1: %\documentclass[preprint,prb,showpacs,floatfix]{revtex4}
2: \documentclass[twocolumn,prb,showpacs,floatfix]{revtex4}
3:
4: \usepackage{graphicx}
5: \usepackage{dcolumn}
6: \usepackage{float}
7: \usepackage{amsmath}
8:
9: \newcommand{\comment}[1]{}
10: \newcommand{\cm}{cm$^{-1}$\/}
11:
12: \begin{document}
13:
14: %\preprint{}
15:
16: \title{Optical self-energy of superconducting Pb in the THz region}
17: % repeat the \author\address pair as needed
18: \author{T. Mori$^1$}
19:
20: \author{E.J. Nicol$^2$}
21: \email{nicol@physics.uoguelph.ca}
22:
23: \author{S. Shiizuka$^{1}$}
24: \author{K. Kuniyasu$^{3}$}
25: \author{T. Nojima$^{3}$}
26: \author{N. Toyota$^{1}$}
27: \author{J.P. Carbotte$^{4}$}
28:
29: \affiliation{$^{1}$ Physics Department, Graduate School of Science,
30: Tohoku University, Sendai 980-8578, Japan}
31: \affiliation{$^2$Department of Physics, University of Guelph,
32: Guelph, Ontario N1G 2W1 Canada}
33: \affiliation{$^{3}$ Institute for Materials Research and Center for
34: Low Temperature Science,
35: Tohoku University, Sendai 980-8577, Japan}
36: \affiliation{$^4$Department of Physics and Astronomy, McMaster
37: University, Hamilton, Ontario N1G 2W1 Canada}
38:
39: \date{\today}
40:
41:
42: \begin{abstract}
43: New THz data on the optical conductivity of Pb are presented as well as
44: a detailed Eliashberg analysis with particular emphasis on phonon-assisted
45: processes not included in a BCS approach. Consideration of the optical
46: self-energy instead of the conductivity itself helps highlight the differences
47: with BCS predictions. Predicted coherence peaks are observed in the optical
48: scattering rates. Impurities enhance the optical effective mass at zero
49: frequency by an order of magnitude and induce a large peak at twice
50: the gap in agreement with theory. This work illustrates the usefulness of
51: the optical self-energy for the analysis of data.
52: \end{abstract}
53: % insert suggested PACS numbers in braces on next line
54: \pacs{74.25.Gz,74.20.Fg,74.25.Nf,74.25.Kc}
55:
56: \maketitle
57: % body of paper here
58:
59: \section{Introduction}
60:
61: Optical properties of superconducting Pb have been measured in the past
62: by several groups,
63: including Palmer and Tinkham\cite{palmer}, who compared their data
64: with BCS predictions. Joyce and Richards\cite{joyce} saw phonon
65: structure in the absorptivity and Farnworth and Timusk\cite{farnworth}
66: analysed similar data to recover the electron-phonon spectral density
67: $\alpha^2F(\omega)$. This latter
68: work was based on approximate equations for the conductivity
69: derived by Allen\cite{allen} from Fermi golden rule considerations.
70: Good agreement with the tunneling-derived electron-phonon spectral density
71: was found\cite{mcmillan}. More elaborate theories of the conductivity
72: based on the Eliashberg equations and a Kubo formula for the current-current
73: correlation function were formulated by Nam\cite{nam} and further
74: elaborated upon in many places\cite{schrieffer} including Scalapino\cite{scalapino} and Marsiglio and Carbotte\cite{marsigliorev,marsiglio}. In this paper
75: we report new high quality THz measurements on Pb.
76: In addition to the new data, we
77: provide a detailed theoretical analysis
78: based on numerical solutions of the Eliashberg equations with
79: the known $\alpha^2F(\omega)$~\cite{mcmillan,rowell}. We emphasize particularly
80: the
81: phonon-assisted
82: processes not included in BCS and describe
83: deviations that these processes introduce.
84: BCS deals only with the coherent part of the single particle Green's function.
85: In Pb, the electron-phonon mass renormalization factor $\lambda=1.55$. In
86: this case, only $1/(1+\lambda)=1/2.55$
87: of the spectral weight resides in the coherent part
88: and $\lambda/(1+\lambda)=1.55/2.55$
89: is found in the incoherent part which describes the
90: phonon-assisted processes.
91:
92: \section{Methods}
93:
94: The Pb films are deposited onto the (0001) surface of a single-crystalline
95: sapphire substrate held at room temperature. Four-terminal resistance measurements
96: on films of 30, 50, and 480 nm in thickness show the superconducting
97: transitions around 7.2K with the width of $\pm 50$ mK, indicating little
98: degradation in $T_c$ from the bulk. The present TDTS (Time Domain TeraHertz
99: Spectroscopy) measurements\cite{Nuss1998} are carried out with the use of a
100: commercial spectrometer (RT-20000, Tochigi Nikon Co. Ltd) based on standard
101: transmission techniques, covering the frequency range of 0.15-2.0 THz. The real
102: part $\sigma_1(\omega)$ and the imaginary part $\sigma_2(\omega)$ of the
103: optical conductivity $\sigma(\omega)=\sigma_1(\omega)+i\sigma_2(\omega)$
104: are directly evaluated by numerically analyzing both the amplitude and
105: phase of transmitted terawaves through the substrate (0.5 mm in thickness)
106: with and without the film. The data will be presented for the 50 nm film most
107: systematically studied.
108:
109: For a correlated electron system, specifically here a coupled,
110: electron-phonon system, it has become standard practice to represent
111: the optical conductivity $\sigma(T,\omega)$ [with $T$, the
112: temperature,
113: and $\omega$, the photon energy] in terms of an optical self-energy in
114: analogy to the quasiparticle self-energy which is introduced to
115: include
116: interactions in the one-particle Green's function. By definition,
117: \begin{equation}
118: \sigma(\omega)=i\frac{\omega_p^2}{4\pi}\frac{1}{\omega-2\Sigma^{op}(\omega)},
119: \label{eq:sig2}
120: \end{equation}
121: where $\omega_p$ is the plasma frequency.
122: The imaginary part of the optical self-energy gives the optical
123: scattering rate
124: $1/\tau^{op}(T,\omega)=-2\Sigma_2^{op}(T,\omega)$
125: and its real part gives the optical effective mass ($m^*_{op}(T,\omega)$)
126: according to
127: $\omega[m^*_{op}(T,\omega)/m-1]=-2\Sigma_1^{op}(T,\omega)$.
128: Further, $m^*_{op}(T,\omega)/m-1$ is the optical mass renormalization
129: $\lambda^{op}(T,\omega)$. Here, $m$ is the bare electron mass.
130: Traditionally, it has been common to display in optical papers
131: the real and imaginary part of $\sigma(T,\omega)$ but in the recent
132: literature, it has been recognized that much can be learned from the
133: scattering rate and effective mass renormalization, particularly for
134: comparison
135: with quasiparticle properties. In terms of $\sigma(T,\omega)$
136: \begin{eqnarray}
137: \frac{1}{\tau^{op}(\omega)} &=&\frac{\omega_p^2}{4\pi}
138: {\rm Re}\biggl(\frac{1}{\sigma(T,\omega)}\Biggr)=
139: \frac{\omega_p^2}{4\pi}\frac{\sigma_1}{\sigma_1^2+\sigma_2^2} ,\label{eq:tau}\\
140: \omega\frac{m^*_{op}(T,\omega)}{m} &=&-\frac{\omega_p^2}{4\pi}
141: {\rm Im}\Biggl(\frac{1}{\sigma(T,\omega)}\Biggr) =
142: \frac{\omega_p^2}{4\pi}\frac{\sigma_2}{\sigma_1^2+\sigma_2^2} .
143: \label{eq:lam}
144: \end{eqnarray}
145:
146: The Drude conductivity is recovered when all interactions are
147: neglected
148: and only impurity scattering is included through a frequency and
149: temperature independent scattering rate $1/\tau^{imp}$. In this case,
150: the real part of $\Sigma^{op}(T,\omega)$ which must be related to
151: its imaginary part by Kramers-Kronig is zero. For coupling
152: to a single oscillator of energy $\omega_E$ in the normal state
153: at zero temperature\cite{CSH}
154: \begin{equation}
155: \frac{1}{\tau^{op}(T=0,\omega)}=\frac{2\pi A}{\omega}
156: (\omega-\omega_E)\theta(\omega-\omega_E) ,
157: \label{eq:taueinstein}
158: \end{equation}
159: where $A$ is the coupling between electrons and phonons and
160: \begin{equation}
161: \lambda^{op}(\omega)=-\frac{2A}{\omega}\biggl[\ln\biggl|\frac{\omega_E+\omega}{\omega_E-\omega}\biggr|+\frac{\omega_E}{\omega}\ln\biggl|\frac{\omega_E^2-\omega^2}{\omega_E^2}\biggr|\biggr] .
162: \end{equation}
163: For $\omega\to 0$, $\lambda^{op}(\omega)=2A/\omega_E\equiv\lambda$, the
164: quasiparticle electron-phonon mass renormalization. In this model,
165: $\sigma(\omega)$
166: has two parts. One part is
167: a delta function at $\omega=0$ coming from the region
168: $\omega<\omega_E$ in (\ref{eq:taueinstein}) where there is no scattering. If
169: impurity scattering is included, the $\delta$-function broadens into a
170: Lorentzian
171: \begin{equation}
172: \sigma_L(\omega)=i\frac{\omega_p^2}{4\pi}\frac{1}{\omega(m^*_{op}/m)
173: +i/\tau^{imp}}.\label{eq:siglor}
174: \end{equation}
175: The second part which is finite
176: only for $\omega>\omega_E$ is a boson-assisted part:
177: \begin{equation}
178: \sigma_B(\omega)=i\frac{\omega_p^2}{4\pi}\biggl[\omega\biggl(\frac{m^*_{op}(\omega)}{m}\biggr)
179: +i\biggl(\frac{1}{\tau^{imp}}+\frac{2\pi A}{\omega}(\omega-\omega_E)\biggr)\biggr]^{-1} .\label{eq:sigbos}
180: \end{equation}
181: Further, if we approximate in (\ref{eq:siglor}) $m^*_{op}(\omega)/m$
182: by
183: $1+\lambda$, its zero frequency limit which is exact for small
184: $\omega$,
185: we obtain a coherent Drude part
186: \begin{equation}
187: \sigma_D(\omega)=i\frac{\omega_p^2}{4\pi}\frac{1}{\omega(1+\lambda)
188: +i/\tau^{imp}} ,\label{eq:sigdru}
189: \end{equation}
190: with optical spectral weight of $\omega_p^2/[8(1+\lambda)]$
191: and effective scattering rate of $1/[\tau^{imp}(1+\lambda)]$. The
192: remaining
193: spectral weight is to be found in the incoherent piece
194: (\ref{eq:sigbos})
195: and is $\omega^2_p\lambda/[8(1+\lambda)]$ which follows on application
196: of the optical sum rule
197: \begin{equation}
198: \int_{0}^\infty\sigma_1(\omega)d\omega=\frac{\omega_p^2}{8}.
199: \end{equation}
200:
201: \section{Results}
202:
203: \begin{figure}[t]
204: %\vspace*{-0.8 cm}%
205: \centerline{\includegraphics[width=4.30in]{fig1.eps}}%
206: %\vspace*{-1.8 cm}%
207: \caption{(color online) Real and imaginary part of the optical
208: conductivity, $\sigma_1(T,\omega)$ and $\sigma_2(T,\omega)$,
209: respectively, versus $\omega$ at various reduced temperatures $t=T/T_c$.
210: The points
211: are the data and the lines are theory.
212: }
213: \label{fig1}
214: \end{figure}
215: In Fig.~\ref{fig1}, we present our experimental (points) and
216: theoretical (lines) results for 6 temperatures as labelled
217: by reduced temperature $t=T/T_c$. In
218: the top frame, we show the real part of the conductivity
219: $\sigma_1(\omega)$
220: in $\Omega^{-1}\rm{cm}^{-1}$ versus $\omega$ up to a frequency
221: of $2\times 10^{12}$ Hz while in the lower frame
222: we show the imaginary part $\sigma_2(\omega)$.
223: For the real part, we note a kink in both data and theory at an intermediate
224: frequency which decreases in frequency,
225: with increasing temperature. This arises because there are
226: two separate absorption processes possible. At the lowest frequencies there can
227: be direct absorption from the thermally excited quasiparticles. This contribution
228: provides a Drude-like peak to $\sigma_1(\omega)$. At frequencies above
229: twice the gap, direct absorption through the creation of two excitations out
230: of the condenstate becomes possible. In addition, there is another incoherent
231: phonon-assisted contribution to the absorption and this is the part of
232: most interest in this work. However, it cannot easily be
233: separated out of the real part of the conductivity. We will see that
234: consideration of the optical self-energy $\Sigma^{op}$ defined in
235: Eq.~(\ref{eq:sig2}) helps in this regard. First, however, we note that the fit
236: between theory and the data is very good at low frequency in all cases.
237: We have used the data in the normal state ($t=1.02$) to determine
238: the elastic scattering rate $1/\tau^{imp}$ and the plasma frequency
239: $\omega_p$.
240: Turning next to the imaginary part of $\sigma$ shown in the lower frame of
241: Fig.~\ref{fig1}, we note the same level of agreement and
242: the points for $t=1.02$
243: which are in the normal state also show good agreement with the theory.
244: In general, the superconducting properties of
245: Pb, while very different from those predicted by BCS, have been remarkably
246: well-described by Eliashberg theory\cite{carbotte} over a broad range
247: of experiments
248: and our new data shown here poses no exception. This is important given that
249: in the past, it is has been argued that vertex corrections and other
250: aspects of electromagnetic theory were needed to understand the data in
251: Pb and yet this appears not to be the case here. With this level of agreement
252: between theory and data, we should now be able to proceed to test for more
253: subtle effects due to the phonon-assisted processes in the absorption.
254:
255: To see separately, more clearly, the role of the coherent absorption
256: processes included in BCS theory and described by the coherent
257: part of the electron Green's function $G$, and the incoherent
258: phonon-assisted processes described by the incoherent part of $G$,
259: it is convenient to deal with the real and imaginary parts of the
260: optical self-energy defined in Eq.~(\ref{eq:sig2}). Equations~(\ref{eq:tau})
261: and (\ref{eq:lam}) are for the optical scattering rate and effective mass,
262: respectively. These are shown in Fig.~\ref{fig2}.
263: \begin{figure}[t]
264: %\vspace*{-0.8 cm}%
265: \centerline{\includegraphics[width=4.30 in]{fig2.eps}}%
266: %\vspace*{-1.8 cm}%
267: \caption{(color online) The optical scattering rate $1/\tau^{op}(T,\omega)$
268: and mass renormalization $[1+\lambda^{op}(T,\omega)]$ vs $\omega$.
269: The points are experiment and the lines are theory. The data for
270: $t=1.02$ are in the normal state just above $T_c$ and the other two
271: data sets are below $T_c$ in the superconducting state.
272: }
273: \label{fig2}
274: \end{figure}
275: The top frame gives
276: $1/\tau^{op}(\omega)$ and the bottom $[1+\lambda^{op}(\omega)]$
277: at three temperatures. To agree with the experimental data it was
278: necessary to include an impurity scattering rate of 22 meV. This is
279: the only scattering that would be included in a BCS theory for
280: which the normal state conductivity would be a pure Drude form. The
281: optical scattering rate corresponding to this Drude is a constant
282: flat line independent of frequency and temperature. We see clearly in
283: the curve for the normal state that significant deviations from
284: this constant occur. These are due to the boson-assisted processes which
285: become even more important as the frequency is increased beyond the
286: range of the data shown. This is shown in the top frame of Fig.~\ref{fig3},
287: where we present our theoretical results over an extended frequency range
288: up to 36 meV, well beyond the end of the phonon spectrum. For clarity,
289: only the $t=0.52$ (superconducting state) and $t=1.02$ (normal state)
290: curves are shown.
291: \begin{figure}[t]
292: %\vspace*{-0.8 cm}%
293: \centerline{\includegraphics[width=4.30 in]{fig3.eps}}%
294: %\vspace*{-1.8 cm}%
295: \caption{(color online) The calculated optical scattering rate
296: $1/\tau^{op}(T,\omega)$
297: and optical mass renormalization $[1+\lambda^{op}(T,\omega)]$ for a large
298: energy range.
299: Results are for our Pb calculation with
300: impurity scattering rate of 22 meV. Red (thin line)
301: is in the superconducting state
302: and blue (thick line)
303: is for the normal state, both at $t=0.52$. We compare with BCS
304: (dot-dashed and dashed black curves) and pure Pb (dotted curves). The vertical arrows indicate
305: the peaks in $\alpha^2F(\omega)$ shifted by $2\Delta$.
306: }
307: \label{fig3}
308: \end{figure}
309: Starting with the normal state (thick blue curve), we note a large deviation from
310: the horizontal dashed black line at 22 meV for
311: elastic impurity scattering. The additional
312: frequency-dependent scattering, due to boson-assisted processes, adds on
313: to $1/\tau^{imp}$ according to Eq.~(\ref{eq:taueinstein}) (in the case
314: of a delta function spectral density). For $\omega\to\infty$ this contribution
315: will saturate at $2\pi A$ (see Eq.~(\ref{eq:taueinstein})) even for a
316: distributed $\alpha^2F(\omega)$ spectrum.
317:
318: Turning next to the superconducting state (thin red curve, top frame),
319: we see larger boson structures with the vertical arrows indicating the
320: transverse and longitudinal peaks in the phonon frequency distribution
321: $F(\omega)$ of Pb shifted by $2\Delta$. These are preceded by a coherence
322: peak in the region above $2\Delta$ which reflects the increased density of
323: final states for scattering in the superconducting as compared with
324: normal state. This curve is strikingly different from the dot-dashed
325: black curve
326: which would apply to a BCS superconductor.
327: An interesting
328: comparison is offered in the dotted curves which were also calculated
329: with the Pb $\alpha^2 F(\omega)$ spectrum, but without residual scattering,
330: i.e. $1/\tau^{imp}=0$. An important point to note is that no coherence peak
331: is seen in the dotted thin red curve because this is an impurity effect, but
332: at higher frequencies the dotted curve is very close to the solid
333: one once it is displaced upward by 22 meV. This holds exactly for the
334: normal state data, blue dotted (pure) and solid (with residual scattering).
335: Returning to the lower frame of Fig.~\ref{fig2}, we show the optical
336: effective mass renormalization $[1+\lambda^{op}(T,\omega)]$.
337: The solid blue dots apply to the normal state at $t=1.02$ just above $T_c$.
338: Note that as $\omega\to 0$ $\lambda^{op}\simeq 1.7$ which is just
339: slightly larger than the zero temperature quasiparticle
340: mass renormalization for Pb, which is 1.55. This is expected
341: since it is known that the mass renormalization increases slightly
342: with increasing temperature. Furthermore, the normal state
343: effective mass has only a small frequency dependence predicted and
344: this is observed.
345: There is no dependence on impurity scattering as it drops out of this
346: curve. (This fact can be understood simply. The Kramers-Kronig (KK)
347: transform of a constant is zero so that
348: a constant scattering rate corresponds to no mass renormalization.) This is in
349: striking contrast to the superconducting state for which impurities
350: profoundly alter the optical curves. This is seen in the other two curves
351: for $t=0.52$ (red hexagons) and $t=0.82$ (black squares). Note first the very
352: large peaks which occur at twice the gap edge $2\Delta$. These peaks
353: have not been emphasized before. They would not be there in the clean
354: case (no impurities). Their origin is easily understood from our
355: results for the scattering rate shown in the top frame. We see the sharp,
356: almost vertical, rise
357: in $1/\tau^{op}(\omega)$ at $2\Delta$ with the energy scale of the rise
358: given by the value of $1/\tau^{imp}=22$ meV. The KK transform of a step at $\omega_E$ is
359: a logarithm with a singularity at $\omega_E$. The larger the step, the larger the log
360: singularity. Of course, here we do not quite have a vertical step and hence the
361: corresponding peak in the effective mass is smeared. Another
362: feature to note is the $\omega\to 0$ limit of $\lambda^{op}$.
363: We note that Eq.~(\ref{eq:lam}) can be
364: rewritten as
365: \begin{equation}
366: \frac{m^*_{op}(T,\omega)}{m} =
367: \frac{\omega_p^2}{4\pi}\frac{\omega\sigma_2}{(\omega\sigma_1)^2+(\omega\sigma_2)^2}
368: \label{eq:lam2}
369: \end{equation}
370: and that in the superconducting state in the limit of $\omega\to 0$
371: we get
372: \begin{equation}
373: \frac{m^*_{op}(\omega=0)}{m} =
374: \frac{\omega_p^2}{4\pi}\lim_{\omega\to
375: 0}\frac{1}{\omega\sigma_2}=\frac{n}{n_s},
376: \end{equation}
377: where $n$ is the electron density in the normal state
378: and $n_s(T)$ is the superfluid density at temperature $T$.
379: For $T=0$, direct strong-coupling calculations of $n_s(T=0)$ show
380: that it is very nearly equal to $n/(1+\lambda)$ which leads to
381: $m^*_{op}(T=0,\omega=0)/m\simeq1+\lambda$,
382: same as for the normal state, and reflects the fact that only the
383: coherent part of the optical conductivity participates in the
384: superfluid
385: condensation. Returning to Eq.~(\ref{eq:lam2}), one can verify that
386: in the normal state, substitution of the Drude form (\ref{eq:sigdru})
387: again gives $m^*_{op}/m=1+\lambda$
388: as we expect. If impurities are included, as we have done in our work,
389: $n_s$ is no longer $n/(1+\lambda)$ but in the dirty limit
390: has instead the approximate form (in renormalized BCS)\cite{marsigliorev}
391: \begin{equation}
392: n_s\simeq\frac{n}{1+\lambda}\biggl[\frac{\pi}{2\alpha}
393: -\frac{1}{2\alpha\sqrt{\alpha^2-1}}
394: \ln\biggl|\frac{\alpha+\sqrt{\alpha^2-1}}{\alpha-\sqrt{\alpha^2-1}}
395: \biggr|\biggr]
396: \end{equation}
397: for $\alpha>1$ where $\alpha=1/[2\Delta\tau^{imp}(1+\lambda)]$,
398: which gives $1+\lambda^{op}(T=0,\omega=0)=8.3$.
399: The exact numerical result (Fig.~\ref{fig2}, bottom frame) is 9.5
400: at $T/T_c=0.52$. Most of the difference can be assigned to the small
401: reduction in $n_s(T)$ between $T=0$ and $t=T/T_c=0.52$.
402:
403: In the lower frame of Fig.~\ref{fig3}, we show additional theoretical
404: results on the optical effective mass which range over a larger frequency interval
405: beyond that covered by the THz data. The dot-dashed
406: black curve is the BCS result
407: (for $1/\tau^{imp}=22$ meV) while the solid thin red curve includes as well
408: the incoherent phonon-assisted Holstein processes. We see clearly in
409: this curve an image of the Pb phonon frequency distribution with transverse
410: and longitudinal phonon peaks shifted by $2\Delta$ indicated by vertical arrows.
411: The thin red dotted lines were obtained for pure Pb (i.e. $1/\tau^{imp}$ is set
412: equal to zero and the only source of scattering comes from
413: the inelastic part described by $\alpha^2F(\omega)$). In this
414: case there is no peak in $[1+\lambda^{op}]$ at twice the gap but at
415: higher frequencies the dotted and solid curves agree. For the normal
416: state at $t=1.02$, the solid thick blue curve applies.
417: Note
418: that for $\omega$ larger than a few times the phonon energy, normal and
419: superconducting curves merge and all tend toward 1, i.e. $\lambda^{op}(\omega)
420: \to 0$ as $\omega\to \infty$. For BCS in the normal state, $\lambda^{op}=0$
421: for all $\omega$ as shown by the long-dashed curve.
422:
423: \section{A Sum Rule}
424:
425: Recently, there have
426: been several discussions of sum rules on $1/\tau^{op}$ in the
427: literature\cite{marsiglio02,chubukov,abanov}. In particular,
428: it was noted that there exists a differential sum rule on the difference
429: in optical scattering rate between the superconducting and normal state.
430: Defining the partial sum $I(\omega)$ as
431: \begin{equation}
432: I(\omega)=\int_0^\omega\biggl[\frac{1}{\tau^{op}_S(\omega^\prime)}-\frac{1}
433: {\tau^{op}_N(\omega^\prime)}\biggr]d\omega^\prime
434: \label{eq:I}
435: \end{equation}
436: it can be shown\cite{chubukov} that $I(\Omega\to\infty)=0$
437: for any value of $1/\tau^{imp}$ and $\Delta$. The energy scale for arriving
438: at the asymptotic limit is set by $\Delta$ even in the dirty limit. As defined,
439: in BCS\cite{chubukov} $I(\Omega)$ is negative and has its maximum absolute
440: value at $2\Delta$ after which it begins to decrease towards zero, but always
441: remaining negative. Our results for the case of Pb are shown in Fig.~\ref{fig4}
442: and deviate significantly from the BCS behaviour just described.
443: \begin{figure}[t]
444: \vspace*{-2.8 cm}%
445: \centerline{\includegraphics[width=4.30 in]{fig4.eps}}%
446: %\vspace*{-1.8 cm}%
447: \caption{(color online) The difference
448: $(1/\tau^{op}_S-1/\tau^{op}_N)$ vs $\omega$ shown up to 35 meV
449: and the partial sum, defined as $I(\omega)$ in the text. The
450: black dashed curves are for the BCS case.
451: In the inset, we compare data with theory.
452: }
453: \label{fig4}
454: \end{figure}
455: The solid thick blue curve gives the result for $1/\tau^{op}_S(\omega)-
456: 1/\tau^{op}_N(\omega)$ at $t=0.52$ and shows a rapid increase
457: through 0 at twice the gap, after which a coherence peak is seen followed
458: by boson structure beyond which the curve crosses the axis again.
459: The solid thin red curve is the result for the partial sum
460: $I(\omega)$. The dashed black curves are the BCS results for comparison.
461: The crossing of the horizontal axis at a frequency slightly above
462: the end of the phonon spectrum in the case for Pb (thin red curve) is due to the
463: inelastic scattering processes and is not seen in the BCS curve. Note also the large positive
464: peak.
465:
466:
467: \section{Conclusions}
468:
469: We have presented new THz data on the optical conductivity of Pb and compared
470: them with Eliashberg results with an emphasis on phonon-assisted (Holstein)
471: processes not part of BCS. We find that these show up most clearly
472: in the optical self-energy rather than in the conductivity itself.
473: In particular, the impurity component of the total scattering is seen
474: directly in the optial scattering rate as a large constant which can easily be
475: subtracted out to reveal directly the phonon-assisted part. In the normal
476: state impurity scattering drops out of the optical effective mass
477: leaving only phonon renormalizations which show excellent agreement between
478: theory and experiment. By contrast, in the superconducting state, the
479: impurities provide a large renormalization of the zero frequency effective mass and cause a peak to appear at twice the gap in the optical
480: mass $[1+\lambda^{op}(T,\omega)]$ which we predict here and observe
481: in the experiment.
482:
483: \begin{acknowledgments}
484: We thank S. Goshima, M. Mori, T. Suzuki and H. Matsui for
485: valuable assistance and advice in TDTS experiments. The experimental
486: work at Sendai has been supported by Grants-in-Aid for Scientific Research
487: (09440141 and 15201019) from the Ministry of Education, Culture, Sports,
488: Science and Technology, Japan.
489: This work has also been supported by the Natural Science and
490: Engineering Council of Canada (NSERC) and the Canadian Institute
491: for Advanced Research (CIFAR).
492: \end{acknowledgments}
493:
494: \begin{thebibliography}{99}
495:
496: \bibitem{palmer} L. H. Palmer and
497: M. Tinkham, Phys. Rev. {\bf 165}, 588 (1969).
498:
499: \bibitem{joyce} R. R. Joyce and P. L. Richards,
500: Phys. Rev. Lett. {\bf 24}, 1007 (1970).
501:
502: \bibitem{farnworth} B. Farnworth and
503: T. Timusk, Phys. Rev. B {\bf 10}, 2799 (1974); ibid. {\bf 14}, 5119 (1976).
504:
505: \bibitem{allen} P. B. Allen,
506: Phys. Rev. B {\bf 3}, 305 (1971).
507:
508: \bibitem{mcmillan} W.L. McMillan and J.M. Rowell
509: in {\it Superconductivity}, Edited by R.D. Parks (Marcel Decker, Inc.,
510: New York, 9169) Vol. 1, p. 449.
511:
512: \bibitem{nam} S. B. Nam,
513: Phys. Rev.{\bf 156}, 470 (1967).
514:
515: \bibitem{schrieffer} J.R. Schrieffer, {\it Theory of
516: Superconductivity},
517: (W.A. Benjamin, New York, 1964).
518:
519: \bibitem{scalapino} D.J. Scalapino
520: in {\it Superconductivity}, Edited by R.D. Parks (Marcel Decker, Inc.,
521: New York, 9169) Vol. 1, p. 449.
522:
523: \bibitem{marsigliorev} F. Marsiglio and J.P. Carbotte
524: in {\it Physics of Conventional and Unconventional
525: Superconductors}, Edited by K. H. Bennemann and J. B. Ketterson
526: (Springer-Verlag, Berlin, 2002) vol. 1, p. 233; F. Marsiglio,
527: J. P. Carbotte, A. Puchkov, and T. Timusk,
528: Phys. Rev. B {\bf 53}, 9433 (1996).
529:
530: \bibitem{marsiglio} F. Marsiglio and J.P. Carbotte, Australian J. Phys.
531: {\bf 50}, 975 (1997); ibid. 1011, (1997).
532:
533: \bibitem{rowell} J.M. Rowell and
534: W.L. McMillan, Phys. Rev. Lett. {\bf 16}, 453 (1966).
535:
536: \bibitem{Nuss1998} M.C. Nuss and J. Orenstein, in \textit{Millimeter
537: and Submillimeter Wave Spectroscopy of Solids}
538: ed. G. Gr\"uner (Springer-Verlag,
539: Berlin Heidelberg, 1998) Ch. 2, pp. 7-50.
540:
541:
542: \bibitem{carbotte} J.P. Carbotte,
543: Rev. Mod. Phys. {\bf 62}, 1027 (1990).
544:
545: \bibitem{CSH} J. P. Carbotte, E. Schachinger, and J. Hwang,
546: Phys. Rev. B {\bf 71}, 054506 (2005).
547:
548: \bibitem{marsiglio02} F. Marsiglio, J. P. Carbotte,and E. Schachinger,
549: Phys. Rev. B {\bf 65}, 014515 (2001).
550:
551: \bibitem{chubukov} A. V. Chubukov, Ar. Abanov, and D. N. Basov,
552: Phys. Rev. B {\bf 68}, 024504 (2003).
553:
554: \bibitem{abanov} Ar. Abanov and A. V. Chubukov,
555: Phys. Rev. Lett. {\bf 88}, 217001 (2002).
556:
557:
558: \end{thebibliography}
559:
560: \end{document}
561: