Applied Surface Science 50 (1991) 415-419 415 North-Holland

Electrical properties of neutron-transmutation-doped InSe

B. M a r l

Centre de Recherches Nucldaires, IN2P3-CNRS/Universitd Louis Pasteur, Groupe "Recherches Physiques et MatOriaux'" BP 20, F-67037 Strasbourg Cedex, France

A . S e g u r a

Departament de Fisica Aplicada, Facultat de Fisica, E-46100 Burjassot, Valencia, Spain

a n d

A . C h e v y

Laboratoire de Physique des Milieux Condensds, 4 Place Jussieu, Tour 13, F-75005 Paris, France

Received 27 November 1990; accepted for publication 16 January 1991

Neutron-transmutation-doping (NTD) has been used to introduce a controlled amount of tin-related shallow donors with a uniform distribution into the layered semiconductor InSe. The electrical properties of transmutation-doped (TD) InSe, between 30 and 300 K have been measured and compared with those of InSe conventionally tin-doped during growth. After annealing at 450 o C, the lattice damage associated with NTD is removed and the donor centres become electrically active. For concentrations of transmutated Sn bigger than 1017 c m - 3 the free-electron concentration saturates and this is interpreted as a result of segregation determined by the Sn solubility limit in the host lattice. The Hall mobility at room temperature is higher in TD samples than in conventionally tin-doped ones for the same electron concentration.

1. Introduction

I n d i u m selenide (InSe) is a l ayered semiconduc- tor of the I I I - V I fami ly whose crys ta l s t ruc ture consists of graphi te - l ike layers b o u n d b y van der Waa l s forces [1]. A m o n g the d i f ferent dop ing agents, t in is the mos t sui table d o n o r in InSe as i t acts as a shal low d o n o r and reduces the res is t ivi ty of the mate r ia l wi thou t s t rongly affect ing the elec- t ron mobi l i ty [2,3]. The ion iza t ion energy of the t in- re la ted shal low donor , as de t e rmined f rom Ha l l effect measurements , is 22 meV [3]. The exis tence of compensa t ing acceptors in t i n -doped InSe is c lear ly suppor t ed b y the Hal l effect [3] and free- carr ier abso rp t ion results [4].

N e u t r o n - t r a n s m u t a t i o n - d o p i n g ( N T D ) is a technique ut i l iz ing the nuc lear reac t ion of the rmal

neu t rons wi th the i so topes p resen t in a semicon- duc to r mater ia l . W h e n a semiconduc to r is i r rad ia- ted wi th the rmal neut rons , the nuclei of the la t t ice a toms cap tu re neu t rons depend ing on the neu t ron cap tu re cross-sec t ion and relat ive a b u n d a n c e of each isotope. The exci ted nuclei decay to s table i so topes which could act as donors or acceptors .

In the case of InSe, the N T D m e t h o d permi t s the i n t r o d u c t i o n of a con t ro l l ed a m o u n t of t in in to the mater ia l . This d o p i n g technique presents two advantages : the i n t r o d u c t i o n of a con t ro l l ed a m o u n t of impur i t i e s due to the knowledge of the f lux of the rmal neu t rons and cross-sect ions of the d i f ferent i so topes and the un i fo rm d i s t r ibu t ion of these impur i t i e s a s suming that the i so topes are h o m o g e n e o u s l y d i s t r i bu t ed in the semiconduc to r mater ia ls . Unfo r tuna t e ly , a cons ide rab le a m o u n t

0169-4332/91/$03.50 © 1991 - Elsevier Science Publishers B.V. (North-Holland)

416 B. Mari et aL / Electrical properties of neutron-transmutation-doped InSe

of crystalline damage occurs during irradiation due to recoil of the 3' and fl particles produced by the N T D process. Thermal annealing is required in order to remove the lattice damage.

This doping method is well understood in the elemental semiconductors Si and Ge, and is nowa- days used to fabricate high-voltage silicon recti- fiers and thyristors [5,6] and low-temperature germanium bolometers [7]. The N T D technique has also been investigated in compound semicon- ductors like GaAs [8,9] and InP [10].

In this work, the N T D method has been ap- plied to the layered semiconductor InSe and the electrical properties of t ransmutat ion-doped (TD) InSe have been evaluated.

2. Experimental

The starting material used in this study is non- intentionally doped InSe, grown by the Bridgman method [11]. Samples of about 300 # m thick were extracted from the same ingot and irradiated with several thermal neutron fluences. Table 1 lists the thermal neutron fluences used (first column) and the Sn concentrations predicted theoretically (sec- ond column).

The resistivity and Hall measurements were carried out using conventional and van der Pauw methods on samples of thicknesses ranging f rom 10 to 40 / tm which were obtained by exfoliation. Ohmic contacts were made by thermal evapora- tion of In in vacuum. A closed He-cycle cryostat was used to do the measurements between 30 and 300 K.

In order to achieve the electrical activity of

donors created by neutron transmutation, the irradiated samples were annealed. The annealings were done in vacuum to avoid surface oxidation.

3. Neutron transmutation doping of InSe

The host atoms of InSe are composed of two In isotopes, USIn a n d 113In, and six Se isotopes. By far the most important nuclear reaction concerns n5In, which is the most naturally abundant iso- tope of In (95.7%) and has a very large thermal neutron capture cross-section of 199 barn. n3In appears in much lower quantities (4.3%) and has a much smaller cross-section of 10.7 barn. Between the six isotopes of Se only three transmutate to different atoms by capturing thermal neutrons. The relative abundance and capture cross-section of the isotopes 74Se, 8°Se and 82Se are 0.9%, 48.8% and 92% and 55 b, 0.08 b and 46 mb, respectively.

When InSe is irradiated with thermal neutrons, the transmutation reaction concerning HSIn is:

nSln( n, T) 116In ~ ll6Sn ( ~ - decay) .

99.70% of all the nuclear reactions transmute In into Sn, about 0.26% transmute Se into As and only 0.04% transmute Se into Br and Kr and In into Cd.

The data of natural abundances and cross-sec- tions were obtained f rom ref. [12].

4. Results and discussion

The third column of table 1 shows the resistiv- ity and the mobility, before annealing and at room

Table 1 Irradiation conditions and electrical properties before and after annealing

Thermal neutron [Sn] calculated Before annealing fluence (cm -2) (cm -3) p (fl cm) /t (cm2/V • s)

After annealing at 450 o C

O (f~ cm) /~ (cm2,/V • S) n (cm -3)

6 x 1014 2 >(1015 1.8 x 1015 6 x1015 16 625 6 x 1015 2 )<1016 323 272 1.8 >( 1016 6 x1016 29 405 6 X1016 2 X1017 376 192 1.5 >( 1017 5.4 × 1017 4770 - 6 x1017 2 ×1018 7470 164

5.8 663 1.6 x 1015 1.24 810 6.2 × 1015 0.43 704 2.1 )< 1016 0.15 715 5.8 × 1016 0.080 655 1.2 )< 1017 0.065 630 1.5 X 1017 0.070 533 , 1.7 3< 1017

B. Mari et aL / Electrical properties of neutron-transmutation-doped lnSe 417

temperature, of the irradiated samples. The trend is that the higher is the thermal neutron fluence used, the higher the resistivity and the smaller the mobility. This effect is associated with the lattice damage produced by the fast neutrons which un- avoidably are part of the total neutron flux and the damage due to the recoil of high-energy 7 and /3 particles produced in the transmutation process of USln. In this case, fast neutrons are about 5% of the thermal neutron flux.

The lattice damage can be removed after an- nealing. Fig. 1 shows the resistivity of one sample irradiated with a thermal neutron fluence of 6 x 1017 n / c m 2 for successive annealings. The first annealing was performed by remaining 10 min at every temperature and about 3 h at the highest temperature (360 o C). No changes in the resistiv- ity were detected for long annealing times. The value of the resistivity achieved after annealing at 360 °C is not stable and the resistivity increases when the sample returns to room temperature. After a second isochronal annealing till 460 ° C, the electrical activity of donors is completely achieved in an irreversible way. Then the electrical resistivity remains nearly constant between room

1o~r,~ , ~ , ,

10-1 I " c ) .. . , ~

10 -z [ I I I I 0 100 200 300 ~,00 500

T(oC)

Fig. 1. Resistivity of an irradiated sample for several successive annealings: (a) 360°C, (b) 460° C , and (c) 480°C . The ther-

mal neutron fluence used was 6 x 1017 n / c m 2.

10 TM I r /I /

/ /

/

1017 /

'3

1016

eared of z~60°C

101s / ~ I I I i0 Is 1016 1017 1018

[Sn]calcu[oted (cm -3)

Fig. 2. Free-electron concentration measured at room tempera- ture for the samples annealed at 460 ° C as a function of the Sn

concentration calculated theoretically.

temperature and 480°C, like in conventionally tin-doped InSe, due to the shallow nature of tin which is completely ionized at room temperature [2,31.

The values of the resistivity, mobility and free- carrier concentration at room temperature after annealing at 450 ° C appear in the fourth column of table 1. The resistivity decreases when the amount of tin added increases. After annealing at 450 o C, the lattice damage has been removed and all the tin atoms occupying substitutional sites become electrically active. The free-electron con- centration at room temperature measured by the Hall effect shows good agreement with the values of Sn concentrations expected from the thermal neutron capture cross-sections and fluence in the 1015-1017 a toms /cm 3 range (fig. 2). The doping effect of tin clearly appears as a monotonous increase of RT electron concentration as the tin content increases. However, for higher concentra- tions of transmutated Sn, the free-electron con- centration saturates. This can be interpreted as the result of a segregation determined by the Sn solu- bility limit in the host lattice. This interpretation is confirmed by the high segregation coefficient of Sn in InSe [11] and the observation of a similar limit (n --- (2-3) X 1017 cm -3) of the free-electron concentration in conventionally tin-doped InSe [2,3]. The tin atoms which are not electrically

418 B. Mari et a L / Electrical properties of neutron-transmutation-doped lnSe

10 ~

>

%

S

10 3

I I

0 0 O 0

° O O o o

c] o

o o

0

0 o

A

&

2 I I 30

i | i i , I I

al o

o o

o

b} 0 0

0

t3

0

~ d l

¢

0

A &

I '

o

0 0

0 0

0 0

O o

o ¢ ¢ o o o ¢

tX ~ & A ~ O o° o

~ o

I I I l t i t I

1 0 0 300 T I K )

Fig. 3. Hall mobility of several annealed TD samples as a function of temperature. The values of the [Sn] calculated are: (a) 6)<1015 cm -3, (b) 6 x l 0 a6 cm -3, (c) 2)<1017 cm -3, (d)

5.4 x 1017 c m - 3.

active probably precipitate in stacking faults where they act as deep donors. Other Sn atoms can occupy acceptor positions [3].

Fig. 3 shows the electron Hall mobility versus the absolute temperature for several samples irradiated with different thermal neutron doses. The evolution of the temperature dependence of the electron mobility, as the tin content increases, corresponds to the expected behaviour and is simi- lar to that observed in conventionally tin-doped samples [2].

In fig. 4, the electron concentration has been represented as a function of the inverse tempera- ture for several irradiated samples. Again the tem- perature behaviour of the electron concentration in the irradiated samples is similar to that ob- served in conventionally tin-doped ones [2,3]. Nevertheless, the apparent electron concentration at low temperature (30 K) is slightly higher ( (6- 11) × 1014 c m - 3 ) than in conventionally tin-doped InSe ( ( 1 - 4 ) × 1014 c m - 3 ) and does not change with the neutron dose. This degenerate behaviour of the electron concentration has been explained through the existence of 2D electron accumulation

layers. Initially, those layers were proposed to be associated to planar precipitates of donor impuri- ties in stacking faults [13]. As the high mobility of 2D electrons in not purposely doped InSe [2] necessarily implies the spatial separation between 2D electrons and ionized donors, a new model has been recently proposed [14]. 2D electric subbands would appear due to the presence of thin regions of c-InSe separated from the bulk ?-InSe by two stacking faults. Electronic levels in c-InSe would be shifted up in energy due to size quantization. Electrons are then transferred to the y-region and the resulting electric field (along with the stacking fault barrier) confines them in 2D subbands. In the framework of this model 2D electrons are present when thin c-InSe regions exist, even if donors are homogeneously distributed, which is the case in N T D InSe. The upper limit of the 2D electron concentration is determined by the fact that the c region becomes positive when electrons are transferred to the 2D subbands, which shifts down the energy levels, resulting in a saturation of the 2D electron concentration even if the donor concentration in the C-regions increases.

1017

' 1016 E

c

ii s

I I

& A

c i °°oo OoO • &

o

0

d]OoOooo °• 0

I I 5 10

I I I I

,o,

0 O o •

O o •

°<> a~ • o

o & A " _ o:

I f I I 15 20 25 30

1000 / T ( 103K -'~ }

Fig. 4. Electron concentration versus inverse temperature for several TD samples. The calculated values of the tin added are: (a) 5.4)<1017 cm -3, (b) 6)<1016 cm -3, (c) 2)<1016 cm -3 and

(d) 6× 1015 cm 3.

B. Mari et al. / Electrical properties of neutron-transmutation-doped lnSe 419

In conventional ly t in-doped InSe, the tempera- ture dependence of the free-electron concentra t ion has been interpreted through a single-donor single-acceptor model f rom which an ionization energy of 22 meV is obtained for the tin-related shallow donor [3]. However, when the same model is applied to T D InSe, an energy of about 15 -17 meV is obtained. The compensat ion ratio N A / N D

obtained with this model is similar in bo th cases (20%). In conventional ly doped InSe the 2D elec- tron concentra t ion is lower and the single-donor single acceptor model applies down to 55 K and through electron concentrat ions going f rom 1015 to 1017 cm-3 . In N T D InSe the model only applies for temperatures higher than 100 K and electron concentrat ions changing by a factor 10. Then the lower value of the tin-related shallow donor ioni- zation energy so obtained can be due to the lower accuracy of the model (reduced temperature and concentra t ion intervals). Otherwise, this result can also be due to a temperature dependence of the shallow donor ionization energy.

5. Conclusion

Neut ron- t ransmuta t ion-doping has been used to introduce a controlled amoun t of tin-related shallow donors into the layered semiconductor InSe. The lattice damage associated with the N T D process is removed by annealing the samples at 4 5 0 ° C . The free-electron concentra t ion shows good agreement with the Sn concentra t ion added till 1017 cm -3. For higher concentrat ion, the free- electron saturates as a consequence of the Sn solubility limit in the host lattice. The electrical properties of t ransmuta t ion-doped InSe between 30 and 300 K have been measured.

Acknowledgements

The authors gratefully acknowledge Dr. Stamp- tier for the neut ron irradiat ions at the Service de la Pile Universitaire (CRN, Strasbourg). One of us (B.M.) acknowledges the Ministerio de Educat ion y Ciencia for financial support .

This work was suppor ted through the Spanish Government : C I C Y T grant number PPA86-0264.

References

[1] A. Likforman, D. Carre, J. Etienne and B. Bohet, Acta Cryst. B 31 (1975) 1252.

[2] A. Segura, F. Pomer, A. Cantarero, W. Krause and A. Chevy, Phys. Rev. B 29 (1984) 5708.

[3] B. Mari, A. Segura and A. Chevy, Appl. Phys. A 46 (1988) 125.

[4] B. Mari, A. Segura and A. Chevy, Phys. Status Solidi (b) 130 (1985) 793.

[5] E.W. Hass and M.S. Scholler, IEEE Trans. Electron De- vices ED-23 (1976) 803.

[6] H.M. Janns and O.M. Marmros, IEEE Trans. Electron Devices ED-23 (1976) 797.

[7] N.P. Palaio, S.J. Pearton and E.E. Hailer, J. Appl. Phys. 55 (1984) 1437.

[8] J.E. Mueller, W. Kellner, H. Kniepkamp, E.W. Haas and G. Fischer, J. Appl. Phys. 51 (1980) 3178.

[9] M.A. Vesaghi, Phys. Rev. B 25 (1982) 5436. [10] B. Lee, N. Pan and G.E. Stillman, J. Appl. Phys. 62 (1987)

1129. [11] A. Chevy, J. Appl. Phys. 56 (1984) 978. [12] Handbook of Chemistry and Physics, 60th ed. (1980-81). [13] E. Kress-Rogers, R.J. Nicholas, J.C. Portal and A. Chevy,

Solid State Commun. 44 (1982) 379. [14] J.P. Martinez Pastor, Tesis Doctoral, Universitat de

Valrncia (1990).