THE BAND GAP OF GLOW-DISCHARGE-PRODUCED AMORPHOUS SiOx

(1)

HAL Id: jpa-00220854

https://hal.archives-ouvertes.fr/jpa-00220854

Submitted on 1 Jan 1981

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

THE BAND GAP OF

GLOW-DISCHARGE-PRODUCED AMORPHOUS SiOx

R. Carius, R. Fischer, E. Holzenkämpfer

To cite this version:

R. Carius, R. Fischer, E. Holzenkämpfer. THE BAND GAP OF GLOW-DISCHARGE-PRODUCED AMORPHOUS SiOx. Journal de Physique Colloques, 1981, 42 (C4), pp.C4-1025-C4-1028.

�10.1051/jphyscol:19814224�. �jpa-00220854�

(2)

CoZZoque C4, suppl6rnent au nO1O, Tome 42, octobre 2981

T H E BAND GAP O F GLOW-DISCHARGE-PRODUCED AMORPHOUS SiO,

R. Carius, R. Fischer and E. HolzenkZmpfer Fachbereich Physik, Universitat Marburg, F.R.G.

Abstract.- Amorphous films of composition SiO (H,N) were prepared in a gas discharge of suitable mixtures of SiH4 wit% N20 or O2 (Ar). The photo- luminescence of these films showed two bands, one of which shifts to higher energy with increasing oxygen content similar to the optical absorption edge. As X increases this luminescence band broadens, and the absorption edge becomes less steep. - This increase of the band gap and the tailing can be understood with a simple model which deals with the role of the Si-Si and Si-0 bonds in SiO,. The band gap of SiO, is chiefly determined by the Si-Si bonding and antibonding bands. Oxygen bridges withdraw elec- tronic states from them and produce new states further below in the Si02 valence band region. The effect on the Si-Si bands is, therefore, mainly indirect in that the Si-Si bond coordination number is reduced. The bands become narrower and the gap widens.

Introduction.- Hydrogenated amorphous silicon appears to have the right band gap for solar cell purposes. There is, however, a wide range of other applications for amorphous semiconducting films. Itwould, therefore, be of great advantage to have the possibility of adjusting the band gap to the special requirement. One way to do this is to alloy silicon with oxygen, thus producing films of the composition SiO,.

In this alloy system, the band gap can be varied between about 1.8 eV (a-Si:lI)' and 9 eV (Si02)'. Here the amorphous phase has a clear advantage over the crystalline:

though oxygen does not enter substitutionally (as e.g. the atoms in the mixed 111-V crystals), the composition can be continuously varied from x = O to X = 2.

Results.- The SiO,(H,N) films used here were deposited from mixtures of SiH4 and

N 2 0 in a gas discharge at a rate of 3 8s-l and a substrate temperature of 2 8 0 ~ ~ .

Very similar films were obtained using, instead of the N20, oxygen diluted in argon.

The oxygen content was determined by proton backscattering. The films contain up to 7 % nitrogen; the hydrogen content was estimated to be on the order of l 0 %. Further details are described elsewhere3.

The samples showed two luminescence bands. The center-of-mass energy of the first changes only little around the position of the well-known a-Si:H band4 upon varying the oxygen content. The second emerges from the first at about x=0.2 and its center-of-mass shifts to higher energy with X as shown in Fig. l (EL>). This band was therefore attributed to the SiOx matrix, while the low-energy band was be- lieved to indicate the presence of Si clusters in these films3. A corresponding structure is seen in the photoconductivity spectra, while none is seen in the absorption edges at a > 1 0 ~ c m - ~ . ~

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:19814224

(3)

C4-1026 JOURNAL DE PHYSIQUE

The absorption edges in the SiOx(H,N) system are rather featureless and there is some arbitrarizess as to the determination of the band gap. We used here the energy E04, where the absorption coefficient a equals 104cm-I (Fig. 1).

The slope dEo4/dx is at first about 1 eV, but for

X larger than about 0.9 it drastically increases.

EL2(x) behaves, at lower energy, in a similar way; above X = 1.2 no luminescence signal can be detected because only a small fraction of the exciting light is absorbed in the films.

In Fig. 2 energy values are plotted, which can be taken as a measure of the tailing of the bands: the reciprocal slope AEa of the (logarith- mic) absorption edge, the width of the second luminescence band (AE1/2 (L2)), and one half of the energy distance between EL2 and E04. These tailing energies increase first rather slowly and then ( X > l) rapidly. This curve goes over a maxi- mum as the value of AEa for SiO2 is again small.' These band edge properties are, naturally, depend- ent on the deposition conditions. For example, the steep rise of Eo4(x) starts only at about

,

^{X =}1.4 in evaporated films; but the qualitative features are the same6.

Model.- The band gap of amorphous semiconducting alloy films was experimentally determined as func- tion of the composition for SixGel-x (H)

'

^and

SixC1-x(H)

*.

In both cases a linear shift with X is obtained. This makes these systems similar to crystalline semiconducting alloys, where the rele- vant band extrema also exhibit an almost linear shift with X, which can be understood with the 2- coherent-potential approximation9. - SiO, behaves

differently: the slope of EO4(x) increases

X AEo - - X

A E (L2) strongly with X. We will show that this can be

,

understood with the aid of a simple tight-Qinding

o model

,

model. For this purpose, the effects of the H or

1.5- '\

'

^N impurities or of Si clusters will not be con-

3 \

^sidered.

\ I

X m I \ \ In the bond-orbital picture, the splitting

E

^{I -} ?'

:

^'\ of the Si 3sp3 hybrid orbitals into bonding and

0

m I \ antibonding orbitals generates the valence and

.-

C

-

.-

^I \

'l

conduction bands of amorphous or crystalline si-

-

0 ^I ^!: Licon (Fig. 3 ) . In a fully coordinated random network the addition of oxygen means that some o f t h e Si-Si bonding and antibonding states are replaced by the corresponding states of the Si-0 bond.

f si

0,

Oxygen further introduces the 2p lone-pair states, which eventually form the upper valence band of

0 -r + ~ i 0 ~1 2 . ~- ~As the energies of the oxygen- ,

0 0.5 1 1.5 related states are largely different from the composition X

energies of the Si-Si bonding and antibonding Fig. 2:

it-^^

dependence states, the effect of oxygen is, in the first

of the tailing of the place, the reduction of the overlap of the states bands (details see text) associated with the Si-Si bonds.

Using the results of the tight-binding approximation in the formulation of Weaire and

(4)

In silicon, one band has 6 nearest neighbors, and so 6B is the total width of one band. In this case the consideration of bonds (instead of atoms) is advantageous since oxygen replaces bonds and not atoms. - The average gap EG of SiO, can be computed by assuming that the oxygen bridges are randomly distributed, - and using an average coordination number z = Z z Q . Here qz is the probability that a given Si-Si bond is surrounded by z Si-Si bonds. Since

we have ;=6-3x, and thus

which is a linear relation. This, however, is not observed in the entire X-range (Fig. 1 ) .

Fig. 3: Bond orbital picture for SiO (schematic)

Calculating the average gap in this way (eq. 3) implies that the wave-functions have the same amplitudes on highly coordinated Si-Si bonds as on bonds with lower coordination. This is certainly not the case as the corresponding energy is lowered by giving the highly coordinated bonds more weight. From this we conclude-that a more localized description of this system should be more appropriate. As a first step in this direc- tion we first determined the band gaps EG,, of the z-fold coordinated bonds. An absorption edge of standard shape15

a, (E) = AZ (E - E ) 2 / ~

G,z (4)

is assigned to these bonds. The total absorption edge is then a weighted superposi- tion of the individual edges (Fig. 4):

X

Here pz=qZ(l

-2)

is the probability to find a z-fold coordinated Si-Si bond in the S i G network. ao(E), the absorption of zero-fold coordinated bonds, was approximated by a Gaussian. For ag(E)=asi(E) and aSi02(E) the respective fits to the absorption edges1 r of a-Si :H and a-Si02 were taken, and

)v2 1

⁼^{5 eV.}^{l 6}

This set of curves agrees qualitatively quite well with what is actually observed

.

Eoq(x) and, approximately, AEa(x) can be directly extracted from this fi- gure and are shown in Figs. 1 and 2, respectively. E04(x) agrees rather well with the experimental results, whereas AEa(x) only reproduces the gross features.

Discussion.- The present model neglects several influences; it was, for example, as- assumed that there is negligible overlap between the original Si-Si states and the states inroduced by the oxygen bridges. This is certainly possible for the valence band, but perhaps not for the conduction band as this lies close to the Si-0 antibonding states. Then, the chemical shift of the Si states due to the ionicity of the Si-0 bond was neglected. In this local treatment, however, this affects the results only, if the Si atoms between which the central bond is formed, have different oxy- dation states. Further, the emphasis on the local configurations is certainly too strong. -

(5)

C4-1028 JOURNAL DE PHYSIQUE

I There are two points,

however, which are made clear by this treatment. One is that the absorption process leading to a = lo4 cm-' is mainly due to

excitations of the Si-Si bonds for almost the entire X-range

(up to X

"

1.9). And second:

the influence of correlation effects is greater in Si& than in silicon, since the ground

(bonding) and excited (antibonding) states of Si-Si bonds are spatially correlated.

0 2 4 6 8 10 Acknowledgement.- We would

l 2 l i k e t o o f . P. Thomas

photon energy /eV for many interesting discus-

sions.

Fig. 4: Absorption edges in the system SiOx according to the model calculations

References

1. see e.g. FERRATON J.P., ANCE C., DONNADIEU A., Thin Solid Films

78

(1981) 207 2. PHILIPP H.R., J.Phys.Chem.Solids

2

(1971) 1935

3. CARIUS R., FISCHER R., HOLZEN~MPFER E., STUKE J., J.Appl.PhyS. (1981) in press 4. see e.g. FISCHER R., in: Amorphous Semiconductors, ed. by M.H. Brodsky,

Springer-Verlag, Berlin 1979, ch. 6

5. HOLZEN~MPFER E., STUKE J., Verhandl. DPG (VI)

15

^{(1980) HL}63

6. HOLZEN~MPFER E., RICHTER F.-W., STUKE J., VOGET-GROTE U., J.Non-Cryst-Solids 32 (1979) 327

7. ~ J S C H I L D T D., FISCHER R., FUHS W., phys.stat.so1. (b)

102

(1980) 563 8. ANDERSON D.A., SPEAR W.E., Phil.Mag. 35 (1977) 1

9. STROUD D., EHRENREICH H., Phys.Rev. B r(1970) 3197 10. BENNETT A.J., ROTH L.M., Phys.Rev. B

4

(1971) 2686 11. LANNOO M., ALLEN G., Solid State Comm.

8

(1978) 733 12. MnRTINEZ E., YNDURAIN F., Solid State Comm.

37

(1981) 979 13. WEAIRE D., THORPE M.F., Phys.Rev. B

4

(1971) 2508

14. HEINE V., J-Phys. C

4

(1971) L 221

15. TAUC J., in: Amorphous and Liquid Semiconductors, ed. by J. Tauc, Plenum Press, London 1974, ch. 4

16. a similar value was reported by SINGH J., Phys.Rev. B

23

(1981) 4156