ELSEVIER Physica C 261 (1996) 167-172
PHYSICA
Influence o f an external electric field on h i g h - T c
superconductivity
W.D. Lee a, J.L. Chen b,,, T.J. Yang b, Bi-Shiou Chiou
aa Institute of Electronic, National Chiao-Tung University, Hsinchu, Taiwan b Department of Electro-Physics, National Chiao-Tung University, Hsinchu, Taiwan
Received 13 March 1995; revised manuscript received 18 January 1996
Abstract
The surface superconductivity induced by applying an electric field perpendicular to the surface of a high-To superconductor covered by an insulator and a gate electrode is studied theoretically. Taking the idea into account that the change of the surface carrier density confined in a thin screening layer modulates the transport properties of superconductors, we develop a simple method based on the Chen and Yang model to study the To shift dependence on the applied voltage in the frameworks of a Thomas-Fermi approximation and BCS theory. In such a case a novel boundary condition is introduced to modify the G - L equations. Solving the modified one, the T c shift derived by the present model is similar in qualitative behavior to the Shapiro result, but much simpler in form. In addition, it is shown to be proportional to the applied field (the square of applied field) for thin film (thick film) for a given material. Based on our simple results, the critical current density Jc and nucleation field He3 as functions of the applied voltage are easily derived.
1. I n t r o d u c t i o n
The surface superconductivity induced b y an elec- tric field is a very important subject [1] for studies in the hope that the transport properties o f superconduc- tors are to be controlled by a p p l y i n g an electric field. F o r m o r e than thirty years, since the first e x p e r i m e n t on the electric-field effect in superconductors was made b y G l o v e r and Sherill as early as 1960 [2], there have been m a n y efforts [ 3 - 1 0 ] to attempt to build a superconducting field-effect transistor for this purpose. However, the o b s e r v e d field effects for conventional superconducting metals were so small
* Corresponding author. Fax: + 886 35 725 230.
as to be neglected. This is ascribed to the long coherence length f and the large concentration o f the charge carriers, which leads to the small electric-field penetration length l D and the small ratio o f l D to ~. Recently the d i s c o v e r y o f h i g h - T c superconductors was expected to open up new possi- bilities to study the electric-field effects because o f the relatively low m o b i l e carrier density and anoma- lously short coherence length. Further, their small coherence lengths are helpful for dealing with the technological difficulties in the fabrication o f ultra- thin superconducting films in which the total carrier density can be m o d u l a t e d b y a substantial proportion. A s s u m i n g that the BCS theory can apply to the present h i g h - T c oxide superconductor [11], Shapiro et al. have calculated the critical temperature in the presence o f an electric field. H o w e v e r , their results 0921-4534/96/$15.00 © 1996 Elsevier Science B.V. All fights reserved
168 W.D. Lee et al./Physica C 261 (1996) 167-172
are complicated in form and thus the numerical calculations are essential to understand the influence of the external electric field on superconductivity. In order to obtain a simple analytic solution, in this paper we reexamine the electric-field effect and at- tempt to develop a new model. Now, we consider the superconducting transport properties to depend sensi- tively on the carrier density according to the BCS model and take them to be possibly modulated by applying an electric field because the carrier density is changed [12] on the surface characterized by the Debye screening length 1D. High-T c oxide supercon-
ductors have a low carrier concentration, giving l D
~ 5 - 1 0 .~ and a very small coherence length. Thus, the critical temperature on the surface called Tcs is different from the intact Tco of the bulk material. Although the situation is similar in outward appear- ance to the assumptions made by Chen and Yang [13], it should be emphasized here that it is different in nature. Namely, from the technical point of view it is valuable to either suppress or enhance Tcs by controlling the direction of the external electric field [7], but not in the Chen and Yang [13] case, which is based on the theoretical results found by Valls et al. [14,15], that is, the pair potential is not depleted but enhanced up to 25% near the surface. To calculate the voltage dependences of the superconducting properties, the changes of the critical temperature resulting from the external electric field are only considered in our analysis, not against the back- ground of the Chen and Yang model. However, it is worth mentioning here that in this work, the T c shift, Jc and Hc3 are derived by the three-dimensional free-electron model and Thomas-Fermi approxima- tion [9-11 ].
2. Theory
For a discussion of potential device applications such as high-T c effect transistors (the basic configu- ration is a structure with a superconducting film of thickness L separated from a gate electrode by an insulating barrier of thickness d), now there is a common point of view [8,16] that the electric-field effect is due to the energy band bending and deple- tion or enhancement of charge carriers near the interface in just the same way as in semiconductors.
So the surface charges induced by the gate voltage Vg is given by
A o - = ~(V~ -- ,h~)/(4-~d)
= e A N , t (1)
where e is the dielectric constant of the insulating barrier, ~b s is the electric potential at the surface of superconductors, and A N is the induced surface carrier density. Here the high-T c superconductor is regarded as a three-dimensional free-electron system because in general the field effects observed are in agreement with the most simple free-carrier predic- tion [12,17,18]. According to the Poisson equation in the framework of the Thomas-Fermi approximation, the potential qb(x) of the field penetrating the sam- ple is described by
~b(x)
= q5 s e x p ( - - X / I D ) , (2)l 2 = e o E F / ( 6 w n o e 2 ) , (3) where l D is the Debye screening length, defining the region where the field-induced effect occurs, n o is the carrier density far from the surface, E F is the Fermi energy, and G 0 is the dielectric constant of the superconductor. Thus the difference in mobile charges can be expressed by
= - n o
= 3n0 e~b ( x ) / 2 E F . (4)
Based on conservation of charge, the whole in- duced charge of the system is zero. This means that
A o - = f ~ e S n ( x ) d x = 3noe2ChslD/2EF. So, we have
(
4 s=
1+
l
and A N / N o = 3 e q b f l D / 2 E F L ,where N o = n o L is the surface carrier density with- out applying electric field. L is assumed to be much larger than l D. In fact, this assumption is in general satisfied with the experimental sample dimensions.
Because the critical temperature is sensitive to the carrier density n and n is nearly the same in the normal and superconducting states [17], the surface
W.D. Lee et a l . / P h y s i c a C 261 (1996) 1 6 7 - 1 7 2 169
parameter T~[n(x)] induced by the external electric field is a function of coordinate x and can be expressed as [19]
= Tc(n0) [1 + (a In Tc/a In n) . . . . ( 8 n / n o ) ]
( s ) Although the mechanism of superconductivity for high-T c oxide superconductors is still unclear, for simplicity, the BCS theory is taken into account in this paper. In the BCS model, superconductivity is based upon the existence of a net attractive interac- tion V between the electrons in a narrow energy range near the Fermi surface. This produces a critical temperature given by
Tco[= Tc(n0) ] = 1.14~o D exp( - I / g 0 ) , (6)
where T~o is the bulk critical temperature, to D is
Debye frequency and go = N(O)V is the intrinsic
electron-phonon coupling constant depending on the materials. Here we assume that V is independent of the electric field, and the density of states N(0) far from the surface is related to the carrier density in the three-dimensional free-electron model. That is to say, N(0) is proportional to n 1/3. Thus,
T~[n( x)] = T~o[1 + (eqbJ2EFgo) e x p ( - - X / I D ) ] .
( 7 ) This indicates that the change of the local transi- tion temperature is only within the screening length l D. Thus, it is reasonable to treat the superconducting film as consisting of two component parts: one is the surface layer affected by the applied field and the other is the intact bulk. In such a case, we naturally introduce the parameter T~ representing the average critical temperature on the surface characterized by l D. Here it is worth noting that in spite of the fact that l D is comparable to the coherence length ~0(T = 0 K) of high-T~ superconductors; for example [1], l D = 5 A, ~0 = 5 A for YBaCuO, and l D = 5 A, ~0 = 3 A for BiBaSrCuO, the order parameter within the screening length from the interface is nearly a constant due to the divergence of the Ginzburg- Landau coherence length ~(T) at the critical temper- ature in the mean-field approximation. Thus, the calculation of the parameter T~s, hereafter called the
surface critical temperature, is easily done by simple averaging Tc[n(x)] over l D. It is expressed as
r c s - = - rco}
dx/ D
-~ ( ed&/2EFgo)T~o. (8)
The last expression indicates the order of magnitude. In such a case, to obtain the T c shift of the whole superconductor, according to our analysis of the effect of the electric field, a novel boundary condi- tion introduced to modify the G - L equations should be considered, [13], as in the theory proposed by Chen et al. Here it should be emphasized again that in the following paragraphs our consideration is valid under the condition that both the sample size L and the GL coherence length ~(T) are much larger
than l D.
First, we assume that the external electric field
only acts on the surface layer characterized by l D
and Tcs. Thus, based on the Chen and Yang model [13], the T c shift dependence on the applied voltage can be described by __~2 d20 2---m- d x - - - 5 + a ( T - - Tc°)~b+/3~b3 = 0' (9) 1 d ~ 2 m y ~ 9 ~ x=0 h2 , (10) and 1 d.~xO x=L ~0 = 0 , (11)
where Y=lDa(T~s--T~o), and or, fl are material
parameters. In the vicinity of the critical temperature T~ the order parameter 0 is very small resulting in the term ~ 3 being neglected. Thus T~ is determined by
tanh{( L / h ) [ 2 m a ( T c --
rco)l '/=}
( y / h ) { 2 m / [ a(r~ 1/2= - Tco)] } , (12)
and ~(T c) = h [ 2 m a ( T c - - Tco)] -1/2, we have
T c = T~o[1 + ( eqbs/2Ev go)Z(/D/~:0) 2]
= + / '
]
170 W.D. Lee et al./Physica C 261 (1996) 167-172 and T~ = Tco[1 + ( e c k s / 2 E F g o ) ( l D / L ) ] = Too 1 + - - for L << ~(Tc). 3g0 (14) However, it is well known that the framework of the G L theory is valid under the condition ( T - T~o)/T~o << 1. Therefore, it must constantly be kept in mind that the basic requirement for our theory to work, is that the ratio o f the applied electric field to the Fermi energy, i.e. e 4 , s / ( 2 E z go), must be much smaller than 1.
On the basis of the above discussion, the T~ shift in Eqs. (13) and (14) in relation to the parameter
ID/~O ( l o / L ) for the thick film (thin film) is physi-
cally reasonable because although the carrier density is changed only in the thin surface layer, all the Cooper pairs within the coherence length will feel the perturbation. A similar result was obtained by Kechiantz [16]; let us discuss this in detail. On the one hand, if the framework considered in this paper is valid, the T c shift is proportional to Tco from Eqs. (13) and (14). On the other hand, the relations between the T~ shift and intrinsic material parameters lo, so0, E F and go would clearly point out w h y we have greater expectations of high-T c superconductors than conventional superconductors for building su- perconducting electronics controlled by an electric field. Finally it should be noted that according to the expression o f sC(T~), if the electric field is weak enough to lead the T c shift AT c (i.e., T c - Tco) approach zero, AT~ is proportional to A N / N o for L << ~(T~). On the other hand, if the electric field is strong enough to lead to L >> ~(Tc), AT c is propor- tional to ( A N ~ N o ) 2. Therefore, we come to the conclusions that
(1) for a given L, the linear relation between AT~
and A N / N o is no longer valid with increasing field,
(2) the larger the film thickness L is, the faster it deviates from the linear relation. This is consistent with the experimental results observed by Xi et al. [12]. Comparing the T~ shift derived by Shapiro et al. [11] with ours, we replot the Fig. 3 in Ref. [11] secundum Eqs. (13) and (14) obtained from our model and present it in Fig. 1. As shown in this
0.21 0.18 , ~ 0.15 Y 0.12 0.09
003
E=O0 52
E Olii !:ili-;:-
025 0.00 0 1 2 3 4 5 6 7 8 9 0 zFig. 1. Reduced critical temperature V=((412Dr)/~02) 1/2 VS. reduced thickness of the film L = L/21D for various reduced external electric fields E = ((4/~ K)~(g~:o2))l/Z,owhere 7 = (T c -
Tco)/Tco, t< =(3e~bs/2EF), / o = 7 A, ~0= 10A, and go = 1 are
considered. These curves (present work) are similar in qualitative behavior to the ones obtained by Shapiro et al. in Ref. [11].
figure, an interesting result is to be noted. Although, according to our assumptions made in the beginning of our analysis, the small-dimension cases, i.e. the ones with sample thickness L comparable to l D, are physically unreasonable, our results for the T c shift extended to L = 2/D are still similar at least in qualitative behavior to Shapiro's, particularly at the low-field regime. In addition, our T~ shift formulae are much simpler in form. Besides the T~ shift, both the response of the critical current density and nucle- ation field He3 to the electric field are also the essential subjects o f study from a practical point of view. Based on the formulae for the Tc shift obtained from our simple model, they are derived as follows. (a) Critical current density in a thin film:
If L << sO(T), we have the critical current density from the de Gennes theory [20,21] as
J c ( T ) = 2 5 / : a 3 / 2 e ( T c - T)3/2/(33/2ml/213 ); (15) then, the ratio of the critical current density after applying the electric field to before applying it is given by
J c ( T ) / J c o ( T )
= [ ( T c - r ) / ( r c o - r ) ] 3/2
= {1 + ( e 4 ) s / 2 E v g o ) [ T c o / ( r c o - T ) ]
× ( l D / L ) } 3/2. (16)
As reported for the field effects in Refs. [22] and [23] it is consistent with Eq. (16) that by applying an
W.D. Lee et al./Physica C 261 (1996) 167-172
171
electric field, a relatively large change of the Jc(T)
values in a thin film occurs in the range where the temperature T approaches the intrinsic bulk critical temperature Tco, even with reduced T~ shift; i.e. (T~ - T~o)/Tco, is very small.
(b) Nucleation field Hca:
The nucleation field He3 should be modified be- cause the surface boundary condition [24] has been changed by applying field. In this case, the G - L
equation with applied magnetic field H = HO z is
given by
h 2
d2~b 1 2 m d x + --mto (2 - = 1 dO x=0 2 m y 0 d xh 2 '
(17) (18)= h k y / ( 2 ell), and here
where to = 2 e H / m and x o
we assume A = (0, H x, 0).
The
term
l m t o 2 ( X - X o )2 approximates zero, while T is close to T c. Therefore the trial function 0 ( x ) is taken as e x p ( - A x ) . Substituting this trial function into Eq. (18) and multiplying Eq. (17) by qt(x)dx, then integrating over x from 0 to ~, we have A = 2 m T / h 2 andr o)
j +
x
× e x p ( 4 m y x ]h2
) d xx [Soexp( 4m x 2 )
Varying x 0 for the minimum of - o r ( T - T c o ) , the result is found that x 0 = h 2 / 4 m y . Then
64m2y 4 -- 32mhZy2ct(r - Tco )
h 6
3 2 m Y 2 1 2 m y 2 -- h20t(T - Tco)]
h 6 (19)
¢.O2~
Let us focus on Hc3 for the thickness of the superconducting film being much larger than £(T~). From Eq. (12), H~3 is given by
Hc3__ maCfls ( ID )
1/2
h e F g °
[T o(T T)]
(20)
In calculating the nucleation field
Hc3
modifiedby applying the electric field, we have obtained Eq. (20) in which Hc3 is proportional to ~b s (here ~b s at
Vg) limited by the condition that e q b J 2 E F g o is
much smaller than one. On the other hand, a special behavior of H¢3 should be pointed out, i.e. the
resulting Hc3 is proportional to (T~ - T) 1/2 different
from the result of de Gennes theory [20,21] in which He3 is proportional to (T~ - T). Finally, experimental confirmation of these theoretical results is necessary to support the research work for the effects of an electric field on superconductivity.
3. Conclusion
To apply an electric field perpendicular to the surface of high-T c superconductors, the state that the surface Tcs is different from the bulk Tco assumed by Chen and Yang [13] can be achieved. Thus according to their idea, a simple model is developed by us to study field effects on superconductivity in the frame- works of BCS theory and the three-dimensional free-electron model. Each of the formulae for the T c shift, critical current density Jc and nucleation field Hc3 depending on the applied voltage with an ana- lytic form is easily obtained. In addition, our formula for the T c shift is much simpler in form than the results derived by Shapiro et al., but they are similar to each other in qualitative nature even if extended to the cases of small sample dimension. On the other hand, the evidence for our model calculations pre- dicting the relations between the T c shift and A N / N o,
and the dependence of the critical current density Jc on temperature has been supplied by results observed in Refs. [12] and [22], respectively. Thus, in such a case we can draw the conclusion from the implica- tions of our formulae that if the framework consid- ered in this paper is valid, it should be possible to achieve higher values of T c by designing compounds
172 W.D. Lee et al./Physica C 261 (1996) 167-172
in which not only both the coherence length and carrier density should be taken as much as possible towards smaller values, but also the product of E F by go should be.
However, the mechanism of superconductivity for high-T~ materials still eludes all research. It has been commonly suggested that the BCS theory cannot apply to the present oxide superconductors, while some other people support it. On the other hand, in quasi-two-dimensional high-T c oxide materials, the validity of the three-dimensional free-electron model for charge screening is also in question. Hence, the basic assumptions of all these discussions in this paper for field effects require further experimental investigation. Very recently, experimental results
showed that T c is proportional to
ns/m
in all theplanar cuprate superconductors and the carrier den- sity is nearly the same [17] in the normal and super- conducting states. Based on these experimental re- sults and assuming that the effective mass is inde- pendent of the field, the same calculations for the T c shift, Jc and He3 can also easily be done according to our proposed model.
To sum up, it should be emphasized that in this paper the BCS theory applied to the present oxide superconductors may be inappropriate, but the method developed in this work may provide a simple and powerful tool to help the understanding of physi- cal properties of high-T c superconductors and the advances in materials sciences to build superconduct- ing three-terminal devices.
Acknowledgement
The authors would like to gratefully acknowledge research supported by National Science Council of Republic of China through the grant number, NSC 83-0212-M009-018.
References
[1] B. Ya. Shapiro, Phys. Rev. B 48 (1993) 16722.
[2] R.E. Glover and M.D. Sherill, Phys. Rev. Lett. 5 (1960) 248. [3] H.L. Stadler, Phys. Rev. Lett. 14 (1965) 979.
[4] A.T. Fiory and A.F. Hebard, Phys. Rev. Lett. 53 (1984) 2057.
[5] M. Gurvitch, H.L. Stonner, R.C. Dynes, J.M. Graybeal and D.C. Jacobson, Extended Abstracts, Proc. Symp. S, eds. J. Bevk and A.I. Braginski (MRS, 1986) p. 47.
[6] V.V. Bogatko and Yu.N. Venevtsev, Sov. Phys. Solid State 29 (1988) 1654.
[7] J. Mannhart, D.G. Schlom, J.G. Bednorz and K.A. Muller, Phys. Rev. Lett. 67 (1991) 2099.
[8] V.V. Lemanov, A.L. Kholkin and A.B. Sherman, Supercond. Sci. Technol. 6 (1993) 814.
[9] S. Sakai, Phys. Rev. B 47 (1993) 9042. [10] X.X. Xi, J. Supercond. 7 (1994) 137.
[11] B.Ya. Shapiro and I.B. Khalfin, Physica C 209 (1993) 99. [12] X.X. Xi, C. Doughty, A. Walkenhorst, C. Kwon, Q. Li and
T. Venkatesan, Phys. Rev. Lett. 68 (1992) 1240. [13] J.L. Chen and T.J. Yang, Physica C 231 (1994) 91. [14] T. Giamarchi, M.T. Beal-Monod and O.T. Valls, Phys. Rew
B 41 (1990) 11033.
[15] S.W. Pierson and O. Valls, Phys. Rev. B 45 (1992) 2458. [16] A.M. Kechiantz, in: Progress in High-Temperature Supercon-
ductivity, vol. 24, eds. W. Gorzkowski, M. Gutowski, A. Reich and H. Szymczak (World Scientific, Singapore, 1990) p. 556.
[17] A.T. Fiory, A.F. Hebard, R.H. Eick, P.M. Mankiewich, R.E. Howard and M.L. O'Malley, Phys. Rev. Lett. 65 (1990) 3441.
[18] J. Mannhart, J.G. Bednorz, K.A. Muller and D.G. Schlom, Zo Phys. B 83 (1991) 307.
[19] V.M. Nabutovski and B.Ya. Shapiro, Sov. Phys. JETP 48 (1978) 480.
[20] P.G. de Gennes, Rev. Mod. Phys. 36 (1984) 225.
[21] P.G. de Germes, Superconductivity of Metals and Alloys (Benjamin, New York, 1966).
[22] A. Walkeuhorst, C. Doughty, X.X. Xi, Q. Li, C-J. Lobb, S.N. Man and T. Venkatesan, Phys. Rev. Lett. 69 (1992) 2709.
[23] X.X. Xi, Q. Li, C. Doughty, C. Kwon, S. Bhattacharya, A.T. Findikoglu and T. Venkatesan, Appl. Phys. Lett. 59 (1991) 3470.
[24] L. Burlachkov, I.B. Khalfin and B.Ya. Shapiro, Phys. Rev. B 48 (1993) 1156.
