a large area absorber into rather small STJ elements has been demonstrated with a Sn/Al/AlxOy/Al de- vice with an energy resolution of 60eV (FWHM) at.
Progress in fabrication and development of Ta-Al/Alx Oy /Al superconducting tunnel junctions as position and energy sensitive X-ray detectors P. Hettl1, G. Angloher1 , M. Bruckmayer2, F. v.Feilitzsch1, J. Jochum3, H. Kraus2 , R.L. Mobauer1 1 Physik-Department E15, Technische Universit at Munchen, D-85748 Garching, Germany 2 University of Oxford, Department of Physics, Oxford OX 1 3RH, UK 3 Center for Particle and Astrophysics, Berkeley, CA
Abstract
of metallic suboxides. Aluminum is known for its accommodating properties in tunnel barrier formation and is therefore used as the STJ material. Absorption of a single X-ray photon in a superconducting absorber generates an excess quasiparticle density to the already existing thermal quasiparticle population. Depending on the quasiparticle lifetime d and the di usion constant D, and thus on the quality of the tantalum, a certain fraction of this quasiparticle density gets trapped in the STJ elements at both sides of the Ta lm, due to a smaller energy gap in the junctions' material. At an operating temperature of about 50 mK, less than 1/10 of the critical temperAl ature Tc of all detector materials (TTa c = 4:46 K; Tc = 1:2 K), thermal contributions to the quasiparticle density are suppressed signi cantly. Thus, the tunnel current of a STJ element is roughly proportional to the fraction of the excess quasiparticle density getting trapped from the tantalum in its electrode. From the charge signals Q collected by two STJs, one can deduce the absorption position x0 and the energy E0 of the incident X-ray photon
A new design for a cryogenic X-ray detector has been developed and veri ed to be successful. Absorption of 5:89 keV X-rays in a 450 m160 m tantalum lm was detected by measurement of coincident charge signals from two aluminum superconducting tunnel junctions (STJs). For a 40 m section in the middle of the tantalum lm the energy resolution was 78 eV (FWHM), consistent with the permeability of the tunnel barriers of the STJs (Rn 2:5 ). We fabricated single Al/Alx Oy /Al STJs using a combined method of photolithography and shadow mask techniques. An energy resolution of E = 27 eV (FWHM) was obtained for 5:89 keV X-rays with a room temperature JFET preampli er. The normal conducting resistance (Rn ) of the devices was of order 1 , the dynamic resistance Rd varies from 50 k to 2 M . A simultaneously recorded electronic tail pulser showed a width of about 19 eV. However, only 11 % of the expected charge signal have been observed. A de cit in the charge yield of such symmetric STJs could be observed at low bias voltages, similar as in heterogeneous STJs. The charge yield has been measured in Al/Alx Oy /Al STJs over a bias voltage range from about 20 V to 300 V.
x0 1 E sinh 2 l with = p l : Q = C e 0 ! sinh D d (1) ! is the mean energy required to excite one quasiparticle in the absorber lm. The charge conversion constant C of our JFET preampli ers can be measured by feeding test pulses from an electronic tail pulser into the junction's readout circuit. The charge yield of the tunnel junction re ects all details of the tunnelling process and is a function of bias voltage. If the trapping electrode of the tunnel junction exhibits a higher gap than the counterelectrode, the signal yield may become zero or even change its sign. This has been measured with proximitized Ta/Al/Alx Oy /Al hetero tunnel junctions [2, 3]. To overcome that problem, we had to design a new X-ray detector layout: the tunnel junctions should exhibit the intrinsic energy gap of aluminum.
1 Introduction We fabricate X-ray detectors based upon superconducting Al/Alx Oy /Al tunnel junctions attached to a tantalum absorber strip. Quasiparticle trapping from a large area absorber into rather small STJ elements has been demonstrated with a Sn/Al/Alx Oy /Al device with an energy resolution of 60 eV (FWHM) at 5.89 keV, and a one dimensional spatial resolution of 5 m in a l = 470 m tin absorber lm [1]. This detector was inapplicable to X-ray astronomy and applied physics due to its degradation after thermal cycling. Tantalum is favored as absorber material because of its high X-ray absorption eciency, suciently long quasiparticle lifetime, thermal stability, and absence A11
2 Fabrication of a tantalum based X-ray detector
age of about 150 V the tantalum oxides are removed smoothly without severe damage to the Ta surface, which was in-situ checked by RHEED analysis. No signi cant changes in RRR, Tc , and Tc have been observed before and after the surface cleaning step [7]. The Al tunnel junctions are grown immediately after the Ta surface has been cleaned. Without breaking the vacuum, the substrate is moved from the IBEchamber to a dedicated aluminum growth chamber. The bottom and top lm of the tunnel junctions are deposited by an electron evaporator there. The oxide layer in between is grown in a third chamber, where the substrate is held into a low energetic oxygen plasma. Finally, the junctions' geometry is de ned using a selective wet-chemical aluminum etch. The tunnel junctions typically show a normal conducting resistance Rn of about 1 { 5 and a dynamical resistance Rd of 0.1{1 M , depending on the chosen oxidation parameters.
The aluminum tunnel junctions reside about 1 ? 2 m nearby the tantalum absorber, deposited onto the sapphire substrate as shown in g. 1. Two tunnel junctions, with an area of 100 m 140 m each, are connected to a Ta absorber strip with their bottom layer touching the Ta surface over 60 m length, separated by a distance of 450 m.
3 Correlated signals from a tantalum absorber strip
Figure 1: Layout of the tantalum based X-ray detector with two tunnel junctions attached at both ends. The size of the STJs of this test detector is 100 m 140 m, with an absorber strip of 160 m 450 m. The tantalum lms are deposited in a dedicated UHV-chamber using an electron beam evaporator. Prior to lm deposition, the system is baked out at 220o C for a period of several days. A base pressure lower than 10?10 mbar is reached. During the evaporation process H2 is dominating the total pressure, which is typically of order 5 10?10 mbar, while the R-plane sapphire substrates are heated in a tantalum oven up to 1100oC. At a deposition rate of 1 A/s the whole substrate wafer is coated with 2 { 6 k A of tantalum. RHEED, X-ray, Channelling, transition curves, and rest resistivity (RRR) measurements have shown high quality and epitaxial thin lm growth [5, 6]. The best RRR was 95 for a 5 k A lm, the critical temperature of the lms is identical to the bulk, and transition curves are 2 mK wide for good lms. Crystal analysis showed epitaxial growth of tantalum with the (100)plane tilted under a slight angle of 1:3o with respect to the sapphire R-plane, which is typical for the growth of a bcc-material as Ta or V onto a hexagonal substrate lattice [4]. In a subsequent photolithographical step the geometry of the absorber strip is de ned: an argon beam ion source is used to etch the tantalum o (IBE). After the removal of the photoresist, the tantalum has to be cleaned from its thermal grown oxide layer before evaporating the Al tunnel junctions. This requires the availability of a low energy ion beam source within the tunnel junction evaporation system. We have mounted a rf driven argon ion beam source to our existing multichamber UHV-system. At a beam volt-
Figure 2: Correlated signals from a test detector consisting of a single tantalum absorber with two aluminum tunnel junctions as shown in gure 1. The inset shows the signals representing a 40 m central region of the tantalum absorber. The 5.89 keV K line is measured with E = 78 eV (FWHM). The whole detector has been irradiated with an 55Fe source through a copper collimator with a 200 m wide slit. X-ray absorption took place over the entire area of the tantalum absorber. Absorption in the tantalum gives rise to correlated charge signals Q in both tunnel junctions (eqn. 1). Fig. 2 shows a A11
pair of curved lines, which re ects the 5.89 keV K and 6.49 keV K line of 55Mn. The quality of the common tantalum absorber is given by the parameter , which can be interpreted as the curvature of the banana style lines in the Q plot. For typical values of RRR = 80 we estimate the di usion constant to be about 0.13 m2/s, and thus obtain a quasiparticle lifetime of d = 0:3 s in the tantalum. From equation (1), we obtain an energy resolution of 78 eV for a 40 m central region of the total 450 m tantalum absorber ( g. 1, inset). Direct measurements of the spatial resolution have not been done yet, but a spatial resolution of about 6 m can be calculated from the measured energy resolution. Both tunnel junctions exhibited a rather high normal conducting resistance of Rn = 2:5 . To improve our detectors, we try to optimize the tunnel junctions' fabrication process further towards more transparent tunnel barriers.
thick. An energy resolution of E = 27 eV (FWHM) at 5.89 keV was measured for energy deposition in the top (2200 A) aluminum lm ( g.4). Due to small absorption of aluminum for 6 keV X-rays, a second pair of K and K lines is measured for absorption in the bottom electrode of the STJ with a width of E = 51 eV. To reduce background from substrate hits as good as possible, we mounted a copper collimator to spare out a slit about 200 m wide, through which the junctions have been illuminated. The conductance characteristic is plotted in g. 5. The bias point VB = 230 V, selected while recording the pulse height spectrum of gure 4, is marked as well. At this bias voltage the di erential resistance was Rd = 600 k .
4 Single Al-STJs as high resolution X-ray detector Single superconducting tunnel junctions are best to study e ects of quasiparticle tunnelling on the detector performance and exhibit high energy resolution as well. No additional layers for contact wirings or absorber lms are attached or in contact to the junctions. Thus, the electrodes of these junctions show the intrinsic energy gap of the used materials. Under these well-de ned conditions, it is feasible to calculate tunnel rates and the charge yield more exactly than for proximitized junction electrodes.
Figure 4: Pulseheight spectrum of a single aluminum tunnel junction illuminated by a 55Fe source. The peak on the right side shows the charge signals from a simultaneously recorded tail pulser with a width of 19 eV.
Figure 5: Measured conductance = (@I=@V )?1 as function of bias voltage. The energy gap of aluminum is indicated as vertical line at VB = 180 V, the bias voltage selected while recording pulses for the spectrum in g. 4 is marked at VB = 230 V.
Figure 3: Single aluminum tunnel junction detector fabricated by a combined method of shadow mask technique and photolithography. We fabricated single aluminum tunnel junctions with a combined method of photolithography and shadow masks. This is a quick and convenient process. These junctions have an area of 100 m 100 m ( g.3), both junction electrodes are about 2 { 5 k A
Nevertheless the substrate rate was high compared to the junction hits. This might have caused the broadening of a simultaneously recorded tail pulser to a width of 19 eV. By calibration with the pulser A11
l ? eVB . Quasiparticles are transferred from
we measure a charge collection eciency of 11 % (8 %) for the top (bottom) electrode. This might be attributed to a slow tunnel rate and fast quasiparticle recombination: the tunnel junction showed a rather high normal conducting resistance of Rn = 2:2 and charge pulse rise time of d = 70 s.
the left side to the right junction electrode. Thus, a positive electron current is resulting. Process 2 is allowed only for quasiparticles with an energy l + eVB above the Fermi level. A quasiparticle is transferred via a cooper pair mediated tunnelling process to the right side, but the electron transfer is opposite, resulting in a negative electron current. The total tunnel currents for electron and quasiparticle ow are determined by the sum of both processes with taking respect of the sign. The contribution of each process is strongly dependent on the distribution of quasiparticles above the energy gap and the density of nal states available for tunnelling, which is a function of the applied bias voltage. Thus, the charge yield de ned in (2) is expected to be a function of bias voltage and is given by
5 Charge signal yield as function of bias for Al-STJs The eciency of converting X-ray excited excess quasiparticles into a charge signal de nes the charge yield of a STJ. contains all details of the tunnel process and is a function of bias voltage. We use a chargesensitive preampli er at room temperature as readout device for the tunnel junctions. The preampli er's output voltage is proportional to the charge signal QT of the tunnel junction. QT re ects the sum of charge, having crossed the oxide barrier and is a fraction of the initially available charge Q0 . The charge yield then is given by the ratio of tunnel rate for the electrons Tel and the total quasiparticle loss rate Vtot [9] Q
= T = el : (2) Q
0
=
Tel :
Tqp + V
(3)
The tunnel rates for electrons ( Tel ) and quasiparticles ( Tqp ) are calculated in Ref. [2, 3] for tunnel junctions with unequal energy gaps. The bias dependance of the charge yield is experimentally veri ed there as well. For the symmetric junctions described in this article, formulas can easily be applied by substituting equal energy gaps (l = r = ). In gure 7 the measured bias dependence of the charge yield in a symmetric Al/Alx Oy /Al tunnel junction is shown. The solid line gives the calculated charge yield, assuming a Boltzmann distribution of the excess quasiparticles with an excess temperature T according to Ref. [2]. At bias voltages greater than 70 V the expected curve ts the measured charge yield for both absorption in top and bottom electrode perfectly. For lower bias the calculated yield obviously deviates form the data, but however the expected suppression due to the in uence of process 2 is reproduced by the assumption of a simple Boltzmann distribution.
Vtot
Fig. 6 shows the two fundamental tunnelling processes which can occur in a situation, where excess quasiparticles are excited in one junction electrode only. Quasiparticles are considered to disappear after having tunnelled once due to fast di usion into the aluminum contact leads ( g. 3). Thus, quasiparticle tunnelling contributes to the total loss rate V tot .
6 Conclusions A new fabrication process has been developed for our X-ray detectors, consisting of a superconducting tantalum absorber with symmetric aluminum tunnel junctions attached. The whole process is now feasible with the equipment of our institute and rst results from X-ray test measurements showed that tantalum is a favourable absorber for our X-ray detector. With single aluminum tunnel junctions we have demonstrated that a charge-sensitive JFET preampli er at room temperature can be used for a high resolution tunnel junction detector. This single junction devices are good candidates for detectors for high resolution X-ray analysis.
Figure 6: Basic tunnel processes in a superconducting tunnel junction with equal energy gaps of both electrodes. Excess quasiparticles are assumed to exist only on the left side. Process 1 is possible for all quasiparticles on the left side with an energy l 0 above the Fermi level, and for all positive bias voltages. Assuming energy conservation in the quasiparticle system for tunnelling, one gets the condition for process 1 to be allowed:
A11
Figure 7: Measured charge yield of a superconducting Al/Alx Oy /Al tunnel junction. The inset contains bias voltages up to 280 V, a second measurement at lower bias voltages is shown in the main frame. The line represents the calculated charge yield according to [2].
References
[8] J. Jochum, H. Kraus, B. Kemmather, M. Gutsche, R.L. Mobauer, "Electronic noise of superconducting tunnel junction detectors", Nuc. Instr. Meth. A, 338 (1994) 458-466. [9] J. Jochum et.al., Ann. Phys., 2 (1993) 631.
[1] H. Kraus et al, Phys. Lett. B 231 (1989) 195. [2] H. Kraus, M. Gutsche, P. Hettl, J. Jochum, and B. Kemmather, "Measurement of the tunnel rate in S-I-S' tunnel junctions as function of bias voltage", Journal of Superconductivity, 9 (1996) 245. [3] M. Gutsche, P. Hettl, J. Jochum, B. Kemmather and H. Kraus, "X-ray detectors with Ta/Al/AlO/Al hetero tunnel junctions" , in: Proceedings of the Sixth International Workshop on Low Temperature Detectors (LTD6), Beatenberg/Interlaken, Switzerland, August 28 - September 1, 1995, Nuc. Instr. Meth. A 370 (1996) 91. [4] M. Gutsche, H. Kraus, J. Jochum, B. Kemmather and G. Gutekust, "Growth and characterization of epitaxial vanadium lms" Thin Solid Films 248 (1994) 18-27. [5] M. Gutsche, PhD Thesis, TU Munich, Germany (1995). [6] P. Hettl, Diploma Thesis, TU Munich, Germany (1995). [7] M. Bruckmayer, Diploma Thesis, TU Munich, Germany (1995). A11
