C:H films prepared by DC unbalanced magnetron sputtering

C:H films prepared by DC unbalanced magnetron sputtering

Thin Solid Films 351 (1999) 151±157 Composite germanium/C:H ®lms prepared by DC unbalanced magnetron sputtering H. Biederman a,*, V. StundzÏia, a, D...

2MB Sizes 2 Downloads 36 Views

Thin Solid Films 351 (1999) 151±157

Composite germanium/C:H ®lms prepared by DC unbalanced magnetron sputtering H. Biederman a,*, V. StundzÏia, a, D. SlavõÂnska a, J. ZÏalman, a, J. PesÏicÏka, a, M. VaneÏcÏek b, J. Zemek, b, W. Fukarek c a

Faculty of Mathematics and Physics, Charles University, Prague, Czech Republic b Institute of Physics, Academy of Sciences, Prague, Czech Republic c Institute of Ion Beam Physics and Materials Research, FZ Rossendorf, Germany

Abstract Composite Ge/C:H ®lms (germanium doped hard plasma polymer (C:H)) have been deposited using unbalanced planar magnetron equipped with germanium/graphite target and operated in argon/n-hexane gas mixture. The composition of the deposited ®lms was determined by Rutherford back scattering (RBS), elastic recoil detection (ERD) and X-ray photoelectron spectroscopy (XPS) analytical methods. Contents of germanium from 0 up to 30 at.% was con®rmed with rather homogenous distribution of germanium through the cross-section of the composite ®lms. Transmission electron microscopy (TEM) investigation of the samples revealed that germanium forms clusters with a maximum diameter of 2 nm embedded in C:H and GeC alloy matrix. An optical gap ranging from 1.9 to 1.0 eV with corresponding refractive index ranging from 2 to 3 were determined. DC electrical properties were measured in the planar electrodes -composite ®lm-con®guration. The electrical conduction is strongly dependent on the germanium content and on the substrate temperature. Current-voltage characteristics are linear at low electrical ®eld and become superlinear at higher ®eld. q 1999 Elsevier Science S.A. All rights reserved. Keywords: Plasma processing and deposition; Composite; Germanium; Carbon

1. Introduction Composite ®lms metal/hard carbon prepared by DC unbalanced magnetron sputtering became the object of attention in recent years [1±4]. Composites of metals, such as Ag, Ni, Mo, with hard carbon (C:H hard plasma polymer) were prepared and their structure, optical and electrical properties described. The same technique can be used for the preparation of composites where the metal is replaced by a semiconductor or even an insulator. Composites of Si and Ge with a-C:H prepared by RF magnetron sputtering have also been recently investigated [5±7]. Composites of Ge that in the past generally received less attention than those with Si were prepared in several studies by plasma polymerization processes using organogermanium compounds [8±12]. In this paper we focused our attention to the preparation and the basic physical characterization of Ge/C:H composite ®lms prepared by DC unbalanced magnetron sputtering. The results of this study are described below.

2. Experimental The deposition con®guration is similar to those used in

* Corresponding author. 0040-6090/99/$ - see front matter q 1999 Elsevier Science S.A. All rights reserved. PII: S 0040-609 0(99)00303-X

Fig. 1. Deposition arrangement.

H. Biederman, V. StundzÏia, / Thin Solid Films 351 (1999) 151±157


Table 1 Deposition parameters of the of samples Ge8, Ge5, Ge6 and Ge7 Sample

Ge5 Ge6 Ge7 Ge8

Flow rate (ccm STP/min) n-Hexan


0.0085 0.0100 0.0110 0.066

1.5 1.5 1.5 1.5

Working pressure (Pa)

U (V)

I (A)

Deposition time (min)

Resistance per square (GV)

4 4 4 4

473 466 460 540

0.2 0.2 0.2 0.2

4 4 4 4

0.8 1.8 . 200 0.10

[1±4] and consisted of an unbalanced magnetron equipped with a graphite target of 80 mm in diameter and 3 mm in thickness with 1 germanium disc 25 mm in diameter and 1 mm thick placed on the center of the graphite target (Fig. 1). The substrate holder of the same size as the magnetron was placed 40 mm above the magnetron and is described in detail in [13]. After pumping down by rotary and diffusion pumps to the ultimate pressure of 10 23 Pa the deposition took place in an Ar/n-hexane mixture at a pressure of 4 Pa. The substrate negative bias reached 223 V during each

deposition carried out at a current of 0.2 A and a voltage between 400 to 550 V. Germanium was physically sputtered off from the target and incorporated into the hard plasma polymer originating from the plasma polymerization process in the Ar/n-hexane mixture under simultaneous mild positive ion bombardment. The preparation conditions of the four selected samples that have been fully characterised are shown in Table 1. These composite ®lms were simultaneously deposited on silica slides for optical measurements, and on silicon and

Fig. 2. Transmision electron micrographs (bright ®eld) of the samples: (a) Ge8, (b) Ge5, (c) Ge6 and (d) Ge7.

H. Biederman, V. StundzÏia, / Thin Solid Films 351 (1999) 151±157

Fig. 3. (a) Optical emission spectra during Ge/C:H composite ®lm deposition. Time dependence of the ratio of the intensities I of Ge line (265.2 nm) and Ar line (420.1 nm) during the sample deposition.

carbon foils on Cu grids (3 mm in diameter) for the study of the structure of the composite ®lms. A transmission electron microscope (Jeol FX 2000) was used for the study of the morphology of the composite ®lms. The accelerating voltage of the electron microscope was 200 kV. Also the changes in the size and the arrangement of the germanium clusters were studied with respect to the intensity and duration of the electron beam irradiation. A current density of 15 pA/cm 2 and time period below 30 s was needed to prepare


the micrograph. This included focusing and stabilisation (25 s), and exposure (2 s) of the picture. Current density of 150 mA/cm 2 for 300 s did not induce any changes in the morphology of the samples meaning that the electron irradiation of the sample during the preparation of the TEM micrographs did not in¯uence the morphology of the sample. A X-ray photoelectron spectrometer (ADES- 400 VG Scienti®c) was used to obtain information about the composition and the chemical bonding of elements in a near surface region of the Ge/C:H composite ®lms. Ka radiation was used for photoelectron spectra excitation. A hemispherical electron energy analyzer was adjusted at 20 or 100 eV pass energy. Spectra were recorded for both as-received and sputter-cleaned surfaces in the regions of Ge 3d, C 1s and O 1s lines at the normal emission angle. Argon ion-cleaning lasting 120 s was accomplished by 4 keV ion beam energy, 2 £ 1025 A/cm 2 ion beam current density at the sample surface, and at 608 to the surface normal. Atomic concentrations were determined semi-quantitatively within a simple model of a semi-in®nite solid homogeneous material in composition [14], accounting for photoelectron crosssections, asymmetry parameters [15], the inelastic mean free paths [16] and the experimentally determined transmission function of the hemispherical energy analyzer [17]. Rutherford Backscattering Spectroscopy (RBS) data were recorded using 1.7 MeV 4He 1 ions from a 2 MeV Van de Graaff accelerator. Usually total charges of 10 mC were used. From RBS spectra the density per unit area (number of atoms per cm 2) as well as depth pro®les of the germanium atoms have been calculated using the RUMP program. As RBS is not suitable for the detection of light elements, elastic recoil detection analysis (ERDA) was used to measure the area density of carbon, oxygen, and hydrogen. Spectra have been recorded using 35 MeV Cl7 1 ions from

Fig. 4. Core level spectra XPS of samples Ge8, Ge5, Ge6 and Ge7: (a) C1s, (b) O1s, (c) Ge 3d.

H. Biederman, V. StundzÏia, / Thin Solid Films 351 (1999) 151±157


Table 2 XPS Analysis of samples Ge8, Ge5, Ge6 and Ge7 on the surface and in the depth of the samples after sputter off. Relative error in determination of the elements concentrations may be estimated to 5% Sample

Ge8 Ge5 Ge6 Ge7

Ge (at. %)C (at.%)O (at.%) On the surface

In the depth

On the surface

In the depth

On the surface

In the depth

13.5 10.3 8.1 1.3

18.8 17.6 13.9 3.2

62.2 69.9 70.0 84.1

68.1 70.6 78.0 91.8

24.3 19.7 21.9 14.6

13.1 11.7 8.1 5.0

the Rossendorf 5 MV tandem accelerator EGP-10-1. The depth resolution of the set-up used was about 20 nm for carbon, allowing the analysis of depth pro®les of about 100 nm thick ®lms. Nuclear reaction analysis (NRA) was used for a detailed analysis of the hydrogen content in the Ge:C:H ®lms employing the nuclear reaction [1H(15N,alphagamma)12C]. The cross-section of this nuclear reaction has a sharp resonant structure in dependence on the energy. As the primary ions loose energy in the solid (stopping) the nuclear reaction occurs in a narrow depth region only. This allows depth-resolved spectra to be recorded by scanning the energy of the primary ions. Transmittance and re¯ectance spectra were recorded in a broad spectral region, 370±1650 nm. Custom made computer controlled single beam spectrometer and standard optical thin ®lm software (Film Wizard, Scienti®c Computing Int.)

has been used for evaluation of thickness, index of refraction n(E) and optical absorption coef®cient a (E), as a function of the photon energy. Tauc gap was evaluated from a (E) spectrum. FTIR absorption spectra in re¯ection mode were measured on samples deposited on Al precoated glass substrate by means of NICOLET 400 Spectrometer. The DC electrical measurements were performed using a Keithley 617 electrometer (PC computer driven) on planar samples. They were prepared by evaporation of silver ®nger-like electrodes, 10 mm apart, onto the surface of the composite ®lm samples.

3. Results and discussion The morphology of samples observed by TEM is shown in Fig. 2. It is obvious that for samples Ge5 and Ge6 germanium exists in the form of clusters of 1 to 2 nm in diameter. For Ge7 the germanium is in inclusions smaller than 1 nm in size. In the case of Ge8 the inclusions (clusters) still exist but they are not easy distinguished. From the electron diffraction micrographs, germanium in the form of clusters is con®rmed. In addition, the continuous character of the diffraction rings is in accordance with the ®nding that the dimensions of the germanium inclusions are 1 nm or less. Optical emission spectroscopy was used for monitoring the ratio of germanium (emission line at 265.2 nm) and argon (emission line at 420.1 nm) in order to get a constant concentration of germanium in the composite ®lms (Fig. 3a). In Fig. 3b the development of the lines ratio with the deposition time is presented, For the sample Ge8 a target partial cleaning is obvious while for the Ge7 sample a target Table 3 NRA and RBS Analysis of samples Ge5, Ge6 and Ge7 (relative error in determination of H concentration may be estimated to 5%) Sample RBS NRA

Fig. 5. (a) Hydrogen concentration in the three samples measured by NRA. (b) Germanium concentration in the three samples measured by RBS.


Ge (at. %)

C (at.%)

H (at.%)

Ge5 Ge6 Ge7

27.0±28.5 25.5±27.5 13.5±4.0

42.0±40.5 42.5±40.5 51.5±61.0

31 32 35

H. Biederman, V. StundzÏia, / Thin Solid Films 351 (1999) 151±157


approximately 200 (if a is in inverse centimeters and photon energy in eV) and its intercept with energy axis gives the Tauc (optical) gap. An exponential deviation below this value is due to the exponential tail in the density of states in this disordered amorphous alloy. Tauc gap is gradually changing for samples Ge8, Ge5 and Ge6 from 1 to 1.2 eV. The abrupt change occurring for sample Ge7 (Tauc gap is not well de®ned about 1.9 eV) could be connected to some structural(morphology) change. This can be also interpreted as follows: as the germanium content decreases we approach a composite structure with Fig. 6. FTIR spectra in re¯ection mode for the three samples.

poisoning is seen. This may cause a variation in the germanium concentration pro®les in these samples. The measured core level spectra (XPS) are shown in Fig. 4a±c . There is no evidence to support a carbide formation. It is obvious that germanium is bonded to carbon (alloy state), hydrogen and oxygen. The Ge, C and O elemental analysis of the surface and near surface region is presented in Table 2. As these data give analysis of the surface only and without information on hydrogen the samples were also analysed by nuclear reaction (NRA) and Rutherford backscattering (RBS). In Fig. 5a the NRA and in Fig. 5b the RBS spectra are shown for samples Ge5, Ge6 and Ge7. Sample Ge7 is graded with a decreasing germanium concentration towards the surface (Fig. 5b). This can be explained by a poisoning effect of the magnetron target with carbon during the deposition process. This results in a decrease of the germanium sputter yield. This is in accordance with the OES shown in Fig. 2b. The hydrogen content is much less affected as the Ge/C ratio (Fig. 5a). The results from NRA and RBS analyses are summarized in Table 3. where oxygen is not shown. However, from ERDA analysis we expect it to be within 1±2 at.%. From the FTIR measurements in re¯ectance mode we could detect C±H, Ge±C and Ge±H bonds (Fig. 6). Most investigators [5±7,19±21]assumed that the so called a-Ge C:H ®lms are homogeneous composites of Ge±C alloy type with some hydrogenation. In [7,8] the authors speak about germanium carbide without any convincing proof. In [22] the carbidic bond was detected in the material prepared under special conditions. This is in accord with the conclusions of [18]. Considering the TEM micrographs and our analyses we can qualitatively describe the ®rst model of our composite Ge/C:H ®lms: Germanium is mainly in small clusters ( , 1 nm in diameter), however, probably some small number of atomic germanium is dispersed in C:H bonded to C (alloy type bond) and also to H. More Ge±C (alloy) and Ge±H bonds will be on the surface of the clusters. The optical gap is plotted in Fig. 7a. The plot of (a E) 1/2 versus E should give the straight line above a value of

Fig. 7. (a) Determination of the optical gap. (b) Index of refraction n in transparency region of light.


H. Biederman, V. StundzÏia, / Thin Solid Films 351 (1999) 151±157

a dark conduction process were calculated from the measured current-temperature dependencies at a constant voltage 2 V (Fig. 8b). Their values range between 0.18 eV for Ge8 (highest Ge content) to 0.83 eV for Ge7 (lowest Ge content). These values coincide well with the measured optical gaps ranging from 1±1.9 eV that also indicate in the ®rst approximation the electronic structure (gap) of the respective samples. Because of `granular' structure of Ge/C:H composites [1± 4,24] where Ge inclusions are separated in the (C:H: (Ge)) matrix the charge carrier transport through this matrix dominates the conduction process. This may be a hopping process in analogy to previous ®ndings [6,23,24]. For a de®nite determination of this process more investigations are needed. 4. Conclusions We have shown that the Ge/C:H ®lms prepared by DC unbalanced magnetron have a granular character containing essentially germanium clusters ,1 nm in diameter. Some Ge is embedded in a C:H matrix. The optical gap ranges between 1±1.9 eV. Current-voltage characteristics presents an expected shape. Activation energies of a dark conductivity (in the temperature range 295±405 K) were determined for samples with germanium content 10±30 at.%. The current±temperature characteristics point out a possibility to apply Ge/C:H composite ®lms as temperature sensors. Acknowledgements We would like to express our thanks to Dr. R. Groetzschel, Dr. U. Kreib ig, Dr. D. Grambole, and to F. Herrmann for ion beam analyses. We are indebted to Prof. P. HlõÂdek and P. Bilkova for infrared spectra measurements. This work was partially supported by the grant ACTION 18P16 and OC NA 100.

Fig. 8. (a) DC current-voltage characteristics of Ge/C:H. (b) Current± temperature characteristics of Ge/C:H with activation energies.

more pronounced germanium clusters embedded in a hard plasma polymer (C:H) matrix. Values of the index of refraction n in the transparency region, below the Tauc gap can be interpreted as another measure of the ®lm composition and density. The spectral dependence n(E) is given in Fig. 7b. We see exactly the same trend as in Fig. 7a. Samples with the lowest Tauc gap has the highest index of refraction in the infrared, as a result of the highest germanium content. Again we see an abrupt decrease for sample Ge7. The current±voltage characteristics (Fig. 8a) measured at 295 K show the expected shape. The activation energies for

References [1] H. Biederman, P. HlõÂdek, J. PesÏicka, V. Stundzia, Vacuum 47 (1996) 11. [2] H. Biederman, P. HlõÂdek, et al., J. PesÏicka. Vacuum 47 (1996) 12. [3] H. Biederman, R.P. Howson, D. SlavõÂnskaÂ, V. Stundzia, J. Zemek, Vacuum 48 (1997) 10. [4] H. Biederman, V. Stundzia, D. SlavõÂnskaÂ, et al., Proc. IUPAC ISPC13, 3 (1997) 1165. [5] N. Saito, T. Yamaguchi, I. Nakaaki, J. Appl. Phys. 78 (1995) 6. [6] N. Saito, I. Nakaaki, T. Yamaguchi, S. Yoshioka, S. Nakamura, Thin Solid Films 269 (1995) 69. [7] Z.T. Liu, J.Z. Zhu, N.K. Xu, X.L. Zheng, Jpn. J. Appl. Phys. 36 (1997) 3625. [8] M.F. Ebel, H. Ebel, M. Mantler, et al., X-ray Spectrometry 21 (1992) 137. [9] R.K. Sadhir, H.E. Saunders, W.A. Byers, Polym. Eng. Sci. 33 (1993) 645.

H. Biederman, V. StundzÏia, / Thin Solid Films 351 (1999) 151±157 [10] M. Gazicki, J. Schalko, F. Olcaytug, M. Ebel, H. Ebel, J. Wernisch, H. Yasuda, J. Vac. Sci. Technol. A 12 (1994) 2. [11] M. Gazicki, M.J. Potrzebowski, J. Tyczkowski, J. Schalko, Thin Solid Films 258 (1995) 10. [12] M. Gazicki, H. Szymanowski, J. Tyczkowski, J. Schalko, F. Olcaytug, J. Vac. Sci. Technol. A 14 (1996) 5. [13] H. Biederman, V. Stundzia, D. SlavõÂnskaÂ, J. GlosõÂk, Vacuum 52 (1999) 415. [14] J.F. Moulder, W.F. Sticke, P.E. Sobol, K.D. Bomben, Handbook of Xray Photoelectron Spectroscopy, Perkin±Elmer Co., Eden Prairie, 1992. [15] I.M. Band, Y. Kharitonov, M.B. Trzhaskovskaya, Atomic Data Nucl, Data Tables 23 (1979) 443. [16] S. Tanuma, C.J. Powel, D.R. Penn, Surf. Sci. 53 (1987) 31.


[17] P. JirõÂcek, Czech. J. Phys. 44 (1994) 261. [18] M. Kumada, A. Masuda, T. Shimira, Jpn. J. Appl. Phys. 37 (1998) 1754. [19] J. Shinar, H.S. Wu, R. Shinar, H.R. Shanks, J. Appl. Phys 62 (1987) 808. [20] S. Kumar, H.J. Trohdal, Thin Solid Films, 193/194 (1990) 72. [21] M. Kumru, Thin Solid Films 198 (1991) 75. [22] T. Bruce, I. Bello, J. Huang, W.M. Lau, M. High, V. Strnad, P. Panchhi, J. Appl. Phys 76 (1994) 552. [23] J. Tyczkowski, E. Odrobina, M. Gazicki, F. Olcaytug, J. Non-Cryst. Solids, 137/138 (1991) 875. [24] V. Stundzia, P. BõÂlkovaÂ, H. Biederman, D. SlavõÂnskaÂ, P. HlõÂdek, Vacuum 50 (1998) 23.