Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.cplire.ru/html/lab234/pubs/2002JETLp507e.pdf Äàòà èçìåíåíèÿ: Mon Feb 28 21:09:05 2005 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 10:11:07 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: arp 220

JETP Letters, Vol. 76, No. 8, 2002, pp. 507­510. Translated from Pis'ma v Zhurnal èksperimental'nooe i Teoreticheskooe Fiziki, Vol. 76, No. 8, 2002, pp. 588­591. Original Russian Text Copyright © 2002 by Tarasov, Fominskii, Kalabukhov, Kuzmin.

Experimental Study of a Normal-Metal Hot Electron Bolometer with Capacitive Coupling
M. Tarasov1, 3, M. Fominskii1, 3, A. Kalabukhov2, 3, and L. Kuzmin2,
1 3

Institute of Radio Engineering and Electronics, Russian Academy of Sciences, ul. Mokhovaya 18, Moscow, 103907 Russia 2 Moscow State University, Vorob'evy gory, Moscow, 119899 Russia 3 Chalmers Technological University, S41296 GÆteborg, Sweden
Received September 19, 2002

Normal-metal hot-electron bolometer with capacitive coupling (CCNHEB) is a further development of the concept of a normal-metal hot-electron bolometer with Andreev mirrors (ANHEB). It was proposed to eliminate the frequency and energy restrictions inherent in ANHEB, in which Andreev mirrors act efficiently only with relatively long absorbers and at energies below the superconducting gap. An important advantage of the CCNHEB is its simple topology, in which the same tunnel junctions provide thermal decoupling, noise protection, temperature measurement, and it can be used for electron cooling. The temperature response of the bolometer was measured at temperatures down to 260 mK. The observed response dV/dT = 1.7 mV/K corresponds to the sensitivity S = 0.4 â 109 V/W. The measured noise at the amplifier output with this sample was found to be Vna = 4 nV/Hz1/2, which corresponds to a noise-equivalent power of 10­17 W/Hz1/2. To measure optical response, black-body radiation was used as a source of signal inside the cryostat. The source was a thin NiCr film sputtered on a thin sapphire substrate and suspended by nylon threads. Optical measurements proved to be in good agreement with the dc measurements. © 2002 MAIK "Nauka / Interperiodica". PACS numbers: 07.57.Kp

1. Topology of the samples and their fabrication. The concept of a normal-metal hot-electron bolometer with capacitive coupling (CCNHEB) was suggested in [1]. The key idea of the design is that the capacitive decoupling replaces Andreev mirrors, which are used for the thermal insulation of an absorber from electrodes in a normal-metal hot-electron bolometer with Andreev mirrors (ANHEB). This allows one to avoid the frequency restriction inherent in the Andreev mirrors, which arises if electrons are superheated above the energy gap. Matching of the absorber impedance at the signal frequency is accomplished through the same capacitors that provide a high potential barrier for hot electrons. For a planar complementary antenna on the silicon substrate, the impedance can be estimated by the formula Zant = 377[2(1 + )]1/2 = 80 . The bolometer impedance is the sum of capacitive, inductive, and resistive components connected in series: Z
bol

2 = --------- + i L + R . iC

(1)

For the planar antenna with real impedance, the inductance of the absorber stripe can be counterbalanced by the tunnel junction capacitance. For an absorber length of 10 µm, ZL = 20 at 300 GHz, the required capacitive impedance is 30 fF, which can easily be obtained with tunnel junctions of size 1 â 0.5 µm2. As a result, an additional band-pass filter forms, which additionally reduces the undesirable action of the thermal back-

ground radiation. Simultaneously, the same tunnel junctions act as temperature sensors and provide electron cooling, similar to the Peltier effect in semiconductors. Samples were prepared on silicon substrates of size 7 â 7 mm with 16 contact pads. Each chip included six CCNHEB structures, of which one was integrated with a log-periodic antenna, four integrated with double dipole antennas for central frequencies of 300 and 600 GHz, and one test structure with two additional tunnel junctions was designed for measuring temperature along the absorber stripe (cf. microphotograph in Fig. 1). The contact pads and antennas were fabricated by UV lithography and sequential thermal evaporation of 10-nm Cr, 40-nm Au, and 10-nm Pd layers. The structure of bolometer was formed in a single vacuum cycle by direct electron-beam lithography and shaded evaporation at different angles through a double mask made from PMMA and COPOLYMER photoresists. In so doing, 60 nm Al was evaporated, oxidized for 2 min in oxygen at a pressure of 10­1 mbar, after which a double-layer Cr/Al film was evaporated at a different angle. The normal metal film in the bolometer was 8 µm long, 0.2 µm wide, and 80 nm in thickness. The resistance of the bolometers with double-layer films consisting of 50 nm Cr and 10 nm Al was about 100 , which was close to the optimal resistance of both log-periodic and double dipole antennas. The results of measuring with a sample containing 50% chromium and 50% aluminum are also presented for comparison. The capaci-

0021-3640/02/7608-0507$22.00 © 2002 MAIK "Nauka / Interperiodica"


508

TARASOV et al.

260 mK. Figure 2 presents the IV curves for one of the samples. These dependences can be used to obtain the temperature dependence of voltage and the sensitivity dV/dT at different temperatures and dc currents (Fig. 3). To determine the bolometer sensitivity and noise for a real microwave signal, a black-body radiation source with modulated temperature (Haller-Beeman Assoc. Inc.; see [4] for detail) was used as a signal source. The source was a thin NiCr film sputtered on a thin sapphire substrate suspended by thin nylon threads. On passing current through the film, it is heated and starts to radiate. The source was mounted on the cold plate of a cryostat at the base temperature of 1.7 K, and its output horn was directed toward the lens with CCNHEB placed at a distance of several centimeters from the source. After applying a power of several milliwatt from the power supply, a radiation with equivalent temperature of up to 100 K arises. The source is capable of producing modulated signals over a wide frequency range. The dependences of the bolometer response on the bias voltage, as measured in the temperature variation regime (Fig. 4, curve DVT) and for the built-in radiation source (curve DV 249 mV), demonstrate a reasonable agreement with the theoretical estimates. To estimate the NEP, low-frequency noise was measured at the bolometer output. The frequency dependence was measured for the voltage fluctuation spectral density, and the noise dependence on bias voltage was measured at several frequencies. In the sample with 50 nm chromium and 10 nm aluminum, the noise of bolometer and amplifier at the operating point with maximal response was 4 nV/Hz1/2. The dependence of noise spectral density on bias voltage in the white-noise region is shown in Fig. 5 for the sample with 30 nm chromium and 30 nm aluminum. It follows from these measurements that the noise spectral density in amplifier can be put equal to 3 nV/Hz1/2, which corresponds to the specifications of an OP27 operational ampli-

Fig. 1. Microphotograph of a bolometer chip with 16 contact pads and six CCNHEB structures. Log-periodic antenna is at the center; two double dipole antennas for the central frequency 300 GHz are on the right and two antennas for 600 GHz are on the left and at the bottom.

tance of tunnel junctions was about 3 fF, so that, for two capacitors connected in series, 100 of bolometer, and intrinsic inductance, the resonance frequency was about 250 GHz. 2. Experimental results. The samples were placed on a flat surface of a hyperhemispheric sapphire lens that was mounted on the He3 evaporator of a closedcycle absorptive cryo cooler (see [2]). The cryo cooler was mounted on the cold plate of a helium cryostat (Infrared Labs [3]) with optical window. Liquid helium vapor was pumped until a temperature of 1.7 K was reached. The minimal temperature in the system was

Fig. 2. Current­voltage characteristics of bolometer, as measured at temperatures from 260 mK to 1.5 K.

Fig. 3. Temperature dependences of the voltage on bolometer for various bias currents. JETP LETTERS Vol. 76 No. 8 2002


EXPERIMENTAL STUDY OF A NORMAL-METAL HOT ELECTRON BOLOMETER DV 249 mV

509

Fig. 4. CCNHEB voltage response (curve DVT) to a change in temperature and (curve DV 249 mV) to the black-body radiation.

Fig. 5. Dependence of the noise in the range 1­10 kHz on the bias voltage on a bolometer with 30 nm chromium and 30 nm aluminum. At biases higher than 500 µV, noise is restricted by the amplifier.

fier--bolometer with an amplifier at the operation point of 200 µV at a level of 6.5 nV/Hz1/2 for 3 kHz and higher, and 8.5 nV/Hz1/2 for 1 kHz. High response for the samples with double-layer absorber films is different from the value predicted by a simple theory for SIN (superconductor­insulator­normal metal) gauge and may be due to the influence of electron cooling. 3. Discussion of results and estimate of the CCNHEB sensitivity. The theoretical values of response for the SIN junction can be obtained using the simple analytic expression 1.76 T c k eV I(V , T ) = ----- 11 TT c exp ­ --------------- sinh ------ , (2) kT eR T which gives a good agreement with experimental data for the SIN junction with a purely chromium absorber. This expression can be used to determine maximal temperature response for the SIN junction at various ambient temperatures. For our base temperature of 270 mK, a single junction should have a response of about 0.5 mV/K; to increase it twofold (to 1 mV/K), electron temperature should be 163 mK. The experimentally observed high values of response can be explained by the combined action of two mechanisms: electron cooling and increase of the junction nonlinearity due to the proximity effect, so that it should be considered as an SINS structure, which approaches the SIS* (superconductor­insulator­superconductor with reduced gap) structure in the zero-temperature limit. The noise in a structure consisting of two tunnel junctions connected in series can be calculated using
JETP LETTERS Vol. 76 No. 8 2002

the following simple expression for the current fluctuation spectral density in a single junction: 2 eV eV S I(V ) = --------- coth --------- . 2 kT R (3)

To obtain voltage fluctuations, one should multiply this expression by the dynamic resistance of junctions. It can be found from Eq. (2). The calculated dependences of the dynamic resistance and noise on the bias voltage proved to be close to the measured dependences. In our first experiments [5], the values of the limiting CCNHEB parameters (NEP = 2.5 â 10­17 W/Hz1/2) were far from the theoretical estimates, because we used non-optimal samples leading to a high noise level in the subsequent signal amplifier. In this work, the slope dV/dT = 1.7 mV/K was obtained for the MF46p616 sample at 260 mK. When estimating the voltage response S = dV/dP = (1/G)dV/dT, we used the values for the heat conduction G = dP/dT = 3T 4, characteristic constant of material = 2.5 â 109 W m­3 K­5 (measured in a separate experiment with ANHEB samples calibrated against the dissipated power of a dc power supply in a copper film; for the samples with doublelayer absorber films, an additional detailed study is needed for the temperature dependence of the characteristic constant of material), and sample volume n = 8 â 10­20 m3, which gives G = 5 â 2.5 â 109 â 8 â 10-20(0.256)4 = 4.3 â 10­12 W/K and S = 1.7 â 103/(4.3 â 10­12) = 0.4 â 109 V/W. For the output noise of bolometer and amplifier, Vna = 4 nV/Hz1/2, the NEP is Vna/S = 6 â 10­18 W/Hz1/2. An example of the measured NEP is presented in Fig. 6. In the absorber film consisting of normal and superconducting layers, all hot electrons are concentrated in the normal metal film because of the Andreev reflection. This reduces effectively the volume


510

TARASOV et al.

Fig. 6. Noise-equivalent power measured for the sample with 30 nm Cr and 30 nm Al in the absorber film.

and, correspondingly, enhances the volt­watt sensitivity. The gain is caused by the effective decrease in the volume of normal metal, the proximity effect, the Andreev reflection, and the electron cooling. An important advantage of a high response is that the requirements to the subsequent amplifier become less stringent. The electrical NEP presented above should be compared with the optical NEP measured using black-body radiation. For the MF46p6-16 sample with a log-periodic antenna, the radiation linewidth was found to be 5° in the E plane and 30° in the H plane. A decrease in the source radiation temperature was = (3/46)(3/7.7) = 1/40. Moreover, one should take into account the reflection loss at the vacuum/insulator interface, the impedance mismatch, the degree of blackness of the blackbody source, the mismatch in the direction of antenna major lobe, and the side lobe loss, which altogether add

no less than 10 dB additional loss. As a result, the radiation weakens more than 400 fold. The received radiation power P = kT f = 10­13 W and a response slope of 0.4 â 109 V/W should provide a value of 40 µV for the voltage response. The value of 20 µV measured for the voltage response to the black-body radiation is in a good agreement with the measured value. 4. Conclusions. Normal-metal hot-electron bolometer with capacitive thermal insulation integrated with planar submillimeter wave antennas have been designed, fabricated, and experimentally investigated. The electrical sensitivity was as high as 1.7 mV/K, the response S = 0.4 â 109 V/W, and the noise-equivalent power was 10­17 W/Hz1/2. The measured response to the black-body radiation is in good agreement with the noise-equivalent electrical power measured at various temperatures. The results obtained give evidence for the efficient operation of the bolometer with capacitive thermal insulation. This work was supported by the STINT, KVA, VR, and INTAS (grant nos. 01-686, 00-384). REFERENCES
1. L. Kuzmin, Physica B (Amsterdam) 284­288, 2129 (2000). 2. L. Duband and B. Collaudin, Cryogenics 39, 659 (1999). 3. Infrared Labs, http://www.irlabs.com/catalog/cdewarindex.html. 4. Haller-Beeman Assoc. Inc, http://www.haller-beeman.com. 5. B. Choi, L. Kuzmin, D. Chouvaev, and M. Tarasov, in Proceedings of 4th International Workshop from Andreev Reflection to the International Space Station, Bjorkliden, Kiruna, Sweden, 2000, p. 289.

Translated by V. Sakun

JETP LETTERS

Vol. 76

No. 8

2002