- Email: [email protected]
Electrical characteristics of silicon nitride on silicon and InGaAs as a function of the insulator stoichiometry
Applied Surface Science 52 (1991) 45-52 North-Holland
applied surface science
Electrical characteristics of silicon nitride on silicon and InGaAs as a function of the insulator stoichiometry P.E. Bagnoli a, A. Piccirillo b, A.L. G o b b i c a n d R. G i a n n e t t i a a Istituto di Elettronica e Telecomunicazioni, Unioersit& di Pisa, Via Diotisaloi 2, 56100 Pisa, Italy b Centro Studi e Laboratori di Telecomunicazioni CSELT, Via Reiss Romoli 274, 10148 Torino, Italy c Centro de Pesquisa e Desenvoloimento TELEBRAS, Campinas, Brazil
Received 7 December 1990; accepted for publication 13 May 1991
The interface state density along the semiconductorenergy gap and the fixed charge were evaluated in SiNx/InGaAs and SiNx/Si interfaces. The insulator layer was deposited by plasma-enhanced chemical vapour deposition (PECVD) using several ammonia/silane gas ratios. In both the samples the measurements revealed two main peaks of interface states whose height is a function of the insulator layer stoichiometry. Further analysis by infrared and Auger electron spectroscopyand electron spin resonance measurements enabled the peaks to be identified as the two silicon-related defects in silicon nitride cited in the literature. The nitrogen dangling bonds were found to affect the fixed charge of the structure. The role of hydrogen in passivating silicon and nitrogen dangling bonds will also be discussed.
1. Introduction Silicon nitride obtained by chemical vapour deposition is now widely used in I I I - V semiconductor device technology. This material has been found to be a promising gate insulator for MISFET transistors on compound semiconductors [1] and a passivating layer for optoelectronic devices [2]. In both the above applications the physical structure and the electrical characteristics of the insulator/semiconductor interface play an important role in determining the reliability and the efficiency of the devices. For instance, the dark current of InGaAs-based PIN photodetectors is strongly affected by the density of states at the passivating insulator/semiconductor interface [2]. Plasma-enhanced chemical vapour deposition technique (PECVD) is particularly attractive as a passivation technique due to the lack of oxygen in the deposition process, its relatively low temperature and the possibility of achieving an in-situ plasma cleaning of the semiconductor surface before the insulator layer deposition.
The electrical characteristics of the insulating layers (resistivity, breakdown strength, interface state density, fixed insulator charge) not only depend on the cleaning procedure before deposition [3] and on the process-induced damages on the semiconductor surface [4], but they are also affected by the insulator stoichiometry resulting from the choice of the deposition parameters such as the substrate temperature, the R F power and the a m m o n i a / s i l a n e gas ratio in the deposition chamber [5-7]. Therefore, in order to obtain an electrically reliable passivating layer, a better knowledge of the origin and nature of the interface states and the other unsuitable parameters of the SiNx/semiconductor interface is required. The aim of this study is the identification of the structural defects of the interface which cause the interface states and the insulator charge. We compared the results of interface state density and fixed charge evaluations in P E C V D SiNx/Si and S i N ~ / I n G a A s MIS capacitors where silicon nitride layers were deposited varying the a m m o n i a / s i l a n e
0169-4332/91/$03.50 © 1991 - Elsevier Science Publishers B.V. All fights reserved
46
P.E. Bagnoli et al. / Electrical characteristics of SiNx / In GaAs and SiNx / Si versus insulator stoichiometry
gas ratio. The other parameters, i.e. the substrate temperature and the R F power, were chosen in order to optimize the resistivity and the breakdown strength of the insulator layer and were discussed in a previous experimental investigation [51. The samples were also characterized by means of physical and structural analyses - electron spin resonance (ESR) measurements, infrared spectroscopy (IR) and Auger electron spectroscopy in order to correlate the electrical properties with the composition of the insulator layers.
capacitance meter. The energy distributions of the interface state density were obtained using the high-frequency C - V method [9]. The fixed charges were evaluated by the relative shift of the C - V curve along the voltage axis at the flat-band condition. Auger electron spectroscopy measurements were carried out on the S i N x / I n G a A s structures in order to investigate the stoichiometry of the insulating film. Furthermore, a semi-quantitative analysis of the, S i - H and N - H bond concentrations was made by using infrared spectroscopy on the SiNx/Si samples, according to the Lanford and Rand method [10].
2. Experiment 3. Results Metal-nitride-semiconductor structures were fabricated on low-doped epitaxial I n G a A s (n = 4 x 1015 cm -3) grown on highly sulfur-doped InP by liquid phase epitaxy (LPE) and on n-doped bulk silicon (100) (n = 3 x 1015 cm-3). The silicon nitride layers were deposited by PECVD using several values of the a m m o n i a / s i l a n e gas ratio in the range 2 - 2 0 and at a constant substrate temperature (300°C) and R F power (20 W). The thickness of the insulator layers was in the range 600-900 ,~. For these samples no thermal treatments were used after the insulator deposition because the annealing procedure can degrade the stoichiometry of the compound semiconductor due to the evaporation of the G r o u p V element [8]. In addition, a second series of SiNx/Si samples were fabricated varying the substrate temperature in the range 200-350°C at a fixed gas ratio ( R = 4) and annealed at 300°C for 10 min in forming gas. Aluminium dots with an area of 1.53 x 10 - 2 cm 2 were deposited on all the samples as top electrodes. On silicon an aluminum film was used as the back contact and on the compound semiconductor alloyed A u - S n . Silicon nitride layers were also deposited on silica glass substrates for the ESR measurements during the same deposition. The electrical characteristics of the interfaces were evaluated on both types of MIS samples by high-frequency (1 M H z ) c a p a c i t a n c e - v o l t a g e measurements, in the bias range from - 3 0 V to + 10 V, using an H P 4280A computer-controlled
The energy distributions o~f the interface state density as a function of the a m m o n i a / s i l a n e gas ratio ( R ) are shown in fig. 1 for the S i N x / I n G a A s samples and in fig. 2 for the SiNx/Si. The two samples show interface state densities with similar shapes and behaviour with respect to the variation of the insulating layer composition. All the curves have two main peaks near the edge of the two semiconductor bands. In the compound semiconductor the peaks are located at 0.15 and 0.65 eV, respectively, from the top of the valence band, whereas in the silicon samples the localization energies are 0.20 and 0.89 eV, respectively.
"7 >
20
•
A
//~ ~' E
R = 2.33 R=4 R=9
15
o
ca 10
5
==
,
0 u.O0
,
,
,
i
,
,
,
,
i
,
,
,
,
i
,
,
0.15 0.30 0.45 Energy from valence band
,
,
i
,
0.60 ( eV )
,
,
i
i
0.75
Fig. 1. Energy distribution of the interface state density in the SiNx/InGaAs samples for several values of the ammonia/ silane gas ratio (R).
P.E. Bagnoli et aL / Electrical characteristics of SiNx / I n G a d s and SiNx / S i versus insulator stoichiometry 10
~-" 25 ¢ ~'
20
•
R = 2.33
~
&
R=4
//
A
R=9
47
•
~'
InGaAs
~
Si
8 e~
o
~o
6
15
v-
W n-
4
t0 X 0
"6
,g
5
e~ _x
2
u.
0
UI
0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . u.o 0.2 0.4 0.6 0.8 1.0
'
'
0
1.2
5 10 AMMONIA / SILANE
15 RATIO
GAS
20
E n e r g y t r o m v a l e n c e b a n d ( eV )
Fig. 2. Energy distribution of the interface state density in the SiNx/Si samples for several values of the ammonia/silane gas ratio (R).
The effect of the variation of the gas ratio in the reactor c h a m b e r can be seen by the height of the peaks and is generally an increasing function of R, as shown in fig. 3. On the contrary, the mid-gap value of the interface state density in all the samples is quite independent of the gas ratio and presents similar values for the two types of samples (4 x 1012 c m -2 eV -1 for the c o m p o u n d semiconductor and ( 2 - 3 ) × 1012 c m -2 eV - ] for the silicon samples). The plots of the fixed charges Qf (cm - 2 ) are shown in fig. 4. The corresponding fiat-band voltages calculated f r o m the C - V curves lie in the ranges f r o m - 5 to - 3 V for the silicon samples
_2'I
>
E
• ....
t---
2O
.... = ' " "
I n G a A s left p e a k I n G a A s r i g h t p.
Fig. 4. Plots of the measured fixed charge for the two types of insulator/semiconductor interfaces.
and f r o m - 0 . 8 to - 0 . 0 1 V for the c o m p o u n d semiconductor. This fixed charge shows the opposite behaviour to the interface states as a function of the gas ratio: it slightly decreases with increasing gas ratio. Moreover, the silicon samples show an average value which is greater than the corresponding value measured in the I n G a A s samples. The results of the physical characterization of the insulating layers are summarized in fig. 5 where the silicon and nitrogen concentrations and the relative a m o u n t s of S i - H and N - H b o n d s are plotted as a function of the gas ratio. N o t e that the s t a n d a r d stoichiometry of the silicon nitride ( S i / N = 0.75) can be obtained using a value of the a m m o n i a / s i l a n e gas ratio close to 9 at which a m a x i m u m of the dielectric layer resistivity has been f o u n d [6].
Si left p. Si right p.
40i
15
C ~
•
¢0 v >-
o v >-
10
P"
5
Z 141 a .
.
_o
¢/) Z IJJ a CI Z O m
a"
.
30
Si-H N-H Si/N
.
,
5
,
,
,
,
i
,
,
,
10
AMMONIA / SILANE
,
i
,
15 GAS
,
,
,
i
i-
z, Z
20
RATIO
Fig. 3. Plots of the left-hand and the fight-hand peak heights of the interface state densities for the two types of samples as a function of the gas ratio.
0 o9
.
0
.
.
.
,
5
.
.
.
.
,
10
.
.
.
.
,
,
15
,
,%~
0
20
AMMONIA / SILANE GAS RATIO
Fig. 5. Plots of the Si-H and N-H bond densities and of the Si/N ratio in the insulator layers.
48
P.E. Bagnoli et al. / Electrical characteristics of SiNx / I n Go.As and SiNx / S i versus insulator stoichiometry
Fig. 5 shows that the composition of the insulating layer strongly depends on the a m m o n i a and silane content during the deposition process: the film was found to be silicon-rich for lower gas ratios and strongly nitrogen-rich for higher ones. Furthermore, this figure shows that the bonded hydrogen concentration in the insulating film, which generally depends on the substrate temperature during the deposition process and on the annealing procedure [11], strongly depends on the gas ratio.
4. Discussion Since the interface states and the fixed charge are related to structural defects at the i n s u l a t o r / semiconductor interface or in the bulk of the dielectric layer, the origin and nature of the electrical parameters of the silicon nitride/semiconductor interfaces can be investigated on the basis of the above experiments by correlating the electrical and physical results. By comparing figs. 1, 2 and 3, it can be seen that (i) the two peaks of the interface state density, the first one close to the conduction band edge and the second one close to the valence band edge, are present in both types of samples, regardless of the semiconductor substrate; (ii) the peak values on the two types of samples are basically of the same order of magnitude for the same deposition conditions, specially for the three lower gas ratios; (iii) the peak heights only seem to depend on the gas ratio used during the insulator layer deposition. On the basis of the above arguments it can be stated that the two peaks in the interface state density are due to structural defects which are related to the silicon nitride layer rather than to the semiconductor surface. In order to identify the origin of the two observed peaks our results can be interpreted by taking into account the theoretical analysis of the main silicon nitride defects performed in ref. [12] using the tight-binding and recursion method. For bulk Si3N 4 two possible levels of dielectric-related states caused by silicon dangling bond defects were found: a lower (-Si ° ) trapped hole level and
a higher ( - S i - ) trapped electron level, separated by a positive correlation energy of 0.66 eV. Both these two trap levels are located in energy near the mid-gap of the silicon nitride and are energetically aligned with the gap of Si. The value of the theoretical correlation energy agrees well with the experimental result obtained from the silicon samples (0.69 eV). Therefore, the two observed peaks in the SiNx/Si can be identified with the theoretical ones calculated by Robertson and Powell [12]. As previously discussed, the peaks of the interface state density of S i N x / I n G a A s samples have similar characteristics with respect to the SiNx/Si sample ones, so they can be ascribed to the same origin, that is silicon dangling bonds. The slightly lower value of the correlation energy (0.5 eV) may be induced by the different semiconductor substrate. Further evidence of the presence of the siliconrelated traps comes from the ESR measurements carried out both at r o o m temperature and at 77 K. The spectra show a peak whose parameters (g = 2.0053 and D H = 15 G at room temperature and g = 2.0047 and D H = 10 G at 77 K) assign this peak to the presence of silicon dangling bond defects such as - S i ° surrounded by Si bonds a n d / o r partially N bonds according to ref. [13]. Several different g values are given in ref. [14]: g = 2.0061 _+ 0.0005 for --Si ° surrounded by Si bonds and g=2.0020+_0.0005 for =Si ° surrounded by N bonds, due to the specific nature of the samples, unnitrided silicon powder and fully nitrided samples. Our values are in this range. If the peak located at higher energy corresponds to the - S i - defect, it cannot be observed using this technique. The variations in the insulating layer composition and the residual hydrogen content as a function of the gas ratio were verified from the Auger electron spectroscopy and infrared analyses, respectively. A complete discussion is reported elsewhere [5]. In particular it was found that for gas ratios less than 4, silicon-rich insulating layers are deposited (for R = 1 the ratio S i / N is 2.13 and the refractive index is 3.38) whereas for gas ratios higher than 9 silicon nitride layers rich in nitrogen ( S i / N less than 0.7) are obtained. At the same
P.E. Bagnofi et al. / Electrical characteristics of SiNx /InGaAs and SiNx / S i versus insulator stoichiometry
time the S i - H bond concentration decreases ( N - H concentration increases) with increasing partial pressure of ammonia, as can be observed in fig. 5. It can also be seen that the relationship between S i / N and S i - H / N - H ratios is not linear. Specifically we point out that the relative percentage of silicon atoms bonded by hydrogen is not constant for all the gas ratios used; in particular, hydrogen preferably bonds to the excess element. On the contrary, some authors [15,16,18] have, found a constant ratio between S i / N and SiH / N - H . However, this discrepancy between the results is probably due to the different RF-power range investigated as the higher values quoted in the above references yield higher dissociation of the reactive gases. In fact, in this work a low RF power value was used in order to reduce the surface damage in the InGaAs, and under these deposition conditions, the composition of the plasma strongly depends on RF power and gas ratio. The dependence of the silicon nitride composition on the gas ratio can be explained by the kinetics of dissociation processes and the freeradical sticking coefficient values [17]. Due to the different binding energies of the H - S i l l 3 (3.8 eV) and N - N H 2 (4.8 eV) bonds, the ammonia dissociation rate is expected to be about 40% lower than the silane one. Then, since at low gas ratios (R < 4) the composition of the plasma is very poor in N H n (n = 1, 2) free radicals, silicon and all Sill,, (rn = 1, 2, 3) free radicals mainly contribute to layer growth, although with low sticking coefficients. It implies that silicon is bonded to even more than one hydrogen atom. This was confirmed by the S i - H peak value range of the IR spectra (2110-2190 cm -1) [19]. For high gas ratio (R > 9) the plasma composition becomes very rich in nitrogen and N H n free radicals with the consequent decrease of the S i / N and S i - H / N H ratios. In particular Si-Hm (m > 1) free radicals, which have low sticking coefficients, are not favoured in contributing to the layer growth. As a consequence, a higher density of silicon hydrogen-free dangling bonds can be found in the insulating layers with low silicon content, obtained using a high ammonia partial pressure. On the other hand, by comparing the results
49
shown in figs. 3 and 5, it is evident that the heights of the two peaks of the interface states increase as the S i - H bond concentration decreases. Owing to this dependence and on the basis of the above argument it is reasonable to suppose that the hydrogen atoms play a passivating role for the silicon unsaturated dangling bond defects which cause the interface states in both the SiNx/Si and SiNx/InGaAs structures. These results are in full agreement with those reported in ref. [11] for SiNx/Si interfaces in which the insulating layer has been deposited by various techniques. As far as the fixed charge is concerned, the large discrepancy between the values shown by the two types of interfaces, as reported in fig. 4, suggests that this parameter is affected also by the nature of the semiconductor substrate. The expression of the flat-band voltage VFB of a MIS structure, from which the values of fig. 4 were calculated, is given by [9]: VFB = ~MS
Qss Ci
1 f0dx N r ( x ) dx, Ci
where ~MS is the difference between the metal and semiconductor work functions, Qss is the charge stored in the interface states, d and Ci are the thickness and the capacitance of the insulator layer, respectively, and Nf (x) ( c m - 3) is the charge density across the insulating layer. The term Nf (x) includes the contributions not only from the charge density in the bulk of the insulator but also from the interface traps located in energy well below or above the semiconductor Fermi level, so that their occupation factor cannot be changed by applying an external voltage. Therefore, since the dielectric layers were deposited at the same time on both the semiconductors, it can be supposed that the different values of the fixed charge shown by the two types of samples are related to the semiconductor/insulator interfaces rather than the bulk of the dielectric layers. The following two factors can be included among the various possible causes of the difference in the fixed charge values: (i) the presence of semiconductive native oxides at the semiconductor surface and (ii) the hole traps in the
50
P.E. Bagnoli et a L / Electrical characteristics of S i Nx / I n G a A s and S i Nx / S i versus insulator stoichiometry 12
12 BEFORE ANN.
10~"
E o v
I.Ll
8
>" I--
6
Z UJ
4
a Z 0 n",
2
:7,
8
6
,,~ {,9 4 (3 1.1.1 X 2
AFTER ANN. i
200
""11
i
i
i
250
300
350
Z
0
DEPOSITION
200
TEMPERATURE
250
300
350
(°C)
Fig. 6. Plots of the fixed charge and the relative N - H bond density measured before (white dots) and after (black dots) the annealing, as a function of substrate temperature in SiNx/Si samples fabricated using fixed ammonia/silane gas ratio (R = 4) and R F power
(20 w). compound semiconductor due to the process-induced damage near the surface [4,20]. Since the measured fixed charge is a net difference between positive and negative charges trapped in the insulator, it is impossible to separate the contributions of the various defect centres. These results suggest that perhaps there is a contribution from the nitrogen dangling bonds. In fact, from the theoretical results of ref. [12] it follows that the nitrogen dangling bond = N - acts as a hole trap which is charged when empty. This trap has a localization energy just above the silicon nitride valence band and well below the silicon valence band. Due to this energy position, the states related to this defect cannot be measured by the C - V method. The negative charges stored in the empty traps partially compensate the fixed charge at the junction. This fact is mainly confirmed by the results obtained from the S i N J S i samples where the insulator layers were deposited varying the substrate temperature at fixed values of the gas ratio and the RF power. Figs. 6a and 6b show plots of the fixed charge and the N - H bond density respectively before (white dots) and after (black dots) annealing in forming gas at 300°C. In these dielectric layers, since they were deposited using the same ammonia/silane gas ratio and the same RF power, the density of the hydrogen radicals depends on the temperature of the
deposition process: higher temperature causes a lower hydrogen content and consequently a higher concentration of free nitrogen dangling bonds. The decreasing plot (with a negative concavity) of the fixed charge before annealing can be explained by the sharp decrease of the N - H bond density and hence by the increase of the = N defects. Consequently the negative charge, which compensates the positive one, is an increasing function of the deposition temperature. Since the annealing procedure induces a desorption of hydrogen and partially a recovery of S i - N bonds, both the fixed charge and the N - H bond concentration decrease. At higher temperature, that is with a more stable dielectric composition, this phenomenon is less evident. On the other hand, in the SiNx/Si and S i N J I n G a A s structures deposited using several gas ratios at a fixed deposition temperature, the density of the --N defects depends on the relative amount of nitrogen within the insulator. Then the decreasing plots of the fixed charge for both the S i N J I n G a A s and SiNx/Si samples (except for the first unbalanced gas ratio value) are due to the increasing nitrogen content and nitrogen dangling bond density as the ammonia partial pressure increases. The slight variation of the fixed charge with increasing gas ratio has to be explained by the large amount of the N H n radicals adsorbed in the dielectric layer, when R is greater than 4 (fig. 5).
P.E. Bagnoli et al. / Electrical characteristics of SiNx / I n G a A s and SiN x / S i versus insulator stoichiometry
So n i t r o g e n defects, such as = N - , w h i c h seem to c o n t r i b u t e negatively to the fixed charge o f the silicon n i t r i d e layers, c a n b e p a s s i v a t e d b y h y d r o gen atoms.
5. Conclusions T h e p r e s e n t s t u d y is m a i n l y b a s e d on the c o m p a r i s o n b e t w e e n the electrical characteristics of P E C V D silicon n i t r i d e / s e m i c o n d u c t o r M I S structures f a b r i c a t e d o n two different s e m i c o n d u c t o r s . O n the basis of the results of b o t h electrical a n d p h y s i c a l m e a s u r e m e n t s p e r f o r m e d on i n s u l a t i n g layers with different c o m p o s i t i o n s , s o m e suggestions a b o u t the origin a n d n a t u r e of the i n t e r f a c e states a n d the fixed c h a r g e can b e m a d e . I n p a r ticular the interface state d e n s i t y in b o t h t y p e s of s a m p l e s has shown the p r e s e n c e of two defect states in the s e m i c o n d u c t o r g a p which c a n be i d e n t i f i e d with the two silicon-related t r a p s theoretically c a l c u l a t e d b y R o b e r t s o n a n d Powell [12]. This h y p o t h e s i s is strongly s u p p o r t e d b y the E S R m e a s u r e m e n t s , carried o u t at r o o m t e m p e r a t u r e a n d at 77 K, which u n a m b i g u o u s l y i d e n t i f i e d silicon d a n g l i n g b o n d defects such as =Si o. T h e d e p e n d e n c e of the interface states o n the c o m p o s i t i o n of the dielectric film d e a r l y shows t h a t the p r o b l e m in m i n i m i z i n g the state d e n s i t y at the p a s s i v a t i n g i n s u l a t o r / c o m p o u n d s e m i c o n d u c tor interface m u s t b e solved t a k i n g i n t o a c c o u n t n o t o n l y the p r e - d e p o s i t i o n surface p r e p a r a t i o n , b u t also the s t o i c h i o m e t r y of the insulator, the p r o c e s s p a r a m e t e r s a n d the resulting h y d r o g e n c o n t e n t of the i n s u l a t i n g film. T h e fixed charge decreases slightly as the gas r a t i o increases, i.e. for R g r e a t e r t h a n 4. U n d e r these c o n d i t i o n s an excess o f N H n r a d i c a l s are a d s o r b e d in the dielectric layer a n d it has b e e n c o n f i r m e d that h y d r o g e n a t o m s p a s s i v a t e the uns a t u r a t e d d a n g l i n g b o n d s . E x p e r i m e n t s , at a fixed gas ratio, c o n f i r m that = N - c o u l d c o n t r i b u t e negatively to the t o t a l fixed charge. In conclusion, the electrical c h a r a c t e r i s t i c s of the two systems S i N x / S i a n d S i N x / I n G a A s were considered. It was f o u n d that the i n t e r f a c e state d e n s i t y a n d the fixed charge d e p e n d o n b o t h the silicon n i t r i d e s t o i c h i o m e t r y a n d its H content.
51
T h e low f l a t - b a n d v o l t a g e values o b t a i n e d for the I n G a A s s e m i c o n d u c t o r suggest t h a t the n a t i v e surface o x i d e s a n d s t r u c t u r a l defects significantly affect the electrical c h a r a c t e r i s t i c s o f the interface.
Acknowledgement T h e a u t h o r s are grateful to M. F e r r a r i s for the ESR measurements.
References [1] M. Renaud, P. Boher, J. Barrier, J. Schneider and J.P. Chane, in: Proc. 17th European Solid State Device Research Conf., University of Bologna (Tecnoprint, Bologna, 1987) p. 129. [2] G. Ripoche, P.H. Decor, C. Blanjot, B. Bourdon, P. Salsac and E. Duda, IEEE Electron Device Lett. EDL-6 (1985) 631. [3] P. Boher, M. Renaud, J.M. Lopez-Villegas, J. Schneider and J.P. Chane, Appl. Surf. Sci. 30 (1987) 100. [4] F. Ducroquet, G. Guillot, J.C. Renaud and A. Nouailhat, in: Proc. 20th European Solid State Device Research Conf., University of Nottingham (IOP, Bristol, 1990) p. 125. [5] A. Piccirillo and A.L. Gobbi, J. Electrochem. Soc., to be published. [6] A. Piccirillo, A.L. Gobbi, M. Ferraris, R. Giannetti and P.E. Bagnoli, Appl. Phys. Lett. 56 (1990) 1661. [7] F. Capasso and G.F. Williams, J. Electrochem. Soc. 129 (1982) 821. [8] S. Krawczyk, K. Schohe, J.Y. Longere, P. Leyral, H.L. Hartnagel and R.Z. Schutz, Le Vide, Les Couches Minces 43 (1988) 193. [9] S.M. Sze, in: Physics of Semiconductor Devices, 2nd ed. (Wiley, New York, 1981). [10] W.A. Lanford and M.J. Rand, J. Appl. Phys. 49 (1978) 2473. [11] S. Van Nguyen, J. Electrochem. Soc. 16 (1987) 275. [12] J. Robertson and M.J. Powell, Appl. Phys. Lett. 44 (1984) 415. [13] D. Jousse, J. Kanicki and J.H. Stathis, Appl. Phys. Lett. 54 (1989) 1043. [14] S. Thorp and T.G. Bushell, J. Mater. Sci. Len. 5 (1986) 1013. [15] W.A.P. Claassen, W.G.J.N. Valkenburg, F.H.P.M. Habraken and Y. Tamminga, J. Electrochem. Soc. 130 (1983) 2419. [16] G.M. Samuelson and K.M. Mar, J. Electrochem. Soc. 129 (1982) 1773. [17] M.J. Kushner, J. Appl. Phys. 63 (1988) 2532.
52
P.E. Bagnoli et al. / Electrical characteristics of S i N x / InGaAs and SiN~ / Si versus insulator stoichiometry
[18] K.M. Mar and G.M. Samuelson, Solid State Technol. 23 (1980) 137. [19] N.B. Colthup, L.H. Daly and S.E. Wiberley, in: Introduction to Infrared and Raman Spectroscopy (Academic Press, New York, 1964).
[20] W.E. Spicer, T. Kendelewicz, N. Newman, R. Cao, C. McCants, K. Miyano, I. Lindau, Z. Liliental and E.R. Weber, Appl. Surf. Sci. 33/34 (1988) 1009.
applied surface science
Electrical characteristics of silicon nitride on silicon and InGaAs as a function of the insulator stoichiometry P.E. Bagnoli a, A. Piccirillo b, A.L. G o b b i c a n d R. G i a n n e t t i a a Istituto di Elettronica e Telecomunicazioni, Unioersit& di Pisa, Via Diotisaloi 2, 56100 Pisa, Italy b Centro Studi e Laboratori di Telecomunicazioni CSELT, Via Reiss Romoli 274, 10148 Torino, Italy c Centro de Pesquisa e Desenvoloimento TELEBRAS, Campinas, Brazil
Received 7 December 1990; accepted for publication 13 May 1991
The interface state density along the semiconductorenergy gap and the fixed charge were evaluated in SiNx/InGaAs and SiNx/Si interfaces. The insulator layer was deposited by plasma-enhanced chemical vapour deposition (PECVD) using several ammonia/silane gas ratios. In both the samples the measurements revealed two main peaks of interface states whose height is a function of the insulator layer stoichiometry. Further analysis by infrared and Auger electron spectroscopyand electron spin resonance measurements enabled the peaks to be identified as the two silicon-related defects in silicon nitride cited in the literature. The nitrogen dangling bonds were found to affect the fixed charge of the structure. The role of hydrogen in passivating silicon and nitrogen dangling bonds will also be discussed.
1. Introduction Silicon nitride obtained by chemical vapour deposition is now widely used in I I I - V semiconductor device technology. This material has been found to be a promising gate insulator for MISFET transistors on compound semiconductors [1] and a passivating layer for optoelectronic devices [2]. In both the above applications the physical structure and the electrical characteristics of the insulator/semiconductor interface play an important role in determining the reliability and the efficiency of the devices. For instance, the dark current of InGaAs-based PIN photodetectors is strongly affected by the density of states at the passivating insulator/semiconductor interface [2]. Plasma-enhanced chemical vapour deposition technique (PECVD) is particularly attractive as a passivation technique due to the lack of oxygen in the deposition process, its relatively low temperature and the possibility of achieving an in-situ plasma cleaning of the semiconductor surface before the insulator layer deposition.
The electrical characteristics of the insulating layers (resistivity, breakdown strength, interface state density, fixed insulator charge) not only depend on the cleaning procedure before deposition [3] and on the process-induced damages on the semiconductor surface [4], but they are also affected by the insulator stoichiometry resulting from the choice of the deposition parameters such as the substrate temperature, the R F power and the a m m o n i a / s i l a n e gas ratio in the deposition chamber [5-7]. Therefore, in order to obtain an electrically reliable passivating layer, a better knowledge of the origin and nature of the interface states and the other unsuitable parameters of the SiNx/semiconductor interface is required. The aim of this study is the identification of the structural defects of the interface which cause the interface states and the insulator charge. We compared the results of interface state density and fixed charge evaluations in P E C V D SiNx/Si and S i N ~ / I n G a A s MIS capacitors where silicon nitride layers were deposited varying the a m m o n i a / s i l a n e
0169-4332/91/$03.50 © 1991 - Elsevier Science Publishers B.V. All fights reserved
46
P.E. Bagnoli et al. / Electrical characteristics of SiNx / In GaAs and SiNx / Si versus insulator stoichiometry
gas ratio. The other parameters, i.e. the substrate temperature and the R F power, were chosen in order to optimize the resistivity and the breakdown strength of the insulator layer and were discussed in a previous experimental investigation [51. The samples were also characterized by means of physical and structural analyses - electron spin resonance (ESR) measurements, infrared spectroscopy (IR) and Auger electron spectroscopy in order to correlate the electrical properties with the composition of the insulator layers.
capacitance meter. The energy distributions of the interface state density were obtained using the high-frequency C - V method [9]. The fixed charges were evaluated by the relative shift of the C - V curve along the voltage axis at the flat-band condition. Auger electron spectroscopy measurements were carried out on the S i N x / I n G a A s structures in order to investigate the stoichiometry of the insulating film. Furthermore, a semi-quantitative analysis of the, S i - H and N - H bond concentrations was made by using infrared spectroscopy on the SiNx/Si samples, according to the Lanford and Rand method [10].
2. Experiment 3. Results Metal-nitride-semiconductor structures were fabricated on low-doped epitaxial I n G a A s (n = 4 x 1015 cm -3) grown on highly sulfur-doped InP by liquid phase epitaxy (LPE) and on n-doped bulk silicon (100) (n = 3 x 1015 cm-3). The silicon nitride layers were deposited by PECVD using several values of the a m m o n i a / s i l a n e gas ratio in the range 2 - 2 0 and at a constant substrate temperature (300°C) and R F power (20 W). The thickness of the insulator layers was in the range 600-900 ,~. For these samples no thermal treatments were used after the insulator deposition because the annealing procedure can degrade the stoichiometry of the compound semiconductor due to the evaporation of the G r o u p V element [8]. In addition, a second series of SiNx/Si samples were fabricated varying the substrate temperature in the range 200-350°C at a fixed gas ratio ( R = 4) and annealed at 300°C for 10 min in forming gas. Aluminium dots with an area of 1.53 x 10 - 2 cm 2 were deposited on all the samples as top electrodes. On silicon an aluminum film was used as the back contact and on the compound semiconductor alloyed A u - S n . Silicon nitride layers were also deposited on silica glass substrates for the ESR measurements during the same deposition. The electrical characteristics of the interfaces were evaluated on both types of MIS samples by high-frequency (1 M H z ) c a p a c i t a n c e - v o l t a g e measurements, in the bias range from - 3 0 V to + 10 V, using an H P 4280A computer-controlled
The energy distributions o~f the interface state density as a function of the a m m o n i a / s i l a n e gas ratio ( R ) are shown in fig. 1 for the S i N x / I n G a A s samples and in fig. 2 for the SiNx/Si. The two samples show interface state densities with similar shapes and behaviour with respect to the variation of the insulating layer composition. All the curves have two main peaks near the edge of the two semiconductor bands. In the compound semiconductor the peaks are located at 0.15 and 0.65 eV, respectively, from the top of the valence band, whereas in the silicon samples the localization energies are 0.20 and 0.89 eV, respectively.
"7 >
20
•
A
//~ ~' E
R = 2.33 R=4 R=9
15
o
ca 10
5
==
,
0 u.O0
,
,
,
i
,
,
,
,
i
,
,
,
,
i
,
,
0.15 0.30 0.45 Energy from valence band
,
,
i
,
0.60 ( eV )
,
,
i
i
0.75
Fig. 1. Energy distribution of the interface state density in the SiNx/InGaAs samples for several values of the ammonia/ silane gas ratio (R).
P.E. Bagnoli et aL / Electrical characteristics of SiNx / I n G a d s and SiNx / S i versus insulator stoichiometry 10
~-" 25 ¢ ~'
20
•
R = 2.33
~
&
R=4
//
A
R=9
47
•
~'
InGaAs
~
Si
8 e~
o
~o
6
15
v-
W n-
4
t0 X 0
"6
,g
5
e~ _x
2
u.
0
UI
0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . u.o 0.2 0.4 0.6 0.8 1.0
'
'
0
1.2
5 10 AMMONIA / SILANE
15 RATIO
GAS
20
E n e r g y t r o m v a l e n c e b a n d ( eV )
Fig. 2. Energy distribution of the interface state density in the SiNx/Si samples for several values of the ammonia/silane gas ratio (R).
The effect of the variation of the gas ratio in the reactor c h a m b e r can be seen by the height of the peaks and is generally an increasing function of R, as shown in fig. 3. On the contrary, the mid-gap value of the interface state density in all the samples is quite independent of the gas ratio and presents similar values for the two types of samples (4 x 1012 c m -2 eV -1 for the c o m p o u n d semiconductor and ( 2 - 3 ) × 1012 c m -2 eV - ] for the silicon samples). The plots of the fixed charges Qf (cm - 2 ) are shown in fig. 4. The corresponding fiat-band voltages calculated f r o m the C - V curves lie in the ranges f r o m - 5 to - 3 V for the silicon samples
_2'I
>
E
• ....
t---
2O
.... = ' " "
I n G a A s left p e a k I n G a A s r i g h t p.
Fig. 4. Plots of the measured fixed charge for the two types of insulator/semiconductor interfaces.
and f r o m - 0 . 8 to - 0 . 0 1 V for the c o m p o u n d semiconductor. This fixed charge shows the opposite behaviour to the interface states as a function of the gas ratio: it slightly decreases with increasing gas ratio. Moreover, the silicon samples show an average value which is greater than the corresponding value measured in the I n G a A s samples. The results of the physical characterization of the insulating layers are summarized in fig. 5 where the silicon and nitrogen concentrations and the relative a m o u n t s of S i - H and N - H b o n d s are plotted as a function of the gas ratio. N o t e that the s t a n d a r d stoichiometry of the silicon nitride ( S i / N = 0.75) can be obtained using a value of the a m m o n i a / s i l a n e gas ratio close to 9 at which a m a x i m u m of the dielectric layer resistivity has been f o u n d [6].
Si left p. Si right p.
40i
15
C ~
•
¢0 v >-
o v >-
10
P"
5
Z 141 a .
.
_o
¢/) Z IJJ a CI Z O m
a"
.
30
Si-H N-H Si/N
.
,
5
,
,
,
,
i
,
,
,
10
AMMONIA / SILANE
,
i
,
15 GAS
,
,
,
i
i-
z, Z
20
RATIO
Fig. 3. Plots of the left-hand and the fight-hand peak heights of the interface state densities for the two types of samples as a function of the gas ratio.
0 o9
.
0
.
.
.
,
5
.
.
.
.
,
10
.
.
.
.
,
,
15
,
,%~
0
20
AMMONIA / SILANE GAS RATIO
Fig. 5. Plots of the Si-H and N-H bond densities and of the Si/N ratio in the insulator layers.
48
P.E. Bagnoli et al. / Electrical characteristics of SiNx / I n Go.As and SiNx / S i versus insulator stoichiometry
Fig. 5 shows that the composition of the insulating layer strongly depends on the a m m o n i a and silane content during the deposition process: the film was found to be silicon-rich for lower gas ratios and strongly nitrogen-rich for higher ones. Furthermore, this figure shows that the bonded hydrogen concentration in the insulating film, which generally depends on the substrate temperature during the deposition process and on the annealing procedure [11], strongly depends on the gas ratio.
4. Discussion Since the interface states and the fixed charge are related to structural defects at the i n s u l a t o r / semiconductor interface or in the bulk of the dielectric layer, the origin and nature of the electrical parameters of the silicon nitride/semiconductor interfaces can be investigated on the basis of the above experiments by correlating the electrical and physical results. By comparing figs. 1, 2 and 3, it can be seen that (i) the two peaks of the interface state density, the first one close to the conduction band edge and the second one close to the valence band edge, are present in both types of samples, regardless of the semiconductor substrate; (ii) the peak values on the two types of samples are basically of the same order of magnitude for the same deposition conditions, specially for the three lower gas ratios; (iii) the peak heights only seem to depend on the gas ratio used during the insulator layer deposition. On the basis of the above arguments it can be stated that the two peaks in the interface state density are due to structural defects which are related to the silicon nitride layer rather than to the semiconductor surface. In order to identify the origin of the two observed peaks our results can be interpreted by taking into account the theoretical analysis of the main silicon nitride defects performed in ref. [12] using the tight-binding and recursion method. For bulk Si3N 4 two possible levels of dielectric-related states caused by silicon dangling bond defects were found: a lower (-Si ° ) trapped hole level and
a higher ( - S i - ) trapped electron level, separated by a positive correlation energy of 0.66 eV. Both these two trap levels are located in energy near the mid-gap of the silicon nitride and are energetically aligned with the gap of Si. The value of the theoretical correlation energy agrees well with the experimental result obtained from the silicon samples (0.69 eV). Therefore, the two observed peaks in the SiNx/Si can be identified with the theoretical ones calculated by Robertson and Powell [12]. As previously discussed, the peaks of the interface state density of S i N x / I n G a A s samples have similar characteristics with respect to the SiNx/Si sample ones, so they can be ascribed to the same origin, that is silicon dangling bonds. The slightly lower value of the correlation energy (0.5 eV) may be induced by the different semiconductor substrate. Further evidence of the presence of the siliconrelated traps comes from the ESR measurements carried out both at r o o m temperature and at 77 K. The spectra show a peak whose parameters (g = 2.0053 and D H = 15 G at room temperature and g = 2.0047 and D H = 10 G at 77 K) assign this peak to the presence of silicon dangling bond defects such as - S i ° surrounded by Si bonds a n d / o r partially N bonds according to ref. [13]. Several different g values are given in ref. [14]: g = 2.0061 _+ 0.0005 for --Si ° surrounded by Si bonds and g=2.0020+_0.0005 for =Si ° surrounded by N bonds, due to the specific nature of the samples, unnitrided silicon powder and fully nitrided samples. Our values are in this range. If the peak located at higher energy corresponds to the - S i - defect, it cannot be observed using this technique. The variations in the insulating layer composition and the residual hydrogen content as a function of the gas ratio were verified from the Auger electron spectroscopy and infrared analyses, respectively. A complete discussion is reported elsewhere [5]. In particular it was found that for gas ratios less than 4, silicon-rich insulating layers are deposited (for R = 1 the ratio S i / N is 2.13 and the refractive index is 3.38) whereas for gas ratios higher than 9 silicon nitride layers rich in nitrogen ( S i / N less than 0.7) are obtained. At the same
P.E. Bagnofi et al. / Electrical characteristics of SiNx /InGaAs and SiNx / S i versus insulator stoichiometry
time the S i - H bond concentration decreases ( N - H concentration increases) with increasing partial pressure of ammonia, as can be observed in fig. 5. It can also be seen that the relationship between S i / N and S i - H / N - H ratios is not linear. Specifically we point out that the relative percentage of silicon atoms bonded by hydrogen is not constant for all the gas ratios used; in particular, hydrogen preferably bonds to the excess element. On the contrary, some authors [15,16,18] have, found a constant ratio between S i / N and SiH / N - H . However, this discrepancy between the results is probably due to the different RF-power range investigated as the higher values quoted in the above references yield higher dissociation of the reactive gases. In fact, in this work a low RF power value was used in order to reduce the surface damage in the InGaAs, and under these deposition conditions, the composition of the plasma strongly depends on RF power and gas ratio. The dependence of the silicon nitride composition on the gas ratio can be explained by the kinetics of dissociation processes and the freeradical sticking coefficient values [17]. Due to the different binding energies of the H - S i l l 3 (3.8 eV) and N - N H 2 (4.8 eV) bonds, the ammonia dissociation rate is expected to be about 40% lower than the silane one. Then, since at low gas ratios (R < 4) the composition of the plasma is very poor in N H n (n = 1, 2) free radicals, silicon and all Sill,, (rn = 1, 2, 3) free radicals mainly contribute to layer growth, although with low sticking coefficients. It implies that silicon is bonded to even more than one hydrogen atom. This was confirmed by the S i - H peak value range of the IR spectra (2110-2190 cm -1) [19]. For high gas ratio (R > 9) the plasma composition becomes very rich in nitrogen and N H n free radicals with the consequent decrease of the S i / N and S i - H / N H ratios. In particular Si-Hm (m > 1) free radicals, which have low sticking coefficients, are not favoured in contributing to the layer growth. As a consequence, a higher density of silicon hydrogen-free dangling bonds can be found in the insulating layers with low silicon content, obtained using a high ammonia partial pressure. On the other hand, by comparing the results
49
shown in figs. 3 and 5, it is evident that the heights of the two peaks of the interface states increase as the S i - H bond concentration decreases. Owing to this dependence and on the basis of the above argument it is reasonable to suppose that the hydrogen atoms play a passivating role for the silicon unsaturated dangling bond defects which cause the interface states in both the SiNx/Si and SiNx/InGaAs structures. These results are in full agreement with those reported in ref. [11] for SiNx/Si interfaces in which the insulating layer has been deposited by various techniques. As far as the fixed charge is concerned, the large discrepancy between the values shown by the two types of interfaces, as reported in fig. 4, suggests that this parameter is affected also by the nature of the semiconductor substrate. The expression of the flat-band voltage VFB of a MIS structure, from which the values of fig. 4 were calculated, is given by [9]: VFB = ~MS
Qss Ci
1 f0dx N r ( x ) dx, Ci
where ~MS is the difference between the metal and semiconductor work functions, Qss is the charge stored in the interface states, d and Ci are the thickness and the capacitance of the insulator layer, respectively, and Nf (x) ( c m - 3) is the charge density across the insulating layer. The term Nf (x) includes the contributions not only from the charge density in the bulk of the insulator but also from the interface traps located in energy well below or above the semiconductor Fermi level, so that their occupation factor cannot be changed by applying an external voltage. Therefore, since the dielectric layers were deposited at the same time on both the semiconductors, it can be supposed that the different values of the fixed charge shown by the two types of samples are related to the semiconductor/insulator interfaces rather than the bulk of the dielectric layers. The following two factors can be included among the various possible causes of the difference in the fixed charge values: (i) the presence of semiconductive native oxides at the semiconductor surface and (ii) the hole traps in the
50
P.E. Bagnoli et a L / Electrical characteristics of S i Nx / I n G a A s and S i Nx / S i versus insulator stoichiometry 12
12 BEFORE ANN.
10~"
E o v
I.Ll
8
>" I--
6
Z UJ
4
a Z 0 n",
2
:7,
8
6
,,~ {,9 4 (3 1.1.1 X 2
AFTER ANN. i
200
""11
i
i
i
250
300
350
Z
0
DEPOSITION
200
TEMPERATURE
250
300
350
(°C)
Fig. 6. Plots of the fixed charge and the relative N - H bond density measured before (white dots) and after (black dots) the annealing, as a function of substrate temperature in SiNx/Si samples fabricated using fixed ammonia/silane gas ratio (R = 4) and R F power
(20 w). compound semiconductor due to the process-induced damage near the surface [4,20]. Since the measured fixed charge is a net difference between positive and negative charges trapped in the insulator, it is impossible to separate the contributions of the various defect centres. These results suggest that perhaps there is a contribution from the nitrogen dangling bonds. In fact, from the theoretical results of ref. [12] it follows that the nitrogen dangling bond = N - acts as a hole trap which is charged when empty. This trap has a localization energy just above the silicon nitride valence band and well below the silicon valence band. Due to this energy position, the states related to this defect cannot be measured by the C - V method. The negative charges stored in the empty traps partially compensate the fixed charge at the junction. This fact is mainly confirmed by the results obtained from the S i N J S i samples where the insulator layers were deposited varying the substrate temperature at fixed values of the gas ratio and the RF power. Figs. 6a and 6b show plots of the fixed charge and the N - H bond density respectively before (white dots) and after (black dots) annealing in forming gas at 300°C. In these dielectric layers, since they were deposited using the same ammonia/silane gas ratio and the same RF power, the density of the hydrogen radicals depends on the temperature of the
deposition process: higher temperature causes a lower hydrogen content and consequently a higher concentration of free nitrogen dangling bonds. The decreasing plot (with a negative concavity) of the fixed charge before annealing can be explained by the sharp decrease of the N - H bond density and hence by the increase of the = N defects. Consequently the negative charge, which compensates the positive one, is an increasing function of the deposition temperature. Since the annealing procedure induces a desorption of hydrogen and partially a recovery of S i - N bonds, both the fixed charge and the N - H bond concentration decrease. At higher temperature, that is with a more stable dielectric composition, this phenomenon is less evident. On the other hand, in the SiNx/Si and S i N J I n G a A s structures deposited using several gas ratios at a fixed deposition temperature, the density of the --N defects depends on the relative amount of nitrogen within the insulator. Then the decreasing plots of the fixed charge for both the S i N J I n G a A s and SiNx/Si samples (except for the first unbalanced gas ratio value) are due to the increasing nitrogen content and nitrogen dangling bond density as the ammonia partial pressure increases. The slight variation of the fixed charge with increasing gas ratio has to be explained by the large amount of the N H n radicals adsorbed in the dielectric layer, when R is greater than 4 (fig. 5).
P.E. Bagnoli et al. / Electrical characteristics of SiNx / I n G a A s and SiN x / S i versus insulator stoichiometry
So n i t r o g e n defects, such as = N - , w h i c h seem to c o n t r i b u t e negatively to the fixed charge o f the silicon n i t r i d e layers, c a n b e p a s s i v a t e d b y h y d r o gen atoms.
5. Conclusions T h e p r e s e n t s t u d y is m a i n l y b a s e d on the c o m p a r i s o n b e t w e e n the electrical characteristics of P E C V D silicon n i t r i d e / s e m i c o n d u c t o r M I S structures f a b r i c a t e d o n two different s e m i c o n d u c t o r s . O n the basis of the results of b o t h electrical a n d p h y s i c a l m e a s u r e m e n t s p e r f o r m e d on i n s u l a t i n g layers with different c o m p o s i t i o n s , s o m e suggestions a b o u t the origin a n d n a t u r e of the i n t e r f a c e states a n d the fixed c h a r g e can b e m a d e . I n p a r ticular the interface state d e n s i t y in b o t h t y p e s of s a m p l e s has shown the p r e s e n c e of two defect states in the s e m i c o n d u c t o r g a p which c a n be i d e n t i f i e d with the two silicon-related t r a p s theoretically c a l c u l a t e d b y R o b e r t s o n a n d Powell [12]. This h y p o t h e s i s is strongly s u p p o r t e d b y the E S R m e a s u r e m e n t s , carried o u t at r o o m t e m p e r a t u r e a n d at 77 K, which u n a m b i g u o u s l y i d e n t i f i e d silicon d a n g l i n g b o n d defects such as =Si o. T h e d e p e n d e n c e of the interface states o n the c o m p o s i t i o n of the dielectric film d e a r l y shows t h a t the p r o b l e m in m i n i m i z i n g the state d e n s i t y at the p a s s i v a t i n g i n s u l a t o r / c o m p o u n d s e m i c o n d u c tor interface m u s t b e solved t a k i n g i n t o a c c o u n t n o t o n l y the p r e - d e p o s i t i o n surface p r e p a r a t i o n , b u t also the s t o i c h i o m e t r y of the insulator, the p r o c e s s p a r a m e t e r s a n d the resulting h y d r o g e n c o n t e n t of the i n s u l a t i n g film. T h e fixed charge decreases slightly as the gas r a t i o increases, i.e. for R g r e a t e r t h a n 4. U n d e r these c o n d i t i o n s an excess o f N H n r a d i c a l s are a d s o r b e d in the dielectric layer a n d it has b e e n c o n f i r m e d that h y d r o g e n a t o m s p a s s i v a t e the uns a t u r a t e d d a n g l i n g b o n d s . E x p e r i m e n t s , at a fixed gas ratio, c o n f i r m that = N - c o u l d c o n t r i b u t e negatively to the t o t a l fixed charge. In conclusion, the electrical c h a r a c t e r i s t i c s of the two systems S i N x / S i a n d S i N x / I n G a A s were considered. It was f o u n d that the i n t e r f a c e state d e n s i t y a n d the fixed charge d e p e n d o n b o t h the silicon n i t r i d e s t o i c h i o m e t r y a n d its H content.
51
T h e low f l a t - b a n d v o l t a g e values o b t a i n e d for the I n G a A s s e m i c o n d u c t o r suggest t h a t the n a t i v e surface o x i d e s a n d s t r u c t u r a l defects significantly affect the electrical c h a r a c t e r i s t i c s o f the interface.
Acknowledgement T h e a u t h o r s are grateful to M. F e r r a r i s for the ESR measurements.
References [1] M. Renaud, P. Boher, J. Barrier, J. Schneider and J.P. Chane, in: Proc. 17th European Solid State Device Research Conf., University of Bologna (Tecnoprint, Bologna, 1987) p. 129. [2] G. Ripoche, P.H. Decor, C. Blanjot, B. Bourdon, P. Salsac and E. Duda, IEEE Electron Device Lett. EDL-6 (1985) 631. [3] P. Boher, M. Renaud, J.M. Lopez-Villegas, J. Schneider and J.P. Chane, Appl. Surf. Sci. 30 (1987) 100. [4] F. Ducroquet, G. Guillot, J.C. Renaud and A. Nouailhat, in: Proc. 20th European Solid State Device Research Conf., University of Nottingham (IOP, Bristol, 1990) p. 125. [5] A. Piccirillo and A.L. Gobbi, J. Electrochem. Soc., to be published. [6] A. Piccirillo, A.L. Gobbi, M. Ferraris, R. Giannetti and P.E. Bagnoli, Appl. Phys. Lett. 56 (1990) 1661. [7] F. Capasso and G.F. Williams, J. Electrochem. Soc. 129 (1982) 821. [8] S. Krawczyk, K. Schohe, J.Y. Longere, P. Leyral, H.L. Hartnagel and R.Z. Schutz, Le Vide, Les Couches Minces 43 (1988) 193. [9] S.M. Sze, in: Physics of Semiconductor Devices, 2nd ed. (Wiley, New York, 1981). [10] W.A. Lanford and M.J. Rand, J. Appl. Phys. 49 (1978) 2473. [11] S. Van Nguyen, J. Electrochem. Soc. 16 (1987) 275. [12] J. Robertson and M.J. Powell, Appl. Phys. Lett. 44 (1984) 415. [13] D. Jousse, J. Kanicki and J.H. Stathis, Appl. Phys. Lett. 54 (1989) 1043. [14] S. Thorp and T.G. Bushell, J. Mater. Sci. Len. 5 (1986) 1013. [15] W.A.P. Claassen, W.G.J.N. Valkenburg, F.H.P.M. Habraken and Y. Tamminga, J. Electrochem. Soc. 130 (1983) 2419. [16] G.M. Samuelson and K.M. Mar, J. Electrochem. Soc. 129 (1982) 1773. [17] M.J. Kushner, J. Appl. Phys. 63 (1988) 2532.
52
P.E. Bagnoli et al. / Electrical characteristics of S i N x / InGaAs and SiN~ / Si versus insulator stoichiometry
[18] K.M. Mar and G.M. Samuelson, Solid State Technol. 23 (1980) 137. [19] N.B. Colthup, L.H. Daly and S.E. Wiberley, in: Introduction to Infrared and Raman Spectroscopy (Academic Press, New York, 1964).
[20] W.E. Spicer, T. Kendelewicz, N. Newman, R. Cao, C. McCants, K. Miyano, I. Lindau, Z. Liliental and E.R. Weber, Appl. Surf. Sci. 33/34 (1988) 1009.