The interplay between antiferromagnetism and superconductivity in disordered ultrathin high-Tc films

Physica A 200 (1993) 287-295 North-Holland SDI:

0378-4371(93)E0120-4

The interplay between antiferromagnetism superconductivity in disordered ultrathin high- TC films K.M.

Beauchamp,

T. Wang,

G.C.

Spalding

and A.M.

and

Goldman

Center for the Science and Application of Superconductivity and School of Physics and Astronomy, University of Minnesota, Minneapolis, MN 55455, USA

A series of ultrathin films of DyBa,Cu,O,_X, with properties spanning the zero-field superconductor-to-insulator transition, were prepared for the purpose of investigating the magnetic field driven superconductor-insulator transition. The resistance of these films was found to exhibit a peak whose magnitude became larger with the weakening of the superconductivity of the films. For a film just on the superconducting side of the zero-field superconductor-to-insulator transition, three successive changes in the sign of the low temperature dRldT with increasing field were found. These have been taken as evidence of the superconducting state being successively exited, reentered, and finally exited with increasing field as the sign changes from positive to negative, to positive, and finally to negative. This unusual behavior may result from an interaction of the carriers with Cu*’ ions on the copper-oxygen sheets. These spins exhibit an increasing tendency to antiferromagnetic ordering as the superconductivity of the films is weakened. The presence of many grain boundaries in the films may be the reason the effect is strikingly large.

There has been considerable interest in studying the interplay between superconductivity and disorder-induced localization in uniform [l] and granular [2] disordered ultrathin films of conventional superconductors. Ultrathin films which realize the two-dimensional limit for both superconductivity and normal electronic transport are of interest because two dimensions is both the lower critical dimension for superconductivity, and the highest dimension in which all electronic states are localized. There is considerable evidence that the superconductor-insulator transition in such systems, either driven by magnetic field or by disorder, is a quantum phase transition. Understanding superconductivity in two dimensions is particularly important in the high temperature superconductors because of the central role of conductive CuO, sheets. Here we report observations which followed from a search for a magnetic-field-driven superconductor-to-insulator transition in a series of disordered ultrathin DyBa,Cu,O,_, (DBCO) films. We found that for films whose sheet resistances fall closely on both sides of the zero-field superconductor-to-insulator transition there is an anomalous magnetoresistance which suggests the need to consider 0378-4371/93/$06.00 0 1993 - Elsevier Science Publishers B.V. All rights reserved

288

K.M.

Beauchamp

et al. I Antiferromagnetism

and superconductivity

in ultrathin films

the role of the antiferromagnetic donor compound in understanding the electrical properties. The transition from metallic and superconducting behavior to insulating behavior has been investigated in a variety of bulk, single crystal, and thin film high temperature superconducting samples by varying doping concentrations [3] and grain boundary resistances [4]. The data have been interpreted using models emphasizing localization, competition between Josephson coupling and intragranular capacitance, and spin disorder scattering. Some time ago we reported a study of a sequence of ultrathin films of DyBa,Cu,O,_, (DBCO), in which a transition from superconducting to insulating behavior at low temperatures was found when the normal sheet resistances rose above a value of about 6450 fl [5]. The transition could be traversed either by studying a series of films of decreasing thicknesses, or by aging a single film by maintaining it in vacuum at room temperature for an extended period. Increased normal state sheet resistances and reduced transition temperatures induced by this processing were correlated with decreased inverse Hall coefficients. In this reduced T, regime the Hall coefficient was temperature independent, and therefore its inverse may give a meaningful measurement of carrier concentration. The decreased carrier concentration associated with the approach to the insulating state suggests that this superconductor-insulator transition develops as oxygen is depleted and as the electronic and magnetic configurations of the insulating donor compound are approached. It should be noted, however, that the carrier concentration at the limit of superconductivity in the ultrathin films was 2 x 10z1/cm3, which is about one order of magnitude higher than the carrier concentration of 2.9 x 102’/cm3 found at the limit of superconductivity in bulk material [6]. Thus, for these films the superconductor-insulator transition does not occur at the particular values of carrier concentration usually associated with such a transition in “123” materials, but instead occurs at a resistance, in zero field, near the quantum resistance for pairs, and at some new critical values in finite field. The work reported here was carried out on ultrathin films of DBCO with thicknesses of about 35 A which were superconducting with reduced transition temperatures. The films were grown on single crystal SrTiO,(lOO) substrates using the technique of ozone-assisted molecular beam epitaxy [7]. X-ray diffraction analysis showed c-axis orientation, and transmission and scanning electron microscope revealed the presence of well-connected islands with characteristic sizes of 500-1000 A [8]. Although films of this type are microscopically disordered as well as granular, they were macroscopically homogeneous as was demonstrated by exchanging current and voltage leads affixed in the van der Pauw configuration [9]. The effects described below were found in measurements of four different sets of DBCO films in addition to the one discussed in detail here.

K.M.

Beauchamp

et al. I Antiferromagnetism

and superconductivity

in ultrathin films

289

The films were patterned into a 0.4 X 0.8 mm* Hall bar configuration by machine-assisted diamond scribing. Thick films of DBCO served as the electrodes. Measurements were made using a Kelvinox 3He-4He dilution refrigerator equipped with a 14 T superconducting magnet with films positioned such that their planes were normal to the direction of the magnetic field. The variation of the sheet resistance with temperature in zero magnetic field of a single film which has been repeatedly aged is shown in fig. 1. Each trace at low corresponds to a distinct aging step, followed by measurements temperatures. The DBCO film studied initially achieved global superconductivity, i.e., zero resistance, at 2 K, but first deviated from its normal state resistance at temperatures of 40 K, a substantially lower temperature than the bulk T, of 90 K. The superconducting transitions were always broad and, except for films of the lowest resistance, only the onset of superconductivity was ever observed. It is possible, although not certain, that zero resistance is achieved below the limiting measuring temperature. With this caveat, films exhibiting positive and negative values of dRldT at the lowest temperatures will be referred to as superconducting and insulating, respectively. This is reasonable given the expectation that for two-dimensional films, except for those right at the superconductor-insulator transition, these are the only possible states in the T+O limit [lo]. In the studies of the superconductorinsulator transition in amorphous composite In,O, films, a similar change in the sign of dRldT was interpreted as evidence of a field-induced superconductor-to-insulator transition [ll]. For temperatures in the transition region the resistance of the least resistive film increased monotonically with magnetic field up to fields of 14 T, but at the lowest temperature such fields were not sufficient to drive the film normal. For

n

u

1

1

2

3

4

5

6

T(K) Fig. 1. Sheet resistance vs. temperature, R(T), for a single 35 8, thick film aged incrementally described in the text. The different symbols correspond to different aging stages.

as

290

K.M.

Beauchamp

et al.

I Antiferromagnetism

and superconductivity

in ultrathin jilms

each higher resistance stage the application of a magnetic field caused a nonmonotonic variation in the resistance with field. This is shown at distinct, fixed temperatures in fig. 2 with each panel corresponding to a separate aging stage. For the nominal superconductor (fig. 2a), the resistance always decreased with decreasing temperature in a given field, and for the insulator (fig. 2c), the resistance always increased with decreasing temperature. For the resistive stage closest to the superconductor-insulator transition (fig. 2b) the curves of R(H) at the lowest temperatures crossed at three well defined values of sheet resistance and magnetic field. These crossings are suggestive of a sequence of magnetic field driven superconductor-to-insulator transitions. The value of the peak in R(H), R,, has been estimated by subtracting a

3.0 lo4

9.0 103 (4)

(a)

6

8

7.0 lo3

;::-:---

3.0

lo3L 0

2

4

10

H(T) Fig. 2. Sheet resistances vs. field, R(H), at various temperatures. Each figure is for a different aging stage. (a) Film with a tendency toward superconductivity. Curves (1) to (4) represent increasing temperatures of 375 mK, 575 mK, 1.1 K and 4.2 K. (b) Film in the transition region between superconducting and insulating behavior. Curves (5) to (9) represent increasing temperatures of 375 mK, 410 mK, 465 mK, 515 mK, 610 mK and 715 mK, respectively. The low temperature crossings at applied magnetic fields of O.lOaO.O5T, 1.08?0.01 T and 7.420.2T suggest field-driven transitions from superconductor to insulator, back to superconductor, and back again to insulator at higher fields. (c) Nominally insulating film. Curves (10) to (15) represent decreasing temperatures of 4.4 K, 1.2 K, 875 mK, 575 mK, 395 mK and 155 mK, respectively.

K.M.

Beauchamp

et al. I Antiferromagnetism

and superconductivity

in ultrathin films

291

linear magnetoresistance background determined by an extrapolation of R(H) from high fields. For each aging stage, R, increases as the temperature is decreased and is correlated with the value of zero-field resistance, as shown in fig. 3. There, R,,determined at the lowest temperature achieved (370 mK) is plotted against the zero-field resistance of the particular aging stage. The temperature of onset of the nonmonotonicity increases with aging as well, from below 1.5 K in the least resistive film to above 5 K in the most insulating film. The field at which the peak occurs varies systematically with temperature and age. A more thorough study is currently being conducted to determine an effective phase diagram, i .e . , the locus of points in field and temperature corresponding to the peak. Understanding of the above phenomena will require taking into account aspects of the granularity of these films. We consider first whether the anomalous magnetoresistance can follow from modeling a film as a disordered two-dimensional Josephson coupled array without any additional assumptions. Periodic modulation of the magnetoresistance is observed in regular arrays due to frustration effects [12], as the field is increased from zero, through incommensurate values, with a non-integer number of flux quanta per plaquette, to a commensurate value, with an integer number of flux quantum per plaquette. The minimum resistances occur in Josephson junction arrays for applied fluxes of 4 = m&/A, where & = 2.07 X 10m7G cm2 is the flux quantum, and A is the area enclosed by a loop. In our samples the average grain size would correspond to a characteristic or typical loop area. The minimum in resistance in our set of films occurs near 10 kG, which in this model would

1000

s 2

100

10 1

1

/ 9

/

/

9’

i

1

0.1 * 0.1

Y

1

10

100

1000

10000

R(H=O) (kQ) Fig. 3. Magnitude of peak in resistance vs. field, resistance for each aging stage. A linear background an extrapolation from high fields.

R,, at 370mK, plotted versus the zero-field mangetoresistance has been subtracted using

292

K.M.

Beauchamp

et al. I Antiferromagnetism

and superconductivity

in ultrathin films

correspond to grain areas of about 450& roughly consistent with the microstructure expected from previous studies where ultra-thin films were grown on pre-thinned, electron-transparent substrates and examined in an electron microscope [8]. Higher order oscillations are not observed in our samples, as would be expected in positionally disordered arrays [13]. A distribution in grain sizes from 180 A to 720 A would cause enough disorder to wipe out all but the first order oscillation in the magnetoresistance. Although this picture may give a qualitative explanation of a magnetoresistance peak, important aspects of the data are not explained by this picture. First, as a film evolves from a superconductor to an insulator, the magnetoresistance should become less pronounced. Instead, it is larger in the insulating state than in the superconducting state. In fact, nonmonotonic magnetoresistance is not observed in the aging stage with the highest superconducting transition temperature. Second, this model does not explain the drop in resistance at high field to below that of zero field in the most insulating film. The field at which the peak occurs also varies with temperature and aging stage, which would not occur if it were due to geometrical effects. We now consider the properties of DBCO itself. In particular, may be doped from an insulator at x = 0, to a superconductor DyBa,Cu,%+, by adding oxygen until x = 1; likewise it may be “undoped” by removing oxygen. The magnetic behavior of the Cu 2+ ions in the conducting planes is also controlled by the oxygen stoichiometry. In the insulating, oxygen deficient within the CuO, phase of DBCO, the Cu*+ ions order antiferromagnetically layers, and couple antiferromagnetically between the CuO, layers at NCel temperatures of near 400 K [14]. As the insulating compound is doped by the addition of oxygen, the NCel temperature drops. Near x = 0.4, the compound becomes superconducting and no longer exhibits long range antiferromagnetic correlations. However, antiferromagnetic spin fluctuations persist into the superconducting regime, and antiferromagnetism and superconductivity may coexist, at least at low doping levels such as those which characterize films close to the superconductor-insulator transition. Peaks in magnetoresistance have been observed in other antiferromagnetic materials such as ErAr [15], a semimetal, and in Er,Si, [16], a metal. These are believed to be due to scattering by critical spin fluctuations [17,18] at a magnetic field driven transition between antiferromagnetic and saturated paramagnetic phases. Yamada and Takada [19] have shown that there will be a peak in the magnetoresistance at this boundary. Since the low temperature phase of the insulating parent compound of DBCO is antiferromagnetic, one might expect to observe a similar peak in its magnetoresistance at the critical field, although such a peak has never been reported. A peak in the derivative with respect to field of the magnetoresistance has been reported in a single

K.M.

Beauchamp

et al. I Antiferromagnetism

and superconductivity

in ultrathin films

293

crystal of La,CuO,, and was associated with a weak ferromagnetic transition [20]. However, in that study the current was directed perpendicular, not parallel, to the CuO, planes. The antiferromagnetic exchange energy in the fully insulating DBCO compound is very high, making the field dependent NCel temperature very high as well. However, right at the superconductorinsulator transition, where the material would have a strongly reduced NCel temperature, the field necessary to induce the antiferromagnetic-paramagnetic phase transition may be substantially reduced. It is likely that the evolution of the magnetotransport properties observed in our ultrathin DBCO films occurs as a result of a combination of effects associated with granularity and with intrinsic magnetic scattering in the DBCO itself. As a film is aged, it is reasonable to postulate that oxygen is depleted preferentially from grain boundaries, reducing the intergranular coupling and the transition temperature. With aging, the contact between grains transforms from a metallic coupling in the most superconducting case, to coupling by tunneling as the grain boundaries become insulating. The evolution of the temperature at which the nonmonotonic magnetoresistance begins is consistent with a continual change in the magnetic nature of the Cu*+ ions as the oxygen concentration is reduced. The data of fig. 3 imply that the increase in the resistance and the increase in the magnetic scattering have a common origin. The changes in the background magnetoresistance can also be understood within the above framework. The positive background magnetoresistance observed in the superconducting films is due to flux-flow resistance. As the film becomes insulating the magnetoresistance is negative at low fields and fieldindependent at higher fields. This is due to damping and saturation of the spin fluctuations. The existence of distinctive values of R and H at which the curves cross, and at which dRldT changes sign, is consistent with there being a succession of transitions with increasing field. The three distinct crossing points occur at fields of 0.10 * 0.04 T, 1.08 _t 0.02 T, and 7.4 + 0.2 T, and sheet resistances of 11.7, 11.7, and 13.5 kfl, respectively. If one grants that the magnetoresistance peak is a consequence of magnetic scattering, the behaviour of the film nearest the zero-field superconductor-insulator transition (fig. 2b) may be understood. The first two field-driven transitions result from the modulation of the resistance, or the effective disorder, caused by the magnetic field. These both occur when the resistance crosses 11.7 kfi, and is identified with the zero-field transition driven by disorder, such as that observed in metal films on a-Ge substrates [l]. The third, or high field transition, is then identified as the field-induced transition, which occurs at a critical vortex density as described by the theory developed for nonmagnetic materials [lo]. This transition occurs at approximately the same field and resistance as was observed in an oxygen

294

K.M.

Beauchamp

et al. I Antiferromagnetism

and superconductivity

in ultrathin films

depleted YBCO single crystal [21]. A nonmonotonic magnetoresistance was not observed in the reduced T, YBCO single crystal, and thus only a high field transition was found. In summary, we have observed a peak in the magnetoresistance of films which span the zero-field superconductor-to-insulator transition. These films were produced in a series of vacuum aging steps which ultimately transform them into the antiferromagnetic donor compound. By analogy with other antiferromagnetic systems, we suggest that the observed peak is the signature of critical scattering at a field-dependent transition from an antiferromagnetic to an aligned paramagnetic phase. This effect has not been seen in other studies of reduced T, single crystals, possibly because of the more granular and greater two-dimensional character of ultrathin films. For granular films the antiferromagnetic material resulting from deoxygenation may be concentrated in the grain boundaries. In such cases the nonmonotonic magnetoresistance would be associated with magnetic effects in intergranular tunneling rather than scattering from antiferromagnetically correlated sites within an island. Although a reduced T, single crystal is anisotropic it may be less twodimensional than an ultrathin film. There is also the possibility that although crystals may be more geometrically homogenous than films, their oxygen doping is not, resulting in an inhomogeneity in the magnetic properties which hides the effects reported here. We may be able to rule out any role played by the magnetic Dy3+ sublattice of DBCO. This sublattice orders antiferromagnetically near 1 K, independent of oxygen concentration [22,23]. The data of fig. 3 demonstrates that the observed magnetic effects depend directly upon the resistance, i.e., the aging process. However, any effect due to the Dy 3+ ions should be independent of aging. Further work is being carried out to definitively determine whether the Dy3+ ions play any role in these results. Finally, we reiterate the importance of studying magnetic effects in the region where superconductivity and antiferromagnetic correlations appear to coexist. Such work is important for a number of proposals [24] which suggest a magnetic basis for superconductivity in the cuprates. The authors would like to thank G. Stewart, D. Huse, S. Doniach, V. Emery, L.I. Glazman, J.W. Halley and O.T. Valls for useful discussions. This work was supported in part by the Materials Research Group program of the National Science Foundation under Grant NSF/DMR-8908094, by the Air Force Office of Scientific Research under Grant F49620-93-1-0076, and by the Department of Administration of the State of Minnesota. Beauchamp acknowledges support by the American Association of University Women Educational Foundation through an American Dissertation Fellowship. Wang

K.M.

Beauchamp

et al. I Antiferromagnetism

acknowledges support by the Graduate through a Dissertation Fellowship.

and superconductivity

in ultrathin films

295

School of the University of Minnesota

References [II D.B. Haviland, Y. Liu and A.M. Goldman, Phys. Rev. Lett. 62 (1989) 2180; Y. Liu, K.A. McGreer, B. Nease, D.B. Haviland, G. Martinez, J.W. Halley and A.M. Goldman, Phys. Rev. Lett. 67 (1991) 2068. PI H.M. Jaeger, D.B. Hailand, A.M. Goldman and B.G. Orr, Phys. Rev. B 34 (1989) 4920. [31 D. Mandrus, L. Forro, C. Kendziora and L. Mihaly, Phys. Rev. B 44 (1991) 2418; M.Z. Cieplak, S. Guha, H. Kojima, P. Lindenfeld, G. Xiao, J.Q. Xiao and C.L. Chien, Phys. Rev. B 46 (1992) 5536. [41 S. Tyc, A. Schuhl, B. Ghyselen and R. Cabanel, Europhys. Lett., submitted. [51 T. Wang, K.M. Beauchamp, D.D. Berkley, B.R. Johnson, J-X. Liu, J. Zhang and A.M. Goldman, Phys. Rev. B 43 (1991) 8623, and references cited therein. Fl Z.Z. Wang, J. Clayhold and N.P. Ong, Phys. Rev. B 36 (1987) 7222. [71 B.R. Johnson, K.M. Beauchamp, T. Wang, J-X. Liu, K.A. McGreer, J-C. Wan, M. Tuominen, Y-J. Zhang, M.L. Mecartney and A.M. Goldman, Appl. Phys. Lett. 56 (1990) 1911. @I V Agrawal, N. Chandrasekhar, Y.J. Zhang, M.L. Mecartney and A.M. Goldman, J. Vat. Sci. Technol. A 10 (1992) 1531. I91 L.J. Van der Pauw, Philips Res. Rep. 13 (1958) 1. [lOI M.P.A. Fisher, Phys. Rev. Lett. 65 (1990) 923; Min-Chul Cha, M.P.A. Fisher, S.M. Girvin, M. Wallin and A.P. Young, Phys. Rev. B 44 (1991) 6883, and references therein. [Ill A.F. Hebard and M.A. Paalanen, Phys. Rev. Lett. 65 (1990) 927. WI R.A. Webb, R.F. Voss, G. Grinstein and P.M. Horn, Phys. Rev. Lett. 51 (1983) 690. (131 M.A. Forrester, H.J. Lee, M. Tinkham and C.J. Lobb, Phys. Rev. B 37 (1988) 5966; M.A. Itzler, A.M. Behrooz, C.W. Wilks, R. Bojko and P.M. Chaikin, Phys. Rev. B 42 (1990) 8319. P41 J.M. Tranquada and G. Shirane, in: Dynamics of Magnetic Fluctuations in High Temperature Superconductors, G. Reitter and P. Horsch, eds. (Plenum, New York, London. 1991) p. 1, and references found therein. P51 S.J. Allen Jr., N. Tabatbaie, C.J. Palmstrom, G.W. Hull, T. Sands, F. DeRosa, H.L. Gilchrist and K.C. Garrison, Phys. Rev. Lett. 62 (1989) 2309. WI J.A. Chroboczek, A. Briggs, W. Joss, S. Auffret and J. Pierre, Phys. Rev. Lett. 66 (1991) 790. P71 P.G. de Gennes and J. Friedel, J. Phys. Chem. Solids 4 (1958) 71. P31 M.E. Fisher and J.S. Langer, Phys. Rev. Lett. 20 (1968) 665. P91 H. Yamada and S. Takada, Prog. Theor. Phys. 49 (1973) 1401. PI Tineke Thio, C.Y. Chen, B.S. Freer, D.R. Gabbe, H.P. Jenssen, M.A. Kastner, P.J. Picone, N.W. Preyer and R.J. Birgeneau, Phys. Rev. B 41 (1990) 231. WI G.T. Seidler, T.F. Rosenbaum and B.W. Veal, Phys. Rev. B 45 (1992) 10162. PI J.M. Tarascon, W.R. McKinnon, L.H. Green, G.W. Hull and E.M. Vogel, Phys. Rev. B 36 (1987) 226. v31 J.A. Hodges, P. Imber, J.B. Marimon da Cunha, J. Hamman, E. Vincent and J.P. Sanchez, Physica C 156 (1988) 143, and references cited therein. [241 P.W. Anderson, G. Baskaran, Z. Zou and T. Hsu, Phys. Rev. Lett. 58 (1987) 2790; V.I. Emery, Phys. Rev. Lett. 58 (1987) 2794; J.R. Schrieffer, X-G. Wen and S.-C. Zhang, Phys. Rev. Lett. 60 (1988) 944; P. Monthoux, A.V. Balatsky and D. Pines, Phys. Rev. Lett. 67 (1991) 3448.