Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.cplire.ru/html/lab234/Tarasov/reprints/PhysicaC2002p161.pdf Дата изменения: Sat Mar 12 19:51:34 2005 Дата индексирования: Sat Dec 22 11:31:04 2007 Кодировка: Поисковые слова: вторая космическая скорость

Physica C 368 (2002) 161­165 www.elsevier.com/locate/physc

Towards a dc SQUID read-out for the normal metal hot-electron microbolometer
M. Tarasov
a

a,b,*

, S. Gudoshnikov a,c, A. Kalabukhov M. Kiviranta e, L. Kuzmin a

a,d

, H. Seppa e,

c

MINA, Department of Physics, Chalmers University of Technology, S-41296 Gothenburg, Sweden b Institute of Radio Engineering and Electronics, Mokhovaya 11, Moscow 101999, Russia Institute of Terrestrial Magnetism, Ionosphere and Radio Wave Propagation, Russian Academy of Sciences (IZMIRAN), 142190 Troitsk, Moscow Region, Russia d Department of Physics, Moscow State University, Moscow 119899 GSP, Russia e VTT Automation, P.O. Box 1304, FIN-02044 VTT, Espoo 15, Finland

Abstract A prototype of SQUID read-out for current measurements at the output of a normal metal hot-electron microbolometer has been developed and tested. The system is based on serial VTT dc SQUID and input l-metal core solenoid transformer. The achieved current resolution is 300 fA/Hz1=2 . Johnson noise of metal resistor and shot noise of tunnel junction were used for current calibration of SQUID read-out. The current noise spectra of 35 kX SIN tunnel junction measured at different bias voltages are presented. с 2001 Elsevier Science B.V. All rights reserved.
Keywords: dc SQUID; SIN tunnel junction; SQUID femtoamperemeter

1. Introduction Normal metal hot-electron microbolometer (NHEB) is a novel superconducting device for radiation detection at ultimately low level. It consists of a normal metal microstrip connected to superconducting electrodes by direct SN contacts, or tunnel SIN junctions [1]. First type is called bolometer with Andreev reflection (ANHEB), and the second--a capacitive reflection bolometer

* Corresponding author. Present address: MINA, Department of Physics, Chalmers University of Technology, S-41296 Gothenburg, Sweden. Fax: +46-31-7723-224. E-mail address: tarasov@fy.chalmers.se (M. Tarasov).

(CNHEB). For normal metal strip temperature measurements an additional SIN tunnel junction is connected to the strip in ANHEB, or inherent contacting SIN junctions are used in CNHEB [2]. In a voltage-bias mode the device is characterized by current responsivity SI ј dIx =dPx , where dPx is the Fourier component of the small variation of the external RF power, and dIx is the current response. Finally, the noise is characterized by a noise equivalent power NEP, which is the net effect of all noise sources referred to the input of the device. In order to reach the NEP ј 10ю18 W/Hz1=2 the noise of read-out system dIx should be below 50 fA/Hz1=2 . The most sensitive current meter is SQUIDbased femtoamperemeter. As it was demonstrated

0921-4534/01/$ - see front matter с 2001 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 1 - 4 5 3 4 ( 0 1 ) 0 1 1 5 9 - 5


162

M. Tarasov et al. / Physica C 368 (2002) 161­165

in Ref. [3] the equivalent input current noise of about 4 fA/Hz1=2 can be reached. Such system is based on a cryogenic current comparator and allows to reach ultra-low current noise with commercial Quantum Design SQUID [4]. The main disadvantages of such approach are relatively big size and weight of device: cryogenic part with superconducting shields has 80 mm diameter and 200 mm high. Another integral SQUID compatible solutions are using a superconducting planar transformer [5] or a current transformer with ferromagnetic core. The last case is easy for realization and can be the first step towards the SQUID read-out for the NHEB.

where Rn is the normal state resistance, R0 the zero-bias resistance, k--Boltzman's constant, T--electron temperature in the normal electrode, D--superconductor energy gap. For noninvasive measurements of the temperature SIN junction should be small enough compared to the strip size that limits the junction dimensions to about 0:2 б 0:2 lm2 . Such tunnel junction optimized for NHEB ultimate performance should have normal resistance in the range 1­10 kX. Variations of current to be read-out are in the range 10­100 fA/ Hz1=2 . To estimate the SQUID read-out performance, one can consider the energy domain. Two reduced expressions
2 em ј Li dIx =2;

П 2ч П 3ч

2. Preliminary estimations In our NHEB bolometer SIN tunnel junction is a temperature sensor. It is dc voltage biased (via resistor Rv ( Rn ) at sensitive bias point. A small variation of the external RF power produce the current response dIx , which goes through a primary winding of SQUID read-out input transformer. The SQUID output signal is proportional to this current. Nonlinear I ­V curve of SIN tunnel junction can be approximated by an analytic relation. For voltages below the energy gap relation can be reduced to [6]: pffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2pDkT V D ю eV Iј Ч exp ю ; П 1ч 2eRn R0 kT

e

SQ

ј SU =2LSQ

give the magnetic energy produced by the noise current dIx and energy resolution eSQ of the SQUID. For the optimal performance when coupling factor k between the input inductance Lin and the SQUID is $1 these energies should be equal. The calculated SQUID parameters and Lin for estimation of current sensitivity are given in Table 1. Table 1 is divided into two parts by energy resolution e. The energy is given in J/Hz1=2 and in Planck's constant h. There are SQUID parameters to the left side and input parameters to the right side. The top left half-row presents the fixed SQUID self-inductances LSQ . The left data field presents the SQUID magnetic flux sensitivity SU , which can be obtained for real device. This value

Table 1 Current sensitivity of SQUID readout for different loop and input coil inductances SQUID parameters LSQ (pH)
1 SU=2

e (h) (J/Hz) 40 6.3 2 0.63 0.2 160 12.6 4 1.2 0.4 640 25 8 2.5 0.8 3220 21 б 10ю31 322 2:1 б 10ю31 32 0:21 б 10ю31 3.2 0:02 б 10ю31

Input coil parameters 316 0.042 0.004 0:4 б 10
ю3

10 3.16

100 0.42 0.042 0.004 0:4 б 10ю
3

31 4.2 0.42 0.042 0.004

10 42 4.2 0.42 0.042

3.1 420 42 4.2 0.42

1 4200 420 42 4.2

dIx (fA/Hz)1

=2

Lin (mH)

lU0 /Hz1

=2

1 0.31 0.1

0:04 б 10ю3


M. Tarasov et al. / Physica C 368 (2002) 161­165

163

depends on SQUID parameters, room-temperature electronics, SQUID temperature, etc. Corresponding energy resolution is given in the middle column. The top right half-row presents the fixed noise current resolution dIx , which is desirable to obtain. The right data field gives the necessary input inductance Lin to get the corresponding dIx . Note that results from Ref. [3] are in a good agreement with energy calculations.

3. SQUID read-out design Cryogenic mount assembly is shown in Fig. 1. It was designed for testing of SQUID, input transformers and NIS junctions. All these parts are placed on a printed circuit plate with 12-pin connector. This plate can be placed in He3 -cryostat in a future. The VTT SQUID [7] is bonded on a special circuit board and placed in a socket on the plate. A small bulk transformer with diameter 2 mm and length 6 mm is placed near the SQUID circuit. The whole probe is shielded with a lead screen.

The transformer has 300 turns primary coil wound of 36 lm diameter copper wire on soft magnetic core 0.4 mm in diameter and 6 mm long. The second superconducting coil with five turns wound above primary is connected to input coil of the SQUID. The SQUID and integrated input coil have inductances LSQ ј 8 pH and Lin ј 340 nH, respectively. The input current of approximately 3 lA yields a SQUID response of one flux quantum. The block diagram of the SQUID read-out is shown in Fig. 2. VTT nonmodulation electronics is used to operate SQUID [8]. The preamplifier intrinsic noise is about 1 nV/Hz1=2 . The operation in closed loop mode is limited within one flux quantum in order to avoid output level hopping due to external disturbances. To verify current-to-flux transfer coefficient it was used standard sinusoidal signal source.

4. Experimental results To test and calibrate SQUID read-out as the first step we have measured the spectral density of

Fig. 1. Cold part of the dip probe with plate for SIN sample.


164

M. Tarasov et al. / Physica C 368 (2002) 161­165

Fig. 2. Block diagram of the SQUID read-out.

the SQUID read-out with opened SQUID input coil. The resulting spectrum is shown in Fig. 3, curve 1. White flux noise level in flux locked loop mode turned out to be 3 lU0 /Hz1=2 . Intrinsic energy resolution corresponding to this noise is approximately eSQ ј 3 б 10ю30 J/Hz1=2 . The bandwidth is over 100 kHz. As the next step we attached l-metal transformer to the input coil and measured the currentto-flux transfer coefficient applying a small ac current to the primary winding of the transformer. The input current of approximately 100 nA yields a SQUID response of one flux quantum. The test noise measurements were made with three cooled

resistors 10 kX, 100 and 10 X. Corresponding spectra are presented in Fig. 3 as curves 2, 3 and 4 respectively. Lines 5, 6 and 7 correspond to values of Johnson current noise of these resistors dIx ј 1=2 П4kT =Rч at 4.2 K. In all cases the increase of white noise level is in good agreement with the theory. The curve 2 obtained with 10 kX resistor demonstrates noise of the SQUID read-out system, because the Johnson noise of 10 kX is twice as low compared to that of SQUID read-out. The equivalent current white noise level is below 0.3 pA/Hz1=2 . An additional low-frequency noise at frequencies below 800 Hz can be due to additional noise produced by permalloy core. This noise can be attributed to the Barkhausen effect in core made of soft ferromagnetic material [9]. For 10 X resistor (curve 4) the frequency bandwidth is decreased, because the cut-off frequency of R­L input circuit becomes less than frequency bandwidth of the SQUID electronics. The calculation of equivalent Lin brings 66 lH. At 300 K the Lin value was 260 lH. The next measurements were performed with standard Al­AlOx ­Cu SIN temperature sensors that bring the linear I ­V curve at LHe temperature. Our SIN junctions have a normal resistance about 35 kX. The noise spectra of this SIN junction were measured at different bias voltages Vb . Results are presented in Fig. 4. The curve 1 was

Fig. 3. Equivalent current/flux noise spectra of the SQUID read-out: curve 1--noise of the system without transformer; curve 2, 3, 4--SQUID read-out with transformer and input resistors 10 kX, 100 and 10 X, respectively; lines 5, 6, 7--calculated Johnson's current noise of the 10 kX, 100 and 10 X resistors.

Fig. 4. Current noise spectra of the 35 kX SIN tunnel junction measured using SQUID read-out at Vb ј 0, 11, 15, 25, 38 mV, (curves 1­5). Curve 6--noise spectrum with 35 kX metal film resistor at the Vb5 ј 38 mV.


M. Tarasov et al. / Physica C 368 (2002) 161­165

165

obtained at Vb ј 0 and equivalent the second curve in Fig. 3. At Vb1 ј 11 mV, at which current I through the SIN junction was 320 nA the noise of SIN junction exceeds the SQUID read-out noise. At Vb2 ј 15 mV the SIN junction white noise was 0.5 pA/Hz1=2 (curve 3). In this bias region Vb the white noise level of the measured spectra is in agreement with Shottky formula: SI ј 2eI : Above Vb3 ј 16 mV the excess noise was observed in SIN junction. The corresponding spectra measured at Vb4 ј 25 mV and Vb5 ј 38 mV are given by curves 4 and 5, respectively. For testing of our system in voltage bias mode the additional spectrum with 35 kX metal film resistor (spectrum 6) and the same Vb5 ј 38 mV was measured. This noise did not exceed the SQUID read-out noise. 5. Conclusion The tests of SQUID read-out with standard VTT SQUID and small permalloy transformer demonstrate the equivalent current resolution of 300 fA/Hz1=2 in more than 100 kHz bandwidth. This result is in a agreement with the energy resolution of the system calculated from SQUID energy resolution and input transformer selfinductance. An improving of the current sensitivity of the SQUID read-out down to 30 fA/Hz1=2 can

be achieved by increasing of the input transformer self-inductance to 600 lH and decreasing of the SQUID flux noise down to 1 lU0 /Hz1=2 .

Acknowledgements Financial support from INTAS, KVA, STINT is gratefully acknowledged.

References
[1] L. Kuzmin, D. Chouvaev, et al., IEEE Trans. Appl. Supercond. 9 (2) (1999) 3186­3189. [2] D. Golubev, L. Kuzmin, J. Appl. Phys. 89 (11) (2001) 6464­ 6472. [3] F. Gay, F. Piquemal, G. Geneves, Rev. Sci. Instrum. 71 (12) (2000) 4592­4595. [4] Quantum Design Inc., 11578 Sorrent Valley Road, San Diego, California, 92121-1311. [5] H. Seppa, A. Satrapinski, et al., IEEE Trans. Instrum. Meas. 9 (2) (1999) 4436­4439. [6] M. Tinkham, Introduction to Supercond., second ed., McGraw-Hill, New York, 1996, ISBN 0-07-064878-6. [7] H. Seppa, M. Kiviranta, et al., IEEE Trans. Appl. Supercond. 3 (1) (1993) 1816­1819. [8] M. Kiviranta, H. Seppa, IEEE Trans. Appl. Supercond. 5 (1995) 2146. [9] L. Callegaro, E. Puppin, S. Ricci, J. Appl. Phys. 90 (5) (2001) 2416­2421.