cond-mat0204096/ahe.tex
1: %\documentclass[aps,twocolumn]{revtex4}
2: \documentclass[aps,preprint]{revtex4}
3: %
4: \usepackage{graphicx}% Include figure files
5: \usepackage{dcolumn}% Align table columns on decimal point
6: \usepackage{bm}% bold math
7: %%%%%%%%%%%%%%%%%%%%%%%%%%
8: \begin{document}
9: %%%%%%%%%%%%%%%%%%%
10: %\newcommand{\kv}{\mbox{\boldmath{$k$}}}
11: \newcommand{\Hso}{H_{\rm so}}
12: \newcommand{\Hp}{H'}
13: \newcommand{\sigmav}{\bm{\sigma}}
14: \newcommand{\kv}{{\bf{k}}}
15: \newcommand{\qv}{{\bf{q}}}
16: \newcommand{\Sv}{{\bf{S}}}
17: \newcommand{\Mv}{{\bf{M}}}
18: \newcommand{\Jv}{{\bf{J}}}
19: \newcommand{\Xv}{{\bf{X}}}
20: \newcommand{\xv}{{\bf{x}}}
21: \newcommand{\av}{{\bf{a}}}
22: \newcommand{\bv}{{\bf{b}}}
23: \newcommand{\cv}{{\bf{c}}}
24: \newcommand{\rv}{{\bf{r}}}
25: \newcommand{\nv}{{\bf{n}}}
26: \newcommand{\omegaell}{\omega_{\ell}}
27: \newcommand{\DOS}{\nu}
28: \newcommand{\vimp}{v_{\rm imp}}
29: \newcommand{\nimp}{n_{\rm imp}}
30: \newcommand{\nm}{n_{\rm m}}
31: \newcommand{\azero}{a_0}
32: \newcommand{\chiuni}{\chi_{0}}
33: \newcommand{\chiv}{\bm{C}}
34: %%%%%%%%%%%%%%%%%%%
35: \title{
36: Chirality driven anomalous Hall effect in weak coupling regime
37: } 
38: \author{Gen Tatara}
39: \author{Hikaru Kawamura}
40: \affiliation{
41: Graduate School of Science, Osaka University, Toyonaka, Osaka 560-0043, 
42: Japan
43: }
44: \date{\today}
45: %%%%%%%%%%%%%%%
46: \begin{abstract}
47: Anomalous Hall effect arising from non-trivial spin configuration 
48: (chirality) is studied based on the $s$-$d$ model.
49: Considering a weak coupling case, the interaction is treated perturbatively.
50: Scattering by normal impurities is included. 
51: Chirality is shown to drive locally Hall current and leads to overall
52: Hall effect if there is a finite uniform chirality.
53: This contribution is independent of the conventional 
54: spin-orbit contribution and shows distinct low temperature
55: behavior.
56: In mesoscopic spin glasses, 
57: chirality-induced anomalous Hall effect is  
58: expected below the spin-glass transition temperature.
59: Measurement of Hall coefficient would be useful 
60: in experimentally confirming the chirality ordering. 
61: \end{abstract}
62: \pacs{72.15.-v,75.30.-m,Ê75.50.Lk}
63: %72.15.-vÊÊÊ Electronic conduction in metals and alloys 
64: %75.30.-mÊÊ Intrinsic properties of magnetically ordered materials 
65: %75.50.LkÊÊ Spin glasses and other random magnetsÊ 
66: \maketitle
67: 
68: Hall effect in ferromagnetic metals has long been known to have 
69: anomalous component which does not vanish at zero external magnetic 
70: field.
71: Theories have explained this as due to the spontaneous magnetization 
72: and spin-orbit interaction\cite{Karplus54,Smit55,Luttinger58}. 
73: It was also shown based on the $s$-$d$ model that the high temperature 
74: (close to the critical temperature) behavior of the anomalous Hall 
75: effect is understood in terms of the fluctuation of the 
76: magnetization coupled with spin-orbit 
77: interaction\cite{Kondo62,Maranzana67}.
78: 
79: Recently some manganites were found to exhibit at high temperatures
80: abnormal behavior\cite{Snyder96,Asamitsu98} 
81: which is not explainable by previous theories. 
82: This behavior was explained by Berry phase effect 
83: associated with thermally driven non-trivial 
84: background spin configuration (chirality)\cite{Ye99,Chun00,Geller01}. 
85: Finite chirality results in a finite Berry phase and leads to 
86: Hall effect if chirality is non-vanishing as a net.
87: In contrast to manganites, anomalous Hall coefficient in 
88: ferromagnetic pyrochlores was found to remain finite at low 
89: temperatures\cite{Taguchi99,Katsufuji00}.
90: This behavior was discussed to be due to finite chirality of the 
91: ground state, which is originating from geometrical 
92: frustration\cite{Ohgushi00,Taguchi01,Shindou01,Onoda01}.
93: Behavior which is not explained solely by chirality 
94: theories was reported recently in some Mo-based 
95: pyrochlores\cite{Yoshii00,Yasui01,Kageyama01}.
96: These recent 
97: theories\cite{Ye99,Chun00,Ohgushi00,Taguchi01,Geller01,Shindou01,Onoda01} 
98: have exclusively 
99: dealt with a strong Hund-coupling limit, considering a half-metallic 
100: nature of the experimental systems.
101: In this limit, the electron spin aligns perfectly to the local 
102: spin and feels the same Berry phase as the local spin carries, and 
103: Hall conductivity has a topological meaning\cite{Thouless82}.
104: The weak coupling region, which would be the case of most common 
105: transition-metal magnets, has never been explored from 
106: the viewpoint of chirality.
107: 
108: In this paper, we study anomalous Hall effect due to chirality based 
109: on the $s$-$d$ model in the weak coupling case. 
110: Using Kubo formula and taking account of the impurity scattering, 
111: we will demonstrate that the chirality drives 
112: local Hall current in the perturbative regime.  
113: To compare with experiments, both the chirality mechanism and the 
114: conventional one 
115: due to the spin-orbit interaction (corresponding to the 
116: result by Karplus and Luttinger\cite{Karplus54,Luttinger58}) need to 
117: be taken account. 
118: We will show that the chirality contribution is independent of the 
119: conventional one, and they simply add up at lowest approximation.
120: In order for the chirality contribution to Hall effect to be finite, 
121: there needs to be a net uniform component of the chirality.
122: The possible effect of the spin-orbit interaction 
123: to induce a uniform component of the chirality in the presence of 
124: uniform magnetization, originally 
125: claimed to be present in the strong Hund-coupling limit\cite{Ye99}, 
126: is examined in our weak-coupling scheme. 
127: It turns out that the spin-orbit interaction induces a vector 
128: chirality if there is a uniform magnetization, and that this indeed  
129: results in a Hall effect in the bulk. 
130: This effect, however, would depend much on the band structure.
131: 
132: Our theory is applicable to a wide class of magnetic systems including 
133: canonical spin glasses, in which the 
134: conduction electron is only weakly coupled to the local spin.
135: In spin glasses, chirality order develops at low temperature leading 
136: to the spin-glass transition\cite{Kawamura92}.
137: Meanwhile, the chirality 
138: order there is spatially random  without a uniform component, making 
139: the experimental detection of the chirality-driven anomalous Hall 
140: effect rather difficult due to the inherent cancellation effect.
141: Even in this case, however, 
142: average of the squared chirality remains finite in small (mesoscopic)
143: samples\cite{Weissman93,Neuttiens00}, and 
144: enhancement of the anomalous Hall coefficient is expected below 
145: the spin-glass transition.
146: Thus, measurement of fluctuation of Hall conductivity in mesoscopic 
147: samples may be useful 
148: in experimental confirmation of the chirality ordering. 
149: Depending on the band structure, the chirality-driven Hall effect 
150: might be observable even in bulk spin-glass samples if the sample possesses a 
151: uniform magnetization, which is induced by applied fields or is 
152: generated spontaneously (as in case of reentrant spin 
153: glasses\cite{Mydosh93}). 
154: 
155: 
156: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
157: We consider electron on lattice whose Hamiltonian is given by 
158: \begin{equation}
159:     H=\sum_{\kv\sigma}\epsilon_{\kv\sigma}c^\dagger_{\kv\sigma}c_{\kv\sigma}
160:     +\Hp+H_{\rm imp},
161:     \label{eq:Hdef}
162: \end{equation}
163: where $\sigma=\pm$ denotes electron spin. 
164: Electron energy is 
165: $\epsilon_{\kv\sigma}=\kv^2/2m-\sigma\Delta-\epsilon_{F}$, where 
166: $m$ is electron  mass, 
167: $\Delta$ is a uniform polarization of conduction electron 
168:  due to magnetization or external field, and
169: $\epsilon_{F}$ is the Fermi energy. 
170: The interaction with localized spin $\Sv_\Xv$ (treated as classical) is represented 
171: by the exchange interaction $\Hp$,
172: \begin{equation}
173:     \Hp=\frac{J}{N}\sum_{\kv\kv'} \Sv_{\kv'-\kv}  (c^\dagger_{\kv'}\sigmav c_{\kv}),
174:     \label{eq:Hp}
175: \end{equation}
176: where $\sigma^\alpha$ ($\alpha=x,y,z$) are Pauli matrices, and $N$ is 
177: the total number of lattice sites.
178: The sign of the exchange coupling $J$ depends on the system we consider; it is 
179: positive 
180: if the interaction is the $s$-$d$ exchange as in case of canonical 
181: spin glasses, and is negative if it is the Hund-coupling as in case of 
182: manganites.
183: Configuration of $\Sv_{\Xv}=(1/N)\sum_{\qv}e^{i\qv \Xv} \Sv_{\qv}$ is 
184: fixed. 
185: We note that localized spins $\Sv_{\Xv}$ do not necessarily occupy all $N$ 
186: sites here: An example is a dilute magnetic alloy such as canonical spin 
187: glass\cite{Mydosh93}.
188: $\Hp$ is assumed not to contain uniform component $\Mv\equiv \Sv_{q=0}$, 
189: since it is taken into account in $\Delta=JM$ with $M=|\Mv|$. 
190: The scattering by normal impurities is represented by 
191: $H_{\rm imp}=({\vimp}/{N})\sum \sum_{i}^{\nimp} e^{i(\kv-\kv')X_{i}}
192: c^\dagger_{\kv'}c_{\kv}$, where $\nimp$ is the number of nonmagnetic 
193: impurities (at sites $X_{i}$) and $\vimp$ is a constant.
194: 
195: The electronic current is given by 
196: $\Jv=\frac{e}{m}\sum_{\kv\sigma}\kv c^\dagger_{\kv'\sigma} c_{\kv\sigma}$.
197: Based on Kubo formula, the anomalous Hall conductivity is obtained as
198: $
199: \sigma_{xy}=\lim_{\omega\rightarrow0}\frac{1}{\omega}{\rm Im}
200: \left( Q_{xy}(\omega+i0)-Q_{xy}(i0) \right)
201: $
202: with 
203: $    Q_{xy}(i\omegaell)\equiv\frac{1}{\beta V} 
204:     <J_{x}(i\omegaell)J_{y}(-i\omegaell)>
205: $, 
206: where the bracket $<>$ denotes averaging over electrons and 
207: impurities, 
208: $\omega_{\ell}\equiv 2\pi\ell/\beta$ being the Matsubara frequency.
209: We treat $\Hp$ perturbatively.
210: 
211: As is obvious, the first and second order contribution vanish since the spatial  
212: asymmetry due to
213: current vertices $J_{x}$ and $J_{y}$ in the correlation function cannot be 
214: deleted. 
215: The first term which can possibly be finite is the third-order term 
216: (Fig. \ref{FIGsigxy});
217: \begin{eqnarray}
218:   \sigma_{xy}^{(3)} &=& \frac{1}{2\pi V} 
219:   \left(\frac{e}{m}\right)^2 \left(\frac{J}{N}\right)^3 
220:   \sum_{\kv\kv'\kv''} 
221:   \sum_{\alpha\beta\gamma} k_{x} k'_{y} 
222:   S^{\alpha}_{\kv''-\kv} S^{\beta}_{\kv'-\kv''} S^{\gamma}_{\kv-\kv'}  
223:   {\rm tr}[
224:   G_{\kv}^R \sigma^\alpha G_{\kv''}^R \sigma^\beta
225:   G_{\kv'}^R G_{\kv'}^A \sigma^\gamma G_{\kv}^A ]
226:    \nonumber\\  &&   
227:    +{\rm c.c.},
228:   \label{eq:Q03def}
229: \end{eqnarray}
230: where $V$ is the total volume, trace is over spin indices, 
231: and $\alpha,\beta,\gamma$ runs over $x,y,z$.
232: $G_{\kv\sigma}^R(\equiv [{\frac{i}{2\tau}-\epsilon_{\kv\sigma}}]^{-1})$ and 
233: $G_{\kv\sigma}^A(=(G_{\kv\sigma}^R)^*)$ are retarded and advanced Green 
234: functions in the $\omega\rightarrow0$ limit, 
235: respectively, which include lifetime due to impurities, $\tau\equiv 
236: 2\pi\DOS\nimp\vimp^2$, $\DOS$ being the density of states per site. 
237: We consider the case where the polarization of conduction electron is small 
238: ($\Delta\sim 0$).
239: The summation over spin indices in eq. (\ref{eq:Q03def}) is then  
240: carried out as 
241: ${\rm tr}[\sigma^\alpha \sigma^\beta \sigma^\gamma]= 
242: 2i\epsilon^{\alpha\beta\gamma}$, where $\epsilon^{\alpha\beta\gamma}$ 
243: is the totally antisymmetric tensor. 
244: By use of the partial derivative, the Hall conductivity reduces to a 
245: compact form;
246: \begin{equation}
247:   \sigma_{xy}^{(3)}   = 
248:   \frac{N}{\pi V}
249:   \left(\frac{e}{m}\right)^2 (2\pi\DOS J)^3\tau^2 \chiuni
250:    =  (4\pi)^2 \sigma_{0} J^3 \DOS^2 \tau {\chiuni}, 
251:     \label{eq:Q0331}
252: \end{equation}
253: where $\sigma_{0}$ is the Boltzmann conductivity, 
254: $\sigma_{0}\equiv \frac{N}{2V}\left(\frac{e}{m}\right)^2\DOS 
255: k_{F}^2\tau$ ($k_{F}$ is the Fermi wavenumber).
256: We see that $\sigma_{xy}^{(3)}\propto \tau^2 \propto 
257: \rho_{0}^{-2}$ ($\rho_{0}=\sigma_{0}^{-1}$ is the resistivity).
258: The uniform chirality $\chiuni$ is given by 
259: \begin{eqnarray}
260:   \chiuni &\equiv&  \frac{1}{N}\sum_{\Xv_{i}}
261:     \Sv_{\Xv_{1}}\cdot(\Sv_{\Xv_{2}}\times\Sv_{\Xv_{3}})  
262:     \nonumber  \\
263:      &  & \times 
264:      \left[ 
265:      \frac{(\av\times\bv)_{z}}{ab} I'(a) I'(b) I(c) 
266:      +\frac{(\bv\times\cv)_{z}}{bc} I(a) I'(b) I'(c)
267:      +\frac{(\cv\times\av)_{z}}{ca} I'(a) I(b) I'(c)
268:      \right] , \label{chiunidef}
269: \end{eqnarray}
270: where $\Xv_{i}$ runs over all the positions of local spins, 
271: while $\av\equiv\Xv_{1}-\Xv_{2}$, 
272: $\bv\equiv\Xv_{2}-\Xv_{3}$ and $\cv\equiv \Xv_{3}-\Xv_{1}$ are the 
273: vectors representing sides of the triangle 
274: ($a\equiv |\av|$ e.t.c.).
275: $I(r)\equiv 
276: \frac{1}{2\pi N\DOS\tau} \sum_{\kv}e^{i\kv\cdot\rv} G_{\kv}^R 
277: G_{\kv}^A$ and 
278: $I'(r)=\frac{d I(r)}{dr}$.
279: It is seen that Hall current is driven by three spins which form 
280: a finite solid angle in spin space (i.e., finite local chirality 
281: $\chi_{123}\equiv
282: \Sv_{\Xv_{1}} \cdot ( \Sv_{\Xv_{2}} \times \Sv_{\Xv_{3}}) $)
283: spanning a finite area in 
284: coordinate space (as seen from $(\av\times\bv)_{z}$ etc.).
285: Note that the factor in the square bracket in the definition of 
286: $\chiuni$ specifies the coupling between the spin- and 
287: the coordinate-space. 
288: In the case three spins ($1,2$ and $3$) align right-handed in 
289: spin space 
290: ($\chi_{123}>0$),  $\chi_{123}$ contributes positively to $\chiuni$ 
291: if these three spins are 
292: located anti-clockwise in real space, and is negative if they are 
293: located clockwise.
294: In the case three spins align left-handed in spin space ($\chi_{123}<0$), this 
295: assignment is reversed.
296: Noting $I(r)=\frac{\sin k_{F}r}{k_{F} r}e^{-r/2\ell}$, 
297: where $\ell$ is elastic mean free path,
298: contribution from largely separated three spins
299: with the scale of $r$ decays rapidly as 
300: $\sim e^{-3r/2\ell}/(k_{F}r)^3$, and the Hall 
301: effect is dominantly driven by chiralities of spins on small triangles.
302: Note that the large-$r$ behavior of the weight function has 
303: resemblance to the RKKY interaction.
304: The expression of the uniform chirality 
305: derived in our weak coupling scheme, eq. (\ref{chiunidef}), 
306: contains contribution from large triangles, and is a natural 
307: extension of the conventional (and naive) definition of the 
308: chirality in terms of spins on adjacent sites only. 
309: Eqs. (\ref{eq:Q0331}) and (\ref{chiunidef}) are main results of 
310: the present paper, which gives a direct relation between the Hall 
311: conductivity and the spin configuration.
312: 
313: Conventional theories of anomalous Hall effect is based on the 
314: spin-orbit interaction, 
315: $H_{\rm so}\equiv
316: i\lambda\sum_{\kv\kv'}(\kv'\times\kv)\cdot 
317:     (c^\dagger_{\kv'}\sigmav c_{\kv})
318: $, where $\lambda$ is the spin-orbit coupling 
319: constant\cite{Karplus54,Smit55,Luttinger58}. 
320: Thus we have also analyzed the contribution of $H_{\rm so}$ on the same 
321: footing as that of $\Hp$ performing a double power series expansion.
322: At the lowest (first) order, the contribution of $H_{\rm so}$ 
323: is made up of so-called skew scattering and side-jump
324: ones\cite{Berger70}, which are calculated as 
325: $\sigma_{xy}^{\rm so}=-\lambda M (A'\tau+B)$\cite{Dugaev01}. 
326: Here $A'$ and $B$ are constants independent of $\tau$, each term 
327: corresponding to skew scattering and side-jump processes, respectively.
328: We note that $A'$ and $B$ are positive in the present single band 
329: approximation, but their signs actually 
330: depend on the band structure in real materials. 
331: At the lowest order, spin-orbit ($\sigma_{xy}^{\rm so}$) and chirality 
332: (eq. (\ref{eq:Q0331})) contributions are 
333: independent, and the total Hall conductivity is simply their sum.
334: These can mix as higher order corrections, but we neglect such small 
335: contributions.
336: The total Hall resistivity,  
337: $\rho_{xy}=\sigma_{xy}\rho_0^2$, 
338: then behaves as
339: \begin{equation}
340: \rho_{xy}\simeq -\lambda M(A \rho_{0}+B \rho_{0}^2) +C J^3 \chiuni,
341: \label{rhoxy}
342: \end{equation}
343: where $A=A'\rho_{0}\tau$ is independent of $\tau$, while
344: $C=\frac{1}{e^2}\frac{2V}{N}(\frac{m}{k_{F}})^2>0$ in our single band 
345: approximation. 
346: The sign of chirality contribution depends 
347: on whether the coupling is of the $s$-$d$ type ($J>0$) or of 
348: the Hund type ($J<0$). 
349: It is seen that the three terms in eq. (\ref{rhoxy}) 
350: depend differently on the impurity 
351: concentration. The chirality contribution is dominant in the 
352: clean regime and at low temperatures.
353: It should be noted that the analysis in the strong Hund-coupling 
354: case which does not consider impurities yields  
355: $\rho_{xy}\propto \rho_{0}^2 \chi$\cite{Ohgushi00}.
356: These different dependences on $\rho_{0}$, which indicate different 
357: behavior as a function of temperature, would be useful in interpreting 
358: the experimental results.
359: 
360: The chirality contribution to Hall coefficient is finite only if 
361: there is a net uniform chirality, $\chiuni\neq0$.
362: Finite net chirality, however, may not be very easy to realize on regular 
363: lattices with simple nearest-neighbor exchange interaction, 
364: since the chirality on adjacent plaquettes usually tends to cancel each 
365: other due to symmetry\cite{Ohgushi00,Shindou02}.
366: One possible mechanism to realize a finite net chirality has been 
367: proposed in Ref. \cite{Ye99}, where it was argued in the strong 
368: coupling case that 
369: the spin-orbit interaction induced a net chirality in the 
370: presence of magnetization as
371: $\chiuni^{\rm so}=-\alpha\lambda M$ ($\alpha$ is a positive 
372: constant).
373: Inspired by this observation, 
374: we have examined whether such a mechanism works in the present weak 
375: coupling case.  
376: To examine the possible coupling between the uniform chirality and the 
377: magnetization $M$, we look into the expectation value (effective 
378: Hamiltonian) of the spin-orbit 
379: interaction, $<H_{\rm so}>$, where $<>$ denotes the thermal averaging over 
380: electrons,  treating $\Hp$ as perturbation.
381: We identify two types of terms as possible candidates. One is the 
382: term linear in $M$ for small $M$ and comes from the third-order 
383: contribution in $\Hp$,
384: \begin{eqnarray}
385:     H_{\rm so}^{(3)\chi} &=&  2\lambda\left(\frac{J}{N}\right)^3 
386:      \sum_{\xv\Xv_{i}} 
387:     \Sv_{\Xv_{1}} \cdot ( \Sv_{\Xv_{2}} \times \Sv_{\Xv_{3}}) 
388:     [(\Xv_{1}-\xv)\times (\Xv_{3}-\xv)]_{z} 
389:     \nonumber\\
390:     && \times \frac{1}{\beta}\sum_{\omega_{n}}
391:      g_{\omega_{n}}'(|\xv-\Xv_{1}|)
392:  g_{\omega_{n}}(a)  g_{\omega_{n}}(b) \Delta'_{\omega_{n}}(|\Xv_{3}-\xv|),
393:      \label{eq:Hsoeff2}
394: \end{eqnarray}
395: where $\xv$ denotes the site at which the spin-orbit interaction acts 
396: ($a$ and $b$ are defined after eq. (\ref{chiunidef})), and the uniform 
397: magnetization is assumed to be in $z$-direction.
398: The thermal Green functions are defined here as  
399: $g_{\omega_{n}}(r)\equiv \frac{1}{2}(G_{\omega_{n} +}+G_{\omega_{n} -})$, 
400: $\Delta_{\omega_{n}}(r)\equiv (G_{\omega_{n} +}-G_{\omega_{n}-})$, where 
401: $G_{\omega_{n}\sigma}(\Xv)=\sum_{\kv}{e^{-i\kv \Xv}} 
402: [{i(\omega_{n}+\frac{{\rm 
403: sgn}(\omega_{n})}{2\tau})-\epsilon_{\kv}+\Delta\sigma}]^{-1}$, and 
404: $G'(r)\equiv\frac{dG}{dr}$.
405: Note that 
406: $\Delta_{\omega_{n}}$ is proportional to $\Delta$, and hence to $M$,  for 
407: $\Delta/\epsilon_{F}\ll1$.
408: It is seen that 
409: $H_{\rm so}^{(3)\chi}$ apparently describes the coupling between the uniform 
410: magnetization and the local 
411: chirality, but without referring to the spatial spin configuration.
412: In fact, since the factor of $((\Xv_{1}-\xv)\times (\Xv_{3}-\xv))_{z}$ 
413: specifies only the angles between 
414: the two spins $S_{\Xv_{1}}$ and $S_{\Xv_{3}}$ when looked from the  
415: position $\xv$ irrespective of spatial configuration 
416: of $S_{\Xv_{2}}$, $H_{\rm so}^{(3)\chi}$ does not contain the 
417: component inducing the uniform chirality $\chiuni$.
418: 
419: The other term is quadratic in $M$ for small $M$ and comes from the 
420: second-order contribution in $\Hp$, which reads as
421: \begin{eqnarray}
422:     H_{\rm so}^{(2)\chi} &=& 
423:     -2\lambda\left(\frac{J}{N}\right)^2 
424:     \sum_{\xv\Xv_{i}} (\Sv_{\Xv_{1}}\times \Sv_{\Xv_{2}})_{z} 
425:     \frac{[(\Xv_{1}-\xv)\times(\Xv_{2}-\xv)]_{z}}
426:        {|\Xv_{1}-\xv| |\Xv_{2}-\xv|}  \nonumber\\
427:        && \times
428:    \frac{1}{\beta}\sum_{\omega_{n}} 
429:     \Delta_{\omega_{n}}'(|\Xv_{1}-\xv|) g_{\omega_{n}}(|\Xv_{1}-\Xv_{2}|) 
430:     \Delta_{\omega_{n}}'(|\Xv_{2}-\xv|).
431:     \label{hso2}
432: \end{eqnarray}
433: In the presence of uniform magnetization $M$, 
434: $z$-component of the vector chirality, defined 
435: by two spins as $\chiv_{12}\equiv(\Sv_{\Xv_{1}}\times\Sv_{\Xv_{2}})$, 
436: plays essentially the same role as the scalar chirality 
437: ($\chi_{123}\simeq M\chiv_{12}^{z}$).   
438: Then, eq. (\ref{hso2}) above describes the coupling between the vector 
439: chirality and the magnetization via the spin-orbit interaction.
440: After summation over the thermal frequency, 
441: the Fourier transform of the electron part becomes
442: \begin{equation}
443:     \frac{1}{\beta}\sum_{\omega_{n}} 
444:     \Delta_{\omega_{n}\kv}g_{\omega_{n}\kv'}\Delta_{\omega_{n}\kv''}=
445:     \frac{1}{8}\sum_{\sigma\sigma'\sigma''}
446:     \frac{\sigma\sigma''}{\epsilon_{\kv\sigma}-\epsilon_{\kv'\sigma'}}
447:     \left[ 
448:  \frac{f_{\kv\sigma}-f_{\kv''\sigma''}}
449:  {\epsilon_{\kv\sigma}-\epsilon_{\kv''\sigma''}}
450: - \frac{f_{\kv'\sigma'}-f_{\kv''\sigma''}}
451:  {\epsilon_{\kv'\sigma'}-\epsilon_{\kv''\sigma''}}
452: \right],
453:  \label{eq:sumzero}
454: \end{equation}
455: where $f_{\kv\sigma}$ is the Fermi distribution function.
456: If there is a particle-hole symmetry (invariance under
457: $\epsilon\rightarrow-\epsilon$), which is equivalent to assuming that 
458: only excitations close to the Fermi level dominates, 
459: eq. (\ref{eq:sumzero}) after summation over $\kv$'s turns out to vanish.
460: This symmetry, however, is not necessarily observed in real material,  
461: for instance due to 
462: the existence of the bottom of the band, and then, a finite contribution 
463: is expected to remain in general. 
464: Thus, this term could serve as a ``symmetry-breaking field'' 
465: ($H_{\rm so}^{(2)\chi}\propto M^2\sum_{\Xv_{i}}\chiv_{12}^{z} 
466: \propto M \chiuni$) inducing a uniform chirality, which results 
467: in a Hall effect in the bulk. 
468: 
469: %%%%%%%%%%%%%%%%%%%%%%%%%
470: Even in the case the chirality does not contain uniform component,  
471: $\chiuni$ could still 
472: be finite if the system size is sufficiently small. 
473: Let us consider the case of spin glasses in zero external field, 
474: in which the chirality is randomly ordered. 
475: The number of triangles which contribute to $\sigma_{xy}$ 
476: is given roughly as 
477: $N_{\chi}\simeq N \nm^3 (\ell/\azero)^4$  ($\nm$ is the concentration 
478: of localized spins, and $\azero$ is lattice constant). 
479: Hence, the sum of random chirality is
480: $\chiuni\propto\frac{1}{\sqrt{N}}\nm^{3/2} (\ell/\azero)^2$. 
481: Although this quantity decays as $\propto 1/\sqrt{N}$, 
482: we expect that detection of Hall effect 
483: may be possible by high-sensitivity-measurements on mesoscopic samples. 
484: This chirality-driven Hall effect of random sign in mesoscopic spin glasses
485: would be measurable\cite{Weissman93,Neuttiens00}
486: by looking at the sample-dependent or 
487:  thermal-cycle-dependent fluctuations, 
488: $\delta\sigma_{xy}\equiv \sigma_{xy}-[{\sigma_{xy}}]_{s}$ 
489: ($[\;]_{s}$ denotes average over spin configurations), 
490: whose squared average is given as
491: $\sqrt{[{ (\delta \sigma_{xy})^2}]_{s}} /\sigma_{0}
492:  = (4\pi)^2 J^3 \DOS^2 \tau
493:   \sqrt{ [{\chi^2}]_{s} } $, 
494: where
495:   $\sqrt{[{\chi^2}]_{s}}\equiv 
496: { \frac{1}{N} \{ \sum_{ijk} 
497: (\chi_{ijk} F_{ijk})^2 }\}^{1/2}
498: \propto\frac{1}{\sqrt{N}}\nm^3 (\ell/\azero)^4$ 
499: ($F_{ijk}$ represents the spatial weight given in the square bracket 
500: in eq. (\ref{chiunidef})).
501: 
502: 
503: %%%%%%%%%%%%%
504: 
505: In the 70's, anomalous Hall effect in spin glasses was experimentally 
506: investigated\cite{Mydosh93,McAlister76}. 
507: It was found that the Hall resistivity 
508: of canonical spin glasses, i.e., dilute magnetic alloys 
509: such as AuFe and AgMn, measured in weak
510: applied fields were negative and exhibited a cusp-like anomaly 
511: around the spin-glass transition 
512: temperature whose behavior was quite similar to that of the 
513: magnetic susceptibility, $\chi_{m}$; i.e., 
514: $\rho_{xy}/H\propto-\chi_{m}$\cite{McAlister76}.
515: Although this behavior can be explained by  
516: the standard spin-orbit contribution  (the first two terms in eq. 
517: (\ref{rhoxy})),  it could also be explainable 
518: by the contribution of a uniform chirality 
519: induced by $H_{\rm so}^{(2)\chi}$ described above.
520: In order to experimentally resolve the spin-orbit contribution of
521: Karplus-Luttinger type  from the spin-orbit induced chirality 
522: contribution, one might possibly examine the dependence on 
523: $\rho_{0}$ (Eq. (\ref{rhoxy})), or measure  
524: the response to external magnetic fields applied  
525: in various directions.
526: In this connection, reentrant spin-glass systems\cite{Mydosh93}, which exhibit 
527: successive phase transitions, first from para to ferro 
528: and then from ferro to spin glass at lower temperature, would be of 
529: much interest. In the ferromagnetic regime, only 
530: the conventional spin-orbit mechanism is expected to work, while in 
531: reentrant spin-glass regime, the chirality contribution 
532: sets in due to the spin canting giving rise to distinct contribution 
533: at lower temperatures. 
534: 
535: 
536: %%%%%%%%%%%%%%%%%%%%%%%%%%
537: 
538: To summarize, we have demonstrated based on the $s$-$d$ model in the weak 
539: coupling 
540: regime that a topologically nontrivial spin configuration (chirality) induces 
541: Hall current.
542: The chirality-driven Hall effect as a bulk appears if uniform component 
543: of the chirality is finite.
544: This contribution is independent of the conventional 
545: contribution from the spin-orbit 
546: coupling exhibiting different dependence on resistivity from the 
547: spin-orbit contribution and other predictions based on chirality 
548: mechanism.
549: If chirality is ordered randomly, as in spin glasses below the spin-glass 
550: transition temperature,
551: sample-to-sample or thermal-cycle-dependent fluctuations of Hall conductivity
552: in mesoscopic samples is expected to show an anomalous enhancement.
553: Without a direct method of magnetic detection of the chirality 
554: available so far, 
555: measurement of (fluctuation of) Hall conductivity would be powerful tool  
556: in experimental confirmation of the chirality ordering. 
557: 
558: %%%%%%%%%%%%%%%%%%%
559: %\acknowledgements
560: \begin{acknowledgements}
561: The authors are grateful to H. Kohno, R. Shindou, H. Yoshino, S. 
562: Yoshii, N. Nagaosa and M. Sato
563: for useful comments and discussion.
564: This work is supported by Priority Areas Grants from the 
565: Ministry of Education, Culture, Sports, Science and Technology, Japan.
566: G. T. thanks The Mitsubishi Foundation for financial support.
567: \end{acknowledgements}
568: 
569: 
570: 
571: %%%%%%%%%%%%%%%%%%%%
572: \begin{thebibliography}{99}
573:     \bibitem{Karplus54} R. Karplus and J. M. Luttinger, Phys. Rev.  
574:          {\bf 95}, 1154 (1954).
575:     \bibitem{Smit55} J. Smit, Physica {\bf 21}, 877 (1955);  
576:           {\bf 24}, 39 (1958). 
577:     \bibitem{Luttinger58} J. M. Luttinger, Phys. Rev.  
578:          {\bf 112}, 739 (1958).
579:     \bibitem{Kondo62} J. Kondo, Prog. Theor. Phys. {\bf 27}, 772 (1962).
580:     \bibitem{Maranzana67} F. E. Maranzana, Phys. Rev. {\bf 160}, 421 
581:     (1967).
582:     \bibitem{Snyder96}
583:     G. J. Snyder {\it et al}, Appl. Phys. Lett. {\bf 69}, 4254 (1996).
584:      \bibitem{Asamitsu98}
585:      A. Asamitsu and Y. Tokura, Phys. Rev. {\bf B58}, 47 (1998).
586:      \bibitem{Ye99} J. Ye {\it et al}, Phys. Rev. Lett. {\bf 83}, 3737 (1999). 
587:      \bibitem{Chun00} S. H. Chun {\it et al}, Phys. Rev. Lett. {\bf 84}, 757 
588:        (2000). 
589:      \bibitem{Geller01} Y. Lyanda-Geller {\it et al}, 
590:      Phys. Rev. {\bf B63}, 184426 (2001). 
591:     \bibitem{Taguchi99} Y. Taguchi and Y. Tokura, Phys. Rev. {\bf 
592:     B60}, 10280 (1999).
593:     \bibitem{Katsufuji00} T. Katsufuji, H. Y. Hwang and S. -W. Cheong, 
594:     Phys. Rev. Lett. {\bf 84}, 1998 (2000).
595:     \bibitem{Ohgushi00} K. Ohgushi, S. Murakami and N. Nagaosa, Phys. 
596:     Rev. {\bf B62}, R6065 (2000).
597:     \bibitem{Taguchi01} Y. Taguchi
598:     %, Y. Oohara, H. Yoshizawa, N. Nagaosa and Y. Tokura,
599:     {\it et al}, Science {\bf 291}, 2573 (2001).
600:     \bibitem{Shindou01} R. Shindou and N. Nagaosa, Phys. Rev. Lett. 
601:     {\bf 87}, 116801 (2001).
602:     \bibitem{Onoda01} M. Onoda and N. Nagaosa, J. Phys. Sco. Jpn. {\bf 
603:     71} 19 (2002).
604:     \bibitem{Yoshii00} S. Yoshii
605:     %, S. Iikubo, T. Kageyama, K. Oda, Y. 
606:     %Kondo, K. Murata and M. Sato, 
607:     {\it et al}, J. Phys. Soc. Jpn. {\bf 69} 3777 (2000). 
608:     \bibitem{Yasui01} Y. Yasui
609:     %, Y. Kondo, M. Kanada, M. Ito, H. 
610:     %Hirashina, M. Sato and K. Kakurai,
611:     {\it et al}, J. Phys. Soc. Jpn. {\bf 70}, 284 (2001).
612:     \bibitem{Kageyama01} T. Kageyama
613:     %, S. Iikubo, S. Yoshii, Y. Kondo, 
614:     %M. Sato and Y. Iye, 
615:     {\it et al}, J. Phys. Soc. Jpn. {\bf 70}, 3006 (2001).
616:     \bibitem{Thouless82} D. J. Thouless
617:     %, M. Kohmoto, M. P. Nightingale and M. den Nijs, 
618:     {\it et al}, Phys. Rev. Lett. {\bf 49}, 405 (1982).
619:     \bibitem{Kawamura92} H. Kawamura, Phys. Rev. Lett {\bf 68}, 3785 
620:     (1992); Phys. Rev. Lett {\bf 80}, 5421 (1998).
621:  \bibitem{Weissman93}  M. B. Weissman, Rev. Mod. Phys. {\bf 65}, 829 (1993).
622:     \bibitem{Neuttiens00}  G. Neuttiens
623:     %, C. Strunk, C. Van Haesendononck and Y. Bruynseraede, 
624:     {\it et al}, Phys. Rev. {\bf B62}, 3905 (2000).
625: \bibitem{Mydosh93} J. A. Mydosh, ``Spin glasses: an experimental 
626: introduction'', Tayer \& Francis (1993).
627:     \bibitem{Berger70} L. Berger, Phys. Rev. {\bf B2}, 4559 (1970).
628:     \bibitem{Dugaev01} V. K. Dugaev, A. Crepieux and P. Bruno, 
629:      Phys. Rev. {\bf B64}, 104411 (2001).
630:     \bibitem{Shindou02} R. Shindou, private communication.
631:      \bibitem{McAlister76} S. P. McAlister and C. M. Hurd, Phys. Rev. 
632:      Lett. {\bf 37}, 1017 (1976).
633: \end{thebibliography}
634: 
635: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
636: \begin{figure}[bthp]
637: \includegraphics{sigmaxy.eps} 
638: %%
639: \begin{minipage}[t]{70mm}
640: \caption{Three spin contribution to $\sigma_{xy}$. The interaction 
641: with the local spin, $\Sv$, is denoted by a shaded small circles. 
642: The two processes are complex congugate to each other. 
643: Other contributions vanish due to symmetry. 
644: \label{FIGsigxy}}
645: \end{minipage}
646: %
647: %
648: \end{figure}
649: %
650: \end{document}