1: \documentclass[letterpaper,twoside,twocolumn,american,pra,aps,showpacs,floatfix]{revtex4}
2: %%\documentclass[letterpaper,twoside,american,prl,aps,showpacs,preprint,onecolumn,floatfix]{revtex4}
3: \usepackage[T1]{fontenc}
4: \usepackage[latin1]{inputenc}
5: \usepackage{amsmath}
6: \usepackage{graphicx}
7: \usepackage{amssymb}
8:
9: \makeatletter
10:
11: \usepackage{babel}
12: \makeatother
13: \begin{document}
14:
15: \title{Nuclear spin qubits in a pseudo-spin quantum chain}
16:
17:
18: \author{E. \surname{Novais}}
19:
20: \affiliation{Department of Physics, Boston University, 590 Commonwealth Ave.,
21: Boston, MA, 02215}
22:
23: \author{A.~H. \surname{Castro~Neto}}
24:
25: \affiliation{Department of Physics, Boston University, 590 Commonwealth Ave.,
26: Boston, MA, 02215}
27:
28:
29: \date{\today{}}
30:
31:
32: \pacs{03.67.Lx, 33.25.+k, 75.10.Pq}
33:
34: \begin{abstract}
35: We analyze a quantum computer (QC) design based on nuclear spin
36: qubits in a quasi-one-dimensional (1D) chain of non-Kramers doublet
37: atoms. We explore the use of spatial symmetry breaking to obtain control
38: over the local dynamics of a qubit. We also study the decoherence mechanisms
39: at the single qubit level and the interactions mediated by the magnetic
40: media. The design can be realized in $\textrm{PrBr}_{\textrm{3}-x}\textrm{F}_{x}$
41: with nuclear magnetic
42: resonance (NMR) techniques.
43: \end{abstract}
44: \maketitle
45:
46: \section{Introduction}
47:
48: Nuclear magnetic resonance is the framework of
49: a very promising quantum computing architecture \cite{NAG97}.
50: NMR is a natural choice because nuclei are protected from many sources
51: of decoherence, and therefore produce robust qubits. Successful
52: realizations of quantum algorithms implemented on a NMR quantum computer
53: have been realized in liquid solutions of
54: molecules \cite{LMKV01,LMKV02}.
55: Nevertheless, a liquid NMR QC is not easily scalable, that is,
56: there is a practical limit in the number of qubits that can be constructed
57: in a molecule. From a handful of qubits already achieved one must scale
58: the QC to several thousands before a non-trivial algorithm
59: can be run \cite{LMKV01,LMKV02}. Though other limitations can
60: also be argued to the use of NMR \cite{NLSP01}, scalability is
61: an undeniable problem.
62:
63: A possible route to deal with the scalability problem is to
64: consider NMR in crystals \cite{TDL+00}.
65: There are several different proposed designs,
66: but all of them share two common elements.
67: Firstly, a gradient magnetic field is used to shift the nuclear resonance
68: frequencies of different nuclei, allowing qubits to be addressed independently.
69: Secondly, as the number of qubits increases, a second decoherence channel is
70: introduced by the low energy excitations of the interacting qubits.
71: In any crystal, the direct dipolar interaction between nuclei produces
72: secular broadening.
73: To a certain extent this broadening can be reduced by NMR techniques.
74: Thus, it is usually assumed that a \emph{perfect selective decoupling}
75: of the qubits from the dipolar interaction can be achieved.
76:
77: Although very promising, there are technical problems with the use
78: of NMR in crystals.
79: For example, in the proposed materials $\textrm{CaF}_{\textrm{2}}$
80: and $\textrm{MnF}_{\textrm{2}}$, qubits are the nuclear spin $1/2$
81: of the F ions \cite{TDL+00}. To obtain a measurable frequency shift
82: from one qubit to another a homogeneous gradient field of more than
83: $1T/\mu m$ is required. The obvious solution is to separate qubits
84: from each other. However, by distancing the qubits to work with an experimentally
85: feasible value of the field gradient, another problem is created by weakening the
86: qubit-qubit interactions.
87:
88: Interacting qubits are a necessary condition for quantum computation.
89: A quantum algorithm is a sequence of unitary transformations
90: in the Hilbert space spanned by all the qubits.
91: A given transformation in a subspace of
92: \emph{n} qubits is called a \emph{n}-qubit gate.
93: A quantum computing scheme must provide a complete
94: set of such quantum gates, in other words, it must be possible
95: to construct any unitary transformation with a sequence of building
96: block operations provided by the design.
97: One of the most useful results
98: in quantum information theory is that from all \emph{one}-qubit gates
99: and \emph{almost} any \emph{two}-qubit gate is possible to find
100: a complete set of gates \cite{SL95}.
101: In a NMR QC, the \emph{one}-qubit gates are easily produced.
102: The \emph{two}-qubit gate is the time evolution of two qubits under
103: an interaction.
104:
105: The viability of a solid state NMR QC
106: relies on interactions available to construct the \emph{two}-qubit gate
107: and the correspondent decoherence times.
108: On the one hand,
109: in $\textrm{CaF}_{\textrm{2}}$, the only available interaction is
110: the direct dipolar coupling between nuclear moments. In most cases
111: this interaction is effectively short ranged for quantum computational
112: purposes.
113: The small nuclear moments and the $1/r^{3}$ dependence makes the
114: operation time of
115: a gate (composed by two qubits far apart) much larger than the
116: decoherence times.
117: On the other hand, in
118: $\textrm{MnF}_{\textrm{2}}$,
119: the relevant interaction is the
120: Suhl-Nakamura coupling \cite{HS58}.
121: This is an indirect coupling of nuclear spins mediated by magnons of the
122: $\textrm{Mn}$ electronic spins. Below its N\'{e}el temperature the magnon spectrum has
123: a gap. At the same time that a gap reduces decoherence,
124: it implies that the interaction strength has an exponential decay with the distance.
125: Thus, it is unlikely that a considerable separation between qubits
126: can be obtained in both cases.
127: The search for long range interactions has motivated several recent
128: publications \cite{DMVP01,AGTS+03}.
129: Unfortunately, long range interactions
130: are tied to low energy modes and, consequently, short decoherence times.
131:
132: In this paper, we discuss nuclear-nuclear interactions mediated
133: by an anisotropic quantum pseudo-spin chain. We analyze how the breaking
134: of spatial symmetries in a system of non-Kramers ions can be used
135: to gain control over local properties of a QC. We show that one
136: can reduce decoherence and/or construct different \emph{two}-qubit gates
137: as a function of external electromagnetic fields.
138: Although our ideas are general, we
139: propose a specific realization in the compounds
140: $\textrm{PrCl}_{\textrm{3-x}}\textrm{F}_{\textrm{x}}$
141: and $\textrm{PrBr}_{\textrm{3-x}}\textrm{F}_{\textrm{x}}$.
142: Both materials are equally suitable
143: to our discussion, but we use the parameters of the latter in
144: our estimates. We start by summarizing the properties of the parent compound,
145: $\textrm{x=0}$. Subsequently, we discuss the chemical doping
146: with F. Finally, we explore the use of the nuclear spin from the F
147: ions as qubits.
148:
149: \section{the physics of $\textrm{PrBr}_3$ and the construction of qubits}
150:
151:
152: $\textrm{PrBr}_{\textrm{3}}$ is a 1D ionic insulator made out of
153: Pr chains separated by $5\textrm{Å}$. The Pr ions are subjected to
154: a crystal field with $C_{3h}$ symmetry. Their ground state is a
155: non-Kramers doublet that is separated from the first excited state by
156: a gap of $17\textrm{K}$ \cite{BSBH+87}. A Jahn-Teller transition
157: takes place at $0.1\textrm{K}$ \cite{SSRLA92}, it lifts the doublet
158: degeneracy, and sets a low temperature limit to the applicability of this material
159: to our design.
160: A convenient way to model this system is via a pseudo-spin
161: $1/2$ representation \cite{AABB70}. We focus on
162: the physics of two adjacent chains and we label the pseudo-spins
163: of each of these chains as $\tau ^{z}$ and $\sigma ^{z}$
164: (see Fig.~\ref{cap:PrBr3-xFx}).
165: The single ion Hamiltonian at site $i$ is written as
166: \begin{eqnarray*}
167: H_{\textrm{ion}} & = & \sum _{i}\hslash \gamma _{s}^{z}B_{z}S_{i}^{z}+g_{s}^{x}E_{x}S_{i}^{x}+g_{s}^{y}E_{y}S_{i}^{y},
168: \end{eqnarray*}
169: where $\overrightarrow{S}=\left\{ \overrightarrow{\sigma },\overrightarrow{\tau }\right\} $,
170: $\gamma _{s}^{z}=1.4\times 10^{11}\textrm{T}^{\textrm{-1}}\textrm{s}^{\textrm{-1}}$,
171: $\vec{B}$ is an external magnetic field and $\vec{E}$ an applied
172: electric field \cite{BSBH+87}. We are unaware of published values
173: for the electric dipolar constants in $\textrm{PrBr}_{\textrm{3}}$,
174: however they should not be very different from the ones in $\textrm{PrCl}_{\textrm{3}}$
175: where $g_{s}^{x,y}=4.0\times 10^{-31}\textrm{Cm}$ \cite{JPH76}. It is important
176: to stress that there is no off-diagonal matrix element that couples
177: the doublet state to the magnetic field. Therefore, a magnetic field
178: cannot induce transitions between the doublet states.
179: The ionic magnetic moments are coupled
180: by a dipolar term, however the most relevant contribution to the interaction
181: Hamiltonian comes from transitions due to the transverse electric
182: dipoles that are strongly coupled to the lattice. Although the only real
183: magnetic moment is oriented along the chain ($z$ direction), this family of compounds is regarded
184: as $\textrm{XY}$ chains described by the Hamiltonian
185: \begin{eqnarray}
186: H_{\textrm{xy}} & = & J_{\perp }\sum _{i}S_{i}^{x}S_{i+1}^{x}+S_{i}^{y}S_{i+1}^{y},\label{eq:Hxy}
187: \end{eqnarray}
188: where $J_{\perp }\approxeq 3\textrm{K}$ \cite{SSRLA+91}.
189:
190: \begin{figure}[bth]
191: \includegraphics[ width=0.9\columnwidth ]{fig6-b}
192: \caption{\label{cap:PrBr3-xFx}Two adjacent chains in
193: $\textrm{PrBr}_{3-x}\textrm{F}_{x}$.}
194: \end{figure}
195:
196: In order to construct a qubit, we propose the use of the nuclear spins
197: of $\textrm{F}$ ions in the diluted salt $\textrm{PrBr}_{\textrm{3-}x}\textrm{F}_{x}$.
198: There are two main components to nuclear decoherence, connected with
199: the two strongest interactions that a nucleus is subjected to: the
200: electric quadrupolar and the magnetic dipolar \cite{AA}.
201: We are ultimately interested in the decoherence channels in a F
202: nucleus in $\textrm{PrBr}_{\textrm{3-x}}\textrm{F}_{\textrm{x}}$.
203: Nuclear quadrupole
204: resonance experiments have measured $T_{1,2}$
205: for the $\textrm{Br}$ nuclei in the parent compound.
206: They established fairly well that
207: the spin-lattice relaxation time, $T_{1}$, is due to magnetic interactions
208: \cite{MDRLA83,SSRLA+91},
209: and it is of order of $100 ms$ at 1K.
210: Moreover, the nuclear spin-spin decoherence time, $T_{2}$, was found to be
211: $\approxeq 40\mu s$ at 1K. The decoherence sources that lead to this
212: value for $T_{2}$ are not yet well understood \cite{SSRLA92}.
213: If we use the Van Vleck formula \cite{AA} to estimate
214: the secular broadening of resonance
215: lines, we find that the direct dipolar interaction among the nuclei leads to a
216: broadening of the order of $10^{2}\mu \textrm{s}$.
217: Further considering the quadrupolar effects
218: it is clear that the direct
219: dipolar interaction gives a sizable contribution to decoherence.
220: Thus, as usual in solid state NMR designs, we can conclude that
221: decoupling is very important
222: in order to make this family of compounds useful to a QC.
223:
224: Each F introduces a local lattice distortion, hence lowering the crystal
225: field symmetry at neighboring Pr ions.
226: The distortion introduced by the $\textrm{F}$
227: ion has its strongest effect on the Pr ions labeled $2$, $4$, $5$
228: in Fig.~(\ref{cap:PrBr3-xFx}). In the pseudo-spin representation,
229: a local symmetry breaking corresponds to the addition of
230: transverse fields, $\bar{\Delta }$ and $\Delta $, on each one of these
231: sites. Moreover, the $\textrm{Pr}_{\left(\textrm{4,5}\right)}$ no
232: longer have a plane of inversion perpendicular to the chain axis.
233: Thus, these ions can develop electric dipoles perpendicular to
234: that plane. The Hamiltonian for the pseudo-spin chains
235: can be written as
236: \begin{eqnarray}
237: H_{\textrm{Pr}} & = & H_{\textrm{ion}}+H_{\textrm{x}y}+\Delta \sigma _{0}^{x}+\bar{\Delta }\left(\tau _{0}^{x}+\tau _{1}^{x}\right)\nonumber \\
238: & + & g_{s}^{z}E_{z}\left(\tau _{0}^{z}-\tau _{1}^{z}\right).\label{eq:fullhamispin}
239: \end{eqnarray}
240:
241: We consider the case where
242: $\left(\bar{\Delta },\, \Delta \right)\ll
243: \max \left(\hslash \gamma _{s}^{z}B_{z},k_{B}T\right)<J_{\perp }$,
244: otherwise the moments at $\textrm{Pr}_{\left(2,4,5\right)}$ would be completely
245: quenched by the symmetry breaking and the analysis below would need
246: to be extended to include next near neighbor interactions.
247: Notice that in Eq.~(\ref{eq:fullhamispin})
248: the transverse fields introduce matrix elements between the two magnetic
249: states of $Pr_{\left(2,4,5\right)}$. Thus, an oscillating magnetic
250: field parallel to the chain axis would reveal two distinct resonant
251: lines, $\omega _{\bar{\Delta }}$ and $\omega _{\Delta }$, associated
252: with the splitting of the $\textrm{Pr}$ doublet state.
253:
254:
255: \section{the qubit Hamiltonian}
256:
257:
258: The use of $\textrm{F}$ as a qubit has two advantages.
259: There is no decoherence due to electric field gradients because
260: it does not have a quadrupolar moment.
261: In addition, there is only one isotope of $\textrm{F}$ in nature,
262: so all qubits experiencing the same magnetic field are identical.
263: By assuming perfect decoupling, we can disregard
264: the direct dipolar interaction between nuclei.
265: This is a much less stringent condition than in other NMR QC schemes
266: because the qubit resonance frequency is very distinct from the other ions.
267: Therefore, straightforward pulse sequences can be used to perform the decoupling.
268: The remaining contribution to the
269: nuclear Hamiltonian comes from the magnetism of the surrounding Pr
270: atoms. Hence, the nuclear hyperfine interaction
271: of each F ion in first approximation can be written as
272: \begin{eqnarray}
273: H_{\textrm{F}} & = & \left[\hslash \gamma _{N}B_{z}+d\left(\sigma _{0}^{z}-\frac{\tau _{0}^{z}+\tau _{1}^{z}}{2}\right)-\tilde{d}\left(\sigma _{-1}^{z}+\sigma _{1}^{z}\right)\right]I^{z}\nonumber \\
274: & + & 3d\left(\tau _{0}^{z}-\tau _{1}^{z}\right)I^{x}+\sqrt{2}\tilde{d}\left(\sigma _{-1}^{z}-\sigma _{1}^{z}\right)I^{y},\label{eq:hf-1}
275: \end{eqnarray}
276: where $d=(\mu _{0}\hslash ^{2}\gamma _{s}^{z}\gamma _{N})/(4\pi
277: r_{0}^{3})\approxeq 10^{-4}\textrm{K}$,
278: $\tilde{d} \approx d/5$, $\gamma _{N}=25\times 10^{7}\textrm{T}^{\textrm{-1}}\textrm{s}^{\textrm{-1}}$,
279: and $\vec{I}$ is the nuclear spin-$1/2$ operator of the $\textrm{F}$
280: nucleus.
281:
282: The pseudo-spin physics described by Eq.~(\ref{eq:fullhamispin})
283: presents us with a very interesting situation. An applied magnetic
284: field with frequency $\omega _{\bar{\Delta }}$ and/or an electric
285: field, $E^{z}$, only affect the $\textrm{Pr}_{\left(\textrm{4,5}\right)}$,
286: and therefore can be used to act locally in the qubit. For example,
287: a sufficiently large electric field forces $\tau _{0}$ and $\tau _{1}$
288: into a singlet configuration, freezing their dynamics. The net result
289: is decoupling of the F ion from the $\tau $-chain. In this case the
290: hyperfine Hamiltonian simplifies to
291:
292: \begin{eqnarray}
293: H_{\textrm{F}} & \approxeq & \left[\hslash \gamma _{N}B_{z}+d\sigma _{0}^{z}-\tilde{d}\left(\sigma _{-1}^{z}+\sigma _{1}^{z}\right)\right]I^{z}\nonumber \\
294: & + & \sqrt{2}\tilde{d}\left(\sigma _{-1}^{z}-\sigma _{1}^{z}\right)I^{y}.\label{eq:hype2}
295: \end{eqnarray}
296: This is a particularly interesting effect. It cancels the strongest
297: transverse part of Eq.~(\ref{eq:hf-1}), and consequently, corresponds
298: to a reduction in the dissipation rates $T_{1,2}^{-1}$.
299:
300:
301: \subsection{Dissipation rates}
302:
303: In order to estimate the dissipation rates due to the Pr magnetic moments,
304: we will focus on the low energy physics of
305: Eq.~(\ref{eq:fullhamispin}). Therefore, we can use Abelian bosonization\cite{Ian} to obtain
306: simple analytical expressions for $T_{1,2}^{-1}$.
307:
308: Bosonization is a well stablished method to study spin chains.
309: In a concise way, we first use the Jordan-Wigner transformation,
310: mapping the pseudo-spins in spinless fermions.
311: Then, we linearizing the dispersion relations around the two Fermi points,
312: $p_{F}=\arccos \left(\hslash \gamma_{s}^{z}B_{z}/J_{\perp }\right)$,
313: and define the
314: Fermi velocity $v=J_{\perp }\sin \left(p_{F}\right)$.
315: The result is that $H_{\textrm{xy}}$ can be re-written
316: as a free bosonic Hamiltonian.
317: In this language, it is straighforward to evaluate the
318: pseudo-spin correlation function at zero temperature\cite{Ian}
319:
320: \begin{eqnarray}
321: \left\langle S^{z}_{j}(\tau)S^{z}_{0}(0\right\rangle & = & \frac{1}{2\pi^{2}}
322: \frac{x^2-(v\tau)^2}{(x^2+(v\tau)^2)^2}\nonumber\\
323: & & \frac{\cos(2p_{K}x)}{2\pi^{2}}\frac{1}{x^2+(v\tau)^2}\label{correlation},
324: \end{eqnarray}
325: where $\tau$ is the imaginary time, $x=a_{0}j$ and $a_0 \approx 4.4\textrm{Å}$
326: is the lattice spacing.
327:
328: For a sufficiently large magnetic field ($B_{z}\gg 0.1\textrm{T}$),
329: $T_{1}^{-1}$ is given by\cite{slichter}
330: \begin{eqnarray}
331: T_{1}^{-1} & = & \frac{1}{2}\int_{-\infty}^{\infty}dt^{\prime}
332: \left\langle H_{\perp}(t) H_{\perp}(t+t^{\prime})
333: \right\rangle e^{-i\omega_{0}t^{\prime}},
334: \label{slichter}
335: \end{eqnarray}
336: where, if we focus in the regime described by Eq.~(\ref{eq:hype2}), we defined
337: \begin{eqnarray*}
338: \omega_{0} & = & \gamma_{N} B_{z},\\
339: H_{\perp} & = & \sqrt{2} \tilde{d}\left(\sigma _{-1}^{z}-
340: \sigma _{1}^{z}\right).
341: \end{eqnarray*}
342: An equivalent expression for $T_{2}$ is obtained when we match
343: the results of a random phase approximation (RPA) calculation for
344: the transverse suceptibility with the solution
345: of the Boch's equations\cite{AJL+87}.
346: Using Eq.~(\ref{correlation}) into Eq.~(\ref{slichter})
347: or the RPA result,
348: we evaluate the zero temperature decoherence rates
349: due to the pseudo-spins as
350: \begin{eqnarray}
351: T_{1}^{-1}=T_{2}^{-1} & \approxeq & 8 \pi^{-1} \gamma _{N}\left(\hslash \tilde{d} \gamma _{s}^{z}\right)^{2}B_{z}^{3} J_{\perp }^{-4}.\label{eq:t2}
352: \end{eqnarray}
353: The unusual dependence of the relaxation
354: time with the magnetic field, scaling like $B_{z}^{3}$, can be used
355: to assert Eq.~(\ref{eq:hf-1}-\ref{eq:hype2}). Finally, an applied transverse
356: electric field ($E_{x,y}\neq 0$) can be used to open a gap in the
357: pseudo-spin spectrum.
358: This further isolates the qubit by quenching the pseudo-spins magnetic moments,
359: and therefore, even smaller values of $T_{1,2}^{-1}$ can be achieved.
360:
361: In general, nuclear spins interacting with a
362: gapless spin chain
363: would have super-ohmic dissipation.
364: However, the hyperfine Hamiltonian, Eq.~(\ref{eq:hype2}), that we derive
365: depends exclusively on the $z$ component of the pseudo-spins.
366: This restricted dipolar interaction implies an ohmic dissipation.
367: We emphasize that this is somewhat unique feature of pseudo-spins.
368: If Eq.~(\ref{eq:hype2}) would have
369: flip-flop terms, then the transverse correlations of the
370: spseudo-spins would imply a super-ohmic behavior.
371:
372:
373: \subsection{Construction of quantum gates}
374:
375: \begin{figure}[tbh]
376: \includegraphics[ width=0.9\columnwidth]{scheame3}
377: \caption{\label{cap:interaction.}Interaction between two nuclear spins of
378: $\textrm{F}$ ions is mediated by the pseudo-spin chain of $\textrm{Pr}$.
379: $d$ and $\tilde{d}$ are the strength of the hyperfine coupling
380: defined in Eq.~(\ref{eq:hf-1}).}
381: \end{figure}
382:
383: Now that we have studied the single qubit problem, we turn our
384: attention to the qubit-qubit interaction. We focus in the regime described by
385: Eq.~(\ref{eq:hype2}) because it is the most favorable for QC.
386: Consider a second $\textrm{F}$ atom along the
387: chain as shown in Fig.~(\ref{cap:interaction.}).
388: By integrating out the $\sigma $-spins we obtain a retarded interaction
389: between the two nuclei. This is very similar to the RKKY interaction,
390: but mediated by the pseudo-spins \cite{RKKY}.
391:
392: Exactally as in the RKKY problem, the $\textrm{F}$ nuclear spins have a
393: much slower dynamics than the pseudo-spins
394: ($\gamma_{N}B_z \ll J_\perp / \hslash$).
395: Therefore, it is reasonable to consider an instantaneous approximation to
396: the interaction.
397: At zero temperature, we use
398: Eq.~(\ref{correlation}) to calculate its form.
399:
400: For the RKKY, finite temperature corrections are usually irrelevant
401: because the
402: Fermi energy is much larger than the temperatures under consideration.
403: However, in the pseudo-spin
404: chain we are assuming temperatures only one order of magnitude smaller
405: than $J_{\perp}$.
406: We can easily re-write the zero temperature correlation function,
407: Eq.~(\ref{correlation}), in its finite temperature form by using the
408: conformal invariance of the $\textrm{XY}$ model \cite{SEIA94}.
409: The final result is the effective
410: interaction between to qubits
411: \begin{eqnarray}
412: H_{\textrm{eff}} & \approxeq & f_{zz}I_{1}^{z}I_{2}^{z}+f_{yz}\left(I_{1}^{y}I_{2}^{z}+I_{1}^{z}I_{2}^{y}\right)+f_{yy}I_{1}^{y}I_{2}^{y},\label{eq:Heff}
413: \end{eqnarray}
414: where we have defined the nuclear exchange couplings
415: %\begin{subequations}
416: \begin{eqnarray*}
417: f_{zz} & = & d^{2}G\left(\Delta x\right)-d\tilde{d}\left[G\left(\Delta x-1\right)+G\left(\Delta x+1\right)\right]\nonumber \\
418: & + & \tilde{d}^{2}\left[2G\left(\Delta x\right)+G\left(\Delta x+2\right)+G\left(\Delta x-2\right)\right],\\
419: f_{yz} & = & \sqrt{2}\left\{ d\tilde{d}\left[G\left(\Delta x-1\right)-G\left(\Delta x+1\right)\right]\right.\nonumber \\
420: & + & \left.\tilde{d}^{2}\left[G\left(\Delta x+2\right)-G\left(\Delta x-2\right)\right]\right\} ,\\
421: f_{yy} & = & 2\tilde{d}^{2}\left[2G\left(\Delta x\right)-G\left(\Delta x+2\right)-G\left(\Delta x-2\right)\right].
422: %\label{exchange}
423: \end{eqnarray*}
424: %\label{exchange}
425: %\end{subequations}
426: $G(\Delta x)$ is the finite temperature pseudo-spin propagator given by
427: \begin{eqnarray}
428: G\left(\Delta x\right) & \approxeq & \frac{1-\cos \left(2p_{F}\Delta
429: x\right)}{2\pi ^{2}v^{3}\beta ^{2}}\left[\sinh \left(\frac{\Delta
430: x}{v\beta }\right)\right]^{-2} \, ,
431: \label{propagator}
432: \end{eqnarray}
433: where $\Delta x$ is the distance between qubits in units of lattice spacing
434: $a_{0}$ and $\beta =1/\left(k_{B}T\right)$.
435: For distances smaller than the thermal coherence length, $\xi _{T}=v\beta $,
436: the interaction decays as a power law, $G\left(\Delta x\right)\approxeq
437: (1-\cos \left[2p_{F}\Delta x\right])/(2\pi ^{2}v\Delta x^{2})$,
438: leading to long range interaction between qubits.
439: It is also interesting to consider the consequences of applying transverse
440: electric fields.
441: Since the pseudo-spin propagator acquires a gap, there is an additional exponential
442: decay in Eq.~(\ref{eq:Heff}) which is a function of $E_{(x,y)}$.
443: Thus, we can use transverse fields to switch on and off the interaction between qubits.
444:
445: Equation~(\ref{eq:Heff}) is a two-qubit gate. In conjunction with the
446: possibility to perform arbitrary rotations, it generates a complete set of
447: quantum gates \cite{SL95}.
448: The inverse of the gate operation time is
449: given by $T_{G}^{-1}\left(\Delta x\right)=\hslash ^{-1}\min \left(\left|f_{zz}\right|,\left|f_{yz}\right|\right)$.
450: In order to compare $T_{G}$ with $T_{1,2}$ we consider
451: a particular case. Take $B_{z}\approx 2\textrm{T}$ and a temperature
452: $T=0.1\textrm{K}$,
453: so that the pseudo-spin chain is partially polarized.
454: Low temperature corrections
455: to Eq.~(\ref{eq:t2}) are very small, and we use it as an upper
456: bound estimate to the decoherence times,
457: %in the case
458: %of \emph{perfect decoupling} and strong $E^{z}$:
459: $T_{1,2}^{-1}\sim 10^{-2}\textrm{s}^{\textrm{-1}}$.
460: These values are much smaller than the rates in $\textrm{PrBr}_{\textrm{3}}$
461: % the parent compound
462: due to three facts: the absence of quadrupolar effects, the reduction of pseudo-spin fluctuation
463: in $\tau _{0,1}$ and the assumption of decoupling.
464: Two qubits separated by $13\textrm{Å}$
465: have $T_{G}\left(3\right)\sim 10^{-1}\textrm{s}$, thus leading to a quantum
466: gate at the edge of the error correction threshold of $10^{-4}$
467: \cite{LMKV02,errorcode}.
468:
469: Another important aspect of Eqs.~(\ref{eq:fullhamispin}) and (\ref{eq:hf-1})
470: is that several different gates can be constructed as a function of
471: the magnetic field $B_{z}$, the resonance frequencies
472: $\omega _{\Delta ,\bar{\Delta }}$
473: and the electric fields $E_{x,z}$.
474: For instance,
475: the pseudo-spin propagator,
476: Eq.~(\ref{propagator}), has an oscillatory behavior with $B_{z}$.
477: This can be used to change the relative
478: strength of $f_{\textrm{ij}}$ in Eq.~(\ref{eq:Heff}).
479: In order to make this point clear, we now pause and consider a concrete
480: example.
481:
482: One of the most simple quantum circuits is the one that creates
483: entangle pairs of qubits (Bell's states). From the quantum-logic perspective,
484: this is accomplished by the use of a Hadamard gate follow by a CNOT
485: gate \cite{NCh}. Since the production of entangle pairs
486: is fundamental to perform quantum computation and quantum comunication,
487: this straightforward circuit is a conerstone in any design.
488: The key element here is the CNOT gate.It is a two qubit gate and,
489: consequently its implementation depends upon the avaiable interaction.
490: In liquid state NMR the strongest component in the
491: Hamiltonian that a pair of qubits is subjected is \cite{LMKV02}
492: \begin{eqnarray}
493: H_{z} & \cong & JI_{1}^{z}I_{2}^{z}.\label{eq:gate1}
494: \end{eqnarray}
495: This Hamiltonian can also be approximated by Eq.~(\ref{eq:Heff}).
496: For the sake
497: of argument, let us assume two F atoms separated by four lattice sites
498: ($\sim 18\textrm{Å}$). In addition, let us consider the external conditions
499: that we considered before: a large $E^{z}$ to freeze the pseudo-spin dynamics
500: in $\tau _{0,1}$ and $T=0.1K$.
501:
502:
503: \begin{figure}[htb]
504: \includegraphics[
505: width=1\columnwidth,
506: keepaspectratio]{gates.eps}
507: \caption{\label{cap:the-coupling-constants}the coupling constants
508: $f_{zz}$,$f_{yz}$
509: and $f_{yy}$ as a function of the magnetic field $B^{z}$ in units
510: of $\tilde{d}^{2}$ for two F atoms separated by four lattice spacing.}
511: \end{figure}
512:
513:
514: From Eq.~(\ref{eq:Heff}) and the definition of the pseudo-spin
515: propagator we can plot Fig.~(\ref{cap:the-coupling-constants}),
516: where we can see that for $B^{z}\cong 1.01T$ the effective coupling
517: constants are
518: \begin{eqnarray*}
519: \frac{f_{zz}}{\tilde{d}^{2}} & \cong & 0.06,\\
520: \frac{f_{yz}}{\tilde{d}^{2}} & \cong & 0.001,\\
521: \frac{f_{yy}}{\tilde{d}^{2}} & \cong & 0.001.
522: \end{eqnarray*}
523:
524:
525: Thus, as in liquid NMR, the strongest component in the interaction
526: is given by Eq.~(\ref{eq:gate1}). In order to produce a a CNOT gate with
527: this Hamiltonian in an NMR setup\cite{LMKV02}, one first apply
528: a radio frequency
529: pulse to rotate $I_{2}$ about $\hat{x}$ ( $+\hat{z}$ goes to $-\hat{y}$).
530: Then the spin system evolves with Eq.~(\ref{eq:gate1}) for a time
531: $t=\pi \hslash /4J$.
532: Then, a second pulse is sent to rotate $I_{2}$ by $90^{\circ }$ about the
533: $-\hat{y}$ axis. Finally, an additional phase shift on both spins
534: is used to obtain the CNOT gate. If we add the initial Hadamard gate,
535: it is necessary to use a total of five radio frequency pulses
536: (one-qubit gates) and
537: the time evolution of the Hamiltonian Eq.~(\ref{eq:gate1}) .
538:
539: Let us analyze another possibility. Consider the same condictions as before,
540: but with an external magnetic field $B^{z}=2.02T$. In this case, the effective
541: coupling constants are
542:
543: \begin{eqnarray*}
544: \frac{f_{zz}}{\tilde{d}^{2}} & \cong & -0.005,\\
545: \frac{f_{yz}}{\tilde{d}^{2}} & \cong & \phantom +0.0003,\\
546: \frac{f_{yy}}{\tilde{d}^{2}} & \cong & -0.025.
547: \end{eqnarray*}
548:
549:
550: In contrast with the previous case, the strongest part of the interaction is
551:
552: \begin{eqnarray}
553: H_{y} & \cong & f_{yy}I_{1}^{y}I_{2}^{y}.\label{eq:gate2}
554: \end{eqnarray}
555:
556:
557: If we allow a free evolution of the system by Eq.~\ref{eq:gate2} for
558: a time $t=\pi \hslash /2f_{yy}$, the unitary transformation that
559: is implemented is:
560:
561: \begin{eqnarray*}
562: R & = & \left[\begin{array}{cccc}
563: 1 & 0 & 0 & -i\\
564: 0 & 1 & i & 0\\
565: 0 & i & 1 & 0\\
566: -i & 0 & 0 & 1\end{array}
567: \right]
568: \end{eqnarray*}
569:
570:
571: Acting on the computational basis with this rotation we automatically
572: generate the entangled states
573:
574: \begin{eqnarray*}
575: \left|\beta _{1}\right\rangle & = & \frac{\sqrt{2}}{2}\left(\left|00\right\rangle -i\left|11\right\rangle \right),\\
576: \left|\beta _{2}\right\rangle & = & \frac{\sqrt{2}}{2}\left(\left|01\right\rangle +i\left|10\right\rangle \right),\\
577: \left|\beta _{3}\right\rangle & = & \frac{\sqrt{2}}{2}\left(\left|01\right\rangle -i\left|10\right\rangle \right),\\
578: \left|\beta _{4}\right\rangle & = & \frac{\sqrt{2}}{2}\left(\left|00\right\rangle +i\left|11\right\rangle \right).
579: \end{eqnarray*}
580:
581:
582: Hence, one can fine tune the experimental setup to obtain a desired
583: quantum circuit using less resources.
584: In the above example, the simple tuning of the
585: magnetic field replace the one qubit gates on the previous setting.
586: However, this is just one of many possible ways to control
587: the interaction Hamiltonian. A more subtle (and potentially more interesting
588: way) is related to the frequency $\omega _{\Delta }$ and
589: $\omega _{\bar{\Delta }}$.
590: In presence of a gradient magnetic field they have a
591: site index
592: ($\hslash \omega _{\Delta }\cong \sqrt{\left[\hslash \gamma _{s}^{z}B^{z}\left(\vec{x}\right)\right]^{2}+\Delta ^{2}}$).
593: Thus, one could act in the magnetic environment of
594: each individual qubit.
595:
596:
597: \section{Discussion and Conclusions}
598:
599: Until this point we discussed how single qubits can be constructed and how
600: a pair of qubits can interact.
601: We now discuss how to use these building blocks in a QC.
602:
603: The natural geometry is to consider a magnetic field gradient applied
604: along the chain direction.
605: Nuclei in the same equipotential line belong to different copies of the QC,
606: and we assume that they can be periodically arranged (see below).
607:
608: Initialization is a very hard problem in QCs based on nuclear spin qubits.
609: However, there are some possible solutions already available in the
610: literature \cite{AGTS+03,LJSUVV99}.
611: At first sight one could imagine that the initialization could
612: be done by optical pumping (Pound-Overhauser effect) with the pseudo-spins,
613: as it is done in $MnF_{2}$ with electronic spin. Unfortunately,
614: the same property that gives a lower decoherence rate
615: than in other gapless magnetic systems hinders this option.
616: Since there
617: is no flip-flop term ($S^{+}I^{-}$) in the hyperfine Hamiltonian,
618: one cannot use the pseudo-spins to pump the nuclear spins.
619: There are two other possible {}``hardware'' solutions that can be
620: used to solve the initialization problem.
621: A diluted set of magnetic
622: impurities can be used to refrigerate the qubits. The general idea
623: is to add a small amount of an ion with a large magnetic moment (such as
624: $Gd$ replacing some $Pr$) to the sample. This set of impurities
625: can be used to pump energy out of the nuclear systems and after
626: some polarization is achieved a sufficiently large magnetic field
627: would {}``freeze'' the impurities. There are two setbacks
628: in this approach. Firstly, the $Gd$ ion would {}``break'' the pseudo-spin
629: chains and the F ions in each side might not interact. Secondly, virtual
630: flips of the $Gd$ spin could introduce an additional decoherence
631: channel. The second {}``hardware'' solution is based on the fact
632: that the crystals can be grown on a semiconductor substrate. By
633: exciting the electron gas in the semiconductor, it is possible to
634: use {}``cross-polarization-coherent transfer techniques''. The latter
635: is the solution found in ref.~\cite{AGTS+03} to the initialization
636: procedure in a QC based on 1-d organic molecules. Finally, if only partial
637: polarization is obtained by one of the {}``hardware'' methods cited above,
638: the Schulman-Vazirani
639: procedure\cite{LJSUVV99} can be used as a {}``software''
640: method to initialize the state.
641:
642: The final element in a QC design is the read-out mechanism.
643: All QCs based on NMR of impurities have the common problem of
644: low signal due to the small density of qubit copies.
645: However, nuclear polarization can increase considerably the NMR sensitivity.
646: In this case, the read-out of a qubit with only $10^{12}$ copies
647: is possible with current NMR technology \cite{AGTS+03}.
648:
649: There are some relevant experimental questions that are open
650: and can foster new theoretical work.
651: In the first place, the simplest way to produce crystals of a salt such as
652: $\textrm{PrBr}_{3-x}\textrm{F}_{x}$ is through dehydration of a liquid solution \cite{CMU69}.
653: This straightforward process creates samples with
654: the F ions in random positions.
655: Although this is sufficient to infer our results for a single qubit,
656: further developments
657: in ionic crystal growth should be accomplished
658: before the full range of possibilities
659: that we discuss can be experimentally studied.
660: One possible research avenue is a molecular-beam
661: epitaxial growth (MBEG).
662: MBEG is a well established
663: technique in semi-conductors and metals. Although from a historical
664: perspective the growth of ionic crystals is an old field,
665: the technology is much less mature.
666: Nevertheless, it shows unique characteristics that are worth
667: exploring \cite{MBEG1,YF90, MBEG2}.
668: The most interesting feature
669: is that the incoming molecule has a very weak bound with the surface terrace
670: and strong bounding to the ledge.
671: This can be simply understood in electrostatic terms, and as a consequence,
672: leads to a large surface diffusion until the molecule reaches the
673: ledge. We speculate that this fact can be used to obtain a higher
674: degree of control in the impurity placement than in any other kind
675: of material. Another interesting characteristic is that large lattice
676: misfits are also allowed in the growth of layers. Thus, it is
677: natural to propose experiments with a crystal composed of a super-lattice
678: of $PrBr_{3}$ and layers of $PrBr_{2}F$. This setup
679: is feasible with the current technology and many of our results for the
680: qubit-qubit interaction can be experimentally tested. Another possibility
681: is a super-lattice of $PrBr_{3}$ with layers of $PrF_{3}$, however
682: the large lattice misfit will probably prevent the layer growth\cite{MBEG2}.
683: A final remark
684: is that ionic crystals grow well on semi-conductors surfaces. This
685: has two main consequences: 1) the semiconducting substrate can be integrated
686: in other quantum
687: computer schemes (similar to Si/P proposals) and with current electronics,
688: 2) a semi-conductor substrate can be used to initialize the quantum
689: computer by optical pumping as we argued above.
690:
691: There isa another issue that is common to all solid state NMR designs:
692: it is unlikely that \emph{perfect decoupling} can be achieved.
693: Therefore, the experimental value of
694: $T_{2}$ is potentially smaller than the prediction of Eq.~(\ref{eq:t2}).
695: Although we are probably overestimating $T_{2}$, we are also underestimating
696: the gate time $T_{G}$. In order to derive Eq.~(\ref{eq:hf-1}),
697: we assumed an specific form to the hyperfine interaction. Following
698: the experimental results in $PrBr_{3}$ and $PrCl_{3}$, we assumed
699: that the dipolar part is the most relevant component in the hyperfine
700: Hamiltonian.
701: This conclusion arises from the hypothesis that the chemical bound is
702: truly ionic.
703: In general there are some covalent components to the bound and this
704: leads to a much stronger interaction with the electronic moments of
705: adjacent ions. For instance, this is preciselly what happens in $MnF_{2}$
706: \cite{AABB70,MF2}.
707: Whereas our hypothesis is based on the experimental facts in
708: $\textrm{PrBr}_{\textrm{3}}$
709: %the parent compound
710: \cite{SSRLA92,SSRLA+91,MDRLA83},
711: a thorough experimental study should be done to assert the hyperfine
712: Hamiltonian.
713:
714: In summary, we showed how a non-Kramers ionic crystal has unique properties
715: that can be exploited in a solid state NMR QC.
716: We propose that chemical substitutions in such system can be used to encode
717: quantum information
718: and, at the same time, break the spatial symmetries.
719: This \emph{controllable} symmetry-breaking can be used to act locally
720: in the magnetic environment of the qubit, thus, having important consequences
721: to decoherence and the construction of quantum gates.
722: We based our discussion in a well known family of materials.
723: However, the general principle that we put
724: forward can be applied in a much broader context.
725: In $\textrm{PrBr}_{\textrm{3-x}}\textrm{F}_\textrm{x}$, we
726: showed that a QC based on our ideas is scalable,
727: the decoherence rates are low, the interactions between qubits can be long ranged
728: and the qubits can be individually accessed with moderate magnetic field gradients.
729:
730: \begin{acknowledgments}
731: We would like to thank I.~Affleck, C.~Chamon, N.~Curro, C.~Hammel,
732: D.~Loss, E.~Mucciolo, M.~Silva Neto, R. de Sousa, and D.~Taylor
733: for illuminating discussions.
734: %and the referees for the critical review of our manuscript and suggestions.
735:
736: \end{acknowledgments}
737:
738: \begin{thebibliography}{25}
739:
740: \bibitem{NAG97}
741: N.A.~Gershenfeld and I.L.~Chuang, Science {\bf 275}, 350 (1997);
742: D.G.~Cory, A.F.~Fahmy, and T.F.~Havel,
743: Proc.~Nat.~Acad.~Sci.~USA {\bf 94}, 1634 (1997);
744: B.E.~Kane, Nature {\bf 393}, 133 (1998);
745: I.L.~Chuang, L.M.K.~Vandersypen, X.L.~Zhou, D.W~Leung and S.Lloyd,
746: Nature, {\bf 393}, 143 (1998).
747:
748: \bibitem{LMKV01}
749: L.M.K.~Vandersypen, M.~Steffen, G.~Breyta, C.S.~Yannoni, M.H.~Sherwood and
750: I.L.~Chuang,
751: Nature {\bf 414}, 883 (2001).
752:
753: \bibitem{LMKV02}
754: L.M.K.~Vandersypen, quant-ph/0205193.
755:
756: \bibitem{NLSP01}
757: N.~Linden and S.~Popescu, Phys.~Rev.~Lett. {\bf 87}, 047901 (2001).
758:
759: \bibitem{TDL+00}
760: T.D.~Ladd J.R.~Goldman, F.~Yamaguchi and Y.~Yamamoto, Appl.~Phys.~A
761: {\bf 71}, 27 (2000).
762:
763: \bibitem{SL95}
764: S.~Lloyd, Phys.~Rev.~Lett. {\bf 75}, 346 (1995);
765: D.~Deutsh, A.~Barenco, and A.~Ekert,
766: Proc.~R.~Soc.~Lond.~A {\bf 449}, 669 (1995) .
767:
768: \bibitem{HS58}
769: H.~Suhl, Phys.~Rev. {\bf 109}, 606 (1958);
770: T.~Nakamura, Prog.~Theor.~Phys. {\bf 20}, 542 (1958).
771:
772:
773: \bibitem{DMVP01}
774: D.~Mozyrsky, V.~Privman, and M.L.~Glasser,
775: Phys.~Rev.~Lett {\bf 86}, 5112 (2001).
776:
777:
778: \bibitem{AGTS+03}
779: A.~Goto, T.~Shimizu, K.~Hashi, H.~Kitazawa and S.~Ohki, Phys.~Rev.~A
780: {\bf 67}, 022312 (2003).
781:
782: \bibitem{BSBH+87}
783: B.~Schmid , B.~Hälg, A.~Furrer W.~Urland and R.~Kremer, J.~Appl.~Phys.
784: {\bf 61}, 3426 (1987).
785:
786: \bibitem{SSRLA92}
787: S.~Su, R.~L.~Armstrong, and W.~Wei,
788: J.~Phys.: Condens.~Matter {\bf 4}, 4301 (1992).
789:
790: \bibitem{AABB70}
791: A.~Abragam and B.~Bleaney, in {\em Electron Paramagnetic Ressonance of
792: Transition Ions}, W.~Marshall and D.H.~Wilkinson, eds.
793: (Oxford Univ. Press, London, 1970).
794:
795: \bibitem{JPH76}
796: J.P.~Harrison, J.P.~Hessler, and D.R.~Taylor,
797: Phys.~Rev.~B {\bf 14}, 2979 (1976).
798:
799: \bibitem{SSRLA+91}
800: S.~Su, R.L.~Armstrong, W.~Wei, and R.~Donaberger,
801: Phys.~Rev.~B {\bf 43}, 7565 (1991).
802:
803: \bibitem{AA}
804: A.~Abragam, in {\em The Principles of Nuclear Magnetism}, N.~F.~Mott,
805: E.~C.~Bullard, and D.~H.~Wilkinson, eds. (Oxford Univ. Press, London, 1961).
806:
807: \bibitem{MDRLA83}
808: M.~D'Iorio, R.L.~Armstrong, and D.R.~Taylor,
809: Phys.~Rev.~B {\bf 27}, 1664 (1983).
810:
811: \bibitem{Ian}
812: I.~Affleck,
813: in \emph{Les Houches 1988 - Session XLIX: fields, strings and critical
814: phenomena},
815: chap. Field Theory Methods and
816: Quantum Critical Phenomena, pp. 563--640, (North-Holland, Amsterdam, 1990).
817:
818: \bibitem{slichter}
819: C.P.~Slichter, in \emph{Principles of Magnetic Ressonance},
820: (Springer-Verlag, New York, 1990).
821:
822: \bibitem{AJL+87}
823: A.J.~Leggett, S.~Chakravarty, A.T.~Dorsey, M.P.A.~Fisher, A.~Garg,
824: and W.~Zwerger, Rev.~Mod.~Phys. {\bf 59}, 1 (1987).
825:
826: \bibitem{RKKY}
827: M.A.~Ruderman and C.~Kittel, Phys.~Rev. {\bf 96}, 99 (1954).
828:
829: \bibitem{SEIA94}
830: S.~Eggert, I.~Affleck, and M.~Takahashi,
831: Phys.~Rev.~Lett {\bf 73}, 332 (1994).
832:
833: \bibitem{errorcode}
834: E.~Knill, R.~Laflamme, W.H.~Zurek, Science {\bf 279} (5349), 342 (1998).
835:
836:
837: \bibitem{NCh} A.~Nielsen and I.L.~Chuang,
838: in \emph{Quantum Computation and Quantum Information},
839: (Cambridge University Press, New York, 2000).
840:
841: \bibitem{LJSUVV99}
842: L.J.~Schulman and U.V.~Vazirani,
843: in {\em Proceedings of the $31^\textrm{st}$ ACM
844: Symposium on Theory of Computing}, 322 (1999);
845: D.E.~Chang, L.M.K.~Vandersypen and M.~Steffen,
846: Chem.~Phys.~Lett. {\bf 338}, 337 (2001).
847:
848: \bibitem{CMU69}
849: J.H.~Colwell, B.W.~Mangum and D.B.~Utton,
850: Phys.~Rev. {\bf 181}, 842 (1969).
851:
852: \bibitem{MBEG1}
853: M.H.~Yang and C.P.~Flynn, Phys.~Rev.~Lett.{\bf 62},
854: 2476-2479 (1989).
855:
856: \bibitem{YF90}
857: M.H.~Yang and C.P.~Flynn, Phys.~Rev.~B {\bf 41}, 8500 (1990).
858:
859: \bibitem{MBEG2}
860: K.~Saiki, App.~Surf.~Sci. {\bf 113/114}, 9-17 (1997).
861:
862: \bibitem{MF2}
863: F.~Freyne, Phys.~Rev.~B {\bf 9}, 4824 (1974).
864: \end{thebibliography}
865:
866:
867: \end{document}
868: