Anda di halaman 1dari 3

866

IEEE TRANSACTIONS ON MAGNETICS, VOL. MAG-23, NO.2, MARCH 1987

HIGH-RESISTANCE SNS SANDWICH-TYPE


JOSEPHSON JUNCTIONS
A. S. Barrerat and M. R. Beasley
Department of Applied Physics Stanford University,
Stanford, Ca. 94305

Abstract

Nb and theupperNb
films were2000
or IOOO'A, and 1000 A or
500A, respectively. Depending on the substrate holder orientation
relative to that of the sources, we could either vary linearly the barrier
We have investigated the behavior of SNS and SIS junctions with
barriers in the vicinity of the Metal-Insulator (MII) transition. We use Nb for concentration along a line of substrates or keep it constant. Inan
orthogonal direction we could also vary the thickness of the barrier from,
the superconductor and amorphous-(Nb-Si)for the barrier. All junctions
one substrate to another or keep it constant. Therefore, it was posible to
were made by meansof electron-beam evaporationon AI203 substrates.
evaporate sets of samples in which either the barrier composition or the
The junction areas were defined by the Selective Niobium Anodization
barrier thickness was systematically varied. The compositionof the alloy
Process (SNAP). We have made junctions downto 8x8 pm2 and barrier
layer was calibratedin terms of the individual evaporation rates
for Nb and
thicknesses in range from 250 A to 600 A. The compositionof the barrier
Si via microprobe determinations.
was varied from 5 O h Nb to 1 5 % Nb. Working on the metallic side of the
M/I transition we observe pbridge-like I-V characteristics with RNA 0.3
The junction areas were defined using standard photolithography
and an adaptation of the SNAP process.
A schematic cross-sectional
Q - p 2 and ICRN 200 pV.
view of a complete junction is shownin Fig. 1. Note that an extra layerof
photoresist is included as a safeguard against shorts through the
Introduction
insulator.

For superconducting electronic circuit applications, it is desirable to


have Josephson junctions with high resistances. High resistances are
needed, for example, in order to match the junctions to transmission
lines or to conventional semiconductor electronics. At the same time,
NONANODIZED
depending on the particular application, junctions with low capacitance( or
,
,
,UPPER CONTACl
UPPER Nb
( N b or Sn)
at least a low PC=(4~er@N2)/h) are desired. Such junctions have the
ELECTRODE
smooth, non-hysteretic current-voltage characteristics that are required
I
for SQUID and RF applications. In the past, non-hysteretic junctions have
-PHOTORESIST
been possible only with shunted tunnel junctions (at the expense of the
-ANODIZED N b
ICRN -product) or with microbridges, which are difficult to fabricate. A
--O-(Nb-Sl)
I
~-Nb
more attractive alternative for non-hysteretic devices is sandwich-type
SNS Josephson junctions. Historically such Josephson junctions have
been only of scientific interest, due to their exceedingly low resistances.
In this paperwe report a new approachto the fabricationof sandwich-type
SNS
Josephson
junctions
that
holds
the
potential
to yield
Figure 1. Cross-sectional view of a complete junction.
high-resistance SNS junctions (R~>l!2) with conventional (1 pm level)
Each substrate was patterned with four junctions of equal area.
photolithography. The new idea isto use an alloy neara M/I transition as
the barrier material. Specifically,we have used amorphous-(Nb-Si) as the
Junctions of three different basic areas (67x67 pm2, 18x18 pm2, and
barrier material, although any alloy near a M/I transition is a potentially
8x8 pm2) were studied. Asseen in Fig. 1, weused either e-beam
useful barrier material. As shown by Hertel et aL1, a-(Nb-Si)undergoes a
deposited Sn or Nbas the contact to the upper Nb layer. Before
deposition of the contact layer, the surface of the upper Nb film was
continuous M/I transition at 11.5 % Nb, thus very high resistances are
cleaned by argon ion milling to eliminate any oxide layer tunnel junction
possible in principle. Barriers on the insulating side of the M/I transition
between the Nb and the upper contact film.
are also of interest because the decay length of the wave junction is
believed to diverge as the M/I transitionis approached from the insulating
Electrical chara&ristica
side 2. In the remainder of this paper we describe the fabrication and
preliminary electrical characteristics of these newSNS Josephson
The I-V characteristics were studiedin two different sets of samples.
junctions.
8, was kept constant and the Nb
In the first set the barrier thickness, 250
concentration in the barrier was varied from 8.1
% Nb to 13.0 Yo Nb. A M/I
transition is known to occur at 11.5 % Nb l . For these junctions the
Fabrication
thickness of the base Nb electrodeis 1000 A and the upper Nb electrode
is 500 A thick. In the second set we made junctions with different barrier
Our sandwich-type SNS Josephson junctions were fabricated on
thicknesses (250 A, 300 A, 450 A,550 A. and 600 A) and kept the barrier
1/4x1/4" polished AI203 substrates. Prior to deposition the substrates
composition fixed at 15 % Nb. In this case the thickness of the base Nb
were cleaned chemically and then again with
ion milling in the evaporator.
electrode is2000 A and the upper Nb electrode is 1000 A. Some
The entire Nb/a-(Nb-Si)/Nbtrilayer structure was evaporatedin situ using
materials parameters for representative junctions from both sets are given
electron beam evaporation. Deposition on
as many as 20 substratesin a 2
in table I 4.
x10 matrix could be carried out in a single evaporation run. The
evaporation rates were0.7 k s for the Nb and1 &s for the a-(Nb-Si). The
Fig. 2 shows the I-V characteristics of two typical junctions(A14 and
a-(Nb-Si) alloy was deposited using coevaporation from two sources. The
A32)as a function of temperature. At 4.2 K the I-V characteristics
pressure during evaporation was less than 7 ~ 1 0 torr.
- ~ The substrate
resembles those of a common S-N-S Josephson junction, but at lower
temperature was maintained at 750 "C for the deposition of the Nb base
temperatures
they begin to develop constant voltage steps. The
origin of
electrode but was reduced to 110 "C for the deposition of the a-(Nb-Si)
these steps is not yet clear, but they appear to be a form of zero-field
barrier layer and the Nb counter electrode. The thicknesses of the base
vortex resonances 5, since the presence of the steps correlates with
increasing w/hJ. In any event, our experience indicates these steps are
Manuscript received September 30, 1986.
not present in sufficiently small junctions.

0018-9464/87/0300-0866$01.00@1987 IEEE

867
------__--__-____---____I___

Table 1. Materials parametersfor representative Josephson junctions.All


temperature dependent quantities quoted at 1.9 K.
jcn. amp. thick.
A
RFC4
IC%
JC
wAJ

(A)

(10 Nb)
A34
A32
A22
A14
A12
B54
B42
B34
B21
B12

11.7
11.9
12.4
12.7
13.0
15
15
15
15
15

6JJn-2)

250
250
250
250
250
250
350
450
550
600

64
64
64
64
64
64
324
324
4489
4489

(awl-?, (PV

( m-?)

2.560
2.176
1.280
1.024
0.832
0.256
0.360
0.463
0.584
0.718

375
656
3125
7031
11875
62500
15432
9259
1559
1113

10
14
40
72
99
160
56
43
9
8

0.4
0.5
1.0
1.6
2.0
5.6
6.4
5.1
7.9
6.8

We also measured the critical current of the junctions of Table1 as a


function of the applied magnetic field (magnetic diffraction pattern) in
order to investigate the uniformity of the current density. Fig. 3 shows a
typical magnetic diffraction pattern for one of these junctions (A14) at
3.04 K. Semiquantitative agreement with theoretical predictions for
uniform junctions6 is evident. This result is
typical of our first set of
junctions.
Due to the larger values of the ratio of the width of the
junction to the
Josephson penetration depth (w/hJ) in the second set of junctions, the
magnetic diffraction patternsfor these junctions (not shown) are more like
those expected for symetrically-fed wide junctions6, i.e., these magnetic
diffraction patterns are symetric but the critical current does not reach
zero for any valueof the applied magnetic field.
Fig. 4 shows In(lCRN) as a function of temperature
for junctions A14
and A32. From thefigure we see thatICRN for the junction with the more
metallic barrier (i.e., larger Nb concentration) decays more slowly as a
function of temperature. Such behavior is to be expected because the
more metallic barriershould have a larger normal coherence length(CN).

I
_

T = 4.2K
3.IK
2.3K
I

2.OK
I

i.6K

On the other hand the temperature dependenceof ICRN is not what one
would expect, according to the theory of the proximity effect, in the dirty
limit7. These theories predict CN-T-* and hence Ic=exp[-d/CN] should
vary exponentially as T112 not T as is observed. The origin of this
difference is not known, but may be a consequence of the barrier being
near the metallinsulator transition where the usual theory of the
proximity effect mayneed revision,

12.7 70Nb

VO LTAGE

T=4.2K
3.lK
2.3K

1.6K

2.OK

I mV

VOLTAG E

Figure 2. I-V characteristics of two typical junctions, A14 and A32, as a


function of temperature.

IC/IC(O)
12.7 % Nb
Ic(0) = 1.83 rnA
T = 3.04 K

T (K)

Figure 4. Logaritmic plot of ICRN as a function of temperature for two


junctions (A14 and A32) with different barrier composition, but constant
barrier thickness(250 A).
The ICRN-product is also found to fall off exponentially as afunction
of thickness, d, as expected. Fig. 5 shows h(lCRN} as a function of
thickness at 1.9 K for a series of barriers containing 15 % Nb. This
behavior is in agreement with Likharevs theory8 where ICRN =
(d/CN)exp[-d/CN]. Frdm the data plotted in Fig. 5 we obtain CN(1 .9K)
100 A.

-30

-20

-10
I10

IIO
O

20

30

MAGNETIC FIELD (Oe)

Figure 3. Typical behavior of the critical current as a function of the


applied magnetic field for the first set of junctions. Thisplot corresponds
to junction A14. Measured at 3.04 K.

ICRN as a function of the barrier composition for a constant barrier


thickness (250 A) is plotted in Fig. 6. All the measurments were carried
out at 1.9 K and are on the metallic sideof the metal/insulator transition.
From Fig. 6 we can see that ICRN grows with the percentage of Nbin the
barrier. This behavior is comprehensible because, as we said before, we
expect that the more metallic barriers have larger CNsand therefore for
fixed d, !CRN = (d/cN)exp[-d/CN] should increase as the barrier
compositions go toa more metallic region.

In order to optimize the electrical characteristics of these junctions,it


will be necessary to examine the behavior of the junctions more
systematically as the metal insulator transition is approached. Specifically
we must establish the best compromise between the higher resistances
possible as the barrier becomes more resistive and the need to keep
d<cN in order tomaximizethe ICRN-product. Since
is found to
decrease as the metal/insulator transition is approached, increasingly
thin
barriers will be required. Given that the present devices have practically
useful characteristics, the prospects for truly excellent characteristics
seem favorable.

TEMPERATURE I .9 K
I5 % Nb

cN

This work supported by the Officeof Naval Research. One of us (A.


B.) would like to thank the "Consejo Nacinal de Ciencia y Tecnologia y
Universidad Nacional Autonomade Mexico" fora graduate scholarship.

e\'

References

t.

Permanent address: Universidad Nacional Autcinoma de MBxico,

I.I.M. Apdo. Postal 70-360, c.p. 04510 Mexico.


I

200

THICKNESS

600

400

(i)

Figure 5. Logaritmic plot of ICRN as a function of the barrier thickness.


The barrier composition is fixed
at 15 % Nb.

1. G. Hertel, D. J. Bishop, E. G. Spencer, J. M. Rowell, and R. C. Dynes,


"Tunneling and Transport Measurments at the Metal-Insulator Transition
of Amorphous Nb:Si", Phys. Rev. Lett.,50,743-746, 1983.

2. See for example, N. Mott, "Metal Insulator Transitions", Taylor and


Francis Ltd., 1974, pg. 34.
H. Kroger, L. N. Smith, and D. W. Jillie, "Selective niobium
anodization process for fabricating Josephson tunnel junctions", Appl.
Phys.Lett., 39(3),280-282,1981.

3.

4. To obtain the effective thickness of the barrier (&),forthe calculation


of the Josephson penetretion depth (AJ), for the first set of junctions we

used the relation 6 = htanh(dS/2h)+ h + d . Where h is the effective


magnetic field penetration depth, d, the thickness of the Nb counter
electrode, and d the thickness of the barrier.Jc is only approximate for
values biggerthan 1x1 O4 Cl-cm2.

5. For a recent discussion, see, M. Cirillo, U. Gambardella, S. Pace, and


B. Savo, "Zero-Field Singularities in Josephson Tunnel Junctions of
lntermadiate length,lEEE Trans. Mag.,21,618,1985.
6 T. Van Duzer, C. W. Turner, "Principles of Superconductive Devices
and Circuits", New 'r'ork: Elsevier, 1981, pp. 152-155;158-164.

T = 1.9 KO
d = 250A
I

L
II

12

14
Nb ( a t . %)

13

15

16

7. M. R . Beasley,
C.
J.
Kircher,
"JOSEPHSON JUNCTION
ELECTRONICS:
MATERIALS
ISSUES
AND FABRICATION
TECHNIQUES", Chap. 9 in SUPERCONDUCTOR MATERIALS
SCIENCE. Simon Foner and Brian B. Schwartz, Ed., Plenum Publishing
Co., 1981.
8. K.K. Likharev, "Superconducting weak links", Rev. Mod. Phys.,
51(1), 101-158,1979.
C. S. Owen, and D.J. Scalapino, "Inductive Coupling of Josephson
Junctions to External Circuits", J. Appl. Phys., 41(5), 2047-2056.

9.

Discussion and Conclusions


Inthispaper
we have demonstrated the practicality of high
resistance, sandwich-type SNS junctions based on the utilization of
barrier materials near a continuous metaVinsulator transition. These
junctions shouldbe of interest for SQUID circuits andfor RF applications,
in particular for Josephson juncions arrays where moderate resistances
for the individual junctions are desirable. The ICRN-products of these
junctions are found to vary systematically with the thickness and the
composition of the barrier in a manner understandable in terms of the
simple theory of proximity effect coupling. The temperature dependence
is apparently anomalous, however. The approach to datehas
demonstrated ICRN-products as high as 190 p V and specific resistances
(i. e., RNA products) greater than 1 Q-pm2,which translates into junction
resistances grater than1 R for a 1 pm diameter junction.

Anda mungkin juga menyukai