Influence of nitrogen doping on photoconductivity properties of a: DLC films

Diamond and Related Materials 6 ( 1997) 1868-1873

ELSEVIER

Influence of nitrogen doping on photoconductivity c films properties of a: L. Klibanov a,*, N. Croitoru a, A. Seidman a, L. Scheffer b, E. Ben-Jacob b a Department of Physical Electronics, Faculty of Engineering, Tel-Aviv University, Ramat-Aviv, Israel ’ School qf Physics and Astronomy, The Raymond and Beverly Sackler Faculty of Exact Sciences, Tel-Aviv University. Ramat-Aviv 69978, Israel Received 21 March 1997; accepted 7 July 1997

Abstract We present photoconductivity, photosensitivity and decay time of photocurrent measured as a function of temperature for both nitrogen-doped and undoped a:DLC films. The a:DLC films were deposited using radio-frequency (RF) glow discharge of methane gas (CH,) as a source of carbon. Several fihns were doped employing nitrogen (N,) as the doping gas. The doped and undoped a:DLC films have shown photoconductivity effects in a wide range of temperatures. All photoconductivity parameters, i.e. spectral response, photosensitivity, decay time and photocurrent, were measured for both undoped and doped iilms. The maximum spectral photosensitivity of doped films shifts to a higher energy, similar to the optical energy-gap measurements. The photocurrent of the doped film is larger by two orders of magnitude than that of undoped film, while the photosensitivity shows an opposite effect. The mobility of doped films (2.43 x 10e5) is larger by two orders of magnitude than that of undoped films (5.64 x lo-‘) at room temperature. In order to provide nanoscale information about rhe morphological properties of the undoped a:DLC films surface, we have used atomic force microscopoy ( AFM ). It was found that the roughness of our films increased with increasing thickness of the films, from 0.3 to 2.5 brn. 0 1997 Elsevier Science S.A. Keywords: Atomic force microscopy: Diamond-like carbon; Doping; Photoconductvity

1. Introduction Hard hydrogenated

amorphous carbon (a-C:H) films, known as a:DLC films, have unique properties such as chemical inertness, high infrared transparency, microhardness and electrical resistivity [l]. Since the photoconductivity of a:DLC films is small, this property has not yet been thoroughly investigated. Recent studies [2] of highly tetrahedral (hydrogen-free) diamond-like amorphous carbon (a-C), have shown a change of about two orders of magnitude of the electrical conductivity under light exposure. It seems that this photoconductivity is related to the high sp3:sp2 ratio in these hydrogenfree a-C films. In Ref. [ 31, the onset of photoconductivity in hydrogenated a-C:H films was observed only for certain deposition conditions. A maximum photosensitivity ratio of 6.5 is reported in this paper. In our previous studies [4], the influence of iodine doping on the photoconductivity properties of a:DLC films was investigated. We found that iodine doping led to an * Corresponding author. 0925-9635/97/$17.00 0 1997 Elsevier Science S.A. All rights reserved. PII SO925-9635(97)00153-2

increase in photoconductivity by about one order of magnitude, At the same time, photosensitivity ratio at room temperature is still less than one. Obviously, a good photoconductivity response of a-C:H materials enhances their range of application in the field of photodetection devices. Scanning tunneling (STM) and atomic force (AFM) microscopy techniques have been increasingly used for observation and measurement oi nanoscale topographic features of the DLC films surface [5] since they provide accurately quantitative information about the surface structure which is not accessible by any other technique. AFM measurements of diamond films, grown on a wide variety of substrates (for example, silicon or natural diamond substrates), were performed 161. In this paper, we have correlated nanoscale surface structure with the photoelectrical properties of films. The aim of the present paper is to investigate the influence of nitrogen doping gas and surface structure on the photoconductivity parameters of a:DLC films. Firstly, we present AFM data. We found that there is a threshold of a:DLC films thickness which determines a drastic increase in rough-

L. Klibarlov et al. / Diawondund Related Mrreriuls 6 ( 1997) 1868-1873

I869

ness. Then we connect this effect with changes of the carrier drift mobility.

2. Experimental set-up LJndopeJ amorphous diamond-like carbon (a:DLC) films were deposited by the RF glow discharge technique using methane gas, CHd, as the carbon source [7,8]. The films (thicknesses of about 0.3, 0.6 and 1.5 pm) were deposited on Corning, Avon, France, (7056) glass substrate. Doping was performed by adding nitrogen gas at partial pressure of 20% to the CH, during film deposition. The thickness of nitrogen-doped samples (N-DLC) was about 0.3 pm. The films’ surface images were obtained using a selfmade contact AFM. The AFM was operated in air, at room temperature, and was equipped with a scanner that allowed a maximum scan range of 1.7 x 1.7 pm and a four-quadrant photodiode as the position sensitive detector (PSD). The probe was a silicon nitride cantilever with an integrated pyramidal tip; the spring constant of the cantilever was 0.12 N/m and the length 200 pm. The images were obtained in the constant force mode, subjected to computer processing and are displayed in three-dimensional modes on the scale of 0.7 x 0.7 pm. Electrical contacts were obtained using a finger planar gap-cell structure consisting of 10 pm wide Al contacts and a spacing of 10 pm between contacts, patterned on the a:DLC films, using photolithography. The currenttemperature (I-T), current-voltage (I-V). phorocurrent-temperature (I,,-T) and decay time (r,) of photocurrent characteristics were measured on the same films. The photoconductivity measurements were carried out using He-Ne and Ar lasers of 5 mW emitting power as the light sources. The spectral photoresponse of the a:DLC film was measured in wavelengths interval of 400-800 nrn. The source of radiation in this case was a halogen lamp in conjunction with monochromator. The spectral resolution of the system was 10 nm. In order to avoid the surface heating from lasers and to detect small signals, the incident light was chopped at a frequency of 90 Hz and the AC component of the photocurrent response was measured using a lock-in amplifier. Measurements of decay time (Q) of photoconductivity were performed using an acousto-optic-modulator (AOM) and argon (488 nm) laser as pulse light source.

3.1. AFM In Fig. 1, we show the surface topography of an a:DLC film of 0.3 pm thickness. The surface is rather

Fig. I. AFM image of undoped DLC: film thickness. 0.3 pm: image size. 0.7 x 0.7 pm.

Fig. 1. AFM image of undoped a:DLC: film-thickness, 1.5 pm; image size. 0.7 r: 0.7 pm.

flat with a surface corrugation of about 0.3 nm. The topographic image consists of many hillocks varying between 20-30 nm in diameter. In Fig. 2, we show the topography of a 1.5 pm thick film. The picture is significantly different from the two previous images. The structure is composed of large clusters and the surface roughness is up to 10 nm. Several measurements were also performed on the a:DLC films of a thickness of about 0.01 pm. The dependence of the surface mean roughness on the deposited a:DLC film thickness is shown in Fig. 3. It can clearly be seen that the increase in the thickness of the films results in a steep increase in the value of the roughness for thickness values higher than 0.6 pm. Since the roughness of the Corning glass substrate is about I-2 pm, the smaller value of the observed roughness of the films is attributed to a deposition mechanism consisting of filling the substrate’s grooves and valleys, which “smoothes” the surface. We believe that in the case of a 1.5 pm thick film, the higher value of the surface corrugation is correlated with the increased value

b

a:DLC, 0.6 pm

(-

a:DLC, 3.3 p N-DLC, 0.3 pm

1o-10

:’

lo-” 0

103

102

IO'

100

r 2

I

3

4

Thickness (pm) Fig. 3. Roughness as a function of films thickness.

of the growing time and not to the influence of the substrate topography. Therefore, the surface topography of the substrate exerts a determinant influence on the deposited film roughness up to a thickness of about 0.3 urn. 3.2. Electrical measurements (I-- V. I-T) The I-V characteristics of both doped and undoped films show linear dependence irrespective of their thickness, which indicates ohmic contact between the a:DLC films and aluminum. The dark conductivity increases with increasing temperature {Fig. 4) in agreement with the Mott relationship b=cro exp(EJkT), which assumes an activation of carriers from localized to extended states. From the I-T measurements, we found that the activation energy, E,, is about 0.26-0.31 eV for a hightemperature range (3 K 1000/T< 4.5) and E, = 0.090.14 eV for a low temperature range (6~ lOOO/T<7.5) in the case of undoped films. The activation energy of the doped films is about E, = 0.17 eV and is independent of temperature. The a:DLC films seem to be of p-type due to structural defects in the amorphous carbon network [9]. If the undoped a:DLC film is considered Fermi level at as p-type, with a defect-pinned 0.26-0.31 eV, it results in the addition of N into the carbon lattice initially compensating for and/or passivating the p-type defect states. A further increase of N doping produces a shift in the Fermi level towards the conduction band; therefore, a reduction of activation energy and resistivity [lo] appears. In accordance with these facts, it is possible to obtain an n-type nitrogendoped a:DLC film and/or a weak p-type undoped a:DLC film for which a photoconductivity study was performed.

5

6

7

8

1000/T [l/K] Fig. 4. I-T characteristics of undoped and doped films, with different thicknesses for temperatures between 323 and 133 K.

3.3. Pizotocurrent (I,,,) nwusurements (I,,,-T, I,,,4 I,,,-T) In Fig. 5, the steady-state photocurrent (I& of an undoped film versus temperature (1000/T), for five values of intensity, given by an Ar laser (488 nm) source can been seen. The nitrogen-doped film shows the same characteristics: however, the value of Iph is larger by two orders of magnitude compared with the undoped films (Fig. 6).

.a

: Dark

-.->._..

I ,c-ICI 3

--- 0.5w

I

I

I

I

4

5

6

7

1OKKIiT [IIK] Fig. 5. Photocurrent versus 1000/T for several values of incident light power for an undoped film with thickness of 0.3 pm. An argon laser was used as the light source.

L. Klihunov ct ul. j Dimnomi und Rdutd 106

Matwids 6

( 1947) 1X68-1873

1871

102

N-DLC, 0.3 pm a:DLC, 0.6 pm -(~&- a:DLC, 0.3 pm IO’ _ -* a:DLC, 1.5 pm

, fJ-+’

-A

1I c.

1o-7 3 E 2 5 IO”

.‘-I..

IO-2

1 mWatt i..

-..-.__.._ 0.5 mWatt

IO’ 3

,

I

,

4

5

6

103 7

8

1000/K [l/K]

3

I

I

4

5

6

1000/T [l&q Fig. 6. Photocurrent versus lOOO!T for several values of incident light power for a doped film with thickness of 3000 A. An argon laser was used as the light source.

For all temperature intervals, we observed a linear dependence of photocurrent on the intensity of incident light. This observation is in agreement with the process of monomolecular recombination (Fig. 7). The nitrogen doping does not change this behavior, but it might indicate that the recombination process in a:DLC films appears through recombination centers within the energy gap, and cannot be compensated by nitrogen doping. Jn order to estimate the influence of doping on

N-DLC,

0.3 pm

y=o.94

-%

/id

*‘EY .’

I/

XDLC

1.:

4

0.3 pm. ym0.99

3

1.6 pm. y=o.95

a

0.6 pm, y=O.98

, 1

10 lnsident Power pVatt]

Fig. 7. Photocurrent versus incident light power for undoped and doped films at a temperature of 300 K. Also shown is the ;*-slope of these characteristics.

Fig. 8. Photosensitivity versus temperature for several thickness values of undoped and doped a:DLC films.

the photoconductivity parameters, we considered that the photosensitivity is defined as the ratio I,,&, where Id is the dark current. Fig. 8 shows that photosensitivity, in case of nitrogen-doped films, decreases by two orders of magnitude compared with the undoped films. In order to understand the photoconductivity mechanism of a:DLC (doped and undoped) films, a similar model may be assumed to that given in Ref. [2], considering !hat the photogenerated current is predominantly cairied by electrons in the extended states of the conduction band. In this case, incident light is exciting electrons from the extended states in the valence band or from filled localized tail states near the valence band edge. The spectral response of photoconductivity may reveal information regarding the values of energy or/and mobility gap [ 11J. Photoresponse (I& W) spectral dependence of undoped and doped films at 300 K was also measured and is shown in Fig. 9. The photoresponse spectrum curve for undoped samples of different thicknesses show a pronounced peak at 450-500 nm. corresponding to photon energies between 2.5 and 2.8 eV. The doped films also shown a peak in the photoresponse spectrum in the same region, but shifted to a higher photon energy. A similar shift, but for an optical band gap. was also found from 1.l- 1.3 eV for undoped and nitrogen-doped films. respectively. in our previous papers [8.4]. Consequently, these results have shown that the mobility gap is larger than that of the optical band gap (1.1-1.5 eV) for a:DLC films. Fig. 10 shows the decay time of the photocurrent as a function of temperature (T) for doped and undoped a:DLC films. At high temperatures ( lOOO/T<4.5) for undoped films. the decay time, rd. increases rapidly with increasing

high-temperature region (I,, < Zd), the possibility of recombination increases together with increased the free carrier concentration in the extended states; therefore, this decreases ZP,,and changes the characteristics of the decay time versus temperature. --b a:DLC, 0.6 pm --Ct - a:DLC, 0.3 pm u a:DLC, 1.5 pm

3.4. Detertnhation

of mobility vulue

The carrier drift mobility was determined from the steady-state photocurrent and transient photocurrent decay time measurements [ 141. The steady-state photocurrent is given as: Zr,, =eZ1drdlZG(1 -e-“d)(

I

I

I

1.6

1.8

2.0

1

2.2

/

/

I

2.6

2.8

3.0

1

2.4

Photon Energy (ev) Fig. 9. Photocurrent spectral dependence of undoped films of several thicknesses and doped films at 300 K.

1000/Z. At low temperatures, (4.5> 1000/T), rd decreases with increasing l/Z’. A similar dependence of rd on T was observed for a:Si, a:Ge [ 12,131 and a:DLC films [4]. The addition of nitrogen as the doping element drastically changes the shape of the rd(T) dependence. In Fig. 10. it can be seen that the rd of the nitrogendoped film increases with decreasing temperature and reaches a maximum only at 1000/Z’= 7.5. This behavior of td can be interpreted by the shift of the point where the dark current is larger than photucurrent to a lower temperature due to nitrogen doping (Fig. 0 1. At the

1 -R)E.lV-d,

(1)

where e is electron charge, /id is the drift mobility of excited carriers, r~ is the quantum efficiency, G is the generation rate of photons, c( is the optical absorption coefficient [ 151, ti is the thickness of the films, R is the front surface reflectance, II*is the width of the films, and E is the electrical field. For the radiation of the Argon laser (Zzv=2.5 eV), the quantum efficiency q may be considered as one. The drift mobility of doped and undoped films may be calculated from Eq. ( 1) using the known Z,,,,, td, and the light intensity of the laser. The drift mobility for the undoped and doped films, as a function of temperature, is shown in Fig. 11. In the case of amorphous materials, the drift mobility may be influenced by occasional trapping on states where the carriers have very low mobility [ 16,111. According to [ 161, we can write for the high(31 lO”/T<5) and low- (5 5 103/T<7) temperature

T cl

e:OLC. 1.5pm

n

e:DLC. 0.6 pm

t)

N-DLC a:DLC, 0.3 pm

a:DLC, 0.3)1m

2

a:DLC, 0.6 pm

100 a 1 E z %

10

1 2

/

I

1

3

4

5

N-DLC. 0.3 pm I I

6

7

i

.-

8

9

1000/K [l/K]

Fig. IO. Decay time of photoconductivity versus temperature for N-DLC and a:DLC films. An argon laser was used as the light source.

3

4

5

6

7

IOOOrr [In<] Fig. I I. Drift mobility of doped and undoped a:DLC films as functions of temperature.

L. Klibunov et cd. I Diurnondmd

regions, respectively, the relationships:

1873

(2) The photocurrent

f2)

!ld = /lhop *

Related Materials 6 (1997) 1868-1873

(3)

where lu.,, and phop are mobility in the extended states and hopping mobility, respectively. Using Eq. (2) for the high-temperature region of mobility as a function of temperature characteristics and assuming that the temperature dependence of ,K_, is much weaker than exponential, we determined that activation energy is E,, =O.12 eV for doped films and 0.13-0.24 eV for undoped films. These values of activation energy are near those of the high-temperature region (lOOO/T<4.5) of the rd characteristics (Fig. 10). The absolute value of the room-temperature mobility is 2.43 x lo-’ cm’/V s for doped and 5.64 x lo-’ cm2/V s for undoped films. Direct measurements of the drift mobility for the a:DLC films gave pd= 1.0~ 8.0 x 10e6 cm’/V s at room temperature [ 171 for undoped films. It is evident that this difference is a result of the quite different structure of a:DLC films. In order to explain difference in p(T) characteristics between a:DLC and N-DLC films we may assume that nitrogen doping compensates dangling bonds in the energy gap and at the same time increases the density of states in the conduction band, N,. This may be based on our earlier optical investigations into nitrogen-doped films 141, where it was shown that nitrogen increases the optical energy gap. Furthermore, the density of states in the conduction band increases [ 161 which may lead to larger value of ~1~ in Eq. (2). However, all these assumptions require further investigation. The decrease in the drift mobility of undoped films might be correlated with the increase in the roughness value, which leads to a higher concentration of defects. As it can seen in Fig. 3, the roughness increases drastically when the thickness increases from 0.6 to 1.5 pm. This result may suggest that the surface structure can affect the transport parameters of a:DLC films [ 181.

4. Conclusions. (1) The nitrogen-doped and undoped a:DLC films showed photoconductivity effects over a wide range of temperatures.

of doped films is larger than that of undoped films by two orders of magnitude, but the photosensitivity displays opposite effect. (3) The mobility of doped films is larger by two order; of magnitude than undoped films at room temperature. The maximum of the spectral photosensitivity in the case of doped films shifted to a higher energy, similar to the optical energy gap measurements. and mobility (4) The decay times of photocurrent depend on the microstructure of the film surface. By increasing the thickness from 0.3 to 1.5 pm and the mean roughness of the surface from 0.38 to 1.4 nm, the mobility decreases from 5.64 x lo-’ to 2.57 x lo-’ cm2/V s.

References [I] Proceedings of the Applied Diamond Conference 1995. USA, August 1995.

NIST,

[?I G.A. Amaratunga, VS. Veerasamy. W.I. Miine, C.A. Davis. S.R.P. Silva, H.S. MacKenzie, Appl. Phys. Lett. 63 ( 1993 ) 370. [3] P.N. Dixit, S. Kumar. D. Saran& R. Bhattacharyya. Solid State Commun. 90 (1994) 421. [4] L. Klibanov, N. Croitoru. A. Seidman. Proc. 19th Conv. or IEEE. Israel. 271-273, 1996. [.5] 1. Rusman. L. Klibanov, L. Burstein. Y. Rosenberg. V Weinstein. E. Ben-Jacob. N. Croitoru, A. Scidman. Thin Solid Films ?S7 (1996) 36. [6] 9. Bhushan. V.V. Subramaniam, B.K. Gupta, Diamond Films Technol. 4 ( 1994) 71. [7] M. Alaluf. J. Appelbaum, L. Klibanov. D. Brinker. D. Scheiman, N. Croitoru. Thin Solid Films 2.56 (1995) I. [8] L. Klibanov. M. Alaluf. A. Seidman. N. Croitoru. Amorphous diamond-like carbon films as a new material for electron devices, Proc. 18th Conv. of IEEE. Israel. 1995. 191 P. Blaudeck. T. Frauenheim. H.G. Busmann. T. Lilt. Phys. Rev. B 49 (1994) 11409. [lo] V.S. Veerasamy, J. Yuan. G.A.J. Amaratunga, W.I. Mihx K.W.R. Gilkes, M. Weiler. L.M. Brown, Phys. Rev. B 48 (1993) 17954. [I I] T.D. Moustakas, W. Paul, Phys. Rev. B I6 ( 1977) 1564. [I?] W. Fuhs. D. Meyer. Phys. Stat. Sol. A 24 (i974) 275. [ 131 C. Main, A. Owen, Proc. 5th International Conf. on Amorphous and Liquid Semiconductors, 2 ( 1974) 783. [ 141 H. Araki, T. Hanawa, Thin Solid Films 169 ( 1989) 187. [ 151 M. Allon-Alaluf, L. Klibanov, N. Croitoru, A. Seidman. Diamond Relat. Mater. 5 (1996) 1497. [l6] A. Madan. M. Shaw. The Physics and Applications of Amorphous Semiconductors. Academic Press. England. 1988. [ 171 W. Mycielski. E. Staryga. A. Lipinski. Thin Solid Films 735 (1993) 13. [IS] D.R. McKenzie, Y. Yin. N.A. Marks. C.A. Davis. E. Kravtchinskaia. B.A. Pailthorpe. G.A.J. Amaratunca. J. Non-Cryst. Solids lh4165166(1993)

1101.