Irradiation damage in GaAs particle detectors

Irradiation damage in GaAs particle detectors

Nuclear lnstrtilllel~ts and methods m Physics Research A 395 i 1997) 88-93 NUCLEAR INSTRUMENTS & METHODS IN PHYSlCS RESEARCH SectlonA ELSEYIER ...

735KB Sizes 0 Downloads 75 Views

Nuclear

lnstrtilllel~ts

and methods

m Physics Research A 395

i 1997)

88-93

NUCLEAR INSTRUMENTS & METHODS IN PHYSlCS RESEARCH SectlonA

ELSEYIER

Irradiation damage in GaAs particle detectors M.R. Brozel

Abstract GaAs particle detectors for use in the LHC at CERN will be subject to a considerable fluence of energetic particles. In this note, possible effects of this i~adiation on the production of damage centres in these detectors will be discussed.

K~vwotd.s: Irradiation damage; GaAs; Particle detectors

1. Introduction The effects of high-energy particle irradiation on virtually all crystalline materials can be expressed in terms of relatively simple rules. These are: (1) The particles will give up energy to the lattice by the creation of interstitial atoms and vacancies (Frenkel defects) by forcing substitutional host atoms off their sites. The energy required to do this in GaAs is about 12 eV. (2) If these primary defects recombine either at the place that they were made or elsewhere in the lattice. following diffusional migration of either or both species, the damage is removed and the host lattice is said to have undergone “selfhealing”. After such recombination of the primary defects the material will show no evidence of the original damage and after any physical assessment it will be concluded that the crystal is “radiation hard’. (3 ) If primary defects are not annihilated in this way, the material will retain evidence of the damage. If the residual damage centres are electrically active (by introducing levels in the forbidden band gap, for instance), it will be concluded by electrical assessment after the i~adiation that the crystal is damaged and that the crystal is not radiation hard. Of course, if these defects are electrically inactive. non-electrical assessment (such as optical absorption) may reveal damage although electrically the material will be “radiation hard’. (4) In semiconductors much of the primary damage is self-healed by the rapid recombination of interstitialvacancy pairs. However, residual damage occurs either by the trapping of some of these defects by impu~ty atoms, their agglomeration into complexes or, in the case of compound semiconductors. by reactions between atoms of one sub-lattice and vacancies of the other. This results in 016X-9002/97i$i PII

7.00 Copyright

SOl68-9002(97)00623-2

@ I997 Elsevier

the semi-permanent effects that are observed externally as radiation-induced damage. Most of these secondary defects are electrically active in some way. (5) Under circumstances of high particle fluence, further agglomeration of primary and secondary defects can result in larger defects such as dislocations (and associated stacking faults), precipitates and voids, In this note, radiation damage effects in silicon are reviewed as it is on this material that most work has been performed. Possible extensions of these observations to GaAs will made. Mechanism (5) is not really applicable with the fluences expected at the LHC and will not be discussed.

2. Radiation effects in silicon Silicon (Si) provides a unique material to study irradiation damage because of the wide applicability of Electron Paramagnetic Resonance (EPR) [ 1,2]. It is this fact that had led workers several decades ago to make a thorough and unambiguous investigation of both primary and secondary damage in this system. The application of this powerful technique was led by Watkins (often with Corbett ) whose work is referenced widely [3]. In essence, the effects of high-energy particle irradiation are as follows

( 1 ) Apparently regardless of the type of incident particle, whether it be electrons or neutrons, the particles most used, Si interstitials and vacancies are formed which then either rapidly annihilate or become mobile by diffusion.

Science B.V. All rights reserved

f2) Si interstitials are mobile in p-type Si below 4 K: the Si interstitial has never been observed because it is always mobile under conditions of i~diation. The properties of this species are known by its interaction with other atoms. Nowever, the identi~cation of the complexes thusformed is unambiguous. (3 ) Vacancies are also very mobile and migrate well below room temperate in the range 70-200 K.

( 1) Mobile Si interstitials are trapped by acceptor atoms (such as B, Al, Ga) and neutral carbon (C). and oxygen (0), atoms. B, Al and Ga are intentional shallow acceptors. C is an impurity often found in float-zone (FZ) Si and is apparently introduced in the reduction of SiOz to Si. 0 occurs at concentrations exceeding 5 x IO" cme3 in Czochralski (Cz) Si due to a reaction between the molten Si and the SiOl crucible from which it is pulled as a single crystal. (2 ) The trapping of Si,,, by acceptors such as B in p-type Si has two effects: (a) There is a resultant loss of electrical conductivity, both because the (Z&-B] pair is no longer a shallow acceptor centre but also because the defect now acts as a scattering centre, reducing the hole mobility as well as acting as a deep level defect. (b) The trapping of a Si,,, atom means that the associated vacancy remains in the lattice. This adds to the residual damage. (3 ) The trapping of Si,,, by 0 or C produces deep electrical levels and leaves a vacancy (see 2(b)). Fig. 1 shows the structures of some Sii”~-impuri~ complexes.

The vacancy is electrically active, states, V-I -I, V-, VO, V- and V” sible via excitation within the band are mobile below 300K they tend f@llows:

having up to five charge all of which are accesgap. However, as they to form complexes, as

(al Al,”

( 1) Vacancies are trapped by donor atoms (such as P. Sb, and As). These produce the so-called “E-centres”, such as the [V-P] close pair. E-centres are not shallow donors and the electrical conductivity is reduced by their fo~ation. Vacancies are also trapped by acceptors (such as B and Al, as before). In addition they are trapped by 0 atoms to produce “A-centres”. All these defects produce deep levels in the band gap. As an example of the secure of these trapped vacancy complexes, Fig. 2 shows the structure of the A-centre. All these defects produce deep levels in the band gap. As an example of the structure of these trapped vacancy complexes, Fig. 2 shows the st~c~re of the A-centre. (2) Vacancies that are not trapped by impurity atoms form divacancies and it is these, rather than single vacancies, that are observed after room temperature irradiation. Divacancies have several charge states, all of which produce deep levels in the band gap. (3) As discussed above with regard to interstitial trapping, damage to the electrical properties of the crystal are associated with the loss of shallow dopant activity, the creation of deep electronic levels and the creation of electrical scattering centres. In addition. of course, the trapping of vacancies impedes the healing process of the lattice by removing vacancies required for the annihilation of interstitials.

Increasing the temperature causes diffusional mobility to increase and also results in the dissociation of complexes into their components. Ideally, this will result in all trapped interstitials and vacancies becoming mobile, followed by their recombination into their original substi~tional positions. Indeed some direct annealing is observed, with several annealing stages being observed up to quite high temperatures. As an example, it is found that E-centres anneal at temperatures up to 400 K, while A-centres anneal at 600 K. However, not ail defects are annihilated in this way.

Ib) B:

Fig. 1. Structures identified by EPR are produced when interstitial silicon is trapped by substitutional (a) afuminium, (b) boron, and (c) carbon. The impurity atoms are shaded (from Ref. 131).

IIt. ~HARACTERISATfO~~MODELLl~G

Fig. 2. Model of a silicon unit cell containing an OV centre. The central atom is removed, the 0 atom (with black point) is bonded to next nearest neighbour Si atoms and next nearest-neigl~bour Si atoms form a reconstructed bond (from Ref. [3]).

For instance, divacancies trap single vacancies that are released by the break up of vacancy complexes and create larger complexes. They may also diffuse, come across another divacancy and produce a four-vacancy complex. Both vacancies and divacancies are trapped by A-centres and 0 atoms to produce “P-centres” which consist of up to three vacancies bonded to up to three 0 atoms. Some of these complexes are stable up to 800 K. These annealing steps have been studied in great detail and compiexes exceeding a dozen vacancies in size have been well characterised 131. Eventually, these complexes merge to form micro-voids or extrinsic dislocation loops according to whether the aggregation takes place on several or on a single lattice plane, respectively. The latter are often detected by tmnsm~ssion electron microscopy (TEM). Such macroscopic features are stable and apparently represent the final stage of the annealing process. While they are large and represent gross disruption to the lattice, they are probably of little electronic activity especially compared to the large number of point defects that are involved in their creation and devices that contain these defects may not be particularly disrupted by them. However, the elevated tempera~res required to render lattice defects inactive in this way will surely destroy a fabricated device.

( 1) The application of electrical techniques such as Hall effect measurements and DLTS has allowed the introduction rates of electrically active defects to be determined with some accuracy [4--71. However neither of these techniques can identify, in an atomic sense, a defect. (2) The EL2 centre is important in SI LEC GaAs. It exhibits a broad infrared absorption spectrum which is characteristic of the defect [S]. This absorption increases substantially on fast particle irradiation and has been used as a measure of damage 191. (3 ) Infrared Absorption by Localised Vibrational Modes (LVM) is useful in providing atomic information about defects and defect complexes. LVM is not a very sensitive method and is limited to atoms of low mass, see [lo], but virtually all of what is known about the atomic identity of impurity interactions in GaAs has been provided by this technique [ 111. Incidentally, LVM absorption was an adjunct technique used for irradiation studies in Si and many of the defect structures derived from EPR measurements were supported by this independent method. Unlike Si, there already exists a high concentration ofnative point defects in bulk GaAs, the choice of material for the vertex detectors for the LHC project [ 121. It follows that a small increase in this concentration following a relatively small fluence of high-energy particles will not be detected by available techniques. This can give the impression that GaAs is radiation hard. As will be demonstrated below, the introduction rate of radiation defects is near that of Si and that GaAs is not expected, on this basis, to be a great improvement on Si.

4. Irradiation damage in GaAs

Because there are two sub-lattices, the interactions of high-energy particles with the GaAs lattice are more complex than in Si. As the masses of Ga and As atoms are similar and the same bonds must be broken to produce Frenkel defects on each sub-lattice, the two sub-lattices are expected to damage at a similar rate. It is thought that both interstitials are mobile below 300 K. However, neither vacancy is expected to be mobile until well above room temperature. 4.2. Secondary damage

3. Experimental considerations for GaAs The application of EPR is very limited in GaAs because of broadening of the resonance lines by the nuclear spins of the Ga and As atoms surrounding the defect. For this reason relatively little is known about the secondary damage process in this important semiconductor. Three approaches have met with some success, however:

It is possible to produce secondary defects by the interactions of interstitials from one sub-lattice with vacancies of the other, in other words to produce antisite defects like Asoa or Ga.hl. The first of these is known to occur in both unirradiated and irradiated bulk GaAs and is the technologically important EL2-centre. Although it has been postulated that the GaAs antisite results in the 78/203 meV double acceptor centre found in GaAs grown from a Ga-rich melt

M.R. BmzellNucl.

Table

Instr. and hletiz. in Phys. Rrs. A 395 11997) 88-93

91

I Introduction Rate (cm-l)

LIE (me\:)

Model

El E? E3

1.5 1.5 0.4

45 140 300

Vi, - or AsFa VA:‘” or ASQ VA, _ Asi

E3 E5

0.08 0.1

760

Y+IS- Asi

960

V,, - As;

[ 131, it has recently been suggested that the defect involved is the B*\ (boron antisite) defect [14]. Thus the occurrence of GaA, defects is unproven. As will be shown later, LVM studies prove that primary defects are trapped by impurity atoms.

5. Electrical data Table 1 shows the in~oduction rates, ionization energies and atomic models of the most ilnpo~ant deep levels introduced in GaAs following electron irradiation at energies exceeding 1 MeV and detected by DLTS [5,7]. Although the defects labelled El and E2 were originally interpreted as being As vacancies, a recent study in parallel with EPR measurements indicates that they may be associated with distorted Asoa antisites 1151. Note the large in~oduction rates of these secondary defects, comparable with those found in Si. Such defects probably result from mobile As interstitials being trapped by an (unknown) defect, possibly associated with a Ga vacancy. El and E2 defects can be annealed near 500 K.

6. EL2 data In a similar context to the above, we note that it has been known for some years that high concen~ations of “ELZ” defects are introduced after electron or neutron irradiation [9]. These defects appear to be different to EL2 defects that are ~WWZ-in to GaAs because they can be annealed out at temperatures between 500 and 600 K (unlike EL2 centres which are stable up to 1200K) and they do not exhibit “photoquenching”, an interesting property whereby EL2 centres can be excited into an inert metastable state by irradiating cooled samples ( < 140 K) with infrared light of photon energy near 1.1 eV. However, they exhibit the usual infrared absorption spectrum associated with EL2, are closely related to EL2, and are probably associated with the El and E2 electronic states. For this reason we label them as “EL2” in this review. Recently, K.riiger et al. [16] performed an irradiation measurement where “EL2” absorption was measured as a function of neutron fluence. Unlike other workers. in

fluence no j1016,m-2f Fig. 3. The introduction of neutral and charged “EL2” centres as a function of neutron fluence. Concentrations of neutral EL2, [ELZ’] are found by measuring the calibrated infrared absorption at a wavelength of 1pm. [EL?+1 is measured by MCDA. AIthough there may exist a systematic error in each of these concentrations. there is a clear, linear increase of each with Ruence.

addition to measuring [EL2’] by optical absorption, they recorded the concentration of ionised “EL?? [EL2+] defects as measured by Magnetic Circular Dichroism Abso~tion (MCDA). Their data are shown in Fig. 3 where it can be seen that [*‘EL2”“] and r‘EL2”+] were produced as irradiation progressed each with an introduction rate of greater than 1 cm-‘. The only reasonable interpretation is that acceptors were produced at the same time as “EL2“ defects. This result seems to explain why the hole signal is seriously degraded in GaAs detectors after irradiation by high-energy pions [ 171: the introduction of hole traps (acceptors) in material that originally contained a small concentration of acceptors would degrade the signal considerably. None of the above defect creation mechanisms involves interactions of primary defects with impurity atoms except possibly the creation of acceptors whose identity is unknown. In order to study impurity reactions we consider LVM data.

7. LVM absorption studies Trapping of As interstitials by both C and B atoms have been demonstrated by this technique. C is the dominant acceptor impurity in SI GaAs and occurs at concentrat~o~is up to lOI cmp3 in commercial material [ 141. C atoms occupy As lattice sites and give rise to a sharp LVM absorption band at an energy of 582 cm-’ [ 181. On electron or neutron irradiation, this line is reduced in

III. CHA~CTERISATION~MODELLING

f

Fig. 4. The structure of the C( 1t centre. The C& atom has trapped an interstitial arsenic atom (from Ref. [ 1931.

7

Fig. 5. The reduction in the magnitude of the LVM absorption lines due to C( I) centres and the concomitant increase in the magnitude of the absorption line due to recovery of CAs acceptors as a function of annealing temperature (from Ref. [IO]). The C( 1) centre has trigonal symmet~ and gives rise to two absorption bands, one at 557cmP’ (open circles) and the other at 606cmP’ (filled circles). CAM acceptors have tetrahedral symmetry and produce a single absorption band at 582 cm-’ (squares).

magnitude

demonstrating

that GS

acceptors

are being

con-

into another defect. New abso~tion features are formed at the same time due to [CA~--AS,,,J compIexes, or C( 1) centres [ 191, Fig. 4. In other words, in a similar fashion to the trapping of mobile interstitials by dopant atoms in Si, CAMacceptors trap As interstitials. C( 1) centres anneal between 500 and 600K when the magnitude of the LVM line due to isolated CAMacceptors returns to its original value, Fig. 5. Boron is an important impurity in GaAs but is rarely considered as an active centre. This is because it is isoelectronic

verted

Fig. 6, The structure of the B( 1) centre. The E& iso-electronic defect has trapped an interstitial arsenic atom (from Ref. [I I]).

and exhibits no electrical activity and also because it is particularly difficult to measure. Although concentrations of B can exceed IO” cme3 [ 141, manufacturers do not quote concentrations in their specifications. Like C atoms, B atoms which occupy Ga lattice sites as neutral Boa defects, can trap mobile As interstitials. Accordingly, on irradiation the LVM line due to Boa reduces while others interpreted as being ]BG~-As,,,~] complexes, or B( 1) centres, are produced, Fig. 6. The ~ll~oduction rates of C( 1) and B( 1) centres cannot be determined from their respective LVM lines as these have not been calibrated in terms of concentration. It is necessary to look at the removal rate of C and B atoms from their substitutional sites to achieve this, as the absorption cross sections of these lines have been calibrated, and assume that all this loss is due to the creation of C( 1) and Bf I ) centres, respectively. When this is done in material containing high concentrations of C, and especially B, the introduction rates can approach 1 cm-‘. demonstrating efficient trapping of mobile interstitials. There is little other reliable information regarding radiation induced defects, one of the reasons being that LVM absorption is limited to light atoms and their complexes. It is known that other impurities such as Si trap primary defects but as these are not present in SI GaAs, they will not be discussed here. It is interesting that all the defects mentioned above involve the trapping of As interstitials rather than Ga interstitials. It may be that Ga interstitials have a strong athnity for Ga vacancies and that the Ga sub-la~ice is self healing to a far greater extent that the As sub-lattice, as has been suggested recently [20,21]. On the other hand, defects produced by the trapping of Ga interstitials may simply be unobservable by present techniques. It is not known whether irradiation damage in GaAs under the influence of a strong electric field (as would be the situation in detectors under bias) is different to that produced in the absence of a field. All the above data were obtained without an electric field.

8. Conclusions Although relatively little is understood about irradiation damage in GaAs, considerable insight can be gained by looking at Si where much data is available. The major damage process in GaAs seems to be associated with the trapping of As interstitials by both impurity atoms and Ga vacancies. The in~oduction rates of these defects are comparable to those found in Si, demonstrating effective trapping. On this basis, there seems to be little reason for saying that GaAs is radiation hard. Of course, if secondary defects are electrically inactive. their effects on particle detectors would be minimized. However, there is a clear indication that damage removes C acceptor atoms, and introduces both Aso,-associated defects and acceptors whose concentrations are propo~ional to the fluence. All these wilt result in electrical degradation of the material. On a positive note, annealing of “ELZ”, B( I ) and C( I ) defects takes place at a relatively low temperature (500400K). Tt follows that annealing of the detectors may provide a solution to these classical irradiation damage problems.

References / I] G.D. Watkins, in: W. Schroeter (Ed.). Electronic

Structure and Properties of Semiconductors, vol. 4, Materials Science and Technology, VCH Verlagsgesellschaft mbH, Wertheim, 1991, pp. 105-141. 121 G.D. Watkins, Semicond. Sci. Technol. 6 (1991) Bl II. 131 C.D. Watkins, Mat. Sci. Forum 143-147 (1994) 9-20 (and the references therein). As an example of the range of defects identified in Si nearly 20 years ago. see J.W. Corbett, J.C. Bourgoin, L.J. Cheng, J.C. Corelli, Y.H. Lee, P.M. Mooney. C. Weigel. Inst. Phys. Conf. Ser. 13 (1977) I-11. 141 D.V. Lang, L.C. Kimerling, Phys. Rev. Lett. 33 (1974) 489: see also an excellent review by D.V. Lang, Inst. Phys. Conf. Ser. 31 (1977) 77-94.

[S] D. Pans, J.C. Bourgoin, J. Phys. C 18 (1985} 3839. [6] D. Stievenard, X. Boddaert, J.C. Bourgoin, Phys. Rev. B 34 ( 1986) 4048. J.C. Bourgoin, Stievenard. X. Boddaert, H.J. van Bardeleben. Phys. Rev. B ?I (1990) 5271. [8] G.M. Martin, Appl. Phys. Lett. 39 (I981 ) 747. [9] See, for example, K.V. Vaidyanathan, L.A.K. Watt, Radiation Effects in Semiconductors, Gordon and Breach, New York, 1971, pp. 293-300. [IO] R.C. Newman, infrared Studies of Cry&al Defects, Taylor & Francis, 1972. [I I] See, for example, M.R. Brozel. R.C. Newman, J. Phys. C I 1 (1978) 3135. [I21 See, for example, M.R. Brozel, Propertres of Gallium Arsenide. 3rd ed., INSPEC, 1996, p. 371. [ 131 K.R. Elliott, D.E. Holmes, R.T. Chen. C.G. Kirkpatrick, Appl. Phys. Lett. 40 ( 19821 898-901. [ 141 R.C. Newman, B.R. Davidson, R. Addinail, R. Murray, J.E. Meert. J. Wagner, W. Gotz, G. Roes, G. Pensl, Mat. Sci. Forum 143-147 (1994) 229-234. [I51 H.J. van Bardeleben, C. Delerue. D. Stievenard. Mat. Sci. Forum 1433147 (1994) 2233228. [I61 J. Kruger, Yan Chin Shih, Liu Xiao, C.L. Wang, J.D. Morse, M. Rogalla, K. Runge, E.R. Weber, Proc. SIMC9, Toulouse, May 1996, IEEE, to be published. [I71 C. Buttar, these Proceedings (4th lnt. Workshop on GaAs Detectors and Related Compounds. Aberfoyle. Scotland, 1996) Nucl. Instr. and Meth. A 395 ( 1997) I. [IS] A review of all LVM lines in GaAs inciuding those produced by fast particle irradiation has been given by R. Murray, Properties of Gallium Arsenide, 3rd ed., INSPEC. 1996, p. 227. [19] J.D. Collins, G.A. Gledhill. R. Murray, P.S. Nandhra, R.C. Newman, Phys. Stat. Sol. R I51 ( 1989) 469-477. [20] A.C. Irvine. D.W. Palmer, J.S. Roberts, Mat. Sci. Forum ~43-14~(1994)289~294. [Zi] K. Karsten, P. Ehrhart. Mat. Sci. Forum 143-147 (1994) 365-370.

171D.

III. CHAR4CT~RIS,4T~~N~M~DELLING