUSSION OF OPEN PROBLEMS}\label{sec:discuss}
1244:
1245: We have studied the induced indirect exchange interaction due to a
1246: bosonic bath which also introduces quantum noise. At certain time
1247: scales the induced two-qubit interaction is sufficiently strong to
1248: produce significant coherent effects. This interaction can be a
1249: factor to be considered in designs of solid state (as well as
1250: ion-trap based) qubit registers. Even more importantly, the fact
1251: that noise-inducing environment can also yield coherent features
1252: in the dynamics of open quantum systems poses several interesting
1253: fundamental challenges.
1254:
1255: The usual approach to
1256: thermalization\cite{VKampen,Louisell,Blum,Abragam,Slichter} has
1257: been to assume that, for large enough times, the time evolution of
1258: the system plus bath is not just covered by the combined
1259: Hamiltonian, but is supplemented by the instantaneous
1260: loss-of-memory (Markovian) approximation, which introduces
1261: irreversibility and imposes the bath temperature on the reduced
1262: system dynamics in the infinite-time limit, which is then
1263: approached as the density matrix elements assume their thermal
1264: values, according to
1265: %/////////////////////////////////////////////////////
1266: \begin{equation} \label{Eq:Dis:STh}
1267: \rho _S (t) \to {{e^{ - H_S /kT} }\Big / {{{Tr}}\left( {e^{ - H_S
1268: /kT} } \right)}} \,,
1269: \end{equation}
1270: %/////////////////////////////////////////////////////
1271: with exponential relaxation (diagonal) and decoherence
1272: (off-diagonal) rates defining the time scales $T_{1,2}$. However,
1273: it turns out that the traditional phenomenological no-memory
1274: approximations, yielding thermalization, the Fermi golden rule for
1275: the transition rates, etc., assume in a way too strong a memory
1276: loss: they erase the possible bath contributions to the coherent
1277: part of the dynamics at shorter times, such as the Lamb shift for
1278: a single system as well as the induced RKKY interactions for a
1279: bi-partite system. Indeed, while relaxation leading to
1280: (\ref{Eq:Dis:STh}) is driven by the ``on-shell'' exchanges with
1281: the bath, it is the memory of (correlation, entanglement with) the
1282: bath modes that drives, via virtual exchanges, the induced
1283: interaction. Actually, the ``on-shell'' condition, imposed by the
1284: so-called secular \red{approximation,\cite{Koppens} also} underestimates
1285: additional decoherence at short times$\,$---$\,$the ``pure'' or
1286: ``adiabatic'' contribution to the off-diagonal
1287: dephasing$\,$---$\,$that has thus been estimated by using other
1288: approaches.\cite{FFPReview,Privman,Privman2,Tolkunov2,Paladino}
1289:
1290: Perhaps the simplest way to recognize the ambiguity is to ask if
1291: the Hamiltonian in (\ref{Eq:Dis:STh}) should have included the
1292: bath-induced interaction terms (not shown)? Should the final
1293: Boltzmann distribution correspond to the energy levels/basis
1294: states of the original ``bare'' system or the one with the
1295: RKKY-interaction/Lamb shifts, and more generally, a
1296: bath-renormalized, ``dressed'' system?
1297:
1298: There is presently no consistent treatment that will address in a
1299: satisfactory way all the expected \textit{physics} of the
1300: bath-mode effects on the dynamics. The issue is partly technical,
1301: because we are after a \textit{tractable}, rather than just a
1302: ``foundational'' answer. It is well accepted that the emergence of
1303: irreversibility cannot be fully treated within tractable and
1304: calculationally convenient approaches derived directly from the
1305: microscopic dynamics: phenomenology has to be appealed to.
1306: However, even allowing for phenomenological solutions, \red{most of the
1307: known tractable Markovian-approximation-involving schemes that allow for
1308: the emergence of the indirect exchange interaction,
1309: treat thermalization in a cavalier way, yielding typically the
1310: noise term corresponding to $T$-dependent relaxation, but
1311: in the strict $t\to \infty$ limit resulting
1312: in the completely random ($T=\infty$) density matrix (proportional
1313: to the unit operator). This then
1314: avoids the issue of which $H_S$ should
1315: enter in (\ref{Eq:Dis:STh}). And, as mentioned, the established
1316: schemes that yield a more realistic, thermalized
1317: density matrix at $t=\infty$, lose some
1318: intermediate- and short-time dynamical effects.}
1319:
1320: Thus, we have discussed the challenges in formulating unified
1321: treatments that will cover all the (or just most of the
1322: interesting) dynamical effects, over several time scales, from
1323: short to intermediate times (for induced interaction effects and
1324: pure decoherence) to large times (for the onset of
1325: thermalization), while providing a tractable calculational
1326: (usually perturbative, many-body) scheme. This discussion also
1327: alludes to several other interesting conceptual challenges in the
1328: theory of open quantum systems.
1329:
1330: Let us presently comment on the issue of the bath-mode
1331: interactions with each other, as well as with impurities, the
1332: latter particularly important and experimentally
1333: relevant\cite{Sakr,JiangPC} for conduction electrons as carriers
1334: of the indirect-exchange interaction. Indeed, the traditional
1335: treatment of open quantum systems has assumed noninteracting bath
1336: modes. When the bath mode interactions had to be accounted for,
1337: \red{the added effects were treated either perturbatively,\cite{MPV}
1338: or, for strong interaction, such as Luttinger-liquid electrons in
1339: a 1D channel, the collective excitations were taken\cite{MDP} as the
1340: new ``bath modes.''}
1341: Generally, however, especially in Markovian
1342: schemes, one has to seek approaches that do not involve certain
1343: double-counting. Indeed, the assumption that the bath modes are at
1344: a fixed temperature, could be possibly considered as partially
1345: accounting for the effects of the mode-mode and mode-impurity
1346: interactions, because these interactions can contribute to
1347: thermalization of the bath, on par with other influences external
1348: to the bath. This may be particularly relevant for phonons that
1349: always have strong anharmonicity for any real material. In a way
1350: this problem fits with the previous one: we are dealing with
1351: effects that can be, on one hand, modeled by added terms in the
1352: total Hamiltonian but on the other hand, may be also mixed in the
1353: process of thermalization that is modeled by actually departing
1354: from the Hamiltonian description and replacing it with Liouville
1355: equations that include noise effects. While all this sounds
1356: somewhat ``foundational,'' recent experimental advances,
1357: interestingly, bring these challenges to the level of application
1358: that requires tractable, albeit perhaps phenomenological solutions
1359: that can be directly confronted with experimental data.
1360:
1361: There are other interesting topics to be considered, for instance,
1362: the question of whether additional sources of quantum noise are
1363: possible? \red{For instance, it has been recently established\cite{MM} that potential
1364: difference between two leads (reservoirs, or baths, of electrons)
1365: can be a source of quantum noise with the potential difference
1366: playing the role of the temperature parameter.}
1367:
1368: As \red{an example} of a more practical issue to be investigated,
1369: let us mention the possible effect of the sample geometry on
1370: phonon and conduction electron induced relaxation and
1371: interactions. \red{The one-dimensional aspect of the electron gas in a
1372: channel has already been explored}.\cite{MDP} Indeed,
1373: electrons are easy to confine by gate potentials. The situation
1374: for phonons, however, is less clear: can geometrical effects
1375: modify, and particularly reduce, their quantum-noise generation
1376: capacity, or change the induced interactions? Our preliminary
1377: studies \red{reviewed} in this article, seem to indicate strong overall
1378: geometry dependence of the exchange interactions. However, the
1379: situation is not entirely clear, especially for the strength of
1380: the noise effects, and requires a full scale exploration because
1381: recent experiments with double dot nanostructures in Si
1382: membranes\cite{ZhangEtAll} suggest that true nanosize confinement
1383: (due to the sample dimensions) of otherwise long-wavelength modes
1384: (in the transverse sample dimensions) is now possible and will
1385: have dramatic effect on the phonon spectrum and, as a result, on
1386: those physical phenomena that depend on the phonon interactions
1387: with electron spins.
1388:
1389: \acknowledgments
1390:
1391: We thank D.\ Tolkunov for collaborations and useful
1392: discussions, and acknowledge funding by the NSF under grant
1393: DMR-0121146.
1394:
1395: \hphantom{A}
1396:
1397: %//////////////////////////////////////////////////////////
1398: %//////////////////////////////////////////////////////////
1399: %//////////////////////////////////////////////////////////
1400:
1401: \begin{thebibliography}{99}{\frenchspacing
1402:
1403: \bibitem{STPs} D. Solenov, D. Tolkunov, and V. Privman, Phys. Lett. A {\bf 359}, 81 (2006).
1404:
1405: \bibitem{STPb} D. Solenov, D. Tolkunov, and V. Privman, Phys. Rev. B \textbf{75}, 035134 (2007).
1406:
1407: \bibitem{Jiang} M. Xiao, I. Martin, E. Yablonovitch, and H. W. Jiang, Nature \textbf{430}, 435
1408: (2004).
1409:
1410: \bibitem{Jiang2} M. R. Sakr, H. W. Jiang, E. Yablonovitch, and E. T. Croke,
1411: Appl. Phys. Lett. \textbf{87}, 223104 (2005).
1412:
1413: \bibitem{Craig} N. J. Craig, J. M. Taylor, E. A. Lester, C. M. Marcus, M. P. Hanson, and
1414: A. C. Gossard, Science \textbf{304}, 565 (2004).
1415:
1416: \bibitem{Elzerman} J. M. Elzerman, R. Hanson, L. H. Willems van Beveren, B. Witkamp, L. M. K. Vandersypen, and L. P. Kouwenhoven, Nature \textbf{430}, 431 (2004).
1417:
1418: \bibitem{Koppens} F. H. L. Koppens, C. Buizert, K. J. Tielrooij, I. T. Vink, K. C.
1419: Nowack, T. Meunier, L. P. Kouwenhoven, and L. M. K. Vandersypen,
1420: Nature {\bf 442}, 766 (2006).
1421:
1422: \bibitem{Petta} J. R. Petta, A. C. Johnson, J. M. Taylor, E. A. Laird, A. Yacoby, M. D. Lukin, C. M. Marcus, M. P. Hanson, and A. C. Gossard, Science {\bf 309}, 2180 (2005).
1423:
1424: \bibitem{MMJ} I. Martin, D. Mozyrsky, and H. W. Jiang, Phys. Rev. Lett. {\bf 90},
1425: 018301 (2003).
1426:
1427: \bibitem{PF} E. Prati, M. Fanciulli, A. Kovalev,
1428: J. D. Caldwell, C. R. Bowers, F. Capotondi, G. Biasiol, and L.
1429: Sorba, IEEE Trans. Nanotechnol. {\bf 4}, 100 (2005).
1430:
1431: %NEW
1432: \bibitem{Nakamura} Y. Nakamura, Y. A. Pashkin, and J. S. Tsai, Nature (London) \textbf{398},
1433: 786 (1999).
1434:
1435: %NEW
1436: \bibitem{Vion} D. Vion, A. Aassime, A. Cottet, P. Joyez, H. Pothier, C. Urbina,
1437: D. Esteve, and M. H. Devoret, Science \textbf{296}, 886 (2002).
1438:
1439: %NEW
1440: \bibitem{Chiorescu} I. Chiorescu, Y. Nakamura, C. J. P. Harmans, and J. E. Mooij,
1441: Science \textbf{299}, 1869 (2003).
1442:
1443: %NEW
1444: \bibitem{Yamamoto} T. Yamamoto, Y. A. Pashkin, O. Astafiev, Y. Nakamura, and J. S.
1445: Tsai, Nature (London) \textbf{425}, 941 (2003).
1446:
1447: \bibitem{Wootters1} S. Hill and W. K. Wootters, Phys. Rev. Lett. \textbf{78}, 5022 (1997).
1448:
1449: \bibitem{Wootters2} W. K. Wootters, Phys. Rev. Lett. \textbf{80}, 2245 (1998).
1450:
1451: \bibitem{Mahan} G. D. Mahan, \emph{Many-Particle Physics\/} (Kluwer Academic, New York, 2000).
1452:
1453: \bibitem{Hasegawa} H. Hasegawa, Phys. Rev. \textbf{118}, 1523 (1960).
1454:
1455: \bibitem{Roth} L. M. Roth, Phys. Rev. \textbf{118}, 1534 (1960).
1456:
1457: \bibitem{SO-Winkler} R. Winkler, \emph{Spin-Orbit Coupling Effects in Two-Dimentional
1458: Electron and Hole Systems\/} (Springer, New York, 2003).
1459:
1460: \bibitem{PGCZ} T. Pellizzari, S. A. Gardiner, J. I. Cirac, and P.
1461: Zoller, Phys. Rev. Lett. \textbf{75}, 3788 (1995).
1462:
1463: \bibitem{MKGB} D. Mozyrsky, Sh. Kogan, V. N. Gorshkov, and G. P. Berman, Phys. Rev. B \textbf{65}, 245213 (2002).
1464:
1465: \bibitem{PBS} M. Asheghi, Y. K. Leung, S. S. Wong, and K. E. Goodson, Appl. Phys. Lett. \textbf{71}, 1798 (1997).
1466:
1467: \bibitem{Leibfried} D. Leibfried, R. Blatt, C. Monroe, and D. Wineland, Rev. Mod. Phys. \textbf{75}, 281 (2003).
1468:
1469: \bibitem{Marquet} C. Marquet, F. Schmidt-Kaler, and D. F. V. James, Appl. Phys. B {\bf 76}, 199 (2003).
1470:
1471: \bibitem{Porras} D. Porras and J. I. Cirac, Phys. Rev. Lett. \textbf{92}, 207901 (2004).
1472:
1473: \bibitem{Leggett} A. J. Leggett, S. Chakravarty, A. T. Dorsey, M. P. A. Fisher, A.
1474: Garg, and W. Zwerger, Rev. Mod. Phys. \textbf{59}, 1 (1987).
1475:
1476: \bibitem{Privman} V. Privman, Modern Phys. Lett. B \textbf{16}, 459 (2002).
1477:
1478: \bibitem{Tolkunov2} D. Tolkunov and V. Privman, Phys. Rev. A \textbf{69}, 062309 (2004).
1479:
1480: \bibitem{Solenov} D. Solenov and V. Privman, Int. J. Modern Phys. B \textbf{20}, 1476 (2006).
1481:
1482: \bibitem{Louisell} W. H. Louisell, \emph{Quantum Statistical Properties of Radiation\/} (Wiley, New York, 1973).
1483:
1484: \bibitem{PALMA} G. M. Palma, K.-A. Suominen, and A. K. Ekert, Proc. R. Soc. London Ser. A \textbf{452}, 576 (1996).
1485:
1486: \bibitem{VKampen} N. G. van Kampen, \emph{Stochastic Processes in Physics and Chemistry\/} (North-Holland, Amsterdam, 2001).
1487:
1488: %NEW
1489: \bibitem{Hanggi} R. Doll, M. Wubs, P. H\"{a}nggi, and S. Kohler, e-print: cond-mat/0703075
1490: (at www.arxiv.org).
1491:
1492: \bibitem{Bennett} C. H. Bennett, D. P. DiVincenzo, J. A. Smolin, and W. K. Wootters, Phys. Rev. A \textbf{54}, 3824 (1996).
1493:
1494: \bibitem{Vedral} V. Vedral, M. B. Plenio, M. A. Rippin, and P. L. Knight, Phys. Rev. Lett. \textbf{78}, 2275 (1997).
1495:
1496: \bibitem{Eberly1} T. Yu and J. H. Eberly, Phys. Rev. B \textbf{68}, 165322 (2003).
1497:
1498: \bibitem{Eberly2} T. Yu and J. H. Eberly, Phys. Rev. Lett. \textbf{93}, 140404 (2004).
1499:
1500: \bibitem{Braun} D. Braun, Phys. Rev. Lett. \textbf{89}, 277901 (2002).
1501:
1502: \bibitem{Blum} K. Blum, \emph{Density Matrix Theory and Applications\/} (Plenum Press, New York, 1996).
1503:
1504: \bibitem{Abragam} A. Abragam, \emph{Principles of Nuclear Magnetism} (Clarendon Press, 1983).
1505:
1506: \bibitem{Slichter} C. P. Slichter, \emph{Principles of Magnetic Resonance} (Springer, 1991).
1507:
1508: \bibitem{FFPReview} A. Fedorov, L. Fedichkin, and V. Privman, J. Comp. Theor. Nanosci. \textbf{1}, 132 (2004).
1509:
1510: \bibitem{Privman2} V. Privman, J. Stat. Phys. \textbf{110}, 957 (2003).
1511:
1512: \bibitem{Paladino} E. Paladino, M. Sassetti, G. Falci, and U. Weiss, Chem. Phys. \textbf{322}, 98 (2006).
1513:
1514: \bibitem{Sakr} M. R. Sakr, E. Yablonovitch, E. T. Croke, and H. W. Jiang, e-print: cond-mat/0504046
1515: (at www.arxiv.org).
1516:
1517: \bibitem{JiangPC} H. W. Jiang, private communication.
1518:
1519: \bibitem{MPV} D. Mozyrsky, V. Privman, and I. D. Vagner, Phys. Rev. B \textbf{63}, 085313 (2001).
1520:
1521: \bibitem{MDP} D. Mozyrsky, A. Dementsov, and V. Privman, Phys. Rev. B \textbf{72}, 233103 (2005).
1522:
1523: \bibitem{MM} D. Mozyrsky and I. Martin, Phys. Rev. Lett. \textbf{89}, 018301 (2002).
1524:
1525: \bibitem{ZhangEtAll} P. Zhang, E. Tevaarwerk, B.-N. Park, D. E. Savage, G. K. Celler, I. Knezevic, P. G. Evans, M. A. Eriksson, and M. G. Lagally, Nature \textbf{439}, 703 (2006).
1526:
1527:
1528:
1529: }\end{thebibliography}
1530:
1531: \end{document}
1532: