cond-mat0207486/h2.tex
1: %{\mbox{\boldmath ${\bf \sigma}$}}
2: %\documentstyle[aps,multicol,epsfig]{revtex}
3: \documentstyle[aps,multicol,epsfig]{revtex}
4: %\documentstyle[aps,prl,preprint]{revtex}
5: \begin{document}
6: \draft
7: \title{Evolution of N{\' e}el order and localized spin moment
8: in the doped two-dimensional Hubbard model} 
9: \author{Takao Morinari}
10: \address{Yukawa Institute for Theoretical Physics,
11: Kyoto University, Kyoto 606-8502, Japan}
12: %\address{Department of Physics, Yale University, P.O. Box 208120,
13: %New Haven, CT 06520-8120}
14: \date{\today}
15: \maketitle
16: \begin{abstract}
17: We investigate effects of doped holes' hopping on N{\' e}el order
18: in the two-dimensional Hubbard model.
19: Semiclassical staggered moments are computed
20: by solving saddle point equations derived from a path-integral
21: formalism.
22: Effects of quantum fluctuations are taken into account 
23: by the Schwinger boson mean field theory.
24: We argue that hopping of doped holes is ineffective
25: in suppressing N{\' e}el order
26: compared to rapid supprestion of N{\' e}el order 
27: in high-temperature superconductors.
28: After destruction of N{\' e}el order, the quantum disordered phase
29: sets in.
30: Taking the strong coupling limit in the quantum 
31: disordered phase leads to a model of spinless fermions and 
32: bosons but no gauge field interaction.
33: \end{abstract}
34: 
35: \pacs{}
36: 
37: \begin{multicols}{2}
38: \narrowtext
39: %
40: % Introduction
41: %
42: \section{Introduction}
43: In high-temperature superconductors, one remarkable feature
44: is rapid destruction of N{\' e}el order by hole doping.
45: In fact, only 2\% doping hole concentration
46: is enought to suppress N{\' e}el order.
47: While critical disorder is 50\% for the bond percolation threshold
48: and 41\% for the site percolation threshold.
49: How do doped holes suppress N{\' e}el order in such an effective
50: way?
51: 
52: Naive expectation is that the hopping process of doped holes 
53: suppresses N{\' e}el oder.
54: In this paper, we examine the effect of disorder brought by
55: hopping of doped holes.
56: As a model, we  take the Hubbard model
57: because at half-filling it is reduced to the 
58: $S=1/2$ antiferromagnetic Heisenberg model
59: that describes the undoped parent compound 
60: of high-temperature superconductors,
61: and perhaps it is the simplest model to see disorder
62: effect by doped holes' hopping on N{\' e}el order.
63: We use path-integral formalism developed by Schulz.
64: \cite{SCHULZ,WTL}
65: We solve saddle point equations to compute
66: the magnitude of the staggered moment.
67: Since saddle point equations are semiclassical equations,
68: solutions do not contation effects of quatum fluctuations.
69: We take into account quantum fluctuation effects 
70: in terms of Schwinger boson mean field theory.\cite{AROVAS_AUERBACH}
71: 
72: 
73: The remainder of this paper is organized as follows:
74: In sec.\ref{sec_sf}, we rewrite the Hubbard model
75: following Schulz.\cite{SCHULZ,WTL}
76: In sec.\ref{sec_Neel}, we dervie saddle point equations.
77: Solving these equations,
78: we compute doping dependence of the semiclassical staggered moment.
79: We examine effects of quantum fluctuations on N{\' e}el order
80: by the Schwinger boson mean field theory.
81: We show that there is the quantum disordered regime 
82: in which quantum fluctuations suppress N{\' e}el order.
83: In sec.\ref{sec_qd}, we discuss the effective action
84: in the quantum disordered phase.
85: Sec. \ref{sec_dis} is devoted to summary and discussion.
86: 
87: 
88: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
89: %
90: % Hubbard model
91: %
92: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
93: 
94: \section{Path Integral formulation of the Hubbard model}
95: \label{sec_sf}
96: In order to investigate doped holes' hopping effects
97: on N{\' e}el order,
98: we first rewrite the model in a convenient form 
99: following Schulz. \cite{SCHULZ,WTL}
100: In the coherent state path-integral formulation, the partition function 
101: of the model is given by
102: ${\cal Z}=\int {\cal D}\overline{c} {\cal D} c \exp (-S)$ with 
103: $S=S_0 + S_U$,
104: where
105: \begin{equation}
106: S_0 =\int_0^{\beta} d\tau \left[ 
107: \sum_j \overline{c}_j \left( \partial_{\tau} - \mu \right) c_j
108: -t\sum_{\langle i,j \rangle} \left( \overline{c}_i c_j + 
109: \overline{c}_j c_i
110: \right) 
111: \right],
112: \label{eq_S0}
113: \end{equation}
114: \begin{equation}
115: S_U = \int_0^{\beta} d\tau  
116: \sum_j U n_{j\uparrow} n_{j\downarrow}.
117: \end{equation}
118: Hereafter $\tau$ dependence of fields is implicit.
119: The summation $\sum_{\langle i,j \rangle}$ is taken over the
120: nearest neighbor sites.
121: Carrier fields are represented in a spinor:
122: $c_i = ^T \left( \begin{array}{cc}
123: c_{i\uparrow} & c_{i\downarrow} \end{array} \right)$ 
124: and 
125: $\overline{c}_i = \left( \begin{array}{cc}
126: \overline{c}_{i\uparrow} & \overline{c}_{i\downarrow}\end{array}
127: \right)$.
128: Using the identity,\cite{MORIYA} the on-site Coulomb interaction term 
129: can be rewritten as,
130: $U\sum_j n_{j\uparrow} n_{j\downarrow} = - (U/4) \sum_j
131: \left[ (n_{j\uparrow} + n_{j\downarrow} )^2
132: + \left( \overline{c}_j 
133: {\mbox{\boldmath ${\bf \sigma}$}} c_j \right)^2 
134: + ( n_{j\uparrow} + n_{j\downarrow} ) \right]$,
135: where the components of the vector 
136: ${\mbox{\boldmath ${\bf \sigma}$}} = (\sigma_x,\sigma_y,\sigma_z)$ 
137: are the Pauli spin matrices.
138: Introducing Hubbard-Stratonovich fields for the charge and spin
139: fluctuations,\cite{SCHULZ,WTL}
140: we obtain ${\cal Z}=\int {\cal D}\overline{c}
141: {\cal D} c {\cal D} \tilde{\bf S} {\cal D}\phi_{c} \exp (-S_0-S_U)$, 
142: where,
143: \begin{eqnarray}
144: S_U &=&
145: \int_0^{\beta} d\tau \left[
146:  U \sum_j \tilde{\bf S}_j^2 
147: - U \sum_j 
148: \tilde{\bf S}_j \cdot
149: \overline{c}_j {\mbox{\boldmath ${\bf \sigma}$}} c_j
150: \right. \nonumber \\ & & \left. 
151: + \frac{U}{4} \sum_j \phi_{cj}^2 
152: - \frac{U}{2} \sum_j \phi_{cj} \overline{c}_j c_j  \right],
153: \end{eqnarray}
154: up to constant.
155: Here the vector $\tilde{\bf S}_j$ represents the localized
156: spin moment.
157: The scalar $\phi_{cj}$ is associated with charge fluctuations.
158: For the charge degrees of freedom, 
159: we take the uniform value at the saddle point:
160: $\phi_{cj}= \langle \overline{c}_j \sigma_0 c_j \rangle 
161: = 1-\delta$, with $\delta$ the doped hole concentration.
162: 
163: 
164: Thus, the approximate action is given by
165: \begin{eqnarray}
166: S &=& S_0 + \int_0^{\beta} d\tau 
167: \left[
168: - U \sum_j 
169: \tilde{\bf S}_j \cdot
170: \overline{c}_j {\mbox{\boldmath ${\bf \sigma}$}} c_j
171: + U \sum_j \tilde{\bf S}_j^2 \right].
172: \label{eq_Sj}
173: \end{eqnarray}
174: Note that the first term in the square brackets has the form of
175: Hund coupling between the localized spin moment and the carrier's
176: spin.
177: 
178: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
179: %   N{\' e}el order
180: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
181: \section{Evolution of N{\' e}el order by hole doping}
182: \label{sec_Neel}
183: Now we consider doped holes' hopping effects on N{\' e}el order.
184: Our strategy is the following:
185: First, we estimate semiclassical staggered magnetic moments
186: by solving saddle point equations.
187: Secondly, we examine stability of N{\' e}el order 
188: against quantum fluctuations.
189: For analysis of quantum fluctuations, we use
190: Schwinger boson mean field theory.\cite{AROVAS_AUERBACH}
191: 
192: 
193: In the N{\' e}el ordered phase, the localized spin moment
194: $\tilde{{\bf S}}_j$ has the following form:
195: \begin{equation}
196: \tilde{\bf S}_j = (-1)^j \tilde{S} \hat{e}_z.
197: \label{eq_Sjs}
198: \end{equation}
199: Substituting this into Eq.\ref{eq_Sj}, and
200: then performing Fourier transformations,
201: we obtain
202: \begin{eqnarray}
203: S &=& {\sum_{\bf k}}' \sum_{i\omega_n} 
204: \left(
205: \begin{array}{cc}
206: \overline{c}_k (i\omega_n) & \overline{c}_{k+Q} (i\omega_n) 
207: \end{array}
208: \right)
209: %
210: \nonumber \\
211: & & \times
212: \left(
213: \begin{array}{cc} -i \omega_n + \epsilon_k - \mu &
214: -U \tilde{S} \sigma_z \\
215: -U \tilde{S} \sigma_z & -i \omega_n - \epsilon_k - \mu
216: \end{array} \right)
217: %
218: \left( \begin{array}{c} 
219: c_k (i\omega_n) \\
220: c_{k+Q} (i\omega_n)
221: \end{array} \right) 
222: \nonumber \\
223: & & 
224: + \beta N U \tilde{S}^2,
225: %
226: \end{eqnarray}
227: where $\epsilon_{\bf k}=-2t (\cos k_x + \cos k_y )$ and 
228: the summation in ${\bf k}$-space is taken over 
229: half of the first Brillouin-Zone. 
230: The energy dispersion of the carriers is given by 
231: $\pm E_{\bf k}$
232: with $E_{\bf k} = \sqrt{\epsilon_k^2+(U \tilde{S})^2}$.
233: After integrating out fermions, we obtain
234: 
235: \begin{eqnarray}
236: S &=& -2 {\sum_{\bf k}}'
237: \left\{
238:   \ln 
239:   \left[ 
240:  1+ {\rm e}^{ -\beta ( E_k - \mu )}
241:   \right]
242: \right. \nonumber \\ 
243: & & \left. + 
244:    \ln 
245:    \left[ 
246:  1+{\rm e}^{-\beta (-E_k - \mu ) }
247:    \right] 
248: \right\}
249: + \beta N U \tilde{S}^2
250: \end{eqnarray}
251: 
252: 
253: At zero temperature,
254: variation of the action with respect to
255: the chemical potential $\mu$ yields
256: \begin{equation}
257: \frac{1}{N} \sum_{{\bf k}}
258: \left[ \theta (E_k +\mu) + \theta (-E_k + \mu)
259: \right]
260: = 1-\delta
261: \label{eq_chemi}
262: \end{equation}
263: and variation with respect to $\tilde{S}$ yields
264: \begin{equation}
265: \frac{1}{2N} \sum_{\bf k}
266: \frac{U}{E_k}
267: \left[ \theta (E_k +\mu)- \theta (-E_k + \mu)
268: \right] = 1,
269: \label{eq_tilS}
270: \end{equation}
271: where $\theta (x) =1$ for $x>0$ and zero otherwise.
272: 
273: We compute $\tilde{S}$ by solving saddle point equations
274: (\ref{eq_chemi}) and (\ref{eq_tilS}).
275: Figure \ref{fig_half} shows $U/t$ dependence of
276: $\tilde{S}$ at half-filling.
277: The same result was obtained in Ref.\cite{SINGH_TESANOVIC}
278: based on the Hartree-Fock approximation.
279: Figure \ref{fig_utdel} shows doping dependence of $\tilde{S}$
280: at $U/t = 6,8,12$.
281: Figure \ref{fig_jeff} shows effective exchange interaction
282: $J(x)/J \equiv (2 \tilde{S})^2$.
283: Temperature dependence of the antiferromagnetic correlation
284: length is given by $\xi_{AF} = 0.26 \exp (1.38 J(x)/T)$
285: according to the renormalization group analysis of non-linear sigma
286: model and numerical simulations.\cite{MANOUSAKIS}
287: 
288: 
289: Note that non-zero values of $\tilde{S}$ do not necessarily
290: imply that there is N{\' e}el order.
291: Because solutions of saddle point equations neglect
292: quantum fluctuation effect.
293: We need to examine stability of N{\' e}el order
294: against quantum fluctuations.
295: For this purpose, we apply the Schwinger boson mean field theory.
296: \cite{AROVAS_AUERBACH}
297: In the Schwinger boson mean field theory,
298: the localized spin moment $\tilde{\bf S}_j$ is represented by
299: \begin{equation}
300: \tilde{\bf S}_j = \frac{1}{2} \overline{z}_j 
301: {\mbox{\boldmath ${\bf \sigma}$}}
302: z_j,
303: \label{eq_SB}
304: \end{equation}
305: with the constratint 
306: $\sum_{{\sigma}=\uparrow,\downarrow}z_{j\sigma}z_{j\sigma}=2\tilde{S}$.
307: We derive the effective action of the localized spin system by
308: integrating out the carrier fields.
309: Detail of calculations is given in Appendix \ref{ap_afh}.
310: The result is,
311: \begin{equation}
312: S_{\rm spin} = \int_0^{\beta} d\tau
313: \left[ \sum_{j\sigma} \overline{z}_{j\sigma}\partial_{\tau}
314: z_{j\sigma} - \frac{J}{2} \sum_{\langle i,j\rangle} 
315: \overline{A}_{ij} A_{ij} \right],
316: \end{equation}
317: where $J=4t^2/U$ and 
318: $\overline{A}_{ij}=\overline{z}_{i\uparrow} \overline{z}_{j\downarrow}
319: - \overline{z}_{i\downarrow} \overline{z}_{j\uparrow}$ and
320: $A_{ij}=z_{i\uparrow} z_{j\downarrow}
321: - z_{i\downarrow} z_{j\uparrow}$.
322: (In terms of the fields $\tilde{\bf S}_j$, $S_{\rm spin}$ is,
323: $S_{\rm spin} = \sum_j S_j^{\rm Berry} + \int_0^{\beta}
324: d\tau J \sum_{\langle i,j \rangle} 
325: \tilde{\bf S}_i \cdot \tilde{\bf S}_j$,
326: where $S_j^{\rm Berry}$ denotes the Berry phase term 
327: for the localized spin moment $\tilde{\bf S}_j$.
328: This is nothing but the 
329: spin $\tilde{S}$ antiferromagnetic Heisenberg model.)
330: 
331: In the Schwinger boson theory, N{\' e}el order is stabilized 
332: if $\tilde{S}$ is larger than $S_c=0.19660$
333: as shown in Ref.\cite{AROVAS_AUERBACH}.
334: (This result is briefly summarized in Appendix \ref{ap_sc}.)
335: The parameter regions of $\tilde{S} > S_c$ and 
336: $0<\tilde{S} < S_c$ are shown in Fig. \ref{fig_doping}.
337: In the $\tilde{S}>S_c$ regime, N{\' e}el order is stabilized.
338: Whereas in the $0 < \tilde{S} < S_c$ regime, N{\' e}el order is 
339: suppressed by quantum fluctuations.
340: The point $\tilde{S} = S_c$ can be taken as 
341: a quantum critical point.\cite{HERTZ_MILLIS}
342: The $0<\tilde{S}<S_c$ regime is called the quantum disordered regime.
343: \cite{CHN}
344: Note that this quantum disordered regime
345: is identical to that in Ref.\cite{CHN}.
346: The condition of $\tilde{S} > S_c$ for the
347: stability of N{\' e}el order turns out to be
348: $g<g_c$ in Ref.\cite{CHN}.
349: 
350: 
351: In Fig.~\ref{fig_doping}, the $\tilde{S}=0$ regime is also shown.
352: Contrary to the other regimes, there is no antiferromagnetic 
353: Heisenberg type correlations in this regime.
354: Therefore, spin fluctuations in this regime are disconnected to
355: the original spin correlations at half-filling.
356: 
357: 
358: \section{Quantum disordered phase in the strong coupling limit}
359: \label{sec_qd}
360: In this section, we discuss
361: the effective action for the quantum disordered regime
362: in the strong coupling limit.
363: The effective action in the quantum disordered regime
364: is given by
365: \begin{eqnarray}
366: S &=& \int_0^{\beta} d\tau
367: \left[ \sum_j \overline{c}_j \left( \partial_{\tau}
368: - \mu \right) c_j 
369: - t\sum_{\langle i,j \rangle} 
370: \left( \overline{c}_i c_j + 
371: \overline{c}_j c_i \right) \right. \nonumber \\
372: & & \left. - \frac{U}{2}
373: \sum_j \left( \overline{c}_j 
374: {\mbox{\boldmath ${\bf \sigma}$}} c_j \right)
375: \left( \overline{z}_j {\mbox{\boldmath ${\bf \sigma}$}}
376: z_j \right)
377: \right. \nonumber \\
378: & & \left. + \sum_{j\sigma}\overline{z}_{j\sigma}
379: \partial_{\tau} z_{j\sigma}
380: + \sum_{j\sigma} \lambda_j
381: \left( \overline{z}_{j\sigma} z_{j\sigma} - \tilde{S}
382: \right) \right. \nonumber \\
383: & & \left. -\frac{J}{2} \sum_{\langle i,j\rangle}
384: \left(\overline{z}_{j\downarrow} \overline{z}_{i\uparrow}
385: - \overline{z}_{j\uparrow} \overline{z}_{i\downarrow}
386: \right)
387: \left( z_{i\uparrow} z_{j\downarrow} 
388: - z_{i\downarrow} z_{j\uparrow} \right) \right].
389: \label{eq_S}
390: \end{eqnarray}
391: This action consists of the free fermion part
392: and the Schwinger boson part that describes
393: the spin $\tilde{S}$ antiferromagnetic Heisenberg model.
394: The interaction between them is of the form of Hund coupling.
395: The action (\ref{eq_S}) is so-called the spin-fermion model.
396: 
397: 
398: Now let us consider the strong coupling
399: limit, $U/t \rightarrow \infty$.
400: Since the Hund coupling term is dominant in this regime,
401: we move to a frame in which the Hund coupling term 
402: is diagonalized.
403: In such a frame, the doped hole's spin at the $j$-site
404: is in the direction of the localized spin moment $\tilde{\bf S}_j$.
405: The transformation to this frame is given by
406: \begin{equation}
407: c_j = U_j f_j,
408: \label{eq_tra1}
409: \end{equation}
410: where the matrix $U_j$ is given by
411: \begin{equation}
412: U_j = \left(
413: \begin{array}{cc} z_{j\uparrow} & -\overline{z}_{j\downarrow} \\
414: z_{j\downarrow} & \overline{z}_{j\uparrow} \end{array} \right).
415: \end{equation}
416: After this transformation, the action reads
417: \begin{eqnarray}
418: S
419: &=& \int_0^{\beta} d\tau
420: \left[ 
421: \sum_j \overline{f}_j \left(
422: \partial_{\tau} - \mu + \overline{U}_j \partial_{\tau} U_j \right) f_j
423: \right. \nonumber \\
424: & & \left. -t\sum_{\langle i,j \rangle} 
425: \left( \overline{f}_i \overline{U}_i U_j f_j
426: + \overline{f}_j \overline{U}_j U_i f_i \right) \right. \nonumber \\
427: & & \left. 
428: -\frac{U}{2} \sum_j (-1)^j \overline{f}_j \sigma_z f_j \right.
429: \nonumber \\
430: & & \left.
431: \sum_{j\sigma}\overline{z}_{j\sigma}
432: \partial_{\tau} z_{j\sigma}
433: + \sum_{j\sigma} \lambda_j
434: \left( \overline{z}_{j\sigma} z_{j\sigma} - \tilde{S}
435: \right) \right. \nonumber \\
436: & & \left. -\frac{J}{2} \sum_{\langle i,j\rangle}
437: \left(\overline{z}_{j\downarrow} \overline{z}_{i\uparrow}
438: - \overline{z}_{j\uparrow} \overline{z}_{i\downarrow}
439: \right)
440: \left( z_{i\uparrow} z_{j\downarrow} 
441: - z_{i\downarrow} z_{j\uparrow} \right)
442: \right].
443: \label{eq_S2}
444: \end{eqnarray}
445: Note that the hopping for the fermions from $j$-site to
446: $i$-site containes a matrix,
447: \begin{equation}
448: \overline{U}_i U_j = 
449: \left( \begin{array}{cc}
450: F_{ij} & -\overline{A}_{ij} \\
451: A_{ij} & \overline{F}_{ij} 
452: \end{array}
453: \right),
454: \end{equation}
455: with $F_{ij} = 
456: \overline{z}_{i\uparrow} z_{j\uparrow} 
457:   + \overline{z}_{i\downarrow} z_{j\downarrow}$
458: and $A_{ij} = 
459: z_{i\uparrow} z_{j\downarrow} -z_{i\downarrow} z_{j\uparrow}$.
460: Note that $F_{ij}$ describes ferromagnetic correlations
461: and $A_{ij}$ describes antiferromagnetic correlations 
462: in the localized spin system.
463: In fact, $\langle F_{ij} \rangle$ is taken for the mean field
464: in the forromagnetic spin system and 
465: $\langle A_{ij} \rangle$ is taken for the mean field
466: in the antiferromagnetic spin system in the 
467: Schwinger boson mean field theory.\cite{AROVAS_AUERBACH}.
468: 
469: 
470: A similar action can be derived in the slave-fermion 
471: mean field theory of the t-J model.
472: However, there is a crucial difference.
473: If we take the strong coupling limit of $U/t \rightarrow \infty$,
474: then one finds that the fermion hopping only couple to 
475: ferromagnetic correlations in the localized spin system.
476: Therefore, coupling between fermions and the gauge field 
477: that describes antiferromagnetic fluctuations is absent.
478: 
479: 
480: 
481: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
482: %
483: %
484: \section{Summary and Discussion}
485: \label{sec_dis}
486: In this paper, we investigate effects of doped holes' hopping
487: on N{\' e}el order.
488: What we have found is that disorder effects induced by
489: holes' hopping on N{\' e}el order is rather small.
490: In fact, the critical doping concentration is
491: $\delta_c \simeq 0.40$ at $U/t=10$.
492: This value is substantially larger than that in 
493: high-temperature superconductors.
494: Therefore, hopping processes of doped holes are 
495: not so effective in suppressing N{\' e}el order.
496: For destruction of N{\' e}el order in high-temperature superconductors,
497: holes must behave like an excitation which suppresses
498: N{\' e}el order more effectively.
499: Such an excitation would be intimately connected with
500: properties of the localized spin system.
501: 
502: 
503: After destruction of N{\' e}el order, the quantum 
504: disordered phase appears.
505: (This quantum disordered phase is special to 
506: the antiferromagnetic correlations.
507: In fact, there is no such phase in case of 
508: the ferromagnetic correlations 
509: because there is no quantum fluctuations that suppress 
510: the semiclassical long-rage order as in the antiferromagnetic
511: case.)
512: If we take the strong coupling limit $U/t\rightarrow \infty$,
513: then the coulping between the doped holes and the antiferromagnetic
514: fluctuations are lost.
515: Here we first derive the spin-fermion model starting from
516: the Hubbard model.
517: After representing the localized spin moments by
518: the Schwinger bosons, we take the strong coupling limit.
519: There is another way of taking this strong coupling limit.
520: If we take the strong coupling limit first at 
521: the spin-fermion model, it is believed that
522: the model is reduced to the t-J model.
523: Applying the slave-boson theory, we obtain a system of 
524: spinless fermions and bosons with a gauge field interaction.
525: In the derivation of the slave-boson representation of the 
526: t-J model, one big assumption is that there is 
527: a deconfinement phase of a U(1) gauge field theory.
528: Whether there is a deconfinement phase or not is still 
529: an unsolved issue.
530: By contrast, there is no need to assume a deconfinement
531: phase when we introduce boson fields to describe 
532: the localized spin moments at the spin-fermion model.
533: After taking the strong coupling limit, we obtain
534: a system of spinless fermions and bosons
535: but there is no gauge field interaction.
536: Our analysis suggests that taking the strong coupling limit
537: is not justified or indirectly suggests
538: that there is no deconfinement phase.
539: A situation such that taking the strong coupling limit
540: is not allowed occurs when there is a term of
541: spin-orbit coupling like $H_{\rm so} 
542: = i\sum_{\langle i,j \rangle} c_i^{\dagger}
543: {\mbox{\boldmath ${\bf \lambda}$}}_{ij} \cdot
544: {\mbox{\boldmath ${\bf \sigma}$}} c_j + {\rm h.c.}$.
545: In the presence of such a term, doped holes rotate
546: their spin at every hopping process.\cite{MORINARI}
547: Therefore, we cannot take the strong coupling limit.
548: 
549: 
550: \acknowledgments
551: I thank Professor S.~Sachdev for his kind hospitality at Yale University
552: where part of this work was done.
553: This work was supported by a Grant-in-Aid from the Ministry of Education,
554: Culture, Sports, Science and Technology of Japan.
555: 
556: 
557: 
558: 
559: 
560: \appendix
561: \section{Derivation of the antiferromagnetic Heisenberg model}
562: \label{ap_afh}
563: In this appendix, we derive the antiferromagnetic Heisenberg model
564: from $S_0 + S_U$, where $S_0$ is defined by Eq.~(\ref{eq_S0})
565: and
566: $S_U$ is defined by Eq.~(\ref{eq_Sj}),
567: by applying second order perturbation theory with respect to $t$.
568: \cite{LACOUR74}.
569: In order to describe the localized spin moments $\tilde{\bf S}_j$,
570: we introduce the Schwinger bosons through Eq.\ref{eq_SB}.
571: with the constraint $\sum_{\sigma=\uparrow,\downarrow}
572: z_{j\sigma}^{\dagger} z_{j\sigma} = 2\tilde{S}_j$.
573: We rotate the carrier's spin in the direction of the localized spin moment 
574: at the same site in terms of the following unitary transformation:
575: \begin{equation}
576: c_j = U_j f_j,
577: \end{equation}
578: where
579: \begin{equation}
580: U_j = \left( 
581: \begin{array}{cc}
582: z_{j\uparrow} & -\overline{z}_{j\downarrow} \\
583: z_{j\downarrow} & \overline{z}_{j\uparrow} 
584: \end{array} \right).
585: \end{equation}
586: The total action reads
587: \begin{eqnarray}
588: S &=& \int_0^{\beta} d\tau 
589: \left[
590: \sum_j \overline{f}_j 
591: \left(
592: \partial_{\tau} - \mu + \overline{U}_j \partial_{\tau} U_j \right) f_j
593: \right. \nonumber \\
594: & & \left. -t \sum_{\langle i,j \rangle}
595: \left( \overline{f}_i \overline{U}_i U_j f_j
596: + \overline{f}_j \overline{U}_j U_i f_i \right)
597: \right. \nonumber \\
598: & & \left. -\frac{U}{2} \sum_j \overline{f}_j \sigma_z f_j
599: \right].
600: \label{eq_a}
601: \end{eqnarray}
602: 
603: We integrate out $\overline{f}_j$ and $f_j$:
604: \begin{eqnarray}
605: S_{\rm eff} &=& -{\rm Tr} \ln \left[
606: \left( \partial_{\tau} - \mu + \overline{U}_j \partial_{\tau} U_j
607: -\frac{U}{2} \sigma_z \right) \delta_{ij} \right. \nonumber \\
608: & & \left. - t_{ij} \overline{U}_i U_j \right],
609: \label{eq_Sln}
610: \end{eqnarray}
611: where $t_{ij}=t$ for the nearest neighbor sites and 
612: $t_{ij}=0$ otherwise.
613: 
614: 
615: 
616: 
617: We expand the logarithm in Eq. (\ref{eq_Sln}) 
618: with respect to $t_{ij}$.
619: The second order term is
620: \begin{eqnarray}
621: S_{\rm eff}^{(2)} &=&
622: \frac12 {\rm Tr} 
623: \left[ \frac{1}{\partial_{\tau} - \mu
624: + \overline{U}_i \partial_{\tau} U_i - \frac{U}{2} \sigma_z}
625: t_{ij} \overline{U}_i U_j 
626: \right. \nonumber \\
627: & & \left. \times \frac{1}{\partial_{\tau} - \mu
628: + \overline{U}_j \partial_{\tau} U_j - \frac{U}{2} \sigma_z}
629: t_{ji} \overline{U}_j U_i  \right].
630: \end{eqnarray}
631: Applying the derivative expansion technique, we obtain
632: \begin{eqnarray}
633: S_{\rm eff}^{(2)}
634: &=& \frac{t^2}{\beta} 
635: \sum_{i\omega_n} \sum_{\langle i,j\rangle} \sum_{\sigma,\sigma'}
636: \frac{1}{i\omega_n + \mu + \frac{U}{2} \sigma}
637: \frac{1}{i\omega_n  + \mu + \frac{U}{2} \sigma'} \nonumber \\
638: & & \times
639: \int_0^{\beta} d\tau 
640: \langle \sigma | \overline{U}_i U_j (\tau) |\sigma' \rangle
641: \langle \sigma' | \overline{U}_j U_i (\tau) |\sigma \rangle
642: \nonumber \\
643: & & + ({\rm higher~derivatives})
644: \end{eqnarray}
645: After the summation over the fermion Matsubara frequencies,
646: we take $\beta U \rightarrow \infty$ limit.
647: Thus, we obtain
648: \begin{equation}
649: S_{\rm eff}^{(2)} = -\frac{J}{2} \int_0^{\beta} d\tau
650: \sum_{\langle i,j \rangle } \overline{A}_{ij} A_{ij},
651: \label{eq_sa}
652: \end{equation}
653: where $J=4t^2/U$ and 
654: $A_{ij} = z_{i\uparrow} z_{j\downarrow} - z_{i\downarrow} z_{j\uparrow}$
655: and $\overline{A}_{ij} = \overline{z}_{i\uparrow} \overline{z}_{j\downarrow}
656: - \overline{z}_{i\downarrow} \overline{z}_{j\uparrow}$.
657: 
658: 
659: On the other hand, the expansion of the logarithm
660: in Eq. (\ref{eq_Sln}) with respect to 
661: $\overline{U}_j \partial_{\tau} U_j$ gives the Berry phase term 
662: for the localized spin moments as follows.
663: The term with the first order of $\overline{U}_j \partial_{\tau} U_j$
664: is,
665: \begin{equation}
666: S_{\rm eff}^{\rm Berry} = -{\rm Tr} \frac{1}{\partial_{\tau}
667: -\mu - \frac{U}{2} \sigma_z} \overline{U}_j \partial_{\tau}
668: U_j.
669: \end{equation}
670: Applying the derivative expansion technique, we obtain
671: \begin{equation}
672: S_{\rm eff}^{\rm Berry} 
673: = \frac{1}{\beta} \sum_{i\omega_n} \sum_{\sigma}
674: \frac{1}{i\omega_n +\mu + \frac{U}{2} \sigma}
675: \sum_j \int_0^{\beta} d\tau
676: \langle \sigma |\overline{U}_j \partial_{\tau} U_j |\sigma
677: \rangle.
678: \end{equation}
679: After the summation over the fermion Matsubara frequency,
680: we take $\beta U \rightarrow \infty$ limit.
681: Thus, we obtain
682: \begin{equation}
683: S_{\rm eff}^{\rm Berry} =
684: \int_0^{\beta} d\tau 
685: \sum_{j \sigma} \overline{z}_{j\sigma} \partial_{\tau} 
686: z_{j\sigma}.
687: \end{equation}
688: This is nothing but the Schwinger boson representation 
689: of the spin's Berry phase.
690: In terms of the original spin moment fields $\tilde{\bf S}_j$,
691: the action 
692: $S_{\rm spin} = S_{\rm eff}^{\rm Berry} + S_{\rm eff}^{(2)}$
693: turns out to be the antiferromagnetic Heisenberg model.
694: 
695: 
696: \section{The computation of $S_c$}
697: \label{ap_sc}
698: Here we briefly summarize the result of Ref.~\cite{AROVAS_AUERBACH}
699: for $S_c = 0.19660$.
700: 
701: In terms of the Schwinger boson fields,
702: which is defined in Eq.~(\ref{eq_SB}), 
703: the antiferromagnetic Heisenberg Hamiltonian 
704: $H=J\sum_{\langle i,j\rangle} \tilde{\bf S}_i \cdot \tilde{\bf S}_j$,
705: reads
706: \begin{equation}
707: H=-\frac{J}{2} \sum_{\langle i,j\rangle}
708: \left( z_{i\uparrow}^{\dagger} z_{j\downarrow}^{\dagger}
709: -z_{i\downarrow}^{\dagger} z_{j\uparrow}^{\dagger} \right)
710: \left( z_{i\uparrow} z_{j\downarrow} - z_{i\downarrow} z_{j\uparrow}
711: \right)
712: +2J \tilde{S}^2 N.
713: \end{equation}
714: We introduce mean fields $A_{ij} = \langle z_{i\uparrow} z_{j\downarrow}
715: -z_{i\downarrow} z_{j\uparrow} \rangle$ 
716: and $A_{ij}^* = \langle z_{i\uparrow}^{\dagger} z_{j\downarrow}^{\dagger}
717: -z_{i\downarrow}^{\dagger} z_{j\uparrow}^{\dagger} \rangle$
718: and assume 
719: the uniform value $A_{ij}=A_{ij}^*=A={\rm const}$.
720: Then, the free energy of the system is given by
721: \begin{equation}
722: {\cal F} = \frac{2}{\beta N}
723: \sum_{\bf k} \ln \left[ 2\sinh \left( \frac{\beta \omega_{\bf k}}{2} \right)
724: \right] + J A^2 - \lambda (2\tilde{S} + 1),
725: \end{equation}
726: where $\lambda$ is a Lagrange multiplier to impose the constraint
727: $\sum_{\sigma} z_{j\sigma}^{\dagger} z_{j\sigma} = 2\tilde{S}$,
728: and $\omega_{\bf k}=\sqrt{\lambda^2-4A^2J^2 \alpha_{\bf k}^2}$
729: with $\alpha_{\bf k}=(\sin k_x + \sin k_y)/2$.
730: 
731: 
732: The variation with respect to $\lambda$ and $A$ yields
733: \begin{equation}
734: \frac{1}{N} \sum_{\bf k} \frac{J\alpha_{\bf k}^2}{\omega_{\bf k}}
735: \coth \frac{\beta\omega_{\bf k}}{2} = \frac12,
736: \end{equation}
737: \begin{equation}
738: \frac{1}{N} \sum_{\bf k} \frac{1}{\omega_{\bf k}}
739: \coth \frac{\beta \omega_{\bf k}}{2} =\frac{2\tilde{S}+1}{\lambda}.
740: \label{eq_aa2}
741: \end{equation}
742: At zero temperature, Eq.(\ref{eq_aa2}) has the following form:
743: \begin{equation}
744: \tilde{S} = \frac{1}{2}
745: \left( \frac{4}{\pi^2} \int_0^1 d\gamma
746: \frac{K(\sqrt{1-\gamma^2})}{\sqrt{1-p^2 \gamma^2}}-1 \right),
747: \label{eq_aa2b}
748: \end{equation}
749: with $K$ the complete elliptic function and $p=2AJ/\lambda$.
750: The right hand side is a monotonically increasing function 
751: with respect to $p$ and it takes the maximum value 
752: $S_c \equiv 0.19660$ at $p=1$.
753: Therefore, Eq.~(\ref{eq_aa2b}) has a solution of $p<1$
754: for $\tilde{S} < S_c$.
755: Solutions of $p<1$ imply that spin wave excitation 
756: have gap, 
757: and there is no Bose-Einstein condensation 
758: of Schwinger bosons.\cite{YOSHIOKA}.
759: Therefore, there is no N{\' e}el order for 
760: $\tilde{S}<S_c$.
761: 
762: 
763: \begin{references}
764: 
765: \bibitem{SCHULZ} H.~J.~Schulz, Phys. Rev. Lett. {\bf 65}, 2462 (1990).
766: 
767: \bibitem{WTL} See, also, Z.~Y.~Weng, C.~S.~Ting, and T.~K.~Lee,
768: Phys. Rev. B {\bf 43}, 3790 (1991).
769: 
770: \bibitem{AROVAS_AUERBACH} D.~P.~Arovas and A.~Auerbach, 
771: Phys. Rev. B {\bf 38}, 316 (1988).
772: 
773: \bibitem{MORIYA} See, for example, 
774: T.~Moriya, {\it Spin Fluctuations in Itinerant Electron
775: Magnetism}, Springer-Verlag, Berlin (1985).
776: 
777: \bibitem{SINGH_TESANOVIC} A. Singh and Z. Te\v{s}anovi\'{c},
778: Phys. Rev. B {\bf 41}, 614 (1990).
779: 
780: \bibitem{MANOUSAKIS} E.~Manousakis, Rev. Mod. Phys. {\bf 63}, 1 (1991).
781: 
782: \bibitem{HERTZ_MILLIS} J.~A.~Hertz, Phys. Rev. B {\bf 14}, 1165 (1976);
783: A.~J.~Millis, Phys. Rev. B {\bf 48}, 7183 (1993).
784: 
785: \bibitem{CHN} S.~Chakravarty, B.~I.~Halperin, and D.~R.~Nelson,
786: Phys. Rev. Lett. {\bf 60}, 1057 (1988).
787: 
788: \bibitem{MORINARI} T.~Morinari, 
789: Phys. Rev. B {\bf 65}, 064513 (2002).
790: 
791: \bibitem{LACOUR74} P.~Lacour-Gayet and M.~Cyrot, 
792: J. Phys. C {\bf 7}, 400 (1974). 
793: 
794: \bibitem{YOSHIOKA} D.~Yoshioka, J. Phys. Soc. Jpn. {\bf 58}, 3733 (1989);
795: S.~Sarker {\it et al}., Phys. Rev. B {\bf 40}, 5028 (1989).
796: 
797: 
798: \end{references}
799: 
800: 
801: %
802: % Figures
803: %1 % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % 
804: \begin{figure}[htbp]
805: \center
806: \epsfxsize=2.5truein
807: \psfig{file=fig1.ps,width=2.5in,angle=270}
808: \vspace{0.1in}
809: \caption{The localized spin moment $\tilde{S}$ versus $U/t$ at half-filling.}
810: \label{fig_half}
811: \end{figure}
812:  
813: 
814: %2 % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % 
815: \begin{figure}[htbp]
816: \center
817: \epsfxsize=3.2truein
818: %\psfig{file=fig2.ps,width=2.5in,angle=270}
819: %\psfig{file=s_del.ps,width=3.2in,angle=0}
820: \psfig{file=fig2.ps,width=3.2in,angle=0}
821: \caption{The semiclassical staggered moment $\tilde{S}$
822: versus the doping concentration $\delta$ for 
823: $U/t=6,8,12$.}
824: \label{fig_utdel}
825: \end{figure}
826: 
827: %3 % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % 
828: \begin{figure}[htbp]
829: \center
830: \epsfxsize=3.2truein
831: %\psfig{file=fig2.ps,width=2.5in,angle=270}
832: \psfig{file=fig3.ps,width=3.2in,angle=0}
833: \caption{The doping dependence of the effective 
834: exchange interaction $J(\delta)/J \equiv (2 \tilde{S})^2$
835: for $U/t=6,8,12$.
836: }
837: \label{fig_jeff}
838: \end{figure}
839: 
840: %4 % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % 
841: \begin{figure}[htbp]
842: \center
843: \epsfxsize=2.5truein
844: \psfig{file=fig4.ps,width=2.5in,angle=270}
845: \caption{The N{\' e}el order regime $\tilde{S}>S_c$ (N{\' e}el) 
846: and the quantum disordered regime 
847: $0< \tilde{S} < S_c$ (QD) on the $U/t$-$\delta$ plane.
848: In the $\tilde{S}=0$ regime, there is no antiferromagnetic correlation
849: that is associated with the Heisenberg antiferromagnetic type correlation.
850: }
851: \label{fig_doping}
852: \end{figure}
853:  
854:  
855: 
856: \end{multicols}
857: \end{document}
858: 
859: