1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2: \documentclass[twocolumn,showpacs,preprintnumbers,prb,fleqn]{revtex4}
3: \usepackage{epsfig}
4: \usepackage{graphicx}
5: \usepackage{amsfonts}
6: %\usepackage[dvips]{color}
7: \begin{document}
8:
9: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
10:
11: \title{Hole and electron dynamics in the triangular-lattice antiferromagnet --- \\
12: interplay of frustration and spin fluctuations}
13: \author{Pooja Srivastava}
14: \email{psriv@iitk.ac.in}
15: \author{Avinash Singh}
16: \email{avinas@iitk.ac.in}
17: \affiliation{Department of Physics, Indian Institute of Technology Kanpur - 208016}
18: \date{\today}
19: \begin{abstract}
20: Single-particle dynamics in the 120$^{\circ}$ ordered antiferromagnetic state
21: of the triangular-lattice Hubbard model is studied using a physically transparent
22: fluctuation approach in terms of multiple magnon emission and absorption processes
23: within the noncrossing approximation.
24: Hole and electron spectral features are evaluated at intermediate $U$,
25: and analyzed in terms of a competition between the frustration-induced direct hopping
26: and the virtual hopping.
27: Finite $U$-induced competing interactions and frustration effects contributing through
28: the magnon dispersion are also discussed.
29: Finite contribution to self-energy correction from long-wavelength (Goldstone) modes,
30: together with the high density of electron scattering states in the narrow, sharp peak
31: in the upper band, result in strong fermion-magnon scattering
32: leading to pronounced incoherent behaviour in the electron dynamics.
33: The fluctuation-induced first-order metal-insulator transition
34: due to vanishing band gap is also discussed.
35: \end{abstract}
36: \pacs{71.10.Fd,75.50.Ee,75.30.Ds,75.10.Lp}
37: \maketitle
38:
39: \section{Introduction}
40: There has been renewed interest in correlated electron systems on triangular lattices,
41: as evidenced by recent studies of antiferromagnetism, superconductivity
42: and metal-insulator transition in the organic systems
43: $\rm \kappa -(BEDT-TTF)_2 X$,\cite{review1,review2}
44: the discovery of superconductivity in $\rm Na_x Co O_2 . y H_2 O$,\cite{watersup}
45: the observation of low-temperature insulating phases
46: in some $\sqrt{3}$-adlayer structures such as K on Si[111],\cite{weitering}
47: and quasi two-dimensional $120^\circ$ spin ordering and spin-wave excitations
48: in $\rm Rb Fe (MoO_4)_2$ (Refs. 5,6)
49: and the multiferroic material $\rm Ho Mn O_3$.\cite{holmium,sw}
50:
51: The recent finding of finite $U$-induced competing interactions and frustration
52: in the $120^\circ$ ordered antiferromagnetic (AF) state of the Hubbard model on a triangular lattice,\cite{tri} resulting in vanishing spin stiffness at $U \approx 6$
53: and a magnetic instability towards a F-AF state at $U \approx 7$,
54: both in the insulating state, adds a new dimension to the intrinsic
55: geometric frustration of the triangular-lattice antiferromagnet.
56: Indeed, strongly enhanced quantum spin fluctuations associated with the magnetic instability may account for
57: the absence of long-range magnetic order in the {\em nearly isotropic}
58: organic antiferromagnet $\rm \kappa -(BEDT-TTF)_2 Cu_2 (CN)_3$,
59: as inferred from recent $^1$H NMR and static susceptibility measurements down to 32 mK,
60: well below the estimated exchange constant $J \sim 250$ K,
61: suggesting the realization of a quantum spin-liquid state.\cite{kanoda}
62: The realization of a non-magnetic insulator at intermediate $U$ is also interesting
63: as it allows, with decreasing $U$, for a Mott-type metal-insulator transition
64: not accompanied by any magnetic symmetry breaking,
65: as seen in the layered system $\rm \kappa -(BEDT-TTF)_2 Cu[N(CN)_2]Cl$,\cite{kagawa}
66: and currently of much theoretical interest.\cite{pirg2,pirg3}
67:
68: Recently,
69: self-energy corrections due to multiple magnon processes
70: in the AF state of the frustrated square-lattice $t-t'$-Hubbard model
71: were evaluated using a fluctuation approach.\cite{self}
72: Quasiparticle dispersion obtained with the same set of Hubbard model cuprate parameters
73: as obtained from a recent magnon spectrum fit\cite{spectrum}
74: was found to yield excellent agreement with ARPES data for
75: $\rm Sr_2 Cu O_2 Cl_2$,\cite{self} thus providing a unified description
76: of magnetic and electronic excitations in cuprates.
77:
78: It is therefore interesting to examine and contrast self-energy corrections
79: and quasiparticle behaviour in the $120^\circ$ ordered AF state of the triangular-lattice Hubbard model which involves non-collinear ordering, intrinsic geometric frustration,
80: and also finite $U$-induced competing interactions and frustration.
81: Indeed, we find that long-wavelength magnon modes yield
82: finite contribution to the fermion-magnon scattering process,
83: unlike the square-lattice case where this contribution was
84: negligible.\cite{self}
85:
86: Frustration and spin fluctuations are involved in an interesting interplay
87: with respect to quasiparticle behaviour.
88: As neighbouring spins are not antiparallel in
89: triangular-lattice and frustrated square-lattice AF states,
90: frustration results in an O($t$) or O($t'$) direct hopping in addition
91: to the O($J$) virtual hopping.
92: Competition between the two dispersions results in band broadening/narrowing,
93: which has a dramatic effect on self-energy corrections
94: due to significantly different density of fermion scattering states.
95: Competition also results in a reduced band gap,
96: thus bringing the system closer to a metal-insulator (MI) transition.
97: The $t-J$ model calculations do not involve this competition
98: as the virtual hopping term is absent.
99:
100: In this paper, we examine the hole and electron spectral functions in terms of
101: self-energy corrections in the $120^\circ$ ordered AF state of the Hubbard model
102: on a triangular lattice.
103: Hole and electron dynamics in an antiferromagnetic background is associated with
104: multiple magnon emission and absorption processes corresponding to
105: broken AF bonds, string states, and scrambling of the AF spin ordering.
106: The fluctuation approach adopted in this paper in terms of a
107: diagrammatical expansion provides a physically transparent picture
108: in which the hole motion is renormalized due to the fluctuating transverse
109: field associated with spin fluctuations.
110: This fluctuation perspective is physically relevant
111: in view of the substantial ordered-moment reduction,
112: which approaches $59\%$ in the strong-coupling (Heisenberg) limit.\cite{capriolti}
113:
114: Single-particle dynamics in the AF state of the triangular-lattice $t-J$ model
115: has been studied in the self-consistent Born approximation
116: (SCBA),\cite{dombre,trumper,apel} exact diagonalization,\cite{dombre}
117: and projection techniques.\cite{vojta,dagotto}
118: One-electron density of states
119: has been examined using the quantum Monte Carlo method,\cite{qmc_green}
120: showing a pseudogap development for intermediate $U$.
121:
122: We consider the Hubbard model
123: \begin{equation}
124: {\cal H} = -t \sum_{i\delta\sigma} a_{i,\sigma} ^\dagger a_{i+\delta,\sigma}
125: + U \sum_i a_{i\uparrow} ^\dagger a_{i\uparrow} a_{i\downarrow} ^\dagger a_{i\downarrow}
126: \end{equation}
127: with nearest-neighbour (NN) hopping on a triangular lattice.
128: The model has particle-hole symmetry under the transformation $t \rightarrow -t$.
129: In this paper we consider the case of positive $t$ and hole (electron)
130: doping in the lower (upper) band;
131: same result holds for negative $t$ and electron (hole) doping in the upper (lower) band.
132: In the following we set the hopping energy $t=1$.
133:
134: The organization of this paper is as follows.
135: The three-sublattice basis is briefly reviewed in section II
136: to introduce the notation and key features of
137: the classical-level fermion dispersion.
138: Transverse spin fluctuations are introduced in section III
139: in terms of magnon amplitudes and energies.
140: Intraband self-energy corrections due to multi-magnon processes
141: are studied in section IV in the non-crossing approximation.
142: Hole and electron spectral functions for the lower and upper bands
143: are discussed in section V and conclusions are presented in section VI.
144:
145: \section{Three-sublattice representation}
146: While the spiral-state description applies only to Bravais lattices,
147: the sublattice-basis description applies to Kagom\'{e} type non-Bravais
148: lattices as well. In general, the $120^\circ$ AF state is characterized by an
149: ordering plane (normal $\hat{n}_1$) and a planar direction ($\hat{n}_2$) in
150: spin space, with reference to which spin orientations are given by
151: \begin{equation}
152: \hat{\alpha} = \cos\phi_\alpha \; \hat{n}_2
153: + \sin\phi_\alpha (\hat{n}_1 \times \hat{n}_2 )
154: \end{equation}
155: corresponding to angles $\phi_\alpha=0^\circ$, $120^\circ$, and $240^\circ$
156: on the three sublattices $\alpha=\rm A,B,C$.
157: A convenient choice is $\hat{n}_1=\hat{z}$
158: (spin-ordering in the $x-y$ plane) and $\hat{n}_2=\hat{x}$,
159: so that the local mean field ${\bf \Delta}_\alpha = \frac{1}{2}
160: U\langle \Psi_i ^\dagger {\mbox{\boldmath $\sigma$}} \Psi_i \rangle_\alpha $
161: in the $120^\circ$ ordered Hartree-Fock state is given by
162: \begin{equation}
163: {\bf \Delta}_\alpha = \Delta \hat{\alpha} \;\;\;\;\; (\hat{\alpha}=\hat{a},\hat{b},\hat{c})
164: \end{equation}
165: on the three sublattices
166: in terms of the three lattice unit vectors
167: \begin{equation}
168: \hat{a} = \hat{x}, \;\;\;\;
169: \hat{b} = -\frac{1}{2}\hat{x} + \frac{\sqrt{3}}{2} \hat{y},
170: \;\;\;\;\;
171: \hat{c} = -\frac{1}{2}\hat{x} - \frac{\sqrt{3}}{2} \hat{y} .
172: \end{equation}
173:
174: %fig1
175: \begin{figure}
176: \vspace*{-22mm}
177: \hspace*{-20mm}
178: \psfig{figure=2triBZ.ps,width=100mm,angle=-90}
179: \vspace{-20mm}
180: \caption{Brillouin zone (BZ) for the triangular lattice
181: and magnetic Brillouin zone (MBZ) for the $120^\circ$ ordered AF state,
182: along with the BZ - MBZ correspondence involving the branch index $l$.}
183: \end{figure}
184:
185: Fourier transformation within the sublattice basis yields
186: \begin{equation}
187: {\cal H}_{\rm HF} = \sum_{\bf k} \Psi_{\bf k} ^\dagger
188: \left [ \begin{array}{ccc}
189: -{\mbox{\boldmath $\sigma$}}.{\bf \Delta}_{\rm A} & \delta_{\bf k} &
190: \delta_{\bf k}^* \\
191: & & \\
192: \delta_{\bf k}^* & - {\mbox{\boldmath $\sigma$}}.{\bf \Delta}_{\rm B} &
193: \delta_{\bf k} \\
194: & & \\
195: \delta_{\bf k} & \delta_{\bf k}^* &
196: -{\mbox{\boldmath $\sigma$}}.{\bf \Delta}_{\rm C}
197: \end{array}
198: \right ]\Psi_{\bf k} \; ,
199: \end{equation}
200: where $\Psi_{\bf k} \equiv
201: (a_{{\bf k}\uparrow}\;a_{{\bf k}\downarrow}
202: \; b_{{\bf k}\uparrow}\;b_{{\bf k}\downarrow}
203: \; c_{{\bf k}\uparrow}\;c_{{\bf k}\downarrow} )$ in terms of the fermion operators
204: $a_{\bf k},b_{\bf k},c_{\bf k}$ defined on the three sublattices A, B, C.
205: Wavevector $\bf k$ lies within the Magnetic Brillouin Zone (MBZ),
206: corresponding to the three inter-penetrating triangular sublattices (lattice parameter $\sqrt{3}a$).
207: The NN hopping term
208: \begin{equation}
209: \delta_{\bf k} = -t \sum_{\hat{\delta}=\hat{a},\hat{b},\hat{c}}
210: e^{i{\bf k}.\hat{\delta} }
211: = -t [e^{i k_x} + 2e^{-ik_x/2} \cos (\sqrt{3} k_y/2) ]
212: \end{equation}
213: mixes AB, BC, and CA sublattices, which are connected by the three lattice
214: unit vectors. The lattice hopping term $\delta_{\bf k}$ yields the
215: triangular-lattice free-fermion energy
216: \begin{equation}
217: \epsilon_{\bf k} = \delta_{\bf k} + \delta_{\bf k} ^\ast
218: \end{equation}
219: and transforms as
220: \begin{equation}
221: \delta_{\bf k\pm Q} =
222: \delta_{\bf k} \, e^{\pm i 2\pi/3}
223: \end{equation}
224: under momentum translation by the ordering wavevector
225: ${\bf Q}=(2\pi/3,2\pi/\sqrt{3})$.
226:
227: The $[6\times 6]$ Hamiltonian matrix obeys the cyclic property
228: $[{\cal H}_{\rm HF}]_{AB} = [{\cal H}_{\rm HF}]_{BC} =
229: [{\cal H}_{\rm HF}]_{CA}$ of the $120^\circ$ ordered state,
230: resulting in the following spin-sublattice structure of the normalized eigenvectors
231: \begin{equation}
232: |{\bf k},l \rangle = \frac{1}{\sqrt{3}} \left ( \begin{array}{l}
233: \alpha_{{\bf k},l} \; e^{-i\phi_\alpha} \\
234: \beta_{{\bf k},l} \; \end{array} \right )_\sigma
235: \otimes \left (
236: \begin{array}{c} 1 \\ e^{i\theta_l} \\ e^{-i\theta_l} \end{array} \right )_\alpha \; ,
237: \end{equation}
238: where the planar spin orientations $\phi_\alpha = 0^\circ,\; 120^\circ,\; 240^\circ$
239: for the three sublattices,
240: and the sublattice phase angle $\theta_l = 2\pi l/3 $ for the three
241: fermion branches $l=0,\pm 1$.
242: Substituting the above eigenvector structure,
243: and contracting over the sublattice sector $\alpha$,
244: the eigenvalue equations of the $[6\times 6]$ Hamiltonian matrix
245: reduce to three $[2\times 2]$ equations in the spin sector corresponding to $l=0,\pm 1$;
246: the six eigenvalues $E_{{\bf k},l}^\mp$ and amplitudes
247: $(\alpha_{{\bf k},l}^\mp \; \beta_{{\bf k},l}^\mp )$
248: for the lower ($-$) and upper ($+$) AF bands are then given by
249: \begin{eqnarray}
250: E_{{\bf k},l}^\mp &=& \eta_{\bf k} \pm \sqrt{\Delta^2 + \xi_{\bf k}^2} \\
251: \alpha_{{\bf k},l}^\mp &=& \pm \frac{1}{\sqrt{2}}\left (1 \pm
252: \frac{\xi_{\bf k}}{\sqrt{\Delta^2 + \xi_{\bf k}^2}} \right )^{1/2} \nonumber \\
253: \beta_{{\bf k},l}^\mp &=& + \frac{1}{\sqrt{2}}\left (1 \mp \frac{\xi_{\bf k}}{\sqrt{\Delta^2 + \xi_{\bf k}^2}} \right )^{1/2} \; ,
254: \end{eqnarray}
255: where
256: $\eta_{\bf k} \equiv (\epsilon_{\bf k} + \epsilon_{\bf k-Q})/2 $ and
257: $\xi_{\bf k} \equiv (\epsilon_{\bf k} - \epsilon_{\bf k-Q})/2$,
258: with momentum values $\bf k$, $\bf k \pm Q$
259: corresponding to fermion branches $l=0,\pm 1$, respectively.
260: The mean field $\Delta$ and magnetization $m$ are related to $U$
261: through the self-consistency condition.\cite{tri}
262:
263: %fig2
264: \begin{figure}
265: \vspace*{-68mm}
266: \hspace*{-35mm}
267: \psfig{figure=compar.ps,width=140mm}
268: \vspace{-82mm}
269: \caption{(color online) Comparison of density of states corresponding to the
270: full HF dispersion and the direct hopping dispersion,
271: showing the band broadening (narrowing) due to the virtual hopping dispersion.}
272: \end{figure}
273:
274: Typical of the AF state,
275: upper band states yield a negative contribution to spin densities
276: due to the negative sign of $\alpha_{\bf k} ^+$.
277: Consequently, when the two bands start overlapping on decreasing $U$,
278: partial occupation of the upper band has an amplified effect on reduction of
279: magnetic order ---
280: reduced sublattice magnetization due to band overlap decreases the mean-field $\Delta$, which further increases the overlap. Typically, the magnetic order
281: therefore falls very rapidly after band overlap sets in.
282:
283: The density of states (DOS) corresponding to the full HF (classical) dispersion in Eq. (10),
284: shown in Fig. 2 for both positive and negative $t$,
285: exhibits a key competition between the two dispersion terms associated with
286: direct hopping $(\eta_{\bf k})$ of order $t$
287: and virtual hopping $(\sqrt{\Delta^2 + \xi^2 _{\bf k}} - \Delta)$,
288: which is of order $J$ in the strong-coupling limit.
289: For positive $t$, both dispersion terms favour same state at the top of the lower band,
290: while they favour different states at the bottom of the upper band,
291: thereby causing band narrowing.
292: The competition results in broadening and narrowing of the
293: lower and upper bands, depending on the sign of $t$,
294: which has a dramatic effect on the self-energy correction
295: due to significantly different density of fermion scattering states.
296: There is no such competition in the $t-J$ model studies where
297: the virtual hopping dispersion term is absent.
298:
299: From Eq. (8), it follows that
300: \begin{equation}
301: \epsilon_{\bf k} = -(\epsilon_{\bf k+Q} + \epsilon_{\bf k-Q})
302: \end{equation}
303: so that
304: \begin{equation}
305: \eta_{\bf k-Q} = -\frac{\epsilon_{\bf k}}{2} \; .
306: \end{equation}
307: Therefore, the direct hopping dispersion $\eta_{\bf k-Q}$,
308: to which the HF quasiparticle dispersion (10)
309: reduces in the strong-coupling limit,
310: is identical to the classical-level dispersion $\epsilon_{\bf k}/2$ taken in
311: earlier $t-J$ model studies, corresponding to effective hopping
312: $t\cos (120^\circ) = -t/2$ associated with the $120^\circ$ ordering,
313: where momentum translation by ${\bf Q}$ connects
314: the real and slave fermions.\cite{dombre}
315: Comparison of DOS corresponding to the full dispersion with that of $\eta_{\bf k}$
316: is also shown in Fig. 2.
317:
318: \section{Transverse spin fluctuations}
319: Including both transverse and longitudinal spin fluctuations,
320: the full spin-fluctuation propagator $\langle \Psi_{\rm G} |{\rm T}[
321: S_i ^\mu (t) S_j ^\nu (t')]\Psi_{\rm G} \rangle$
322: in the $120^\circ$ ordered AF state
323: has been studied recently in the random phase approximation (RPA)
324: in the full $U$ range.\cite{tri}
325: Even in the intermediate-coupling regime,
326: the magnitude of longitudinal fluctuation $\langle S_\alpha ^2 \rangle$
327: along the local ordering directions $\hat{\alpha}$ was found to be quite negligible
328: ($\langle S_{\alpha} ^2 \rangle \sim 10^{-4}$ at $U \approx 7$).
329: We therefore focus on transverse
330: spin fluctuations along the two locally normal spin directions.
331:
332: Spin rotation about the $\hat{z}$ direction
333: \begin{equation}
334: \left ( \begin{array}{c} \sigma_x \\ \sigma_y \\ \sigma_z \end{array} \right )'
335: = \left [ \begin {array}{ccc} \cos \phi_\alpha & \sin \phi_\alpha & 0 \\
336: -\sin \phi_\alpha & \cos\phi_\alpha & 0 \\
337: 0 & 0 & 1 \end{array} \right ]
338: \left ( \begin{array}{c} \sigma_x \\ \sigma_y \\ \sigma_z \end{array} \right )
339: \end{equation}
340: by angles
341: $\phi_\alpha=0^\circ,120^\circ,-120^\circ$ for $\alpha$ = A,B,C
342: renders $x'$ as the spin ordering direction for all three sublattices.
343: In the $2\otimes 3$ spin-sublattice basis of the two transverse spin directions
344: $\mu,\nu=y',z'$ and the three sublattices $\alpha,\beta=A,B,C$,
345: the RPA-level spin-fluctuation propagator is then given by
346: \begin{equation}
347: [\chi ({\bf q},\omega)]_{\alpha\beta} ^{\mu\nu} = \frac{\frac{1}{2}\;[\chi^0 ({\bf q},\omega)]}
348: {{\bf 1} - U[\chi^0 ({\bf q},\omega)]} \; ,
349: \end{equation}
350: where the bare particle-hole propagator
351: \begin{eqnarray}
352: [\chi^0({\bf q},\omega)]^{\mu\nu} _{\alpha \beta} =&& \frac{1}{2}
353: \sum_{{\bf k},l,m} \left [
354: \frac{\langle \sigma_\mu \rangle_\alpha ^{-+}
355: \langle \sigma_\nu \rangle_\beta^{-+*}}
356: {E_{{\bf k-q},m}^+ - E_{{\bf k},l}^- + \omega}
357: \right. \nonumber \\
358: && \left. \; \; \;\; \; \;\; \; \;\; \; \; \;
359: +\frac{\langle \sigma_\mu \rangle_\alpha ^{+-}
360: \langle \sigma_\nu \rangle_\beta ^{+-*}}
361: {E_{{\bf k},m}^+ - E_{{\bf k-q},l}^- - \omega} \right ]
362: \end{eqnarray}
363: involves integrating out the fermions in the broken-symmetry state.
364: In the particle-hole matrix elements of the rotated spins
365: \begin{equation}
366: \langle \sigma_\mu \rangle_\alpha ^{-+} \equiv
367: \langle {\bf k-q},m| \sigma_\mu |{\bf k},l\rangle_\alpha
368: \end{equation}
369: the spin orientation angles $\phi_\alpha$ in the fermion states (Eq. 9)
370: are transformed out.
371:
372: We now discuss the spin-sublattice structure of the $[\chi^0({\bf q},\omega)]$ matrix
373: and its eigenvectors.
374: While the spin-diagonal blocks $[\chi^0({\bf q},\omega)]^{\mu\mu}$ are Hermitian
375: and the off-diagonal blocks $[\chi^0({\bf q},\omega)]^{\mu \bar{\mu}}$
376: are anti-Hermitian, they obey the cyclic symmetry
377: $[\chi^0({\bf q},\omega)]_{AB} = [\chi^0({\bf q},\omega)]_{BC} =
378: [\chi^0({\bf q},\omega)]_{CA}$ of the $120^\circ$ ordered phase.
379: Also, the sublattice diagonal elements $[\chi^0({\bf q},\omega)]_{\alpha\alpha}$
380: are all identical due to sublattice symmetry.
381: Consequently, the normalized eigenvectors of $[\chi^0({\bf q},\omega)]$
382: have the following spin-sublattice structure
383: \begin{equation}
384: |\phi _\lambda \rangle = \frac{1}{\sqrt{3}}
385: \left ( \begin{array}{r} -iu \\ v \end{array} \right )_\mu
386: \otimes \left (
387: \begin{array}{c} 1 \\ e^{i\,2\pi \lambda /3 } \\ e^{-i\,2\pi \lambda /3} \end{array}
388: \right )_\alpha
389: \end{equation}
390: where $\lambda =0,\pm 1$ for the three magnon branches,
391: and the real and normalized amplitudes $u$ and $v$ represent the fluctuation
392: amplitudes in the $y'$ and $z'$ directions, respectively.
393: Contracting over the sublattice index,
394: the eigenvalue equation for $| \phi \rangle$ therefore reduces to
395: \begin{equation}
396: [\chi^0 ({\bf q},\omega)] | \phi \rangle_\lambda =
397: [\chi^0 _\lambda({\bf q},\omega)] \left ( \begin{array}{c} u \\ v \end{array} \right )
398: = \lambda_{\bf q} \left ( \begin{array}{c} u \\ v \end{array} \right ) \; ,
399: \end{equation}
400: where $[\chi^0 _\lambda ({\bf q},\omega)]$ is a $[2\times 2]$ real-symmetric matrix.
401:
402: Solving the pole equation $1-U\lambda_{\bf q}(\omega)=0$
403: for the magnon energy $\omega_{\bf q}$,
404: and expanding $\lambda_{\bf q}(\omega)$ around the poles to obtain the magnon amplitudes,
405: yields the magnon propagator
406: \begin{equation}
407: [\chi ({\bf q},\omega)] = \sum_{\lambda=0,\pm 1} \left [
408: \frac{| {\bf q},\lambda \rangle \langle {\bf q},\lambda |_A }
409: {\omega + \omega_{{\bf q},\lambda} - i\eta}
410: -
411: \frac{| {\bf q},\lambda \rangle \langle {\bf q},\lambda |_R }
412: {\omega - \omega_{{\bf q},\lambda} + i\eta} \right ] \; ,
413: \end{equation}
414: where the magnon eigenvectors for the advanced (A) and retarded (R) modes are given by
415: \begin{equation}
416: |{\bf q},\lambda \rangle = \frac{1}{\sqrt{3}}
417: \left ( \begin{array}{r} \mp i {\cal Y}_{{\bf q},\lambda} \\
418: {\cal Z}_{{\bf q},\lambda} \end{array} \right )_\mu
419: \otimes \left (
420: \begin{array}{c} 1 \\ e^{i\,2\pi \lambda /3 } \\ e^{-i\,2\pi \lambda /3} \end{array}
421: \right )_\alpha
422: \end{equation}
423: in terms of the magnon amplitudes
424: \begin{eqnarray}
425: &&{\cal Y}_{{\bf q},\lambda} = u_{{\bf q},\lambda}
426: /\sqrt{2U^2 |d\lambda_{{\bf q},\lambda}/d\omega| } \;\;\;\;\;\;\;\;\;\;\;\;
427: {\rm and} \nonumber \\
428: &&{\cal Z}_{{\bf q},\lambda} = v_{{\bf q},\lambda}
429: /\sqrt{2U^2 |d\lambda_{{\bf q},\lambda}/d\omega| }
430: \end{eqnarray}
431: in the $y'$ and $z'$ directions.
432: Expressions for magnon energy and amplitudes in the strong-coupling limit
433: are given in the Appendix.
434:
435: \ \\
436: \ \\
437:
438: \section{Self-energy correction}
439: Due to multiple magnon emission and absorption processes
440: associated with fermion motion in the AF state,
441: the fermion self energy matrix in the spin-sublattice basis is obtained as
442: \begin{eqnarray}
443: & & [\Sigma({\bf k},l,\omega)]_{\alpha\beta} \nonumber \\
444: &=&
445: U^2 \int\frac{d\Omega}{2\pi i} \sum_{\mu\nu} [\sigma_\mu] [G_{\alpha\beta}
446: ({\bf k-q},m,\omega-\Omega)]
447: [\sigma_\nu] [ \chi ({\bf q},\Omega)]_{\alpha\beta}^{\mu\nu}\nonumber \\
448: \end{eqnarray}
449: in the non-crossing (rainbow) approximation.
450:
451: %fig3
452: \begin{figure}
453: \vspace*{-25mm}
454: \hspace*{-27mm}
455: \psfig{figure=scortri.ps,angle=-90,width=130mm}
456: \vspace{-30mm}
457: \caption{Self energy in the noncrossing approximation.
458: Wavy lines represent the magnon propagator.
459: The bare fermion-magnon interaction vertex is $U$.}
460: \end{figure}
461:
462: We consider the intraband contribution involving hole (electron) scattering states
463: in the lower (upper) band, which redistributes the spectral function,
464: leaving the integrated spectral weight and the sublattice magnetization unchanged.
465: We obtain for the hole self energy
466: \begin{eqnarray}
467: \Sigma_{{\bf k},l} (\omega) &\equiv & \langle {\bf k},l| [\Sigma({\bf k},l,\omega)]
468: | {\bf k},l \rangle \nonumber \\
469: & = & \sum_{{\bf q},m,\lambda} \frac{|M|^2}
470: {\omega+\omega_{{\bf q},\lambda} - E_{{\bf k-q},m} ^- -
471: \Sigma_{{\bf k-q},m} (\omega+\omega_{{\bf q},\lambda} ) } \nonumber \\
472: \end{eqnarray}
473: where the fermion-magnon scattering matrix element
474: \begin{equation}
475: M = U \sum_{\alpha\mu} \langle {\bf k},l| \sigma_\mu | {\bf k-q},m \rangle_\alpha
476: \otimes | {\bf q},\lambda \rangle_{\alpha\mu} ^{\rm A}
477: \end{equation}
478: involves the advanced magnon mode.
479: Substituting the sublattice structure of the fermion and magnon amplitudes,
480: the sum over sublattice index $\alpha$ yields
481: \begin{equation}
482: 1 + e^{i(m+\lambda -l)2\pi/3} + e^{-i(m+ \lambda -l)2\pi/3} = 3\,
483: \delta_{m+\lambda -l}
484: \end{equation}
485: effectively amounting to a conservation of sublattice polarization
486: at the fermion-magnon interaction vertex.
487: Therefore, the fermion-magnon scattering matrix element
488: reduces to a sum of the $y'$ and $z'$ fluctuation terms
489: \begin{eqnarray}
490: M = \frac{U}{\sqrt{3}}
491: [&-&(\alpha_{{\bf k},l}^- \beta_{{\bf k-q},m}^- -
492: \beta_{{\bf k},l}^- \alpha_{{\bf k-q},m}^- )
493: \, {\cal Y}_{{\bf q},\lambda} \nonumber \\
494: &+& (\alpha_{{\bf k},l}^- \alpha_{{\bf k-q},m}^- -
495: \beta_{{\bf k},l}^- \beta_{{\bf k-q},m}^- )
496: \, {\cal Z}_{{\bf q},\lambda} ] \; .
497: \end{eqnarray}
498:
499: For fermion and magnon states in the matrix element $M$,
500: the sublattice-basis MBZ description of momentum
501: translates to a BZ description according to the correspondence
502: ${\bf k},l \rightarrow {\bf k}+ l{\bf Q}$ and
503: ${\bf q},\lambda \rightarrow {\bf q}+ \lambda {\bf Q}$.
504: With this equivalence, Eq. (26) simply corresponds to momentum conservation in the BZ.
505: Analysis of the fermion-magnon matrix element $M$ in the strong-coupling limit
506: and comparison with earlier results for the $t-J$ model is discussed
507: in the Appendix.
508: %4
509: \begin{figure}
510: \vspace*{-50mm}
511: \hspace*{-70mm}
512: \psfig{figure=iteration.ps,width=170mm,angle=-90}
513: \vspace{-55mm}
514: \caption{Variation of hole self energy with iterations, effectively
515: illustrating the role of successively higher-order magnon processes.
516: Here $\Delta=4$ $(U=8.8)$ and ${\bf k}=0,\; l=-1$.}
517: \end{figure}
518:
519: It is important to note here that long-wavelength magnon modes yield
520: finite contribution to the fermion-magnon scattering process
521: in the triangular-lattice AF, unlike the square-lattice case
522: where this contribution is negligible.\cite{self}
523: For the square-lattice AF,
524: the small-$q$ contribution was suppressed because the fermion-magnon
525: matrix element $M^2 \sim q$ due to destructive interference
526: within sublattice summation.
527: For the triangular lattice also,
528: for $q=0$ and $\lambda=0$ (in-plane mode),
529: the fermion matrix element in Eq. (25) reduces to an expectation value
530: which identically vanishes for $\mu = y'$ as the spins are oriented in the $x'$ direction,
531: yielding similar $M^2 \sim q$ behaviour for small $q$.
532: However, for the out-of-plane $z'$ fluctuation modes $(\lambda=\pm 1)$,
533: the fermion matrix element is finite, resulting in $M^2 \sim 1/q$
534: and a finite contribution of long-wavelength modes within the
535: two-dimensional $(\int q dq )$ momentum summation.
536: %5
537: \begin{figure}
538: \vspace*{-70mm}
539: \hspace*{-35mm}
540: \psfig{figure=dismean.ps,width=140mm}
541: \vspace{-75mm}
542: %
543: \vspace*{-77mm}
544: \hspace*{-35mm}
545: \psfig{figure=dislower.ps,width=140mm}
546: \vspace{-84mm}
547: \caption{(color online) Quasiparticle dispersion $E_{\bf k}^\ast$
548: along different symmetry directions in the MBZ (lower panel),
549: along with the HF dispersion $E_{\bf k}^0$ for comparison (upper panel).}
550: \end{figure}
551:
552: \section{Results and Discussion}
553: The self-consistent numerical evaluation of the self energy (24)
554: was carried out on a $30\times 40$ grid in the MBZ ${\bf k}$ space
555: and a frequency interval $\Delta \omega =0.025$
556: for $\omega$ in the range $-15 < \omega < 10$.
557: The self energy was iteratively evaluated,
558: starting with $\Sigma_{\bf k}(\omega) = 0$.
559: Typically, self-consistency was achieved within ten iterations
560: for the lower band (Fig. 4) and fifteen iterations for the upper band (Fig. 8).
561:
562: Self-energy corrections for an added hole in the
563: broad, flat band near the top of the lower band are qualitatively different
564: from that of an added electron in the narrow, sharp peak near the bottom of the
565: upper band. The low density of hole scattering states and the dominant
566: band-energy denominator suppresses the hole self energy.
567: However, the electron self energy is significantly enhanced due to the
568: sharp peak and the small band energy compared with magnon energy,
569: resulting in the characteristic signature of string states associated
570: with multi-magnon processes.
571: It is convenient to visualize these qualitatively different self-energy corrections
572: in terms of hole (electron) motion in an effective ferromagnetic (antiferromagnetic)
573: spin background projected out of the $120^\circ$ spin ordering.
574: %6
575: \begin{figure}
576: \vspace*{-70mm}
577: \hspace*{-35mm}
578: \psfig{figure=modeT.ps,width=140mm}
579: \vspace{-75mm}
580: %
581: \vspace*{-77mm}
582: \hspace*{-35mm}
583: \psfig{figure=modeK.ps,width=140mm}
584: \vspace{-75mm}
585: %
586: \vspace*{-77mm}
587: \hspace*{-35mm}
588: \psfig{figure=modeM.ps,width=140mm}
589: \vspace{-84mm}
590: \caption{(color online) Hole spectral function $A_{\bf k}(\omega)$ at the special MBZ points
591: $\Gamma$, K, and M. Here $\Delta=4$.}
592: \end{figure}
593:
594: \subsection{Hole Dynamics (Lower Band)}
595: Figure 5 shows renormalized quasiparticle dispersion along different symmetry directions
596: in the MBZ. Comparison with the HF result shows
597: nearly momentum independent shift in the quasiparticle energies,
598: leaving the effective hole mass essentially unchanged.
599: States in upper portion of the band are pushed up,
600: while those in the lower portion are pulled down,
601: in accordance with the formally second-order structure of the self-energy correction.
602: The quasiparticle (hole) energy is maximum (minimum) at the $\Gamma$ point
603: ${\bf k}=(0,0)$ for $l=-1$, corresponding to BZ momentum $-{\bf Q}$.
604: %7
605: \begin{figure}
606: \vspace*{-83mm}
607: \hspace*{-48mm}
608: \psfig{figure=tp3dtk.ps,width=150mm}
609: \vspace{-88mm}
610: %
611: \vspace*{-92mm}
612: \hspace*{-48mm}
613: \psfig{figure=tp3dmt.ps,width=150mm}
614: \vspace{-95mm}
615: \caption{(color online) Hole spectral function $A_{\bf k}(\omega)$ along the $\Gamma-K$
616: and $\Gamma-M$ directions for the lowest-energy branch $l=-1$.}
617: \end{figure}
618:
619: %8
620: \begin{figure}
621: \vspace*{-30mm}
622: \hspace*{-64mm}
623: \psfig{figure=tniter.ps,width=160mm,angle=-90}
624: \vspace{-33mm}
625: \caption{Variation of electron self energy with iterations for the upper band,
626: effectively illustrating the role of successively higher-order magnon processes.
627: Here $\Delta=4$ and ${\bf k}=(-\frac{2\pi}{3},0)$, $l=0$.}
628: \end{figure}
629:
630: %9
631: \begin{figure}
632: \vspace*{-70mm}
633: \hspace*{-38mm}
634: \psfig{figure=tndismean.ps,width=135mm}
635: \vspace{-75mm}
636: %\caption{.}
637: %\end{figure}
638: %\begin{figure}
639: \vspace*{-73mm}
640: \hspace*{-38mm}
641: \psfig{figure=tndisp.ps,width=135mm}
642: \vspace{-81mm}
643: \caption{(color online) Quasiparticle dispersion $E_{\bf k}^\ast$
644: for the lowest-energy branch $l=0$
645: along different symmetry directions in the MBZ (lower panel),
646: along with the HF dispersion $E_{\bf k} ^0$ for comparison (upper panel).}
647: \end{figure}
648:
649: %10
650: \begin{figure}
651: \vspace*{-70mm}
652: \hspace*{-35mm}
653: \psfig{figure=tnmodeM.ps,width=140mm}
654: \vspace{-73mm}
655: %
656: \vspace*{-80mm}
657: \hspace*{-35mm}
658: \psfig{figure=tnmodeK.ps,width=140mm}
659: \vspace{-73mm}
660: %
661: \vspace*{-80mm}
662: \hspace*{-35mm}
663: \psfig{figure=155.ps,width=140mm}
664: \vspace{-73mm}
665: %
666: \vspace*{-80mm}
667: \hspace*{-35mm}
668: \psfig{figure=tnmodeT.ps,width=140mm}
669: \vspace{-85mm}
670: \caption{(color online) Electron spectral function $A_{\bf k}(\omega)$
671: at the special MBZ points for $\Delta=4$ $(U \approx 8.8)$.}
672: \end{figure}
673:
674: Figure 6 shows the spectral function $A_{\bf k}(\omega)$ for the special points
675: $\Gamma$, K, and M.
676: The weak self-energy correction in the broadened lower band
677: of the frustrated $120^\circ$ ordered AF state
678: results in no visible oscillatory structure, typically associated with the
679: string of broken bonds as the hole moves in the AF background.
680: As expected, the spectral function at the $\Gamma$ point
681: ${\bf k}=(0,0)$ and $l=-1$ shows a coherent quasiparticle peak,
682: as this state lies at the top of the lower band.
683: All three branches are degenerate at point K, as at the HF level.
684: The spectral functions at K and M points show well-defined peak structures,
685: with finite quasiparticle damping as these states lie well within the band.
686: The quasiparticle peak broadens and loses intensity
687: along both the $\Gamma-K$ and $\Gamma-M$ directions, as seen in Fig. 7.
688:
689: \subsection{Electron Dynamics (Upper Band)}
690: Self-energy correction in the narrow upper band is relatively large
691: and self consistency is noticeably slower (Fig. 8),
692: illustrating the importance of multi-magnon processes
693: and resulting in the characteristic oscillatory structure
694: in the spectral functions (Fig. 10) associated with string states.
695:
696: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
697: %11
698: \begin{figure}
699: \vspace*{-83mm}
700: \hspace*{-48mm}
701: \psfig{figure=tn3dmt.ps,width=150mm}
702: \vspace{-88mm}
703: %
704: \vspace*{-90mm}
705: \hspace*{-48mm}
706: \psfig{figure=tn3dkm.ps,width=150mm}
707: \vspace{-93mm}
708: %
709: \vspace*{-88mm}
710: \hspace*{-48mm}
711: \psfig{figure=tntk1.ps,width=150mm}
712: \vspace{-95mm}
713: \caption{(color online) Evolution of the electron spectral function
714: for the lowest-energy branch $l=0$ along different symmetry directions.}
715: \end{figure}
716:
717: %fig12
718: \begin{figure}
719: \vspace*{-32mm}
720: \hspace*{-35mm}
721: \psfig{figure=bz.ps,width=120mm,angle=-90}
722: \vspace{-35mm}
723: \caption{(color online) Hole (blue) and electron pockets (red) corresponding to
724: lowest-energy states in the Brillouin zone.
725: $\bf Q_1,Q_2,Q_3$ are equivalent AF ordering wave vectors.}
726: \end{figure}
727:
728: Figure 9 shows the quasiparticle dispersion along the $\Gamma-K-M-\Gamma$ directions
729: for the lowest-energy branch $l=0$.
730: The lowest-energy state at point $M$ shows a well-defined quasiparticle peak
731: at $\omega = 1.3$, along with a long incoherent tail, as seen in Fig. 10.
732: All branches are degenerate at $K$, second and third branches at $M$,
733: first and second branches at $\Gamma$, while at $N=(2\pi/5,0)$
734: all branches are non-degenerate, exactly as at the HF level.
735: %13
736: \begin{figure}
737: \vspace*{-68mm}
738: \hspace*{-35mm}
739: \psfig{figure=dos4.ps,width=140mm}
740: \vspace{-82mm}
741: %
742: \vspace*{-70mm}
743: \hspace*{-35mm}
744: \psfig{figure=dos3.ps,width=140mm}
745: \vspace{-82mm}
746: \caption{(color online) Renormalised density of state for one added hole and electron for
747: (a) $\Delta=4$ and (b) $\Delta=3$, showing the vanishing of energy gap.}
748: \end{figure}
749:
750: The states $K$, $\Gamma$, and $N$, which are well inside the band,
751: are strongly damped and yield dominantly incoherent spectral functions,
752: along with small quasiparticle peaks at nearly same frequency
753: $\omega \approx 2$ (Fig. 10).
754: It is the strong negative peak in the self energy at $\omega \approx 2$ (Fig. 8)
755: which leads to nearly same quasiparticle energy for all ${\bf k}$ points,
756: resulting in drastically reduced quasiparticle bandwidth (Fig. 9)
757: and enhanced effective mass.
758: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
759: Figure 11 shows that the well-defined quasiparticle peak at $M$ rapidly diminishes
760: in intensity and disappears as one moves along the $M-\Gamma$ direction.
761: However, in the $M-K$ direction,
762: the quasiparticle peak is discernible in the full $k$ range.
763: %initially decreases as $k$
764: %and broadens when we move away from $M$ of MBZ.
765: In the $K-\Gamma$ direction,
766: the quasiparticle peak marginally increases and then rapidly disappears
767: as one moves towards $\Gamma$.
768:
769: The lowest-energy hole and electron pockets are shown in Fig. 12.
770: The one-particle density of states is shown in Fig. 13 for $\Delta=4$ and $3$.
771: The classical-level asymmetry strongly influences the quantum corrections
772: and the characteristic signature of multi-magnon processes
773: is dominant for the upper band corresponding to the narrow sharp
774: peak at the classical level.
775: As expected within the many-body expansion,
776: the string-state signature of multi-magnon processes
777: is more prominent for higher $U$.
778: The renormalized gap vanishes for $U \approx 7$,
779: whereas the HF band gap $2(\Delta-2)$ vanishes at $U \approx 5$.
780: The band gap is indirect,
781: with the lowest-energy hole and electron states corresponding to momenta
782: $(4\pi/3,0)$ and $(-2\pi/3,0)$, respectively.
783:
784: It is interesting to note that
785: in the intermediate-coupling regime of interest,
786: the interaction strength effectively controls the fermion-magnon scattering
787: through three different aspects.
788: With decreasing interaction strength,
789: besides the explicit reduction in the fermion-magnon scattering matrix element
790: due to $U$ in Eq. (25),
791: the classical-level fermion and magnon dispersions are also significantly modified.
792: Due to enhanced virtual hopping and competition with the direct hopping term
793: the sharp peak in the upper band DOS (Fig. 2) shifts towards higher energy.
794: Also, the magnon energy $\omega_M$ at momentum ${\bf Q}/2$ decreases rapidly
795: and vanishes at $\Delta = 2.9$.\cite{tri}
796: Fig. 13 shows the modifications in the upper band DOS
797: due to these intermediate-$U$ effects.
798:
799: \section{Conclusions}
800: In conclusion, we have studied the hole and electron dynamics
801: in the $120^\circ$ ordered AF state
802: of the Hubbard model on a triangular lattice using a physically transparent
803: fluctuation approach involving the dynamical spin fluctuations,
804: which interpolates between the weak and strong coupling limits.
805: Finite-$U$, double occupancy effects, neglected in earlier $t-J$ model studies,
806: have been incorporated naturally in terms of classical level fermion and spin dynamics.
807: Intrinsic features of the frustration-induced direct hopping dispersion
808: associated with the $120^\circ$ ordering - the broad flat band and the narrow sharp peak
809: in the fermion DOS corresponding to lowest-energy hole and electron states -
810: are characteristics of ferromagnetic and antiferromagnetic ordering, respectively.
811: The qualitatively different self-energy corrections for hole and electron can therefore
812: be conveniently visualized in terms of hole (electron) motion in an
813: effective ferromagnetic (antiferromagnetic) spin background
814: projected out of the $120^\circ$ spin ordering.
815:
816: For an added hole in the broad lower band,
817: the reduced density of scattering states suppresses the fermion-magnon interaction
818: resulting in nearly coherent quasiparticle peak for all ${\bf k}$ states.
819: No signature of string states in the spectral function
820: reflects an effective F background seen by holes near the top of the lower band.
821: Quasiparticle dispersion shows a nearly momentum-independent shift of
822: hole energies, implying no mass renormalization.
823:
824: For an electron in the lowest-energy branch $l=0$ of the upper band,
825: quasiparticle peak is observed only near the MBZ boundary ($M-K$),
826: and rapidly vanishes away from it.
827: Strong incoherent behaviour and clear signature of string states in the spectral function
828: is a consequence of an effective AF background seen by electrons
829: near the bottom of the upper band.
830: The strong and nearly momentum-independent peak in the self energy
831: leads to nearly same quasiparticle energy for all $\bf k$ points,
832: resulting in drastically reduced bandwidth and enhanced effective mass.
833:
834: The renormalized band gap was found to vanish for $U \approx 7$,
835: yielding a first-order M-I transition, as also obtained earlier
836: for the frustrated square-lattice AF.\cite{self}
837: On the other hand, for the unfrustrated AF the band gap
838: never vanishes for any finite $U$.
839: The vanishing of band gap at moderate $U$ for both frustrated antiferromagnets
840: due to frustration-induced band broadening
841: thus highlights the role of frustration in M-I transition.
842:
843: Finally, frustration and spin fluctuations are involved in an interesting interplay
844: with respect to stability of the insulating state.
845: Frustration generally enhances spin fluctuations and magnetic disordering.
846: It will therefore be interesting to also examine the interband self-energy contribution
847: which reduces magnetic order due to interband spectral weight transfer,
848: and also widens the band gap and thereby stabilizes the insulating state.
849: The first-order interband contribution exactly cancels for negligible AF bandwidth,
850: as for the unfrustrated AF in the strong-coupling limit,\cite{self}
851: but will survive for finite frustration-induced bandwidth,
852: thus highlighting the interplay between frustration and spin fluctuations,
853: and providing deeper insight into the nature of the Mott insulator.
854:
855: \appendix*
856: \section{Fermion-magnon matrix element in the strong-coupling limit}
857: We show that to leading order in the $U/t \rightarrow \infty$ limit,
858: the intraband fermion matrix elements reduce to order $t/\Delta$,
859: and the fermion-magnon matrix element explicitly reduces to order $t$,
860: as within the $t-J$ model.
861:
862: For the magnon energy and magnon amplitudes in the $y'$ and $z'$ directions
863: we have
864: \begin{eqnarray}
865: \omega_{\bf q} & = & 3JS [(1-\gamma_{\bf q})(1+2\gamma_{\bf q})]^{1/2} \nonumber \\
866: {\cal Y}_{\bf q} ^2 &=&
867: \frac{3JS}{4\omega_{\bf q}} ( 1 + 2\gamma_{\bf q} ) \nonumber \\
868: {\cal Z}_{\bf q} ^2 &=&
869: \frac{3JS}{4\omega_{\bf q}} ( 1 - \gamma_{\bf q} ) \\
870: {\rm where} \;\; \gamma_{\bf q} & = & \frac{1}{3}\left
871: [\cos q_x + 2 \cos \frac{q_x}{2} \cos \frac{\sqrt{3}}{2}q_y \right ] \; .\nonumber
872: \end{eqnarray}
873:
874: For the lower $(-)$ and upper $(+)$ bands, the fermion amplitudes
875: given in Eq. (11) reduce to
876: \begin{equation}
877: \alpha_{\bf k}^\mp \simeq \pm \frac{1}{\sqrt{2}}
878: \left (1 \pm \frac{\xi_{\bf k}}{2\Delta} \right ) \;\;\; , \;\;\;
879: \beta_{\bf k}^\mp \simeq + \frac{1}{\sqrt{2}}
880: \left (1 \mp \frac{\xi_{\bf k}}{2\Delta} \right ) \; .
881: \end{equation}
882: Hence in terms of the spin raising and lowering operators defined below,
883: the intraband fermion matrix elements
884: \begin{eqnarray}
885: \langle \sigma^+ \rangle & \equiv &
886: \langle {\bf k},l| \sigma_{z'} + i \sigma_{y'} | {\bf k-q},m \rangle
887: \simeq \frac{\xi_{{\bf k},l}}{\Delta} \nonumber \\
888: \langle \sigma^- \rangle & \equiv &
889: \langle {\bf k},l| \sigma_{z'} - i \sigma_{y'} | {\bf k-q},m \rangle
890: \simeq \frac{\xi_{{\bf k-q},m}}{\Delta}
891: \end{eqnarray}
892: for lower band states are explicitly of order $t/U$.
893:
894: In terms of corresponding magnon amplitudes
895: \begin{equation}
896: \Phi^\pm _{{\bf q},\lambda} \equiv \sqrt{3} [ |{\bf q},\lambda \rangle_{z'}
897: \pm i |{\bf q},\lambda \rangle_{y'} ]
898: = {\cal Z}_{{\bf q},\lambda} \pm {\cal Y}_{{\bf q},\lambda}
899: \end{equation}
900: for the advanced mode, the fermion-magnon matrix element
901: \begin{eqnarray}
902: \sqrt{3} M & = & U
903: [-i \langle \sigma_{y'} \rangle {\cal Y} + \langle \sigma_{z'} \rangle {\cal Z}]
904: \nonumber \\ \nonumber \\
905: & = &U [\langle \sigma^+ \rangle \Phi^- + \langle \sigma^- \rangle \Phi^+ ] /2
906: \nonumber \\ \nonumber \\
907: & = & -(\xi_{{\bf k},l} - \xi_{{\bf k-q},m}) {\cal Y}_{{\bf q},\lambda}
908: + (\xi_{{\bf k},l} + \xi_{{\bf k-q},m}) {\cal Z}_{{\bf q},\lambda}
909: \nonumber \\ \nonumber \\
910: & = & \xi_{{\bf k},l} \Phi^- _{{\bf q},\lambda} + \xi_{{\bf k-q},m}
911: \Phi^+ _{{\bf q},\lambda} \;\;\;\;\;\; ({\rm where} \;\;m=l-\lambda)
912: \nonumber \\
913: \end{eqnarray}
914: explicitly reduces to order $t$.
915:
916: The above expression for the fermion-magnon matrix element has exactly same structure
917: as the result $\sqrt{3} t M(k,q)=
918: \sqrt{3} t [v_{\bf q} h_{\bf k} - u_{\bf q} h_{\bf k+q} ]$
919: obtained within the $t-J$ model.\cite{dombre}
920: Indeed, magnon amplitudes $u_{\bf q}$ and $v_{\bf q}$ in Ref. [17]
921: exactly correspond to ${\cal Z}_{{\bf q}} \pm {\cal Y}_{{\bf q}}$,
922: and $(\sqrt{3}t)h_{\bf k} \equiv \sqrt{3}t \sum_{\hat{\delta}}
923: \sin {\bf k}.\hat{\delta} = - \xi_{\bf k-Q}$.
924: However, the Goldstone-mode contribution appears to be different.
925: While our fermion-magnon matrix element vanishes for the Goldstone mode
926: $q \rightarrow 0,\; \lambda=0$,
927: for which $\gamma_{\bf q} \rightarrow 1$
928: and the magnon amplitude ${\cal Z} \rightarrow 0$
929: representing rigid spin rotation about $z$ axis,
930: the matrix element $M({\bf k,q})$ in Ref. [17] does not vanish.
931: The other two Goldstone modes $q \rightarrow 0,\; \lambda=\pm 1$,
932: for which $\gamma_{\bf q} \rightarrow -1/2$ and ${\cal Y} \rightarrow 0$,
933: do yield non-vanishing matrix elements with $M^2 \sim 1/q$,
934: although resulting in a finite contribution $\int q dq /q $ from long-wavelength modes.
935: Thus, for the triangular-lattice AF,
936: long-wavelength modes do contribute to the fermion-magnon scattering.
937:
938: %\newpage
939: \begin{thebibliography}{05}
940:
941: \bibitem{review1}
942: K. Kanoda, Physica C {\bf 282-287}, 299 (1997);
943: K. Kanoda, Hyperfine Interact. {\bf 104}, 235 (1997).
944:
945: \bibitem{review2}
946: R. H. McKenzie, Science, {\bf 278}, 820 (1997).
947:
948: \bibitem{watersup}
949: K. Takada {\em et al.},
950: Nature {\bf 422}, 53 (2003).
951:
952: \bibitem{weitering}
953: H. H. Weitering, X. Shi, P. D. Johnson, J. Chen, N. J. Dinardo, and S. Kempa,
954: Phys. Rev. Lett. {\bf 78}, 1331 (1997).
955:
956: \bibitem{res1}
957: T. Inami, Y. Ajiro, and T. Goto,
958: J. Phys. Soc. Jpn. {\bf 65}, 2374 (1996).
959:
960: \bibitem{resonance}
961: L. E. Svistov, A. I. Smirnov, L. A. Prozorova, O. A. Petrenko, L. N. Demianets,
962: and A. Ya. Shapiro,
963: Phys. Rev. B {\bf 67} 094 434 (2003).
964:
965: \bibitem{holmium}
966: O. P. Vajk, M. Kenzelmann, J. W. Lynn, S. B. Kim, and S.-W. Cheong,
967: cond-mat/0502006 (2005).
968:
969: \bibitem{sw}
970: S. Ghosh and A. Singh, cond-mat/0506475.
971:
972: \bibitem{tri}
973: A. Singh, Phys. Rev. B {\bf 71}, 214406 (2005).
974:
975: \bibitem{kanoda}
976: Y. Shimizu, K. Miyagawa, K. Kanoda, M. Maesato, and G. Saito,
977: Phys. Rev. Lett. {\bf 91}, 107001 (2003).
978:
979: \bibitem{kagawa}
980: F. Kagawa, T. Itou, K. Miyagawa, and K. Kanoda,
981: Phys. Rev. B {\bf 69}, 064 511 (2004);
982: F. Kagawa, T. Itou, K. Miyagawa, and K. Kanoda,
983: cond-mat/0409437 (2004).
984:
985: \bibitem{pirg2}
986: H. Morita, S. Watanabe, and M. Imada,
987: J. Phys. Soc. Jpn. {\bf 71}, 2109 (2002).
988:
989: \bibitem{pirg3}
990: M. Imada, T. Mizusaki, and S. Watanabe, cond-mat/0307022 (2003).
991:
992: \bibitem{self}
993: P. Srivastava and Avinash Singh, Phys. Rev. B, {\bf 70}, 115103 (2004).
994:
995: \bibitem{spectrum}
996: A. Singh and P. Goswami,
997: Phys. Rev. B {\bf 66}, 92402 (2002).
998:
999: \bibitem{capriolti}
1000: L. Capriolti, A. E. Trumper and S. Sorella, Phys. Rev. Lett., {\bf 82}, 3899
1001: (1999)
1002:
1003: \bibitem{dombre}
1004: M. Azzouz and T. Dombre, Phys. Rev. B, {\bf 53}, 402 (1996).
1005:
1006: \bibitem{trumper}
1007: A. E. Trumper, C. J. Gazza and L. O. Manuel, Phys. Rev. B, {\bf 69}, 184407 (2004).
1008:
1009: \bibitem{apel}
1010: W. Apel, H.-U. Everts and U. K$\ddot{o}$rner, Eur. Phys. J. B, {\bf 5}, 317 (1998).
1011:
1012: \bibitem{vojta}
1013: Matthias Vojta, Phys. Rev. B {\bf 59}, 6027 (1999).
1014:
1015: \bibitem{dagotto}
1016: M. Vojta, E. Dagotto, Phys. Rev. B, {\bf 59}, 713 (1999).
1017:
1018: \bibitem{qmc_green}
1019: M. C. Refolio, J. M. L\'{o}pez Sancho, and J. Rubio,
1020: cond-mat/0103459 (2001).
1021:
1022: \end{thebibliography}
1023:
1024: \end{document}
1025:
1026: