RAPID COMMUNICATIONS PHYSICAL REVIEW B VOLUME 61, NUMBER 14 1 APRIL 2000-II Periodic magnetization instabilities in a superconducting Nb film with a square lattice of Ni dots A. Terentiev, D. B. Watkins, and L. E. De Long Department of Physics and Astronomy, University of Kentucky, Lexington, Kentucky 40506-0055 L. D. Cooley Applied Superconductivity Center, University of Wisconsin, Madison, Wisconsin 53706-1687 D. J. Morgan and J. B. Ketterson Department of Physics and Astronomy, Northwestern University, Evanston, Illinois 60208 共Received 23 September 1999兲 Isothermal magnetization curves of a superconducting Nb film perforated by a square lattice of Ni dots exhibit quasiperiodic instabilities below ⬃4 K, with a field-dependent period equal to the first, second, or third matching fields. The instabilities are found in a range of applied fields well above the saturation matching value, and under conditions where a continuous Nb film with the same dimensions and magnetization remained stable. The results suggest that terraces of matched flux density exist at the border of a flux-depleted zone created by a geometric barrier near the film edge. Geometric effects may thus play an important role in determining the presence or absence of matching effects. Thin-film superconductors patterned with lattices of artificial pinning centers 共APC’s兲 exhibit a variety of interesting mesoscopic and quantum phenomena.1–3 Lattices of antidots 共holes兲,3–5 or dots of normal6,7 or magnetic6–11 metals are known to greatly enhance the low-field critical current density J c (H), as well as provide different opportunities to precisely control flux pinning and flow, as desired in practical applications of superconducting films.12 The magnetization and electrical resistivity of patterned films are of particular interest, since they exhibit strong anomalies when the temperature is very close to the critical temperature T c and the perpendicular applied field is very near a ‘‘matching field’’ H n , at which the equilibrium number of Abrikosov flux lines 共FL’s兲 共expected to penetrate an unpatterned film in the mixed state兲 is an integral multiple n of the number of APC.1–3 However, the isolated Abrikosov FL are energetically unstable to the formation of multiply quantized (n⭓1) fluxoids at each APC.3,13–16 At intermediate applied fields H n ⬍H n⫹1 , the equilibrium state is a mixture of fluxoids 关 n⌽ 0 or (n⫹1)⌽ 0 ] that penetrate the specimen,3–5 where ⌽ 0 is the flux quantum. The magnetization curve M (H) is then similar to the logarithmic reversible flux penetration above the lower critical field H c1 of pristine type-II superconductors;16 however, the transfer of flux 共e.g., via hopping of singly quantized FL兲 between APC is irreversible and increases the magnetic hysteresis well above that of unpatterned films.3–5 From an applications viewpoint, more interesting behavior occurs when n is equal to a ‘‘saturation index’’ s ⫽D/4 (T), where D is the APC diameter and (T) is the temperature-dependent coherence length.14 When H⬎H s , it is not possible to add fluxoids with flux (s⫹1) 0 to the saturated pins,13–15 so that additional FL are instead repelled into the interstitial region between the APC.17–19 This results in a sharp transition to high flux-flow rates15,18 and greatly reduced magnetization hysteresis,4 because enhanced FL motion is permitted by relatively weak interstitial FL pinning 0163-1829/2000/61共14兲/9249共4兲/$15.00 PRB 61 very near T c . On the other hand, random pinning on interstitial defects, which grows in strength with decreasing temperature, could trap interstitial FL to produce quite disordered arrangements, even when H⬍H s . 12,17–21 It is therefore not surprising that matching anomalies previously have been observed in superconducting films with APC only when the temperature is very near to T c . 3–12,18,19 The present work reports the unexpected observation of matching anomalies in Nb films patterned with square Ni dot lattices near and below 3 K, which is well below the zerofield T c ⬇8.8 K of our Nb film. The anomalies are ‘‘quasiperiodic’’ magnetization jumps whose period may abruptly change from ⬃15 Oe, to ⬃28 Oe, to ⬃42 Oe, as the external field is increased. These periods are remarkably close to the first three matching fields (1⭐n⭐3) previously detected7,10,11 very close to T c , strongly suggesting that the quasiperiodicity is a result of field matching well below T c , in clear contrast to the general rule that matching anomalies rapidly become ill-defined for T⭐0.9T c . 12 Sample films for this study were fabricated using electron-beam lithography and electron-beam deposition. Blocks of square Ni dot lattice of 90 m⫻90 m were stitched together to produce an overall film area of 1 mm⫻1 mm. A Nb film of 95 nm thickness was then deposited around Ni dots of 120 nm diameter and 110 nm height; thus the Ni dots completely perforated the Nb layer. A control film of Nb was fabricated with the same spatial dimensions and zero-field T c ⫽8.8 K as the patterned film, which implies there was no macroscopic reduction of T c due to magnetic depairing or proximity effects. Detailed descriptions of the fabrication and physical properties of sample films were given elsewhere.7,10,11 dc magnetization and ac magnetic susceptibility measurements were done using a Quantum Design MPMS5 superconducting quantum interference device 共SQUID兲 magnetometer, using applied fields aligned perpendicular to the plane of the films. Demagnetization of the dots was accomplished R9249 ©2000 The American Physical Society RAPID COMMUNICATIONS R9250 A. TERENTIEV et al. PRB 61 FIG. 2. dc magnetic moment versus applied magnetic field for an unpatterned Nb film held at a temperature of 2.5 K. Note that narrow intervals of unstable magnetization as the last flux exits the film on either positive or negative field sweeps. FIG. 1. 共a兲 Magnetic moment versus applied magnetic field for a Nb film perforated with Ni dots, and for an unperforated Nb film at a temperature of 8.7 K (T c ⫽8.8 K兲. First and second matching fields of ⬃13 Oe and 26 Oe are clearly seen, as well as a weak matching anomaly near 39 Oe. 共b兲 Magnetic moment versus applied magnetic field for the films of 共a兲 taken at a slightly lower temperature of 8.5 K, where matching anomalies are no longer well defined. Note that there is a small field offset ⬇1 Oe in both plots due to a residual field trapped in the magnetometer solenoid. by oscillating the field of the superconducting magnet with a gradual decrease of field amplitude, starting at 1 T, while the sample was held at room temperature. After demagnetization, the sample was cooled in zero field and isothermal dc magnetization was measured at several temperatures. Additional magnetization measurements were conducted using an electromagnet fitted with a cryostat insert that accommodated a Hall-probe 共HP兲 sensor. The HP experiments were conducted to eliminate possible effects of sample motion through inhomogeneous magnetic field regions, and to achieve better thermal stability by immersing the sample in a coolant bath. The HP magnetometer data confirmed the general features of the SQUID experiments, and details of these results are discussed elsewhere.22 dc magnetization data for the patterned and control films at a temperature of 8.7 K⬃0.99 T c are shown in Fig. 1共a兲. Matching anomalies appear at the first (H 1 ⬃13 Oe, second (H 2 ⬃26 Oe兲, and third 共 H 3 ⬃39 Oe兲 matching fields for the perforated film. These values are roughly consistent with the calculated values H n ⫽n⫻(14.4 Oe兲 for the 1.2 m dot spacing 共recent data taken on another Quantum Design magnetometer indicate the low matching fields may be due to an approximate 10% scaling error in the field/current calibration of our solenoid兲. These matching anomalies previously were found7,10,11 to be asymmetric, depending on the direction of polarized Ni moments. The magnetization curve shown in Fig. 1共a兲 is close to symmetric, consistent with a low remanent magnetization of Ni dots in demagnetized samples. The data for the unpatterned film at 8.7 K are smooth, and the hysteresis width is lower than that of the patterned film only at low fields. This indicates that Ni dots act to enhance FL pinning in the matching regime H⬍H 3 ⬇H S , but induce no significant enhancement of magnetic hysteresis at higher fields at temperatures very close to T c . Lower temperature data11 indicate that there is little difference between the magnetization hysteresis for Nb films with and without APC well below T c , suggesting that spatial correlation between the FL and APC is lost at low temperatures. This is consistent with the smearing out of matching anomalies by 8.5 K ⬃0.97T c , as shown in Fig. 1共b兲 and a large body of experimental results3–12,18,19 on other patterned films. Further decreases of temperature reduce thermal stability and result in flux jumps in unpatterned Nb films,11,22 similar to the behavior of commercial superconducting wire.23 We observed this effect over a narrow range of fields close to zero when flux was leaving an unpatterned film, as shown in Fig. 2. In contrast to Fig. 2, unexpected and remarkable lowtemperature behavior is exhibited by the film patterned with Ni dots, as shown in Fig. 3. The collapse of the usual zerofield peak in magnetization of the patterned film yields a ‘‘fish tail’’ pattern recently attributed to random thermomagnetic instabilities in unpatterned Nb films.24 Nevertheless, the anomalies shown in Fig. 3 are quite unique in that the magnetization jumps are quasiperiodic over a large field range at the lowest experimental temperature of ⬃2.5 K, which is emphasized in Fig. 4. The saw-tooth oscillations first exhibit a 14–15 Oe (⬇H 1 ) period at fields above 5H 1 ⬇70 Oe, then abruptly switch to 27–29 Oe 共⬇H 2 兲 near 5H 2 ⬇140 Oe, followed by another switch to ⬃41 Oe 共⬇H 3 兲 near 5H 3 ⬇205 Oe. Moreover, our unpatterned control film magnetization, shown in Fig. 2, is essentially the same as that of the patterned film below 4 K, even though it exhibits no quasiperiodic structure at high field. The presence of in- RAPID COMMUNICATIONS PRB 61 PERIODIC MAGNETIZATION INSTABILITIES IN A . . . FIG. 3. Development of a saw-tooth structure in the magnetic moment versus applied magnetic field of a Nb film perforated with Ni dots at a temperature of 2.5 K. stabilities in the patterned film data, and their absence at high field in the control film data, suggest that the Ni dot lattice exerts a crucial influence on the appearance and nature of the instabilities. Standard stability arguments23 predict stable flux fronts for 0 J 2c a 2 / C(T c ⫺T)⬍3, assuming that J c has linear temperature dependence. The hysteresis width ⌬H⫽aJ c yields25 an estimated critical current density J c ⬇109 – 1010 A/m2 for FIG. 4. Expanded view of the data of Fig. 3 共negative applied field-top, positive applied field-bottom; arrows indicate field sweep direction兲. Note the regions of pronounced field periodicity in the field increasing data 共previously unmagnetized, virgin film兲 with period switching from ⬃14 Oe at fields below 140 Oe, to ⬃28 Oe at higher fields. The period ⬃42 Oe observed near 200 Oe corresponds to the third matching field. R9251 both patterned and unpatterned samples. Taking into account the high demagnetization factor (103 – 104 ), the film width a⫽1 mm, the density ( ⫽8550 kg/m3兲 and specific heat 关 C(T c )⫽0.8 J/kg K兴 for Nb, and inserting J c ⫽109 A/m2, yields the condition a⬍0.3 mm, which is close to the actual sample sizes. This indicates that flux gradients should be close to the instability threshold. Large instabilities abruptly cease above a reproducible field H f , whose temperature dependence (H f ⬇200 Oe at T⫽4.0 K, 230 Oe at 3.5 K, and 800 Oe at 3.0 K兲 is consistent with a strong reduction of C and increase in J c as temperature decreases.22 Explanation of the quasiperiodic instabilities clearly requires the introduction of additional mechanisms beyond random thermomagnetic events. The fact that the initial lowfield slope 共i.e., the perfect diamagnetic susceptibility of the virgin film兲 shown in Fig. 3 is very close to the slopes observed at fields just above quasiperiodic instabilities, strongly suggests that these instabilities take place in the screening currents located near the film edge region that is traditionally expected to experience local magnetic fields close to the applied field in the critical state.25 Therefore, the strong quasiperiodic behavior that persists to at least 350 Oe in Fig. 3 seems to contradict the general experimental trend that matching anomalies are limited to very low-field strengths 共e.g., n⭐8 for Nb films with comparably dimensioned antidot lattices19兲. Alternatively, matching correlations could be retained within interior domains whose sizes are comparable to the magnetic penetration depth (T). A ‘‘terraced critical state,’’ or staircase flux profile was predicted20 in the case of moderate interstitial pinning, as a compromise between uniform matching and the usual linear Bean critical state profile. Molecular-dynamic simulations21 suggest that the FL lattice is plastically deformed in domain walls between wellmatched zones. However, (T) decreases with decreasing temperature, and the well-ordered flux region within a given domain should become small compared to the total flux threading the sample well below T c . Therefore, complex flux profiles are expected to form deep inside the film, and no matching anomalies characteristics of large correlated domains in the film interior would be anticipated. A detectable signature of matching then might only occur for supercurrents running through relatively homogeneous flux arrangements near the film edge, where their contribution to the overall magnetization is maximal. Taking these observations into account, we postulate that the quasiperiodic magnetization anomalies observed for the patterned film are the result of matching in flux-depleted regions of the film that experience local magnetic fields on the order of the lowest matching fields H n (n⫽1,2,3). In particular, rarified regions in the FL density near sample edges are expected at low fields, due to the ‘‘geometric barrier’’ and the strongly penetrating Meissner currents that act to force flux toward the center of plate-like samples that have strong demagnetizing effects.26–29 Indeed, effects due to the ‘‘geometric barrier’’ should be ubiquitous in patterned films, but this scenario generally has not been considered in the literature to date. The field range where quasiperiodic anomalies are most clearly observed, from ⬃70 to ⬃250 Oe, is consistent with the field range within which the geometric barrier should be RAPID COMMUNICATIONS R9252 A. TERENTIEV et al. influential. Current lithography techniques limit the APC diameter D⬎0.1 m and the APC separation to d⬃1 m, which restricts H 1 to ⬃10–30 G. At high temperatures, where H c1 is low, we expect the enforcement of uniform matching across the entire pin array. However, far below T c , this ‘‘single-terrace’’ matching will be lost because of the inhomogeneities generated in FL density due to the combined effects of the geometrical barrier and interstitial pinning. Using the low critical field H c1 of bulk Nb⬃2000 G, and the given width W⬇1 mm and thickness t⬇0.1 m of our films 共in the absence of pinning兲, we estimate26 that FL first penetrate the edge of the film when the applied field is above H p ⫽H c1 (t/2W) 1/2⬇14 Oe 共as observed in Fig. 2兲. As H is increased above H p , a domelike distribution of FL is initially formed by the inward force of Meissner currents. This ‘‘compression region’’ spreads outward from the center of the sample into a flux-depleted region,30 and reaches the film edge at H e ⬇5H p ⬇70 Oe. When H⬎H e , the geometric barrier is overcome and FL are redistributed more uniformly throughout the film. Note that pattern in Fig. 4 共bottom plot for a virgin magnetization sweep兲 suggests this process may ‘‘reset’’ or repeat itself for 2H e or 3H e , implying that each matching field functions as a ‘‘effective lower critical field scale,’’ since the geometrical barrier might reestablish itself at these threshold fields where complete filling of the depletion region is just achieved at successively higher-field 1 A. T. Hebard, A. T. Fiory, and S. Somekh, IEEE Trans. Magn. 13, 589 共1977兲. 2 P. Martinoli, Phys. Rev. B 17, 1175 共1978兲. 3 V. V. Metlushko et al., Solid State Commun. 91, 331 共1994兲. 4 M. Baert et al., Phys. Rev. Lett. 74, 3269 共1995兲. 5 Y. Bruynseraede et al., Proc. SPIE 2697, 328 共1996兲. 6 A. F. Hoffmann, Ph.D. thesis, University of California, San Diego, 1999. 7 Y. Jaccard et al., Phys. Rev. B 58, 8232 共1998兲. 8 D. J. Morgan and J. B. Ketterson, Phys. Rev. Lett. 80, 3614 共1998兲. 9 J. I. Martin, M. Velez, J. Nogues, and I. K. Schuller, Phys. Rev. Lett. 79, 1929 共1997兲. 10 David J. Morgan, Ph.D. thesis, Northwestern University, 1998. 11 A. Terentiev et al., Physica C 324, 1 共1999兲 12 V. V. Moshchalkov et al., Phys. Scr. T55, 168 共1994兲. 13 A. Buzdin, Phys. Rev. B 47, 11 416 共1993兲. 14 G. S. Mkrtchyan and V. V. Schmidt, Zh. Eksp. Teor. Fiz. 61, 367 共1971兲 关Sov. Phys. JETP 34, 195 共1972兲兴. 15 I. V. Khalfin and B. Ya. Shapiro, Physica C 207, 359 共1993兲. 16 V. V. Moshchalkov et al., Phys. Rev. B 54, 7385 共1996兲. 17 K. Harada et al., Science 274, 1167 共1996兲. PRB 61 matching densities. It is then plausible that at low enough applied fields the flux profile near the film edge has terraces that consist solely of strongly pinned fluxoids 共i.e., no interstitial FL兲. Flux jumps then occur when the terrace edges become unstable, due to the increasing gradients associated with the overall FL profile. We conclude the quasiperiodic instabilities, which we observe for applied fields approaching 1 kOe, are related to a ‘‘geometric barrier’’26–29 that produces steep flux gradients. We identify these instabilities as unexpected lowtemperature matching anomalies, most probably initiated near the edge of the sample where the flux density can be much lower than in the film interior. This interpretation suggests that the geometrical barrier plays an important role in determining the FL distribution in films with APC lattices, and that sample geometry should be a crucial consideration in developing patterned films for applications. Research at the University of Kentucky was supported by the U.S. Department of Energy, Office of Basic Energy Science, Division of Materials Science, Grant No. DE-FG0297ER45653. Research at Northwestern University was supported by National Science Foundation Grant No. DMR9309061. Research at the University of Wisconsin was supported by the U.S. Department of Energy, Division of High-Energy Physics, Grant No. DE-FG02-96ER40961 and the NSE MRSEC for Nanostructured Materials. V. V. Metlushko et al., Europhys. Lett. 41, 333 共1998兲. V. Metlushko et al., Phys. Rev. B 60, R12 585 共1999兲. 20 L. D. Cooley and A. M. Grishin, Phys. Rev. Lett. 74, 2788 共1995兲. 21 C. Reichhardt et al., Phys. Rev. B 54, 16 108 共1996兲. 22 A. Terentiev, B. Watkins, L. E. De Long, L. D. Cooley, D. J. Morgan, and J. B. Ketterson, Physica C 共to be published兲. 23 M. N. Wilson, Superconducting Magnets 共Oxford University Press, New York, 1983兲, Chap. 7, pp. 131–158. 24 Y. Kopelevich and P. Esquinazi, J. Low Temp. Phys. 113, 1 共1998兲. 25 C. P. Bean, Phys. Rev. Lett. 8, 250 共1962兲. 26 E. Zeldov et al., Phys. Rev. Lett. 73, 1428 共1994兲. 27 M. Benkraouda and J. R. Clem, Phys. Rev. B 53, 5716 共1996兲. 28 Th. Schuster et al., Phys. Rev. Lett. 73, 1424 共1994兲. 29 E. Zeldov, J. R. Clem, M. McElfresh, and M. Darwin, Phys. Rev. B 49, 9802 共1994兲. 30 If pinning is strong, the FL distribution also contains a central region of reduced density 共‘‘doughnut hole’’兲 共Refs. 26–28兲. In any event, the critical state region is predicted to have a highly nonlinear profile until the applied field is well above H e 共Ref. 27兲. 18 19