cond-mat0203526/v3.tex
1: \documentclass{epl}
2: 
3: 
4: \title{Comparative study between two quantum spin systems 
5: KCuCl$_{3}$ and TlCuCl$_{3}$ }
6: 
7: \author{ T. Saha-Dasgupta\inst{1} and Roser Valent\'\i\inst{2}} 
8: 
9: \institute{
10: \inst{1} S.N. Bose National Centre for Basic Sciences, 
11: JD Block, Sector 3,
12:  Salt Lake City, Kolkata 700098, India.\\
13: \inst{2} Fakult\"at 7, Theoretische Physik,
14:  University of the Saarland,
15: 66041 Saarbr\"ucken, Germany.}
16: 
17: \pacs{75.10.Jm}{Quantized spin models}
18: \pacs{75.30.Et}{Exchange and superexchange interactions}
19: \pacs{71.20.-b}{Electron density of states and band structure of
20:                 crystalline solids}
21: 
22: 
23: \begin{document}
24: 
25: 
26: \maketitle
27: 
28: 
29: \begin{abstract}
30: We have performed an {\it ab initio} study
31: of the electronic structure of two isostructural
32:  quantum spin systems, KCuCl$_{3}$ and TlCuCl$_{3}$,
33: which have recently attracted much attention  due to 
34: their unconventional magnetic properties. Our first-principles analysis shows
35: unambiguously the role of Tl, as opposed to structural differences
36: between the two compounds, in making TlCuCl$_{3}$ a {\it strongly} coupled
37:  s=$\frac{1}{2}$ dimer 
38: system compared to KCuCl$_{3}$ which shows a {\it weakly}
39: coupled  s=$\frac{1}{2}$ dimer behavior. 
40: Good agreement with the existing analysis of  
41: inelastic neutron scattering results has been observed.
42: 
43: \end{abstract}
44: 
45: 
46: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
47: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
48: 
49: %\vspace*{1cm}
50: 
51: {\it Introduction.-} The behavior of low-dimensional quantum spin systems is
52:   dominated by 
53: zero-point spin fluctuations.  As a consequence, they show a variety
54:  of  interesting  magnetic properties which are not present in their
55:  classical counterparts. In the last years a great amount of effort,
56:  both theoretically and experimentally, has been dedicated to
57:  the study of these quantum spin systems.  An important example
58:  of such systems are materials
59: having a singlet ground state with a finite energy 
60:  gap in the spin excitation spectra like
61: even-leg ladder compounds\cite{lad}, s= $\frac{1}{2}$
62:  alternating chain compounds
63: \cite{ac}, spin-Peierls \cite{sp} or 
64: spin-dimer compounds \cite{dimer}. 
65:  In particular, the family of coupled spin-dimer systems
66:  ACuCl$_3$ (A = K, Tl) is being currently intensively studied in connection to
67:  its magnetic properties.   KCuCl$_{3}$  has
68:    a spin excitation gap $\Delta$ $\sim$ 2.6 meV 
69:  and the  analysis of inelastic neutron scattering
70:  (INS)\cite{cavadini_00,kcucl} data indicate that this system is 
71:  a spin-dimer system  with {\it weak} three-dimensional interdimer
72: couplings.  TlCuCl$_3$,   though being isostructural to 
73:  KCuCl$_{3}$,  has
74:    a spin excitation gap  $\Delta$ $\sim$ 0.65meV  about four times
75: smaller than that of  KCuCl$_{3}$ and a  saturation field 
76: twice as large as that for KCuCl$_3$. Accordingly, 
77: the behavior of this system
78:  has been proposed\cite{INS2,INS1}
79:  to be that of a three-dimensional {\it strongly} coupled
80:  spin-dimer system. The small size of the spin gap makes these 
81:  compounds very suitable for studying the effects of the application of
82:  an external magnetic field on the quantum properties of the system.
83:  Recently \cite{mag},
84:  magnetic field-induced N\'eel ordering  has been observed in 
85:  TlCuCl$_3$ for magnetic fields $H$ higher than the
86:   gap field $H_g$ = $\Delta /g \mu_B$ $\sim$ 6 T 
87: and has been interpreted
88:  as a Bose-Einstein condensation
89:  of  excited triplets (magnons) \cite{BEC}. New nuclear magnetic
90: resonance (NMR) investigations\cite{Vyaselev_02}
91:  have also shown that this
92:  phase transition is accompanied by changes in the crystal
93: lattice parameters,
94:  what is an indication of an important spin-phonon coupling in this
95:  material. The observation of field-induced phase transition
96:  has been also reported in KCuCl$_3$ for a magnetic field
97: $H > H_g$ $\sim$ 22 T \cite{Oosawa_01}.
98:  Magnetization measurements\cite{dis} were also performed in 
99:  the mixed spin system (Tl$_{1-x}$K$_x$)CuCl$_3$ where the $x$ substitution
100:  creates bond randomness in the system and the field-induced
101: phase transition was studied as a function of the K concentration.
102:   Motivated by the great amount of exotic phenomena 
103:  observed in these materials, we have carried out first principles study
104:  of the electronic properties of these systems
105:  in order to give a microscopic foundation 
106:  to their behavior and analyze
107:  the similarities and differences between the two systems.
108: 
109: 
110: 
111: 
112: %\vspace*{0.5cm}
113: {\it Crystal Structure.-} Both KCuCl$_{3}$ and TlCuCl$_{3}$ crystallize  in the monoclinic $P2_{1}/c$
114: space group\cite{struc} with four formula units per unit cell.
115:  The lattice parameters
116: for KCuCl$_{3}$ are given by $a = 4.029 \AA$, $b = 13.785 \AA$, $c= 8.736 \AA$
117: and $\beta = 97.20^{o}$, while those for TlCuCl$_{3}$ are $a = 3.982 \AA$, 
118: $b = 14.144 \AA$, $c= 8.890 \AA$ and $\beta = 96.32^{o}$. Therefore, compared
119: to KCuCl$_{3}$, the TlCuCl$_{3}$ lattice is somewhat compressed along
120: the $a$-axis and enlarged  in the $bc$ plane.  The degree of monoclinicity
121: also varies by some extent. Both structures are formed by edge-sharing 
122: CuCl$_{6}$ octahedra which build zig-zag chains running along the $a$-axis
123: (fig. 1).  These chains are located at the corners and
124:  center of the unit cell in the
125: $bc$ 
126: plane. K$^{+}$/Tl$^{+}$ ions are distributed in between these chains.
127:  The CuCl$_{6}$ octahedra are highly distorted due to 
128: the Jahn-Teller active
129: Cu$^{2+}$ ion with large elongation along one direction resulting into
130: nearly planar Cu$_{2}$Cl$_{6}$ dimer clusters  which are the basic
131: unities in the structure (marked with the box in fig. \ref{structure}).
132: 
133: 
134: %\vspace*{0.5cm}
135: {\it Band Structure.-} 
136: In panels (a) and (b) of fig.\ \ref{bands}
137:  we show the band-structures of KCuCl$_{3}$ and 
138: TlCuCl$_{3}$ along various symmetry directions. First-principles Density
139:  Functional Theory (DFT)
140: calculations have been carried out within the framework of the 
141: state-of-art full-potential
142:  linearized augmented plane wave (LAPW)
143: method\cite{lapw} and the
144:  linearized muffin tin orbital (LMTO) method\cite{lmto} based on
145: the local-density-approximation (LDA). The 
146: band-structures obtained by the two methods are in overall
147:  agreement with each other.
148: The predominant features of the band structure are the complex of four 
149: narrow bands close to the Fermi level formed by Cu d$_{x^2-y^2}$
150:  orbitals (in the local frame of reference) contributed
151: by each Cu atom in the unit cell, admixed with Cl p-states. 
152: These  bands are half-filled and the insulating groundstate observed
153:  in these compounds should be explained by the effect of strong 
154: correlations which are not fully considered in the LDA approximation. 
155:  This set of bands is separated from the low-lying valence bands by a gap of
156:  about 0.3 to 0.5 eV  \cite{explanation} and from the high-lying
157:   excited
158:   bands, which are
159: dominated by states from the K/Tl atoms, by another 
160: gap of about 3 eV. Concentrating on the low-energy bands close to
161: the Fermi-level,  we see that the bands are dispersive along all the symmetry
162: directions suggesting the compounds to be three-dimensionally
163:  coupled systems in
164: agreement with the findings of INS measurements
165:  \cite{cavadini_00,kcucl,INS2,INS1}.
166:  Since the dispersion behavior 
167: is similar between KCuCl$_{3}$ and TlCuCl$_{3}$,  
168: the predominant interaction pathways should be  alike for both 
169: compounds. 
170:  The total bandwidth of the four-band complex close to the 
171: Fermi level
172: is somewhat larger for TlCuCl$_{3}$ ($\approx$ 0.5 eV)  compared to that
173: of KCuCl$_{3}$ ($\approx$ 0.4 eV) suggesting the intradimer coupling 
174: to be stronger  in
175: TlCuCl$_{3}$ than in
176:  KCuCl$_{3}$. 
177:  We also observe here that the
178:  bands in TlCuCl$_{3}$ are in general more dispersive  than in   KCuCl$_{3}$
179:  what indicates larger interdimer interactions in  TlCuCl$_{3}$
180:  compared to KCuCl$_{3}$ \cite{explanation_2}.
181: 
182: 
183: 
184: 
185:  This change in the
186: strength of the intra- and interdimer couplings between the two compounds
187:  could be caused by (i)
188: the changes in the structural parameters
189: between the two compounds or by (ii) the role of the Tl$^{+}$ ion compared to
190: that of the K$^{+}$ ion. In order to have an  understanding of which of these
191:  effects
192:  determines predominantly the behavior of these systems,
193:  we have carried out model calculations of  KCuCl$_{3}$
194:  by considering the  structural parameters  of the Tl-compound and
195:  viceversa, we  computed TlCuCl$_{3}$
196:  by considering the structural parameters  of the K-compound.
197:  The corresponding
198: band structures are shown in panels (c) and (d) of fig.\ \ref{bands}.  
199:  As we see, changing
200: the lattice parameters has a minor effect on the band structure (compare
201: fig.\ \ref{bands}(a) and fig.\ \ref{bands}(c))
202:  while substituting K$^{+}$ by Tl$^{+}$ in the
203: same lattice has a significant effect making the band dispersions comparable
204: to that of the Tl-compound.
205:  This exercise already points unambiguously to the role
206: of the Tl$^+$
207:  ion in enhancing the strength of the coupling as compared
208: to that of changes in the structural parameters. In the next section
209:  we will make
210: this analysis more quantitative in terms of various hopping
211: integrals.
212: 
213: 
214: %\vspace*{0.5cm}
215: {\it Low-energy Hamiltonians: Hopping integrals.-}
216: We have used the LMTO-based downfolding method, which has been proposed, 
217: implemented\cite{newlmto} and applied to a number of cases\cite{example} in
218: recent years, in order to obtain the low-energy effective Hamiltonians
219: that describe the behavior of a system.
220:  This method of deriving the low-energy Hamiltonians by 
221: integrating out (downfolding)
222:  the high energy degrees of freedom results in  
223: Hamiltonians defined in the basis of effective orbitals. The process takes
224: into account the proper effective contribution
225: from the orbitals that are being downfolded.
226: For the present compounds, we have derived the low-energy Hamiltonians
227: defined in the basis of effective Cu d$_{x^2-y^2}$ orbitals, by keeping only 
228: the d$_{x^{2}-y^{2}}$ orbital for each Cu atom in the unit cell 
229: and integrating
230: out all the rest. The Fourier transform of this
231:  few-orbital downfolded Hamiltonian
232: provides the various hopping integrals, $t_{ij}$,
233:  between these effective orbitals and the corresponding tight-binding (TB)
234: Hamiltonian can be written as
235: $H = \sum_{(i,j)} t_{ij} (c_i^{\dagger} c_j + h.c.)$
236: where $i$ and $j$ denote a pair of Cu$^{2+}$ ions.
237: These hopping integrals thus, form the first-principles set of parameters
238: obtained without any fitting procedure containing the fingerprint of the
239: pathways involved in the hopping processes.
240: 
241: 
242: 
243: 
244: 
245: 
246: Table I shows the various hopping integrals for KCuCl$_{3}$, 
247: TlCuCl$_{3}$ and the two model compounds  KCuCl$_{3}$ considering the 
248:  lattice parameters of
249: TlCuCl$_{3}$ and TlCuCl$_{3}$ in the lattice of KCuCl$_{3}$. The various
250: hopping pathways are marked in fig. 1. 
251: The hoppings are named according to the
252:  various magnetic couplings shown in fig. 1 of ref.  \cite{INS1} i.e.
253:  $t(l\ m\ n)$
254:  ($t'(l\ m\ n)$)
255:  denote the hopping parameters between two equivalent (nonequivalent)
256: sites  (in terms of the corresponding spins for magnetic
257: interaction) in dimers separated by a
258:  lattice vector $\vec{r} = l a \hat{x} + m b \hat{y} + n c \hat{z}$.  In 
259: Table I we also show the magnetic couplings obtained in various references
260: by fitting the experimental  INS data.
261: 
262: 
263: Although,  due to the complexity of the exchange
264:  pathways, there are no
265: simple relationships connecting the hopping and magnetic
266:  exchange integrals,
267: the relative strength of the hopping parameters in the
268:  K- and Tl-compound can be easily noticed to be
269: in good qualitative agreement with the relative strength
270: shown by the magnetic exchange parameters.
271: 
272:  Considering first the intradimer coupling, 
273:  the distortion in the nearest neighbor Cu-Cl-Cu dimer
274: bond (see the marked box in fig. 1)
275: from 90$^o$ towards a linear bonding of 180$^o$ angle increases the 
276: importance of the antiferromagnetic (AF) coupling over the  ferromagnetic
277: (FM) coupling. The Cu-Cl-Cu bond angles in these compounds are
278:  $\approx$ 96$^{o}$ which is larger
279: than 94$^{o}$, the  {\it upper bound} angle  given  in the
280: edge-shared cuprates\cite{af+fm}
281:  for a 
282: possible ferromagnetic coupling, thereby providing the intradimer
283:  exchange coupling J
284:   of AF nature.  Considering the superexchange mechanism to be
285: the effective mechanism of exchange in the dimer, and therefore
286: relating the magnetic exchange integral J with the hopping
287:  integral $t$ through J=$4t^2/U$ where $U$ is the effective on-site
288: Coulomb repulsion, we observe that the ratio
289: t$^{2}_{kcucl_3}$/t$^{2}_{tlcucl_3}$ $\sim$  0.75 obtained from the
290:  {\it ab initio} analysis compares very well with the ratio of
291: J's obtained by fitting the INS data, 0.76 in ref. \cite{INS1} 
292: and 0.79 in ref. \cite{INS2}.
293: 
294: Focusing now on the interdimer interactions,
295: note that
296:  the values of the hopping matrix elements in 
297:  TlCuCl$_{3}$ are larger than those for  KCuCl$_{3}$, thus concluding 
298:  that TlCuCl$_{3}$ will be a strongly coupled dimer system compared to  
299: KCuCl$_{3}$ which can
300: be characterized as a weakly coupled dimer system in agreement with
301: the conclusions drawn from INS data and already conjectured from
302: band dispersions. Quantitatively, the largest enhancement of the
303: interdimer coupling between TlCuCl$_{3}$ and KCuCl$_{3}$ occurs
304: for neighboring dimers in the  (1,0,-2) plane, i.e.
305:  $(l\ m\ n) = (2\ 0\ 1)$. The other
306: dominant interdimer hoppings are $t(1\ 1/2\ 1/2)$ and $t(1\ -1/2\
307: 1/2)$\cite{explanation_3}, the last one
308: being 
309: significantly enhanced in the TlCuCl$_{3}$ system compared to that of
310: KCuCl$_{3}$. The LDA band dispersions of the four-band complex 
311: are reproduced reasonably well by considering the intradimer hopping
312: matrix element
313: and the following interdimer hoppings $t^{'}(2\ 0\ 1)$ and $t(1\ 1/2\ 1/2)$ for
314: KCuCl$_{3}$ and the additional hopping of $t(1\ -1/2\ 1/2)$
315: for TlCuCl$_{3}$.
316: 
317: 
318:     
319:  The analysis of the two model calculations (panel (c) and (d) of
320:  fig.\ \ref{bands}) shows
321:  the evident role of the Tl$^+$ ion  in enhancing
322: the hopping matrix elements between the Cu$^{2+}$ ions. Comparing the 2nd and 5th column
323: of Table I, the hopping matrix elements turn out to be similar
324: between KCuCl$_{3}$ and the model system KCuCl$_{3}$ in the lattice
325: of TlCuCl$_{3}$, while keeping the lattice parameters fixed
326: and substituting K by Tl, the hoppings are affected 
327: significantly (4th column) with larger intra-  and interdimer
328: hoppings \footnote{Considering the model calculation of TlCuCl$_{3}$ in
329: the lattice of KCuCl$_{3}$, the lattice shrinks along the $b-$axis and
330: expands along the $a-$axis (by a factor 8 smaller
331:  than the shrinkage along the $b-$axis)
332:  which is a distorsion similar to the recently
333: observed changes in the crystal lattice when the system undergoes
334: a phase transition to a N\'eel state induced by the application
335: of an external magnetic field\cite{Vyaselev_02}. Although the relative
336: change in the $b-$parameter for this model calculation is much larger
337: than that reported for the field-induced transition, this study may
338: provide indications of the changes in the couplings caused by this
339: strain effect. }.  In order to quantify the action of Tl$^+$, we
340: show 
341: in fig. 3  the Tl and K partial density
342: of states (PDOS) in TlCuCl$_{3}$ and  KCuCl$_{3}$ respectively. As one sees,
343: the contribution of Tl in the four narrow low energy bands is much
344: larger compared to that of K which implies larger hybridization of
345: Tl with the Cu d$_{x^2-y^2}$ orbitals in TlCuCl$_{3}$ compared to that
346: of K in KCuCl$_{3}$. This relatively large hybridization effect
347:  from the Tl atoms
348: situated in between the double chains   can be understood in
349:  terms of the proximity of 
350: the Cu d$_{x^2-y^2}$ and the Tl  6$s$, 6$p$ energy states as well as the
351: extended nature of these orbitals in comparison to K 4$s$.  This behavior
352:  leads to a
353: strongly coupled 
354: network of Cu-dimers in TlCuCl$_{3}$,  while   KCuCl$_{3}$
355:  behaves as a collection of Cu-dimers
356:  weakly interacting in three-dimensions. This description is
357:  also supported by electron density analysis of the LDA results.
358: Interestingly, this conclusion is opposite to the case of CaV$_{2}$O$_{5}$ and
359: MgV$_{2}$O$_{5}$ where the subtlety in structural differences turn
360: out to be responsible for the differences in the 
361: behavior of the two systems\cite{cavo+mgvo}.
362: 
363: 
364: 
365: 
366: 
367: %\vspace*{0.5cm}
368: {\it Conclusions.-}
369:   We have presented a comparative {\it ab initio} study of the electronic
370:  properties of KCuCl$_3$ and TlCuCl$_3$, two isostructural
371:  quantum spin systems which have been the subject of recent interest
372:  due to their unconventional magnetic properties. 
373:  It has been the purpose of this study: (i) to obtain a microscopic
374: description
375: of the properties of these systems and its comparison with the behavior
376:  predicted from INS. 
377:  (ii) to understand  the  microscopic origin of the different behavior
378:       between  these two compounds.
379: 
380: By means of the downfolding-TB analysis which reduces the 
381: information provided by LDA to an effective low-energy Hamiltonian
382: in tight-binding basis,
383: we could succesfully corroborate the coupled dimer behavior predicted
384: for these systems by INS. In agreement with the predictions from
385: the INS data analysis, we found in KCuCl$_3$ the interdimer couplings
386:  to be  weak, while TlCuCl$_3$ manifests itself as a strongly coupled
387: dimer system.
388: We also showed that  the origin of the differences in the interdimer
389: coupling between the two systems is related to
390:  the relative contribution of Tl compared to that of K to 
391:  the chemical bonding between two Cu$^{2+}$ ions.
392: Our study is also indicative of possible changes thay may occur
393: in the coupling constants due to lattice changes caused by strain
394: effects\cite{Vyaselev_02}. 
395: 
396: 
397: 
398: 
399: 
400: \acknowledgments
401: 
402: 
403: 
404: 
405: 
406: We acknowledge useful discussions with Peter Lemmens and H. Tanaka
407:  and R.V. thanks
408: the Deutsche Forschungsgemeinschaft for finantial support through
409:  a Heisenberg fellowship.
410: 
411: 
412: 
413: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
414: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 
415:  \begin{thebibliography}{99}
416: 
417: \bibitem{lad} Dagotto E. and Rice T. M., Science {\bf 271}, (1996) 618.
418: 
419: \bibitem{ac} Garrett A. W., Nagler S. E., Tennant D. A.,
420:              Sales B. C. and Banes T., Phys. Rev. Lett {\bf 79} (1997)
421:              745.
422: 
423: \bibitem{sp} Hase M., Terasaki I. and Uchinokura K.,
424: 	     Phys. Rev. Lett {\bf 70} (1993) 3651.
425: 
426: \bibitem{dimer} Tennant D. A., Nagler S. E., Garrett A. W., Barnes T.,
427:                 Torardi C. C., Phys. Rev. Lett. {\bf 79}, 745 (1997).
428: 
429: \bibitem{cavadini_00} Cavadini N., Heigold G., Henggeler W., Furrer A.,
430: 		G\"udel H. -U., Kr\"amer K., Mutka H.,
431:                 J. Phys.:Condens. Matter {\bf 12} (2000) 5463.
432: 
433: \bibitem{kcucl} M\"uller M. and Mikesha H. -J., J. Phys.:
434:               Condens. Matter {\bf 12}, (2000) 7633.
435: 
436: \bibitem{INS2}  Cavadini N., Heigold G., Henggeler W., Furrer A.,
437: 		G\"udel H. -U., Kr\"amer K., Mutka H., Phys. Rev
438: 		B {\bf 63} (2001) 172414.
439: 
440: 
441: 
442: \bibitem{INS1}  Oosawa A., Kato T., Tanaka H., Kakurai K., 
443:                 M\"uller M. and Mikesha H. -J., Phys. Rev. B
444:                 {\bf 65} (2002) 094426.
445: 
446: 
447: 
448: 
449: \bibitem{mag} Oosawa A., Ishii M. and Tanaka H., J. Phys.:
450:               Condens. Matter {\bf 11}, (1999) 265.
451: 
452: \bibitem{BEC} T. Nikuni, M. Oshikawa, A. Oosawa and H. Tanaka,
453:               Phys. Rev. Lett {\bf 84}, 5868 (2000); M. Matsumoto, 
454:               B. Normand, T. M. Rice, and M. Sigrist, to appear in
455:               Phys. Rev. Lett (2002)
456: 
457: \bibitem{Vyaselev_02} Vyaselev O., Takigawa M., Vasiliev A.,
458:                       Oosawa A., and Tanaka H.
459:                       {\it to be published}.
460: 
461: 
462: \bibitem{Oosawa_01} Oosawa A., Tanaka H., Takamasu T.,
463:                     Abe H., Tsujii N., and Kido G.,
464:                     Physica B {\bf 294-295}, (2001) 34.
465: 
466: \bibitem{dis}    Oosawa A. and Tanaka H., {\it cond-mat/0203293}, to
467:                  appear in Phys. Rev. B (2002).
468: 
469: \bibitem{struc}  Willett R. D., Dwinggins C., Kruh R. F. and 
470:                  Rundle R. E., J. Chem. Phys.,{\bf 38}, (1963) 2429; 
471:                  Tanaka H., Oosawa A., Kato T., Uekusa H., Ohashi Y.,
472:                  Kakurai K. and Hoser A., J. Phys. Soc. Jpn {\bf 70}
473:                  (2001) 939. 
474: 
475: \bibitem{lapw}   Blaha P., Schwarz K. and Luitz J., WIEN97,
476:              {\it A Full Potential Linearized Augmented Plane 
477:         Wave Package for Calculating Crystal Properties}, (Karlheinz
478:          Schwarz, Techn. Univ. Wien, Vienna 1999). ISBN 3-9501031-0-4.
479:        [Updated version of Blaha P., Schwarz K., Sorantin P.,
480:          and Trickey S. B., Comp. Phys. Commun. {\bf 59}, (1990) 399].
481: 
482: \bibitem{lmto}  Andersen O. K., Phys. Rev. B {\bf 12}, (1975) 3060.
483: 
484: \bibitem{explanation} The gap between the occupied valence states
485: and the four-band complex at the Fermi level is underestimated in the  LMTO
486: method compared to the LAPW method, what can be attributed to the
487: atomic sphere approximation invoked in LMTO. This discrepancy
488: however does not affect the low-energy physics at the Fermi level
489: that we will be
490: concerned with.
491: 
492: \bibitem{explanation_2} This interpretation can be understood by the
493:  fact that the bandstructure of an isolated dimer system ({\it i.e.} 
494:  systems having only non-zero 
495:  intradimer interaction) is formed by two dispersionless bands, the separation
496:  of which is proportional to the strength of the intradimer
497:  interaction. Switching on the interactions between isolated dimers
498:  (inter-dimer interactions) brings in dispersion to the otherwise
499:  dispersionless band structure. Therefore, the bandwidth of the
500:  four-band complex in the present case gives us
501:  an estimate of the intradimer coupling-strength while the 
502:  degree of dispersion
503:  of these bands can be related to the interdimer couplings.
504: 
505: \bibitem{newlmto} Andersen O. K. and Saha-Dasgupta T.,
506: Phys. Rev. {\bf B62}, R16219 (2000) and references there in.
507: 
508: \bibitem{example} Pavarini E., Dasgupta I., Saha-Dasgupta T.,
509: Jepsen O.,
510: Andersen O. K., 
511: Phys. Rev. Lett. {\bf 87} (2001) 047003; Valent\'\i~ R.,
512: Saha-Dasgupta T., 
513: Alvarez J. V., Pozgajcic K., Gros C.,
514: Phys. Rev. Lett. {\bf 86}, (2001) 5381;  M\"uller T.,
515: Anisimov V.I., Rice T. M., Dasgupta I., Saha-Dasgupta T.,
516: Phys. Rev. {\bf B57}, (1998) R12655.
517: 
518: 
519: \bibitem{af+fm} Mizuno Y., Tohyama T., Maekawa S., Osafune T.,
520: Motoyama N., Eisaki H. and Uchida S., Phys. Rev. {\bf B57}, (1998) 5326 .
521: 
522: 
523: 
524: \bibitem{explanation_3} While the interdimer distances $t^{'}(1\
525: 1/2\ 1/2)$ and $t^{'}(1\ -1/2\ 1/2)$, $t^{'}(0\
526: 1/2\ 1/2)$ and $t^{'}(0\ -1/2\ 1/2)$ are all equivalent, $t(1\
527: 1/2\ 1/2)$ and $t(1\ -1/2\ 1/2)$ are not equivalent since
528: they  involve different
529: pathways.
530: 
531: 
532: 
533: \bibitem{cavo+mgvo} Korotin M. A., 
534: Anisimov V. I., Saha-Dasgupta T. and Dasgupta I.,
535: J. Phys. (Condens. Matter) {\bf 12} (2000) 113.
536: 
537: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
538: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
539: 
540: 
541: \end{thebibliography}
542: 
543: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
544: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
545: 
546: 
547: \newpage
548: 
549: {\bf FIGURE CAPTIONS} \\ \\
550: 
551: 
552: \noindent
553: {1. Structure of KCuCl$_3$ (TlCuCl$_{3}$). Small black and white atoms
554: represent Cu and Cl respectively while the large gray atoms
555: represent K(Tl). The middle double chain is shifted with respect
556: to other two double chains along $b-$direction by half a lattice
557: constant. Various hoppings are marked (see text for details).}\\ \\
558: 
559: \noindent
560: {2. Band structure of KCuCl$_{3}$ (panel (a)), TlCuCl$_{3}$
561: (panel (b)), KCuCl$_{3}$ in the lattice of TlCuCl$_{3}$ (panel(c))
562: and TlCuCl$_{3}$ in the lattice of KCuCl$_{3}$ (panel(d)) along
563: the symmetry directions $\Gamma$ = (0 0 0),  B=(-$\pi$,0,0),
564: D=(-$\pi$,0,$\pi$),
565: Z=(0,0,$\pi$), $\Gamma$,
566: Y=(0,$\pi$,0), A = (-$\pi$,$\pi$,0),
567: E=(-$\pi$,$\pi$,$\pi$).
568: }\\ \\
569: 
570: \noindent
571: {3. Tl(solid) and K(dotted) contributions to
572: the density of states of TlCuCl$_{3}$ and KCuCl$_{3}$ respectively. 
573: }\\
574: 
575: {\bf TABLE CAPTIONS} \\ \\
576: 
577: 
578: %\noindent
579: {\it I. Hopping integrals obtained from our ab initio analysis
580: for TlCuCl$_{3}$,
581:  KCuCl$_{3}$ and the two model systems considered in the text and magnetic interactions from ref.\ \cite{INS2} and ref.\ \cite{INS1}
582:  for TlCuCl$_{3}$ and
583:  KCuCl$_{3}$.
584:  The negative sign denotes a ferromagnetic coupling.}
585: 
586: \newpage
587: 
588: %%%%%%%%%% FIGURES %%%%%%%%%%%%%%%%%%%%%%%%%%%
589: 
590: \begin{figure}
591: \includegraphics[width=8cm]{fig_ac1.eps}
592: %\onefigure{fig_ac1.eps}
593:  \caption{.
594:  }
595:  \label{structure}
596: \end{figure}
597:   
598: 
599: 
600: \begin{figure}
601: \includegraphics[width=16cm]{bands.eps}
602: %\onefigure{fig_ac1.eps}
603:  \caption{  }
604:  \label{bands}
605: \end{figure}
606: 
607: 
608: \begin{figure}
609: \includegraphics[width=8cm]{pdos.eps}
610: %\onefigure{fig_ac1.eps}
611:  \caption{  }
612:  \label{dos}
613: \end{figure}
614: 
615: 
616: \begin{table}
617: \caption{  }
618: \begin{tabular}{|c|c|c|c|c|}
619: t's and J's & KCuCl$_{3}$ & TlCuCl$_{3}$ & TlCuCl$_{3}$ in & KCuCl$_{3}$ in \\ 
620: in meV &  &  & KCuCl$_{3}$
621: & TlCuCl$_{3}$ \\ 
622: & & & lattice & lattice \\ \hline
623: t & 116 & 134 & 144 & 117 \\
624: J & 4.34 ref\cite{INS1} & 5.68 ref\cite{INS1} & & \\
625:  &  4.29 ref\cite{INS2} & 5.42 ref\cite{INS2} & & \\\hline
626: t(100) & 2.72 & 6.81 & 4.08 & 1.00 \\
627: t'(100) & 9.52 & 21.76 & 24.49 & 5.40 \\
628: J$_{eff}$(100)= & -0.21 ref\cite{INS1} & -0.46 ref\cite{INS1} & & \\
629: 1/2[2J(100) - J'(100)] &  -0.21 ref\cite{INS2} & -0.47 ref\cite{INS2} & & \\\hline
630: t'(201) & 36.73 & 65.31 & 72.11 & 35.37 \\
631: J$_{eff}$(201) = & -0.45 ref\cite{INS1} & -1.53 ref\cite{INS1} & & \\
632: -1/2J'(201) &  -0.34 ref\cite{INS2} & -1.43 ref\cite{INS2} & & \\\hline
633: t(1 1/2 1/2) & 40.82 & 36.73 & 42.17 & 35.40 \\
634: t(1 -1/2 1/2) & 4.08 & 32.18 & 40.50 & 2.72 \\
635: t'(1 1/2 1/2) & 5.19 & 12.24 & 10.88 & 8.16 \\
636: J$_{eff}$(1 1/2 1/2)= & 0.28 ref\cite{INS1} & 0.49 ref\cite{INS1} & & \\
637: 1/2[J(1 1/2 1/2) - J'(1 1/2 1/2)] &  0.37 ref\cite{INS2} & 0.62 ref\cite{INS2} & & \\\hline
638: t(0 1/2 1/2) & 5.44 & 8.16 & 6.80 & 8.01 \\
639: t'(0 1/2 1/2) & 14.97 & 19.05 & 16.33 & 17.63 \\
640: J$_{eff}$(0 1/2 1/2)= & -0.003 ref\cite{INS1} & -0.06 ref\cite{INS1} & & \\
641: 1/2[J(0 1/2 1/2) - J'(0 1/2 1/2)] &  &  & & \\\hline
642: \end{tabular}
643: \end{table}
644: 
645: 
646: 
647: 
648: \end{document}
649: 
650: