Materials Science and Engineering C 22 (2002) 359 – 366 www.elsevier.com/locate/msec
Spectro-ellipsometry on cadmium stearate Langmuir–Blodgett films G. Gonella a,b,*, O. Cavalleri a, I. Emilianov a, L. Mattera a,b, M. Canepa a,b, R. Rolandi a a
INFM UdR di Genova and Dipartimento di Fisica, Universita` di Genova, Via Dodecaneso 33, Genoa 16146, Italy b Laboratorio LAMIA, C.so Perrone 24, Genoa 16152, Italy
Abstract We have studied cadmium stearate Langmuir – Blodgett (LB) films by spectroscopic ellipsometry (SE) and we have compared the results with those obtained by using X-ray reflectivity and atomic force microscopy. The measurements on cadmium stearate LB films deposited on silicon wafers capped by a native oxide layer provide sound results for films formed by 1 – 25 molecular layers. The experimental data are well fitted by a Cauchy-type model over the whole examined wavelength range (245 – 725 nm). The model, which takes into account the anisotropy of the film, uses the film thickness of several mono- and multilayers films, together with the in-plane and the out-of-plane components of the refraction index as free parameters. The obtained mean thickness of the monolayer (ML), 2.50 F 0.02 nm, very well agrees with X-rays reflectivity measurements and with literature data. D 2002 Elsevier Science B.V. All rights reserved. Keywords: Spectro-ellipsometry; Cadmium stearate; Langmuir – Blodgett
1. Introduction Spectroscopic ellipsometry (SE), which is frequently used to determine the refractive index and the thickness of thin films of various nature [1], is still rarely used to study organic films [2]. In the last 10 years, out of about 2400 papers dealing with SE, only a minor fraction deals with organic films, notwithstanding the increasing interest into the field. This last figure sounds even more exiguous if we consider that the papers concerning with thin organic films are about 5500. On the basis of these figures, it seems to us that the SE capability to characterise organic films is not fully exploited and we intend to explore it systematically. In this paper, we report our first results of SE measurements on organic films well characterised by other techniques to test our experimental systems and to establish procedures for future measurements. Organic films can be prepared with techniques such as molecular beam deposition, evaporation, self-assembly (SA) and Langmuir –Blodgett (LB) technique. According to the last one, surfactant molecular layers, formed at the air – water interface, are sequentially deposited on suitable solid substrates. The film thickness is easily tailored by * Corresponding author. Dipartimento di Fisica, Universita` di Genova, Via Dodecaneso 33, I-16146 Genoa, Italy. Tel.: +39-10-353-6287; fax: +3910-311-066. E-mail address:
[email protected] (G. Gonella).
choosing the number of deposited layers since the monolayer (ML) ordered stacking extends for almost the entire film thickness: for this reason, LB films are very suitable to test SE measurements [1]. So we have chosen to use mono- and multi-molecular LB films of cadmium stearate, [CH3(CH2)16COO]2Cd2 + (CdSt2), deposited on silicon substrates. Atomic force microscopy (AFM) and X-ray reflectivity measurements (XRR) have been also performed for useful comparison. SE measures the change of the light polarization upon reflection at an interface [1]. The effect is best described using the incidence plane, defined by the propagation direction of the beam and the normal of the reflecting surface as reference frame. The light with the electric field vector oscillating within, p-light, and perpendicularly, s-light, to the plane of incidence, remains linearly polarized after reflection. For incident light not exactly in p- or s- state, the change of the polarization is caused by the change of the phase and amplitude of the p and s components according to the Snell reflection rules. The reflection properties of a sample, according to the Jones formalism [1], are described by the reflection matrix: 0 1 rpp rps @ A ð1Þ rsp rss where ri,j are the complex reflection Fresnel coefficients.
0928-4931/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 8 - 4 9 3 1 ( 0 2 ) 0 0 2 0 8 - 4
360
G. Gonella et al. / Materials Science and Engineering C 22 (2002) 359–366
In the case of isotropic materials or materials with uniaxial anisotropy and optical axis perpendicular to the surface, the Jones matrix is diagonal 0 @
rpp
0
0
rss
1 A
ð2Þ
where rpp and rss are the reflection coefficients for p-light and s-light, respectively. Standard SE measures the ellipsometric parameters W and D [1], which are related to rpp and rss by the relation:
q¼
rpp ¼ tgWeiD rss
ð3Þ
Assuming, in the most general case, that the reflecting surface is made up by a stack of optical layers, U depends on the incidence angle, the radiation wavelength, and the thickness and refraction index of each layer. The thickness and the refraction index of the examined film are obtained from measurements of D and W as function of the wavelength by fitting the experimental data with suitable optical model of the system under consideration.
2. Experimental methods 2.1. Substrate preparation and film deposition We used commercial (ACM) silicon wafers, 2 cm in diameter, capped by an oxide layer. The wafers were cleaned by the following procedure: 30 min in ‘‘piranha’’ solution (30% H2O2 (BDH Analar, purity>30%): 70% concentrated H2SO4 (Fluka, purity 95– 97%)) at 80 jC, 3 min in HF (Fluka, purity>48%), 30 min again in piranha solution at 80 jC. Each step was followed by careful rinsing in Milli-Q water. This cleaning procedure assures a good hydrophilic substrate. The morphological quality of the prepared samples has been checked by AFM. The spectroscopic characterisation will be described in the next section. LB films deposition was performed by using a computer-controlled Nima Technology trough. A 1 mg/ml solution of stearic acid (Fluka, purity>99.5%) in chloroform (BDH, Aristar) was spread on a purified water (MilliQ) subphase containing 2.5 10 4 M CdCl2 (Sigma, ACS Reagent) adjusted to pHf6.5 by using NaHCO3 (Sigma, purity>99.5%). Before spreading the stearic acid, the purity of the subphase was checked by controlling that the pressure during isothermal compression did not increase. The transfer was performed at 27 mN/m and the dipping speed was 10 mm/min. We have deposited films of 1, 3, 7, 13, 25 ML thickness.
2.2. SE The ellipsometric study has been performed by means of a M-2000 (J.A. Woollam) rotating compensator spectroscopic ellipsometer [3]. This instrument is computer controlled and can measure simultaneously at 225 different wavelengths in the range 245– 725 nm and variable angle of incidence (40 –90j). The M-2000 SE is equipped with WVASE32 (J.A. Woollam), a powerful and flexible software for data analysis [3]. In this paper, spectra at three different angles of incidence (50j, 60j and 70j), both for substrates and films, have been measured in order to collect a large data set for fitting purposes. 2.3. XRR X-ray reflectivity measurements have been performed with a butterfly-type X-ray reflectometer set-up at the Department of Physics of Genoa University, by using mechanical parts provided by JJ X-Ray (Denmark). The reflectometer is equipped with a conventional X-ray glass tube (PW 2273/20, Long Fine Focus, Cu anode, Philips Analytical). A Ni-filter cuts the CuKh line and lets the CuKa line (0.15417 nm) pass through an attenuator made up of Cu foils to suitably reduce the beam intensity. A couple of slits before the energy-dispersive Si-detector (AMPTEK) cuts off the diffuse scattering background. The incident angle can be varied from 0j to 7j with minimum steps of 0.0003j. The set-up allows reflectivity measurements down to R f 10 7. The experimental data were analysed with IMD software developed by David L. Windt, provided by ESRF (Grenoble, France) as a part of the XOP 2.0 (X-ray Oriented Programs) software package [4]. 2.4. AFM AFM measurements have been performed by using a Dimension 3000 equipped with a ‘‘G’’ scanner head (scan range: 92.8 Am) and controlled by a Nanoscope III (Digital Instruments, Santa Barbara, CA, USA). The images have been obtained in contact mode using 200 Am Si3N4 cantilevers with spring constants of about 0.06 Nm 1.
3. Results and discussion In each deposition, only half of the wafer was dipped into the solution. For each sample, we have then measured W and D both on the bare and on the covered portion of the wafer. A representative example of the data obtained on the bare substrate is shown in Fig. 1. The data are qualitatively consistent with measurements on similar substrates [5]. In order to obtain a quantitative description of the substrate, which is indeed fundamental for a reliable anal-
G. Gonella et al. / Materials Science and Engineering C 22 (2002) 359–366
361
Fig. 1. Ellipsometric Spectra of a substrate: (a) W and (b) D as a function of the wavelength for incidence angles of 50j, 60j and 70j; symbols: experimental data; continuous line: calculated values obtained by using the model described in the main text.
ysis of the deposited films, we have modelled the substrate as an optically thick silicon layer covered by a silicon oxide layer with an intermediate layer mimicking the Si/SiO2 interface. Tabulated values of the Si and SiO2 optical parameters have been adopted [6]. The Si/SiO2 interface was modelled through an effective medium approximation. Under
the assumption that the interface is composed of 50% of Si and 50% of SiO2, the Bruggeman relation [7] reads: eA e eB e ¼ eA 2e eB 2e
ð4Þ
where e is the complex dielectric function of the mixed material, and A and B refer to the constituent materials.
362
G. Gonella et al. / Materials Science and Engineering C 22 (2002) 359–366
The model accurately reproduced the experimental data, with the thickness of the mixed layer of (0.32 F 0.06) nm and that of the oxide layer of (2.09 F 0.04) nm. The calculated values are shown as continuous lines in Fig. 1. The fit is rather satisfactory (mean square error MSE = 3) along the whole wavelength range with small discrepancies only in the N-UV region, especially for incidence angles of 50j and 60j.
Fig. 2 shows the SE patterns obtained for films of different thickness. For the sake of clarity, only the data collected at 70j are reported. In Fig. 3, we show the whole set of data related to the 25-ML film. In order to reproduce the data, we have assumed an uniaxial model [1] with the optical axis perpendicular to the film surface, accordingly to the cylindrical symmetry of the molecular constituents.
Fig. 2. Ellipsometric Spectra of 1, 3, 7, 13 and 25 ML of CdSt2 LB films at the same angle of incidence (h = 70j): (a) W and (b) D.
G. Gonella et al. / Materials Science and Engineering C 22 (2002) 359–366
363
Fig. 3. Ellipsometric Spectra of the 25 ML sample at incidence angles of 50j, 60j and 70j: (a) W and (b) D. Symbols and continuous line as in Fig. 1.
Because the CdSt2 films are transparent in the visible range [1,2], we have considered real refractive indexes. The following Cauchy [8] dispersion relation: n¼Aþ
B k2
ð5Þ
has been used both for nx,y (in-plane refractive index) and nz (out-of-plane refractive index). We have also assumed the refractive indexes independent from the thickness, so that
only the Cauchy parameters (Ax,y , Bx,y , Az, Bz) and the film thickness d were the free parameters of the model. From a multi-sample analysis [9], using as free parameters, the thickness of the five different films and Ax,y, Bx,y, Az, Bz (the same for all the samples), a satisfactory fit of the whole ensemble of data (MSE = 3.9) has been obtained. The values of the parameters produced by the best fit of the experimental data are reported in Table 1 and the calculated spectra are represented in Fig. 3 as continuous lines. The resulting
364
G. Gonella et al. / Materials Science and Engineering C 22 (2002) 359–366
Table 1 Cauchy parameters and thickness values of the LB films obtained with the fit procedure described in the main test (MSE = 3.9) Ax,y Bx,y Az Bz d (1 ML) d (3 ML) d (7 ML) d (13 ML) d (25 ML)
1.4932 F 0.0007 (0.0066 F 0.0001) Am2 1.5377 F 0.0007 (0.0065 F 0.0001) Am2 (2.164 F 0.009) nm (7.34 F 0.02) nm (16.62 F 0.01) nm (31.59 F 0.02) nm (62.34 F 0.02) nm
refractive indexes as a function of the wavelength k are reported in Fig. 4. The slope of the straight line fitting the thickness of the film as a function of the number of deposited layers provides the average thickness per monolayer (2.50 F 0.02) nm (Fig. 5). The thickness of the 1-ML film appears neatly lower than the above-mentioned average value. This finding is in agreement with literature results and it is likely due to a relative molecular disorder of the monolayer film that leads to a larger tilt of the molecules [10 – 13]. The results of the SE measurements have been compared with those of XRR measurements. Fig. 6 shows the reflectivity curve obtained on the CdSt2 13-layers film. The most pronounced maxima are the main Bragg peaks caused by the interference of the waves reflected by the electronically denser, hydrophilic parts of the film. The spacing between these layers is provided by the equation: d ¼ k=ð2sinðhnþ1 hn ÞÞ
ð6Þ
where k is the X-ray wavelength, hn + 1 and hn are the angular positions of two adjacent Bragg peaks. This spacing turns out
Fig. 4. Calculated values of nx,y and nz of the CdSt2 LB films as a function of the wavelength.
Fig. 5. Thickness values provided by the best-fit procedure as a function of the number of molecular layers in a film. The slope of the straight line fitting the data, 2.50 F 0.02 nm, gives the thickness per ML.
to be (4.96 F 0.05) nm. As expected for a well ordered Y-type film, this value is very close to twice the longitudinal length of the stearic acid molecule. This value is fully consistent with the thickness per ML obtained by SE. Immediately after the critical angle and around main Bragg peaks, less pronounced maxima, the so-called Kiessig oscillations, are visible. They are caused by the interference of the signals reflected by the substrate – film interface and the film –air interface. Their angular distance provides the film total thickness according to Eq. (6). We note that as the Kiessig oscillations are generally damped by surface roughness, their observation suggests that the film surface is relatively smooth. The resulting total film thickness, 31.65 F 0.05 nm, is consistent with the SE thickness determination of the 13-ML film (cf. Table 1) as well.
Fig. 6. X-rays reflectivity curves of the 13 ML sample: reflectivity, in arbitrary units, is plotted as a function of the incidence angle. From the angular distance between Kiessig and Bragg peaks, values of 31.65 F 0.05 and 4.96 F 0.05 nm can be deduced for the film and bilayer, respectively.
G. Gonella et al. / Materials Science and Engineering C 22 (2002) 359–366
365
Fig. 7. AFM (upper part) image of the border region between bare (on the left) and 1 ML covered (on the right) substrate; (lower part) z profile obtained sectioning the image along the white line. The monolayer thickness results to be approximately 2.4 nm.
The value of the thickness per monolayer obtained from SE and the value of the repeating unit in XRR data are nicely consistent with previous X-rays reflectivity measurements on the same system [10,11]. AFM does not provide such a reliable and accurate determination of thickness as XRR and SE do. However, it allows one to inspect the sample topography and to compare local measurements with the mean values provided by the spectroscopic techniques. Fig. 7 (upper part) shows an AFM image of the border region between the bare (on the left) and the monolayer covered substrate (on the right). The monolayer thickness estimated from z profiles, whose an example is shown in Fig. 7 (lower part), is about 2.4 nm. Taking into account the noise level of the background signal and considering that this measurement has been obtained in a particular region of the sample, this value can be considered in satisfactory agreement with the SE result. Finally, it is interesting to attempt a comparison with generalized spectroscopic ellipsometry results recently obtained on a closely related system such as cadmium behenate (CdBe2) LB films [14]. Though CdBe2 films are slightly thicker than CdSt2, the optical properties of the two systems are expected to be relatively similar. In fact the authors, Lecourt et al. [14] found values for the refractive index, which are close to our determinations. However, they claim also a relatively weak in-plane anisotropy (nx differing from ny), which we could not resolve in the present standard spectroscopic ellipsometry
measurements. A generalized spectroscopic ellipsometry experiment is in progress.
4. Conclusions The aim of this work was to establish standard procedures for SE measurements on thin organic films. In summary, we have used SE, in the 245– 725nm wavelength range, to investigate the optical properties and thickness of CdSt2 LB films on silicon substrate. The substrate properties have been obtained by using a Si/ intermixed layer/SiO2 model, while for the organic film, a Cauchy model with uniaxial anisotropy has been adopted. We have determined the in-plane (nx,y), out-of-plane (nz) refractive indexes and the thickness of different films made up of 1 –25 ML. XRR and AFM measurements have confirmed these results that are in excellent agreement with the literature data. We intend to develop this research to study other kinds of organic films, in particular self-assembled monolayers. Preliminary measurements have been performed on alkanethiols, L-cysteine and azurin on gold.
Acknowledgements The authors thank Thomas Wagner (LOT-Oriel) for the advice in the spectro-ellipsometric data analysis.
366
G. Gonella et al. / Materials Science and Engineering C 22 (2002) 359–366
References [1] R.M.A. Azzam, N.M. Bashara, Ellipsometry and Polarized Light, North-Holland, Amsterdam, 1977. [2] A. Ulman, An Introduction to Ultrathin Organic Films, Academic Press, San Diego, 1991. [3] http://www.jawoollam.com. [4] D.L. Windt, Comput. Phys. 12 (1998) 360. [5] H.G. Thompkins, W.A. McGahan, Spectroscopic Ellipsometry and Reflectometry, Wiley, New York, 1999. [6] E.D. Palik (Ed.), Handbook of Optical Constants of Solids, Academic Press, New York, 1985. [7] D.A.G. Bruggeman, Ann. Phys. (Leip.) 24 (1935) 636.
[8] L. Cauchy, Bull. Des. Sci. Math. 14 (1830) 9. [9] T.U. Kampen, A.M. Paraian, U. Rossow, S. Park, G. Salvan, T. Wagner, M. Friedrich, D.R.T. Zahn, Phys. Status Solidi A. 188 (2001) 1307. [10] D.K. Schwartz, Surf. Sci. Rep. 27 (1997) 241. [11] Y. Sasanuma, H. Nakahara, Thin Solid Films 261 (1995) 280. [12] D.J. Ahn, E.I. Franses, J. Phys. Chem. 96 (1992) 9952. [13] F. d’Acapito, I. Emilianov, A. Relini, P. Cavatorta, A. Gliozzi, V. Morante, S. Morante, P.L. Solari, R. Rolandi, Langmuir 18 (2002) 5277. [14] B. Lecourt, D. Blaudez, J.M. Turlet, Thin Solid Films 313 – 314 (1998) 790.
