Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.cplire.ru/html/lab234/pubs/ASC98/eob-05.pdf Дата изменения: Mon Feb 28 21:19:28 2005 Дата индексирования: Tue Oct 2 14:57:23 2012 Кодировка: Поисковые слова: п п п п п п п п п п п п п п р п р п р п р п р п р п р п р п р п р п р п р п р п р п р п р п р п р п р п р п р п р р п п р п п р п

ASC'98, Palm Desert. Report EOB-05

"This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice, after which this version may no longer be accessible"

Niobium Tunnel Junctions with Multi-Layered Electrodes
Pavel N. Dmitriev, Andrey B. Ermakov, Alla G. Kovalenko, Valery P. Koshelets
Institute of Radio Engineering and Electronics, Russian Academy of Sciences, Mokhovaya 11, 103907 Moscow, Russia

Nickolay N. Iosad,
Department of Applied Physics and Materials Science Centre, University of Groningen, Nijenborgh 4, 9747 AG Groningen, the Netherlands

Alexander A. Golubov
Low Temperature Division, Department of Applied Physics, University of Twente, P.O. box 217, 7500 AE Enschede, The Netherlands

Michael Yu. Kupriyanov
Institute of Nuclear Physics, Moscow State University, Vorobyovy Ghory, 119899, Moscow, Russia Abstract - The current-voltage characteristics of the niobium aluminum oxide - niobium tunnel junctions have been studied systematically and compared with numerical simulations based on the microscopic theory of the proximity effect. The thickness of the base niobium layer is varied from 35 to 500 nm while the thickness of the aluminum layer is kept constant (about 9 nm). In a separate series of experiments the aluminum thickness is varied from 2 to 30 nm for two fixed thickness of the base electrode: 50 and 200 nm. The appropriate conditions for a full suppression of the so called "knee" structure at the gap voltage in the current-voltage characteristic are experimentally determined and theoretically interpreted in the framework of the microscopic theory. The influence of the additional layer of aluminum in a composite base electrode on the properties ofthe tunnel junction have been studied in dependence on the aluminum thickness and distance of this layer from the barrier. The obtained results demonstrate that the current-voltage characteristics of tunnel junction can be engineering by an appropriate layer thickness of compound base electrode.

appears between Nb and isolator barrier and the tunnel structure is Nb/Al/AlOx/Nb. It results in suppression of the Nb gap and appearance of the so-called knee structure due to proximity effect. In this report we present the study of the knee dependence on the thickness of the base Nb electrode and additional Al layer. The experimental results are compared with numerical calculations based on microscopic theory of the proximity effect. II. JUN
CTION FABRICATION AND MEASUREMENTS

I. I

NTRODUCTION

Nowadays the Nb-AlOx-Nb tunnel junctions are basic elements of most low-Tc superconducting electronic devices and circuits. In particular the SIS-mixers based on the high quality Nb-AlOx-Nb tunnel junctions have the noise temperature limited only by the fundamental quantum value hf/2k; these devices are currently used in the most mm and submm radio-telescopes. To realize quantum limited performance the SIS tunnel junctions with small leakage current Il(V) under the gap voltage and minor energy gap spreading Vg are required. It is especially important for relatively low frequency devices (f ~ 100 ­ 300 GHz) since Vg has to be much smaller than the frequency quantum hf/e and the leakage current at a bias voltage of about Vg ­ hf/2e determines the noise of the mixer. Any additional structure on the IVC of the junction considerably decreases the operation range of the mixer. The fabrication technology of Nb-AlOxNb tunnel junctions is based on the fact that a thin Al layer
Manuscript received September 14, 1998

cTne wolrlywcs suer thee iNparbasethelRussion Program[2];Basics eselarchyer a h fu k aov pport d n b ts by e ectr ade [1], for thi R A la a b he Rus tly SSP "Su ize ondu s a y". sundstequensian is oxidpercd. Activitconsequence a residual Al layer

The SIS tunnel junctions were fabricated by using Selective Niobium Etching and Anodization Process (SNEAP) [3], [4] on the crystalline Si substrates covered by a buffer layer of Al2O3 (d = 80 nm). A trilayer structure Nb-Al/Al2O3-Nb was deposited in single vacuum run by using dc-magnetron sputtering for both Nb and Al films (PAr = 1.10-2 and 5.10-3 mbar, deposition rate was of about 2 and 0.2 nm/s for Nb and Al correspondingly) [5]. The substrates were thermally attached to the holder under temperature control. Pure oxygen at appropriate pressure was used for the formation of the tunnel barrier (oxidation temperature 21 C, time = 20 min). The SIS junction area was defined by RIE followed by anodization, the thermally deposited SiO layer of about 270 nm is used as insulator. A computer based data acquisition system is employed to measure the SIS tunnel junction IVCs. All measurements were done at T = 4.2 K. The system operates in the constant current mode. This system collects the measured data, and computed junction parameters (up to 17 for a single sample). All data will be stored in a data base for further use (optimization of technological procedure, using real IVCs for computation and so on). The definition of the knee value is illustrated in Fig. 1. The knee current Ik is defined as the point of maximum deflection of the IVC from the Rn line. Ik is normalized to the quasiparticle current jump Ig at the gap voltage Vg. The value of Ig is evaluated as a current at Vg between lines of Rn and leakage resistance Rj. The value of Vg is determined at crossing of the bisector between Rn and Rj with measured IVC (see Fig. 1).


I
Ik

Rn

S' metals, whereas B describes the effect of the potential barrier and/or Fermi velocities mismatch between these layers. Given the values of N s ,s '( 0) , the value of can be estimated from RRR measurements of thin S/S' films, while the B is an adjustable parameter. In practice both and B may be determined from the fit to the data for IVC and for T c of S/S' bilayer as a function of layer thickness' d s , d s ' . The best fit with the experiment gives us the following set of the parameters: Nb = 15 nm, Al = 40 nm, = 0.3, B = 1. These parameters were used for IVC calculations, see Fig. 2, 3. Calculations in the above model show that DOS in the S' l layer has an energy gap g < bubk with large weight of filled N
l subgap states in the energy range g < E < bubk . That leads N

Rv Ig Rj Vg
Fig. 1 Definition of the main parameters for the model SIS IVC.

V

The knee value strongly depends on many technological parameters and can be suppressed considerably by edge effects. The properties of the tunnel barrier on the perimeter of junction are different from the central part. It is caused by suppression of the gap due to plasma etching and especially anodization. As a result of averaging of the currents in the different parts of the junction the gap voltages is smeared and the knee is suppressed, especially for micron size junctions. The contribution of the outer part is decreased with increasing junction area A up to an A of about 1500 µm2 for our technology. The knee does not changed at further increase of the junction size since the contribution of perimeter becomes negligible. The external interference can also considerably suppress the knee for small junctions. Junction' areas from 120 to 7200 µm2 were used for this study; to minimize the influence of edge effect the data for A > 1500 µm2 are presented below. III. T
HEORY

to th e shown layers. before with th

appearance of the knee structure on the IVC. As is below, the knee disappears in the regime of thin S, S' While some theoretical predictions have been made [7], no systematic experimental study and comparison e data was performed to date.
2,5 5 2,0 20 10 4 3 dNb/
Nb

Current IeRn/

Nb

1,5 1,0 0,5 0,0 1,7

=2
Param eter s:
Nb = 15nm, Al = 40nm
*

dAl/ Al =0.2

Nb /A l

*

=0.3,

Nb/Al B

=1

1 ,8

1,9

2 ,0

2,1
Nb

2 ,2

2,3

V oltag e e V/

Current IeRn/Nb

According to the Werthamer tunnel theory the IVC of the Nb-Al/Al2O3-Nb (S-S'-I-S) tunnel junction depend on the quasiparticle density of states (DOS) in the S' layer (Al). We have calculated the DOS on the basis of the microscopic proximity effect model for S-S' bilayers described in [6]. The model assumes short electron mean free path (dirty limit conditions) both in S (Nb) and S' (Al) materials. The parameters of the problem are:
= s s s 's*'
= D D
s' s

Fig. 2 Calculated IVCs for the S-S'/I/S structure at parameters corresponding to dAl = 8 nm. The thickness of the base electrode varies from 30 to 300 nm.

2,0

1,5
dNb/Nb=3 from bottom to top dAl=(2,3,4,5,6,7,8 9,10,15,20) - 1 nm Nb=15 nm,
Al =40 nm
*

N s '( 0 ) , N s (0)
cs



B

=

RB . s 's*'
s ,s '

1,0

Here s = D s / 2T cs , s*' = D s ' / 2T
N
s ,s '(

,D

, s

,s '

and

0,5

0) are the coherence lengths, the diffusion coefficients,
0,0 1,8 2,0 Voltage eV/Nb
Fig. 3 Calculated IVCs for the S-S'/I/S structure at parameters, corresponding to dNb = 45 nm. The thickness of the Al layer varies from 2 to 20 nm.

the normal state resistivities and the electronic densities of states in the normal state of S and S' metals, T cs is the critical temperature of S metal, and R B is the product of the resistance of the S-S' boundary and its area. These parameters can be understood as follows: is a measure of the strength of the proximity effect between S and

2,2


IV. RESULT

S AND DISCUSSION.

The experimentally measured IVCs at different thicknesses of the base Nb electrode for dAl = 9 nm are presented in Fig. 4, the currents are normalized to I(4 mV). The values of the knee determined from both theoretical and experimental curves, as well as the measured values of Vg are listed in the Table 1.
TABLE I PARAMETERS OF Nb-AlOx-Nb JUNCTIONS (A=7200 µ2) for dAl = 9 nm dNb, nm 35 50 75 100 150 200 350 500 Ik*/Ig* (theory for dAl= 8 nm) 0.04 0.85 0.145 0.185 0.245 0.285 0.325 0.325 Ik/Ig 0.055 0.075 0.105 0.17 0.195 0.245 0.21 0.225 Vg,mV 2. 2. 2. 2. 2. 2. 2. 2. 75 77 79 82 84 86 86 86

The normalized knee value KNb = Ik(dNb)/Ik(200 nm) is shown in Fig. 5. One can see that the experimental dependence coincides well with the theory up to dNb = 200 nm. At further increase of the Nb thickness the surface morphology of the sputtered Nb films changes considerably [8]. As a result the Al layer is not uniform and the measured knee (averaged over the junction area) is lower than the calculated one. To avoid the morphology effect a thin Nb base electrode (dNb = 50 nm) was used to study the knee dependencies on the Al thickness. The experimental IVC's for different Al thickness are shown in Fig. 6. It should be noted that the Al thickness is decreased at the oxidation, so 1 nm was subtracted from the initial value in the calculations (see Fig. 3). The obtained data are summarized in Table 2, the knee value as a function of Al thickness is shown in Fig. 7.
TABLE I PARAMETERS OF Nb-AlOx-Nb JUNCTIONS (A=1700 µ2) for dNb = 50 nm dAl, nm 2 3 4 5 6 7 8 9 10 15 Ik*/Ig* (theory) 0 0.016 0.031 0.041 0.05 0.056 0.060 0.062 0.065 0.07 Ik/Ig 0 0.01 0.02 0.03 0.04 0.05 0.07 0.085 0.115 Vg, mV 2.86 2.85 2.84 2.83 2.81 2.78 2.77 2.73 2.66

1,0 3

Normalized Current

0,8 1 0,6 0,4 0,2 0,0

2

dNb = 35 nm 75 nm 200 nm

2,5

3,0 Voltage (mV)

3,5

Fig. 4 Experimentally measured IVCs at dAl = 9 nm for 3 different thickness of the base electrode.

The experimental dependency has a different slope as compared with the calculated curve. Furthermore at dAl of about 8 nm the measured knee value abruptly increases and exceeds the theoretical one. An identical dependence is experimentally obtained for the thicker dNb = 200 nm. This discrepancy can not be explained by uncertainty in the Al thickness.
1,00

Normalized knee value K

Nb

1,2 1,0 0,8 0,6 0,4 0,2 0,0 0 100 200 300 400 500 Thickness of Nb layer (nm) Experiment, dAl = 9 nm Theory, dAl = 8 nm

Normalized Current

0,75
dAl =

0,50

0,25

15 10 8 6 3

nm nm nm nm nm

0,00

2 ,6

2 ,8

3,0

3 ,2

Voltage (mV)
Fig. 5 Calculated and measured values of the knee (normalized to the value at dNb = 200 nm) versus thickness of the Nb base electrode. Fig. 6 Experimentally measured IVCs at dNb = 50 nm for 5 different thickness of the Al layer.


120
0,12 0,10 Experiment, dNb = 50 nm Theory, dNb = 45 nm

100

Knee value Ik / Ig

Current (µA)

0,08 0,06 0,04 0,02 0,00 0 2 4 6 8 10 12 14 16

80 60 40 20 0 0

Nb-Al-Nb-Al/AlOx-Nb Rj/Rn = 40 Ik/Ig = 0.02 Vg = 2.6 mV, Vg = 150 mV

Thickness of Al layer (nm)
Fig. 7 Dependence of the knee value on the Al thickness (calculations and experiment).

1
a a

2

3

4

5

Voltage (mV)
Fig. 9 IVC of the Nb/Al /Nb -Al/AlOx-Nb junction at thickness of the additional Al layer dAla = 5 nm and dNba = 50 nm.

This disagreement could be caused by a number of reasons: i) the transition from the surface electron scattering in the Al film to the bulk one; ii) additional DOS broadening due to inelastic scattering and/or Nb gap inhomogeneity along the junction. The theory is not strictly applicable to ultrathin Al films with surface scattering. The crossover from surface to bulk scattering takes place at a certain dAl, the theory becomes valid and describes the increase of the knee value. According to the obtained results thin Al layers (dAl < 5 nm) should be used to realize a "knee-free" IVC. This thin layer does not cover completely the Nb surface for thick Nb films (dNb 200 nm) because of its morphology. As a result the Rj/Rn ratio is considerably decreased with a reduction of the Al thickness (Rj/Rn is 40, 25, 12, 3 for dAl = 7, 5, 4, 3 correspondingly). Thin base Nb (d = 50 nm) is completely covered by Al down to dAl = 3 nm (see Fig. 6); Rj/Rn is of about 40 for all used Al thickness. The SIS junctions with thin Nb base electrode have almost ideal IVC but are not suitable for high frequency application since dNb < LNb = 90 nm, that considerably increases the inductance of the microwave elements. To overcome this problem an additional Al layer is introduced in the Nb base electrode to realize a "knee-free" IVC for reasonably thick base Nb (see Fig. 8).

The introduction of an additional Al interlayer into Nb/Al/AlOx/Nb structures leads [8] to steeper IVC and disappearance of the knee structure. The reason is that with introduction of such a layer the order parameter in thin Nb-Al bilayer near the barrier becomes spatially homogeneous and thus the density of states in this bilayer becomes BCS-like with smaller energy gap. The experimental IVC for Nb/Ala/NbaAl2/AlOx-Nb structure is shown in Fig. 9. One can see that this IVC is very close to the "ideal" one with slightly reduced gap voltage. Authors would like to thank Willem Luinge for useful comments REFEREN
CES



Nb Si O
2 3

A l/A l 2 O Nb Al Nb Al 2 O Si
3

A n od i z at i o n

Fig. 8 Schematic cross-section of the Nb/Ala/Nba-Al/AlOx-Nb structure with an additional Al interlayer.

[1] J. M. Rowell, M. Gurwitch, and J. Geerk, "Modification of tunneling barriers on Nb by a few monolayers of Al", Phys. Rev. B., vol. 24, pp. 2278-2281, 1981. [2] M. Gurwitch, W. A. Washington, and H. A. Huggins, "High quality refractory Josephson tunnel junctions utilizing thin aluminum layers", Appl. Phys. Lett., vol. 42, pp. 472-474, 1983. [3] S. Morohashi, and S. Hasuo, "Experimental investigations and analysis for high-quality Nb/Al-AlOx/Nb Josephson junctions", J. Appl. Phys., vol. 61, pp. 4835-4849, 1987. [4] T. Imamura, T. Shiota, and S. Hasuo, "Fabrication of High Quality Nb/AlOx-Al/Nb Josephson Junctions: I ­ Sputtered Nb Films for Junction Electrodes", IEEE Trans. on Appl. Supercond., vol. 2, pp. 1-14, March 1992. [5] V.P. Koshelets, S.A. Kovtonyuk, I.L. Serpuchenko, L.V. Filippenko, and A.V. Shchukin, "High Quality Nb-AlOx-Nb Tunnel Junctions for Microwave and SFQ Logic Devices", IEEE Transactions on Magnetics, vol. 27, pp. 3141-3144, 1991. [6] A.A.Golubov, E.P.Houwman, J.G.Gijsbertsen, V.M.Krasnov, J. FLokstra, H.Rogalla "Proximity effect in superconductor-insulator-superconductor Josephson tunnel junctions: theory and experiment", Physical Review B, vol. 51, 2 pp. 1073-1089, 1995 . [7] A.A.Golubov, A.W.Hamster, M.Yu.Kupriyanov, J.Flokstra, H.Rogalla, "Characterization of junctions based on multilayer electrodes for application as X-ray detectors", Proceedings of the LTD-7 conference, Munchen, pp. 16-17, 1997. [8] T. Imamura, T. Shiota, and S. Hasuo, "Fabrication of High Quality Nb/AlOx-Al/Nb Josephson Junctions: II ­ Deposition of Thin Al Layers on Nb Films", IEEE Trans. on Appl. Supercond., vol. 2, pp. 84-94, June 1992.