1: \documentstyle[prl,aps]{revtex}
2: \begin{document}
3:
4: PACS Numbers: 74.20.-z, 74.62.Dh, 74.62.Yb, 74.62.-c, 74.72.Dn
5:
6: \vskip 4mm
7:
8: \centerline{\large \bf Isotope Effect in the Presence of Magnetic and
9: Nonmagnetic Impurities}
10:
11: \vskip 4mm
12:
13: \centerline{L. A. Openov and I. A. Semenihin}
14:
15: \vskip 2mm
16:
17: \centerline{\it Moscow State Engineering Physics Institute (Technical
18: University), 115409 Moscow, Russia}
19:
20: \vskip 2mm
21:
22: \centerline{R. Kishore}
23:
24: \vskip 2mm
25:
26: \centerline{\it Insituto Nacional de Pesquisas Espaciais, C.P. 515,
27: S. J. Campos, S. P., 12201-970, Brasil}
28:
29: \vskip 6mm
30:
31: \begin{quotation}
32:
33: The effect of impurities on the isotope coefficient is studied theoretically
34: in the framework of Abrikosov-Gor'kov approach generalized to account for
35: both potential and spin-flip scattering in anisotropic superconductors. An
36: expression for the isotope coefficient as a function of the critical
37: temperature is obtained for a superconductor with an arbitrary contribution
38: of spin-flip processes to the total scattering rate and an arbitrary degree
39: of anisotropy of the superconducting order parameter, ranging from isotropic
40: $s$ wave to $d$ wave and including anisotropic $s$ wave and mixed
41: ($s$+$d$) wave as particular cases. It is found that both magnetic
42: and nonmagnetic impurities enhance the isotope coefficient, the enhancement
43: due to magnetic impurities being generally greater than that due to
44: nonmagnetic impurities. From the analysis of the experimental results on LSCO
45: high temperature superconductor, it is concluded that the symmetry of the
46: pairing state in this system differs from a pure $d$ wave.
47:
48: \end{quotation}
49:
50: \vskip 6mm
51:
52: The isotope effect has played an important role in the development of phonon
53: mediated mechanism of electron pairing in superconductors. Its
54: discovery gave rise to the
55: theory of phonon mediated mechanism of electron pairing by Bardeen, Cooper
56: and Schrieffer \cite{BCS}. The BCS theory gave the isotope coefficient
57: $\alpha$, defined by the relation $T_{c}\propto M^{-\alpha}$,
58: or equivalently,
59: $\alpha = -\frac{\partial \ln T_{c}}{\partial \ln M}$,
60: equal to 1/2, in agreement with the experiments for mercury
61: \cite{max}. For simple superconducting metals like Hg, Zn, S, Pb,
62: etc., the values of $\alpha$ were found to be very close to the BCS value
63: 1/2. The deviations from the BCS theory found in superconducting transition
64: metals and their compounds were reasonably well explained by taking into
65: account the effects of Coulomb interactions \cite{morel},
66: left out in the BCS theory, and by more realistic treatments based on
67: Eliashberg equations \cite{carbotte}.
68:
69: The discovery of high temperature superconductors (HTSCs) brought a serious
70: challenge to the BCS theory which had firmly established the phonon mediated
71: mechanism of electron pairing as a dominant cause of superconductivity in
72: most previously known superconductors. For conventional superconductors, both
73: theory and experiment agree with the fact that $\alpha$ approaches the BCS
74: value of 1/2 as $T_{c}$ becomes larger. However, this trend is violated in
75: copper oxide HTSCs, where $\alpha$ has been found to vary with doping in
76: different ways \cite{Franck}. For the optimum doping, corresponding to the
77: highest critical temperature, the value of $\alpha$ is usually quite small
78: compared to the BCS value of 1/2. For a certain doping level, e.g. in some
79: La$_2$CuO$_4$ based HTSCs, the value of $\alpha$ becomes greater than 1/2
80: \cite{Franck}. Such a behavior of $\alpha$ has been considered as one of the
81: strongest evidence for a new mechanism of superconductivity in HTSCs.
82: However, this argument is not as strong as it appears. For example, it was
83: shown that, within the BCS picture or within the more realistic Eliashberg
84: approach, these deviations can be understood \cite{Kishore} by considering
85: the effects of anharmonicity, energy dependence of the electronic density of
86: states, pair breaking effects, isotope mass dependence of the carrier concentration,
87: etc.
88:
89: The presence of impurities strongly affects various characteristics of HTSCs,
90: including the isotope
91: coefficient \cite{soerensen,babushkina,morris}. Earlier theoretical attempts
92: to describe the isotope effect in impure HTSCs were based on either the
93: Abrikosov-Gor'kov formula for $T_c$ of an isotropic $s$-wave superconductor
94: containing magnetic impurities \cite{babushkina,kresin1} or the
95: Abrikosov-Gor'kov-like formula for $T_c$ of an anisotropic superconductor
96: that contains nonmagnetic impurities only \cite{buzea,scattoni}. An
97: increase in $\alpha$ with impurity concentration has been demonstrated,
98: in qualitative agreement with the experiment. However, the theory predicted
99: the universal dependence of the normalized isotope coefficient,
100: $\alpha/\alpha_0$, on the normalized critical temperature, $T_c/T_{c0}$,
101: where $\alpha_0$ and $T_{c0}$ are, respectively, the values of $\alpha$ and
102: $T_c$ in the absence of impurities. This prediction does not agree with the
103: experimental findings \cite{soerensen,babushkina,morris}.
104: It was shown by Kresin {\it et al.} \cite{kresin1} that the universal
105: behavior of $\alpha/\alpha_0$ versus $T_c/T_{c0}$ is restricted to the case
106: that magnetic impurities are the only cause for the increase of
107: $\alpha/\alpha_0$,
108: and that other effects like the nonadiabaticity of the apex oxygen in
109: YBa$_2$Cu$_3$O$_7$ could break the universality. Here we provide an
110: alternative explanation of how the nonuniversality of the dependence of
111: $\alpha/\alpha_0$ on $T_c/T_{c0}$ arises. Our approach is based on taking
112: into account the combined effect of both nonmagnetic and magnetic
113: scatterers on the critical temperature of a superconductor with anisotropic
114: superconducting order parameter.
115:
116: To derive the expression for the isotope coefficient, we make use of the
117: equation for $T_c$ of an anisotropic superconductor
118: containing both nonmagnetic and magnetic impurities \cite{openov}:
119: \begin{equation}
120: \ln\left(\frac{T_{c0}}{T_c}\right)=(1-\chi)\left[\Psi\left(\frac{1}{2}+%
121: \frac{1}{2\pi T_c\tau_m^{ex}}\right)-\Psi\left(\frac{1}{2}\right)\right]+%
122: \chi\left[\Psi\left(\frac{1}{2}+\frac{1}{4\pi T_c\tau}\right)-%
123: \Psi\left(\frac{1}{2}\right)\right],
124: \label{Tc}
125: \end{equation}
126: where $\Psi$ is the digamma function; $\tau_m^{ex}$ is the electron
127: relaxation time due to exchange scattering by magnetic impurities; $\tau$ is
128: the total electron relaxation time due to potential scattering by both
129: magnetic and nonmagnetic impurities, as well as due to exchange scattering by
130: magnetic impurities.
131: Eq. (\ref{Tc}) was obtained in Ref. \cite{openov} within the weak-coupling
132: limit of the BCS model in the framework of Abrikosov-Gor'kov approach.
133: It generalizes the well-known expressions \cite{Abrikosov1,Abrikosov2} for
134: the critical temperature of impure superconductors to the case of combined
135: effect of both nonmagnetic and magnetic impurities on the critical
136: temperature of anisotropic superconductors.
137: The coefficient
138: $\chi=1-\frac{\langle\Delta({\bf p})\rangle^2_{FS}}%
139: {\langle\Delta^2({\bf p})\rangle_{FS}}$
140: quantifies the degree of
141: anisotropy of the order parameter $\Delta({\bf p})$ on the Fermi surface (FS),
142: where the angular brackets $\langle ... \rangle_{FS}$ stand for a FS average.
143: For isotropic $s$-wave pairing, $\langle\Delta({\bf p})\rangle^2_{FS}%
144: =\langle\Delta^2({\bf p})\rangle_{FS}$, and hence $\chi=0$. For a
145: superconductor with $d$-wave pairing we have $\chi=1$ since
146: $\langle\Delta({\bf p})\rangle_{FS}=0$. The range $0<\chi<1$ corresponds to
147: anisotropic $s$-wave or mixed $(d+s)$-wave pairing. The higher the
148: anisotropy of $\Delta({\bf p})$ (e.g., the greater the partial
149: weight of a $d$-wave in the case of mixed pairing), the closer to unity is
150: the value of $\chi$. Note that in two particular cases of ($i$) magnetic
151: scattering in an isotropic $s$-wave superconductor ($\chi =0$) and ($ii$)
152: nonmagnetic scattering only in a superconductor with arbitrary in-plane
153: anisotropy of $\Delta({\bf p})$ ($1/\tau_m^{ex}=0, 0\leq\chi\leq 1$),
154: Eq. (\ref{Tc}) reduces to the well-known expressions
155: \cite{Abrikosov1,Abrikosov2}.
156:
157: It is convenient to specify the relative contribution of spin-flip
158: scattering rate, $1/\tau_m^{ex}$, to the total scattering rate, $1/\tau$, by
159: the dimensionless parameter $\gamma$ defined as
160: $\frac{1}{\tau_m^{ex}}=\gamma\frac{1}{\tau}$.
161: The greater is the relative contribution from exchange scattering by magnetic
162: impurities to $1/\tau$, the higher is the value of $\gamma$ ($\gamma$ ranges
163: from 0 in the absence of exchange scattering to 1 in the absence of
164: non-spin-flip scattering). In general, the value of $\gamma$ depends on the
165: scattering strengths of individual nonmagnetic and magnetic impurities, as
166: well as on their concentrations. At relatively low doping level, one can
167: expect $\gamma$ to depend only on the type of the host material and doping
168: elements, not on the impurity concentration \cite{openov}.
169:
170: Differentiating Eq. (\ref{Tc}) for $T_c$ with respect to the isotopic mass
171: $M$ under a reasonable assumption that electron relaxation times and the
172: anisotropy coefficient $\chi$ do not depend on $M$, taking the definition of
173: the parameter $\gamma$ into account and using the definition of the isotope
174: coefficient $\alpha$, one has
175: \begin{equation}
176: \frac{\alpha}{\alpha_0} = \left[1-
177: (1-\chi)\frac{\gamma}{2\pi T_c\tau}%
178: \Psi^{\prime}\left(\frac{1}{2}+\frac{\gamma}{2\pi T_c\tau}\right)-%
179: \chi\frac{1}{4\pi T_c\tau}%
180: \Psi^{\prime}\left(\frac{1}{2}+\frac{1}{4\pi T_c\tau}\right)\right]^{-1}.
181: \label{alpha/alpha_0}
182: \end{equation}
183: Equation (\ref{alpha/alpha_0}) is obviously more general than equations
184: used previously for the analysis of the pair breaking effect on the isotope
185: coefficient in HTSCs \cite{babushkina,kresin1,buzea,scattoni}.
186: Indeed, Eq. (\ref{alpha/alpha_0}) does not rely on particular assumptions
187: about the symmetry of the superconducting state and the nature of impurities
188: ({\it either} magnetic {\it or} nonmagnetic). This equation can be used for
189: an impure superconductor with {\it arbitrary} anisotropy of the
190: superconducting order parameter and {\it arbitrary} relative contributions of
191: potential and spin-flip scattering to the electron relaxation time. Such an
192: approach is of particular importance for HTSCs doped with various chemical
193: elements since, first, there is a strong evidence for a dominant $d$-wave
194: (i.e., highly anisotropic) order parameter in HTSCs \cite{tsuei2} with a
195: subdominant $s$-wave component \cite{schurrer}, and, second, a
196: lot of experiments give evidence for the presence of magnetic scatterers
197: (along with nonmagnetic ones) in doped HTSCs
198: \cite{Xiao,Mahajan,Williams,Awana}.
199:
200: At given values of $\chi$ and $\gamma$, Eqs. (\ref{Tc}) and
201: (\ref{alpha/alpha_0}) define the dependence of $\alpha/\alpha_0$ on
202: $T_c/T_{c0}$. One can see from
203: Eqs. (\ref{Tc}) and (\ref{alpha/alpha_0}) that the dependence of
204: $\alpha/\alpha_0$ on $T_c/T_{c0}$ has a universal shape for both an isotropic
205: $s$-wave superconductor ($\chi=0$) with nonzero contribution of exchange
206: scattering to the total scattering rate ($0<\gamma\le 1$) and a $d$-wave
207: superconductor ($\chi=1$) with an arbitrary ratio of spin-flip and potential
208: scattering rates ($0\le\gamma\le 1$), as it is seen from Fig. 1.
209:
210: Quite a different picture is realized in the case $0<\chi<1$, i.e., for a
211: mixed $(d+s)$-wave or an anisotropic $s$-wave superconductor.
212: In this case the behaviour of $\alpha/\alpha_0$ as a
213: function of $T_c/T_{c0}$ essentially depends on the value of $\gamma$. As
214: $T_c/T_{c0}$ goes to zero, i.e., in dirty superconductors, the value of
215: $\alpha/\alpha_0$ tends to $1/(1-\chi)$ in the absence of exchange scattering
216: ($\gamma=0$), while $\alpha/\alpha_0$ grows as
217: $\alpha/\alpha_0 \propto T_{c0}/T_c$ in the case $\gamma\ne 0$ for all values
218: of $\chi$. Thus, the universality of $\alpha/\alpha_0$ versus $T_c/T_{c0}$
219: curve breaks down for $0<\chi<1$. This is consistent with experimental
220: observations. Indeed, the studies of isotope effect in impure HTSCs
221: \cite{soerensen,babushkina,morris} show that the curves of $\alpha/\alpha_0$
222: versus $T_c/T_{c0}$ vary with the type of impurities, i.e., with the value of
223: $\gamma$ (since different chemical elements contribute differently to
224: spin-flip and potential scattering rates of charge carriers). So, our results
225: seem to be in a qualitative agreement with the experiment if one assumes that
226: the superconducting state in HTSCs differs from a pure $s$ or $d$-wave, i.e.,
227: $\chi\ne 0$ and $\chi\ne 1$, as is also indicated by several experimental
228: observations \cite{schurrer,klemm}.
229:
230: However, our theoretical consideration doesn't account for a
231: number of factors that could influence the isotope coefficient in impure
232: superconductors, e.g., the energy dependent electronic density of states,
233: nonadiabaticity, anharmonicity, etc. In particular,
234: the change in the carrier concentration $n_h$ upon chemical substitution can
235: have a strong influence on $T_c$ (and hence on $\alpha$) along with the
236: pair-breaking effect. This is why, to compare our calculations with the
237: experiment, we have taken the data on the isotope effect in the system
238: La$_{1.85}$Sr$_{0.15}$Cu$_{1-x}$M$_x$O$_4$ (M = Ni, Zn, Co, Fe), see
239: Ref. \cite{babushkina}. In this system, the critical temperature decreases
240: rapidly as the impurity content $x$ increases, and falls down to
241: $\approx 0.3T_{c0}$ already at $x=0.02-0.03$.
242: At such low impurity concentration, the change in $T_c$ due to
243: the change in $n_h$ can be neglected in the first approximation (as compared
244: with the pair-breaking effect) since the value of $T_{c0}\approx 40$K in the
245: impurity-free HTSC La$_{1.85}$Sr$_{0.15}$CuO$_4$ corresponds to the maximum
246: on the curve $T_c(n_h)$, and hence changes insignificantly with $n_h$ as long
247: as the change in $n_h$ is small. Note that such an approach (i.e., ignoring
248: the change in the carrier concentration upon doping) may not be appropriate
249: in case of YBa$_2$(Cu$_{1-x}$Zn$_x$)$_3$O$_7$ where $T_c/T_{c0}\approx 0.3$
250: at $x=0.06$ and all the more for Y$_{1-x}$Pr$_x$Ba$_2$Cu$_3$O$_7$ where
251: $T_c/T_{c0}\approx 0.3$ at $x=0.5$, see Ref. \cite{soerensen}.
252: A reasonable explanation of the isotope effect in these HTSCs has been given
253: by Kresin {\it et al.} \cite{kresin1} who considered the interplay between
254: the changes in the carrier concentration and nonadiabaticity of the apex
255: oxygen.
256:
257: Fig.2 shows the experimental data along with theoretical graphs calculated
258: for $\chi=0.5$ and different values of $\gamma$, ranging from 0 to 1. One can
259: see that at low impurity content $x<0.008$ ($T_c/T_{c0}>0.75$) there is a
260: good agreement between the theory and the experiment. However, at higher
261: values of $x$ (i.e., at lower values of $T_c/T_{c0}$) the experimental points
262: lie well above the theoretical curves for all impurity elements, except for
263: M = Ni. The same is true for other values of $\chi$ since the upper curve
264: in Fig. 2 (for $\gamma=1$) changes insignificantly with $\chi$, at least for
265: $T_c/T_{c0}>0.2$.
266:
267: To bring the theory closer to the agreement with the experiment, one can
268: assume that the value of $T_{c0}$ in the impurity-free HTSC
269: La$_{1.85}$Sr$_{0.15}$CuO$_4$ is strongly depressed relative to its
270: "intrinsic" value \cite{Kresin2} because of the scattering of charge
271: carriers by inhomogenities produced by substitution of Sr for La. Recently it
272: was suggested \cite {baskaran} that the anomalous response of the
273: anisotropic superconducting
274: state to the development of low energy dynamical charge stripes
275: can also cause the suppression of the "intrinsic" $T_{c0}$. Taking $T_{c0}$
276: to be equal to its "intrinsic" value, it is possible to reach a qualitative
277: agreement with the experiment on the isotope effect in impure
278: HTSCs even for heavily doped samples with low $T_c$. Since the value of
279: "intrinsic" $T_{c0}$ in La$_{1.85}$Sr$_{0.15}$CuO$_4$ is not known
280: {\it a priori}, we take the suggested value \cite{baskaran} of 90 K.
281:
282: Fig. 3 shows the results for the set of $T_{c0}=90$K and $\chi=0.5$. For such
283: a choice of the values of $T_{c0}$ and $\chi$, the theoretical curves for
284: $\gamma=0-1$ are closer to the experimental data \cite{babushkina}. Since
285: the value of $\alpha_0$ at "intrinsic" $T_{c0}$ is not known, we assumed for
286: simplicity that the value of $\alpha$ is the same at $T_{c0}\approx$ 40 K and
287: 90 K. This assumption can be the reason for a discrepancy between the
288: theory and the experiment in the region $T_c/T_{c0}=0.3-0.4$, see Fig. 3,
289: since one could expect the value of $\alpha$ at $T_{c0}=90$K to be somewhat
290: lower than at $T_{c0}\approx 40$K. It should be noted that in our theory
291: different types of impurities correspond to different values of $\gamma$. The
292: magnetic elements Fe and Co can be thought of being characterized by
293: $\gamma$ in the range 0.1 - 1, while Ni, whose magnetic moment is reduced
294: considerably by doping \cite{babushkina}, can be assigned somewhat lower
295: value of $\gamma\approx 0.01$, see Fig. 3. Similiarly, the doping by the
296: nonmagnetic element Zn induces magnetic moments
297: \cite{Xiao,acquarone}. One can see from Fig. 3 that Zn can be
298: assigned the value of $\gamma$ in the range 0.01 - 0.05, i.e. lower than in
299: the case of Fe and Co but greater than in the case of Ni.
300:
301: Finally, we note that the agreement between the theory and the experiment
302: becomes worse if one takes the value of $\chi$ closer to unity. This implies
303: that the symmetry of the pairing state in La-based HTSCs can differ
304: considerably from a pure {\it d}-wave. As far as we know, up to now there
305: were only indirect arguments in favor of {\it d}-wave symmetry of the
306: superconducting state in La$_{1.85}$Sr$_{0.15}$CuO$_4$ \cite{Takanaka}, while
307: phase-sensitive experiments are unknown to us. The value of $\chi\approx 0.5$
308: is however close to that expected for the two-dimensional order parameter of
309: the type $\Delta ({\bf k})$ = $\Delta_0$ [cos($k_x a$)+cos($k_y a$)],
310: see Ref. \cite{Openov2}.
311:
312: In conclusion, we have studied theoretically the effect of both magnetic and
313: nonmagnetic impurities on the isotope coefficient in the framework of a
314: generalized Abrikosov-Gorkov approach for the anisotropic superconductors. We
315: have shown how the interplay between the potential and spin-flip impurity
316: scattering gives rise to the nonuniversal dependence of $\alpha/\alpha_0$
317: versus $T_{c}/T_{c0}$ in mixed $(d+s)$ wave or anisotropic $s$ wave
318: superconductors. Our main result is that if the impurities are viewed as the
319: only cause for the increase in the isotope coefficient in
320: La$_{1.85}$Sr$_{0.15}$Cu$_{1-x}$M$_x$O$_4$, then the symmetry of the
321: superconducting order parameter in La$_{1.85}$Sr$_{0.15}$CuO$_4$ appears to
322: be different from a pure $d$-wave.
323: Indeed, even if one takes into account relatively large
324: experimental errors in the values of $\alpha$, it is clearly seen that
325: experimental points do not lie on a single "universal" curve of
326: $\alpha/\alpha_0$ versus $T_c/T_{c0}$ as one could expect in the case of a pure
327: $d$-wave symmetry of the superconducting state. According to our calculations,
328: the difference in $\alpha/\alpha_0$ versus $T_{c0}/T_c$ curves for different
329: impurity elements can be attributed to different contributions from the
330: exchange scattering to the total scattering rate of charge carriers in the
331: mixed $(d+s)$-wave or anisotropic $s$-wave superconducting state. The agreement
332: with the experiment is much more better if one assumes that the "intrinsic"
333: value of $T_{c0}$ in La$_{1.85}$Sr$_{0.15}$CuO$_4$ reaches $\approx 90$ K, more
334: than twice greater than the experimental value $\approx 40$ K.
335: It would be interesting to check if it is possible to explain the
336: nonuniversality of $\alpha/\alpha_0$ versus $T_c/T_{c0}$ in
337: La$_{1.85}$Sr$_{0.15}$Cu$_{1-x}$M$_x$O$_4$ within a pure $d$-wave framework.
338:
339: \vskip 2mm
340:
341: L.A.O and I.A.S acknowledge the support from the Russian Federal Program
342: "Integration", projects No A0133 and No A0155. A part of the work was carried
343: out during the stay of R. K. at ICTP, Trieste, Italy as a Research Associate.
344:
345: \vskip 6mm
346:
347: \begin{references}
348:
349: \bibitem{BCS} J. Bardeen, L. N. Cooper, and J. R. Schrieffer,
350: Phys. Rev. {\bf 108}, 1175 (1957).
351:
352: \bibitem{max} E. Maxwell, Phys. Rev. {\bf 78}, 477 (1950).
353:
354: \bibitem{morel} P. Morel and P. W. Anderson,
355: Phys. Rev. {\bf 125}, 1263 (1962);
356: N. N. Bogolyubov, V. V. Tolmachev, and D. V. Shirkov,
357: {\it A New Method in the Theory of Superconductivity},
358: Cons. Bureau, New York (1959).
359:
360: \bibitem{carbotte} J. P. Carbotte, Rev. Mod. Phys. {\bf 62}, 1027 (1990).
361:
362: \bibitem{Franck} J. P. Franck, in {\it Physical Properties of High
363: Temperature Superconductors}, edited by D. M. Ginsberg (World Scientific,
364: Singapore, 1994 ), Vol. IV, p. 189; A. Bill, V. Z. Kresin, and S. A. Wolf,
365: in {\it Pair Correlations in Many-Fermion Systems} (Plenum Press, New York,
366: 1998).
367:
368: \bibitem{Kishore} R. Kishore, in {\it Studies of High Temperature
369: Superconductors}, edited by A. V. Narlikar (Nova Science Pub.,
370: New York, 1999), Vol. 29, p. 23.
371:
372: \bibitem{soerensen} G. Soerensen and S. Gygax, Phys. Rev. B {\bf 51},
373: 11848 (1995).
374:
375: \bibitem{babushkina} N. A. Babushkina {\it et al.},
376: Physica C {\bf 272}, 257 (1996).
377:
378: \bibitem{morris} D. E. Morris {\it et al.}, Physica C {\bf 298}, 203 (1998).
379:
380: \bibitem{kresin1} V. Z. Kresin {\it et al.},
381: Phys. Rev. B {\bf 56}, 107 (1997).
382:
383: \bibitem{buzea} C. Buzea {\it et al.}, Physica C {\bf 270}, 317 (1996).
384:
385: \bibitem{scattoni} M. Scattoni, C. Grimaldi, and L. Pietronero,
386: Europhys. Lett. {\bf 47}, 588 (1999).
387:
388: \bibitem{openov} L. A. Openov, Pis'ma Zh. \'Eksp. Teor. Fiz. {\bf 66},
389: 627 (1997) [JETP Lett. {\bf 66}, 661 (1997)];
390: Phys. Rev. B {\bf 58}, 9468 (1998).
391:
392: \bibitem{Abrikosov1} A. A. Abrikosov and L. P. Gor'kov, Zh. \'Eksp. Teor.
393: Fiz. {\bf 35}, 1558 (1958); {\bf 36}, 319 (1959); {\bf 39}, 1781 (1960) [Sov.
394: Phys. JETP {\bf 8}, 1090 (1959); {\bf 9}, 220 (1959); {\bf 12}, 1243 (1961)].
395:
396: \bibitem{Abrikosov2} A. A. Abrikosov, Physica C {\bf 214}, 107 (1993).
397:
398: \bibitem{tsuei2} C. C. Tsuei {\it et al.}, Nature (London) {\bf 387}, 481 (1997).
399:
400: \bibitem{schurrer} I. Sch\"urrer, E. Schachinger, and J. P. Carbotte,
401: J. Low Temp. Phys. {\bf 115}, 251 (1999).
402:
403: \bibitem{Xiao} G. Xiao {\it et al.},
404: Phys. Rev. B {\bf 42}, 8752 (1990).
405:
406: \bibitem{Mahajan} A. V. Mahajan {\it et al.},
407: Phys. Rev. Lett. {\bf 72}, 3100 (1994).
408:
409: \bibitem{Williams} G. V. M. Williams, J. L. Tallon, and R. Meinhold,
410: Phys. Rev. B {\bf 52}, 7034 (1995).
411:
412: \bibitem{Awana} V. P. S. Awana {\it et al.}, Mod. Phys. Lett. B {\bf 10}, 619 (1996).
413:
414: \bibitem{klemm} R. A. Klemm, C. T. Rieck, and K. Scharnberg,
415: Phys. Rev. B {\bf 61}, 5913 (2000).
416:
417: \bibitem{Kresin2} V. Z. Kresin, S. A. Wolf, and Yu. N. Ovchinnikov,
418: Phys. Rev. B {\bf 53}, 11831 (1996); J. Supercond. {\bf 9}, 431 (1996).
419:
420: \bibitem{baskaran} G. Baskaran, Cond-mat/9910161 (1999).
421:
422: \bibitem{acquarone} M. Acquarone, in
423: {\it High Temperature Superconductivity - Models and Measurements},
424: Ed. M. Acquarone (World Scientific Press, Singapore, 1996),
425: p.281 and p.335.
426:
427: \bibitem{Takanaka} K. Takanaka and K. Kuboya, Phys. Rev. Lett. {\bf 75},
428: 323 (1995).
429:
430: \bibitem{Openov2} L. A. Openov, V. F. Elesin, and A. V. Krasheninnikov,
431: Physica C {\bf 257}, 53 (1996).
432:
433: \end{references}
434:
435: \newpage
436:
437: \centerline{\bf FIGURE CAPTIONS}
438:
439: \vskip 2mm
440:
441: Fig. 1. Universal dependence of the normalized isotope coefficient
442: $\alpha/\alpha_0$ on the normalized critical temperature $T_c/T_{c0}$ in an
443: impure isotropic $s$-wave superconductor ($\chi=0$) with a finite
444: concentration of magnetic scatterers and an impure $d$-wave superconductor
445: ($\chi=1$) with an arbitrary ratio of spin-flip and potential scattering
446: rates.
447:
448: Fig. 2. Same as in Fig. 1 for $\chi=0.5$ (a specific case of anisotropic
449: pairing) for different values of the coefficient $\gamma$ specifying the
450: relative contribution to the total scattering rate from exchange scattering.
451: $\gamma=0$ (dot-dashed curve), 0.01 (thin solid curve), 0.05 (dashed curve),
452: 0.1 (dotted curve), 1 (thick solid curve). Experimental data from
453: Ref. \cite{babushkina} for isotope effect in
454: La$_{1.85}$Sr$_{0.15}$Cu$_{1-x}$M$_x$O$_4$ with different
455: $x$ and M = Ni (triangles), Zn (open squares); Co (closed circles);
456: Fe (closed squares). Experimental values of $T_c$ as a function of $x$ are
457: normalized to the value of $T_{c0}$ = 37.5 K at $x=0$.
458:
459: Fig. 3. Same as in Fig. 2 for $T_{c0}$ = 90 K, see text for details.
460:
461: \end{document}
462: