Reactive unbalanced magnetron sputtering of hydrogenated amorphous silicon and silicon oxide

I/acuum/volume44/numbers 3/4/pages 227 to 230/1993

0042-207X/9356.00+.00 © 1993 Pergamon Press Ltd

Printed in Great Britain

Reactive unbalanced magnetron sputtering of hydrogenated amorphous silicon and silicon oxide Department of Physics, Loughborough University of Technology, Ashby Road, Loughborough, Leicestershire LE1 1 3TU, UK G W Hall and R P Howson, and A C h e w , Institute of Surface Science and Technology, Loughborough University of Technology, Ashby Road,

Loughborough, Leicestershire LE11 3TU, UK

Hydrogenated amorphous silicon (a-Si : H) films have been deposited using an unbalanced magnetron, giving film deposition rates of around 3 nm s-1, a substrate bias of 30 V and an ion current of 2.50 mA cm -2. A polycrystalline silicon target was reactively sputtered in argon at 3 mtorr and hydrogen partial pressures ranging between 0 and 1 mtorr. Dopants of antimony and indium were also sputtered at the same time as the silicon in order to tailor the absorption coefficient and band gap of the resulting films. The analysis of dopant concentrations was undertaken using SIMS. A novel technique for calculating the absorption coefficients of the a-Si : H films by eliminating interference effects, using three measured optical parameters, is reported here. By replacing hydrogen with oxygen as the reactive gas, it was found to be possible to produce silicon oxides as films with a range of compositions, having optical properties ascribed to the monoxide through to the dioxide, refractive indices from 2 down to 1.45, with optical absorption occurring in the blue region of the spectrum for those with the highest indices. 1. Introduction A m o r p h o u s silicon, a-Si, is an important semiconducting material, with applications as diverse as photoreceptors t,2 image sensors 3 and photovoltaics. A m o r p h o u s silicon photovoltaics are almost exclusively thin film, and hold the greatest promise for inexpensive cells at present. Following research into a-Si photovoltaics begun in the late 1960s 4, the first devices were produced in 19745, by the glow discharge decomposition of silane. These films contained hydrogen as a result of the decomposition, which acts to saturate dangling bonds at the microvoids and at other defects in the silicon film. Silicon dioxide is also an important material in photovoltaics, with uses as anti-reflection layers, passivation coatings and barrier layers 6. Therefore, in the field of photovoltaic conversion, parameters such as the absorption coefficient, energy gap and refractive index need to be closely controlled in order for successful devices to be made.

To ensure the formation of silicon hydride complexes rather than interstitial hydrogen atoms, and to increase photovoltaic efficiency, previous work s'9 has identified the optimum deposition temperature. In the system of Figure 1 the heating station

2. Experimental Planar dc magnetron sputtering was identified as a technique ideally suited to varying the hydrogen content in the films. The magnetron was designed to provide a wide erosion zone combined with the benefits of an unbalanced magnetic field directing the plasma towards the growing film 7. The magnetron target was boron doped polycrystalline silicon ; the power supplied to it was maintained at 400 W and the magnetic field was optimized to avoid contamination from the front poles. Glass substrates were loaded singly into the load lock, before being transferred to the main chamber, having a 3 m t o r r partial pressure of argon. Once inside the main chamber, substrates locate on a rotatable cage, so that they can be moved to the deposition zone in front of the silicon source.

F-

l: Drivewheel 2: Guide wheel 3: Siliconsputtering source 4: Dopant sputteringsource 5: Pre-depositionheating station 6: Argoninlet 7: Reactivegas manifold 8: Cage Figure 1. The deposition system.

--7

9: Substrate & holderin depositionzone 10: Rotatingcage support 11: Gate valve 12: Separatelypumpedairlock 13: Substrate & holderon loading arm 14: Hydrogeninjectionline

227

G W Hall et al: Reactive unbalanced magnetron sputtering ~-d~4

~-d 4

The expression for the transmittance of a film giving interference, attached to a large substrate is J

RI~ Ra.~_

Ram~ -

--

AR the difference between R, and Rs.

__

T = n2 no

lo

-

[]

E

Hi[]

Tm

[]

Film side illumination

(1)

Tm -

[]

where t and r are the amplitude coefficients, 6 and/7 represent the real and imaginary parts of the phase change in the beam passing through the film. Rigorous manipulation of equation (1) enables the calculation of ~ from the simple measured parameters of transmittance, and the difference in reflectance from both sides of the substrate, AR.

Substrate side illumination

Figure 2. Interface parameters of a film/substrate combination illuminated from both sides. Parameters in bold are measurable.

is a halogen lamp which heats the substrate to 250°C before rotating the substrate to the deposition zone in front of the silicon cathode.

3. a-Si production The doping of silane deposited a-St using phosphine or arsine is well documented, but for the dc sputtering technique, a safe, repeatable technique was sought. Following work by Street e t a l ~o, solid source doping was used. The dopant source in Figure 1 was either an indium or antimony target, sputtered during deposition from the silicon source. A partial pressure of hydrogen was admitted through the injection line so that sputtered dopant atoms could either diffuse or be carried by the hydrogen to arrive at the substrate as silicon was being deposited on it. A key parameter for high efficiency solar cells is the absorption coefficient, a.. The conventional method for calculating e is to measure transmittance and reflectance, calculate absorptance, and hence ~. This simple method, however, is difficult to use for thin films showing interference, because the measured parameters are of the film plus substrate, not of the film alone. An alternative technique is derived here with reference to Figure 2. Assume a film, thickness d, is deposited onto a much thicker substrate. The combination is illuminated by a spectrophotometer beam, firstly from 'the film side, then from the substrate side. Components of the incident radiation I 0 are defined in the following.

3.1. Measured parameters. Tm the measured transmittance through both film and substrate ; Ram the measured reflectance (containing all components) from the film side ; and Rms the measured reflectance (containing all components) from the substrate side. 3.2. Derived parameters. T the actual transmittance through the film ; R~ interface reflectance between film and air; R2 interface reflectance between substrate and air, R. the actual reflectance in the film with illumination from the film side ; Rs the actual reflectance in the substrate with illumination from the substrate side ; and 228

t2t~ e 2~

2acos2&+rlr222e 4B,

1--2rlr2e

AV = \ ~ 1 "

1~ ' W

(2)

The refractive index of the film, n~, is measured using ellipsometry, but absolute accuracy is not important because equation (2) is not sensitive to n~ owing to the exponential multiplier. Figure 3 shows c~calculated using this technique as a function of incident light energy E, for a-St : H, where the hydrogen partial pressure was varied from 0 to 1 mtorr in 0.2 mtorr steps. The effect of increasing the hydrogen dilution is to reduce ~ at constant energy (wavelength), and to increase the band gap Eg, since the extrapolation of x/(c~E) to low energies on a ~/(~E)vs E plot approximates the band gap. These results are in agreement with the previous work of Shirafuji e t a l t 2, Van den Heuvel e t a l ~3 and Pinarbasi e t a l ~4 Figure 4 shows the absorption coefficient for a-Si:H doped with indium. In this instance the indium source was run at 200 W ; half the power of the silicon source. As with Figure 3, increasing the hydrogen content results in an increased band gap, but there is no significant difference between values of either c~or Eg for undoped or indium doped a-St. SIMS profiles showed the presence of indium in the films but it is suggested that because indium does not form a hydride, its incorporation is as interstitials. The hydrogen saturates dangling bonds as in the a-St with no dopant, and the indium plays no beneficial role. Unlike indium, antimony does form a hydride, so a-Si:H

4.5x1074,0x1073.5x I 0 7-

0.2roTaTor

~=~3.0x107

rH

@ 2.5x107t~ o= 2.0x 107~ 1.5x10 71.0xl07 5.0x 106 0.0x 1001.5

2

2.5

3

3.5

E (eV)

Figure 3. Absorption coefficientof undoped a-Si : H, varying hydrogen partial pressure.

(3 W H a l l et al : Reactive unbalanced magnetron sputtering

4.5x107

•

4'0x107

T

3.5x107.

3.5x107

0.6 m ~ Q ~

~

3.0x107

/

J/A/

g 25x,o7

~ 3.0xlO7.~ 2.5X107. : :a

0.4 mTor,'.~H

0.4 m

/////~

~

orr

0.8mTorrH

"5 2.0xlO 7-

H

0.2 mTorr H

.~ |.5xlO 70.2 retort H

o

~ 1.5xlO7

~

/

/1

I[

7

I~)mTo~H

1.Ox107. 5.0x | 06

5"0x106

~

O.OxlO (

1.5

2

2.5

3

3.5

0.0xl 00

....

J

_

.

i .... 1.5

E (eV)

~

"0.8 ,,Tor,'H

i .... i .... 2 2.5 E (eV)

i .... 3

3.5

Figure 4. Absorption coefficient of a-Si : H doped with indium. Indium target run at half the power of the silicon target.

Figure 6. Absorption coefficient of a-Si : H doped with antimony. Antimony target run at twice the power of the silicon target.

was deposited with varying hydrogen contents,• firstly with the antimony source running at half the power of the •silicon source (200 W), then with it at twice the power (800 W). Figures 5 and 6 show the absorption coeffÉcient in the visible for the above antimony dopant source powers, respectively. Once again, the shift to higher band gaps with increasing hydrogen contents is seen with both antimony source powers. The main difference between Figures 5 and 6 is in the values of ~. As a reference, take a mid-range hydrogen partial pressure of 0.6 mtorr. A t 2.5 eV, c~ is 5.2 × 106 m i for the 800 W antimony target, whilst for the 200 W antimony source ~ is 1.2 × 10 7 m - i. This is in agreement with the work of Ellis and Delahay 15, but contradicts the work of Schmid 16 who found ~ to increase with increasing hydrogen with using arsenic as a dopant. In heavily doped semiconductors such as ours (SIMS giving antimony concentrations of 1 x 1019 cm-3), valence band filling occurs and the drop in ~ at a given energy is due to the decrease in the number of available final states due to the Burstein-Moss shift. Steeper absorption coefficient gradients at higher energies are seen in Figure 6 compared with Figure 5, in addition to less

random variations in slope and a m o v e m e n t of Eg to higher energies. Qualitatively, SIMS shows there to be less antimony in the 200 W antimony doped a-Si samples, with approximately 1 x l019 antimony atoms c m - 3 ion the 800 W antimony sample produced at 0.6 mtorr hydrogen pressure. Hence it is assumed that the increased antimony content of the a-Si is increasing ~, steepening the slope, and widening the band gap. R o o m temperature conductivity measurements gave the result shown in Table 1. The hydrogen increases the conductivity from that of unhydrogenated sample, and the antimony doping increases it further. This conductivity of 9.7 x 10- 5 (f~ m ) - ~ is comparable with similar work of Street et allo but is lower than other workers' silane deposited 17 and rf sputtered films 18. The two latter authors report conductivities of approximately 10 s ([2 cm) - ~at 280°C, whereas our conductivity measurements were at r o o m temperature. The n-type antimony may have actually doped the a-Si more than is indicated above, because the silicon target material was itself boron doped. Hence a progression from p-type silicon through insulator would have to be overcome before n-type a-Si : Sb was reached, The combination of increased Eg with dopant, and lower conductivity than other similar work leads us to suggest that the antimony is assisting the process of hydrogenation more than acting as a conductivity enhancing dopant, with the parameters used in our work.

4.5x107 4.0x107~ 3.5x107-

4. Silicon dioxide

~"~E3.0x 107

/S

°1

~ 2.5x107

2.0X107 ]

~:~1.5X107~ 1.OxlO7-

Figures 7 and 8 show the results obtained for oxide film production. When less than 5 sccm oxygen was admitted through a piezo valve, films with refractive indices close to that of bulk

Table 1. Dark conductivity of a-Si, a-Si : H and a-Si : H : Sb

5.0xlO60.0xl00 ~ . . . .

, .... 1.5

~ .... 2

~ .... 2.5

, .... 3

3.5

E (eV) Figure 5. Absorption coefficient of a-Si : H doped with antimony. Antimony target run at half the power of the silicon target.

Sample

Hydrogen partial pressure (mtorr)

Sb target power (W)

Dark conductivity (f~ m)- 1

a-Si a-Si: H a-Si : H : Sb

-0.6 0.6

--800

4.5 x 10 5 8.9 x 10 5 9.7 × 10- 5 229

G W H a l l et a l . Reactive unbalanced magnetron sputtering

3.5-

measurable absorptance. D e p o s i t i o n rates o f m o n o x i d e a n d dioxide are approximately 2 a n d 0.8 n m s - ~, respectively.

5. Conclusions 2.5

2

•' l -

~

=

1.52

n = 1.48

n=1.49

1

....

4

r ....

4.5

i ....

5

J ....

5.5

i ....

6

i ....

6.5

I ....

7

n=l.4~

i ....

7.5

i ....

8

i ....

'

8.5

Oxygen flow (sccrn) Figure 7. Reactive indices of silicon oxides as a function of oxygen flow.

(1) T h e energy gap o f a-Si p r o d u c e d by reactive dc sputtering can be varied between 1.7 a n d 2.2 eV, by the addition of up to 1 m t o r r o f h y d r o g e n during deposition. (2) Sputtering indium d u r i n g deposition of a-Si has no m e a s u r a b l e effect o n the a b o v e optical parameters. (3) I n c o r p o r a t i o n o f different a n t i m o n y c o n c e n t r a t i o n s into a-Si : H has been achieved by sputtering such a target at different powers during deposition o f a-Si. (4) The doping with a n t i m o n y increased Eg further, a n d increased r o o m t e m p e r a t u r e d a r k conductivity by a factor of 2 over a-Si w i t h o u t h y d r o g e n a t i o n . (5) Reactive sputtering using oxygen has been s h o w n to offer the o p p o r t u n i t y to tailor refractive index a n d t r a n s m i t t a n c e o f silicon oxides whilst m a i n t a i n i n g low a b s o r p t i o n , by the simple control o f the deposition conditions.

100 n=1.45

References

n = 1.52

90 80

n=1.66 "-""~

702

50

,,=,_2o

/

/

~

40 ,1:3.3

7/

30 . . . . , . . . . ~ . . . . , . . . . , . . . . ~ . . . . r . . . . I . . . . 300 350 400 450 500 550 600 650 700 Wavelength(rm0 Figure 8. Optical transmittance of silicon oxides.

silicon a n d t r a n s m i t t a n c e o f less t h a n 6 0 % resulted. Between 5 a n d 9 sccm the t r a n s i t i o n between m o n o x i d e a n d dioxide occurs ; the refractive indices c h a n g i n g from 2 t h r o u g h to 1.45 with n o

230

1 Shimizu, S Sirai and E Inoue, J A p p l Phys, 52, 2776 (1981). 2 M Ondris, G Van der Voort and W den Boer, J Non- Cryst Solids, 59/60, 1243 (1983). 3F Okumaura, S Kaneko and H Uchida, Extended Abstracts of 15th Conference on Solid State Devices and Materials, pp 201-204 (1983). 4H F Sterling and R C G Swann, Solid State Electron, 8, 653 (1965). 5D E Carlson, US Patent 4,065,521 (1977). 6W A Anderson, A E Delahoy and R A Milano, J Appl Phys, 45, 3913 (1974). 7R P Howson, H A Ja'fer and A G Spencer, Thin Solid Films, 193/194, 127 (1990). 8 R Sridhar et al, J Non-Cryst Solids, 199, 331 (! 990). 9j M Asensi et aL 10th European Photovoltaic Solar Energy Conference Book of Abstracts, 1B09 (1991). ~°R A Street, N M Johnson, J Walker and K Winer, Phil Ma9 Lett, 60, 177 (1989). ]] R P Howson, J d e Physique, 25, 212 (1964). t2T Shirafuji et al, Solar Energy Mater, 23, 256 (1991). ]3j C Van den Heuvel et al, Solar Energy Mater, 22, 185 (1991). ~4M Pinarbasi, M J Kushner and J R Abelson, J Vac Sci Technol, A8, 1369 (1990). 15F B Ellis Jr and A E Delahoy, Solar Energy Mater° 13, 109 (1986). 16p E Schmid, Phys Rev, B23, 553 (1981). ]TW J Varhue and H Chao, Thin Solid Films, 169, 179 (1989). t~ M C Ozturk and M G Thompson, Appl Phys Lett, 44, 916 (1984).