1: \documentclass[aps,prb,twocolumn,showpacs,amsmath]{revtex4}

2: %\documentclass[preprint,aps,prb,showpacs,amsmath]{revtex4}

3: \usepackage{latexsym}

4: \usepackage{amssymb}

5: \usepackage{graphicx}

6: \usepackage{bm}% bold math

7: \usepackage[english]{babel}

8: \usepackage[latin1]{inputenc}

9: \usepackage{amsmath}

10: \usepackage{amsfonts}

11: 

12: \def\stackunder#1#2{\mathrel{\mathop{#2}\limits_{#1}}}

13: \newcommand{\pdag}{{\phantom{\dagger}}}

14: \newcommand{\bq}{\begin{equation}}

15: \newcommand{\eq}{\end{equation}}

16: \newcommand{\bn}{\begin{eqnarray}}

17: \newcommand{\en}{\end{eqnarray}}

18: 

19: 

20: \begin{document}

21: 

22: \title{Ac-cotunneling through an interacting quantum dot in a magnetic field}

23: 

24: \author{Bing Dong}

25: \affiliation{Department of Physics, Shanghai Jiaotong University,

26: 1954 Huashan Road, Shanghai 200030, China}

27: 

28: \author{X.L. Lei}

29: \affiliation{Department of Physics, Shanghai Jiaotong University,

30: 1954 Huashan Road, Shanghai 200030, China}

31: 

32: \author{N. J. M. Horing}

33: \affiliation{Department of Physics and Engineering Physics, Stevens Institute of Technology, %%@

34: Hoboken, New Jersey 07030, USA}

35: 

36: \begin{abstract}

37: 

38: We analyze inelastic cotunneling through an interacting quantum dot subject to an ambient %%@

39: magnetic field in the weak tunneling regime under a non-adiabatic time-dependent %%@

40: bias-voltage. Our results clearly exhibit photon-assisted satellites and an overall %%@

41: suppression of differential conductance with increasing driving amplitude, which is %%@

42: consistent with experiments. We also predict a zero-anomaly in differential conductance under %%@

43: an appropriate driving frequency.       

44: 

45: \end{abstract}

46: 

47: \date{\today}

48: 

49: \pacs{72.40.+w, 73.23.Hk, 73.63.Kv, 03.65.Yz}

50: 

51: \maketitle

52: 

53: Recently, cotunneling\cite{Averin} through discrete levels, i.e., a quantum dot (QD), has %%@

54: attracted much attention since it determines the intrinsic limitation of accuracy of %%@

55: single-electron transistors due to leakage, and since it also involves correlation effects, %%@

56: such as the Kondo effect.\cite{cotunnelingExp} It has also been reported %%@

57: experimentally\cite{Kogan} and theoretically\cite{Ng,Goldin,Lopez,Kaminski} that external %%@

58: microwave irradiation can induce the occurrence of Kondo satellites and an overall %%@

59: suppression of the Kondo peak. However, there are few studies so far concerning %%@

60: time-dependent second-order cotunneling in the weak tunneling regime at temperatures above %%@

61: the Kondo temperature. About ten years ago, Flensberg presented an analysis for coherent %%@

62: photon-assisted cotunneling in a double-junction Coulomb blockade device in the adiabatic %%@

63: limit.\cite{Flensberg}

64: In this letter, we will further study the cotunneling in an interacting QD when an ac %%@

65: bias-voltage is applied between two electrodes in the non-adiabatic regime.  

66: 

67: We employ the s-d exchange Hamiltonian to model inelastic cotunneling through a QD in an %%@

68: ambient magnetic field, $B$, in the weak-coupling regime:

69: \begin{align}

70: H=& \,H_{0}+H_{\mathrm{I}}, & &  \label{Hamiltonian} \\

71: H_{0}=& \,\sum_{\eta \mathbf{k}\sigma }\varepsilon _{\eta \mathbf{k}}(t) c_{\eta %%@

72: \mathbf{k}\sigma }^{\dag }c_{\eta \mathbf{k} \sigma}^{{\phantom{\dagger}}}-\Delta_0 S^{z},\cr 

73: H_{\mathrm{I}}= & \,\sum_{\eta ,\eta ^{\prime },\mathbf{k}

74: ,\mathbf{k}^{\prime }}J \bigl [\bigl(c_{\eta \mathbf{k}

75: \uparrow }^{\dag }c_{\eta ^{\prime }\mathbf{k}^{\prime } \uparrow}^{{\phantom{\dagger}}} - %%@

76: c_{\eta \mathbf{k}\downarrow }^{\dag }c_{\eta ^{\prime }

77: \mathbf{k}^{\prime }\downarrow }^{{\phantom{\dagger}}}\bigr)S^{z}\cr&

78: \,+c_{\eta \mathbf{k}\uparrow }^{\dag }c_{\eta ^{\prime }\mathbf{k}^{\prime} %%@

79: \downarrow}^{{\phantom{\dagger}}}S^{-} + c_{\eta \mathbf{k}\downarrow}^{\dag} %%@

80: c_{\eta^{\prime} \mathbf{k}^{\prime }\uparrow }^{{\phantom{\dagger}}} S^{+}\bigr] + H_{\rm %%@

81: dir}, \cr

82: H_{\rm dir}= &\, J_{\rm d} \sum_{\sigma} \bigl ( c_{L {\bf k} \sigma}^\dagger + c_{R {\bf k} %%@

83: \sigma}^\dagger \bigr ) \bigl ( c_{L {\bf k} \sigma}^\pdag + c_{R {\bf k} \sigma}^\pdag \bigr %%@

84: ),  \notag

85: \end{align}

86: where $c_{\eta \mathbf{k}\sigma }^{\dagger }$ ($c_{\eta \mathbf{k}\sigma }$)

87: is the creation (annihilation) operator for electrons with momentum $\mathbf{k}$, %%@

88: spin-$\sigma$ in lead $\eta$ ($=\mathrm{L,R}$). The energies $\varepsilon _{\eta %%@

89: \mathbf{k}}(t)= \varepsilon _{\eta \mathbf{k}}^0+eV_{\eta}(t)$ include a rigid shift of the %%@

90: Fermi energy of the electrons in the leads due to the applied time-dependent bias-voltage %%@

91: $V_{\eta}(t)=V_{\eta}^0 + v_{\eta}\cos(\Omega t)$ with $V_{\eta}^0$ ($v_{\eta}$) being the %%@

92: amplitude of the dc(ac) part of the bias-voltage. Here, we assume that the Fermi energies of %%@

93: two leads are zero at equilibrium and $V_{L}^0=-V_{R}^0=eV_0/2$.   

94: $\Delta_0=g_e\mu _{B}B$ is the static magnetic-field $B$-induced Zeeman energy. 

95: ${\bf S}\equiv (S^x, S^y, S^z)$ are Pauli spin operators of electrons in the QD %%@

96: [$S^{\pm}\equiv S^x \pm i S^y$], and $J$ is the exchange coupling constant. 

97: $H_{\rm dir}$ is the potential scattering term with $2J_{\rm d}=J$.      

98: As in our previous paper,\cite{Dong} we can rewrite the tunneling term, $H_{\rm I}$ in %%@

99: Eq.~(\ref{Hamiltonian}), as a sum of three products of two variables: 

100: \bq

101: H_{\rm I}=Q^z S^z+ Q^+ S^- + Q^- S^+ + Q^{\hat 1},

102: \eq

103: with the generalized coordinates $Q^{z(\pm)}$ of reservoir variables as 

104: \begin{align}

105: Q^{z}=&\, \sum_{\eta,\eta'} Q_{\eta\eta'}^{z}= \sum_{\eta,\eta', {\bf k},{\bf k}'} J \bigl( %%@

106: c_{\eta {\bf k} \uparrow}^\dag c_{\eta' {\bf k}' \uparrow}^\pdag - c_{\eta {\bf k} %%@

107: \downarrow}^\dag c_{\eta' {\bf k}' \downarrow}^\pdag \bigr), \label{Qz} \\

108: Q^{+}=&\, \sum_{\eta,\eta'} Q_{\eta\eta'}^{+}= \sum_{\eta, \eta', {\bf k},{\bf k}'} J c_{\eta %%@

109: {\bf k} \uparrow}^\dag c_{\eta' {\bf k}' \downarrow}^\pdag, \label{Q+}\\

110: Q^{-}=&\, \sum_{\eta,\eta'} Q_{\eta\eta'}^{-}= \sum_{\eta, \eta', {\bf k},{\bf k}'} J c_{\eta %%@

111: {\bf k} \downarrow}^\dag c_{\eta' {\bf k}' \uparrow}^\pdag, \label{Q-}

112: \end{align}

113: and $Q^{\hat 1}=H_{\rm dir}$.

114: In the following, we will use units where $\hbar=k_{B}=e=1$.

115: 

116: As in our previous studies of inelastic cotunneling through an interacting QD in the weak %%@

117: tunneling limit, we employ a generic quantum Langevin equation %%@

118: approach\cite{Ackerhalt,Cohen,Smirnov} to establish a set of quantum Bloch equations for the %%@

119: description of the dynamics of a single spin [modeled by Eq.~(\ref{Hamiltonian})] explicitly %%@

120: in terms of the response and correlation functions of free reservoir variables. This %%@

121: procedure provides explicit analytical expressions for the nonequilibrium magnetization and %%@

122: cotunneling current for arbitrary dc bias-voltage and temperature.\cite{Dong,Dong1} Here, we %%@

123: generalize our previous derivations to the time-dependent case in the non-adiabatic and %%@

124: high-frequency regime. 

125: 

126: In the derivation, we proceed with the Heisenberg equation of motion for the Pauli spin %%@

127: operators and the lead operators, and then formally integrate these equations from the %%@

128: initial time $0$ to $t$ exactly to all orders of $J$. Next, under the assumption that the %%@

129: time scale of decay processes is much slower than that of free evolution, we replace the %%@

130: time-dependent

131: operators involved in the integrals of these EOM's approximately in terms of their free %%@

132: evolutions. Thirdly, these EOM's are expanded in powers of $J$ up to second order, resulting %%@

133: in non-Markovian dynamic equations for the time evolution of the QD spin variables in a %%@

134: compact form:

135: \bn

136: \dot S^{z}(t)&=& -2 \int_{-\infty}^t dt' \left ( e^{-i\Delta_0 \tau} + e^{i\Delta_0 \tau} %%@

137: \right ) C(t,t') S^{z}(t') \cr

138: && - \int_{-\infty}^t dt' \left ( e^{-i\Delta_0 \tau} - e^{i\Delta_0 \tau} \right ) R(t,t'), %%@

139: \label{eom:sz} \\

140: \dot S^{\pm}(t)&=& \mp i \Delta_0\, S^\pm(t) - 2 \int_{-\infty}^t dt' C(t,t') S^{\pm}(t') \cr

141: && - 2 \int_{-\infty}^t dt' e^{\mp i\Delta_0 \tau} C(t,t') S^{\pm}(t'), \label{eom:spm}

142: \en

143: with $\tau=t-t'$. The correlation function, $C(t,t^{\prime })$, and 

144: the response function, $R(t,t^{\prime })$,

145: of free reservoir variables (tagged by subscript ``$o$") are defined as:

146: \bq

147: C(R)(t,t^{\prime }) ={\frac{1}{2}}\theta (\tau )\langle [ Q^{\pm}_{o}(t), %%@

148: Q^{\mp}_{o}(t^{\prime })]_{+(-)}\rangle.  \label{randcf}

149: \eq

150: 

151: Special attention must be paid to the free reservoir variables due to the time-dependent %%@

152: energies $\varepsilon _{\eta \mathbf{k}}(t)$:

153: \bq

154: c_{\eta {\bf k} \sigma}^o(t) = e^{-i\int_{t'}^t \varepsilon_{\eta {\bf k}}(\tau) d\tau} %%@

155: c_{\eta {\bf k} \sigma}(t').

156: \eq

157: Therefore, the kernels $C(R)(t,t')$ become {\em double-time-dependent} functions due to the %%@

158: lack of time-translation-invariance stemming from the ac-bias:

159: \bn

160: C(R)(t,t') &=& {1\over 2}\theta(\tau) \langle [Q_{o}^{+}(t), Q_{o}^{-}(t')]_\pm \rangle \cr

161: &=& {1\over 2} \theta(\tau) J^2 \sum_{\eta,\eta',\xi,\xi'} \sum_{{\bf k},{\bf k}',{\bf %%@

162: q},{\bf q}'} \langle [c_{\eta{\bf k} \uparrow}^\dagger (t) c_{\eta' {\bf k}' %%@

163: \downarrow}^\pdag (t), \cr

164: && c_{\xi {\bf q} \downarrow}^\dagger(t') c_{\xi' {\bf q}' \uparrow}^\pdag (t')]_\pm \rangle %%@

165: \cr

166: &=& {1\over 2} \theta(\tau) J^2 \cr

167: && \times \sum_{\eta,\eta',\xi,\xi'} \sum_{{\bf k},{\bf k}',{\bf q},{\bf q}'} e^{i\int_{t'}^t %%@

168: d\tau [\epsilon_{\xi'{\bf q}'}(\tau)- \epsilon_{\xi {\bf q}}(\tau)] } \cr

169: && \times \left [ \langle c_{\eta {\bf k} \uparrow}^\dagger (t) c_{\xi' {\bf q}' %%@

170: \uparrow}^\pdag (t) \rangle \langle c_{\eta' {\bf k}' \downarrow}^\pdag (t) c_{\xi {\bf q} %%@

171: \downarrow}^\dagger (t) \rangle  \right. \cr

172: && \left. \pm \langle c_{\xi {\bf q} \downarrow}^\dagger (t) c_{\eta' {\bf k}' %%@

173: \downarrow}^\pdag (t) \rangle \langle c_{\xi' {\bf q}' \uparrow}^\pdag (t) c_{\eta {\bf k} %%@

174: \uparrow}^\dagger (t) \rangle \right ]\cr

175: &=& \frac{1}{2}\theta(\tau) \sum_{\eta} g \int d\epsilon d\epsilon' e^{i(\epsilon-\epsilon') %%@

176: \tau} \cr

177: && \times \left \{ f_{\eta}(\epsilon) \left [ 1-f_{\eta}(\epsilon') \right ] \pm %%@

178: f_{\eta}(\epsilon') \left [ 1-f_{\eta}(\epsilon) \right ] \right \} \cr

179: && + \frac{1}{2}\theta(\tau) g \int d\epsilon d\epsilon' \left [ %%@

180: e^{i(\epsilon-\epsilon')\tau} \right. \cr

181: && \times e^{i\int_{t'}^t d\tau' V_{\rm ac} \cos(\Omega \tau')} \pm  %%@

182: e^{-i(\epsilon-\epsilon')\tau} \cr

183: && \left. \times e^{-i\int_{t'}^t d\tau' V_{\rm ac} \cos(\Omega \tau')} \right ] 

184: \left \{ f_{L}(\epsilon) \left [ 1-f_{R}(\epsilon') \right ] \right. \cr

185: && \left. \pm f_{R}(\epsilon') \left [ 1-f_{L}(\epsilon) \right ] \right \},

186: \en

187: with $g\equiv J^{2} \rho_{0}^{2}$, $V_{\rm ac}=v_L-v_R$, and the Fermi-distribution function %%@

188: is $f_{\eta}(\epsilon)=\left [1+ e^{(\epsilon-\mu_{\eta})/T} \right ]^{-1}$ ($T$ is the %%@

189: temperature). Here we assume the two electrodes to be Markov-type reservoirs with a constant %%@

190: density of states $\rho_0$. The kernels reduce exactly to our previous results, Eq.~(B8) in %%@

191: Ref.~\onlinecite{Dong}, if there is no ac-bias or with the same ac amplitude in the left and %%@

192: right leads. 

193: 

194: In the presence of a periodic ac-bias, the spin variables $S^{z(\pm)}(t)$ naturally depend %%@

195: periodically on $t$ with a period ${\cal T}_{\rm ac}=2\pi/\Omega$. As a result, the full %%@

196: solutions of Eqs.~(\ref{eom:sz}) and (\ref{eom:spm}) can be formally written as a %%@

197: superposition of all harmonics

198: \bq

199: S^{z(\pm)}(t)=\sum_{n=-\infty}^{\infty} S_{(n)}^{z(\pm)} e^{-i n \Omega t}.

200: \eq

201: Employing this expansion in the dynamic equations (\ref{eom:sz}) and (\ref{eom:spm}), an %%@

202: infinite set of linear equations results in which the $S_{(n)}^{z(\pm)}$ are coupled with %%@

203: each other via the kernels. To obtain a solution for the spin variables, one has to terminate %%@

204: this infinite chain at a chosen order and then solve the resulting equations in a recursive %%@

205: way. However, in the limit of high frequencies, $\Omega\gg \Gamma$ (tunneling rate) and $T$ %%@

206: (temperature), of interest in this letter, the ac-bias oscillates so fast that an electron %%@

207: experiences many cycles of the ac-bias during its presence inside the dot, and thus can not %%@

208: sense the details of the dynamics within one period ${\cal T}_{\rm ac}$. In this %%@

209: non-adiabatic limit, one can approximately replace the kernels by a time-average with respect %%@

210: to the center-of-mass of time ${\bar \tau}=t+t'$:\cite{Tien,Vicari}

211: \bq

212: C(R)(t,t')\approx C(R)^{(0)}(\tau) = \frac{1}{{\cal T}_{\rm ac}} \int_0^{{\cal T}_{\rm ac}} d %%@

213: {\bar \tau} C(R)(\tau,\bar \tau).

214: \eq

215: By the same token, one can retain only the stationary part of the spin variables and neglect %%@

216: the rapidly oscillatory parts, leading to detailed balance equations in a Markov %%@

217: approximation by making the replacement $\int_{-\infty}^t d\tau\Rightarrow %%@

218: \int_{-\infty}^\infty d\tau$ in Eqs.~(\ref{eom:sz}) and (\ref{eom:spm}):

219: \bn

220: 0&=& -2 \left (C_\omega^{(0)}(-\Delta_0) + C_\omega^{(0)}(\Delta_0)\right ) S_{(0)}^{z} \cr

221: && + R_\omega^{(0)}(\Delta_0) - R_\omega^{(0)}(-\Delta_0),  \label{fun:sz} \\

222: 0 &=& \mp i \Delta_0\, S_{(0)}^\pm - 2 \left [ C_\omega^{(0)}(0) + C_\omega^{(0)}(\mp %%@

223: \Delta_0) \right ] S_{(0)}^{\pm}. 

224: \en

225: Here, $C(R)_\omega^{(0)}(\omega)$ are the Fourier transforms of the time-averaged kernels %%@

226: $C(R)^{(0)}(\tau)$:     

227: \bn

228: C_\omega^{(0)}(\omega)&=& \pi g T \varphi \left ( \frac{\omega}{T} \right ) + \frac{\pi}{2} g %%@

229: T \sum_{n=-\infty}^{\infty} J_n^2\left ( \frac{V_{\rm ac}}{\Omega} \right ) \cr

230: && \times \left [ \varphi \left ( \frac{\omega + V+ n\Omega}{T}\right ) + \varphi \left ( %%@

231: \frac{\omega- V-n\Omega}{T}\right ) \right ], \cr

232: && \\

233: R_\omega^{(0)}(\omega)&=& \pi g \left [ 1 + \sum _{n=-\infty}^{\infty} J_n^2\left ( %%@

234: \frac{V_{\rm ac}}{\Omega} \right ) \right ] \omega, \label{response}

235: \en

236: with $\varphi(x)\equiv x \coth  ( x/2 )$ and $J_n(x)$ is the Bessel function of order $n$. To %%@

237: derive these equations (\ref{fun:sz})-(\ref{response}), we use the relation

238: \bq

239: e^{ix \sin (\omega t)}=\sum_{n=-\infty}^{\infty} J_{n}(x) e^{in \omega t}.

240: \eq

241: 

242: The solution of Eq.~(\ref{fun:sz}) yields the nonequilibrium magnetization of the QD subject %%@

243: to an ac-bias voltage as

244: \bq

245: S_{(0)}^z = \frac{R_{\omega}^{(0)}(\Delta_0)}{2C_{\omega}^{(0)}(\Delta_0)}. %%@

246: \label{magnetization}

247: \eq

248: This formula is our central result, which can be regarded as a direct generalization of the %%@

249: dc nonequilibrium magnetization\cite{Dong,Parcollet,Paaske2} of a QD under a non-adiabatic %%@

250: high-frequency field. Obviously, it reduces exactly to previous results in absence of %%@

251: ac-bias, $V_{\rm ac}=0$.\cite{Dong,Parcollet,Paaske2} As an illustration, we exhibit in %%@

252: Fig.~1(a) the dependence of the magnetization, $S_{(0)}^z$, on dc bias-voltage for the %%@

253: driving frequency $\Omega/\Delta_0=0.5$. It should be noted that $S_{(0)}^z$ exhibits %%@

254: different behaviors with increasing ac-amplitude $V_{\rm ac}$. For small dc bias voltage, the %%@

255: QD spin is fully polarized due to the nonzero external magnetic field, and it is gradually %%@

256: quenched with increasing dc bias voltage. Application of an ac bias tends to quench the spin %%@

257: polarization more rapidly. This tendency suggests that the ac bias plays a role in dephasing %%@

258: the electronic tunneling processes in a QD, as analyzed in Ref.~\onlinecite{Kaminski}.   

259: 

260: \begin{figure}[htb]

261: \includegraphics [width=8.5cm,height=4cm,angle=0,clip=on]{diac1}

262: \caption{Nonequilibrium magnetization, $S_{(0)}^z$ (a); and the differential conductance, %%@

263: $dI/dV_0$ (in units of $2ge^2/h$) (b); as functions of dc bias-voltage, $V_0$, for various %%@

264: amplitudes of the ac-bias with a fixed driving frequency $\Omega/\Delta_0=0.5$. The %%@

265: temperature we use in the calculation is $T/\Delta_0=0.01$.} \label{fig1}

266: \end{figure}

267: 

268: Proceeding to the calculation of tunneling current, the current operator through the QD is %%@

269: defined as the time rate of change of charge density $N_{\eta}=\sum_{{\bf k},\sigma} %%@

270: c_{\eta{\bf k}\sigma}^\dagger c_{\eta{\bf k} \sigma}^\pdag$ in lead $\eta$: $I_{\eta}(t) = %%@

271: \dot N_{\eta }$. From linear-response theory we have

272: \bq

273: I(t) = \langle I_{\eta }(t) \rangle =-i\int_{-\infty}^t dt' \langle [ J_{\eta }(t), H_{\rm %%@

274: I}(t')]_- \rangle. \label{currentsigma}

275: \eq

276: Because the dc component of current is easily measurable experimentally, we compute the %%@

277: time-averaged current $I = \frac{1}{{\cal T}_{\rm ac}} \int_0^{{\cal T}_{\rm ac}} dt I(t)$. %%@

278: Performing the same high-frequency and Markov approximations as above, we obtain the %%@

279: time-averaged currents:

280: \bn

281: I &=& 4\pi g \sum_{n=-\infty}^{\infty} J_n^2\left ( \frac{V_{\rm ac}}{\Omega} \right ) ( V + %%@

282: n\Omega) + 2 \pi g T S_{(0)}^z \cr

283: && \times \sum_{n=-\infty}^{\infty} J_n^2\left ( \frac{V_{\rm ac}}{\Omega} \right ) 

284: \left [ \varphi \left ( {\Delta_0 -V -n\Omega\over T} \right ) \right. \cr

285: && \left. - \varphi \left ( {\Delta_0 +V +n\Omega\over T} \right ) \right ], \label{ic}

286: \en

287: which is a generalization of Tien-Gordon-type formula in cotunneling current.\cite{Tien} It %%@

288: should be noted that our expression for photon-assisted cotunneling current is valid in the %%@

289: high frequency limit, whereas Flensberg derived a perturbative current under pump conditions %%@

290: in the low-frequency limit.\cite{Flensberg}   

291: 

292: \begin{figure}[htb]

293: \includegraphics [width=8.5cm,height=4cm,angle=0,clip=on]{diac2}

294: \caption{Calculated differential conductance vs. dc-bias. (a) results for a relatively high %%@

295: driving frequency $\Omega/\Delta_0=1.5$ for various ac-amplitudes, as in Fig.~1; (b) the %%@

296: ac-frequency dependence of $dI/dV_0$ for a fixed driving amplitude $V_{\rm ac}/\Delta_0=1.0$. %%@

297: Other parameters are the same as in Fig.~1.} \label{fig2}

298: \end{figure}

299: 

300: We plot the dc bias-voltage-dependent differential conductance, $dI/dV_0$, in Figs.~1(b) and %%@

301: 2. Obviously, the differential conductance shows some satellites at $V_0=\pm %%@

302: (\Delta_0-n\Omega)$ superimposed on the characteristic jump at $V_0=\pm \Delta_0$ in the %%@

303: presence of an ac-bias. These satellites arise physically from photon-assisted spin-flip %%@

304: cotunneling, i.e., albeit $|V_0|<\Delta_0$, the spin-flip cotunneling process can still %%@

305: become energetically activated by an electron absorbing photon quanta to compensate for the %%@

306: energy difference. Moreover, an overall suppression is observed with increasing ac-amplitude, %%@

307: which is qualitatively consistent with the experimental results.\cite{Kogan} More %%@

308: interestingly, we find that $dI/dV_0$ exhibits a transition from peak-splitting to %%@

309: zero-bias-anomaly if the driving frequency is higher than $\Delta_0$. This behavior can be %%@

310: ascribed to the fact that a driving field with an appropriately high frequency can spur %%@

311: spin-flip cotunneling, notwithstanding $|V_0|<(\Omega-\Delta_0)$; in contrast, when the %%@

312: dc-bias increases to $\Delta_0>|V_0|>(\Omega-\Delta_0)$, spin-flip events become inactive %%@

313: instead. The $dI/dV_0$ curve recovers peak-splitting behavior if $\Omega \geq 2\Delta_0$, as %%@

314: shown in Fig.~2(b).   

315: 

316: In summary, we have generalized the generic Langevin equation approach to study %%@

317: photon-assisted cotunneling through an interacting QD in the non-adiabatic and high frequency %%@

318: regime, deriving explicit analytic expressions for the dc components of nonequilibrium %%@

319: magnetization and current with a generalized Tien-Gordon-type form. Our results show that %%@

320: applying an ac-bias is an important method for tuning the $I$-$V$ characteristics. %%@

321: Considering experiments\cite{cotunnelingExp} in which a static magnetic field $B=11$ T is %%@

322: applied, $\Delta_0\simeq 0.1$ meV, and the ac-frequency is $\Omega\sim 12-50$ GHz with %%@

323: ac-amplitude $V_{\rm ac}\leq 0.15$ mV, all the parameters are easily accessible %%@

324: experimentally and they satisfy the non-adiabatic condition $\Omega\gg \Gamma$ ($\sim 30$ %%@

325: $\mu$eV).   

326:                    

327: 

328: This work was supported by Projects of the National Science Foundation of China, the Shanghai %%@

329: Municipal Commission of Science and Technology, the Shanghai Pujiang Program, and Program for %%@

330: New Century Excellent Talents in University (NCET).

331: 

332: 

333: \begin{thebibliography}{99}

334: 

335: \bibitem{Averin}{D.V. Averin and A.A. Odintsov, Phys. Lett. A {\bf 140}, 251 (1989); D.V. %%@

336: Averin and Yu.V. Nazarov, Phys. Rev. Lett. {\bf 65}, 2446 (1990).}

337: 

338: \bibitem{cotunnelingExp}  {S.De Franceschi, S. Sasaki, J.M. Elzerman, W.G.

339: van der Wiel, S. Tarucha, and L.P. Kouwenhoven, Phys. Rev. Lett. \textbf{86}, 878 (2001); A. %%@

340: Kogan, S. Amasha, D. Goldhaber-Gordon, G. Granger, M.A. Kastner, and H. Shtrikman, Phys. Rev. %%@

341: Lett. \textbf{93}, 166602 (2004); D.M. Zumb\"{u}hl, C.M. Marcus, M.P. Hanson and A.C. %%@

342: Gossard, Phys. Rev. Lett. \textbf{93}, 256801 (2004).}

343: 

344: \bibitem{Kogan}{A. Kogan, S. Amasha, and M.A. Kastner, science {\bf 304}, 1293 (2004).}

345: 

346: \bibitem{Ng}{T.-K. Ng, Phys. Rev. Lett. {\bf 76}, 487 (1996).}

347: 

348: \bibitem{Goldin}{Y. Goldin and Y. Avishai, Phys. Rev. Lett. {\bf 81}, 5394 (1998).}

349: 

350: \bibitem{Lopez}{R. L\'opez, R. Aguado, G. Platero, C. Tejedor, Phys. Rev. Lett. {\bf 81}, %%@

351: 4688 (1998).}

352: 

353: \bibitem{Kaminski}{A. Kaminski, Yu.V. Nazarov, and L.I. Glazman, Phys. Rev. Lett. {\bf 83}, %%@

354: 384 (1999); Phys. Rev. B {\bf 62}, 8154 (2000).}

355: 

356: \bibitem{Flensberg}{K. Flensberg, Phys. Rev. B {\bf 55}, 13118 (1997).}

357: 

358: \bibitem{Dong}{Bing Dong, N.J.M. Horing, and H.L. Cui, Phys. Rev. B {\bf 72}, 165326 (2005).}

359: 

360: \bibitem{Ackerhalt}{J.R. Ackerhalt and J.H. Eberly, Phys. Rev. D {\bf 10}, 3350 (1974).}

361: 

362: \bibitem{Cohen}{C. Cohen-Tannoudji, J. Dupont-Roc, and G. Grynberg, {\em Atom-Photon %%@

363: Interactions: Basic Processes and Applications}, Wiley, New York, 1992 (Complements Av, pp. %%@

364: 388 ff and C$_{IV}$, pp. 334 ff).}

365: 

366: \bibitem{Smirnov}{G.F. Efremov and A.Yu. Smirnov, Zh. \'Eksp. Teor. Fiz. {\bf 80}, 1071 %%@

367: (1981) [Sov. Phys. JETP {\bf 53}, 547 (1981)].}

368: 

369: \bibitem{Dong1}{Bing Dong, X.L. Lei, and N.J.M. Horing, cond-mat/0509098.}

370: 

371: \bibitem{Tien}{P.K. Tien and J.R. Gordon, Phys. Rev. {\bf 129}, 647 (1963); C. Bruder and H. %%@

372: Schoeller, Phys. Rev. Lett. {\bf 72}, 1076 (1994).}

373: 

374: \bibitem{Vicari}{M. Vicari, A. Braggio, E. Galleani d'Agliano, M. Sassetti, Eur. Phys. J. B %%@

375: {\bf 25}, 115 (2002).}

376: 

377: \bibitem{Parcollet}{O. Parcollet and C. Hooley, Phys. Rev. B {\bf 66}, 85315 (2002).}

378: 

379: \bibitem{Paaske2}{J. Paaske, A. Rosch, and P. W\"olfle, Phys. Rev. B {\bf 69}, 155330 %%@

380: (2004).}

381: 

382: \end{thebibliography}

383: 

384: \end{document}

385: