LIGHT BACKSCATTERING ANALYSIS of Textured Silicon SAMPLES 8TH EUROREGIONAL WORKSHOP ON “ THIN SILICON DEVICES” SALERNO, ITALY MARCH 6TH-8TH, 2002 LIGHT BACKSCATTERING ANALYSIS of Textured Silicon SAMPLES A. Parretta, E. Bobeico*, I. Nasti and F. Varsano ENEA Centro Ricerche Portici, I-80055 Portici (Na), Italy INTRODUCTION The light collection capabilities of silicon solar cells are strongly influenced by the particular surface structure. The total reflectance of silicon surface, can be reduced, for example, from ~ 30% down to 3-4% when pyramid-like structures are formed [1]. Besides reducing reflection at surface, texture scatters light into the semiconductor, thereby promoting multiple internal reflections and improving the overall absorption process. Surface texture is a key factor in the modern silicon cells technology, as the actual tendency is to reduce cell thickness down to few tens of microns. METHOD and APPARATUS We investigate the interaction between a parallel light beam from a laser source, incident normally onto a silicon sample, and the silicon surface textured in different ways, by measuring the spatial distribution of the intensity of the light backscattered on the front hemisphere. Backscattered light incident beam Silicon sample Fig.1 shows the schematics of the optical apparatus used to measure the backscattered light. A light beam from the laser source (l) (λ0 = 543 nm or λ0 = 633 nm) crosses the integrating sphere (is) and strikes on the sample (c). Purpose Surface texture determines light scattering towards the semiconductor and back into air. The basic idea of the present work is that light backscattered from the silicon surface brings information on the way light is scattered into the semiconductor and then on the light collection process. Purpose of this work is to look for a correlation between backscattered light spatial distribution and total reflectance (total optical loss) of the cell. Light is backscattered from the sample towards the front hemisphere, and the part confined in the solid angle W is captured by the sphere through window (w). The light collected and integrated by the sphere is measured by the photodiode (d). The intensity of light at direction , averaged with respect to the azimuth angle f, can be obtained by a differentiation vs.  of the measured current. q (is) (c) W (ch) (lock-in) x (g)  D (d) f Flat silicon High Reflectance Textured silicon Low Reflectance sample Preparation An array of pyramids with the same pitch (4m) and different square size was produced on the surface of a crystalline silicon wafer. The patterned area is 1.5 cm2. The process is articulated into four steps: Deposition of a SiO2 thin film ~ 1000 Å Photolithographic process Silicon oxide etching with a buffered HF solution Anisotropic etching of silicon using an alkaline (KOH) solution at 80°C The mask used in the photolitographic process is made of an array of circles of 2μm diameter with a 4μm pitch. Playing with the exposition and developing of the resist, as well as with the oxide etching time, it is possible to open in the oxide circles of different diameter. In this way we got square-based pyramids of 2 × 2 μm2, 3 × 3 μm2 and 4.0 × 4.0 μm2 size. (l) Fig 1- Apparatus BASALT(Backscattered light topographer) We have: I()  light collected within solid angle W (); I(, f)  intensity of light backscattered towards direction (, f); I() = The average intensity of light backscattered at angle  is obtained by differentiating I() and normalizing with respect to the cross section area: RESULTS SiO2 etch with buffered HF solution Photolitographic process mm mm mm Patterned sample Total hemispherical reflectance at l=543 nm of the textured samples, as obtained at a Lambda 9 Perkin Elmer spectrophotometer. The textured samples show quite similar reflectance values, smaller than that of flat silicon. Measured current by photodiode (d) as function of sample-sphere distance. The current, proportional to the collected light by the sphere, show different behaviours, corresponding to different distributions of the backscattered light. SEM views of the pyramids array Optical microscope views of the pyramids array CONCLUSIONS Even if the smallest pyramids, with the largest relative flat area, have shown, as expected, the highest reflectance, the total reflectance of the differently textured silicon samples is resulted, at a wavelength of 543 nm, not sufficiently differentiated in order to make the searched correlation. As regards the light scattering, moreover, the use of samples textured with the same pitch has determined light diffractions characterized by similar spatial distributions. As the spectral reflectance measurements show, the reflectance values divergence significantly at l>600 nm. The actual experimental conditions, therefore, seems to be quite unfavourable to carry on the correlation between total loss and light backscattering figure of textured silicon samples. Future experiments should be performed at higher wavelengths and at different pitch values of textures. Nevertheless, we have developed a procedure for the analysis of backscattered light which has demonstrated to give correct results, and which can be considered valid, therefore, in the future investigations. mm mm Measured current as function of the cone semi-aperture angle q. The curves were smoothed by applying an adjacent averaging (10 points) procedure. Scattered light intensity as function of angle q, calculated by applying Eq. (2). The high peaks at  7° correspond to the expected 1st order diffraction pattern. Smaller peaks are easily distinguishable, corresponding to the higher orders of diffraction. [1] A. Parretta et al., Optics Communications, 172 (1999) 139-151. * Permanent address: Institute of Applied Physics, Academy of Sciences of Moldova