1: \documentclass[superscriptaddress,floats,a4paper,floatfix,prb]{revtex4}
2: \usepackage{graphicx}% Include figure files
3: \usepackage{dcolumn}% Align table columns on decimal point
4: \usepackage{bm}% bold math
5:
6: \newcommand{\lpcmo}{(La$_{1-y}$Pr$_{y}$)$_{0.7}$Ca$_{0.3}$MnO$_3$}
7: \newcommand{\lpcmoa}{(La$_{0.5}$Pr$_{0.5}$)$_{0.7}$Ca$_{0.3}$MnO$_3$}
8: \newcommand{\lpcmob}{(La$_{0.25}$Pr$_{0.75}$)$_{0.7}$Ca$_{0.3}$MnO$_3$}
9: \newcommand{\lcmo}{La$_{1-x}$Ca$_{x}$MnO$_3$}
10: \newcommand{\tc}{$T_C$}
11: \newcommand{\tg}{$t_{2g}$}
12: \newcommand{\eg}{$e_g$}
13: \newcommand{\cm}{cm$^{-1}$}
14: \newcommand{\sdc}{$\sigma_{DC}$}
15: \newcommand{\sig}{$\sigma_1(\omega)$}
16: \newcommand{\eps}{$\epsilon_1(\omega)$}
17:
18: \begin{document}
19:
20: \title{Evidence for differentiation in the iron-helicoidal-chain in GdFe$_{3}$(BO$_{3}$)$_{4}$ }
21:
22: \author{S.A. Klimin}
23: \affiliation{Material Science Center, University of
24: Groningen, 9747 AG Groningen, The Netherlands.}
25: \affiliation{Institute of Spectroscopy, RAS, 142190, Troitsk, Moscow
26: Region, Russia.}
27: \author{D. Fausti}
28: \author{A. Meetsma}
29: \affiliation{Material Science Center, University of
30: Groningen, 9747 AG Groningen, The Netherlands.}
31: \author{L.N. Bezmaternykh}
32: \affiliation{L.V.Kirensky Institute of Physics, Siberian branch
33: of RAS, Krasnoyarsk 660036, Russia.}
34: \author{P.H.M. van Loosdrecht}
35: \affiliation{Material Science Center, University of
36: Groningen, 9747 AG Groningen, The Netherlands.}
37: \author{T.T.M. Palstra}
38: \affiliation{Material Science Center, University of
39: Groningen, 9747 AG Groningen, The Netherlands.}
40:
41: %\date{\today}
42:
43: \begin{abstract}
44: We report on a single-crystal X-ray structure study of
45: GdFe$_3$(BO$_3$)$_4$ at room temperature and at T=90~K. At room
46: temperature GdFe$_3$(BO$_3$)$_4$ crystallizes in a trigonal space
47: group $R$32 (No. 155), the same as found for other members of
48: iron-borate family RFe$_3$(BO$_3$)$_4$. At 90 K the structure of
49: GdFe$_3$(BO$_3$)$_4$ transforms to the space group $P$3$_{1}$21
50: (No. 152). The low-temperature structure determination gives new
51: insight into the weakly first-order structural phase transition at
52: 156~K and into the related Raman phonon anomalies. The presence of
53: two inequivalent iron chains in the low temperature structure
54: provides a new perspective on the interpretation of the
55: low-temperature magnetic properties.
56: \end{abstract}
57: \maketitle
58:
59:
60: \section{Introduction}
61: The family of borates RM$_3$(BO$_3$)$_4$ with a rare earth (RE) or
62: yttrium as R and Al, Ga, Fe, or Sc as M crystallize in the
63: huntite, CaMg$_3$(CO$_3$)$_4$, structure type with space group
64: $R$32 \cite{Juob,Belo,Camp}. The interest in this family of
65: crystals arises both from a fundamental point of view and from
66: already realized and proposed applications. Crystals of
67: YAl$_3$(BO$_3$)$_4$ and GdAl$_3$(BO$_3$)$_4$ doped with Nd were
68: widely studied during recent years and have been used in optical
69: devices, such as self-frequency doubling and self-frequency
70: summing lasers (see, e.g., Ref. \cite{Bren} and references
71: therein). Concentrated NdAl$_3$(BO$_3$)$_4$ crystals are efficient
72: media for minilasers \cite{che}.
73: \\ Apart from the interesting optical properties arising chiefly
74: from the lack of inversion
75: symmetry, the 'sub-family' of RFe$_3$(BO$_3$)$_4$ also attracts
76: considerable attention due to their structure peculiarities. The
77: presence of magnetic order at temperatures less then 37 K was
78: attributed to magnetic Fe-Fe or Fe-O-Fe interactions inside quasi
79: one-dimensional (1D) iron chains \cite{Camp,bala}. Recent works
80: \cite{levi,bala}, focusing on low temperature magnetism in
81: GdFe$_3$(BO$_3$)$_4$, revealed two magnetic phase transitions. The
82: second-order magnetic ordering phase transition at
83: $T_{\mathrm{N1}}=$37~K (antiferromagnetic ordering of Fe atoms) is
84: followed by a first-order spin-reorientational phase transition at
85: $T_{\mathrm{N2}}=$10~K. Additionally specific heat and Raman
86: measurements on single crystals \cite{levi} of
87: GdFe$_3$(BO$_3$)$_4$ revealed a weakly first-order structural
88: phase transition at $T_{\mathrm{s}}=$156~K. This structural phase
89: transition is observed in almost all members of the RE
90: ferro-borates family \cite{hina,levi,loos}. Recently, Hinatsu et
91: al. \cite{hina} have shown that DyFe$_3$(BO$_3$)$_4$ undergoes
92: such structural phase transition at 340~K, by measuring the
93: temperature dependence of the lattice parameters on powder
94: samples. Here too, a peak in the specific heat was observed at
95: this temperature. Similar peaks in specific heat vs temperature
96: dependence were found for other heavy RE (R= Eu-Ho, Gd, Tb)
97: ferroborates. These peaks were ascribed to structural phase
98: transitions. The transition temperatures $T_{\mathrm{s}}$ were
99: found to depend linearly on the ionic radius of the RE. To date,
100: there are no single crystal data available nor has the
101: low-temperature (LT) space group been determined.
102:
103: The high temperature (HT) $R$32 structure of powder
104: RFe$_3$(BO$_3$)$_4$ compounds was first determined in Ref.
105: \cite{Juob} for R = La, Nd, Sm-Ho, and Y. X-ray experiments on
106: single crystals with R=Nd$_{0.5}$Bi$_{0.5}$ \cite{Belo} and R=La,
107: Nd, and Y$_{0.5}$Bi$_{0.5}$ \cite{Camp} confirmed this structure.
108: Moreover, Raman measurements for LaFe$_3$(BO$_3$)$_4$ \cite{alic},
109: NdFe$_3$(BO$_3$)$_4$ \cite{alic,loos} and GdFe$_3$(BO$_3$)$_4$
110: \cite{loos} are in good agreement with group theoretical analysis,
111: based on the $R$32 structure.
112:
113: Summarizing, whereas the HT crystal structure of most of the
114: ferroborates is known, the LT space group, lattice parameters, and
115: atomic positions are unknown. In this work we report on a
116: single-crystal X-ray diffraction structure study of
117: gadolinium-iron borate at room temperature (RT) and at 90~K . It
118: is found that the LT-structure has the $P$3$_1$21 symmetry. Two
119: nonequivalent spin chains exist in the LT-structure which gives
120: new insight in the low-temperature magnetic properties. The
121: $P$3$_1$21 symmetry of the LT-phase also confirms the
122: interpretation of the reported IR absorption by crystal-field
123: transitions \cite{levi,chuk} and leads to a better understanding
124: of some peculiarities observed in Raman spectra \cite{levi,loos}.
125:
126: \section{Experiment and result of structure determination}
127:
128: Crystals of GdFe$_3$(BO$_3$)$_4$ were grown using a
129: K$_{2}$Mo$_{3}$O$_{10}$ - based flux, as described in Ref.
130: \cite{bala}. Big transparent single crystals of gadolinium
131: ferroborates were light green in color and had a good optical
132: quality. A block-shaped crystal ('broken-fragment') with the
133: dimensions of 0.22 $\times$ 0.15 $\times$ 0.11~mm$^{3}$~was
134: mounted on top of a glass fiber and aligned on a Bruker
135: \cite{Bru1} SMART APEX CCD diffractometer. The crystal was cooled
136: to 90(1)~K using a Bruker KRYOFLEX. Intensity measurements were
137: performed using graphite monochromated Mo-K$\bar{\alpha}$
138: radiation. Generator settings were 50~KV/ 40~mA. SMART \cite{Bru1}
139: was
140: used for preliminary determination of the unit cell constants and data
141: collection control. The intensities of reflections of a hemisphere
142: were collected by a combination of 6 sets of exposures (frames).
143: Each set had a different angle for the crystal and each exposure
144: covered a range of $0.3^{\circ}$ in $\omega$. A total of 3600
145: frames were collected with an exposure time of 10.0 seconds per
146: frame. The overall data collection time was 16 hours. Data
147: integration and global cell refinement was performed with the
148: program SAINT. The final unit cell was obtained from the xyz
149: centroids of 4767 and 5439 reflections after integration for RT
150: and 90 K, respectively. Intensity data were corrected for Lorentz
151: and polarization effects, scale variation, for decay and
152: absorption (an analytical absorption correction was applied), and
153: reduced to $F_{o}^{2}$. The program suite SHELXTL was used for
154: space group determination (XPREP).
155: \begin{figure}[htb]
156: \includegraphics[width=85mm]{fig1}
157: \caption{The structure of GdFe$_{3}$(BO$_{3}$)$_{4}$ in two
158: different projection. The left panels (a) and (b) show the RT
159: structure, and the right panels (c) and (d) show the structure at
160: 90~K. The Fe atoms are arranged in chiral chains parallel to the
161: \textbf{\textit{c}}-axis. Different chains are separated by
162: GdO$_{6}$ prisms and BO$_{3}$ groups. The unit cell outline for
163: $R$32 is shifted by (1/3, 1/3, 0) for comparison with the LT
164: $P$3$_{1}$21 structure.} \label{structure}
165: \end{figure}
166:
167: At both temperatures of RT and 90~K the unit cell was identified
168: as trigonal: reduced cell calculations did not indicate any higher
169: metric lattice symmetry \cite{X3}. Space groups $R$32 and
170: $P$3$_1$21 were derived for RT and for 90~K, respectively, from
171: the systematic extinctions and discriminated from other candidate
172: space groups, which comply with the same extinctions conditions,
173: during the structure determination process. Examination of the
174: final atomic coordinates of the structure did not yield extra
175: crystallographic or metric symmetry elements \cite{X4,X5}.
176:
177: The polarity of the structure of the crystal actually chosen was
178: determined by Flack's $\mathbf{x}$-refinement
179: \cite{flack1,flack2,flack3,Herbst1998}; refinement resulted in a
180: $\mathbf{x}$ value of 0.50(1), so ultimately an inversion twin is
181: used in the refinement.
182:
183: \begin{table}[p]
184: \caption{Crystallographic parameters}
185: \tiny
186: \label{roomlowTcomparison}
187: \begin{tabular}{lll}
188: &&\\
189: \multicolumn{3}{l}{\footnotesize \textbf{a-Crystal data and details of the structure determination.}}\\
190: & Room Temperature = 293(1)~K& T = 90(1)~K \\
191: Moiety Formula &\multicolumn{2}{c}{GdFe$_{3}$(BO$_{3}$)$_{4}$} \\
192: Formula Weight $(g.mol^{-1})$ &\multicolumn{2}{c}{560.04} \\
193: Crystal System & \multicolumn{2}{c}{Trigonal} \\
194: Space group. & $R$32, 155 & $P$3$_{1}$21, 152\\
195: $a$\ (\AA) & 9.5203(6) & 9.5305(3)\\
196: $c$\ (\AA) & 7.5439(5) & 7.5479(2)\\
197: $V$\ (\AA$^{3}$) & 592.15(7) & 593.73(3)\\
198: $\Theta$ range unit cell:min.-max., &3.66-38.70;4767 & 3.66-38.55; 5439\\
199: deg; reflections & & \\
200: Formula\_Z & 3 & 3\\
201: Space group\_Z & 18 & 6\\
202: Z'(=Formula\_Z/SpaceGroup\_Z) & 1/6 & 0.5\\
203: $\rho (g/cm^{3})$ & 4.712 & 4.699\\
204: F(000) electrons & 774 & 774\\
205: $\mu(Mo K\alpha) (cm^{-1})$ & 137.74 & 137.37\\
206:
207: &&\\
208: \multicolumn{3}{l}{\footnotesize \textbf{b-Data collection.}}\\
209: $\lambda (MoK\bar{\alpha})$, \AA & \multicolumn{2}{c}{0.71073}\\
210: Monochromator & \multicolumn{2}{c}{Graphite}\\
211: Measurements device type &\multicolumn{2}{c}{Bruker SMART APEX,}\\
212: &\multicolumn{2}{c}{CCD area-detector diffractometer}\\
213:
214: Detector area resolution (pixels/mm) & \multicolumn{2}{c}{$4096\times 4096/62\times 62$~(binned 512)} \\
215: Measurement method & $\varphi$- and $\omega$-scans & $\varphi$- and $\omega$-scans\\
216: $\theta$ range; min. max.,deg & 3.66, 38.70 & 2.47, 38.58\\
217: Index ranges & $h:-16\rightarrow16;k:-16\rightarrow16$; & $h:-16\rightarrow16;k:-16\rightarrow16$;\\
218: & $l: -13\rightarrow13$ & $l: -13\rightarrow13$ \\
219: Min.- Max. absorption transmission factor & 0.0563-0.2519& 0.0620-0.2513\\
220: X-ray exposure time, h & 16.0 & 16.0\\
221: Total data & 4700 & 13868\\
222: Unique data & 755 & 2217\\
223: Data with criterion: ($F_{o}\geq 4.0\sigma(F_{o})$) & 755 & 1956\\
224: $R_{int} = \sum[|F_{o}^{2} -F_{o}^{2} (mean)|]/\sum[F_{o}^{2}]$ & 0.0339 & 0.0338\\
225: $R_{sig} = \sum\sigma(F_{o}^{2}) / [F_{o}^{2}]$ & 0.0203 & 0.0189\\
226: &&\\
227: \multicolumn{3}{l}{\footnotesize \textbf{c-Refinement.}}\\
228: %\multicolumn{}\\
229: Number of reflections & 755 &2217\\
230: Number of refined parameters & 35 &95\\
231: Isotropic secondary-extinction coefficient, g & 0.0371(12) &0.0359(10)\\
232: Final agreement factors: & 0.0374 &0.0452\\
233: $wR(F^{2}) = [\sum [w(F_{o}^{2}-F_{c}^{2})^{2}] /[w(F_{o}^{2})^{2}]]^{1/2}$ & & \\
234: Weighting scheme: a, b & 0.0272, 0.0 &0.0312, 0.0\\
235: $w=1/[\sigma^{2}(F_{o}^{2})+(aP)^{2}+bP]$ & &\\
236: And $P=[max(F_{o}^{2},0) + 2F_{c}^{2}]/3$ & &\\
237: $R(F)=\sum(||F_{o}|- |F_{c}||)/\sum|F_{o}|
238: For F_{o}>4.0\sigma(F_{o})$ &0.0151 & 0.0167\\
239: Absolute-Structure parameter Flack's x &0.50 &0.50\\
240: $GooF=S=[\sum[w(F_{o}^{2}-F_{c}^{2})^{2}]/(n-p)]^{1/2}$& 1.075 &0.855\\
241: $n$\ = number of reflections & &\\
242: $p$\ = number of parameters refined & &\\
243: Residual electron density in final &-1.0, 0.6(1) &-0.5,0.8(1)\\
244: Difference Fourier map, $e/$\AA$^{3}$ & &\\
245: Max. (shift/s) final cycle &0.001 &0.003\\
246: Average (shift/$\sigma$) final cycle &0.000 &0.000\\
247: \end{tabular}
248: \end{table}
249:
250:
251: Crystal data and numerical details on data collection and
252: refinement are given in Table \ref{roomlowTcomparison}. Final
253: fractional atomic coordinates, equivalent displacement parameters
254: and anisotropic displacement parameters are given in Table
255: \ref{roomt-par-tab} for data at RT and in Table \ref{lowT-par-tab}
256: for the 90 K data.
257:
258: \begin{table}[htb]
259: \caption{Fractional atomic coordinates and Anisotropic Parameters
260: at RT} \label{roomt-par-tab}
261: \begin{tabular}{ccccccc}
262: \hline Atom &Wyckoff & Schoenflies& x & y & z & $U_{eq}$(\AA) \\
263: \hline
264: Gd & 3 a &D$_{3}$ & 0 & 0 & 0 & 0.00867(5)\\
265: Fe & 9 d &C$_{2}$ & 0.21659(5) & 1/3 & 1/3 & 0.00654(9)\\
266: O1 & 9 e &C$_{2}$ & 0.1442(2) & x & 1/2 & 0.0083(3)\\
267: O2 & 9 e &C$_{2}$ & 0.4087(3) & x & 1/2 & 0.0132(5)\\
268: O3 & 18 f& C$_{1}$ & 0.0254(2) & 0.2125(2) & 0.1824(2) & 0.0095(3)\\
269: B1 & 3 b &D$_{3}$ & 0 & 0 & 1/2 & 0.0066(6)\\
270: B2 & 9 e &C$_{2}$ & 0.5526(3) & x & 1/2 & 0.0077(4)\\
271: \hline\\
272: &&&&&&\\
273: \end{tabular}
274:
275: \begin{tabular}{ccccccc}
276: \hline
277: Atom & $U_{11}$ & $U_{22}$ & $U_{33}$ & $U_{23}$ & $U_{13}$ & $U_{12}$ \\
278: \hline
279: Gd & 0.00888(8) & 0.00888(8) & 0.00825(9)& 0.0000(-)& 0.0000(-)& 0.00444(4)\\
280: Fe & 0.00591(12)&0.00635(14) & 0.00750(18)&0.00009(8)&0.00004(4)&0.00317(7)\\
281: O1 & 0.0058(5) &0.0058(5) & 0.0109(7) &0.0014(3) &0.0014(3)&0.0011(6)\\
282: O2 & 0.0076(6) &0.0076(6) & 0.0182(11) &0.0047(5) &0.0047(5)&0.0009(8)\\
283: O3 & 0.0065(5) &0.0112(5) & 0.0111(5) &0.0025(5) &0.0021(4)&0.0046(4)\\
284: B1 & 0.0058(9) &0.0058(9) &0.0081(14) &0.0000(0) &0.0000(-) &0.0029(4)\\
285: B2 & 0.0070(7) & 0.0070(7) &0.0093(8) & 0.0007(4) & 0.0007(4) &0.0036(11)\\
286: \hline
287: \end{tabular}
288: \end{table}
289:
290: \begin{table}
291: \caption{Fractional atomic coordinates and Anisotropic Parameters
292: at T=90~K} \label{lowT-par-tab}
293: \begin{tabular}{ccccccc}
294: \hline Atom &Wyckoff & Schoenflies& x & y & z & $U_{eq}$(\AA) \\
295: \hline
296: Gd & 3 a &C$_{2}$ & -0.33342(1) & x & 0 & 0.00406(4)\\
297: Fe1& 3 a &C$_{2}$ & 0.11536(5) & x & 0 & 0.00360(6)\\
298: Fe2& 6 c &C$_{1}$ & -0.21420(6) &-0.54975(4) & 0.34154(2) & 0.00366(6)\\
299: O1 & 3 b &C$_{2}$ & 0 & -0.07819(15)& 1/6 & 0.0072(3)\\
300: O2 & 6 c &C$_{1}$ & -0.5832(2) & -0.2709(1) & 0.13774(12) &0.00692(17)\\
301: O3 & 6 c &C$_{1}$ & -0.1194(3) & -0.30445(26)& -0.17980(18)& 0.0057(3)\\
302: O4 & 6 c &C$_{1}$ & -0.1467(3) & -0.36234(16)&0.18479(18)& 0.0058(2)\\
303: O5 & 6 c &C$_{1}$ & 0.4755(2) & 0.1451(2) & -0.15980(8) & 0.0057(3)\\
304: O6 & 3 b &C$_{2}$ & 0.1877(2) & 0 & 5/6 & 0.0049(3)\\
305: O7 & 6 c &C$_{1}$ & -0.5235(3) & -0.53811(17)&-0.18533(18) & 0.0056(2)\\
306: B2a& 6 c &C$_{1}$ & -0.4473(4) & -0.1201(-) & 0.15617(14) & 0.0053(3)\\
307: B2b& 3 b &C$_{2}$ & 0 & -0.2223(3) & 1/6 & 0.0049(4)\\
308: B1 & 3 b &C$_{2}$ & 0.33204(14) & 0 & 5/6 & 0.0045(5)\\
309: \hline\\
310: &&&&&&\\
311: \end{tabular}
312:
313: \begin{tabular}{ccccccc}
314: \hline
315: Atom & $U_{11}$ & $U_{22}$ & $U_{33}$ & $U_{23}$ & $U_{13}$ & $U_{12}$ \\
316: \hline Gd & 0.00409(6) & 0.00409(7) & 0.00408(7) & 0.00010(2)& 0.00009(1) &0.00210(3)\\
317: Fe & 0.00355(10)&0.00355(10) & 0.00355(10)&0.00006(4) &0.00006(4) &0.00166(10)\\
318: Fe & 0.00349(9) &0.00356(11) & 0.00378(9) &0.00013(7) &0.00003(5) &0.00166(6)\\
319: O1 & 0.0081(6) &0.0054(4) & 0.0089(4) &0.00121(19)&0.0024(4) &0.0040(3)\\
320: O2 & 0.0058(3) &0.0062(3) & 0.081(3 &0.0012(3) &0.0014(3) &0.0025(3)\\
321: O3 & 0.0062(4) &0.0040(5) & 0.0065(4) &0.0006(3) &0.0007(3) &0.0023(4)\\
322: O4 & 0.0054(4) &0.0045(4) & 0.0067(4) &0.0007(3) &0.0003(3) &0.0019(4)\\
323: O5 & 0.0051(4) &0.0050(4) & 0.0063(5) &0.0007(3) &0.0003(3) &0.0019(5)\\
324: O6 & 0.0044(4) &0.0058(6) & 0.0062(5) &0.0019(5) &0.0010(2) &0.0029(3)\\
325: O7 & 0.0047(4) &0.0063(4) & 0.0064(4) &0.0008(3) &0.0001(3) &0.0033(4)\\
326: B2a & 0.0050(6) &0.0067(7) &0.0055(5) &0.0004(5) &0.0001(4) &0.0040(4)\\
327: B2b & 0.0018(8) & 0.0061(6) &0.0055(7) & 0.0004(3) &0.0008(6) &0.0009(4)\\
328: B1 & 0.0042(8) & 0.0048(9) &0.0046(11) & 0.0009(5) &0.0004(2) &0.0024(4)\\
329: \hline
330: \end{tabular}
331: \end{table}
332:
333: \section{Discussion of the structure}
334:
335: After a brief introduction describing the main features of
336: GdFe$_3$(BO$_3$)$_4$, we will focus on the difference between the
337: RT and the 90~K (LT) structure. First, we will try to understand
338: the consequences that the structural changes have on the
339: interpretation of Raman spectra anomalies at the weak first-order
340: phase transition reported in the literature
341: \cite{levi}. Second, we will give a new perspective for the
342: interpretation of low temperature magnetic data
343: \cite{levi,bala}.
344:
345:
346: The room temperature structure of GdFe$_3$(BO$_3$)$_4$ belongs to
347: the $R$32 space group (Fig. \ref{structure}a and
348: \ref{structure}b). Our measurement confirms the structure reported
349: previously for powder samples
350: \cite{Juob} and for several single crystals \cite{Belo,Camp} from
351: the family RFe$_3$(BO$_3$)$_4$. The structure consists of
352: alternating layers (parallel to $ab$-plane) of Fe-Gd and BO$_{3}$
353: groups (see Fig. 1b). The main features of this structure are
354: already described in literature \cite{Camp}: the BO$_{3}$ groups
355: are arranged in layers nearly perpendicular to the C$_{3}$ axis
356: and the Fe atoms are arranged in helicoidal chains parallel to
357: this axis. Different chains are connected by GdO$_{6}$ and
358: BO$_{3}$ groups, where each individual BO$_3$ and GdO$_{6}$ group
359: connects three chains. The distance between Fe atoms in the same
360: chain ($3.1669(4)$~\AA) is shorter than the Fe-Fe distances for
361: two nearest chains, that varies along the chain, $4.8308(5)$~\AA,
362: being the shortest one. The main exchange interaction between
363: Fe$^{3+}$ is therefore of quasi-1D nature.
364:
365: \begin{figure}[htb]
366: \centerline{\includegraphics[width=85mm]{fig2}}
367: \caption{Coordination polyhedra for the RT ($R$32-(a)) and the LT
368: ($P$3$_{1}$21-(b)) structures of GdFe$_{3}$(BO$_{3}$)$_{4}$. At RT
369: the Fe atoms with C$_{2}$-symmetry are coordinated by three
370: different types of oxygen atoms. AT 90~K there are two
371: inequivalent positions for the Fe atoms: Fe1 (the same symmetry as
372: RT) and Fe2, which is in a general position (surrounded by five
373: different types of oxygen). At RT the Gd atoms are surrounded by
374: six oxygens of one type (D$_{3}$-symmetry). At 90~K they are
375: coordinated by three different types of oxygen atoms
376: (C$_{2}$-symmetry). For the B atoms there are two kinds of
377: coordination at RT. B1 (in D$_{3}$-symmetry) is surrounded by
378: three oxygen atoms of the same type, the B1O$_{3}$-group is thus
379: an equilateral triangle. The B2 atoms are surrounded by two types
380: of O, B2O$_{3}$ is an isosceles triangle (C$_{2}$-symmetry). At
381: 90~K there are three kinds of coordinations for the B atoms:
382: B2bO$_{3}$ and B1O$_{3}$ are isosceles triangles (C$_{2}$
383: symmetry) and B2aO$_{3}$ is general triangle. } \label{corr}
384: \end{figure}
385:
386: \begin{table}[htb]
387: \caption{Symmetry position of BO$_{3}$ groups, angle between the
388: group and the \textbf{\textit{c}}-axis and their flatness of them
389: The flatness is expressed as distance of the B atoms from the
390: plane defined by the three oxygen ligands.} \label{tilt}
391: \begin{tabular}{cccccccl}
392: \multicolumn{4}{c}{Room Temperature } & \multicolumn{4}{c}{Low Temperature } \\
393: & Symm. & Angle [degr.]& Flatness & & Symm. & Angle [degr.]& Flatness\\
394: B1 & D$_{3}$& 90 & 0 & B1 & C$_{2}$ & 87.52(4)& 0.00002(4)\\
395: B2 & C$_{2}$ & 84.37(11) &0 & B2a & C$_{1}$ & 81.89& 0.0055(11)\\
396: & & & & B2b & C$_{2}$ & 83.55& 0.0000(2)\\
397: \end{tabular}
398: \end{table}
399:
400: Upon lowering the temperature the GdFe$_3$(BO$_3$)$_4$ crystal
401: reduces the symmetry from $R$32 to $P$3$_{1}$21, in the trigonal
402: system. Fig. \ref{corr} shows the coordination polyhedra
403: (GdO$_{6}$, FeO$_{6}$, and BO$_{3}$) for two different structures
404: $R$32 and $P$3$_{1}$21. At RT the BO$_{3}$ groups occupy two
405: inequivalent positions, B1 (D$_{3}$) and B2 (C$_{2}$). At 90~K the
406: site symmetry of the B1 atoms is reduced to C$_{2}$, whereas the
407: site symmetry of the B2 atoms differentiates into a B2b (C$_{2}$)
408: and B2a (C$_{1}$) (see also Table \ref{tilt}). In the LT-phase,
409: the angle between BO$_{3}$ groups and \textbf{\textit{c}}-axis is
410: changed. Moreover, in the LT-phase the BO$_{3}$ groups in C$_{1}$
411: position are considerably distorted and no longer flat (see
412: Tab.\ref{tilt}).
413:
414: \begin{figure}
415: \centerline{\includegraphics[width=85mm]{fig3}}
416: \caption{Temperature dependence of Raman frequency of the low
417: energy mode appearing at the structural phase transition
418: $T_{\mathrm{s}}$. Insert shows the low-frequency part of Raman
419: spectrum at two close temperatures before and after phase
420: transition.} \label{softmode}
421: \end{figure}
422:
423: The LT structure is in agreement with the Raman data
424: \cite{levi,loos}. The group theoretical vibrational analysis,
425: based on the 90~K structure, shows that, due to the lowering of
426: the symmetry, new librational modes of the BO$_{3}$ become Raman
427: active ($R_{x}$ and $R_{y}$). The observed structural change is
428: compatible with an appearance of the librational $R_{y}$ mode of
429: the BO$_{3}$ group upon approaching $T_{s}$ from above, and a
430: subsequent hardening of this mode in the LT phase (See
431: Fig.\ref{softmode}). Also the anisotropic displacement parameters
432: (Table \ref{roomt-par-tab}) $U_{33}$ of the oxygen atoms suggest
433: that the borate groups are relatively free to oscillate around the
434: $y$-axis.
435:
436: Concerning the magnetic structure, the main peculiarity of both
437: the RT and the 90~K structures is the existence of magnetically
438: quasi-1D helicoidal iron chains (see Fig. \ref{structure}). The
439: intra-chain exchange interaction between the Fe ions is expected
440: to be dominated by Fe-Fe direct exchange and Fe-O-Fe
441: superexchange, depending respectively on Fe-Fe distance and two
442: Fe-O-Fe angles (Fig.\ref{exchange}). Nevertheless there are some
443: significant differences.
444:
445: At RT all the Fe atoms are in equivalent positions (C$_{2}$). All
446: Fe-chains are equivalent as are the Fe-O-Fe angles
447: ($102.40^{\circ}$(12) and $103.65^{\circ}$(8)) and Fe-Fe distances
448: ($3.1669(4)$~\AA). Therefore, the exchange interactions between
449: neighboring iron ions within a chain are also equivalent. At LT
450: (Fig. \ref{structure}), as shown in Table \ref{tilt}, the BO$_{3}$
451: groups form one general triangle (gray C$_{1}$ position) and two
452: isosceles triangles (white and dark, C$_{2}$ position). At RT
453: "gray" and "white" groups become also equivalent, with C$_{2}$
454: symmetry, and the "dark" one is a regular triangle with the
455: D$_{3}$ symmetry. The reduction of the symmetry of the borate
456: groups changes the surrounding of the Fe atoms (Fig. \ref{corr}),
457: yielding two inequivalent positions (C$_{2}$ and C$_{1}$).
458: Therefore, the Fe-Fe distances are different for the two chains:
459: one is stretched ($3.1828(4)$~\AA) and in the other one is
460: compressed ($3.1554(4)$~\AA). The angles Fe-O-Fe for the first
461: chain are $101.24^{\circ}$(5) and $103.71^{\circ}$(9), while those
462: for the second are $102.46^{\circ}$(6) and $103.91^{\circ}$(8).
463: Therefore also the intra-chain exchange interaction is different
464: for the two chains.
465:
466: The GdO$_{6}$ prism connects three chains, one containing
467: Fe1-atoms with C$_{2}$ symmetry and two with Fe2-atoms in a
468: general position. There is only one inequivalent position for the
469: Gd-ions in both the RT and the 90~K structures. However, the site
470: symmetry of the Gd-ions changes from D$_{3}$ at RT to C$_{2}$ at
471: 90~K. This confirms the interpretation of the authors of Ref.
472: \cite{levi}, who interpreted low-temperature infrared spectra of
473: Nd$_{0.01}$Gd$_{0.99}$Fe$_{3}$(BO$_{3}$)$_{4}$ in terms of Kramers
474: doublets of Nd, assuming only one structural position for the Nd
475: ions.
476:
477: \begin{figure}
478: \centerline{\includegraphics[width=50mm]{fig4}} \caption{Two
479: Fe-chains and the most important exchange paths: the intra-chain
480: exchange is via Fe-Fe direct exchange or Fe-O-Fe super-exchange,
481: while the inter-chain exchange is through Fe-O-Gd-O-Fe and
482: Fe-O-O-Fe.} \label{exchange}
483: \end{figure}
484:
485: The detailed structure of the Fe-chains and their interconnection
486: are shown in Fig. \ref{exchange}. The intra-chain interactions
487: between Fe-atoms go through two Fe-O-Fe superexchange pathways.
488: The magnetic inter-chain interaction arises from the Fe-O-O-Fe,
489: and possibly the Fe-O-Gd-O-Fe superexchange path. The role of the
490: first superexchange path is important because the substitution of
491: Gd for non-magnetic Y does not lead to a disappearance of 3D
492: magnetic ordering. On the contrary, the N\'eel temperature for
493: YFe$_3$(BO$_3$)$_4$, $T_{\mathrm{N1}}$=40~K, is larger than the
494: one found for Gd-compound ($T_{\mathrm{N1}}$=37~K), and for
495: NdFe$_3$(BO$_3$)$_4$ ($T_{\mathrm{N1}}$=30~K) \cite{chuk,loos}.
496: Considering the different RE$^{3+}$ ionic radii ($0.983$~\AA\ for
497: Nd, $0.938$~\AA\ for Gd, and $0.900$~\AA\ for Y) it is clear that
498: the Ne\'el temperature of RFe$_3$(BO$_3$)$_4$ depends strongly on
499: the ionic radii: a smaller ionic radius results in a higher
500: $T_{\mathrm{N1}}$. Moreover, $T_{\mathrm{N1}}$ does not seem to be
501: affected by the spin of the RE. In this sense it is clear that the
502: main superexchange path of the interaction between different
503: chains is the Fe-O-O-Fe path, and that a small distortion of this
504: path changes substantially the magnetic properties of the system.
505: It is therefore clear that the existence of two nonequivalent Fe
506: chains with different Fe-Fe distances and Fe-O-Fe angles could
507: lead to substantially different intra-chain exchange constant for
508: the two chains \cite{goodenough}, therefore it should be taken
509: into account for the interpretation of the magnetic properties.
510:
511: \section{Conclusions}
512:
513: In summary, we have determined the crystal structure of
514: GdFe$_3$(BO$_3$)$_4$ at RT and 90~K. At RT GdFe$_3$(BO$_3$)$_4$
515: exhibits $R$32 structure, in agreement with Refs.
516: \cite{Juob,Camp}. Below the structural phase transition
517: ($T_{\mathrm{s}}=156$~K) the structure has the $P$3$_{1}$21 space
518: group. The main difference of LT-structure compared to the RT one
519: is the lowering of the symmetry and the tilt of the BO$_{3}$
520: groups. This confirms the interpretation of the Raman spectra of
521: this phase transition. The main conclusion resulting from the LT
522: structure determination is the presence of two inequivalent
523: positions for the Fe atoms giving rise to two different iron
524: helicoidal chains.
525:
526: \section{Acknowledgments}
527: The authors are very grateful to M.N. Popova for valuable
528: discussions. This work was partially supported by the Stichting
529: voor Fundamenteel Onderzoek der Materie (FOM, financially
530: supported by the Nederlandse Organisatie voor Wetenschappelijk
531: Onderzoek (NWO)). One of the authors (S.K.) acknowledges the
532: support of the Russian Foundation for Basic Research, grant
533: ¹04-02-17346, and the Russian Academy of Sciences under the
534: Programs for Basic Research.
535: \bibliography{gfbo}
536:
537: \end{document}
538: