1: \documentclass[amsmath,amssymb,aps,prb,twocolumn,groupedaddress,showpacs,floatfix]{revtex4}
2: \usepackage{graphicx}
3: \begin{document}
4:
5: \newcommand {\be}{\begin{equation}}
6: \newcommand {\ee}{\end{equation}}
7: \newcommand {\bea}{\begin{eqnarray}}
8: \newcommand {\eea}{\end{eqnarray}}
9: \newcommand {\nn}{\nonumber}
10:
11: \title{Quantum Phase Transitions in Coupled Dimer Compounds}
12:
13: \author{Omid Nohadani$^{(1)}$}
14: \author{Stefan Wessel$^{(2)}$}
15: \author{Stephan Haas$^{(1)}$}
16:
17: \affiliation{$^{(1)}$Department of Physics and Astronomy, University of Southern California, Los Angeles, CA 90089-0484}
18: \affiliation{$^{(2)}$Institut f\"ur Theoretische Physik III, Universit\"at Stuttgart, 70550 Stuttgart, Germany}
19:
20: %\date{\today}
21: \begin{abstract}
22: We study the critical properties in cubic systems of antiferromagnetically coupled spin dimers
23: near magnetic-field induced quantum phase transitions.
24: The quantum critical points in the zero-temperature phase diagrams are determined from quantum Monte Carlo simulations.
25: Furthermore, scaling properties of the uniform magnetization and the staggered transverse magnetization
26: across the quantum phase transition in magnetic fields are calculated.
27: The critical exponents are derived from Ginzburg-Landau theory.
28: We find excellent agreement between the quantum Monte Carlo simulations and the analytical results.
29: \end{abstract}
30:
31: \pacs{75.10.Jm, 73.43.Nq, 75.40.Cx}
32:
33: \maketitle
34:
35: \section{\label{int_sec}Introduction}
36: Recent improvements in high magnetic field technology have made detailed investigations of quantum phenomena in strong magnetic fields possible. In particular, they allow for
37: studies of quantum critical properties induced by high magnetic fields in weakly coupled spin dimer compounds, such as
38: $\rm TlCuCl_3$ \cite{rsttktmg,tlcucl,rtokuokh}, $\rm KCuCl_3$ \cite{rsttktmg,kcucl},
39: $\rm BaCuSi_2O_6$ \cite{bacusio}, and $\rm Sr_2Cu(BO_3)_2$ \cite{srcubo}.
40: The ground state of these materials consists of local spin singlets.
41: If a magnetic field is applied that exceeds their singlet-triplet excitation gap,
42: they undergo a transition
43: into a magnetically ordered state.
44: This quantum phase transition into a regime with transverse antiferromagnetic (AF) order
45: can be described as a Bose-Einstein condensation (BEC) of triplet excitations, which behave as bosonic
46: quasiparticles, called triplons.
47: In the corresponding Bose-Hubbard model, this is analogous to the transition from the Mott-insulating phase to the superfluid condensate,
48: where the magnetic field translates into the chemical potential.
49: At very high magnetic fields,
50: there is a saturation threshold, beyond which the spins fully align
51: along the field direction, and the transverse antiferromagnetic order is destroyed.
52: A schematic zero-temperature phase diagram is shown in Fig.\ref{phase_diagram}.
53: Because these compounds are three-dimensional,
54: field-induced transverse antiferromagnetic order persists up to a
55: finite transition temperature, $T_c(h)$, between the upper and lower critical field ($h_c\le h\le h_s$).
56: Approaching the critical field from the partially polarized phase, $h\rightarrow h_c$,
57: the critical temperature is expected to vanish as
58: $T_c (h) \propto |h - h_{c}|^{1/\alpha}$, with a universal power-law exponent that is predicted to be
59: $\alpha=3/2$.~\cite{sachdevbook,kawashima_qcp}
60: While early experimental \cite{rtokuokh} and numerical \cite{wessel} studies reported non-universal, i.e. coupling dependent, values of $\alpha>2$, it was recently shown in Ref. \onlinecite{universality} that
61: careful fitting of both experimental and numerical data indicates an effective exponent $\alpha(h)>3/2$, which approaches $3/2$
62: as $h\rightarrow h_c$. Motivated by this observation, Shindo and Tanaka\cite{shindo_tanaka} performed a
63: fit of specific heat data in $\rm TlCuCl_3$ in the close vicinity of the lower critical field, yielding an exponent of $\alpha = 1.67\pm0.07$. This value is closer to the
64: universal value of $\alpha$ found in Ref. \onlinecite{universality} than previous
65: experimental reports.
66:
67: While numerical and experimental results on critical properties of
68: the field-induced ordering transition agree well with
69: analytical results at finite temperatures\cite{bec,kawashima_qcp}, there are no
70: comparable scaling predictions for the zero-temperature quantum phase transitions.
71: In this paper, we present a numerical analysis of critical properties
72: in cubic systems of coupled dimers. In particular,
73: zero-temperature scaling properties of the uniform and staggered magnetization
74: in the partially polarized phase are studied, based
75: on large-scale quantum Monte Carlo (QMC) simulations.
76: \begin{figure}[h]
77: \includegraphics[width=8cm]{fig1}
78: \vspace{-2.2cm}
79: \caption{\label{phase_diagram}
80: Schematic zero-temperature phase diagram of a three-dimensional coupled dimer compound
81: with intra-dimer couplings $J$ and inter-dimer couplings $J'$.
82: $h$ denotes the strength of the external magnetic field.
83: }
84: \end{figure}
85: These systems, upon increasing the inter-dimer coupling, enter an antiferromagnetically ordered state at a quantum critical point
86: $(J'/J)_c$.~\cite{sachdevbook}
87: Magnetic exchange constants are known to depend sensitively on the
88: distance between the magnetic sites. If their relative magnitudes are
89: altered by the application of external pressure, such magnetic ordering transitions can be induced, as recently observed for $\rm TlCuCl_3$.
90: \cite{tanaka_pressure,tanaka_pressure2}
91: In this numerical study, the quantum critical point is determined for the
92: structures shown in Fig.~\ref{structures}. Furthermore,
93: the quantum criticality induced by an applied magnetic field is studied.
94: In particular, the scaling behavior of the uniform magnetization and
95: the antiferromagnetic order parameter upon entering the partially polarized phase is determined. The associated critical exponents
96: can be accessed experimentally and thus allow for a quantitative comparison.
97: \begin{figure}[h]
98: \includegraphics[width=9cm]{fig2}
99: \vspace{-6.0cm}
100: \caption{Layers of coupled dimers, with staggered (left panel) and aligned (right panel) arrangements of dimers. The
101: dimer bonds ($J$) are denoted by solid bars, whereas the inter-dimer bonds ($J'$) are denoted by dashed lines. In the
102: three-dimensional crystal, these layers are coupled by interlayer couplings of the same strength as $J'$.
103: The arrangement of the dimers in neighboring layers is staggered (left panel), or aligned (right panel), respectively.}
104: \label{structures}
105: \end{figure}
106:
107: The paper is organized as follows:
108: in the next section, the model used in the numerical simulations is defined and details about the quantum Monte
109: Carlo method are presented. In Sec. \ref{phasediag_sec}, the zero-temperature phase diagram of the systems is discussed, and the
110: quantum critical points at zero magnetic field are determined.
111: A detailed scaling analysis of the magnetic properties
112: at the quantum phase transitions in a magnetic field is presented in Sec. \ref{qpt_sec}. Finally, conclusions are given in Sec.\ref{sum_sec}.
113: The Appendix contains
114: a phenomenological derivation of the mean-field exponents, observed in Sec. \ref{qpt_sec}, within Ginzburg-Landau theory.
115:
116: \section{\label{num_sec}Model and Method}
117:
118: We consider the spin-1/2 Heisenberg antiferromagnet on the lattice structures shown in Fig.~\ref{structures}. The Hamiltonian is given by
119: \bea
120: H = \sum_{\langle i,j \rangle} J_{ij} {\bf S}_i \cdot {\bf S }_j - h \sum_i S^z_i ,
121: \eea
122: where the ${\bf S}_i$ denote localized spin-1/2 moments, and
123: $J_{ij}$ indicates the coupling constant between sites $i$ and $j$, which takes values $J$ for the dimer, and J'
124: for the inter-dimer couplings. Furthermore, $h$ denotes the applied magnetic field.
125:
126: Since the lattice structures that are considered here, are bipartite
127: (c.f. Fig.~\ref{structures}), antiferromagnetism in these systems is
128: not frustrated,
129: so that their properties can be studied
130: using large-scale QMC, without a sign-problem~\cite{sgn}.
131: Here, we use the stochastic series expansion QMC
132: method~\cite{sse,sandvik_green}
133: with directed loop updates~\cite{directed_loop,alet}.
134: This update scheme results in
135: higher sampling efficiency than other methods based
136: on the conventional operator-loop
137: update. Close to criticality,
138: autocorrelation times are reduced by up to an order of magnitude.
139: This allows detailed simulations of ground state properties on
140: clusters with up to 10$\;\!$000 sites, even in the presence of large magnetic fields.
141: Furthermore, off-diagonal observables such as the transverse magnetic structure factor
142: can be measured efficiently during the directed loop construction.\cite{troyer,alet}
143: Therefore, the
144: order parameter in the partially polarized phase, i.e. the staggered transverse magnetization perpendicular to the magnetic field direction, can be calculated by using
145: \begin{equation}
146: \label{perp_stagg_mag}
147: m_{\rm s}^{\perp} = \sqrt{\frac{S^\perp_{\rm s}}{L^3}}.
148: \end{equation}
149: Here, $L$ denotes the linear system size and $S^\perp_{\rm s}$ the
150: transverse staggered structure factor,
151: \begin{equation}
152: \label{stagg_structure_factor}
153: S^\perp_{\rm s} = \frac{1}{L^3}\sum_{\langle i,j \rangle} (-1)^{i+j}
154: \langle S^{x}_i S^{x}_j \rangle .
155: \end{equation}
156: These observables have been shown to saturate in QMC simulations at temperatures below $T=J'/2L$, which is used throughout this study to ensure zero-temperature behavior.
157:
158: \section{\label{phasediag_sec}Zero-Temperature Phase Diagram}
159:
160: The zero-temperature
161: phase diagram of coupled dimer systems in a magnetic field
162: can be obtained using the bond-operator mean-field theory~\cite{matsumoto,matsumoto_pressure}.
163: A schematic phase diagram is shown in Fig.~1.
164: It consists of three
165: phases: (i) at low fields and small coupling ratios $J'/J$, the system
166: is in a magnetically disordered phase, i.e. a dimer spin liquid; (ii) at intermediate fields
167: and/or sufficiently large values of $J'/J$, the ground state is partially spin
168: polarized and has an antiferromagnetic long-range order transverse to the magnetic field direction; and
169: (iii) at large fields, $h>h_s=J+5J'$ all spins are fully polarized.
170: These phase separations occur in both dimer arrangements, shown
171: in Fig.~\ref{structures}.
172:
173: While bond-operator theory provides a reliable description of the phase diagram~\cite{universality},
174: a more precise estimate of the critical inter-dimer coupling strength, $(J'/J)_c$, is required for the
175: study of critical properties in finite magnetic fields, presented below.
176: In order to determine this zero-field quantum critical point,
177: we perform a finite-size scaling analysis of
178: the staggered magnetization obtained from QMC simulations for various system sizes. Defining the dimensionless coupling ratio $g=J'/J$,
179: the relevant finite-size scaling is obtained as follows.
180: The correlation length $\xi$ diverges near the quantum critical point $g_c$ as $\xi \propto |g-g_c|^{- \nu}$. The correlation time $\tau_c$
181: during which fluctuations relax and decay (equilibration),
182: is related to the correlation length $\xi$ via
183: $\tau_c \propto \xi^z \propto |g-g_c|^{- \nu z}$,
184: with the dynamical critical exponent $z$.\cite{vojta_qpt}
185: In the vicinity of the critical point and at zero-field,
186: the staggered magnetization $m_{\rm s}^\perp = m_{\rm s}$ for $g>g_c$ scales as
187: \be
188: \label{stag_mag_scale}
189: m_{\rm s} \propto (g-g_c)^\beta ,
190: \ee
191: defining an exponent $\beta$.\cite{troyer_exponent}
192: In general, this implies a finite-size scaling relation of the
193: order parameter at the quantum critical point, $g=g_c$:
194: \be
195: \label{stag_mag_L_scaling}
196: m_{\rm s} \propto L^{-\beta/\nu}.
197: \ee
198: Since these systems are explicitly dimerized, the Berry-phase contributions to the path-integral cancel.~\cite{science}
199: Therefore, this QPT
200: belongs to the universality class of the classical $O(3)$ Heisenberg model in $3+1$ dimensions, with the dynamical critical exponent $z=1$.
201: Since the effective classical model is at the upper critical dimension ($d_c=4$),
202: the critical exponents $\nu$ and $\eta$ take on mean-field values
203: $\nu=1/2$, $\eta=0$, and the above scaling laws hold up to logarithmic corrections.\cite{sachdevbook}
204:
205: A scaling plot of $m_{\rm s}^\perp L$ is shown in Fig.~\ref{crit_coupl}, using data from system sizes $L=10$ to $20$.
206: On the scale of the main part of Fig.~\ref{crit_coupl}, a common intersection point of the finite-size data appears to exist.
207: However, on a smaller scale, the insets of Fig.~\ref{crit_coupl} exhibit the logarithmic corrections to Eq.(\ref{stag_mag_L_scaling}).
208: With increasing system size, we observe that the
209: intersections for neighboring system sizes move towards increasing values of $(J'/J)$.
210: As seen
211: in inset (b) of Fig.~\ref{crit_coupl}, the systematic increase of the crossing points scales well as a function of $1/L^2$, which allows
212: us to extrapolate the thermodynamic limit value of
213: $(J'/J)_c = 0.2492\pm0.0002$ for the quantum critical point.
214: These logarithmic corrections are best accessible when system-sizes
215: of different orders of magnitude are compared.
216: However, because of computational restrictions, we follow the approach described above.
217:
218: \begin{figure}[h]
219: \includegraphics[width=8cm]{fig3}
220: \vspace{-0.3cm}
221: \caption{\label{crit_coupl}
222: Scaling plot of the zero-temperature staggered magnetization in the aligned arrangement of Fig.~2~(b) obtained
223: from quantum Monte Carlo simulations using systems of linear sizes $L=10$ to $20$. $J$ denotes the intra-dimer coupling.
224: At the critical inter-dimer coupling $(J'/J)_c$, the
225: different curves intersect each other (main panel). Inset (a) shows a magnification
226: of the intersection region. Inset (b) exhibits the corrections to the mean-field scaling behavior, arising from logarithmic corrections in the upper critical dimension.
227: The statistical error bars fall within the symbol size.}
228: \end{figure}
229:
230:
231: Performing a similar analysis for the staggered configuration of dimers in Fig.~\ref{structures}(a),
232: we find the quantum critical point at $(J'/J)_c =0.2170\pm0.0002$.
233: This lower critical inter-dimer coupling for the staggered configuration indicates
234: a reduced tendency towards the formation of dimer singlets for this arrangement of dimers, as compared
235: to the aligned dimer configuration.
236: \begin{figure}[h]
237: \includegraphics[width=8.5cm,height=7cm]{fig4}
238: \vspace{-3cm}
239: \caption{\label{m_j}
240: Staggered magnetization $m_s$ of coupled dimer arrays as a function of the inter-dimer coupling strength $J'$,
241: for the two different dimer arrangements of Fig. 2. Dashed lines show the mean-field scaling behavior with
242: exponent $\beta=\frac{1}{2}$ close to the quantum critical point.
243: }
244: \end{figure}
245:
246: Beyond the critical inter-dimer coupling, the staggered magnetization increases with a mean-field exponent $\beta=1/2$, as in Eq.~(\ref{stag_mag_scale}). The QMC data of $m_s$, shown in Fig.~\ref{m_j}, are consistent with this scaling behavior.
247: However, due to restricted system sizes, we are not able to extract
248: the logarithmic corrections to the scaling law.
249:
250: \section{\label{qpt_sec}Field-Induced Quantum Phase Transition}
251:
252: Having determined the quantum critical points of coupled-dimer arrays, we proceed to study the
253: effects of a magnetic field in the different regimes. In particular, we determine the scaling behavior of the uniform magnetization
254: and the order parameter upon entering the partially polarized region. The corresponding scaling exponents
255: are accessible experimentally by direct measurement of the magnetization, and by neutron scattering.
256: Thus, the numerical results can be compared with analytical predictions and with measurements on the materials mentioned above.
257: Here, simulations for the aligned configuration of dimers, shown in Fig.~\ref{structures}(b), are presented.
258: For the staggered configuration of dimers, the same scaling exponents are obtained. This is due to the underlying universality.
259:
260:
261: \subsection{\label{mu_sec}Scaling of the Uniform Magnetization}
262:
263: First, we discuss the behavior of the uniform magnetization $m_u(h)$ as a function of the applied magnetic field.
264: In the dimer spin liquid phase, i.e. for coupling ratios smaller than
265: $(J'/J)_c$, a finite magnetic excitation gap $\Delta$ separates the ground state
266: singlet and the lowest triplet state. Thus, a finite magnetic field $h_{c}=\Delta$ is required to close this gap, and to induce
267: a finite uniform magnetization.
268: The Ginzburg-Landau approach presented in Appendix A predicts $m_u$ to increase linearly, $m_u\propto (h-h_{c})$
269: for $h>h_{c}=\Delta$ and for $J'<J'_c$.
270: When the inter-dimer coupling is increased, the gap $\Delta$ becomes smaller, until it vanishes at the critical
271: inter-dimer coupling $J'_c$. At the quantum critical point, $J'=J'_c$, the Ginzburg-Landau approach
272: predicts the uniform magnetization to scale as $m_u\propto h^3.$
273: For larger values of $J' > J'_c$, the excitation
274: gap remains zero, and the finite uniform susceptibility $\chi_u$ results in a linear response $m_u=\chi_u h$ of the uniform
275: magnetization in the N\'eel-ordered regime beyond $J'_c$.
276:
277: Let us now compare these predictions with the QMC results obtained from simulations
278: of systems with linear sizes up to $L=30$. For such large system sizes, we did not detect finite-size effects in the uniform
279: magnetization. In Fig.~\ref{umag}, results for various
280: values of the inter-dimer coupling strength are presented.
281: The main part of Fig.~\ref{umag} shows the uniform magnetization over the full range of magnetic field strengths, up to
282: the saturation field $h_s=J+5J'$. Inset (a) of Fig.~\ref{umag} focuses on the region close to the critical field $h_c$ for various
283: inter-dimer couplings $J'$.
284: The linear scaling in $h$, which is predicted on both sides of the quantum critical point, is clearly observed in this small-$m_u$ region.
285: In contrast, at the quantum critical point, $J' = J'_c$, the uniform magnetization increases non-linearly with $h$, as shown in Fig.~\ref{umag}~(a).
286: Indeed, we observe a scaling $m_u\propto h^3$, presented in Fig.~\ref{umag}~(b).
287: This is expected from the Ginzburg-Landau theory
288: (c.f. Appendix A) and the bond-operator mean-field theory\cite{matsumoto_pressure}.
289: Using QMC, these scaling exponents of the uniform magnetization are verified.
290:
291: Finally, we note that the field dependence of $m_u$ for the entire region between $h_{c}$ and $h_{s}$ is linear for weakly coupled
292: dimers (see e.g. Fig.~\ref{umag} for $J'=0.07J$), consistent with
293: bond-operator mean-field theory~\cite{matsumoto_pressure}.
294: For larger values of $J'$, $m_u$ shows deviations from this linear behavior in high magnetic fields.
295: This non-linear behavior can be accounted for within bond-operator
296: theory by
297: including the contributions of higher-energy triplet modes to the ground state~\cite{matsumoto_pressure}.
298:
299: \begin{figure}[h]
300: \includegraphics[width=8.5cm,height=7cm]{fig5}
301: \vspace{-0.4cm}
302: \caption{\label{umag}
303: Zero-temperature uniform magnetization of aligned dimer arrays for different inter-dimer couplings as a function
304: of the magnetic field $h$. Inset (a) focuses on the low-field region, $h/J<1$.
305: The scaling $m_u\propto h^3$ at $J'=J'_c$ is demonstrated in inset (b).
306: }
307: \end{figure}
308:
309:
310: \subsection{\label{ms_sec}Scaling of the Order Parameter}
311:
312: Next, we discuss the scaling properties of the order parameter in the partially polarized phase,
313: i.e. the staggered transverse magnetization perpendicular to the
314: magnetic field direction, $m_s^{\perp}(h)$, as a function of the applied magnetic field $h$.
315: Before presenting numerical data, the expectations from the Ginzburg-Landau theory from Appendix A can be summarized as following.
316: For $J'<J'_c$ and magnetic fields $h>h_c=\Delta$, $m_s^{\perp}\propto
317: \sqrt{(h-h_c)}$ is expected, consistent with bond-operator mean-field theory~\cite{matsumoto_pressure}.
318: At the critical inter-dimer coupling, $J'=J'_c$, a linear relation $m_s^{\perp}\propto h$ is expected.
319: Within the antiferromagnetically ordered regime, $J'>J'_c$, a finite staggered magnetization $m_s^{\perp}=m_s$ exists.
320: For small fields, the order parameter scales with $m_s^{\perp}-m_s\propto h^2$, as presented in Ref.~\onlinecite{matsumoto_pressure}.
321:
322: QMC calculations of $m_s^{\perp}$ for different values of $J'/J$ in a system with $L=16$ are shown in Fig.~\ref{smag}.
323: This Figure demonstrates that for $J'\le J'_c$, the staggered magnetization
324: is largest for $h=(h_c+h_s)/2$ and decreases upon approaching $h_c$ and $h_s$. In the thermodynamic limit, $m_s^{\perp}$ vanishes in the fully polarized regime, $h>h_s$ and below $h_c$ for weakly coupled dimers.
325: In contrast to the uniform magnetization, the field dependence of $m_s^{\perp}$ shows a strong finite-size dependence.
326: Finite-size effects are most pronounced for magnetic field regimes outside the partially polarized phase, i.e. for $h<h_c$ and
327: $h>h_s$, where
328: $m_s^{\perp}=0$ in the thermodynamic limit. On the finite systems accessible to QMC, the values of $m_s^{\perp}$, defined in Eq. (\ref{perp_stagg_mag}) do not vanish. However, they scale to zero upon increasing the system size.
329: In order to extract the scaling behavior of $m_s^{\perp}$, we thus need to perform a
330: careful finite-size scaling analysis of the numerical data.
331:
332: \begin{figure}[h]
333: \includegraphics[width=8cm]{fig6}
334: \vspace{-0.3cm}
335: \caption{\label{smag}
336: Zero-temperature staggered transverse magnetization $m_s^{\perp}$
337: in coupled aligned dimer arrays as a function of an applied
338: magnetic
339: field $h$. Quantum Monte Carlo data of a $16\times 16\times 16$ system are shown for different values of the inter-dimer coupling $J'$. The
340: intra-dimer coupling is denoted by $J$.
341: The off-sets of $m_s^{\perp}$ outside the range $h_c\le h\le h_s$
342: are due to finite-size effects.
343: }
344: \end{figure}
345:
346: Such an analysis can be directly applied in the weak coupling regime, for $J'<J'_c$. Here, the field-induced ordering
347: transition is known
348: to constitute a Bose-Einstein condensation of triplons,\cite{sachdevbook}
349: with a dynamical critical exponent $z=2$, reflecting the
350: quadratic dispersion relation of these bosonic excitations.
351: Therefore, the quantum phase transition at the critical field $h_c$ is in the universality class of the classical 5-dimensional $O(2)$ model, and a finite-size scaling analysis similar to the one performed for the
352: zero-field quantum critical point in Sec. \ref{mu_sec} can be applied. Since the classical theory is now above $d_c=4$, mean-field scaling without logarithmic corrections applies.
353: In particular, at the critical field $h_c$, one finds
354: \be
355: \label{m_l_scale}
356: m_{\rm s}^{\perp} \; L \propto F(L/\xi)\;,
357: \ee
358: where $F$ is a scaling function that depends on the ratio of the correlation length $\xi$ and the system size $L$.
359: In order to determine $h_c$ for a given value of $J'<J_c$, we can thus construct a scaling plot of $m_s^{\perp} L$ as a function of $h/J$.
360: Fig.~\ref{smag_02} illustrates an example of such a plot for
361: an inter-dimer coupling $J'=0.2 J$. In Fig.~\ref{smag_02}(a),
362: data for $m_s^{\perp} L$ are shown for system sizes $L$ = 10, 14, 18, and 20.
363: The intersection point of the different finite-size data allows the extraction of the critical field $h_c=0.383\pm 0.001$ for this value of
364: $J'=0.2 J$.
365:
366: \begin{figure}[h]
367: \includegraphics[width=8.5cm]{fig7}
368: \vspace{-0.3cm}
369: \caption{\label{smag_02}
370: Scaling plot (a) and data collapse (b) for the zero-temperature staggered transverse magnetization $m_s^{\perp}$
371: of weakly coupled aligned dimers as a function of the applied magnetic field $h$ for a coupling-ratio of $J'/J=0.2$.
372: Results from quantum Monte Carlo simulations are shown for systems sizes of $L=10$ to $20$.
373: }
374: \end{figure}
375: Once the critical field $h_c$ is determined, data collapse of the finite-size data
376: is verified in the vicinity of $h_c$.
377: Namely, from the scaling of the correlation length close to the critical field, $\xi\propto |h-h_c|^{- \nu}$, one obtains
378: \be
379: \label{data_collapse}
380: m_{\rm s}^{\perp} \; L \propto \tilde{F}(L^{1/\nu}\; |h-h_c|\;)\;,
381: \ee
382: with a new scaling function $\tilde{F}$ and $\nu=1/2$.
383: An example of the data collapse for $J'=0.2 J$ is shown in Fig.~\ref{smag_02}~(b).
384:
385: In order to extract the behavior of $m_s^{\perp}(h)$ for $h>h_c$ in the thermodynamic limit,
386: we perform an extrapolation of the finite-size data $m_s^{\perp}(L)$ as a function
387: of $1/L$.
388: An example of such an extrapolation, again for $J'/J=0.2$ is shown in Fig.~\ref{smag_ex_02} for various values of $h$.
389: While for $h$ close to $h_c$, a linear scaling in $1/L$ is obtained, extrapolation away from $h_c$ requires
390: the use of higher order polynomials. Using these extrapolated values of $m_s^{\perp}(h)$, representing the thermodynamic limit,
391: we then obtained the scaling behavior of $m_s^{\perp}(h)$ shown for different values of $J'/J$ in Fig.~\ref{m_h_inf}.
392:
393: \begin{figure}[h]
394: \includegraphics[width=8.5cm,height=8cm]{fig8}
395: \vspace{-2.0cm}
396: \caption{\label{smag_ex_02}
397: Extrapolation of finite-size data of staggered magnetization for different
398: magnetic fields close to the transition. At $h_c$, thermodynamic limit is obtained
399: from a $1/L$-extrapolation. Higher order polynomials are required away from
400: criticality.
401: For all data presented here, up to the third order is used.
402: Dashed lines are for $h<h_c$.
403: }
404: \end{figure}
405:
406: For weakly coupled dimers with $J'=0.07J$ Fig.~\ref{m_h_inf}(a) exhibits a scaling of $m_{\rm s}^{\perp}\propto \sqrt{h-h_c}$.
407: The scaling exponent $\frac{1}{2}$ is also found for stronger
408: inter-dimer couplings, as shown in Fig.~\ref{m_h_inf}(b) for $J'=0.2J$.
409: This square root dependence $m_{\rm s}^{\perp} \propto \sqrt{h - h_{c}}$
410: is consistent with the analytical results and expected from the mean-field value $\beta=1/2$.
411:
412: Next, let us consider the scaling of $m_s^{\perp}(h)$ for a critical inter-dimer coupling $J'=J'_c$. In Fig.~\ref{m_h_inf}(c),
413: we show results from simulations at the critical point $J'=0.249$.
414: Here, a linear scaling
415: $m_{\rm s}^{\perp} \propto h $ is observed, consistent with the Ginzburg-Landau calculations from Appendix A.
416: Finally, we consider the N\'eel-ordered phase ($J'>J'_c$). The results from the extrapolated QMC data is shown for $J'=0.3 J$ in
417: Fig.~\ref{m_h_inf}(d). Increasing the magnetic field, a non-linear increase of $m_s^{\perp}(h)$ from its zero-field value $m_s$ is found,
418: which can
419: be fitted well to the analytical prediction, $m_s^{\perp}(h)-m_s\propto h^2$.
420: We thus obtain agreement of the QMC data with the analytical results based on Ginzburg-Landau and bond-operator mean-field
421: theory also for the scaling of the order parameter.
422: \vspace{-2mm}
423: \begin{figure}[h]
424: \includegraphics[width=8.0cm]{fig9ab}
425: \includegraphics[width=8.0cm]{fig9cd}
426: \vspace{-0.1cm}
427: \caption{\label{m_h_inf}
428: Scaling behavior of the zero-temperature staggered transverse magnetization
429: as a function of the applied magnetic field $h$. Quantum Monte Carlo data
430: are shown after extrapolation of finite-size data ($L=10-20$) to the thermodynamic limit.
431: (a)(b) in the
432: dimer spin liquid phase ($J' < J'_c$), $(m_{\rm s}^{\perp})^2$ scales linear
433: with $h - h_c$.
434: (c) near the critical coupling ($J'_c$),
435: $m_{\rm s}^{\perp}$ scales linear with $h - h_c$.
436: (d) for $J' > J'_c$, $m_{\rm s}^{\perp}-m_s$ increases quadratically with $h$.
437: }
438: \end{figure}
439:
440: %\vspace{-1cm}
441: \section{\label{sum_sec}Summary}
442: We examined quantum phase transitions in three-dimensional coupled dimer arrays.
443: The zero-temperature phase diagrams of these systems feature a low-field dimer spin liquid phase at weak inter-dimer couplings, a partially polarized regime with
444: long-range transverse magnetic order for intermediate magnetic fields
445: $h_c\le h\le h_s$, and a fully polarized phase at high magnetic fields.
446:
447: The critical exponents associated with quantum phase transitions
448: between these regimes were extracted using
449: finite-size scaling analysis of quantum Monte Carlo data.
450: The numerical values of these exponents compare well with
451: Ginzburg-Landau calculations and bond-operator mean-field theory.
452: In particular, for small inter-dimer coupling, the order parameter is found to scale
453: as $(h - h_{c})^{1/2}$, whereas the uniform magnetization scales linear in $h - h_{c}$. Moreover, at the quantum critical point $(J'/J)_c$, we demonstrated linear scaling for $m_{\rm s}^{\perp} \propto h$.
454: In a recent magnetization study on the $\rm TlCuCl_3$ system
455: under hydrostatic pressure~\cite{tanaka_pressure2},
456: a cubic scaling of the uniform magnetization $m_u\propto h^3$ at the
457: critical value of the applied pressure was observed. This
458: is in perfect agreement with numerical results and analytical predictions based on Ginzburg-Landau calculations and bond-operator theory.
459:
460: We found that corrections to mean-field scaling emerge at the
461: zero-field pressure-induced quantum phase
462: transition, such as those in
463: Ref.~\onlinecite{tanaka_pressure,tanaka_pressure2}.
464: For example, in
465: Ref. \onlinecite{tanaka_pressure2}, the pressure-dependence of the gap was fitted to a power law with an exponent 0.445, while
466: the mean-field theory predicts an exponent 0.5, without taking logarithmic corrections into account.\cite{matsumoto_pressure}
467: Assuming a linear scaling of the pressure $P$ with the inter-dimer interactions,~\cite{tanaka_pressure2} close to the critical point $P_c$,
468: we find the scaling of the
469: gap $\Delta$ to be
470: \be
471: \Delta \propto 1/\xi \propto
472: (P-P_c)^{1/2}\left(-\ln\left|\frac{P-P_c}{P_c}\right|\right)^{1/6}.
473: \ee
474: from a renormalization group analysis of the classical $\phi^4$-theory at
475: $d_c=4$~\cite{zinn_justin}.
476: Therefore, careful fitting of experimental data can exhibit the logarithmic
477: corrections near quantum critical points in three-dimensional compounds.
478:
479: \section*{Acknowledgment}
480: We thank Andreas Honecker,
481: Bruce Normand, Tommaso Roscilde, Subir Sachdev
482: Manfred Sigrist, Hidekazu Tanaka, Matthias Troyer, and Matthias Vojta for useful discussions.
483: Furthermore, we acknowledge financial support from NSF Grant No.
484: DMR-0089882. Computational support was provided
485: by the USC Center for High Performance Computing and Communications.
486: Parts of the numerical simulations were performed using the ALPS project library.~\cite{ALPS}
487:
488: \appendix
489:
490: \section{\label{gb_sec}Ginzburg-Landau Theory}
491: In this Appendix, we present a phenomenological
492: Ginzburg-Landau description of weakly coupled dimers in a magnetic field~\cite{ehrenfest},
493: and derive the scaling exponents of the uniform magnetization $m_u$ and
494: the order parameter $m_s^{\perp}$ in the partially polarized phase.
495:
496: Due to rotational symmetry
497: within the plane perpendicular to the external field, the free
498: energy of the system can be expanded as
499: \begin{equation}
500: \label{free_e}
501: F = a(h,J') \; (m_{\rm s}^{\perp})^2 + b \; (m_{\rm s}^{\perp})^4 ,
502: \end{equation}
503: neglecting higher orders of $m_{\rm s}^{\perp}$
504: in the vicinity of the critical field.\cite{GL}
505: Here, $a(h,J')$ vanishes for $h$ equal to the critical magnetic field $h=h_c(J')$.
506: Taking into account time reversal
507: symmetry, $a$ can be expressed as
508: $a = \tilde{ a}\cdot({h}^2 -h_c^2(J'))$.
509: The staggered magnetization
510: $m_{\rm s}^{\perp}$ and the uniform magnetization $m_u$ is obtained
511: from Eq.~\ref{free_e} as
512: \be
513: \frac{\partial F}{\partial m_{\rm s}^{\perp}} \Bigg|_{h,J'}= 0 \;\;
514: \mbox{and} \;\;\; m_u = -\frac{\partial F}{\partial h} \Bigg|_{J'}.
515: \ee
516: From this, one finds
517: \begin{eqnarray}
518: \label{gl_ms_mu}
519: m_{\rm s}^{\perp} &=& \sqrt{\frac{\tilde{a}(h^2-h_c^2)}{2b}} \\
520: m_u &=& 2 \tilde{a} h (m_{\rm s}^{\perp})^2\;.
521: \end{eqnarray}
522:
523: To proceed further, we distinguish two regions, namely weak inter-dimer
524: couplings ($J'<J'_c$) and the quantum critical point ($J' = J'_c$). For the
525: first region, (\ref{gl_ms_mu}) can be expanded in terms of the distance to the
526: lower critical field, i.e. $h-h_{c}$. One obtains
527: \begin{eqnarray}
528: \hspace{-2cm} \mbox{for } J' < J'_c\; \mbox{:} \hspace{0.5cm}&&
529: \hspace{2cm} \nonumber \\
530: m_{\rm s}^{\perp} &=& \sqrt{
531: \frac{\tilde{a}(h+h_{c})(h-h_{c})}{2b}}, \nonumber \\
532: \Rightarrow m_{\rm s}^{\perp} &\propto& (h-h_{c})^{1/2}\;,\\
533: \Rightarrow m_u &\propto& 2 \tilde{a} h (h-h_{c})^{1}\;.
534: \end{eqnarray}
535:
536: At the critical inter-dimer coupling the critical field is zero.
537: Therefore,
538: \begin{eqnarray}
539: \hspace{-2cm} \mbox{for } J' = J'_c\; \mbox{:} \hspace{0.8cm}&&
540: \hspace{3cm} \nonumber \\
541: m_{\rm s}^{\perp} &=& \sqrt{
542: \frac{\tilde{a}}{2b}} \; \; h\;, \\
543: \Rightarrow m_u &\propto& 2 \tilde{a} h^{3} .
544: \end{eqnarray}
545:
546: These exponents agree with the numerical data
547: presented in the preceding sections, as well as
548: previous bond-operator mean-field theory results~\cite{matsumoto_pressure}.
549:
550:
551:
552: \begin{thebibliography}{99}
553:
554: \bibitem{rsttktmg} W. Shiramura, K. Takatsu, H. Tanaka, K. Kamishima,
555: M. Takahashi, H. Mitamura, and T. Goto, J. Phys. Soc. Jpn. {\bf 66},
556: 1900 (1997).
557:
558: \bibitem{tlcucl} A. Oosawa, M. Ishii, and H. Tanaka, J. Phys. Condens.
559: Matter {\bf 11}, 265 (1999); N. Cavadini, G. Heigold, W. Henggeler, A.
560: Furrer, H.-U. G\"udel, K. Kr\"amer, and H. Mutka, Phys. Rev. B {\bf 63},
561: 172414 (2001); Ch. R\"{u}egg, N. Cavadini, A. Furrer, H.-U. G\"{u}del, K. Kr\"{a}mer, H. Mutka, A. Wildes, K. Habicht, and P. Vorderwisch, Nature
562: (London) {\bf 423}, 62 (2003).
563:
564: \bibitem{rtokuokh} H. Tanaka, A. Oosawa, T. Kato, H. Uekusa, Y. Ohashi,
565: K. Kakurai, and A. Hoser, J. Phys. Soc. Jpn {\bf 70}, 939 (2001).
566:
567: \bibitem{kcucl} T. Kato, K. Takatsu, H. Tanaka, W. Shiramura, M. Mori,
568: K. Nakajima, and K. Kakurai, J. Phys. Soc. Jpn. {\bf 67}, 752 (1998);
569: A. Oosawa, T. Takamasu, K. Tatani, H. Abe, N. Tsujii, O. Suzuki,
570: H. Tanaka, G. Kido, and K. Kindo, Phys. Rev. B {\bf 66}, 104405 (2002).
571:
572: \bibitem{bacusio} M. Jaime, V. F. Correa, N. Harrison, C. D. Batista, N. Kawashima, Y. Kazuma, G.A. Jorge, R. Stern, I. Heinmaa, S.A. Zvyagin, Y. Sasago, and K. Uchinokura, Phys. Rev. Lett. {\bf93}, 087203 (2004).
573:
574: \bibitem{srcubo} S. Sebastian, D. Yin, P. Tanedo, G. A. Jorge, N. Harrison, M. Jaime, Y. Mozharivskyj, G. Miller, J. Krzystek, S. A.
575: Zvyagin, and I. R. Fisher, Report cond-mat/0403334.
576:
577: \bibitem{sachdevbook}
578: S. Sachdev, Quantum Phase Transitions, (Cambridge University Press, 1999).
579:
580: \bibitem{kawashima_qcp} N. Kawashima, Report cond-mat/0408649.
581:
582: \bibitem{wessel} S. Wessel, M. Olshanii, and S. Haas, Phys. Rev. Lett.
583: {\bf 87}, 206407 (2001).
584:
585: \bibitem{universality}
586: O. Nohadani, S. Wessel, B. Normand, and S. Haas, Phys. Rev.
587: B {\bf 69}, 220402 (2004).
588:
589: \bibitem{shindo_tanaka}
590: Y. Shindo and H. Tanaka, J. Phys. Soc. Jpn. Vol.{\bf 73} No.10 (2004).
591:
592:
593: \bibitem{bec} T. Giamarchi and A. M. Tsvelik, Phys. Rev. B {\bf 59},
594: 11398 (1999); T. Nikuni, M. Oshikawa, A. Oosawa, and H. Tanaka,
595: Phys. Rev. Lett. {\bf 84}, 5868 (2000).
596:
597: \bibitem{tanaka_pressure}
598: A. Oosawa,, M. Fujisawa, T. Osakabe, K. Kakurai, and H. Tanaka,
599: J. Phys. Soc. Jpn. Vol.{\bf 72} No.5 (2003).
600:
601: \bibitem{tanaka_pressure2}
602: K. Goto, M. Fujisawa, T. Ono, H. Tanaka, and Y. Uwatoko, J. Phys. Soc. Jpn. Vol.{\bf 73} No.12 (2004).
603:
604: \bibitem{sgn} M. Troyer and U.-I. Wiese, Report cond-mat/0408370.
605:
606: \bibitem{sse} A. W. Sandvik, Phys. Rev. B {\bf 59}, R14157 (1999).
607:
608: \bibitem{sandvik_green} A. W. Sandvik, Phys. Rev. B {\bf 56}, 11678 (1997); A. W. Sandvik, Phys. Rev. Lett. {\bf 80}, 5196 (1998).
609:
610: \bibitem{directed_loop} O. F. Sylju{\aa}sen and A. W. Sandvik,
611: Phys. Rev. E {\bf 66}, 046701 (2002).
612:
613: \bibitem{alet} F. Alet, S. Wessel, and M. Troyer, Report cond-mat/0308495.
614:
615: \bibitem{troyer} A. Dorneich and M. Troyer, Phys. Rev. E {\bf 64},
616: 066701(2001).
617:
618: \bibitem{matsumoto} M. Matsumoto, B. Normand, T. M. Rice, and M. Sigrist,
619: Phys. Rev. Lett. {\bf 89}, 077203 (2002).
620:
621: \bibitem{matsumoto_pressure}
622: M. Matsumoto, B. Normand, T. M. Rice, and M. Sigrist, Phys. Rev. B {\bf 69}, 054423 (2004); B. Normand, M. Matsumoto, O. Nohadani, S. Wessel, S. Haas, T.M. Rice, and M. Sigrist, J. Phys.: Condens. Matter {\bf 16}, No 11, S867-S873 (2004).
623:
624: \bibitem{vojta_qpt} M. Vojta, Rep. Prog. Phys. {\bf 66}, 2069 (2003).
625: A. V. Chubukov, S. Sachdev, and J. Ye, Phys. Rev. B {\bf 49}, 11919 (1994).
626: S. Chakravarty, B. I. Halperin, and D. R. Nelson, Phys. Rev. Lett. {\bf 60}, 1057 (1988); Phys. Rev. B {\bf 39}, 2344 (1989).
627: Troyer and M. Imada, Computer Simulations in
628: Condensed Matter Physics X, eds. D. P. Landau {\it et al.}, (Springer
629: Verlag, Heidelberg, 1997).
630:
631: \bibitem{troyer_exponent}
632: M. Troyer, M. Imada
633: and K. Ueda, J. Phys. Soc. Jpn. {\bf 66}, 2957 (1997).
634: M. Troyer and M. Imada, Computer Simulations in
635: Condensed Matter Physics X, eds. D. P. Landau {\it et al.}, (Springer
636: Verlag, Heidelberg, 1997).
637:
638: \bibitem{science} T. Senthil, {\it et al.} Science {\bf 303}, 1490 (2004); and refernces therein.
639:
640: \bibitem{zinn_justin}
641: J. Zinn-Justin, {\it Quantum Field Theory and Critical Phenomena},
642: Oxford University Press, 2002.
643:
644:
645: \bibitem{ALPS} M. Troyer {\it et al.}: Lecture Notes in Computer
646: Science {\bf 1505} (1998) 191; F. Alet {\it et al.}, Report cond-mat/0410407. Sources of this library are available at
647: {\tt http://alps.comp-phys.org/}.
648:
649: \bibitem{ehrenfest}
650: M. Matsumoto and M. Sigrist, Report cond-mat/0401411.
651:
652: \bibitem{GL}
653: I. S. Aranson and L. Kramer, Rev. of Mod. Phys. {\bf 74}, 99-143 (2002).
654:
655: \end{thebibliography}
656:
657: \end{document}
658:
659:
660:
661:
662: