improvement of superconductive properties of mesoscopic …3) so far, such structures have been...
TRANSCRIPT
Improvement of Superconductive Properties of Mesoscopic Nb Wires
by Ti Passivation Layers
Kohei Ohnishi1, Takashi Kimura1;2, and Yoshichika Otani1;2
1Institute for Solid State Physics, University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa, Chiba 277-8581, Japan2Frontier Research System, RIKEN, 2-1 Hirosawa, Wako, Saitama 351-0198, Japan
Received December 17, 2007; accepted December 28, 2007; published online February 1, 2008
We have investigated superconductive properties of nano-scale Nb wires fabricated by a simple lift-off process with magnetron sputtering.
The superconductive properties of the Nb wires were remarkably improved by employing highly plasma resistant electron-beam resist
ZEP520A combined with a thin Ti passivation layer. This optimized fabrication process yielded a 300-nm-wide Nb wire with the same
transition temperature as that of the reference Nb film. Thereby, a highly transparent Nb/Cu junction was successfully fabricated.
# 2008 The Japan Society of Applied Physics
DOI: 10.1143/APEX.1.021701
The experimental studies on nano-structured super-conductors are of great importance from fundamentalas well as technological viewpoints.1,2) In particular,
characteristics such as proximity effect and charge imbalancehave been extensively studied in normal metal/supercon-ductor hybrid nanostructures.3) So far, such structures havebeen based on nano-structured Al with very low operationtemperatures below 1.5K.4,5) However, using Nb with ahigher superconducting transition temperature may enablethe device operation above 1.5K. The large superconduct-ing gap of Nb is also an advantage for operating quantumcomputing devices and single electron transistors, etc.6,7)
However, conventional lift-off processes using electron beamevaporators cannot be applied easily to refractory Nb with themelting point of 2700K because thermal radiation from aheated Nb target causes outgassing from the organic resist,deteriorating the superconductivity. To prevent such con-tamination, novel procedures using nonorganic masks8) andthermostable polymer polyethersulfone9) have been pro-posed. However, these methods complicate fabrication.
Multi-angle deposition techniques with electron beamevaporators are well utilized to fabricate the superconduct-ing junctions.7,9–11) This technique has the advantage ofrealizing clean interface and self-alignment. However, thedevice geometry is restricted by the undesired structuresdeposited during the process. Therefore, establishing asimple fabrication process for Nb nanostructures is animportant issue. Here, we show a simple lift-off process forfabricating submicron Nb wires using magnetron sputtering.The influence of the reduced size of Nb wires is minimizedby optimizing the process in terms of the width. A trans-parent Nb/Cu nanowire junction with two-step lift-offtechnique is also presented.
Micron and submicron wire structures with current andvoltage leads, as shown in Fig. 1, were patterned on aSi/SiO2 substrate with a positive resist layer by means ofelectron beam lithography with an acceleration voltageof 75 kV. Methyl methacrylate (MMA)/poly(methyl meth-acrylate) (PMMA) 950K bilayer resist and ZEP 520A mono-layer resists were chosen for the present study. A Nb film60 nm in thickness was deposited by an rf magnetron sputtermethod. The base pressure was 4 105 Pa and an Ar pres-sure during the deposition was 8 101 Pa. The depositionrate for the Nb film was 0.6 nm/s at an rf power of 200W.Finally, a lift-off process yielded the narrow wire structureas in Fig. 1 after soaking the film in the organic solution
agitated by ultrasonic vibrations. A reference Nb thin filmwas simultaneously deposited on the Si/SiO2 substrate,whose transition temperature was evaluated to be 7.9K. Thereason of the transition temperature reduced from that of thebulk, i.e., 9.4K, may be that the Nb film is sputtered onSi/SiO2 substrate at room temperature without annealing.12)
First, a PMMA/MMA bilayer was used to fabricate Nbwires because a undercut structure can easily be formedbecause of higher sensitivity of the bottom MMA resist.However, the width of the fabricated wire was much widerthan the designed value. This means that the narrow patternof the top resist does not work well as an aperture becausethe sputtered particles’ beam is not collimated due to lowvacuum process. Thus, the width of the wire is decided bythe bottom MMA resist pattern. Moreover, important to noteis that the transition temperature is very much lowered evenin micron-wide Nb wires. Figure 2(a) shows the temperaturedependence of the resistivity for the Nb wire 3 m in width,the transition temperature of which is 5.3K, much lowerthan that of the reference film. The critical temperaturemonotonically decreases with decreasing the width and thenreduces below 2.5K for the width of 1 m. We also like topoint out that the temperature range where the transitiontakes place is as wide as 1K, implying that the wire isgradually transformed into the superconductive state. Inorder to evaluate the superconductive properties, the dif-ferential resistance for the 3-m-wide wire is measured as afunction of the dc current at 3.6K. For the measurementof the differential resistance, the ac lock-in technique, with1 A ac excitation current for every 2 A dc current, isemployed. The critical current density is found to be1:0 106 A/cm2, which is much lower than that for bulkNb wires.13) Interestingly, many resistance anomalies are
V-
V +I +
I-
260 nm
780 nm
Fig. 1. Scanning electron microscope (SEM) image of the Nb wire
260 nm in width fabricated by the lift-off technique with the magnetron
sputtering.
Applied Physics Express 1 (2008) 021701
021701-1 # 2008 The Japan Society of Applied Physics
observed in the descending sweep, possibly due toinhomogeneous superconductivity.14) Thus, we concludethat the Nb wires fabricated using PMMA based resistshow poor superconductive properties.
ZEP520A resist was then used as an alternative resist forthe Nb wire fabrication. Although ZEP520A is a monolayerresist, a undercut structure is naturally formed because of thehigh sensitivity, and is much shallower than that in PMMA/MMA bilayer resist. The transition temperature for theobtained 1-m-wide Nb wire is 6K, which is much higherthan for the wire that prepared using PMMA, and is 3K evenfor the wire 0.3 m in width. In this way, the quality ofsuperconductivity is drastically improved by changing theresist from PMMA/MMA to ZEP520A. The origin of theimprovement is as follows: ZEP520A is highly resistant toAr plasma etching, and is utilized as the mask for the dryetching process. In contrast with ZEP520A, PMMA, andMMA are easily removed by the plasma. Therefore, duringthe sputtering, a large amount of admixtures of PMMAand MMA are co-deposited in the Nb wire. The transitiontemperature is thus strongly reduced since the admixturesare well known contaminators deteriorating the super-conductive properties of Nb films.9–11)
Although ZEP520A resist improves the superconductingproperty of the Nb wire, the transition temperature for thesubmicron-wide Nb wire is still not high enough. We expectthat this is also due to contaminators from ZEP520A. Tofurther improve the superconductive properties, prior to thesputtering of Nb films, a thin Ti layer less than 5 nm issputtered to passivate the resist surface. Because of the lowvacuum, such Ti passivation layers should be deposited notonly on top of the resist and Si substrate but also on the sideedge of the resist. Thus the Ti layer blocks contaminatorspenetrating into Nb sputtered films. In the way, the Nb wireswith less contaminators than without the passivation layercan be fabricated by a simple magnetron sputtering process.
As can be seen in Fig. 3(a), Nb wires using this techniqueexhibit the high transition temperature, i.e., 7.9K for a
1-m-wide wire same as that for a reference 2D film. How-ever, the transition temperature for the Nb wire graduallydecreases with wire width below 1 m. This reduction can beunderstood as follows: Figure 3(b) shows the width depen-dence of the deposited thickness of the Nb, which isevaluated from atomic force microscope analyses. As shownin the figure, the thickness of the Nb wire starts to decreasesharply below 1 m, which is due to a narrowed apertureformed on the ZEP resist hindering the deposition. Thesimilar tendency is also observed in the width dependenceof the transition temperature. These experimental resultssuggest that the decrease in the transition temperature shouldbe due to the reduction of the deposited thickness. To confirmthis assumption, the Nb wire 300 nm in width and 60 nmin thickness is fabricated. As in Fig. 4(a), the transitiontemperature is as high as that of the 1 m wide wire, i.e.,7.9K. It is thus concluded that the decrease in the transitiontemperature is due to not the width but the actual thickness.15)
We then measured the dc current dependence of thedifferential resistivity d in order to evaluate the qualityof superconductivity. The current dependence of d inFig. 4(b) is measured at 7K. In spite of the nearby transitiontemperature, a single sharp transition from the super-conductive to normal states is observed. This indicates thehomogeneous superconductivity of the Nb wire. The criticalcurrent density is also drastically improved compared withthat in the Nb wire fabricated using PMMA. Moreover, wesee no resistance anomalies, meaning that the fabricated Nbwire is a defect-free homogeneous superconductor. Thecurrent density vs voltage (J–V ) characteristics is shown inFig. 4(c). The critical current at the descending sweep ismuch lower than that at the ascending sweep. Such ahysteretic J–V characteristic is often observed in a quasi-onedimensional superconductive wire.16) Two possible mechan-isms are condsidered as the origin of the hysteretic behavior.One is self heating; Joule heating in the normal state raisesthe sample local temperature.17,18) The other is a phaseslipping dynamics at phase slip centers acting as under-
2
4
6
8
10
0.1 0.5 1 5 10
20
40
60
80
Width (µm)
Tc
(K)
Thi
ckne
ss(n
m)
(a)
(b)
Fig. 3. (a) Width dependences of the superconducting transition
temperatures Tc for the Nb wires fabricated by using MMA/PMMA
(triangles), ZEP520A (circles), and ZEP520A with the Ti passivation
(squares). (b) Width dependence of the average deposited thickness
for the Nb wires fabricated by a magnetron sputtering.
0 5 10 150
20
40
60
T (K)
Tc
(a)
(b)
ρµΩ
µΩ(
cm)
5 6 7 8
0
20
40
60
J (10 A/cm2)5
T = 3.6 K
ρ d(
cm)
Fig. 2. (a) Temperature dependence of the resistivity for the Nb
wire 3 m in width fabricated by using PMMA/MMA resists. The arrow
indicates the transition temperature Tc. (b) Differential resistivity d(¼ dv=di) for the Nb wire as a function of dc current density Jmeasured at 3.6K. The arrows indicate the positions of resistivity
anomalies.
K. Ohnishi et al. Appl. Phys. Express 1 (2008) 021701
021701-2 # 2008 The Japan Society of Applied Physics
damped systems.13,19) In the present case, as in Fig. 4(b), thevalue of the differential resistance is the same as that of thezero-bias resistance in the normal state and stays constantafter the transition from the superconductive to normal states.Moreover, the critical current in the descending sweep has anonzero offset current. These results mean that the super-current exists even in the dissipative regime and suggests thatthe present origin is the dynamics of the phase motion.13)
Finally, a highly transparent Nb/Cu wire junction isfabricated as shown in the inset of Fig. 5, in order to verifythe applicability of the present method. The structure isproduced by repeating electron beam lithography and lift-offprocedures twice. First, the Nb wire is fabricated by theabove sputtering process. The thickness of the Nb wire is30 nm. Then, the Cu wire is deposited by using a Jouleevaporator. Prior to the Cu deposition, the Nb surface iscleaned for 30 s by Ar-ion milling with 600V accelerationvoltage. The size of the Nb/Cu junction is 100 260 nm2.Figure 5 shows the temperature dependence of the contactresistance. The probe configuration is shown in the inset ofFig. 5. The value of the contact resistance is as low as thatof a highly tranparent NiFe/Cu interface.20) Therefore, webelieve that the present Nb/Cu interface is also highlytransparent. The resistance peak just above the transitiontemperature is due to the charge imbalance induced by thequasi-particle injection. The injected quasi-particles have thechemical potential different from that of the cooper pair. Thequasi-particle and cooper-pair potentials respectively varyon the charge relaxation length and the coherent length.
Since the charge relaxation length is much longer than thecoherent length, the Cu–Nb voltage probes can detect thedifference between the nonequilibrium quasi-particle andpair electrochemical potentials. The difference decreaseswith the temperature as the charge relaxation lengthdecreases with the temperature. Therefore, the resistanceshows the anomaly around the transition temperature. Thesefeatures strongly assure a high transparent interface.21)
In short, the simple fabrication process of the nano-structured Nb wire enabled to fabricate the superconductingdevices with high temperature operation and multi-terminalnormal metal/superconducting hybrid structures withoutusing multi-angle deposition. Moreover, this techniquemay be able to be applied to the lift-off process of othersuperconductors such as Nb-based alloys and MgB2.
1) H. Takayanagi and J. Nitta: Toward the Controllable Quantum States:
Mesoscopic Superconductivity and Spintronics (World Scientific,
Singapore, 2003) p. 253.
2) M. Tinkham: Introduciton to Superconductivity (Dover, New York,
2004) p. 248.
3) G. E. Blonder, M. Tinkham, and T. M. Klapwijk: Phys. Rev. B 25
(1982) 4515.
4) C. J. Chien and V. Chandrasekhar: Phys. Rev. B 60 (1999) 3655.
5) P. Charlat, H. Courtois, Ph. Gandit, D. Mailly, A. F. Volkov, and B.
Pannetier: Phys. Rev. Lett. 77 (1996) 4950.
6) Y. Nakamura, Yu. A. Pashkin, and J. S. Tsai: Nature 398 (1999) 786.
7) Y. Harada, D. B. Haviland, P. Delsing, C. D. Chen, and T. Claeson:
Appl. Phys. Lett. 65 (1994) 636.
8) R. E. Howard: Appl. Phys. Lett. 33 (1978) 1034.
9) P. Dubos, P. Charlat, Th. Crozes, P. Paniez, and B. Pannetier: J. Vac.
Sci. Technol. B 18 (2000) 122.
10) R. Dolata, H. Scherer, A. B. Zorin, and J. Niemeyer: Appl. Phys. Lett.
80 (2002) 2776.
11) H. Im, Y. A. Pashkin, T. Yamamoto, O. Astafiev, Y. Nakamura, and
J. S. Tsai: J. Vac. Sci. Technol. B 25 (2007) 448.
12) Y. Saito and T. Anayama: J. Low Temp. Phys. 21 (1975) 169.
13) A. Rogachev and A. Bezryadin: Appl. Phys. Lett. 83 (2003) 512.
14) M. Park, M. S. Isaacson, and J. M. Parpia: Phys. Rev. B 55 (1997)
9067.
15) M. S. M. Minhaj, S. Meepagala, J. T. Chen, and L. E. Wenger: Phys.
Rev. B 49 (1994) 15235.
16) R. Tidecks: Current-Induced Nonequilibrium Phenomena in Quasi-
One Dimensional Superconductors (Springer, Berlin, 1990), p. 211.
17) A. VI. Gurevich and R. G. Mints: Rev. Mod. Phys. 59 (1987) 941.
18) M. Tinkham, J. U. Free, C. N. Lau, and N. Markovic: Phys. Rev. B 68
(2003) 134515.
19) Y. Song: J. Appl. Phys. 47 (1976) 2651.
20) T. Kimura and Y. Otani: J. Phys.: Condens. Matter 19 (2007) 165216.
21) P. Cadden-Zimansky and V. Chandrasekhar: Phys. Rev. Lett. 97
(2006) 237003.
0 20 40 60
0
5
10
15
0
5
10
15
6 7 8 9 10
0
5
10
15
T (K)
ρµΩ
(
cm)
ρµΩ
d(
cm)
V(m
V)
J (10 A/cm2)5
(a)
(b)
(c)
Tc
T = 7 K
T = 7 K
Fig. 4. (a) Temperature dependence of the resistivity for the
300 nm wide Nb wire fabricated with ZEP520A resists with the
Ti passivation. The arrow indicates the transition temperature Tc.
(b) Differential resistivity d (¼ dv=di) for the Nb wire as a function of
the dc current density J measured at 7K. (c) Current density vs
voltage V characteristics of the Nb wire measured at 7K.
0 10 20
0
0.5
1
T (K)
V/I
(Ω)
I+
V+
V-I-
CuNb
Tc
Fig. 5. Temperature dependence of the interface resistance V=I fora mesoscopic Nb/Cu junction. The inset shows the SEM image of the
fabricated structure together with the probe configuration for the
resistance measurement.
K. Ohnishi et al. Appl. Phys. Express 1 (2008) 021701
021701-3 # 2008 The Japan Society of Applied Physics