1
VOLUME 82, NUMBER 8 PHYSICAL REVIEW LETTERS 22 FEBRUARY 1999 Bruder and Imry Reply: In our recent Letter [1], we pointed out that there is an additional contribution to the magnetic response of a normal-metal-coated supercon- ducting cylinder (as compared to a planar proximity sam- ple). Semiclassical trajectories that are localized at the outer surface of the cylinder (glancing states) fail to “see” the superconductor; they lead to a paramagnetic correc- tion of the magnetic response at low temperatures. These states are more stable with respect to scattering than orbits in a normal cylinder because the presence of the super- conductor modifies the spectrum of the normal layer and eliminates final states to be scattered into at low energies. Fauchère et al.’s [2] Comment addresses the size of the effect that we describe. They argue that the ratio of the paramagnetic current produced by the glancing states and the Meissner current given in Eq. (7) of our paper [1] is reduced by screening. We agree; in fact, we discuss the screening in the paragraph following Eq. (7). At the end of the paragraph, we mention a reduction by l eff yd where l eff is the effective penetration depth in the proximity layer and d is the thickness of the normal layer. The effective penetration depth can be estimated [3] to be l eff 3 q l 2 L ,, where , is the elastic mean free path. For the (almost) clean samples that we have in mind, l eff is significantly larger than the London penetration depth l L , i.e., our estimate is much more favorable than the reduction factor given in the Comment. In our opinion, it is the effective screening length that has to appear in the estimate: In a perfectly clean system, there is no (local) screening [4], but a small concentration of impurities leads to screening on a length scale l eff [3]. A crucial point that we explain in the paragraph below Eq. (7) in our paper is that the paramagnetic current I P is to be compared not to the full diamagnetic current I D but to the real Meissner current (also reduced by screening) I M . Therefore the final magnetic moment compared by our paramagnetic current can easily be on the order of percents of that produced by I M . Since all of the other available theoretical treatments lead to a saturation of the magnetic response at low temperatures, we think that this is a significant correction. However, a true self-consistent calculation of the magnetic response is needed to settle the numbers. We do not agree with Fauchère et al. that our idea has been presented before in Ref. [5]. In this work, the in- fluence of curvature in a proximity system has been stud- ied within the quasiclassical theory of superconductivity. Since the effect that we describe is similar in nature to a normal persistent current (amplified by the presence of the superconductor), it cannot be obtained within the frame- work of the quasiclassical theory of superconductivity. In the last part of their Comment, Fauchère et al. address the temperature dependence of the correc- tion. They argue that it is not permissible to use the temperature dependence of normal persistent currents in this connection. We do not understand this argument, since the glancing states are a special type of normal persistent currents. We also point out that there is another source of temperature dependence caused by the loss of “protection” of the glancing states at higher temperatures. Fauchère et al. are right that ordinary normal-metal persistent currents can have either diamagnetic or para- magnetic signs. However, we have discussed the special properties of the glancing states and we have shown that their orbital response is predominantly paramagnetic. This is due to their relatively large azimuthal velocities, as is discussed following Eq. (5) in our paper. Finally, we would like to address the question of the relation of our work to the experiment [6]. The experiment by Mota and co-workers [6] has remained unexplained for quite a number of years now. Whereas the effect that we discuss in our Letter may not account for a complete disappearance of the diamagnetic response that was found in some samples (this is clearly spelled out in our Letter; explaining the experiment of Ref. [6] is not our main purpose), we do predict a paramagnetic correction to the magnetic susceptibility of nonsingly connected proximity systems that should show up at low temperatures. As far as we know, this is the first time such a reentrance effect was found theoretically. In our opinion, the effect that we describe is a first step towards an explanation of the experiments, and we suggested two crucial experimental tests that can be used to check the relevance of our idea. Christoph Bruder Departement Physik und Astronomie, Klingelbergstrasse 82 Universität Basel, CH-4056 Basel, Switzerland Yoseph Imry Weizmann Institute of Science Department of Condensed Matter Physics IL-76100 Rehovot, Israel Received 25 September 1998 [S0031-9007(99)08506-3] PACS numbers: 74.50. + r, 73.23. – b [1] C. Bruder and Y. Imry, Phys. Rev. Lett. 80, 5782 (1998). [2] A. Fauchère, V. B. Geshkenbein, and G. Blatter, preceding Comment, Phys. Rev. Lett. 82, 1796 (1999). [3] W. Belzig, C. Bruder, and A. L. Fauchère, Phys. Rev. B 58, 14 531 (1998). [4] A. D. Zaikin, Solid State Commun. 41, 533 (1982). [5] W. Belzig, Diploma thesis, Universität Karlsruhe, 1995. [6] P. Visani, A. C. Mota, and A. Pollini, Phys. Rev. Lett. 65, 1514 (1990); A. C. Mota, P. Visani, A. Pollini, and K. Aupke, Physica (Amsterdam) 197B, 95 (1994). 0031-9007y 99 y 82(8) y 1797(1)$15.00 © 1999 The American Physical Society 1797

Bruder and Imry Reply:

  • Upload
    yoseph

  • View
    216

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Bruder and Imry Reply:

VOLUME 82, NUMBER 8 P H Y S I C A L R E V I E W L E T T E R S 22 FEBRUARY 1999

tcmtererpaniehthai

n

yeb

o

pthne

o

w

nb

r

ttee

intuitot

rrr

hes innt,

malthers ofes.lara-cial

thattic.s,

ofeedas

untnse

led]ticlyow

eourrdstwothe

82

-3]

ing

B

t.

Bruder and Imry Reply: In our recent Letter [1], wepointed out that there is an additional contributionthe magnetic response of a normal-metal-coated superducting cylinder (as compared to a planar proximity saple). Semiclassical trajectories that are localized atouter surface of the cylinder (glancing states) fail to “sethe superconductor; they lead to a paramagnetic cortion of the magnetic response at low temperatures. Thstates are more stable with respect to scattering than oin a normal cylinder because the presence of the suconductor modifies the spectrum of the normal layereliminates final states to be scattered into at low energ

Fauchèreet al.’s [2] Comment addresses the size of teffect that we describe. They argue that the ratio ofparamagnetic current produced by the glancing statesthe Meissner current given in Eq. (7) of our paper [1]reduced by screening. We agree; in fact, we discussscreening in the paragraph following Eq. (7). At the eof the paragraph, we mention a reduction byleffyd whereleff is the effectivepenetration depth in the proximitlayer andd is the thickness of the normal layer. Theffective penetration depth can be estimated [3] to

leff ­3

ql

2L,, where, is the elastic mean free path. F

the (almost) clean samples that we have in mind,leffis significantly larger than the London penetration delL, i.e., our estimate is much more favorable thanreduction factor given in the Comment. In our opinioit is the effective screening length that has to appin the estimate: In a perfectly clean system, thereno (local) screening [4], but a small concentrationimpurities leads to screening on a length scaleleff [3].A crucial point that we explain in the paragraph beloEq. (7) in our paper is that the paramagnetic currentIP isto be compared not to the full diamagnetic currentID butto the real Meissner current (also reduced by screeniIM . Therefore the final magnetic moment comparedour paramagnetic current can easily be on the ordepercents of that produced byIM . Since all of the otheravailable theoretical treatments lead to a saturation ofmagnetic response at low temperatures, we think thatis a significant correction. However, a true self-consistcalculation of the magnetic response is needed to sthe numbers.

We do not agree with Fauchèreet al. that our idea hasbeen presented before in Ref. [5]. In this work, thefluence of curvature in a proximity system has been sied within the quasiclassical theory of superconductivSince the effect that we describe is similar in nature tnormal persistent current (amplified by the presence ofsuperconductor), it cannot be obtained within the framwork of the quasiclassical theory of superconductivity.

In the last part of their Comment, Fauchèet al. address the temperature dependence of the co

0031-9007y99y82(8)y1797(1)$15.00

oon--

he”

ec-se

bitser-ds.

eend

sthed

e

r

the,arisf

g)yof

hehisntttle

-d-y.a

hee-

eec-

tion. They argue that it is not permissible to use ttemperature dependence of normal persistent currentthis connection. We do not understand this argumesince the glancing states are a special type of norpersistent currents. We also point out that there is anosource of temperature dependence caused by the los“protection” of the glancing states at higher temperaturFauchère et al. are right that ordinary normal-metapersistent currents can have either diamagnetic or pmagnetic signs. However, we have discussed the speproperties of the glancing states and we have showntheir orbital response is predominantly paramagneThis is due to their relatively large azimuthal velocitieas is discussed following Eq. (5) in our paper.

Finally, we would like to address the questionthe relation of our work to the experiment [6]. Thexperiment by Mota and co-workers [6] has remainunexplained for quite a number of years now. Wherethe effect that we discuss in our Letter may not accofor a complete disappearance of the diamagnetic respothat was found in some samples (this is clearly spelout in our Letter; explaining the experiment of Ref. [6is not our main purpose), we do predict a paramagnecorrection to the magnetic susceptibility of nonsingconnected proximity systems that should show up at ltemperatures. As far as we know, this is the first timsuch a reentrance effect was found theoretically. Inopinion, the effect that we describe is a first step towaan explanation of the experiments, and we suggestedcrucial experimental tests that can be used to checkrelevance of our idea.

Christoph BruderDepartement Physik und Astronomie, KlingelbergstrasseUniversität Basel, CH-4056 Basel, Switzerland

Yoseph ImryWeizmann Institute of ScienceDepartment of Condensed Matter PhysicsIL-76100 Rehovot, Israel

Received 25 September 1998 [S0031-9007(99)08506PACS numbers: 74.50.+r, 73.23.–b

[1] C. Bruder and Y. Imry, Phys. Rev. Lett.80, 5782 (1998).[2] A. Fauchère, V. B. Geshkenbein, and G. Blatter, preced

Comment, Phys. Rev. Lett.82, 1796 (1999).[3] W. Belzig, C. Bruder, and A. L. Fauchère, Phys. Rev.

58, 14 531 (1998).[4] A. D. Zaikin, Solid State Commun.41, 533 (1982).[5] W. Belzig, Diploma thesis, Universität Karlsruhe, 1995.[6] P. Visani, A. C. Mota, and A. Pollini, Phys. Rev. Let

65, 1514 (1990); A. C. Mota, P. Visani, A. Pollini, andK. Aupke, Physica (Amsterdam)197B, 95 (1994).

© 1999 The American Physical Society 1797