2 Department of Physics and Astronomy, University of Rochester, Rochester NY, USA

Direct comparison of performance has been done for a Pb_1-xSn_xTe(In) photodetector, a Si(Sb) BIB structure and a state of the art Ge(Ga) photoconductor in the integration cavity. The Pb_1-xSn_xTe(In) photodetector shows several orders of magnitude higher responsivity S_I then the Si(Sb) BIB at the wavelength l ~ 14 m m. Persistent photoresponse with S_I ~ 10³ A/W at 40 mV bias and 1 s integration time at the wavelengths of l = 90 and 116 m m has been observed for the first time for the Pb_1-xSn_xTe(In) photodetector. This value is by a factor of ~ 100 higher than the respective parameters of the Ge(Ga) photoconductor in the same conditions.

Most of the sensitive photodetecting systems operating in the far infrared wavelength range (20-200) m m are based on germanium or silicon doped with shallow impurities [1]. The highest cutoff wavelength reported l _co » 220 m m corresponds to the uniaxially stressed Ge(Ga) [2]. The main advantage of germanium and silicon is very well developed growth technology that allows receiving materials with perfect crystalline quality and extremely low uncontrolled impurity concentration.

An alternative possibility for construction of sensitive far-infrared photodetectors is provided by unique features of a narrow-gap semiconductor system indium-doped Pb_1-xSn_xTe. Exciting results of fundamental research on the group III - doped lead telluride - based alloys [3] provided possibilities for construction of far-infrared photodetectors based on basically new physical principles [4]. Results of these activities are very promising since performance of far-infrared photodetectors based on doped lead-tin tellurides is comparable or better than for analogs based on doped Ge or Si. On the other hand, some important parameters of these photodetectors, for instance, their spectral characteristics, are not yet determined explicitly.

We have performed direct comparison of performance of a Pb_1-xSn_xTe(In) photodetector, a Si(Sb) blocked impurity band (BIB) photodetector and a Ge(Ga) photodetector using the same cryogenic equipment and measuring electronics. Besides that, we have done first direct measurements of photoresponse of Pb_1-xSn_xTe(In) photodetectors at wavelengths 90 and 116 m m.

Doping of the lead telluride with indium in amount exceeding concentration of other impurities and defects results in the Fermi level pinning effect [5]. Paradoxial consequence of this effect is homogenization of electrical properties of the semiconductor despite it is heavily doped and posesses high number of growth defects. This feature results in almost absolute reproducability of sample parameters independently on the growth technology and degree of purification of initial components. High sample homogeneity gives rise to high carrier mobility reaching 10⁵-10⁶ cm²/V× s at low temperatures.

Position of the pinned Fermi level E₀ may be tuned by alloying [6]. Conductivity of material being in the dielectric state (0.22 < x < 0.28) is defined by activation from the impurity local level E₀, and therefore free carrier concentration is very low n,p < 10⁸ cm^-3 at temperatures T < 10 K. It should be noted that the free carrier concentration in the undoped alloys is defined by the electrically active growth defects. Their concentration is never smaller than 10¹⁵ cm^-3. So a very unusual situation is realized, when a heavily doped narrow-gap semiconductor with high number of the growth defects acts as an almost ideal semiconductor with practically zero background free carrier concentration and very high electrical homogeneity. This makes very attractive the idea to use this material as an infrared photodetector.

External infrared illumination leads to the substantial increment of material conductivity at the temperatures T < 25 K [7] . High amplitude of photoresponse at the low temperatures is a consequence of the persistent photoconductivity effect: the photoresponse increases linearly in time providing a kind of «internal integration» of the incident radiation flux. When the radiation is switched off, the conductivity relaxes very slowly. The chatacteristic relaxation time t > 10⁴ s at 4.2 K < T < 10 K, then it sharply decreases with the temperature rising, and t ~ 10^-2 s at T » 20 K. The effect is defined by the specifics of impurity states that form DX-like centers [8].

If sample temperature is so that t is higher that the operation time required, then a photoresistor may operate only if there exists a possibility to return to the initial "darkness" state, i.e. to quench quickly the persistent photoconductivity. Moreover, periodical accumulation and successive fast quenching of the photosignal leads to the substantial gain in the S/N ratio with respect to the case of ordinary single photodetectors.

The most efficient way of quenching of the persistent photoconductivity in Pb_1-xSn_xTe(In) is application of strong microwave pulses to sample contacts [4]. Using this technique it is possible localize the long-living photoexcited free electrons for 10 m s. Therefore it became possible to operate in the regime of periodical accumulation and successive fast quenching of photoresponse. Moreover, application of microwave pulses in some special regime results in giant increment of the quantum efficiency up to ~ 100 [4].

Microwave quenching however requires some special equipment. More easy way of this quenching is just heating up a sample above T_c = 25 K and successive cooling down in darkness.

We have performed direct comparison of performance of a Si(Sb) BIB, a state of the art Ge(Ga) photoconductor in the integration cavity and a Pb_0.75Sn_0.25Te(In) photodetector. In Pb_1-xSn_xTe(In) with x = 0.25 the Fermi level is pinned in the gap at ~ 20 meV below the conduction band bottom [6]. The thermal activation energy of the ground impurity state corresponds to the wavelength l » 60 m m. The photodetectors were mounted on a sample holder on the backside of a helium vessel (Fig.1).

Fig.1. Experimental setup: 1 - 300 K or 77 K blackbody; 2 - 300 K window; 3 - 77 K filter; 4 - 4.2 filter and a stop aperture; 5 - cold filter on the filter wheel; 6 - filter wheel; 7 - sample; 8 - liquid helium can. Lines with arrows indicate multiple radiation bouncings resulting in radiation «leak».

A thermal screen was attached to the helium vessel to minimize the background radiation. Another screen had the liquid nitrogen temperature. A window separated the vacuum space of the dewar and the atmosphere at the room temperature level. A 300 K or 77 K blackbody was used as an external radiation source. This blackbody radiation passed on its way to a sample through a series of filters including a 300 K window, a 77 K filter, a 4.2 K filter and a filter on a filter wheel inside the helium screen. A stop aperture on the helium screen was restricting the field of view of a photodetector to the input window. The solid angle – detector area product was
1.3× 10^-4 cm²/srad for the Ge(Ga) photoconductor, and 2.7× 10^-6 cm²/srad for the
Pb_1-xSn_xTe(In) photodetector and the Si(Sb) BIB, that is by a factor of ~ 50 smaller.

Quenching of the persistent photoconductivity could not be provided in this setup, so a Pb_1-xSn_xTe(In) sample could be transferred to the ground «dark» state only by heating to T_c ~ 25 K and successive cooling. Such cooling was taking 3 - 5 min.

The «dark» current-voltage characteristics of the 1.8 m m thick Pb_0.75Sn_0.25Te(In) MBE-grown film of 1.5´ 0.2 mm² area is shown in the Fig.2, curve 1. It was taken in the conditions of complete screening of background radiation with a piece of metal instead of the «helium» filter. The curve 2 at the same figure corresponds to the «dark» I-V characteristics of a Si(Sb) BIB. After that the filter was returned to its place.

Fig.2. I-V curves of the Pb_0.75Sn_0.25Te(In) photodetector (curve 1) and of the Si(Sb) BIB (curve 2) taken in darkness and the I-V curve for the Si(Sb) BIB under the action of the 14 m m radiation «leak» (curve 3).

The standard experimental procedure used for Si and Ge – based photodetectors is the following. First, the I-V curve is measured when a filter on the filter wheel is away from a sample. Then the filter wheel is rotated to the position when the filter is in front of a sample. The characteristic time of this manipulation is several seconds. Then the blackbody radiation passing through the system of filters affects the sample directly. The I-V curves are taken for 300 K and 77 K blackbody temperature. The transmission spectra of filters are known, and it allows to calculate the incident photon flux and radiation intensity for any given set of filters. Consequently, the photodetector parameters like quantum efficiency, responsivity and some others can be calculated for every bias.

The results of this experiment for a Si(Sb) BIB of 0.4´ 0.4 mm² area and a system of filters providing narrow transmission line at l » 14 m m are shown in the Fig.3. The normal operating bias for this photodetector is (2 - 2.5) V.

Fig.3. I-V cuves of the Si(Sb) BIB taken under the action of blackbody radiation. Figures near the curves correspond to the blackbody temperature.

Repeating the same procedure with the Pb_0.75Sn_0.25Te(In) film led to immediate amplifier overload (sample current more than 0.25 m A) even for the lowest bias applied 40 mV for at least less than the characteristic time of the filter wheel rotation. Moreover, even when the filter was away from the sample, the sample current increased in time indicating that there is a radiation «leak» through the gap between the filter wheel and the helium screen due to multiple bouncing of the incident radiation (arrows in the Fig.1). It should be noted that the helium screen was covered by black paper providing very low reflectivity. The spectral composition of this leak corresponds to transparence of all filters except for the one on the filter wheel. It is a relatively wide band with the maximum at 14 m m and a halfwidth of
2 m m. The long-wavelength «tail» of transparence goes up to 23 m m. The I-V curves taken at different time after the sample cooldown are shown in the Fig.4 together with the «dark» curve.

Fig.4. I-V curve of the Pb_0.75Sn_0.25Te(In) film taken in darkness (curve D), and its evolution under the action of the 14 m m radiation «leak». Figures near the curves correspond to time in min. after the sample cooldown to 4.2 K.

This radiation leak could be detected by the Si(Sb) BIB as well. The photocurrent provided by this leak for the highest normally used bias of 2.5 V in this BIB is only ~ 10^-12 A (Fig.2, curve 3), whereas the photocurrent in the Pb_1-xSn_xTe(In) film measured just after the cooldown at the same bias is ~ 10^-7 A. Existing calibration of the BIB has allowed to estimate the photon flux corresponding to the radiation «leak» ~ 10⁸ photons/s. Estimates of the quantum efficiency of the Pb_1-xSn_xTe(In) photodetector in this spectral range gives the value h ~ 1. More accurate result requires more exact data concerning the accumulation time.

This kind of data were obtained for two other sets of filters providing
90 m m and 116 m m line transmission. In this case the radiation «leak» is also present, but it is much smaller than in the 14 m m case (Fig.5).

Fig.5. I-V curve of the Pb_0.75Sn_0.25Te(In) film taken in darkness (curve D), and its evolution under the action of the 83 m m radiation «leak». Figures near the curves correspond to time in min. after the sample cooldown to 4.2 K.

The spectral composition of this leak is a wide band with the blue cutoff at l » 83 m m going to the longer wavelenghts.

The photorespose increase rate is considerably smaller: for 2 V and 4 V bias the amplifier overloads in several seconds after the sample is exposed to radiation, and for the 40 mV bias that is the usual bias for Ge(Ga) BIBs operating in this spectral region, the amplifier overload occurs in several dosens of seconds (Fig.6).

Fig.6. Photoconductivity kinetics of Pb_0.75Sn_0.25Te(In) film exposed to blackbody radiation filtered at l = 90 and 116 m m. Figures near the curves correspond to the blackbody temperature and the wavelength. T = 4.2 K. Illumination starts at the moment t = 0 (with accuracy of (3-4) s).

The last circumstance allows to make more accurate estimates of the quantum efficiency h » 1 % for l = 90 m m and h » 2 % for l = 116 m m. These figures look reasonable since this spectral range corresponds to the Restrahlen band, and up to 99 % of incident radiation is reflected by the lattice. Antireflection coatings and operation in the microwave stimulation regime should increase the quantum efficiency. The responsivity at the 40 mV bias is S_I ~ 10³ A/W for the accumulation time of 1 s. The I-V curves taken for the same spectral composition of radiation flux, but with with a factor of 50 higher intensity (see above), for the Ge(Ga) photoconductor in the integrating cavity are shown in the Fig.7. The responsivity estimates give a value of (3.3 – 3.5) A/W depending on the wavelength, that is by at least two orders of magnitude lower than for the Pb_1-xSn_xTe(In) photodetector.

Fig.7. I-V curves of the Ge(Ga) photoconductor exposed to blackbody radiation filtered at l = 90 and 116 m m. Figures near the curves correspond to the blackbody temperature and the wavelength. T = 4.2 K. The label L corresponds to the I-V curve measured under the 83 m m radiation «leak».

The energy of a radiation quantum corresponding to 90 and 116 m m wavelengths is considerably smaller than the thermal activation energy of the ground impurity state in Pb_0.75Sn_0.25Te(In) . It means that the main contribution to the effect is provided by electrons being not in the ground, but in the metastable local state. Population of these levels by application of microwave pulse (microwave stimulation of the quantum efficiency), injection of electrons from contacts, background thermal radiation - result in the red shift of l _co compared to the ground «darkness» state. The value of l _co is still undefined for this case. However some indications exist that the red cut-off wavelength may substantially exceed 220 m m - the highest l _co obtained so far [2].

Another important note is the following. Mobility of photoexcited electrons in the film used in our experiment is very low - only 1200 cm²/V s at T = 4.2 K, that is by two orders of magnitude lower than in the good-quality bulk samples [3]. However an attempt to measure the photoresponse characteristics of such a bulk sample performed in our experiment failed. The sample cooled in complete darkness shows semiinsulating behavior, however the same sample cooled down in the intstallation used in our experiment with closed helium schield has resistance at least lower than 2 kW at T = 4.2 K after annealing to 25 K. This may be due either to the presence of very small radiation leak even in the closed schield that «overloads» the sensitive sample, or to residual radiation of the shield itself at very long wavelengths. Estimates of the residual radiation flux done with the help of the state-of the art Ge(Ga) BIBs show that it is at least lower than several dosens of photons per second. Microwave quenching of the persistent photoconductivity is needed to calculate parameters of this Pb_1-xSn_xTe(In) photodetector.

Direct comparison of the Pb_0.75Sn_0.25Te(In) photodetector, the Si(Sb) BIB structure and the ge(Ga) photodetector using the same cryogenic equipment and measuring electronics shows that the responsivity of the Pb_0.75Sn_0.25Te(In) photodetector is by (2-7) orders of magnitude higher depending on the bias applied, the accumulation time and the incident radiation wavelength.

We have performed first direct observation of persistent photoresponse in the Pb_0.75Sn_0.25Te(In) photodetector at the wavelengths of 90 and 116 m m that are considerably longer than the wavelength corresponding to the thermal activation energy of the ground impurity state. Indications exist that the red cutoff wavelength for this photodetector may exceed 220 m m - the highest l _co observed so far.

The research described in this paper was supported in part by the NATO Collaborative Research Grant HTECH.EV 971387.

1. J.Wolf and D.Lemke, Infrared Phys. 25, 327 (1985).

2. E.E.Haller, M.R.Hueschen and P.L.Richards, Appl. Phys. Lett. 34, 495 (1979).

3. B.A.Akimov, A.V.Dmitriev, D.R.Khokhlov, and L.I.Ryabova, Phys. Stat. Sol. A: 137 9 (1993).

4. S.N.Chesnokov, D.E.Dolzhenko, I.I.Ivanchik, and D.R.Khokhlov, Infrared Phys. 35, 23 (1994).

5. V.I.Kaidanov, R.B.Mel'nik, and I.A.Chernik, Fiz. Tekh. Poluprov. 7, 759 (1973) [Sov. Phys. Semicond. 7, 522 (1973)].

6. B.A.Akimov, L.I.Ryabova, O.B.Yatsenko, and S.M.Chudinov, Fiz. Tekh. Poluprovodn. 13, 752 (1979) [Sov. Phys. Semicond. 13, 441 (1979)].

7. B.M.Vul, I.D.Voronova, G.A.Kaluzhnaya, T.S.Mamedov, and T.Sh.Ragimova, Pis'ma v Zh.Eksp.Teor.Fiz. 29, 21 (1979); B.A.Akimov, N.B.Brandt, S.O.Klimonskiy, L.I.Ryabova, and D.R.Khokhlov, Phys.Lett. 88A, 483 (1982).

8. D.R.Khokhlov and B.A.Volkov, Proc. 23 Int. Conf. Phys. Semicond., Berlin, Germany, July 21-26, 1996, ed. M.Scheffer, R.Zimmermann, World Sci. (Singapore), 4 2941 (1996).