RELATION BETWEEN MICROSCOPIC DEFECTS AND MACROSCOPIC CHANGES IN SILICON DETECTOR PROPERTIES

FIRST INTERNATIONAL WORKSHOP ON DEFECT ENGINEERING OF ADVANCED SEMICONDUCTOR DEVICES

of the ENDEASD held in Santorini -Greece, 21-22 APRIL 1999

RELATION BETWEEN MICROSCOPIC DEFECTS AND

MACROSCOPIC CHANGES IN SILICON DETECTOR PROPERTIES

AFTER HADRON IRRADIATION

E.Fretwurst, M.Kuhnke, M.Moll, G.Lindström

University of Hamburg

II. Institut für ExperimentalphysikROSE (CERN R&D 48) - Collaboration

ABSTRACT

Silicon detectors produced from materials with different doping and oxygen concentra-tion have been irradiated with 27 MeV and 23 GeV protons, 192 MeV pions and 5.3MeV neutrons. In isothermal annealing studies at 60°C the microscopic defect evolution(measured with DLTS) and the changes in the detector leakage current were monitored inparallel. It is shown that the annealing of a broad DLTS peak located at about 0.42eVbelow the conduction band is correlated with a decrease of leakage current. This electrontrap is at least partly attributed to the singly charged divacancy. At the same time anelectron trap level at about EC-0.24eV which is commonly attributed to the doublycharged divacancy level is growing with increasing annealing time. The annealing proc-ess of both electron traps (EC-0.42 and EC-0.24 eV) and the annealing of the current re-lated damage parameter α have been found to be independent of the oxygen and dopingconcentration in the materials under investigation (2×1014 < [O] < 1018cm-3 and1012 < [P] < 4×1013cm-3).

Furthermore the introduction rate of the defect level at EC-0.42 eV and the current relateddamage parameter α are independent of the type of used particle, if the particle fluence isnormalized to the non ionizing energy loss (NIEL). However, the introduction rates of theobserved point defects VO and CiCs were found to depend on the particle type. Thus clearindication is given that the generation of point defects does not scale with the NIEL.Compared to the neutron irradiated samples the introduction rate of point defects afterirradiation with charged hadrons was found to be higher by a factor between 1.6 and 2.2.1. INTRODUCTION

The long term operation of silicon detectors in the extremely intense hadronic radiationfields of high energy physics experiments leads to a degradation of the detector propertiesdue to radiation damage in the silicon bulk. An increase of leakage current, a change ofeffective doping concentration and a decrease in charge collection efficiency are wellknown effects of hadron irradiation [1]. However, only little is known about the nature ofthe radiation induced defects in the silicon bulk that lead to the different changes in themacroscopic detector properties [2-4]. Without such a knowledge, attempts to produceradiation harder silicon by changing the impurity contents are based on a try and errorstrategy only. Thus, a more complete understanding of the relation between the micro-scopic defects and the macroscopic detector properties would very much support suchefforts.

It has been shown that the increase in leakage current as well as its annealing after irra-diation with neutrons is independent of the used silicon material [5] and that an electron

trap at about EC-0.42eV is correlated with the leakage current [3]. Furthermore it wasshown that in oxygen rich silicon detectors the radiation induced changes of the effectivedoping concentration is suppressed in charged hadron damage while after irradiation withneutrons it is not [6]. This gave us the impulse for first microscopic investigations onoxygen rich materials irradiated with charged and uncharged hadrons.

2. EXPERIMENTAL PROCEDURE

The device parameters are listed in Tab. 1. The given oxygen and carbon concentrationshave been measured by IR absorption and/or SIMS prior to the processing of diodes. Asindicated in some of the cases these values are below the detection limits of the usedmethods. The broad range of [O] and [C] in the epitaxial p-type material produced byITME reflects the inhomogeneous depth profile derived from SIMS-measurements.

The samples were exposed to different particle fields. Irradiation experiments have beenperformed with fast neutrons of the d+Be source (mean energy 5.3 MeV) at the PTBBraunschweig/Germany [7], reactor neutrons at the Ljubljana/Slovenia reactor facility[8], 192 MeV pions at the PSI/Villigen [9], 27 MeV protons at the Legnaro/Italy accel-erator [10] and 23GeV protons at the PS/CERN [11]. All fluences given in this paper arenormalized to a 1 MeV equivalent neutron beam using the appropriate hardness factors[12] if not explicitly mentioned otherwise. For the investigation of the annealing processan isothermal heat treatment at 60°C was chosen to accelerate the process with respect toroom temperature. Between the isothermal heating steps the I/V characteristics weremeasured at room temperature. Further details about the determination of the bulk gen-eration current can be found in [5]. A commercially available DLTFS-apparatus [13] wasused for the defect characterization which is described in more detail in [14]. The dis-played spectra correspond to the sinus correlator function (b1) and were obtained with atime window of 200ms. The quiescent reverse bias was 10V and during the 100 ms last-ing filling pulse 0V was applied to the sample.

AcronoymWM-3kWE-7-25kWI-4kWI-400TS-7kII-800II-120IH-130Cz-140ID-400ID-2kID-4kCrystaltypen-FZn-FZn- FZn- FZn- FZn- FZn- FZn-FZn-Czp-EPIp-EPIp-EPICrystalproducerWacker a)Wacker a)Wacker a)Wacker a)Topsil b)ITME c)ITME c)ITME c)Polovodice d)ITME c)ITME c)ITME c)Resistivity[kΩcm]2.710 - 204.00.426.60.780.110.130.140.41.63.9b)d) [O][10 cm-3]16[C][10 cm-3]16< 5< 5< 0.02< 10< 517< 10< 10904-203-204-60< 0.5< 0.5< 3< 2< 0.5< 2220.51-21-21-2 a) Wacker AG, Burghausen, Germany c) Institute of Electronic Materials Technology, Warsaw, Poland Topsil, Frederikssund, DenmarkPolovodice, Prague, Czech RepublicTable 1: Silicon material of investigated devices.- 2 -

3. EXPERIMENTAL RESULTS

3.1 ISOTHERMAL ANNEALING STUDIES

3.1.1 REVERSE CURRENT

The radiation induced increase of the reverse current ∆I is given by the difference be-tween the currents measured at total depletion of the device before and after irradiation. Ithas been shown for numerous samples and different radiation fields that the increase ofthe current per unit of depleted volume ∆I/V for a well defined annealing state after irra-diation is proportional to the fluence Φeq and can thus be described by (see e.g.[5, 15,16])

∆I/V = α×Φeq

(1)

where the proportionality factor α is called current related damage rate. This relationwas found to be valid for a wide fluence range from 1010 cm-2 to 1015 cm-2 and thereforeindependent on the fact that some samples are inverted in conduction type after exposureto high fluences [5].

The annealing behavior of the current related damage rate α after irradiation is displayedin Fig. 1 for various temperatures. The α value, respectively the leakage current, is con-tinuously decreasing and only for the highest temperature, after a 2 months lasting an-nealing at 106°C, a saturation is indicated at a value of about 6×10-18 A/cm.

For room temperature the annealing curves were usually fitted to a sum of exponentialsand a saturation value α∞ as indicated in Fig. 1 by the dashed-dotted line [15,16,17].However such parameterization does not represent sufficiently the long term annealing

101 hour1day1 month1 yearWunstorf (92) α∞=2.9x10-17A/cmα [ 10-17 A/cm ] 86420106oC49C60oC80oCo21oC(α∞)101102time [ min ]103104105106Fig 1: Current related damage rate α as function of cumulated annealing time at dif-ferent temperatures. Solid lines represent fits according to Eq. (2), dashed-dotted linerepresents a simulation according to [15].- 3 -

data at room temperature for cumulated annealing times t>1 year. This is most obviouslyseen in the figure from the annealing at higher temperatures where no saturation value α∞can be seen. Instead, in the long term annealing the α value seems to follow a logarithmicfunction in time. Thus the long term annealing at room temperature and the annealing atelevated temperatures was chosen to be described by one exponential and a logarithmicterm:

α(t) = α0×exp(-t/τI) + α1 − α2 × ln(t/t0).(2)

For each temperature the data were fitted according to Eq.(2) with t0 set to 1 min. Thecorresponding parameters are displayed in Tab. 2. For the exponential term the weightedaverage value for the amplitude α0 was found to be α0 = (1.23±0.06)×10-17 A/cm and anArrhenius plot of the time constant τI revealed [14]:

1/τI = k0I×exp(-EI/kBTa)

(3)

with k0I = 1.2×1013 s-1 and EI = (1.11±0.05) eV. Thus, with a first order kinetics and afrequency factor k0I close to the most abundant phonon frequency, it is very likely that a

dissociation of a defect is

Taα0τIα1α2responsible for the expo-[°C]10-17 A/cmmin10-17 A/cm10-18 A/cmnential part of the leakage

211.237.073.291.4×104current annealing.

491.282605.363.11The interpretation of the601.26944.873.16logarithmic part of Eq.(2) is801.1394.232.83far more complicated and it106--3.382.97is noted here that the pre-Table 2: Parameters of current annealing at different tem-sented parameterizationperatures Ta (Eq. 2). For the fit the parameter t0 was set to 1does not claim to be based

min.on a physical model. The

average value of the parameter α2 is given by α2 = (3.07±0.18)×10-18 A/cm, whereas theparameter α1 clearly displays a temperature dependence which can be described by [14]:

α1(Ta) = α10 + α11×1/Ta (4)

with α10 = -(8.9±1.3)×10-17 A/cm and α11 = (4.6±0.4)×10-14 AK/cm. It has to be noted thatthe given parameterization only holds for the time and temperature range indicated inFig.1.

3.1.2 DLTS STUDIES

In parallel to the measurements of the current annealing described above the isothermalannealing at 60°C of neutron induced defects has been studied by the DLTS method.Fig. 2 demonstrates the evolution of the DLTS spectrum within a period of 85 hours for adetector fabricated from oxygen rich Cz material (Cz-140, see Tab. 1). Typically fiveelectron traps are observed in the temperature range between 50K and 250K which arelabeled by the letter E and a number. In this special device the peak E1 can be attributedto a superposition of a thermal donor TD(0/+) and the carbon interstitial Ci(-/0) whose con-centration decreases very quickly at 60°C while the concentration of the TD defect doesnot change with proceeding annealing time. The DLTS signal E2 is assigned to be thewell known A-center VOi(-/0) and an increasing part which is associated with the

- 4 -

TD(0/+) + Ci(-/0) (E1)0.6 ? + VV(-/0) + ? (E4)DLTS signal (b1) [pF] 5 min at 60oC82000 min at 60oC0.4VO + CiCs (E2)E4b0.2VV(--/-)E4a (E3)050100150200250Temperature [K]Fig.2: Evolution of the DLTS spectrum at 60°C for a neutron irradiated sample produced fromCz silicon. The main part of the huge peak at 60K (E1) is due to thermal donors (TD) observedin the material already before irradiation.CiCs(-/0)(A) trap induced by the reaction Ci + Cs → CiCs . The dynamic of such processesin high resistivity silicon has been described elsewhere [18, 19].

In the following we will concentrate on the evolution of the peaks E3 and E4. Both peakscannot be attributed to isolated point defects. E4 is usually assigned to be the singlycharged divacancy VV(-/0) and E3 to the doubly charged divacancy VV(--/-) inside ornearby disordered regions or clusters [3, 20, 21]. Although both levels are due to thesame defect their observed peak heights are different. According to a model described in[20] this is due to lattice strain in the vicinity of clusters which reduces the signal of thedoubly charged divacancy (E3).

It is conspicuous that the peak height of the trap E4 is not only decreasing with cumulatedannealing time but the peak maximum shifts also in its position to lower temperatures.This indicates that a trap level hidden in the right hand side shoulder of the peak anneals

0.3E4b∆ (5 min - 82000 min)∆( DLTS signal (b1) [pF] )0.2E4a0.10-0.1VO+CiCs (E2)100VV(--/-) (E3)Temperature [K]150200250Fig. 3. Evolution of the \"difference spectra\" corresponding to the spectra shown in Fig.2.- 5 -

out. More information on this hidden trap level can be gained if one analyses the “differ-ence spectra” which have been extracted by subtracting each spectrum measured after acertain annealing time from the spectrum taken after the first heat treatment of 5 min.Fig. 3 shows the corresponding evolution of the “difference spectra”. Obviously, there isa trap level visible with a peak maximum at T = 212 K (E4b). The displayed increasewith cumulated annealing time corresponds to a decrease in defect concentration. Thetrap parameters are included in Tab. 3. A similar annealing behavior is observed for thebroad peak E4a. In contrast to E4b the peak shape of E4a cannot be described by a singledefect level. The difference signal of trap E3 has an opposite sign which means that thedefect concentration is increasing with progressing heat treatments. The “differencespectra” were used for determination of the defect parameters and trap concentrations ofE4a and E4b (see Tab. 3) since they allow a more accurate analysis than the spectrashown in Fig. 2 in which E4a and E4b are superimposed with further levels. In Fig. 4 thetrap concentrations normalized to the 1MeV neutron equivalent fluence Φeq as function ofcumulated annealing time are shown for all three electron traps. For comparison the an-nealing of the current related damage rate α is included. The observed time dependencefor the concentration of the E4b and E4a is quite similar to that found for the leakage cur-rent particularly for the short term component. A more direct correlation of the leakagecurrent with the trapping center E4b can be seen if one plots the differences ∆α = α(5min) - α(ta) versus the differences in the trap concentration ∆NT/Φeq = NT/Φeq(5 min) -NT/Φeq(ta) taken at the same annealing time ta (see Fig. 5). The proportionality betweenboth quantities hold for the short term annealing period up to 640 min and the resultingslope is found to be (4.18±0.08)×10-17 A. Such correlation between the electron trap E4band the leakage current has also been found in isochronal annealing experiments [3].Since the hole capture cross section of the level E4b is unknown it is not possible to cal-culate the generation current produced by the defect from the level parameters measuredby DLTS. However, using the standard Shockley-Read-Hall model and the relation be-tween the defect concentration and the generation current given above the hole capturecross section can be estimated to be about 1×10-13cm2. Thereby one has to keep in mindthat other current generation mechanisms like the inter-center charge transfer model sug-gested in [22] can enhance the current generation and therefore the calculated capturecross section may be to large.

1.552.52.01.51.00.50.01α0.53210010101annealing time [ min ]102103104105∆α [ 10-17 A/cm ]α [ 10-17A/cm ]NT/Φeq [ cm-1 ]E4bE4aE345.3 MeV neutronsTrap E4bEC-ET= 0.46 eV00.10.20.30.4∆NT/Φeq [ cm-1 ]0.50.6Fig.4.Annealing of defect concentration (E3,E4a, E4b) and α value (see text).Fig.5.Correlation between trap E4b and leak-age current (α value).- 6 -

DefectE3E4aE4bE4∆Et [eV]0.240.360.460.41

σt [cm2]≈10-14≈10-141×10-141.5×10-15Introduction rate [cm-1]

0.24→0.41 (a)0.26 (b)0.62 (b)1.04 (c)

Table 3: Defect parameters. The given introduction rates refer to certain annealing states at 60°C:(a) after 5 min and after 82000 min, (b) annealed in the period 5min to 10000 min and (c) afteranneal to 8200 min.

3.2 MATERIAL DEPENDENCE

2TD+/0Introduction rate [cm-1]FZ-120Ωcm, [O] < 5 1016 cm-3FZ-800Ωcm, [O] = 1.7 1017 cm-3CZ-100Ωcm, [O] = 9 1017 cm-3VOi-/0+CiCs(A)-/0 (E2)VV-/0+ ? (E4)1VV=/- (E3)050100Temperature [K]150200250Fig. 6. DLTS spectra (normalized to introduction rate at 200K) obtained after irradiation withneutrons and a subsequent 80min heat treatment at 60°C for different materials (see legend).

Similar studies have been performed on devices fabricated from different materials (seeTab. 1). As an example in Fig. 6 DLTS spectra for such different samples after neutronirradiation and annealing for 80 minutes at 60°C are shown. Here the peak amplitudes ofthe individual E4 signals are normalized to corresponding introduction rates NT(E4))/Φeq.A comparison of the introduction rates for all three trapping centers E2, E3 and E4 isdemonstrated in Fig. 7 and shows that the rates for the traps E3 and E4 are independenton the impurity or doping concentration of the material under investigation. On the otherhand the introduction rate of the point defects VOi and CiCs seen in peak E2 show a dis-tinct dependence on the impurity concentration of the material. For devices with highoxygen concentration the introduction rate is about a factor two smaller compared to ma-terial with a standard oxygen content. This effect is related to the dependence of the de-fect kinetics on the impurity concentration [14, 23]. A high oxygen concentration leads toa shift of the Ci sharing between the two reactions Ci + Cs → CiCs and Ci + Oi → CiOi infavor of the second one, i.e. the CiCs signal is reduced in oxygen rich material comparedto standard material.

For numerous detectors fabricated from the materials mentioned above and additionalp-type material (see Tab. 1) the isothermal annealing of the leakage current at 60°C hasbeen measured.

- 7 -

2.55FZ, 120 Ωcm, [O] < 5 1016cm-3FZ, 800 Ωcm, [O] = 1.7 1017cm-3Cz, 140 Ωcm, [O] = 9 1017cm-3VV-/0 + ?VOi + CiCsp-type n-typeintroduction rate [ cm-1 ]2.01.51.00.50.0damage rate α [ 10-17 A/cm ]43[O] = 4-20 1016 cm-3[O] = 3-20 1016 cm-3[O] = 4-60 1016 cm-32VV=/-1[O] < 5 1016 cm-3[O] < 5 1016 cm-3[O] < 2 1014 cm-3[O] < 5 1016 cm-3[O] = 2 1017 cm-3[O] < 5 1016 cm-3[O] = 9 1017 cm-3[O] < 5 1016 cm-3[O] < 5 1016 cm-3E2 E3 E4trap number-4.10130-2.10130Neff [ cm-3 ]2.10134.1013Fig.7. Material dependent introduction rates Fig. 8. Damage rate α as function of effectiveof defects E2, E3 and E4. doping concentration.

The extracted α values after annealing for 80 minutes are plotted in Fig. 8 as function ofthe effective doping concentration Neff . In the legend of Fig. 8 the oxygen concentrationsof the different materials are given. It is very striking that the current related damage rateα does not depend on the material properties of the silicon, although the oxygen concen-tration is varying by about 4 orders of magnitude (1014 cm-3 < [Oi] < 1018 cm-3 ) and theresistivity is ranging from about 400 Ωcm to 4 kΩcm in p-type and from about 100 Ωcmto 25 kΩcm in n-type materials. Since such strong variations in impurity content have aninfluence on the defect kinetics of migrating vacancies and interstitials (see above) it isconcluded here, that after irradiation with neutrons mainly intrinsic defects composed ofvacancies and interstitials inside or close to the clusters (DLTS peak E4) are responsiblefor the increase of the generation current but not impurity related defects.3.3 DEPENDENCE ON PARTICLE TYPE

In this section we present new DLTS measurements performed on oxygen enriched FZmaterial which has been irradiated with 23 GeV protons, 27 MeV protons, 192 MeV pi-ons and 5.3 MeV neutrons. The results will be compared with observed changes in theleakage current induced by the same particles. The aim of these studies was twofold.First, it should be clarified whether the correlation between the current damage rate α andthe DLTS signal E4 could also be established for charged hadrons and whether the NIEL-hypothesis (non ionizing energy loss) for the scaling of the damage parameter α and theintroduction rate of the trap E4 is valid for all hadrons under study. Secondly, it was re-cently found by the ROSE collaboration [1, 6] that the radiation induced change of theeffective doping concentration Neff after exposure to charged hadrons is strongly reducedfor oxygen rich material compared to material with standard oxygen concentration. How-ever, after neutron irradiation no difference was observed between oxygen rich and stan-dard material.

The extracted current damage rates α as function of fluence for the different particles aresummarized in Fig. 9. The α values correspond to the annealing state of 80 min at 60°C.On the right hand side ordinate of Fig. 9 the so called hardness factors κ of the differentparticle fields are given which coincide with values calculated from published NIEL-

- 8 -

10α80/60 [10-17A/cm]27 MeV protons2.228<5.3 MeV> neutrons 421.14192 MeV pions0.5123 GeV protons101110121013Φ [cm-2]1014Fig. 9. α values for samples of same type irradiated with 27 MeV protons, 5.3 MeV neutrons192 MeV π+ and 23 GeV protons measured after a 80min lasting heat treatment at 60°C. Thescale on the right hand side indicates the corresponding experimental hardness factors κ.functions [12]. All plotted values are derived from individual devices irradiated to thegiven fluences. These results demonstrate that the radiation induced increase of the bulkgeneration current scales with the displacement damage cross section (NIEL) for chargedand neutral hadrons.

DLTS measurements have been performed on detectors fabricated from oxygen enrichedsilicon (II-800, see Tab. 1) after irradiation with different particles as mentioned above.Also in this study all samples have been annealed to 80 min at 60°C. The spectra shownin Fig. 10 are normalized in such a way that the peak heights of the E4 signals correspond

1FZ-800Ωcm, [O] = 1.7 1017cm-3DLTS-signal [ arb. units ]VOi-/0 +CiCs(A)-/0 (E2)n 5.3 MeV p+ 23 GeVπ+ 192MeVp+ 27 MeVVV-/0+ ? (E4)0.5VV =/- (E3)050100Temperature [K]150200250Fig. 10. DLTS spectra obtained on samples from the same wafer obtained after irradiationwith different particles (see legend) and a 80min lasting heat treatment at 60°C. The spectraare normalized to the 1 MeV neutron equivalent introduction rate at 200 K.- 9 -

κ61.47to the defect introduction rate NT/Φeq using the hardness factors presented above. As canbe seen the peak height of the E4 center is nearly independent of the damaging particletype. In Fig. 11 the introduction rates of E4 are compared with their corresponding α val-ues. A nice correlation between both quantities can be stated which substantiates the as-sumption that the E4 center can be attributed to intrinsic defects in clusters and that theintroduction rate scales with the non ionizing energy loss.

The significantly larger E3 signal (see Fig.10) for charged hadrons indicates that the lat-tice strain is less pronounced in charged particle damage. A more pronounced differenceis observed for point defects as demonstrated by the different peak height of the E2 signalwhich is associated with the VOi and CiCs defects. The introduction rates of all 3 trappingcenters are summarized in Fig. 12. The ratio of the E2 introduction rates for 23 GeVprotons and 192 MeV pions with respect to neutrons is 1.6 and for 27 MeV protons 2.2.Such an enhanced introduction of point defects by charged particles compared to neu-trons lead to the suggestion that this effect is caused by the coulomb interaction ofcharged particles with silicon atoms in the lattice. Coulomb interaction will create pri-mary knock-on atoms (PKAs) with much smaller recoil energies as the interaction withMeV neutrons leading finally to an enhanced generation of isolated point defects.

Since it is still unknown which defects are responsible for the change of Neff it is difficultto understand why the change in Neff is suppressed in oxygen rich material after irradia-tion with charged hadrons. So far we can only speculate. Since the change in Neff ismainly due to the generation of negatively charged defects in the space charge region onemay think e.g. of an enhanced generation of oxygen related positively charged donorsafter charged hadron irradiation partly compensating the negative space charge.

4.0E4 introduction rate [ cm-1 ]E4 introduction ratecurrent damage rate α2.5α80/60 [ 10-17A/cm ]3.0introduction rate [ cm-1 ]8.06.02.01.51.0VOi + CiCs5.3 MeV neutrons192 MeV pions23 GeV protons27 MeV protonsVV-/0 + ?2.04.01.02.00.0VV=/-0.50.00.0 neutrons pions protons protons 5.3 MeV 192 MeV 23 GeV 27 MeVE2 E3 E4trap numberFig. 11. Correlation between defect level E4 andFig.12.: Particle dependent introduction

rate of defects E2, E3 and E4.the α value.

4. CONCLUSION

Systematic isothermal annealing studies on the radiation induced increase of the leakage

current and defect formation measured by the DLTS technique have been presented fordetectors fabricated from silicon materials with different impurity and doping concentra-tions after exposures to neutral and charged hadrons. It has been proven that the currentrelated damage rate α and the introduction rate of the electron trap E4 which is usuallyattributed to the VV(-/0) defect does not depend on the material properties. We have alsodemonstrated that both quantities scale properly with the non ionizing energy loss andthat a nearly perfect correlation between the short term annealing component of α and theannealing of an electron trap (E4b) at EC – Et = 0.46 eV exist. From all these results we

- 10 -

conclude that the observed defects E4, E4a and E4b induced by neutral or chargedhadrons are intrinsic defects composed of vacancies and possibly interstitials inside orclose to clusters and that they are essentially responsible for the increase of the generationcurrent.

The recently observed effect that in charged hadron fields oxygen rich silicon is moreradiation hard with respect to the change of the effective doping concentration than stan-dard material and the fact that this effect is not present in neutron damage initiated firstDLTS studies on this subject. There is clear evidence that the introduction of point defectby charged particles is much larger compared to the introduction by neutrons. However,there is so far no clear understanding of the underlying defect processes which can ex-plain the observed reduction of the change of Neff in oxygen rich silicon induced bycharged hadrons. In order to clarify this problem further studies on the formation andkinetic of defects are needed.ACKNOWLEDGEMENTS

We greatfully appreciate the manifold assistance of R.Böttger, H.J.Brede and H.Klein forproviding the irradiation facility at the Physikalisch-Technische BundesanstaltBraunschweig, V.Cindro and M.Mikuz for the exposures at the Ljubljana reactor,G.Casse, B.Dezillie, M.Glaser, F.Lemeilleur and A.Ruzin for providing the irradiationfacility at CERN and the pion irradiation at the Paul Scherrer Institut Villigen andD.Bisello and J.Wyss for the exposures at the Legnaro proton irradiation facility. Finan-cial support is acknowledged from the CERN LHCC to the RD48 collaboration and fromthe BMBF to our group under contract 05 7HH17I.REFERENCES

ROSE collaboration, 2nd RD48 STATUS REPORT, CERN/LHCC 98-39, October 1998.H.Feick et al., IEEE Trans. Nucl. Sci. NS-44 (1997) 825.S.J.Watts et al., IEEE NS-43 (1996) 2587.M.Moll et al., NIMA 409 (1998) 194.

M.Moll et al., \"Leakage current of irradiated silicon detectors - material dependence\publication in Nucl. Instr. and Meth A.[6] A.Ruzin et al., this proceedings.

[7]H.J.Brede et al., Nucl.Instr. and Meth. A274 (1989) 332.

[8]Jozef Stefan Institute, University of Ljubljana , Sl-1000 Ljubljana, Slovenia[9] Paul Scherrer Institut, CH-5232 Villigen PSI, Switzerland,

[10]Laboratori Nazionali di Legnaro - Via Romea 4 - 35020 - Legnaro (Padova), Italy[11]European Laboratory for Particle Physics (CERN), CH-1211 Geneva 23[12]A.Vasilescu, Technical Note ROSE TN97/3, 1997.[13]Dr.L.Cohausz, Halbleitermeßtechnik Gmbh, Moosburg.

[14]M.Moll, PhD thesis, University of Hamburg, to be published in 1999.

[15]R.Wunstorf, PhD thesis, University of Hamburg, see also DESY FH1K-92-01 (1992).[16]K.Gill et al., Nucl. Instr. and Meth. A322 (1992) 177.

[17]A.Chilingarov et al., Nucl. Instr. and Meth A360 (1995) 432.[18]B.Schmidt et al., J.Appl.Phys. 76(7) (1994) 4072.

[19]H.Feick and M.Moll, Solid State Phenomena 57-58 (1997) 233, Scitec Publications, Switzerland.[20] B.G.Svensson et al., Phys. Rev. B 43 (1991) 2292.

[21]E.Fretwurst et al., Nucl. Instr. and Meth. A377 (1996) 258.[22] K.Gill et al., J.Appl.Phys. 82 (1997) 126.

[23] B.C.MacEvoy, Ph.D. thesis, Imperial College London, RAL-TH-97-003 (1997).[1] [2] [3] [4][5]

- 11 -