1: %
2: % This is a paper to be submitted to Phys. Rev. B
3: %
4: % Typeset using REVTeX 4
5: %
6: % Please address all correspondence to:
7: %
8: % Dr. C.C. Homes, Rm. 1-17
9: % Department of Physics - Bldg. 510B
10: % Brookhaven National Laboratory
11: % P.O. Box 5000
12: % Upton, NY 11973-5000
13: %
14: % Tel: (631) 344-7579
15: % FAX: (631) 344-2739
16: % Email: homes@bnl.gov
17: %
18: % Last edited on June 27, 2002
19: %
20: %\documentclass[prb,preprint,showpacs,byrevtex,floatfix]{revtex4}
21: %
22: %
23: %
24: \documentclass[prb,showpacs,byrevtex,twocolumn,floatfix]{revtex4}
25: \usepackage{graphicx}
26: \usepackage{dcolumn}
27: \usepackage{bm}
28:
29:
30: %
31: \begin{document}
32:
33: \bibliographystyle{apsrev}
34:
35: \preprint{Draft copy --- not for distribution}
36: %
37: % The title and the list of authors
38: %
39: \title{Infrared optical properties of \boldmath Pr$_2$CuO$_4$ \unboldmath}
40: %
41: %
42: %
43: \author{C. C. Homes}%
44: \email[Electronic address: ]{homes@bnl.gov}%
45: \affiliation{Department of Physics, Brookhaven National Laboratory, Upton, NY
46: 11973}
47: %
48: % Materials Science
49: %
50: \author{Q. Li}%
51: \affiliation{Department of Materials Science, Brookhaven National Laboratory,
52: Upton, NY 11973}
53: %
54: % Sherbrooke
55: %
56: \author{P. Fournier}%
57: \altaffiliation{Present address: D\'{e}partment de Physique,
58: Universit\'{e} de Sherbrooke, Sherbrooke, Qu\'{e}bec, J1K 2R1 Canada.}
59: %
60: % Maryland
61: %
62: \author{R. L. Greene}
63: \affiliation{Center for Superconductivity Research, Department of Physics and
64: Astronomy, University of Maryland, College Park, MD 20742}%
65: \date{\today}
66: %
67: % The abstract
68: %
69: \begin{abstract}
70: %
71: The {\it ab}-plane reflectance of a Pr$_2$CuO$_4$ single crystal has been
72: measured over a wide frequency range at a variety of temperatures, and the
73: optical properties determined from a Kramers-Kronig analysis.
74: %
75: Above $\approx 250$~K, the low frequency conductivity increases quickly with
76: temperature; $\rho_{dc}\approx 1/\sigma_1(\omega\rightarrow 0)$ follows the
77: form $\rho_{dc}\propto \exp(E_a/k_BT)$, where $E_a\approx 0.17$~eV is much less
78: than the inferred optical gap of $\approx 1.2$~eV. Transport measurements show
79: that at low temperature the resistivity deviates from activated behavior and
80: follows the form $\rho_{dc}\propto \exp [(T_0/T)^{1/4}]$, indicating that the
81: {\it dc} transport in this material is due to variable-range hopping between
82: localized states in the gap.
83: %
84: The four infrared-active $E_u$ modes dominate the infrared optical properties.
85: Below $\approx 200$~K, a striking new feature appears near the low-frequency
86: $E_u$ mode, and there is additional new fine structure at high frequency. A
87: normal coordinate analysis has been performed and the detailed nature of the
88: zone-center vibrations determined. Only the low-frequency $E_u$ mode has a
89: significant Pr-Cu interaction. Several possible mechanisms related to the
90: antiferromagnetism in this material are proposed to explain the sudden
91: appearance of this and other new spectral features at low temperature.
92: %
93: \end{abstract}
94: \pacs{74.25.Kc, 74.25.Gz, 74.72.Jt}%
95: \maketitle
96:
97: %
98: % Page header
99: %
100: %\makeatletter%
101: %\global\@specialpagefalse%
102: %\def\@oddhead{BNL \hfil }%
103: %\let\@evenhead\@oddhead%
104: %\makeatother
105:
106: %
107: % Introduction
108: %
109: \section{INTRODUCTION}
110: Within the family of high-temperature cuprate superconductors, the Ce-doped
111: series $R_{2-x}$Ce$_x$CuO$_{4-\delta}$, where $R$=Nd, Sm, Eu, Gd, etc., are the
112: only materials which appear to be electron doped.\cite{tokura89} The undoped
113: Nd$_2$CuO$_4$ and Pr$_2$CuO$_4$ are antiferromagnetic insulators, which become
114: ``bad metals'' with Ce doping until the sudden onset of superconductivity at
115: $x\approx 0.14$. The region of superconductivity in the electron-doped
116: materials is quite narrow ($x\approx 0.14-0.18$).\cite{takagi89a,takagi89b}
117: There is essentially no ``underdoped'' region for the superconductivity, with
118: the maximum value for the $T_c$'s of $23\,$K and $19\,$K in the Nd and Pr
119: systems, respectively, occurring at $x\approx 0.15$. At higher dopings $T_c$
120: decreases rapidly vanishing above $x\approx 0.18$. At the solubility limit
121: ($x=0.22$) the Nd system is metallic with no evidence of
122: superconductivity.\cite{hagen91} The optimally doped systems become
123: superconducting only after oxygen reduction\cite{izumi89,klamut93,kawashima94}
124: ($\delta\approx 0.01-0.03)$, and some transport measurements suggest that both
125: electrons and holes participate in the charge transport in the superconducting
126: phase.\cite{jiang94,fournier97} The role played by oxygen in these materials
127: may be more complex than in the hole-doped cuprates. It is the interesting
128: behavior of these superconducting systems that motivates an examination of the
129: optical properties of one of the parent compounds, Pr$_2$CuO$_4$.
130:
131: The $T^\prime$ structure of Pr$_2$CuO$_4$ is similar to the $T$ structure of
132: the hole-doped La$_{2-x}$Sr$_x$CuO$_4$; both structures are body-centered
133: tetragonal, space group $I4/mmm$ $(D_{4h}^{17})$ (Ref.~\onlinecite{muller75}).
134: These materials consist of two-dimensional sheets of copper-oxygen layers,
135: which define the {\it a-b} planes, with the {\it c} axis being perpendicular to
136: the planes. The $T$ and $T^\prime$ structures differ in the location of the
137: oxygen atoms between the copper-oxygen sheets.
138: %
139: In the $T$ structure the copper atoms have octahedral coordination, surrounded
140: by four oxygen atoms in the {\it a-b} plane, and two apical oxygens along the
141: {\it c} axis.
142: %
143: However, in the $T^\prime$ structure shown in Fig.~\ref{unitcell}, the apical
144: sites are empty and the out-of-plane oxygen atoms are not chemically bonded to
145: the copper atoms in the planes, which, as a result, have a square
146: coordination.\cite{crawford90a} While the difference between the $T$ and
147: $T^\prime$ structures results in different Raman active modes, the same number
148: of infrared active modes $3A_{2u}+B_{2u}+4E_u$ are expected for
149: each.\cite{crawford90a} (The doubly degenerate $E_u$ modes are active in the
150: {\it a-b} planes, the singly degenerate $A_{2u}$ modes are active only along
151: the {\it c} axis, and the $B_{2u}$ mode is silent.)
152:
153:
154: %
155: % Figure 1
156: %
157: \begin{figure}
158: \vspace*{-0.7in}%
159: \centerline{\includegraphics[width=2.7in]{figure1.eps}}%
160: \vspace*{-1.0in}%
161: \caption{The unit cell of the Pr$_2$CuO$_4$ in the $I4/mmm$ space group showing
162: the square planar coordination of the Cu-O(1) layers, as well as the corrugated
163: structure of the Pr-O(2) layers. The unit cell dimensions are $a=b=3.943$~\AA\
164: and $c=12.15$~\AA\ (Ref.~\onlinecite{sumarlin95}).}%
165: \label{unitcell}
166: \end{figure}
167:
168: %
169: % Cu ordering
170: %
171: The strong Cu-O bonding in the {\it a-b} plane of this material gives rise to
172: two-dimensional electronic and magnetic behavior. The weak out-of-plane
173: coupling induces long range antiferromagnetic (AFM) order in the Cu spins at
174: the relatively modest temperature of $T_{N,Cu}\approx 250-280$~K (Refs.
175: \onlinecite{cox89,matsuda90,sumarlin95}), which is similar to the values of
176: $T_{N,Cu}\approx 250-300$~K observed in Nd$_2$CuO$_4$
177: (Refs.~\onlinecite{matsuda90,endoh89}), Eu$_2$CuO$_4$
178: (Ref.~\onlinecite{chattopadhyay94}), and La$_2$CuO$_4$
179: (Refs.~\onlinecite{sun91,hayden91}).
180: %
181: % Pr ordering
182: %
183: The rare earth Pr ions carry localized $4f$ moments, which typically order at
184: very low temperature. However, from symmetry considerations, the Pr moments
185: should order below $T_{N,Cu}$ in this material.\cite{hill95} In fact, due to
186: the significant exchange interactions between the Pr ions which are mediated
187: through the copper-oxygen layers, there is a Pr contribution to the
188: susceptibility\cite{sumarlin95} below $\approx 200$~K.
189:
190: The optical properties of Nd$_2$CuO$_4$ have been investigated in ceramics and
191: single crystals.\cite{degiorgi89,heyen90,zhang91,lupi92,jandl95,jandl96,
192: strach97,calvani96} Single crystals of Pr$_2$CuO$_4$ have been studied at
193: either room temperature \cite{cooper90,tajima91,arima93,jandl97} or at low
194: temperature\cite{crawford90a,crawford90b} ($\approx 10$~K), but there has been
195: no detailed investigation of the temperature dependence of the optical
196: properties of this material. Furthermore, there is some disagreement in the
197: literature with respect to the vibrational parameters of the $E_u$ modes.
198:
199: %
200: % Figure 2
201: %
202: \begin{figure}
203: \centerline{\includegraphics[width=4.0in]{figure2.eps}}%
204: \vspace*{-0.5in}%
205: \caption{The temperature dependence of the {\it ab}-plane resistivity of
206: Pr$_2$CuO$_4$. The resistivity is changing over nearly six orders of magnitude
207: over the temperature interval. Inset: The resistivity vs $1/T$. This plot
208: shows that at high temperatures the resistivity is activated and follows the
209: form $\rho_{dc}=\rho_0 \exp(E_a/k_BT)$; a linear regression yields a value for
210: $E_a=1380\pm 50$~cm$^{-1}$ ($\approx 0.17$~eV) (dotted line).
211: Note the deviation from activated behavior at low temperature.}%
212: \label{resis}
213: \end{figure}
214:
215:
216: In this paper we report on the detailed optical properties of Pr$_2$CuO$_4$ at
217: a variety of temperatures. The conductivity at low-frequency has a strong
218: temperature dependence above room temperature, which when taken with optical
219: estimates of the gap suggests that the transport is due to variable-range
220: hopping due to localized states in the gap. Transport measurements support
221: this conclusion. A strong new vibration appears close to the low-frequency
222: $E_u$ mode below about 200~K, in addition to other weak vibrational structure.
223: A normal coordinate analysis of the zone-center vibrational modes indicates
224: that of all the $E_u$ modes, only the low-frequency mode involves a significant
225: Pr-Cu interaction. It is proposed that the AFM order in this material is
226: responsible for the lifting of the degeneracy of the low-frequency $E_u$ mode;
227: the absence of any interaction between the Pr and Cu atoms in the other $E_u$
228: modes limits this effect to just the low-frequency $E_u$ mode.
229:
230: %
231: % Figure 3
232: %
233: \begin{figure*}[t]
234: \vspace*{-2.0cm}%
235: \centerline{\includegraphics[width=4.8in,angle=270]{figure3.eps}}%
236: \vspace*{-0.60in}%
237: \caption{The reflectance of Pr$_2$CuO$_4$ for light polarized in the {\it a-b}
238: plane for temperatures between 15 and 390~K from $\approx 30$ to
239: $1500$~cm$^{-1}$. Above room temperature, the low frequency reflectance goes
240: rapidly to unity indicating a ``metallic'' behavior, while the rapid appearance
241: of the low-frequency oscillatory structure at low temperature is an indication
242: of the increasingly insulating and transparent nature of the sample. Inset:
243: The reflectance at 295~K over a much wider frequency range from $\approx 30$ to
244: over $16\,000$~cm$^{-1}$.}%
245: \label{reflec}
246: \end{figure*}
247:
248:
249: %
250: % Experimental section
251: %
252: \section{EXPERIMENTAL}
253: Single crystals of Pr$_2$CuO$_4$ were grown using a CuO-based direction
254: solidification technique.\cite{peng91} The crystals are thin platelets $\approx
255: 1.5\,$mm $\times 1.5\,$mm in the {\it a-b} plane, but quite thin along the {\it
256: c} axis ($\approx 50\,\mu$m). The crystals examined had a flat, mirror-like
257: surface which was free of flux or other residue. Transport and optical
258: measurements were performed on the same sample for consistency.
259: %
260: % Transport
261: %
262: The measurements of resistivity in the {\it ab}-plane were carried out using
263: standard four-probe configuration. Gold wires were attached to the specimen
264: directly using silver-loaded epoxy and followed by a spot annealing at the
265: contacts with focused laser beam. The whole specimen remained at room
266: temperature, except the areas of $\sim 70$~$\mu$m wide surrounding the
267: contacts. This process generally resulted in a contact resistance less than
268: 1~$\Omega$. With such a low contact resistance, without changing the oxygen
269: content in these samples, we were able to measure the resistivity over six
270: orders of magnitude from 90~K to 340~K with high accuracy (better than 0.1\%),
271: shown in Fig.~\ref{resis}. The inset shows the resistivity vs the inverse
272: temperature; the linear response at high temperature is an indication of
273: activated behavior, although this is no longer the case for $T\lesssim 160$~K.
274:
275: %
276: % Magnetization measurements
277: %
278: Magnetization measurements were performed on an unoriented 2~mg crystal of
279: Pr$_2$CuO$_4$ in an MPMS magnetometer at an applied field of 5~T. The behavior
280: of $1/\chi_M$ follows a Curie-Weiss form, and linear regression yields a
281: N\'{e}el temperature for Pr of $T_{N,Pr}\approx 44$~K with a Pr magnetic moment
282: of $3.2\,\mu_{\rm B}$. A careful survey of the temperature region in which the
283: antiferromagnetic ordering of copper moments is expected $(200\,{\rm K} <
284: T_{N,Cu} < 320\,{\rm K})$ reveals no apparent anomaly in the magnetization. A
285: feature comparable in size and definition to that observed by Sun {\it et
286: al.}\cite{sun91} for La$_2$CuO$_4$ would have been readily observable.
287: However, based on the scatter in our data, an antiferromagnetic anomaly reduced
288: in size by more than 50\% and/or smeared over a larger temperature region might
289: not be discernible. It is worth noting that the majority of studies of the
290: ordering of the Cu spins in these materials have been neutron scattering
291: measurements on substantially larger samples than the ones examined in this
292: work.
293:
294: For the optical measurements, crystals were mounted in a cryostat on an
295: optically-black cone. The temperature dependence of the reflectance was
296: measured at a near-normal angle of incidence from $\approx 30$ to over
297: $16\,000$~cm$^{-1}$ on a Bruker IFS 66v/S using an {\it in situ} overcoating
298: technique, which has previously been described in detail
299: elsewhere.\cite{homes93} This technique is especially useful when measuring
300: small samples, as it allows the entire face of the sample to be utilized. Above
301: $\approx 5000$~cm$^{-1}$ the reflectance is assumed to be temperature
302: independent.
303:
304: %
305: % Figure 4
306: %
307: \begin{figure*}[t]
308: \vspace*{-2.0cm}%
309: \centerline{\includegraphics[width=4.8in,angle=270]{figure4.eps}}%
310: \vspace*{-0.65in}%
311: \caption{The optical conductivity of Pr$_2$CuO$_4$ from $\approx 30$ to
312: $16\,000$~cm$^{-1}$ at 295~K for light polarized in the {\it a-b} plane. Inset:
313: the temperature dependence of the conductivity in the region of the four
314: infrared-active $E_u$ modes; while the conductivity is dominated by the lattice
315: vibrations, a broad incoherent background is observed to form rapidly above
316: room temperature. Note also the appearance of a new low-frequency vibrational
317: feature at low temperature at $\approx 145$~cm$^{-1}$ (arrow). }%
318: \label{conduc}
319: \end{figure*}
320:
321:
322: %
323: % Implications of transparency for KK analysis
324: %
325: The optical conductivity has been determined from a Kramers-Kronig analysis of
326: the reflectance, for which extrapolations for $\omega \rightarrow 0, \infty$
327: must be supplied. At low frequency, a metallic extrapolation was used for
328: $T\gtrsim 260$~K, $R \propto 1 - \sqrt{\omega}$; while below this temperature
329: the reflectance was assumed to continue smoothly to $\approx 0.45$ at zero
330: frequency. At high frequency, the reflectance was assumed to be constant above
331: the highest measured frequency point to $2\times 10^5$~cm$^{-1}$, above which a
332: free-electron ($R\propto w^{-4}$) behavior was assumed.
333: %
334: At low temperature, the semitransparent nature of the sample has implications
335: for the Kramers-Kronig analysis, which assumes specular reflectance from a
336: single surface only. While the absorption due to the lattice modes assures
337: that the reflectance in those regions is essentially that of the bulk material,
338: the same cannot be said for the high-frequency region. The presence of
339: multiple reflections leads to asymmetries in the line shapes and an optical
340: conductivity less than zero. As a result, for the calculation of the
341: low-frequency conductivity in the region of the lattice modes, the reflectance
342: has been truncated at $\approx 3000$~cm$^{-1}$ and assumed to be constant only
343: to $8000$~cm$^{-1}$, above which a free electron approximation has been
344: assumed. While the lineshapes in the conductivity are symmetric Lorentzians
345: whose positions and widths do not vary greatly with different choices for the
346: high-frequency extrapolations, the amplitudes are somewhat sensitive upon this
347: choice.
348:
349: %
350: % RESULTS - reflectance is shown in Figure 3
351: %
352: \section{RESULTS}
353: The reflectance of Pr$_2$CuO$_4$ for light polarized in the {\it a-b} plane is
354: shown in Fig.~\ref{reflec} from $\approx 20$ to 1500~cm$^{-1}$ at a variety of
355: temperatures, and in the inset from $\approx 30$ to $16\,000$~cm$^{-1}$ at room
356: temperature. The reflectance at low frequency is dominated by structure due to
357: the normally-active infrared modes, while the reflectance at high frequency is
358: relatively featureless, except for some structure at $\approx 11\,000$
359: cm$^{-1}$. The low-frequency vibration at $\approx 130$~cm$^{-1}$ is observed
360: to be a single mode at room temperature. However, at low temperature this mode
361: is resolved as a doublet. The low-frequency reflectance has an interesting
362: behavior; at low temperature the fringes indicate a lack of absorption due to
363: the insulating nature of the sample, but at room temperature and above the
364: fringes have vanished and the reflectance appears to be tending towards unity
365: as $\omega\rightarrow 0$; at 390~K the low frequency reflectance is over 80\%
366: and the system appears to be weakly metallic.
367:
368: %
369: % Figure 5
370: %
371: \begin{figure}[t]
372: \centerline{\includegraphics[width=4.0in]{figure5.eps}}%
373: \vspace*{-0.5in}%
374: \caption{The optical conductivity of Pr$_2$CuO$_4$ at low frequency for light
375: polarized in the {\it a-b} plane from 15 to 390~K showing the rapid formation
376: of the background conductivity, and the extrapolated values for $\rho_{dc} =
377: 1/\sigma_1(\omega\rightarrow 0)$; the different symbols denote the extrapolated
378: values at different temperatures. At low temperature a strong new vibrational
379: feature is observed (arrow). Inset: the extrapolated values for $\rho_{dc}$ vs
380: $1/T$; the linear regression of the points (dotted line) yields
381: $E_a=1630\pm 200$~cm$^{-1}$ ($\approx 0.2$~eV).}%
382: \label{lowfreq}
383: \end{figure}
384:
385: %
386: % Optical conductivity shown in Figure 4
387: %
388: The optical conductivity $\sigma_1(\omega)$ calculated from a Kramers-Kronig
389: analysis of the reflectance at 295~K is shown in Fig.~\ref{conduc}. The low
390: frequency conductivity is dominated by the infrared-active $E_u$ lattice modes,
391: but a careful examination shows a slight asymmetry in the lineshape of the
392: strongest mode. The weak feature in observed in the reflectance at $\approx
393: 11\,000$~cm$^{-1}$ signals the onset of absorption in the conductivity at about
394: $9000$~cm$^{-1}$, which peaks at about $12\,000$~cm$^{-1}$. The optical
395: conductivity in the region of the infrared active modes calculated using the
396: truncated reflectance is shown in the inset of Fig.~\ref{reflec}. The modes
397: have symmetric profiles, and the three high-frequency modes all harden and
398: narrow with decreasing frequency.
399:
400: %
401: % Activation plot shown in Figure 5
402: %
403: The low frequency conductivity is shown in detail in Fig.~\ref{lowfreq} from
404: $\approx 20$ to 180~cm$^{-1}$. At 390~K, there is a considerable amount of
405: background conductivity ($\approx 75$~$\Omega^{-1}$cm$^{-1}$) over most of the
406: observed frequency range, which decreases rapidly with decreasing temperature.
407: The inset in Fig.~\ref{lowfreq} shows the extrapolated value of the resistivity
408: $\rho_{dc}\approx 1/\sigma_1(\omega\rightarrow 0)$ vs $1/T$ ($T\gtrsim 250$~K);
409: the linear behavior indicates that the conductivity is strongly
410: activated.\cite{finite}
411:
412:
413: %
414: % Lattice modes - Figure 6
415: %
416: The behavior of the low-frequency mode is seen clearly in a more detailed plot
417: of the {\it ab}-plane reflectance shown in Fig.~\ref{detail} and the
418: conductivity in upper inset. The low-frequency mode can now clearly be
419: distinguished as fundamental at $\approx 130$~cm$^{-1}$. A new feature appears
420: quickly below room temperature, gaining oscillator strength monotonically with
421: decreasing temperature and hardening to $\approx 145$~cm$^{-1}$ at low
422: temperature.
423:
424: The slightly transparent nature of the sample requires that the reflections
425: from the back of the crystal be considered. This approach is described in
426: Appendix A, and the results of fits to the reflectance using this method are
427: listed in Table~I.
428:
429: %
430: % Table I
431: %
432: \begin{table*}
433: \caption{The results of fitting the phonon parameters to the
434: reflectance of Pr$_2$CuO$_4$ for light polarized in the {\it a-b}
435: plane at 295, 200 and 15~K, using the model for a lamellar
436: plate.\cite{back} A thickness of $d=50\,\mu{\rm m}$ and
437: $\epsilon_\infty = 6.5$ have been assumed. The parameters
438: $\omega_{TO,i}$, $\gamma_i$ and $\omega_{p,i}$ refer to the
439: frequency, width and effective plasma frequency of the $i$th
440: vibration. [All units are in cm$^{-1}$, except for the
441: dimensionless oscillator strength $S_i=\omega_{p,i}^2/
442: \omega_{TO,i}^2$.] }
443: %
444: \begin{ruledtabular}
445: \begin{tabular}{cccc c cccc c cccc}
446: \multicolumn{4}{c}{295~K} & & \multicolumn{4}{c}{200~K} & &
447: \multicolumn{4}{c}{15~K} \\
448: $\omega_{TO,i}$ & $\gamma_i$ & $\omega_{p,i}$ & $(S_i)$ & &
449: $\omega_{TO,i}$ & $\gamma_i$ & $\omega_{p,i}$ & $(S_i)$ & &
450: $\omega_{TO,i}$ & $\gamma_i$ & $\omega_{p,i}$ & $(S_i)$ \\
451: \cline{1-4} \cline{6-9} \cline{11-14}
452: %
453: % First phonon and anomaly
454: %
455: 130.6 & 4.5 & 161 & (1.51) &
456: %
457: $\left\{ \begin{array}{c} \\ \end{array} \right.$ \hspace*{-1.2cm}&
458: \begin{tabular}{c} 130.4 \\ 141.1 \end{tabular} &
459: \begin{tabular}{c} 3.3 \\ 9.9 \end{tabular} &
460: \begin{tabular}{c} 157 \\ 69 \end{tabular} &
461: \begin{tabular}{c} (1.45) \\ (0.24) \end{tabular} &
462: $\left\{ \begin{array}{c} \\ \end{array} \right.$ \hspace*{-1.2cm} &
463: %
464: \begin{tabular}{c} 128.2 \\ 146.1 \end{tabular} &
465: \begin{tabular}{c} 2.7 \\ 2.4 \end{tabular} &
466: \begin{tabular}{c} 144 \\ 124 \end{tabular} &
467: \begin{tabular}{c} (1.26) \\ (0.72) \end{tabular} \\
468: %
469: % Second phonon
470: %
471: 304 & 7.5 & 815 & (7.18) & \ &
472: 305 & 3.0 & 831 & (7.42) & \ &
473: 306 & 1.4 & 834 & (7.43) \\
474: %
475: % Third phonon
476: %
477: 331 & 17.4 & 469 & (2.01) & &
478: 333 & 7.7 & 458 & (1.89) & &
479: 341 & 3.4 & 431 & (1.59) \\
480: %
481: % Fourth
482: %
483: 490 & 27.3 & 511 & (1.08) & &
484: 491 & 19.5 & 508 & (1.07) & &
485: 495 & 8.4 & 503 & (1.03) \\
486: %
487: % Fifth phonon
488: %
489: & & & & &
490: 541 & 38 & 112 & (0.04) & &
491: 542 & 25 & 108 & (0.04) \\
492: %
493: % Strange phonon
494: %
495: & & & & &
496: 688 & 25 & 60 & (0.01) & &
497: 688 & 15 & 90 & (0.02) \\
498: \end{tabular}
499: \end{ruledtabular}
500: %
501: % Table notes, if any.
502: %
503: \label{phonons}
504: \end{table*}
505: %
506:
507:
508: %
509: % Discussion
510: %
511: \section{DISCUSSION}
512: %
513: % Discussion of the background conductivity
514: %
515: \subsection{Electronic properties}
516: The peak in the optical conductivity of Pr$_2$CuO$_4$ at $\approx
517: 12\,500$~cm$^{-1}$ in Fig.~\ref{conduc} has been observed in previous
518: work.\cite{lupi92,cooper90,arima93} The location of this feature is similar to
519: the onset of absorption in La$_2$CuO$_4$ and related materials\cite{perkins98}
520: at $\approx 11\,000$~cm$^{-1}$ ($\approx 1.4$~eV). The peak is characteristic
521: of a semiconducting band edge or a polaronic excitation. In either of these
522: cases, the {\it dc} resistivity is expected to be activated and to follow the
523: form\cite{alexandrov}
524: %
525: \begin{equation}
526: \rho_{dc}\propto e^{E_a/{k_BT}},
527: \label{active}
528: \end{equation}
529: %
530: where $E_a$ is either the half the optical gap $2\Delta$, or half the polaronic
531: level shift $E_p$. The low-frequency reflectance increases quickly above room
532: temperature, which is indicative of a ``metallic'' response in which the
533: reflectance goes to unity at zero frequency. This observation is realized in
534: the optical conductivity, where $\sigma_{dc} \equiv \sigma_1(\omega \rightarrow
535: 0)$ has a very strong temperature dependence for $T \gtrsim 300$~K. The plot of
536: $\rho_{dc}=1/\sigma_{dc}$ vs $1/T$ is shown in the inset in Fig.~\ref{lowfreq},
537: and is described quite well by Eq.~(\ref{active}); a linear regression yields
538: $E_a\approx 1630\pm 200$~cm$^{-1}$ ($\approx 0.2$~eV). The rather large error
539: associated with this estimate is due to the fact that the fit is limited to a
540: narrow interval in the high-temperature region. The transport data shown in
541: Fig.~\ref{resis} and the inset are also described quite well by activated
542: behavior; a linear regression applied in the high-temperature region ($T\gtrsim
543: 160$~K) yields $E_a=1380\pm 50$~cm$^{-1}$ ($\approx 0.17$~eV) and is shown as a
544: dashed line in the inset of Fig.~\ref{resis}; this value for $E_a$ is fairly
545: close to the value determined from the optical conductivity.
546: %
547: % Discussion with respect to polarons.
548: %
549: The optical conductivity of the nickelates, which have a similar appearance,
550: has been fitted using a small-polaron model,\cite{eklund93, katsufuji96,
551: alexandrov} which for $T\gg \omega/2$ produces an asymmetric Gaussian peak with
552: a maximum at $4E_a$. This model would imply that the peak in the conductivity
553: at high temperature should occur at $\approx 6000$~cm$^{-1}$, which is much
554: less than the observed value of $\approx 12\,500$~cm$^{-1}$. A rough estimate
555: of the direct optical gap may be made by extrapolating from the linear part of
556: the leading edge conductivity in Fig.~\ref{conduc} to the abscissa, which gives
557: $2\Delta\approx 9500$~cm$^{-1}$. If the transport were due to the carrier pair
558: density from thermal excitations across the gap, then $\sigma_{dc} \propto
559: n_i\approx \exp(-\Delta/k_BT)$. However, $E_a \ll \Delta$, suggesting that the
560: {\it dc} transport is due to variable range hopping between localized states
561: within the gap.\cite{mott}
562: %
563: % Figure 6
564: %
565: \begin{figure}
566: \centerline{\includegraphics[width=4.0in]{figure6.eps}}%
567: \vspace*{-0.5in}%
568: \caption{The reflectance of Pr$_2$CuO$_4$ for light polarized in the {\it a-b}
569: planes in the region of the low frequency infrared-active $E_u$ mode. The
570: vibration is a single mode at room temperature, but is resolved as a doublet at
571: low temperature. Upper inset: The optical conductivity which shows not only
572: the appearance of the new mode, but also the rapid increase in strength and
573: narrowing of this feature. Lower inset: The temperature dependence of the
574: fundamental mode (filled circles) and the new vibrational feature (open
575: circles).}%
576: \label{detail}
577: \end{figure}
578: %
579: % Arguments to support hopping
580: %
581: At low temperatures, the resistivity due to hopping is not activated, but
582: instead has the form
583: %
584: \begin{equation}
585: \rho_{dc}\propto e^{(T_0/T)^{1/n}},
586: \end{equation}
587: %
588: where $T_0$ is a characteristic temperature, and $n=d+1$ and $d$ is the
589: dimensionality of the system.\cite{mott} The transport measurements have been
590: performed over a wide enough range so that the departure from activated to
591: power-law behavior at low temperature is clearly visible in the insert in
592: Fig.~\ref{resis}. Different power laws were examined, but it appears that the
593: resistivity at low temperature is well described by $\rho_{dc}\propto
594: \exp{[(T_0/T)^{1/4}]}$, suggesting that the hopping is a three-dimensional
595: rather than a two-dimensional phenomena restricted to the CuO$_2$
596: planes.\cite{ambegaokar71} The localized states that result in hopping may
597: arise from defects which produced states within the gap. There is a very weak
598: feature in the optical conductivity in Fig.~\ref{conduc} at $\approx
599: 8000$~cm$^{-1}$ which occurs at $\approx E_a$ below the leading edge of the
600: absorption. This feature may suggest a possible origin for the localized
601: states. However, it not clear if the defects states responsible for this
602: feature are an intrinsic property of the sample, or due to extrinsic effects.
603:
604:
605: %
606: % Vibrational properties
607: %
608: \subsection{Vibrational properties}
609: %
610: \subsubsection{Normal coordinate analysis of the lattice modes}
611: %
612: The unusual behavior of the low-frequency $E_u$ mode requires a detailed
613: understanding not only of the nature of this particular mode, but of the other
614: $E_u$ modes as well. For this reason, a normal coordinate analysis of
615: Pr$_2$CuO$_4$ was undertaken.\cite{normal}
616:
617: The normal coordinate calculations were performed using Wilson's $GF$ matrix
618: method \cite{wilson,shimanouchi61} and a commercially available software
619: package.\cite{vibratz} Initially, a simple valence force field was adopted,
620: consisting of bond stretching and angle bending coordinates; the types and
621: values of the force constants used, and the internal coordinates to which they
622: correspond are listed in Table~II. The force field involves six bond
623: stretching force constants and four angle bending force constants. The bond
624: stretch and angle bends all involve oxygen bonding, with the exception of the
625: bond stretches denoted $f_4$ and $f_6$, which deal with the Pr-Pr and Pr-Cu
626: interactions; while the value for the Pr-Cu bond stretch is relatively small,
627: the refinement is significantly worse if it is omitted. While this initial
628: force field described the vibrations in the Cu-O(1) planes quite well, the
629: corrugated nature of the Pr-O(2) layers shown in Fig.~\ref{unitcell} leads to
630: difficulties in describing the restoring forces within this layer. The
631: agreement with the observed frequencies improved significantly when an
632: interaction force constant between the bond stretch and the angle bend in the
633: Pr-O(2) layer was introduced. The comparison between observed and calculated
634: frequencies, and the potential energy distribution (PED) are listed in
635: Table~III. The PED's gives the relative contribution of the force constants to
636: the potential energy of the normal modes. This treatment yields the atomic
637: displacements and PED for all of the normal modes. Each of the in-plane
638: infrared active $E_u$ modes merits further discussion.
639:
640: %
641: % Table II
642: %
643: \begin{table}
644: \caption{Force constants and internal coordinates for Pr$_2$CuO$_4$. The
645: labeling scheme of the atoms is shown in Fig.~1.}
646: \begin{ruledtabular}
647: \begin{tabular}{cclcc}
648: Force & Internal & & Distance (\AA )/ & \\
649: constant & coordinate & & Angle ($^\circ$) & Value$^a$ \ \\
650: \cline{1-5}
651: & & & & \\
652: \multicolumn{3}{c}{Bond stretch} & & \\
653: \cline{1-3}
654: $f_1$ & Cu-O(1) & [4]$^b$ & 1.98 & 0.355 \\
655: $f_2$ & Pr-O(2) & [8] & 2.12 & 0.532 \\
656: $f_3$ & Pr-O(1) & [8] & 3.03 & 0.785 \\
657: $f_4$ & Pr-Pr & [1] & 4.58 & 0.787 \\
658: $f_5$ & O(2)-O(2) & [4] & 2.80 & 0.300 \\
659: $f_6$ & Pr-Cu & [8] & 3.62 & 0.067 \\
660: & & & & \\
661: \multicolumn{3}{c}{Angle bend} & & \\
662: \cline{1-3}
663: $\alpha_1$ & O(1)-Cu-O(1) & [4] & 90 & 0.572 \\
664: $\alpha_2$ & O(1)-Cu-O(1)$^c$& [4] & 180 & 0.195 \\
665: $\alpha_3$ & O(2)-Pr-O(2) & [4] & 138 & 0.469 \\
666: $\alpha_4$ & O(1)-Pr-O(1) & [4] & 82 & 0.478 \\
667: & & & & \\
668: \multicolumn{3}{c}{Interaction} & & \\
669: \cline{1-3}
670: $i_1$ & $f_2\alpha_3$ & [16] & & 0.074 \\
671: \end{tabular}
672: \end{ruledtabular}
673: %
674: \footnotetext[1] {Bond stretches are in units of md/\AA , angle bends in md-\AA /rad.}%
675: \footnotetext[2] {The number of internal coordinates.}%
676: \footnotetext[3] {In plane and out of plane.}%
677: %
678: \end{table}
679:
680: %
681: % Descriptions of the Eu modes
682: %
683: % 490 cm-1 mode
684: %
685: The $E_u$ mode at 490~cm$^{-1}$ at room temperature involves primarily the
686: in-plane Cu-O angle bending resulting in displacements of the O(1) atoms. In
687: addition, there are small displacements of the atoms in the Pr-O(2) planes; the
688: motions of the Pr and O(1) atoms are coordinated. However, as Table~III shows,
689: there is no interaction between the rather small Pr and the Cu displacements.
690: In general, the eigenvectors determined from the normal coordinate analysis are
691: in good agreement with the calculated zone-center displacements shown in
692: Ref.~\onlinecite{zhang91}. This mode hardens to 495~cm$^{-1}$ at 15~K and
693: narrows dramatically from $\approx 25$ to $\approx 8$~cm$^{-1}$, as shown in
694: Table~I. The oscillator strength of this mode does not vary with temperature.
695: The slight asymmetry that is observed in the optical conductivity of the
696: high-frequency $E_u$ mode in Fig.~\ref{conduc} is most likely produced by the
697: uncertainty in the high-frequency extrapolation used in the Kramers-Kronig
698: analysis; fits to the reflectance are exact and do not suggest a large
699: asymmetry in this feature or others (Appendix A).
700:
701: %
702: % 331 cm-1 mode
703: %
704: The mode at 331~cm$^{-1}$ is a pure Pr-O(2) bond stretch, while the
705: 304~cm$^{-1}$ mode is almost a pure Cu-O(1) bond stretch (Table~III). The mode
706: at 331~cm$^{-1}$ hardens considerably to 341~cm$^{-1}$ at 15~K, and also
707: narrows from $\approx 17$ to 3.4~cm$^{-1}$.
708: %
709: % 304 cm-1 mode
710: %
711: The mode at $\approx 304$~cm$^{-1}$ hardens only slightly with decreasing
712: temperature, but narrows significantly from $\approx 8$~cm$^{-1}$ at room
713: temperature to 1.4~cm$^{-1}$ at 15~K; this is the strongest $E_u$ mode with an
714: oscillator strength of $S \approx 7.4$.
715:
716: %
717: % 130 cm-1 mode
718: %
719: The mode at 130~cm$^{-1}$ is a combination of Cu-O angle bending, resulting in
720: the in-plane displacements of the Cu and O(1) atoms, as well as the
721: out-of-phase motion of the Pr and O(2) atoms in the Pr-O layers; interestingly,
722: this mode has a significant coordination between the displacements of the Pr
723: atoms and the Cu and O(1) atoms, indicated in Table~III. The only other mode
724: that has a strong Pr-Cu interaction is the low-frequency {\it c}-axis $A_{2u}$
725: mode. The low-frequency $E_u$ mode softens to $\approx 128$~cm$^{-1}$ at low
726: temperature, while narrowing. The observed frequencies are in good agreement
727: with previous work,\cite{tajima91,crawford90b} however, the linewidths are all
728: narrower and there is only rough agreement with the reported strengths.
729:
730: %
731: % Subsubsection
732: %
733: % Discussion of the splitting of the Eu mode
734: %
735: \subsubsection{Origin of the low temperature doublet at $\approx 130$~cm$^{-1}$}
736: %
737: The most unusual feature in the reflectance spectra is striking appearance of a
738: new mode at low temperature, which is just above the low frequency $E_u$ mode.
739: The new mode is not simply a peculiar artifact due to fringes in the presence
740: of vibrational structure, as Fig.~\ref{model}(b) demonstrates (Appendix A).
741: While this feature has been previously observed at low
742: temperature\cite{crawford90a,crawford90b} the evolution of this new mode has
743: never been studied or fully explained. The new feature appears suddenly below
744: $\lesssim 250$~K at $\approx 140$~cm$^{-1}$, as seen in Fig.~\ref{detail} and
745: in the insets; this mode hardens dramatically and narrows with decreasing
746: temperature, gaining strength monotonically until it has almost the same
747: oscillator strength as the low-frequency $E_u$ mode (Table I). This feature
748: does not appear to evolve from a strong asymmetry or shoulder in the
749: low-frequency $E_u$ mode.
750: %
751: There is also additional very weak new fine structure in the low temperature
752: reflectance at $\approx 541$ and 688~cm$^{-1}$. These frequencies are well
753: above the high-frequency $E_u$ mode at $\approx 490$~cm$^{-1}$. A
754: symmetry-breaking process would allow Raman-active modes as well as the
755: longitudinal optic (LO) modes to become weakly active. However, the highest
756: observed frequency for a Raman mode is the $E_g$ mode at $\approx
757: 480$~cm$^{-1}$, suggesting that these features are not Raman modes. To
758: determine the positions of the LO modes, the reflectance and optical
759: conductivity have also been fit using a factorized form of the dielectric
760: function\cite{servoin80}
761: %
762: \begin{equation}
763: \tilde\epsilon(\omega)=\epsilon_\infty \prod_j {
764: {\omega_{LO,j}^2-\omega^2-i\gamma_{LO,j}\omega} \over
765: {\omega_{TO,j}^2-\omega^2-i\gamma_{TO,j}\omega} },
766: \end{equation}
767: %
768: % Positions of the LO modes and possible activity
769: %
770: where $\omega_{LO,j}$, $\omega_{TO,j}$, $\gamma_{LO,j}$ and $\gamma_{TO,j}$ are
771: the $j$th LO and the normally-active transverse optic (TO) modes, and LO and TO
772: damping, respectively. The fit at 15~K yields values for the TO modes which are
773: nearly identical to those shown in Table~I, and LO modes at 131, 149, 332, 430
774: and 588 cm$^{-1}$, respectively. None of the LO modes are close to the new
775: features observed at low temperature, making it unlikely that any new structure
776: is due to the LO modes.
777:
778: %
779: % Best explanation of the splitting
780: %
781: A more promising explanation of the splitting of the low frequency mode may
782: involve the fact that of all the $E_u$ modes, only the low-frequency mode
783: involves a significant Pr-Cu interaction. Magnetization measurements in this
784: work show that the Pr atoms are ordered at low temperature ($\lesssim 50$~K),
785: and neutron scattering measurements\cite{sumarlin95} show that there is at
786: least a partial ordering at higher temperatures, indicating the presence of a
787: Pr-Cu exchange interaction. In the case of the low-frequency $E_u$ mode, the
788: displacements of the Pr and Cu atoms are strongly correlated, suggesting that
789: the exchange interaction will have a significant effect upon the nature of this
790: vibration. This can lead to a symmetry-breaking process and the splitting of
791: the doubly-degenerate $E_u$ mode. This view is consistent with the observation
792: that this new feature is {\it not} present in Pr$_{1.85}$Ce$_{0.15}$CuO$_4$,
793: which is in the region of the phase diagram for this class of materials where
794: the AFM order is destroyed.\cite{heyen90} Because the other $E_u$ modes do not
795: involve strong Pr-Cu coupling, no exchange-induced splitting would be expected.
796: In fact, of all the other vibrational modes, only the low-frequency $A_{2u}$
797: mode has a significant Pr-Cu interaction (Table~III), but since this mode is
798: singly degenerate no splitting is expected, nor it is
799: observed.\cite{crawford90b}
800: %
801: % Similar behaviour in Nd2CuO4?
802: %
803: Similar behavior might also be expected in Nd$_2$CuO$_4$, where the Nd moments
804: order below 37~K. In fact, a weak feature is observed in the reflectance of
805: Nd$_2$CuO$_4$ at 10~K at $\approx 165$~cm$^{-1}$, quite close to the
806: low-frequency $E_u$ mode, and it has been suggested that this feature is indeed
807: magnetic in origin.\cite{heyen90} Kramers doublets have also been reported in
808: the Raman spectra at low temperature,\cite{dufour95} which have been attributed
809: to the removal of degeneracy by the Nd-Cu exchange interaction.\cite{jandl93}
810: While this is a reasonable explanation for the strong feature at low frequency,
811: it is less satisfactory for the fine structure observed at high frequency.
812:
813: %
814: % Table III
815: %
816: \begin{table}
817: \caption{Calculated and observed frequencies for the zone-center vibrations of
818: Pr$_2$CuO$_4$ at room temperature, and the potential energy distribution (PED).
819: (All frequencies are in cm$^{-1}$.) }
820: \begin{ruledtabular}
821: \begin{tabular}{cccl}
822: Symmetry & obs.$^a$ & cal. & PED$^b$ ($\geq 10$\% ) \\
823: \hline
824: $A_{1g}$ & 228 & 227 & $f_3(42), f_4(37), \alpha_3(17)$ \\
825: $B_{1g}$ & 328 & 326 & $f_2(28), \alpha_3(73)$ \\
826: $E_g$ & 480 & 481 & $f_2(46), f_5(54)$ \\
827: $E_g$ & 126 & 123 & $f_2(18), f_3(52), f_5(23)$ \\
828: %
829: \hline
830: $A_{2u}$ & 505 & 504 & $f_3(83)$ \\
831: $A_{2u}$ & 271 & 271 & $f_2(43), \alpha_3(113), i_1(-58)$ \\
832: $A_{2u}$ & 135 & 135 & $f_6(37), \alpha_2(58)$ \\
833: $B_{2u}$ & (silent) & 480 & $f_3(82)$ \\
834: %
835: $E_u$ & 490 & 490 & $f_3(60), \alpha_1(29), \alpha_2(10)$ \\
836: $E_u$ & 331 & 332 & $f_2(98)$ \\
837: $E_u$ & 304 & 303 & $f_1(93)$ \\
838: $E_u$ & 131 & 129 & $f_3(28), f_6(22), \alpha_1(30), \alpha_2(10)$ \\
839: %
840: \end{tabular}
841: \end{ruledtabular}
842: %
843: \footnotetext[1] {The observed frequencies for the $A_{2u}$ modes are taken
844: from Ref.~\onlinecite{tajima91}, while the Raman active modes are taken as
845: those for Nd$_2$CuO$_4$ from Ref.~\onlinecite{jandl93}. The frequencies for
846: the $E_u$ modes are from this work (Table I).}%
847: \footnotetext[2] {See Table~II for the identification of the force constants.}
848: %
849: \end{table}
850:
851: %
852: % Is this structure due to zone folding?
853: %
854: The new structure observed at $\approx 688$~cm$^{-1}$ at low temperature is
855: similar in frequency to structure that has been observed the reflectance in
856: other electron-doped materials\cite{calvani96,wang90} at $\approx
857: 690$~cm$^{-1}$. It has been suggested that much of the fine structure observed
858: in the cuprate systems are local vibrational modes due to polaron fine
859: structure.\cite{calvani96,lupi98} Alternatively, a number of weak features are
860: also seen below the N\'{e}el transition in CuO, including a mode at $\approx
861: 690$~cm$^{-1}$, which has been attributed to the activation of a zone-boundary
862: mode due to the reduction of the Brillouin zone resulting from the magnetic
863: order.\cite{chen95,kuzmenko01}
864: %
865: It is possible that the magnetic order is responsible for a weak structural
866: distortion and the activation of zone boundary modes due to the commensurate
867: reduction of the Brillouin zone. However, a difficulty with this explanation
868: is that any reduction of the Brillouin zone should result in a many new modes
869: being activated, rather than just the three that are observed.
870: %
871: One of the difficulties in explaining the vibrational fine structure observed
872: in this system is the failure to observe these results on a consistent basis in
873: different studies.\cite{calvani96,homes97} This raises the distinct
874: possibility that some of these features arise from impurities and/or phase
875: separation, ultimately making it difficult to identify a specific mechanism
876: responsible for the origin of the fine structure.
877: %
878: %The lattice dynamics of Nd$_2$CuO$_4$ have been calculated in detail,\cite{rampf93}
879: %and it is assumed that these results would be similar to those of
880: %Pr$_2$CuO$_4$. The dispersion in the planes of the $E_u$ modes along the
881: %$\Gamma - X$ direction ($\Sigma$ branches), indicates that the $\Sigma_1$ and
882: %the $\Sigma_4$ branches both harden at the zone boundary, making them good
883: %candidates for the new structure observed at low temperature at $\approx 146$
884: %and $\approx 540$~cm$^{-1}$. However, this explanation has several difficulties;
885: %first, this does not help to identify the structure at $\approx 690$~cm$^{-1}$
886: %which does not appear to be related to any of the $\Sigma$ branches.
887: %
888:
889:
890: %
891: % Conclusions
892: %
893: \section{Conclusions}
894: The strong temperature dependence of the low-frequency reflectance above
895: $\approx 250$~K results in a resistivity $\rho_{dc}\approx 1/
896: \sigma_1(\omega\rightarrow 0)$ which is exponential with temperature, following
897: the functional form $\rho_{dc} \propto \exp(E_a/k_BT)$, where $E_a\approx
898: 0.20$~eV; a more detailed transport measurement yields a slightly lower and
899: more accurate value of $E_a\approx 0.17$~eV. The fact that the transport gap
900: $E_a$ is much less than the inferred optical gap $2\Delta\approx 1.2$~eV, and
901: that transport measurements of the resistivity show a departure from activated
902: behavior for $T\lesssim 160$~K to a $\rho_{dc} \propto \exp[(T_0/T)^{1/4}]$
903: power-law behavior suggests that the transport in this material is due to
904: variable-range hopping between localized states in the gap.
905:
906: Due to the unusual vibrational structure in Pr$_2$CuO$_4$ at low temperature,
907: considerable attention was devoted to a discussion of the lattice modes. A
908: normal coordinate analysis has been performed to provide a detailed
909: understanding of the nature of the zone-center vibrations in this material. A
910: prominent feature close to the low-frequency $E_u$ mode develops at low
911: temperature, and additional weak fine structure is observed at high frequency.
912: These new features appear below $T_{N,Cu}$ at roughly the same temperature at
913: which the Pr moments are thought to begin to order ($\approx 200$~K).
914: %
915: Only the low-frequency $E_u$ mode involves a significant Pr-Cu interaction.
916: This suggests that the new vibrational structure is due to the
917: antiferromagnetic order in this material and the removal of degeneracy of this
918: mode.
919:
920: %
921: % Acknowledgements
922: %
923: \begin{acknowledgments}
924: We are grateful to A.R.~Moodenbaugh for performing the magnetization
925: measurements. We would also like to thank D.N.~Basov, V.J.~Emery,
926: J.L.~Musfeldt, M.~Strongin, T.~Timusk, J.M.~Tranquada, and J.J.~Tu for many
927: helpful discussions.
928: %
929: This work was supported by the Department of Energy under contract number
930: DE-AC02-98CH10886; work in Maryland is supported by the NSF Condensed Matter
931: Physics Division under grant No.~DMR 9732736. Research undertaken at NSLS was
932: supported by the U.S. DOE, Division of Materials and Chemical Sciences.
933: \end{acknowledgments}
934:
935:
936:
937: %
938: % Appendix A
939: %
940: \appendix
941: \section{Reflectance of a dielectric slab}
942: %
943: % Phonon fits - Figure 6
944: %
945: Due to the complications introduced by the slightly transparent nature of the
946: sample, the reflectance and the lattice modes have been fit using a model which
947: considers reflections from both the front and back surfaces of the crystal. The
948: frequency-dependent reflectance of a lamellar plate at a normal angle of
949: incidence is $R=\tilde{r}_l \tilde{r}_l^\ast$, where $\tilde{r}_l$ is defined
950: as
951: %
952: % Figure 7
953: %
954: \begin{figure}
955: \centerline{\includegraphics[width=4.0in]{figure7.eps}}%
956: \vspace*{-0.5in}%
957: \caption{(a) The reflectance in the infrared region of Pr$_2$CuO$_4$ for light
958: polarized in the {\it a-b} plane at room temperature (solid line) and the fit
959: to the data from Eq.~(1) (dotted line) using the vibrational parameters in
960: Table~I, with $\epsilon_\infty = 6.5$ and a sample thickness of $d=50$~$\mu$m.
961: The dashed line indicates the calculated behavior the (infinitely thick) bulk
962: material, which is identical to that of the model, indicating the opaque nature
963: of the sample.
964: %
965: (b) The reflectance at 15~K (solid line) with the model fit (dotted line) using
966: the vibrational parameters in Table~I, using the same values for $d$ and
967: $\epsilon_\infty$. The dashed line indicates the calculated behavior of the
968: bulk material. The doublet structure, fringe spacing and amplitude are in good
969: agreement with the reflectance. Note the absence of fringes in regions of
970: strong absorption.}%
971: \label{model}
972: \end{figure}
973: %
974: \begin{equation}
975: \tilde{r}_l={ {\tilde r \left[ 1-e^{i\omega 4\pi\tilde{n}d}\right]} \over
976: {1-r^2e^{i\omega 4\pi\tilde{n}d}} },
977: \end{equation}
978: %
979: where $d$ is the sample thickness, and $\tilde{r}$ is the Fresnel reflectance
980: of the bulk material,
981: \begin{equation}
982: \tilde{r}={ {1-\tilde{n}} \over {1+\tilde{n}} }.
983: \end{equation}
984:
985:
986:
987:
988: The complex refractive index is $\tilde{n}=n+ik$, which is related to the
989: complex dielectric function $\tilde\epsilon = \epsilon_1 + i\epsilon_2 =
990: \tilde{n}^2$, allowing the real and imaginary parts of the refractive index $n$
991: and $k$ to be determined
992: \begin{equation}
993: n= \left\{ {1\over 2} \left[ \sqrt{ \epsilon_1^2 + \epsilon_2^2 } +
994: \epsilon_1 \right] \right\}^{1/2},
995: \end{equation}
996: %
997: \begin{equation}
998: k= \left\{ {1\over 2} \left[ \sqrt{ \epsilon_1^2 + \epsilon_2^2 } -
999: \epsilon_1 \right] \right\}^{1/2}.
1000: \end{equation}
1001: %
1002: The optical properties of the bulk material are modeled using a series of
1003: Lorentzian oscillators
1004: %
1005: \begin{equation}
1006: \tilde\epsilon(\omega)=\epsilon_\infty+\sum_j {{\omega_{p,j}^2} \over
1007: {\omega_j^2-\omega^2-i\gamma_j\omega}},
1008: \label{lorentz}
1009: \end{equation}
1010: %
1011: where $w_j$, $\gamma_j$ and $\omega_{p,j}$ are the frequency, width and
1012: effective plasma frequency of the $j$th vibration, and $\epsilon_\infty$ is the
1013: core contribution to the dielectric function. The dimensionless oscillator
1014: strength is written as $S_j = \omega_{p,j}^2/ \omega_j^2$. The optical
1015: conductivity $\tilde\sigma(\omega) = \sigma_1(\omega) + i\sigma_2(\omega)$ is
1016: related to the complex dielectric function by $\tilde\sigma(\omega) = -i\omega
1017: \tilde\epsilon(\omega) /4\pi$.
1018:
1019: The phonon parameters were refined by a non-linear least-squares fit of the
1020: model for the reflectance of a thin dielectric slab to the experimental
1021: reflectance in Fig.~\ref{reflec}. The results at 295, 200 and 15~K are listed
1022: in Table~I, and a comparison of the experimental reflectance and the model
1023: results are shown in Figs.~\ref{model}(a) and \ref{model}(b) at 295 and 15~K,
1024: respectively. The broad, incoherent electronic background observed at room
1025: temperature is modeled by a zero-frequency Drude term, as well as two
1026: mid-infrared overdamped oscillators (Table~I). The results were then compared
1027: with fits to the features in the conductivity by assuming simple linear
1028: background in the region of the lattice modes, and were found to be in good
1029: agreement. This indicates that in the region of the phonon features where the
1030: sample is opaque, the Kramers-Kronig relation yields acceptable values for the
1031: conductivity.
1032:
1033:
1034:
1035: %
1036: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1037: %
1038: % References
1039: %
1040: \bibliography{pcoref}
1041:
1042: %
1043: \end{document}
1044: