szaczsag

Gcncva 1994 OCR Output

Comactpcrson: D. Forkcl—WirthSpokesperson; D. Forkcl-Wirth

Soarcs’, M. Toulcmondc", P. Wcllmarm", A. Winnackcf', W. WitthuhnJ. Mcicr“, V.I. Mishin’, W. Prostj, E. Rccknagcl‘, U. Rcislochncr°, M. Rucb“, J .C.

Winh, H. Haas, U. Kaufmann, T. Licht, M. Lindroos", R. Magcrlc', J.G. Marques"z s4

N. Achtzigcr", A. Burchard', J. Corrciaw, M. Dcichcr', V. N. Fcdoscycv’, D. Forkcl

Troitzk— CERN Collaborationg 10

Konstanz— Bcrlin—Duisburg’ - Erlangc-:n“— Frciburg- Jcna- Lisboa— Cacn] z s ° 7 g

HYDROGEN IN SEMICONDUCTORS

NUCLEAR, ELECTRICAL AND OPTICAL STUDIES OF

PROPOSAL TO THE ISOLDE COMMITTEE

SC00000195

12 SEPTEMBER 1994r | HHIIHINIIWIIIHINHINHIIIWHWNIW WISC/P65

CERN LIBRARIES, GENEVA CERN/ISC 94-15gl-} ‘ I 5 .3 1;CELQRI —·· tix;.

OT\ EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH(/\`<’.,

gp I 0

Q/ERN

spectroscopy) on the same or identically prepared samples is essential. OCR Output

with electrical (Hall-effect) and optical techniques (photoluminescence, infraredTo get a link to the macroscopic properties of these systems, the combination of PACpotential), since H is trapped and observed at the Cd acceptor.

dynamics of a trapped hydrogen (diffusion around the acceptor due to its Coulombatom. On the other hand, the PAC probe atom Cd/Cd allows to study the""“m

be observed, since no binding mechanism exists anymore between H and the PAC probe

constituents of the semiconductor. As function of temperature, the free H diffusion can

takes place at a daughter isotope (Ge, ln) chemically identical to one of the73ln

act as dopant atoms (As, Cd). being able to trap H whereas the actual measurement73H7

As/Ge in Ge and Cd/ln in lll-V compounds like lnAs or InP. Here, the parents7373 H7I w

(ii) The H diffusion mechanism shall be directly studied by using the PAC probe atoms

of the band gap).

the hydrogen correlated complexes as function of doping and stoichiometry (i.e. the size

technological important systems are microscopic structure, formation and stability of

"representative" of group IIB metal acceptors. The problems addressed in these

(AlGaAs, GalnP, AlGaN) shall be investigated, using the PAC probe atom "‘Cd as am

(I) The H passivation mechanism of acceptors in GaN and temary III-V compounds

The aim of the proposed experiments is twofold:

an atomistic scale using radioactive isotopes provided by ISOLDE.

correlation technique (PAC) has contributed to the understanding of this phenomena on

on the electrical properties of the semiconductors ("passivation"). The perturbed angular

formed complexes, their structural arrangements, and the implications of this interaction

Ge and lIl—V semiconductors has improved considerably conceming the stability of the

During the last years, the understanding of H and its interaction with dopant atoms in Si,

Summary

Contactperson: D.Forkel-WirthSpokesperson: D. Forkel-Wirth

J .C. Soares', M. Toulemonde°, P. Wellmann'°, A. Winnacker°, W. WitthuhnJ. Meier°, V.I. Mishin’, W. Prostj, E. Recknagel', U. Reisloehner, M. Rueb,

°`°

Wirth, H. Haas, U. Kaufmann, T. Licht, M. Lindroos", R. Magerlel, J.G. Marques',j44z

N. Achtziger“, A. Burchard', J. Correia", M. Deicher', V.N. Fedoseyev", D. Forkel

CERN‘° CollaborationKonstanz- Berlin—Duisburg’ - Erlangen" - Freiburg— Jena- Lisboa- Caen- Troitzk

l z s °7 g °

SEMICONDUCTORS

NUCLEAR, ELECTRICAL, AND OPTICAL STUDIES OF HYDROGEN IN

ISC/P65/S

- 2

hydrogen passivation of acceptors in GaN and temary Ill-V semiconductors, mainly OCR Output

In our proposed experiments we intend to perfomi the first PAC experiments on

could be provided on hydrogen passivation of Cd acceptors in Ge [15].

"representative" of group lIB metal acceptors l6—l4]. Later on, also new information

passivation in lll~V compounds could be perfomied, using "‘Cd as a typicalm

the move of the ISOLDE facility, one of the most comprehensive studies on acceptor

compounds [6]. During the last few years and despite a major break of three years due to

(PAC) experiments have been performed at ISOLDE to study H in binary III·V

Triggered by the importance of the topic, in the late 80s the first angular correlation

for Si and for binary and temary lll-V compounds even less information is available [5].

semiconductors: H", H`, neutral H and H2 - not all relevant parameters are yet known

Due to the complex behavior of H - there exist at least four different states in

different processing procedures.

stability of impurity—H complexes, H diffusion) is essential for the optimizing of the

the understanding of the behavior of H in these systems (its interaction with dopants,

optoelectronic devices, e.g. by passivating unwanted deep level impurities [4]. Therefore

and temary lll-V compounds is nowadays used to produce new microwave and

On the other hand, intentional hydrogenation of donors and acceptors in binary

growth is suspected to cause the electrical inactivity of acceptors [3].

desired electronic properties, e.g. in GaN the incorporation of H during the crystal

to the used metalorganic constituents. This fact can lead to unpredictable changes of the

techniques like MOCVD always involves H presence during the growth of the layer due

growth of multilayers involving different stoichiometries (hetero structures). But using

deposition (MOCVD) technique which allows the control of stoichiometry but also the

devices. These materials are usually produced by the metalorganic chemical vapor

the stoichiometry enables the realization of new or high-performance semiconductor

compounds like AlXGa]_xAs or GaxIn]_xAs whose tunability of band-gaps by varying

"new" and most promising candidate for the blue light emitter or temary III-V

particular for semiconducting materials of current technological interest like GaN, the

etching H can be unintentionally introduced into the material. This problem exists in

[1,2]. During different steps of device processing, like crystal growth or chemical

and deep level dopants and impurities or causes new hydrogen related electronic levels

In semiconductors, hydrogen either saturates dangling bonds, passivates shallow

OCR Outputl. Introduction

- 3

more complex configuration of Cd—H2* It is less stable than the Cd-H pairs and OCR Output

An orientation of the corresponding efg in <l l l> direction can be excluded, indicating a

trapping of one hydrogen atom and either additional defects or more hydrogen atoms.

this complex Cd—H2* can be only fomied if H and defects coexist, indicating the

and InAs a second, hydrogen correlated Cd-configuration was found [7,8,14]. In InAs

[16]. The dissociation energy of these complexes depends on the Fermi level. In GaAs

bond center site, fomiing H~P and H-As bonds instead of an acceptor-hydrogen bond

stability of the Cd-H pairs support a model, proposed by Pajot which favors H on the

pairs, oriented in <lll> lattice direction. The crystallographic orientation and the

ll,14]. The corresponding PAC parameters reveal the fomiation of next nearest Cd-H

GaAs, InAs, lnP and GaP after plasma charging and low energy H+ implantation [8

lt was found by PAC, that similar hydrogen correlated complexes are formed in

was performed (s. table I).

acceptors the most comprehensive study of acceptor passivation in lll-V compounds

hydrogen complexes I6-I5]. Using '“Cd as a representative of group IIB metalm

new information conceming formation, stability and microscopic structure of acceptor

acceptors in Ge and lll-V compounds at ISOLDE provided a considerable amount of

During the last few years, the PAC experiments on hydrogen passivation of

2. Present Results

dissociation of H can be compared to the localized motion of H in Cd-H complexes.

mechanism exists any more between H and the PAC probe atoms. In Ge, the free

function of temperature, the free H diffusion can be observed, since no binding

the daughter isotopes represent self atoms (Ge in Ge, 1In in InAs or InP). As1773

or an acceptor (Cd in InAs, InP), being able to trap H. After the radioactive decay,1*7

compounds (InAs, InP), respectively. The parent isotope represents a donor (As in Ge)73

As/Ge and Cd/ln, which would enable such experiments in Ge and III-V7373mm

mechanism 0f H on a microscopic scale. ISOLDE provides two PAC probe atoms,

In some specific cases, PAC offers the possibility to study directly the diffusion

passivation mechanism in function of doping and stoichiometry.

corresponding specialists of our collaboration, thus opening the access to study the

complexes. The crystals will be specially designed for our experiments and grown by the

conceming formation, stability and microscopic structure of the hydrogen correlated

- 4

hydrogen [17]. OCR Output

probe atom As was successfully used to study the passivation of donors in Si by73

of a localized motion of H inside the Cd-H complex [15]. ln first experiments, the PAC

formation of Cd-H pairs and Cd-H2 complexes could be proven, as well as the existence

The mechanism of double acceptor passivation was studied for Cd in Ge and the

*) complex with probably 2 bond hydrogen or an additional lattice defect

13357(2)Cd-H

InSb

not<l1l> l. 141(1)Cd-}-12* 58l(5)

<lll>427(l) 13.3(1)Cd—Hl

InAs

<lll> 13,14.5(l)525(3)Cd—H

Ga?

10<lll> .8(l)484(l)Cd-H

InP

<lll> .35(l) 10,1248l(2)Cd-H2

<lll> .35(1) 10,12457(2)Cd-Hl

GaAs

(MHz) (QV)

Ref.T] <ijk>Complex V

of the complex.the orientation of the symmetry axis of the EFG tensor. ED gives the dissociation energycoupling constant measured at room temperature, 11 describes the symmetry and djk>

Table I. Cd-hydrogen complexes fomaed in III-V semiconductors. vo denotes the quadrupole

defect or H.

converts most likely to Cd-H pairs by detzrapping of the second partner, whether a

- 5

stoichiometry because the position of its levels relative to the band edges may be OCR Output

the behavior of the different charge states of H in this systems can be influenced by

— with increasing Al (AlXGal_xAs) or Ga (GaXlnl_xAs) content, the band gap increases.

a statistical distribution of Al and Ga second nearest neighbors.

Therefore each dopant atom like Cd residing on a group lll lattice site is surrounded by

substituting e.g. Ga by Al or ln by Ga does not result in an ordered structural phase.

semiconductor will change several features of the material:

compounds as function of stoichiometry. Modifying the stoichiometry of a temary III—V

propose PAC experiments using '“Cd to study the Cd-H interaction in ternary IIl—Vm

prerequisite to utilize H in hetero structures based on these systems. Therefore we

necessary due to the used layer growth techniques (MOCVD, MOVPE) and is

application in optical devices. Understanding the behavior of H in these compounds is

between that of AlAs (GaAs) and GaAs (lnAs) can be achieved, enabling e.g. a broad

important property, that by varying the stoichiometry each value of the band gap

Temary lll-V compounds like AxB]_XAs (AlXGal_xAs, GaxIn1_xAs) reveal the

due to Coulomb interaction.

identified donors [3,18] could become possible, if they are trapped at the Cd acceptor

hydrogen complexes. Besides the H studies, also the identification of the residual, not yet

perform first experiments with "‘Cd in GaN to study the fomiation of acceptorm

inactivity of the acceptors results from hydrogen passivation [3]. Therefore we plan to

type dopants apparently remains inactive. At the moment it is suspected that the

donors [3,18] have to be compensated first, but also because a large fraction of the p

was found, that p-type doping of GaN represents a problem, not only because residual

MOCVD technique, involving the presence of H due to the metalorganic constituents. It

moment a multi-national effort sponsored by the EC. The materials are grown by the

temperature electronics application [3]. To develop the basics for such devices is at the

e.g. GaN appears to become a candidate for short wave length lasers and high

very promising for the development of optoelectronic devices in the blue/purple range,

the last years the technological interest has considerably increased. These compounds are

A system almost unexplored are the III-V nitrides like GaN and AlGaN, although during

A. Hydrogen passivation mechanism of acceptors:

3 Physics Motivation and Proposed Experiments

- 6

observing the break-up of the complex microscopically. This resulted in the OCR Output

known. The PAC measurements using "‘Cd also revealed the stability of these pairs by"‘

From our previous work. the fomtation of Cd-H pairs in lll-V semiconductors is well

unique way to study H diffusion on an atomistic scale by PAC in a few selected systems.

The availability of the isotope Cd decaying to In at ISOLDE opens an"7 H7

obtained by these techniques.

band gap, a detailed microscopic picture of the diffusion mechanism can hardly be

and the dependence of their relative concentrations on the position of the H levels in the

measurements |2()|. Due to the different possible charge states of H in these materials

Schottky diodes. the detemiination of deuterium profiles by SIMS or effusion

concentration profile as function of annealing temperature in reverse bias annealed

semiconductors is mostly based on macroscopic techniques like monitoring the carrier

Up to now, the knowledge on the diffusion behavior of H in compound

B. Hydrogen dtyjitsion mechanism:

compounds like GalnAsP.

provide the necessary information to go later on one step further up to quatemary

XP, where the Cd—H formation is already known for GaP and InP. Furthermore it will

to the changing band gap. In parallel, similar experiments shall be performed in GaXInl_

energy or by the change in the relative population of the different charge states of H due

AlGaAs, if it is influenced by a stoichiometric dependence of the dopant—H binding

number of the formed Cd-H pairs. Therefore it will give hints about the H diffusion in

few % Al to GaAs [19]. The PAC measurements will reveal the stability and the total

measurements that deuterium diffusion profiles are strongly influenced by adding only a

the formation probability of Cd·H pairs and their stability. It is known from SIMS

except the Al concentration unchanged will reveal the influence of the stoichiometry on

samples with increasing Al content ending up with pure AlAs. Keeping all parameters

Starting from the known behavior of GaAs we will perform PAC experiments with

Ga (GaxIn1_xAs) content may change the configuration and stability of Cd—H pairs.

- changing the local environment of the acceptor Cd with increasing Al (AlxGal_xAs) or

position of the Fermi level which can be controlled by doping.

changed. In addition, the population of the different H states can be influenced by the

- 7

electronic effects and dynamics caused by the motion of the H or D atom. In Si, such OCR Output

at the Cd-H complex. Using either hydrogen or deuterium should discriminate between

caused by a hopping of the H around the Cd-acceptor or by charge fluctuations localized

experiments we want to clarify the nature of this local dynamics which can either be

fluctuation of the electric field gradient characterizing the Cd-H complexes. In further

PAC signal shows a dynamically relaxation on a nanosecond time scale caused by a

has been observed. At temperatures below the breal<—up temperature of the complex the

function of the measuring temperature a local dynamic behavior of the Cd-H complexes

formation of more then one Cd-H complex has been demonstrated. Moreover, as

of stability and structure of the fomied complexes. Due to the double charged Cd the

acceptor "‘Cd in Ge has been studied. The goal of this experiments was the questionm

In previous experiments, the formation of H complexes with the double

not determined by the nuclear state used for PAC but by the longer lived preceding state.

to Cd/ln, the time window for the observation of the presence or absence of H isH7ln

able to trap H. After the radioactive decay the diffusion of H‘ will be studied. In contrast

as discussed above can be performed in Ge using e-Y PAC. As acts as a donor in Ge,73

Using the isotope As decaying to Ge, the same type of diffusion experiments7373

determined.

in the band gap can be changed and this influence on the diffusion behavior can be

of H diffusion is possible. In a next step, by using temary compounds the hydrogen levels

determined by the measuring temperature. In this way, a direct microscopic observation

nuclear state. The mean time the H atom needs to perform a first diffusion jump is

probe atom within the time-window of about lO() ns defined by the life—time of the used

can now monitor the presence or absence of the H atom in the neighborhood of the

resides near to ln without any attractive binding to this atom. The PAC measurementsm

populated by the decay of 117Cd. That means, at the moment after this decay the H atom

In contrast to ‘“Cd, the nuclear state used for PAC is an excited state of lnulm

experiments using '"Cd and can be confirmed by Hall—effect measurements.m

passivated with H. The efficiency of this passivation is well-known from PAC

involving In with Cd. After electrical activation by annealing, these acceptors can beU-'

In the proposed experiment, we want to dope InP, InAs and temary compounds

migration enthalpy.

determination of an activation enthalpy for these pairs which is the sum of binding and

- 3

experiments involving Cd/ln. OCR Output"7' '7

demonstrated for the decay of ln to Cd in GaAs [22] and should be extended to themm

change of the intensity of photoluminescence peaks becomes possible. This has been

properties of the sample. ln this way, the identification of defects by the characteristic

characteristic half-life of the decay is reflected in the change of electrical or optical

hand, if an elemental transmutation is involved in the decay of radioactive dopants, the

annealing conditions, this will be done at the Konstanz heavy ion implanter. On the other

photoluminescence measurements. Using stable isotopes and identical implantation and

requires the characterization of identically prepared samples by Hall-effect and

linked to the optical and electrical properties of the semiconductor under study. This

The microscopic insights delivered by nuclear techniques like PAC have to be

representative samples.

H experiments with PAC, the annealing has to be also checked by PAC for a few

checked by Hall—effect and photoluminescence measurements. But before performing the

and dose at the Konstanz implanter. The annealing and electrical activation will be

dopants. This process will be optimized by implanting stable isotopes with same energy

annealing of the implantation induced damage and the electrical activation of the

Prerequisite for all further experiments in GaN and temary compounds is the

Erlangen and the Solid-State Electronics Department of the University of Duisburg.

Physics (IAF) in Freiburg, the Institute of Materials Science VI of the University of

compounds). This will be done in collaboration with the Frauenhofer-Institut of Applied

ions) and the planned experiments (availability of GaN, stoichiometry of the temary IH-V

grown to meet the requirements for ISOLDE implantations (penetration depth of the

Besides the use of commercial available crystals, epitaxial layers will be specially

4. Experimental Procedure

observed [21], using the PAC probe atom ln/Cd.mH 1

dynamics of complexes formed with different donors like H, As, P and Cu have been also

- 9

Experiments conceming the local H diffusion in Ge will require 5 shifts. Assuming, the OCR Output

used for studies of the H passivation mechanism in GaN and ternary lll-V compounds.

meets our experimental requirements. The major part of the beam time (35 shifts) will be

already a proton beam intensity of 5·l0" protons/pulse and 4 - 5 pulses per supercycle

worked very reliable for several beam-times at the PSB-ISOLDE and it turned out, that

This isotope is produced by a molten Sn target / plasma ion source. This target has

"'mctt (40 shirts)

other users:

ln total we request 54 shifts for the next two years which all can be shared with

6. Beam time request

implantation chambers.

About l shift of stable beam is required to optimize the transmission into the

Before each beam-time, about one day is necessary to install the equipment.

state laboratory at ISOLDE.

necessary sample treatment before and after implantation will be performed in the solid

spectrometers (about 40 mf) and the low energy implanter INES (about 20 mz). The

During beam—time, we need space in the experimental hall to install the PAC

surface preparation by comparing the results for different implantation energies.

experiments. The platform will be also needed to determine a possible influence of the

least 200 keV is essential. Therefore the high voltage platform is essential for our

structures and samples prepared for photoluminescence) an implantation energy of at

For special experiments (AlAs with protective surface layer, some hetero

temperatures without breaking the vacuum.

the samples. The chamber allows the implantation of up to 10 samples at variable

are equipped with a beam—sweep system necessary to achieve a homogenous doping of

chamber connected either to the GLM or to the GHM bearn-line, since both beam-lines

The collections will be mostly performed using the solid state implantation

5. Experimental requirements

- 10

Appl. Phys. Lett. 55 (1989) 2301 OCR Output[6] A. Bauiichter. M. Dcicher, S. Deubler, D. Forkel, H. Plank, H. Wolf, W. Witthuhn

Vol. 148 -149, Trans Tech Publications, Aedennannsdorf, 1994).[5] S.] . Peanon (ed.): Hydrogen in Compound Semiconductors, (Materials Science Forum

[4] S.A. Stockman and G.E. Stilman, in ref. [5]. D. 501

J. Appl. Phys. 76 (1994) 1363[3] H. Morkoc. S. Stritc, G.B. Gao. M.E. Lin, B. Sverdlov, M. Bums

Semiconductors and Scmimetals vol. 34, (Academic Press, Boston, 1991)[2] Pankove, J .1. and Johnson, N.M. (cds.): Hydrogen in Semiconductors,

Springer Series in Mat. Science vol. lo. (Springer, Heidelberg, 1992)ll] S.J. Pearton, l.W. Corbett. and M. Stavola: Hydrogen in Crystalline Semiconductors,

References

I-l diffusion in Ge.

over 2 beam-times within 2 years. This isotope will be used for the investigation of free

availability (up to now) of only one e·Y spectrometer, these shifts should be distributed

Produced by the Nb foil! hot plasma ion source. Due to the long half-life of As and the73

As (5 shifts)73

beam time will be the study of free H diffusion in lll—V compounds.

experiments to get a total beam-time of about one week for this target. The aim of this

requested shifts (including l shift stable beam) should be combined with other

of this target/ion source combination (setup and tuning of the laser system), the

possible to increase the yield by at least one order of magnitude. Due to the complexity

per sample. As stated by our collaborators — the experts for this ion source · it should be

pulse). Assuming 7 pulses per cycle this results in a minimum collection time of 1 hour

performed. This gave a yield of about 8·l0’ atoms per proton pulse (2-10protons!13

last run of this target, a short collection of Ag - populating Cd - has beenU-’ 117

This isotope requires a UC target with the chemical selective laser ion source. During the

cu (9 sums)m

stable beam in our request for '“Cd.m

shifts will bc distributed 0vcr 4 bcam-timcs within 2 years, we have included 4 shifts of

- 11

[22] R. Magerle et al., to be published

[2]] M. Deicher. Hypcrline lmeruclions 79 (1993) 681, and references [herein

[20] J. Chevallicr, in re1`.15],p. 219

Phys. Rev. B45 (1992) 8803[[9] J. Chevallier, B. Maehayekhi, C.M. Grauepain, R. Rabbi, and B. Theys,

Mat. Sci. Forum 143-147 (1994) 87[[8] M. Kunzer, U. Kaufmann, K. Maier, J. Schneider, N. Herres, L. Akasaki, and H. Amano,

[[7] J .C. Correia, Lo be published

[[6] B. Pajol, lnsi. Conf. Series 95 (1989) 437

[[5] W. Pfeiffer etal., io bc published

press

presented on Conference "Shallow Level Centers in Semiconduct0rs" Berkeley 1994, inW. Wilthuhn,Magerle, J.G. Marques, J. Meier, W. Pfeiffer, U. Resiloehner, M. Rueb, M. Toulemonde,

[[4] D. Forkel—Winh, N. Achtziger, A. Burchard, J .C. Correia, M. Deicher, T. Licht, R.

Appl. Surf. Sci. 50 (1991) 165[13] A. Bauriehter, M. Deicher, S. Deubler, D. Forkel, J. Meier, H. Wolf, and W. Witthuhn,

Wolf, D. Forkel, N. Moriya and R. Kalish, Appl. Phys. Lett. 58 (1991) 1751[12] W. Pfeiffer, M. Deicher, R. Keller, R. Magerle, E. Reeknagel, H. Skudlik, Th. Wichen, H.

Conference "Semi-insulating lll-V Materials", Warsaw 1994, in pressR. Magerle, J. Meier, W. Pfeiffer, U. Reisloehner, M. Rueb, W. Witthuhn, presented on

[1]] D. Forkel—Winh, N. Achtziger, A. Burehard, J .G. Correia, M. Deicher, T. Lieht,

Skudlik, Th Wichen, and H. Wolf, Man. Res. Symp. Proc. 262 (1992) 431[10] N. Moriya, M. Deicher, R. Kalish, R. Keller, R. Magerle, W. Pfeiffer, P. Pross, H.

Wolf, D. Forkel, N. Moriya, R. Kalish, Appl. Phys. Lett. S8 (1991) 1751[9] W. Pfeiffer, M. Deicher, R. Keller, R. Magerle, E. Reeknagel, H. Skudlik, Th. Wichert, H.

Mat. Sci. Forum 83-87 (1992) 593[8] A. Baurichter, M. Deicher, S. Deubler, D. Forkel, J. Meier, and W. Witthuhn,

[7] M. Deicher and W. Pfeiffer, in ref. [5]. P. 481

— 12