Observation of negative persistent photoconductivity In an n-channel GaAs/AI,Ga 1-xAs single heterojunction J. Chen, C. H. Yang, and FL A. Wilson Joint Program for Advanced Electronic Materials, EIectrical Engineering Department, University of Maryland, CoIlege Park Maryland 20742 M. J. Yang Naval Research Laboratory, Washington, DC 203 75 (Received 13 December 1991; accepted for publication 24 February 1992) We report the first observation of negative persistent photoconductivity at 4.2 K in an n-channel modulation doped GaAs/Ale,,Gac6,As single heterostructure, where two-dimensional electrons have a mobility of - 550 000 cm2/V s when density is - 3.0 x 10” cm - 2, Based on extensive magnetotransport measurements, we conclude that the negative persistent photoconductivity effect comes from the time dependence of ( 1) the annihilation of two- dimensional electrons by photoexcited holes, and (2) the trapping and de-trapping of photoexcited electrons by shallow donors in doped Ale,,Gac6,As. A model that quantitatively explains the nonexponential recovery time is presented. Modulation doped GaAs/Al,Gar _ As heterostruc- tures can offer two-dimensional electron gas (2D EG) with high mobility’ at low electric fields and at low tem- peratures, because of spatial separation of the free carriers and ionized impurities. These structures also provide a sample system for studying some important physical phe- nomena. For example, the positive persistent photoconduc- tivity (PPC) effect, where the electrical conductivity en- hancement upon illumination persists after illumination is turned off, has been widely observed in such structures and is interpreted by the existence of DX (deep and complex) centers. The DX donorlike centers, after photoionized, would show a small (electron) recapture probability due to large lattice relaxation at low temperatures.2 The negative PPC effect, on the other hand, had been observed in a 2D hole system3 and in n-channel InAs/AlSb quantum wells.’ Several observations on single heterojunctions were also attributed to the negative PPC effect.5-8 There had been many studies aiming at understanding the origin of the PPC effect.’ We report in this letter the first observation of negative persistent photoconductivity of the 2D EG at a GaAs/ A10.33Gac67A~ single heterojunction, at 4.2 K. The modulation doped GaAs/Alc,33Gac,,As hetero- structure was grown by molecular epitaxy beam (MBE) at a temperature of 600 “C on a semi-insulating [loo] GaAs substrate. It consists of a 20.5 pm undoped GaAs buffer layer, a 500 A A10.33Gaa6TA~ undoped spacer, a Si-doped 400 A Ale3,Gae6,As ( 1.0X 10’*/cm3), and a 50 A GaAs undoped cap layer. The devices are photolithographically defined and etched into Hall bars with two current con- tacts and six potential probes. The magneto-transport mea- surements are performed in a liquid-He dewar at 4.2 K. We ftrst carefully carried out quantum Hall effect and Shubnikov-de Haas (SdH) effect measurements, which al- lows us to investigate spontaneously the 2D carrier con- centration n and mobility p, Then we used red LED to photoexcite the sample for a certain “light dosage” defined by the product of the LEDs driving current and duration. The peak energy of our red LED emission spectrum is around 2 eV. Magnetoresistance was measured by lock-in technique with an 18 Hz constant ac current source (peak to peak 100 nA). Both pn. or pxx are probed when a mag- netic field perpendicular to the heterointerface is slowly swept ( - 10 min) from 0 T to 9 T. Curve a in Fig. 1 shows a typical scan of pxx before illumination (hours after an accumulated light dosage of 670 mA ms), and three con- secutive scans (curves b-d) after only one more additional light excitation of 120 mA ms (the total accumulated light dosage now becomes 790 mA ms). We find that after illu- mination, n decreases suddenly (see the transitions of the SdH oscillations from curve a to b), and then slowly in- creases (see the transitions from curve b-c and then to d). The “recovery” of IZ after the initial decrease upon photo- excitation is on the order of several hours. The fact that pxx turns zero when filling factor v is 2, irrespective of the electron concentration n, indicates that the signal comes solely from- a 2D EG without a parallel channel. This unique feature is a direct result of quantum transport, and thus greatly simplifies our analysis of the experimental re- sults. To summarize the data from red LED illuminations, we plot the 2D EG density IZ against the accumulated light dosage in Fig. 2(a), and log(p) vs log(n) in Fig. 2(b). The n and ,U in dark after initial cooling down are 1.39 X10” cm -’ and 184 278 cm2/V s, respectively. The changes of n and p can be categorized into two parts, see Fig. 2(a). (A) Region I: As we gradually increase the illumination dosage, although n initially decreases upon illumination, the “recovery” time for the 2D EG is fast enough (on the order of several minutes) so that y1 can saturate prior to or during the first sweeping of the mag- netic field. Therefore, the measurements of pxx based on a slowly-scanning magnetic field only lead to the widely ob- served positive persistent photoconductivity. Note that the electron mobility also increases with the accumulated light dosage in this region. (B) Region II: Beyond a certain amount of accumulated dosage (about 1 mA 250 ms, see the dashed line separating part I and II), the negative photoconductivity persists even after 10 h. We find that 2113 Appt. Phys. Lett. 60 (17), 27 April 1992 0003-6951/92/172113-03$03.00 @ 1992 American Institute of Physics 2113 Downloaded 17 Sep 2002 to 129.2.94.235. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/aplo/aplcr.jsp0 0