Quantitative study of C—H bonding in polymerlike amorphous carbon films using in situ infrared ellipsometry T. Heitz,* B. Dre ´ villon, C. Godet, and J. E. Boure ´ e Laboratoire de Physique des Interfaces et des Couches Minces (UMR 7647 CNRS), Ecole Polytechnique, 91128 Palaiseau Cedex, France ~Received 15 April 1998! Polymerlike hydrogenated amorphous carbon ( a -C:H) films have been deposited by plasma CVD at low temperature and low pressure. Vibrational properties have been investigated by in situ infrared ellipsometry as a function of film thickness. Hydrogen distribution within the films has been changed by varying the ion energy impinging on the film surface. Vibrational properties of C—H stretching and bending modes have been analyzed as function of self-bias ( V bias ) in terms of frequency, bandwidth, and intensity. Absorption strengths are associated with effective charges that have been calculated for the different sp 3 CH n groups. In order to make a comparison with values of organic chemistry, a general review of infrared spectra including alkanes, alkenes, and aromatic hydrocarbons is presented. A dipole description taking into account the local environ- ment of CH bonds is developed showing that methyl and methylene group effective charges are similar for polymeric a -C:H and C x H y organic compounds. Line broadening and frequency shift effects are found to depend on the type of CH groups and are explained through a model including strain and dipole-dipole interactions. The sensitivity of effective charges to the local environment and the determination of CH n group densities are used to propose a description of the hydrogenated network structure. @S0163-1829~98!01044-3# I. INTRODUCTION Hydrogenated amorphous carbon ( a -C:H) films show re- markable physical and chemical properties such as high hardness, electrical resistivity, and near-infrared ~IR! trans- parency or room-temperature luminescence. 1,2 These proper- ties are closely related to the film microstructure, depending on the hybridization state of carbon atoms ~sp 2 or sp 3 ! and on hydrogen distribution among these two configurations. Several techniques have been used to investigate hydrogen content and the sp 2 / sp 3 ratio, such as nuclear magnetic resonance 3 ~NMR!, electron energy loss spectroscopy 4 ~EELS!, and infrared absorption spectroscopy 5,6 ~IRAS!. The first technique can provide quantitative results on the sp 2 / sp 3 ratio but needs a large amount of material; the sec- ond one yields inconsistent conclusions in the low-energy range ~0–40 eV!. 4 IRAS is a widespread technique but faces many problems for providing quantitative results, aris- ing from IR oscillator strength changes and band overlap- ping. Using IRAS technique, some authors 7–9 have tried to de- termine the total H content by calculating the whole absorp- tion area of the stretching ~around 2950 cm 21 ! or bending ~around 1450 cm 21 ! bands. However, the results were incon- sistent with other direct techniques like elastic recoil detec- tion ~ERD!. The decrease of the IRAS signal observed with increasing sp 2 / sp 3 ratio had then been interpreted in terms of bound ~IR active! and unbound ~IR inactive! hydrogen 8,9 leading to the conclusion that up to 50% of hydrogen can be trapped as H 2 molecules. Nevertheless, inelastic neutron scattering has shown that the H 2 fraction 10 contributing to the total hydrogen content is low. This contradiction comes from the hypothesis of constant IR absorption strengths that had often been assumed 6,7,11,12 to facilitate IR data analysis. Jacob 13 has shown that the procedure used by Brodsky 14 in the case of amorphous hydrogenated silicon ( a -Si:H) to cal- culate H concentration from IR data, cannot be applied to a -C:H owing to changing absorption strengths. Furthermore, as hydrogen is not equally distributed between the sp 2 and sp 3 phases, and as no quantitative data on the transition mo- ments have been so far available, the evaluation of the sp 2 / sp 3 ratio by IRAS leads systematically to high values, compared with NMR data: 15 IR results have thus to be cor- rected by empirical factors to be in agreement with NMR results. 15 In order to check hydrogen distribution, fitting procedures using Lorentzian or Gaussian shapes are necessary to get the vibrational contribution of each group since the different CH n bands overlap. Nevertheless decomposition has to be cautiously performed so that results do not depend on the choice of the number of bands as well as on the free param- eters selected for each band ~width, frequency!. Finally, no quantitative conclusions can be drawn about hydrogen distri- bution and film microstructure from a band decomposition unless the absorption strengths for each CH n unit are known. Previous IR analyses 16,17 have revealed that, in the field of organic compounds, some transition dipole moments can be considered as a constant. The aim of this paper is, in particu- lar, to check whether this assumption is valid in the case of polymerlike a -C:H films. The experimental deposition setup described in Sec. II allows to obtain a large range of poly- meric carbon films having different sp 2 / sp 3 ratios. In Sec. III, the general formalism of IR absorption is presented in terms of effective charges and local field corrections. In or- der to perform reliable calculations, a specific method, which will be presented in Sec. IV, is used to analyse a -C:H film vibrational properties. This procedure is partially based on the method used by Shanks et al. 18 and Langford et al. 19 for a -Si:H thin films: transition dipole moments are calculated by combining absorption intensity measurements and deter- PHYSICAL REVIEW B 15 NOVEMBER 1998-II VOLUME 58, NUMBER 20 PRB 58 0163-1829/98/58~20!/13957~17!/$15.00 13 957 ©1998 The American Physical Society