cond-mat0701703/ms2.tex
1: \documentclass[prl,twocolumn,aps,superscriptaddress,showpacs,amsmath]{revtex4}
2: \usepackage{graphicx,bm}
3: \begin{document}
4: 
5: 
6: \title{Interchain-Frustration-Induced Metallic State
7:  in Quasi-One-Dimensional Mott Insulators}
8: 
9: 
10: 
11: \author{M.\ Tsuchiizu}
12:  
13: \affiliation{Department of Physics, Nagoya University, 
14: Nagoya 464-8602, Japan}
15: 
16: \author{Y.\ Suzumura}
17:  
18: \affiliation{Department of Physics, Nagoya University, 
19: Nagoya 464-8602, Japan}
20: 
21: \author{C.\ Bourbonnais}
22: 
23: \affiliation{D\'{e}partement de Physique, Universit\'{e} de Sherbrooke, 
24:  Sherbrooke, Qu\'{e}bec, Canada J1K-2R1}
25: 
26: 
27: \date{September 28, 2007}
28: 
29: \begin{abstract}
30: The mechanism that drives a metal-insulator transition
31:  in an undoped quasi-one-dimensional Mott insulator
32:  is examined in the framework of the Hubbard model
33:  with two different hoppings $t_{\perp1}$ and $t_{\perp2}$
34:  between nearest-neighbor chains.
35: By applying an $N_\perp$-chain renormalization group method
36:   at the two-loop level, 
37:  we show how a metallic state emerges
38:   when both $t_{\perp1}$ and $t_{\perp2}$ exceed critical values. 
39: In the metallic phase, the quasiparticle weight becomes finite
40:  and  develops a strong momentum dependence.
41: We discuss the temperature dependence of the resistivity
42:  and the impact of our theory in the understanding
43:  of recent experiments on half-filled molecular conductors.
44: \end{abstract}
45: 
46: \pacs{71.10.Fd, 71.10.Pm, 71.30.+h}
47:  
48: \maketitle
49: 
50: The remarkable properties of 
51:   strongly correlated systems near   a metal to Mott  insulator (MI) transition
52:   stand out as one of the    richest  parts    of  the physics  of
53:   strongly correlated systems \cite{Imada1998}. 
54: This takes on particular importance in low dimensional  materials at
55:   half-filling, and especially in one dimension, 
56: where spin  and  charge  excitations are well  known to be invariably
57:   decoupled, an effect that receives experimental confirmation in
58:  the one-dimensional (1D) oxide material
59:   SrCuO$_2$ \cite{Kim2006}. 
60: How this picture  modifies when the  hopping of electrons in more than
61:  one spatial direction  progressively   grows
62: and a  higher dimensional metallic ground state becomes possible
63:  is a key issue  that remains   poorly understood.   
64:   This problem finds concrete applications in organic molecular
65:   compounds,  which constitute  very close realizations  
66: of  1D   systems. This is the case notably of 
67:   (TTM-TTP)I$_3$ and (DMTSA)BF$_4$ 
68:   which, as genuine half-filled  band materials
69:   due to the monovalent  anions I$_3^-$ and BF$_4^-$,
70:  see their Mott insulating state being gradually
71:   suppressed under pressure \cite{Mori1997,Mori2004}. 
72: Quite recently, resistivity measurements on
73:    (TTM-TTP)I$_3$ at   high pressure
74:  revealed that the temperature scale for the Mott insulating  behavior 
75:    is suppressed by an order of magnitude down to 
76:  $T_\mathrm{MI}\approx $ 20 K at $P =$ 8 GPa of pressure,
77:    with a  metallic ground state expected
78:   to  occur   above 10 GPa \cite{Yasuzuka}.
79:  It is the objective of this letter to propose a theoretical description 
80:  of this transition.
81: 
82: 
83: 
84:  Bosonization, renormalization group (RG) approaches to  the 1D Hubbard
85:  model in weak coupling and its  exact solution from  the Bethe ansatz  
86:  \cite{Dzyaloshinskii72,Bourbonnais2003,Giamarchi_book},
87:  show that electron-electron umklapp-scattering processes    
88:     are a key ingredient that promotes  the existence of a Mott
89:  insulating ground state at half-filling. 
90: The difficulty underlying the mechanism of the MI transition in the
91:  quasi-1D case resides in the fact that the decoupling of spin and
92:  charge excitations  on the metallic  side of the transition  at higher
93:  dimension 
94: does not occur,
95:  namely  when   Fermi-liquid quasiparticles excitations  appear.
96: This issue has been addressed theoretically
97:   by an RPA treatment of 
98:      interchain hopping \cite{Essler2002},
99:  in which  its feedback effect   on
100:   the Mott gap is neglected for the self-energy
101:   of the one-particle Green's function.
102:   The RPA results   shows  the existence of a Fermi-liquid metallic state
103:   with electron and hole Fermi-surface pockets when the interchain 
104:   hopping exceeds a critical value.
105: The important role of 
106: this feedback effect has been pointed out  recently from the use  of dynamical mean-field theory,
107:   extended   to include the influence of 1D fluctuations 
108:   (the chain-DMFT) \cite{Giamarchi2001} 
109:  -- an approach expected to be workable in the strong-coupling regime.
110: In the weak-coupling regime, however, the electron-electron umklapp scattering 
111:    and the characteristics of its coupling to nesting of  the whole  Fermi surface,  play a key role 
112:     \cite{Dzyaloshinskii72,Bourbonnais2003,Giamarchi_book},  both for
113:  the Mott insulating state  and for the possibility of the MI transition
114:  that emerges in the quasi-1D case.  
115: Transverse dispersion 
116:   which has a tendency towards the realization of the
117:  deconfined metallic states
118:   \cite{Bourbonnais2003,Giamarchi_book,Giamarchi2001} 
119:    is in general overlooked  in the calculation of one-particle Green's function from mean-field like approaches. 
120:     The two-loop RG approach    avoids  such an approximation  and treats the Fermi-surface nesting  conditions properly. These 
121:  can be strongly   altered in the quasi-1D case,
122: especially in the presence of frustration 
123: in the electron kinetics. 
124:  
125:  
126: 
127: In this Letter, we apply an  $N_\perp$-chain RG approach  
128:   to  quasi-1D  half-filled system as developed recently at
129:   the two-loop level \cite{Tsuchiizu2006,Tsuchiizu2006b}. The  momentum
130:   dependence for 
131:    nesting of  the whole Fermi surface 
132:   and  that for   couplings, including   umklapp scattering  
133:   \cite{Duprat2001,Doucot2003,Rohe06}, 
134: are  taken into account in a systematic way
135: in the calculation of one-particle Green's function  and four-point
136:   vertices  \cite{Tsuchiizu2006}.
137:  The combined impact of correlations and  nesting frustration
138:   induced by interchain hopping on the  transverse momentum dependence of the 
139:  quasiparticle weight is obtained in  
140:      weak-coupling regime.
141: 
142: 
143: 
144: We consider the quasi-1D half-filled Hubbard model 
145:  on  an anisotropic 
146: triangular lattice  [Fig.\ \ref{fig:model}(a)], with
147:   the transfer energies
148:   $t_\parallel \gg |t_{\perp 1}|,|t_{\perp 2}|$ 
149:   ($t_\parallel$ is  the energy along chains and 
150:   $t_{\perp1}$  and  $t_{\perp2}$  
151:   are those between chains).
152: Our Hamiltonian, with the on site Coulomb   repulsion $U$
153: is given by 
154: %====================================================================
155: \begin{eqnarray}
156: H &=& 
157: -t_\parallel \sum_{j,\ell,s} 
158: \left(c_{j,\ell,s}^\dagger c_{j+1,\ell,s}^{}+\mathrm{H.c.} \right)
159: \nonumber \\ && {}
160: -t_{\perp 1} \sum_{j,\ell,s} 
161: \left(c_{j,\ell,s}^\dagger c_{j,\ell+1,s}^{}+\mathrm{H.c.} \right)
162: \nonumber \\ && {}
163: -t_{\perp 2}\sum_{j,\ell,s} 
164: \left(c_{j,\ell,s}^\dagger c_{j+1,\ell+1,s}^{}+\mathrm{H.c.} \right)
165: \nonumber \\ && {}
166: + U \sum_{j,\ell} n_{j,\ell,\uparrow} n_{j,\ell,\downarrow}
167: -\mu \sum_{j,\ell,s}  c_{j,\ell,s}^\dagger c_{j,\ell,s}^{}.
168: \label{eq:model}
169: \end{eqnarray}
170: %====================================================================
171: The operator  $c_{j,\ell,s}$ denotes the annihilation  of an electron 
172:   on the $j$th site in the $\ell$th chain with spin $s$, and 
173:  $n_{j,\ell,s}=c_{j,\ell,s}^\dagger c_{j,\ell,s}^{}-\frac{1}{2}$.
174: Here  $\ell=1,\cdots,N_\perp$ is the chain index.
175: We take the continuum limit along the chain direction, whereas
176:  in the transverse direction we consider a finite system
177:  having an even number of chains $N_\perp$ with  
178:  the boundary condition  $c_{j,N_\perp+1,s}=c_{j,1,s}$.   
179: By applying a  Fourier transform,
180: the kinetic term can be rewritten as
181:  $H_0 = \sum_{\bm k,s}
182: \varepsilon(\bm k) \,
183: c_s^\dagger (\bm k) \,c_s^{} (\bm k)$, 
184: where $\bm k \equiv (k_\parallel,k_\perp)$ and
185:    the energy dispersion is given by
186: $\varepsilon(\bm k) =
187:   -2t_\parallel \cos k_\parallel 
188:   -2t_{\perp 1} \cos k_\perp
189:   -2t_{\perp 2} \cos (k_\parallel + k_\perp) - \mu$.
190: For  $|t_{\perp i}| \ll t_\parallel$,
191:   the right- and left-moving electrons in the 1D case are well-defined 
192:   due to  an open Fermi surface [Fig.\ \ref{fig:model}(b)] and 
193:   the shape of the Fermi surface can be parametrized  by the 
194:     transverse momentum  $k_\perp$ \cite{Tsuchiizu2006}.
195: To lowest order in the  interchain hoppings 
196:   $t_{\perp 1}$ and $t_{\perp 2}$,
197: the Fermi surfaces for right $(+)$ and left $(-)$ moving electrons, 
198:   as a function of $k_\perp$,
199:   are given by 
200: $k_F^\pm(k_\perp)
201: = \pm  (\pi/2) 
202: \pm (t_{\perp 1}/t_\parallel) \cos k_\perp
203: - (t_{\perp 2}/t_\parallel) \sin k_\perp$
204: and  the chemical potential is   $\mu=O(t_{\perp i}^2)$.
205: By considering the weak-interacting case 
206:  and neglecting the $k_\perp$ dependence of the velocity,
207:     the linearized dispersion used in  the RG method takes the simple form:
208: $\varepsilon_p (\bm k)= p v[k_\parallel -k_F^p(k_\perp)]$
209: with $v=2t_\parallel$ and $p=\pm$. 
210: 
211: %====================================================================
212: \begin{figure}[t]
213: \includegraphics[width=7cm]{fig1.eps}
214: \caption{
215: (Color online).
216: (a) Lattice geometry of the present model.
217: (b) The corresponding Fermi surface where
218: the case for $N_\perp =16$  is shown.
219: }
220: \label{fig:model}
221: \end{figure}
222: %======================================================================
223: 
224:   
225: 
226: We follow the formulation of the two-loop RG 
227:   method of Ref.\ \cite{Tsuchiizu2006}
228: and first introduce the $g$-ology coupling constants
229:  [see Eq.\ (2.8) in Ref.\ \cite{Tsuchiizu2006}]:,
230:  namely   the backward ($g_{1\perp}$), forward  ($g_{2\perp}$),
231:  and umklapp  ($g_{3\perp}$) scatterings with  opposite spins, and 
232:    the forward scattering  ($g_\parallel$)  and umklapp
233:   ($g_{3\parallel}$) scatterings for parallel spins.
234: The coupling constants are 
235:    renormalized differently, developing an external -- transverse -- momenta 
236:   dependence in the vertex corresponding to a  patch-index  dependence;
237:  that is,  $g_\nu \to g_{\nu(q_\perp,k_{\perp 1},k_{\perp 2})}$,
238:   where $k_{\perp 1}$ and $k_{\perp 2}$ are the transverse momenta
239:   for the right-going fermions,  and $q_\perp$ is the 
240:   momentum transfer \cite{Tsuchiizu2006}.  
241: The magnitude  of the  initial couplings are
242:   given by   
243:   $g_{1\perp}=g_{2\perp}=g_{3\perp}=U$ and 
244:   $g_{\parallel}=g_{3\parallel}=0$.
245: The RG equations   are derived by 
246:  scaling the bandwidth cutoff $\Lambda(\simeq 2\pi t_\parallel )$  as
247:    $\Lambda_l= \Lambda e^{-l}$,  where $l$ is the scaling parameter 
248:   \cite{Tsuchiizu2006}.
249: The explicit forms of the two-loop RG equations 
250:   for all   coupling constants 
251: in the case  $t_{\perp2}=0$ are 
252:  given in Ref.\ \cite{Tsuchiizu2006}.
253: We solve here these  two-loop RG equations numerically for 
254:  a system with $N_\perp=16$.
255: For even $N_\perp$, the solution of the RG flows
256:   indicates the existence of a finite spin gap in the low-energy limit, 
257:    which is  expected to vanish 
258:   in the infinite $N_\perp$ limit.  
259: The characteristic scale $l_{N_\perp}$ above which
260:   finite size effect would appear can be 
261:   roughly estimated  to be 
262:   $l_{N_\perp} \approx 
263:    \ln [\Lambda/|t_{\perp i} \sin (2\pi/N_\perp)|]$.
264: For   $N_\perp=16$ 
265: and $t_{\perp 1}/t_\parallel=0.1$ 
266:    this gives   $l_{N_\perp}\approx 5$, 
267:   which is sufficient to obtain the MI transition 
268:  without  finite size effects 
269: for the present choice of parameters $U/t_\parallel=2$ or $1.5$.
270: 
271: 
272: For the bipartite lattice ($t_{\perp2}=0$),
273:    the system remains always insulating 
274:   even for large $t_{\perp1}$ \cite{Tsuchiizu2006}.
275: This is due to   
276:  perfect nesting condition for the Fermi surface.
277: In this case the interchain hopping $t_{\perp1}$ is relevant and becomes 
278:    large under scaling procedure, and 
279:   some umklapp
280:   couplings become small. However, 
281:  a macroscopic number of these
282:  couplings  remains relevant
283:   due to   perfect nesting at   vector 
284:    $\bm Q=(\pi,\pi)$.
285: In the case of large $t_{\perp2}$,
286:    these relevant umklapp 
287:  couplings
288:  are strongly reduced 
289:  and all umklapp terms
290:  remain weak.
291: The charge gap then collapses 
292:  and  a metallic state 
293:  with a Fermi surface emerges as a consequence of 
294:     nesting deviations   that are introduced
295: by large interchain frustration.
296: 
297: 
298: 
299: 
300: One-particle properties are best studied  from the 
301:  quasiparticle  weight   $z_{k_\perp}$.
302: In the RG method at the two-loop level \cite{Tsuchiizu2006}, 
303:   the $k_\perp$ dependence of the 
304:   self-energy is taken into account in a non perturbative way.  This contrasts  with the chain-DMFT,  which treats 
305:   the one-particle self-energy, as a one-chain
306:    $k_\perp$-independent quantity  of 
307:    the one-particle Green's function  
308:   \cite{Giamarchi2001}.    In the RG formalism
309:  the transverse momentum dependence \cite{Tsuchiizu2006} 
310:   is taken into account  explicitly, and  
311:     the quasiparticle weight of the one-particle Green's function
312:    takes the form
313:       $z_{k_\perp}\equiv z_{k_\perp}^{n} z_{k_\perp}^{u}$,
314:   where $z_{k_\perp}^{n}$ and $z_{k_\perp}^{u}$
315:   are the contributions  
316:  coming from the normal and umklapp parts of the scattering,
317:   respectively.
318: Here, 
319:   one can  focus   on the umklapp contribution
320:  $z_{k_\perp}^{u}$, 
321:   since $z_{k_\perp}^{n}$ remains finite for
322:  $l<l_{N_\perp}$  in the metallic state.
323: The explicit form of $z_{k_\perp}^{u}$ reads
324: \cite{Tsuchiizu2006} 
325: %====================================================================
326: \begin{eqnarray}
327: && \hspace*{-.5cm}
328: z_{k_{\!\perp}}^{u} 
329: \! = \!
330: \exp
331: \left[
332: -\frac{1}{2N_\perp^2} \sum_{q\!_\perp,k'_{\!\perp}} \int
333: dl \,
334:  G_{\Sigma u(q_{\!\perp},k_{\!\perp},k'_{\!\perp})}^2 \,
335:   J^{u}_{1(q_{\!\perp},k_{\!\perp},k'_{\!\perp})}
336: \right],
337: \nonumber \\
338: \label{eq:z}
339: \\
340: && \hspace*{-.5cm}
341: G_{\Sigma u(q_\perp,k_\perp,k'_\perp)}^2  \equiv
342:   G_{c(q_\perp,k_\perp,k'_\perp)}^2
343: + G_{c(\pi-q_\perp+k_\perp+k'_\perp,k_\perp,k'_\perp)}^2
344: \nonumber \\ && {}\qquad\qquad
345: - G_{c(q_\perp,k_\perp,k'_\perp)}  G_{c(\pi-q_\perp+k_\perp+k'_\perp,
346:       k_\perp,k'_\perp)},
347: \end{eqnarray}
348: %====================================================================
349: where $G_{c(q_\perp,k_\perp,k'_\perp)}
350:   \equiv g_{3\perp(q_\perp,k_\perp,\pi-k'_\perp)}/(2\pi v)$.
351: The quantity
352:   $J^{u}_{1(q_\perp,k_\perp,k'_\perp)}$ is a
353:   cutoff function depending on the transverse dispersion 
354:  \cite{Tsuchiizu2006}
355:   given by 
356:   $J^{u}_{1(q_\perp,k_\perp,k'_\perp)} =1$ for
357:   $|A_{q_\perp,k_\perp,k'_\perp}'|\ll \Lambda$ and 
358:    $J^{u}_{1(q_\perp,k_\perp,k'_\perp)} \approx 0$ for
359:   $|A_{q_\perp,k_\perp,k'_\perp}'|\gg \Lambda$ where
360: $A_{q_\perp,k_\perp,k'_\perp}' \equiv
361:   2t_{\perp 1} [\cos k _\perp + \cos (k_\perp-q_\perp)
362:                 - \cos k'_\perp - \cos (k'_\perp-q_\perp)]
363: - 2t_{\perp 2} [\sin k_\perp - \sin (k_\perp-q_\perp)
364:                 + \sin k'_\perp - \sin (k'_\perp-q_\perp)]$.
365: The $k_\perp$ dependence of 
366:  the umklapp-scattering contribution to $z_{k_\perp}$
367:   for several values of $t_{\perp2}$
368:    at  fixed $t_{\perp1}/t_\parallel=0.1$
369:    is shown in  Fig.\ \ref{fig:phasediagram} (a).
370: For weak frustration $t_{\perp2}(<t_{\perp2}^{\mathrm{c}})$, this quantity 
371: is small    showing the existence of an insulating phase 
372: with no  low-energy quasiparticles.
373: On the other hand, 
374:   for strong frustration $t_{\perp2}(>t_{\perp2}^{\mathrm{c}})$,
375:  it takes sizable values  and shows  a strong $k_\perp$ dependence.
376: In the proximity of the insulating phase,
377:  the quantity $z_{k_\perp}^{u}$  presents a broad maximum
378:  around $k_\perp \approx \pm \pi /2$, 
379: which behavior implies the emergence of
380:  sections of Fermi surface 
381:  \cite{Essler2002}, or  ``cold'' regions
382:  around $\bm k \approx  (k_F^p(\pm\pi/2),\pm\pi/2)$
383:  \cite{Duprat2001}. 
384:  The dips or ``hot spots'' in the quasiparticle weight correspond to
385:  regions of the Fermi surface where the nesting, 
386:  within  $2\pi/N_\perp$ accuracy, remain
387:   the most favorable.
388: In the  perfectly nested case    no  Fermi surface is found for  
389:        any value of $U$.  The overall profile is then intrinsically linked to
390:   the momentum dependent nesting properties 
391:   over the whole Fermi surface \cite{Rohe06}. These are not taken 
392:   into account in previous analyses 
393:    \cite{Giamarchi2001,Essler2002},
394:    which reach different conclusions in the weak-coupling regime.      
395: 
396: %====================================================================
397: \begin{figure}[b]
398: \includegraphics[width=8.5cm]{fig2.eps}
399: \caption{
400: (Color online).
401: (a) Umklapp-scattering contribution of the
402:    quasiparticle weight 
403:    $z_{k_\perp}^{\mathrm{u}}$  
404:  for $U/t_\parallel=2$ and
405:   $t_{\perp1}/t_\parallel=0.1$, 
406:  with several $t_{\perp2}$.
407: (b) Ground-state phase diagram 
408:   on the plane of $t_{\perp1}/t_\parallel$ and $t_{\perp2}/t_\parallel$.
409: The dotted line denotes the condition of Eq.\ (\ref{eq:boundary}).
410: }
411: \label{fig:phasediagram}
412: \end{figure}
413: %======================================================================
414: 
415: 
416: 
417: The boundary for the metal-insulator transition 
418:   can be determined analytically
419:   by noting that
420:   the metallic phase  of the RG analysis  is linked to the irrelevance 
421:  of umklapp scattering. 
422: The metallic phase boundary  can thus be obtained when 
423:   the energy scale of  imperfect nesting  
424:   for  umklapp scatterings
425:   becomes comparable to the energy scale 
426:   $\Delta_\rho^{\mathrm{1D}}$ [$\propto \sqrt{t_\parallel U} \exp(-2\pi t/U)$]
427:  of  the 1D  Mott gap.
428: By noting that the nesting vector of the particle-hole loop,
429:  which couples to umklapp,
430:  is given by
431:   $\bm Q = (\pi, \pi \pm 2\alpha )$ with $\tan \alpha=t_{\perp2}/t_{\perp1}$,
432: the degree of the imperfect nesting (the amplitude 
433:   of the quantity $A'_{\pi\pm 2\alpha,k_\perp,k'_\perp}$) becomes 
434: $\sqrt{t_{\perp 1}^2 + t_{\perp2}^2}  \sin 2 \alpha$.
435: The   phase boundary  for the MI transition  is
436:  then determined by the condition    
437: %====================================================================
438: \begin{equation}
439: \frac{t_{\perp1}t_{\perp2}}{\sqrt{t_{\perp 1}^2 + t_{\perp2}^2}}
440: = c  \Delta_\rho^{\mathrm{1D}},
441: \label{eq:boundary}
442: \end{equation}
443: %====================================================================
444: where  $c $ is a numerical constant being of the order of unity.
445: The small difference between the numerical results and 
446:   the above analytical expression comes from  
447:   the renormalization of   interchain hopping due to 
448:   the normal scattering processes. The Mott gap is
449:   also renormalized by   interchain hopping.
450: The ground-state phase diagram in the $(t_{\perp1}/t_\parallel, t_{\perp2}/t_\parallel)$ plane   is shown in 
451:  Fig.\ \ref{fig:phasediagram}(b), where 
452:  the dotted line denotes Eq.\ (\ref{eq:boundary}).
453:   The ambiguities in the  cutoff functions of the RG
454:   \cite{Doucot2003,Tsuchiizu2006},
455:       prevent  us  from obtaining a precise location of the phase
456:       boundary. 
457: 
458:       
459: 
460: The Mott transition  has also been 
461:   addressed in two-dimensional Hubbard model 
462:   on an anisotropic triangular lattice 
463:  with   nearest-neighbor hopping $t$ and 
464:   next-nearest neighbor hopping $t'$
465:    \cite{Imada_Kyung}.   
466: The present model (\ref{eq:model}) can be connected to the
467:   two-dimensional Hubbard model
468:   by taking 
469:   $t_{\parallel}\to t$, $t_{\perp1}\to t$, and
470:   $t_{\perp2}\to t'$. 
471: While our approach  is restricted to the small 
472:  interchain hopping,
473:    the    metal-insulator transition obtained here 
474: for finite frustration  is consistent with the numerical results
475:   in two dimensions. 
476: 
477: 
478: 
479: 
480: From the solution of the scaling flows of the umklapp scatterings,
481:   the temperature dependence of the resistivity can be qualitatively
482:  calculated 
483:  from the  memory function approach
484:   combined with the RG method
485:   by using $l=\ln (\Lambda/T)$ \cite{Giamarchi_book,Giamarchi1991}.
486:  By extending the approach to the quasi-1D case,
487:    the perturbative expression of the conductivity reads 
488: $\rho(T) \propto  N_\perp^{-3}  \sum_{q_\perp,k_\perp,k'_\perp} 
489: G_{c(q_\perp,k_\perp,k'_\perp)}^2(l)  \, e^{-l}$.
490: While this  formula is not valid for large $t_{\perp i}/T$
491:   \cite{Giamarchi1991}, 
492:  it will depict the qualitative  temperature dependence
493:  of resistivity. 
494: Typical behaviors of the resistivity 
495:  for  $\Delta_\rho^{\mathrm{1D}}< t_{\perp 1}$,
496:  obtained 
497:   from this formula, are shown in Fig.\ \ref{fig:resis}(a).
498: At high temperature ($T>t_{\perp 1}$, $t_{\perp2}$),
499:   the effects of interchain kinetics are masked due to the 
500:   thermal fluctuations where the Tomonaga-Luttinger (TL) liquid behavior 
501:    is reproduced.
502: At low temperature, the insulating behavior can be seen 
503:   for small $t_{\perp2}$, 
504:  while the metallic behavior is found for strong
505:   frustration (large $t_{\perp2}$).
506: A finite-temperature phase diagram, 
507:  as a function of $t_{\perp2}$, 
508: can be obtained from this behavior,
509:   as  shown schematically 
510:  in Fig.\ \ref{fig:resis}(b).
511: At high temperature, 
512: the TL liquid state is realized,
513:  which is followed 
514:     in the crossover region by the development of a 
515:   Fermi surface.
516:  The effect of the frustration 
517:   is yet masked by thermal fluctuations,  i.e.,
518:   this region  can be described effectively by a nested Fermi surface.  
519: At further low temperature, the state moves to the Mott insulator 
520:   or the Fermi liquid (FL) state   depending on $t_{\perp2}$.
521: In the FL state, the effect of $t_{\perp2}$ becomes prominent and
522:  the nesting conditions of the full Fermi surface 
523:   are altered.
524: It can also be seen from  Fig.\ \ref{fig:resis}(a) that 
525:   the characteristic temperature $T_{\mathrm{FL}}$ at which the 
526:    crossover   to the metallic FL state  takes place 
527:    increases with $t_{\perp 2}$.
528: For $\Delta_\rho^{\mathrm{1D}} >  t_{\perp 1}$,
529:  on the other hand,  a direct crossover from the TL liquid
530:   state to the Mott insulator would be seen, and neither the FL 
531:   state nor the crossover region appears in the phase diagram.
532: 
533: %====================================================================
534: \begin{figure}[t]
535: \includegraphics[width=8.5cm]{fig3.eps}
536: \caption{
537: (Color online).
538: (a)Temperature dependence of the resistivity 
539:   for $U/t_{\parallel}=1.5$ and $t_{\perp1}/t_{\parallel}=0.1$,
540:   with fixed $t_{\perp2}/t_\parallel=$ 0, 0.02. 0.04, 0.06, 0.08,
541:   0.10 (from top to bottom), and $\rho_0=\rho(\Lambda)$.
542: $T_\rho$
543:   is the characteristic temperature
544:   of the Mott insulator, at which the resistivity takes a minimum.
545: $T_{\mathrm{FL}}$ is the crossover temperature
546:   to the metallic Fermi liquid, where the power of $\rho(T)$ exhibits
547:   a change (guided by the eyes).
548: (b) Schematic illustration of the $t_{\perp2}$-$T$ phase diagram.
549:  $T_{\mathrm{FL}}^0$ is
550:   the bare temperature scale for imperfect nesting 
551:   [left of Eq.\ (\ref{eq:boundary})], which is reduced to 
552:   $T_{\mathrm{FL}}$ due to the correlation effects.
553: }
554: \label{fig:resis}
555: \end{figure}
556: %======================================================================
557: 
558: 
559: 
560: 
561: 
562: We would like to  discuss here the impact of our results on the understanding 
563: of  the phase diagram of  the quasi-1D molecular compound
564:  (TTM-TTP)I$_3$ under  pressure
565:   \cite{Yasuzuka}. 
566: The extended H\"uckel 
567:   calculations \cite{Mori1994}
568:    indicate that 
569: there are two kinds of interchain
570:    transfer integrals,  namely  $t_{\perp1}\approx 9$ meV, 
571: and $t_{\perp 2}\approx 6$ meV, 
572:  which are  small compared to $t_\parallel\approx 260$ meV.
573:  This  emphasizes the pronounced hopping frustration of this quasi-1D compound.
574: The Coulomb repulsion   between electrons 
575:   on the same molecular orbital of TTM-TTP is weak,
576:   and  the estimated magnitude,  $U \approx$ 0.57 eV \cite{Mori1997},
577:   leads to the magnitude 
578:   $U/t_\parallel \approx 2.2$ 
579:     and then the small 1D Mott gap  
580:   $\Delta_\rho^{\mathrm{1D}}/t_\parallel \approx 0.24$. 
581: From these figures, 
582:   the energy scale for  imperfect nesting 
583:   [left-hand side of Eq.\ (\ref{eq:boundary})] turns out to be 
584:   about 5 meV. As for  the magnitude 
585:   of the charge gap 
586:   at ambient pressure, we obtain 
587:   $\Delta_\rho^{\mathrm{1D}} \approx 60$ meV,  
588:  in fair agreement with  the measured 
589:   activation energy of   resistivity 
590:   at ambient pressure \cite{Mori1997}.
591: The bandwidth goes up under pressure, which decreases  
592:  the ratio  $U/t_\parallel$ and in turn the charge  gap. 
593:   To reduce the gap down to the
594:       scale  of  imperfect nesting,
595:  one has roughly to double   the hopping amplitudes. 
596: This  represents a reasonable increase  of band parameters under 8 GPa
597: of pressure,
598: and is consistent with the existence of a MI transition
599:  in (TTM-TTP)I$_3$  \cite{Yasuzuka}.  
600:   Finally, we note that at ambient pressure, below
601:   $T_{\mathrm{c}}\simeq$ 120 K, this compound develops  a nonmagnetic
602:   state (likely of the spin-Peierls type), which comes with an
603:   \textit{intra-molecular} charge disproportionation  \cite{Mori2004}.
604: However, since the resistivity already shows
605:    an insulating behavior above $T_{\mathrm{c}}$,
606: the present system  can be indeed  considered as  a 
607:    Mott insulator rather than a charge-ordered insulator, and
608: the charge disproportionation 
609: can be accounted for by dimerization of the tilted TTM-TTP molecules.
610: 
611: 
612: 
613: In summary, we have examined the effect of  interchain frustration 
614:    on  the half-filled quasi-1D Hubbard chains
615:   by applying an $N_\perp$-chain two-loop RG method.
616: The triangular lattice  geometry of the system 
617:  is found to be a key factor in the stability of the 
618:   Mott insulating state  and whenever the alteration 
619: of nesting conditions due to frustration 
620:   in the transverse hoppings reaches some threshold 
621: a metallic state is restored.
622: Our results find  direct application to  the  description of 
623:  the frustrated quasi-1D half-filled compounds.  
624: 
625: 
626: One of the authors (M.T.) thanks S.\ Yasuzuka, T.\ Kawamoto,
627:   T.\ Mori, A.\ L\"auchli, T.\ Giamarchi, and C.\ Berthod
628:   for valuable discussions.
629: This work was supported in part by a Grant-in-Aid for
630: Scientific Research on Priority Areas of Molecular Conductors
631: (No.\ 15073213) from the Ministry of Education, Science, Sports, and Culture,
632: Japan.
633: 
634: 
635: 
636: 
637: 
638: 
639: \begin{thebibliography}{10}
640: %--------------------------
641: \bibitem{Imada1998}
642: M.\ Imada, A.\ Fujimori, and Y.\ Tokura,
643: Rev.\ Mod.\ Phys.\ \textbf{70}, 1039 (1998).
644: %--------------------------
645: %--------------------------
646: \bibitem{Kim2006}
647: B.J.\ Kim \textit{et al.},
648: Nature Phys.\ \textbf{2}, 397 (2006).
649: %--------------------------
650: %--------------------------
651: \bibitem{Mori1997}
652: T.\ Mori \textit{et al.},
653: Phys.\ Rev.\ Lett.\ \textbf{79}, 1702 (1997).
654: %--------------------------
655: %--------------------------
656: \bibitem{Mori2004}
657: T.\ Mori, Chem.\ Rev.\ \textbf{104}, 4947 (2004).
658: %--------------------------
659: %--------------------------
660: \bibitem{Yasuzuka}
661: S.\ Yasuzuka \textit{et al.},
662: J.\ Phys.\ Soc.\ Jpn.\ \textbf{75}, 053701 (2006);
663: J.\ Low.\ Temp.\ Phys.\ \textbf{142}, 197 (2006).
664: %--------------------------
665: %--------------------------
666: \bibitem{Dzyaloshinskii72}
667: I.\ E.\ Dzyaloshinskii and A.\ I.\ Larkin, 
668: Sov.\ Phys.\ JETP, \textbf{34}, 422 (1972).
669: %--------------------------
670: %--------------------------
671: \bibitem{Bourbonnais2003}
672: C.\ Bourbonnais, B.\ Guay, and R.\ Wortis,
673: in \textit{Theoretical Methods for Strongly Correlated Electrons}
674: eds.\  D.\ S\'en\'echal, A.\ M.\ Tremblay, and C.\ Bourbonnais
675: (New York, Springer, 2003) p.\ 77.
676: %--------------------------
677: %--------------------------
678: \bibitem{Giamarchi_book}
679: T. Giamarchi,
680: \textit{Quantum Physics in One Dimension}
681: (Oxford University Press, New York, 2004).
682: %--------------------------
683: %--------------------------
684: \bibitem{Essler2002}
685: F.\ H.\ L.\ Essler and A.\ M.\ Tsvelik,
686: Phys.\ Rev.\ B \textbf{65}, 115117 (2002). 
687: %--------------------------
688: %--------------------------
689: \bibitem{Giamarchi2001}
690: S.\ Biermann  \textit{et al.},
691: Phys.\ Rev.\ Lett. \textbf{87}, 276405 (2001); 
692: %
693: T.\ Giamarchi \textit{et al.},
694: J.\ Phys.\ IV France \textbf{114}, 23 (2004);
695: %
696: C.\ Berthod \textit{et al.},
697: Phys.\ Rev.\ Lett.\ \textbf{97}, 136401 (2006). 
698: %--------------------------
699: %--------------------------
700: \bibitem{Tsuchiizu2006}
701: M.\ Tsuchiizu,
702: Phys.\ Rev.\ B \textbf{74}, 155109 (2006).
703: %--------------------------
704: %--------------------------
705: \bibitem{Tsuchiizu2006b}
706: M.\ Tsuchiizu, Y.\ Suzumura, and C.\ Bourbonnais,
707: J.\ Phys.\ Cond.\ Mat.\ \textbf{19}, 145228 (2007).
708: %--------------------------
709: %--------------------------
710: \bibitem{Duprat2001} 
711: R.\ Duprat and C.\ Bourbonnais,
712: Eur.\ Phys.\ J.\ B \textbf{21}, 219 (2001);
713: C.\ Bourbonnais and R.\ Duprat,
714: J.\ Phys.\ IV France \textbf{114}, 3 (2004).
715: %--------------------------
716: %--------------------------
717: \bibitem{Doucot2003} 
718: S. Dusuel and B.\ Dou\c{c}ot,
719: Phys.\ Rev.\ B \textbf{67}, 205111(2003).
720: %--------------------------
721: %--------------------------
722: \bibitem{Rohe06}
723: D.\ Rohe and A.\ Georges, cond-mat/0608032.
724: %--------------------------
725: %--------------------------
726: \bibitem{Imada_Kyung}
727: T.\ Kashima and M.\ Imada,
728: J.\ Phys.\ Soc.\ Jpn.\ \textbf{70}, 3052 (2001);
729: B.\ Kyung and A.\ -M.\ S.\ Tremblay, 
730: Phys.\ Rev.\ Lett.\ \textbf{97}, 046402 (2006).
731: %--------------------------
732: %--------------------------
733: \bibitem{Giamarchi1991}
734: T.\ Giamarchi,
735: Phys.\ Rev.\ B \textbf{44}, 2905 (1991).
736: %--------------------------
737: %--------------------------
738: \bibitem{Mori1994}
739: T.\ Mori \textit{et al.},
740: Bull.\ Chem.\ Soc.\ Jpn.\ \textbf{67}, 661 (1994).
741: %--------------------------
742: 
743: \end{thebibliography}
744: 
745: \end{document}
746: 
747: 
748: 
749: 
750: 
751: 
752: 
753: