Manuscript Received The Formation of Laser-Induced Ripple Structures N. C. Kerr, S. E. Clark and D. C. Emmony Loughborough University of Technology, Loughborough, Leics, LE11 3TU We report the results of an investigation into the formation mechanism Key words: Laser damage; Ripple structures; Schlieren imaging; Transient heating. 1. Introduction Many authors, as a result of performing investigations into the interaction of intense laser beams with solids, have noted the formation of well defined periodic damage patterns [1-8]. It is generally accepted, as first proposed by Emmony et al [2], that the cause of these structures is interference between the incident laser light and some form of induced 'surface wave'. Attempts have been made to formally model the mechanism responsible for the formation of LIRS, notably by Emel'yanov et al [7], Temple and Soileau [8], and in particular by Sipe et al [9,10,11]. Inherent in all theories of LIRS is the basic premiss that in order to form such structures there must be a periodic temperature profile induced on the surface. This study is an attempt to validate or otherwise this assumption. It is based upon using a high resolution Schlieren imaging system linked to a short pulsed dye laser to monitor the test surface, with the aim of capturing images of the surface whilst the transient heating induced by a KrF excimer laser is still present. 2. Experimental Technique The damage facility at Loughborough University [5] centres around a Lambda Physik model EMG 200 excimer laser that is operated with a KrF gas mix and produces nominal 1 J pulses of 20 ns duration at 249 nm. Our intention was to image any transient heating effects induced on the surface before the number of excimer pulses required to produce permanent ripples had been incident on the surface. Once such data had been obtained the surface would be permanently rippled by exposure to further pulses, and a comparison between the images of transient heating and permanent rippling made. Central to the requirements to be able to perform these experiments is a high resolution Schlieren [12] imaging system. This system is based upon using a computer controlled video framestore linked to CCD type video cameras. Hardcopy was obtained by using a 3M Dry Silver Imager. Figure 1 schematically illustrates the experimental arrangement used to obtain our results. Two video cameras are used in the system, one to monitor the surface in real time, the other linked to the framestore and used to record transient events. A 10 mW cw HeNe laser was used to monitor the surface in real time for any obvious signs of damage or LIRS. The light source for imaging the transient effects was a FL2000 dye laser synchronously pumped by the excimer beam and operated at 514 nm. Previous work [5] has shown that the heating effect at the surface was greatest immediately after the excimer pulse, and consequently by suitably reflecting the dye laser beam around the work area a 20ns delay between the arrival of the excimer and dye beams on the test surface was obtained. Narrow bandpass optical interference filters were used to ensure that each camera saw either the HeNe or dye beam but not both. In order to separate the transient effects from dust and other scattering sites already on a surface, a sequence of 3 dye laser images was used. The first image (labelled as a) was of the unheated surface. This was obtained by having the dye beam incident upon the surface but not the excimer pulse. The second (labelled b) was an image of the surface immediately after an excimer pulse. This image was obtained by having both the excimer and then the dye beam incident upon the surface. This image shows the transient heating caused by the excimer pulse. Finally the third image (labelled as c) was taken a long time after the incident excimer pulse with just the dye laser incident on the surface. These images were then computer processed. This processing consisted of intensity normalisation followed by subtraction and enhancement. Computer processing of images a and b enabled the net transient effects to be separated from those associated with dust etc that was on the surface, whilst processing a and c enabled any permanent changes to be detected. Once the surface had been suitably permanently rippled by exposure to further excimer pulses, a dye laser image of the surface was obtained to allow comparisons of transient and permanent effects to be made. S polarised excimer light was obtained by using the reflected light from a single quartz beam splitter orientated at Brewster's angle. Typically the excimer fluence was chosen so as to produce permanent ripples somewhere in the beam target interaction area after about 20 pulses. A sequence of transient images was recorded for each incident excimer pulse until the surface was permanently rippled over its whole area. This overall pattern developed in stages as subsequent excimer pulses were incident upon the surface. The surfaces used were polished Ge and GaAs mirrors, which prior to irradiation with the excimer laser had no special treatment except a wipe using Methanol to remove dust etc. 3. Results The excimer laser induced transient surface heating was found to be uniform until approximately five excimer pulses before the production of permanent ripples. At this point there were the first indications of periodic heating. In general, the best results were obtained in the interval between the surface being just about to permanently ripple (i.e. within 1-2 excimer pulses) and being substantially covered in them. Figures 2a-e are results which were obtained on GaAs irradiated by S polarised excimer light incident at 60° where the ripple spacing is ≈2.2 μm. Figure 2a was obtained after the surface first showed permanent ripples with only the dye laser incident whilst figure 2b is an image of the transient surface heating induced by the next excimer pulse obtained by having the dye incident on the surface immediately after the excimer laser. Figure 2c is the surface a long time after the pulse of figure 2b with only the dye beam incident. Inspection of figures 2a and c shows small permanent ripples on the right hand side of the image with essentially no ripples left of centre. It is just apparent from figure 2b that substantial transient periodic heating appears to the left of centre. Processing of figures 2a and c yields 2d the net permanent change whilst figures 2a and b yield 2e the net transient change. Comparison of these two images readily shows that the induced transient heating is periodic with the same spacing as the permanent ripples. Figures 3a and 3b are processed images again for S polarised light incident at 60° on GaAs, but show transient periodic heating without any permanent ripples being initially present or formed as a result of the incident excimer pulse. The transient nature of the pattern can be clearly seen. Figure 4 is another sequence of processed images on GaAs with S polarised light incident at 60°, and shows that initially transient heating patterns give rise to permanent ripples after sufficient incident pulses. Figure 4a is a processed image showing the permanent effect on the surface after 30 pulses have been incident and shows that no permanent rippling has occured upto this point. Figure 4b shows the transient effect caused by the next incident excimer pulse. Figure 4c is an image of the surface after 4 more incident excimer pulses and shows that permanent ripples have formed. Close inspection of figures 4b and c reveals that the areas of transient periodic surface heating towards the right hand edge of the images have become well defined permanent ripples. Results on Ge were very similar to the above except that the quality i.e. the definition of the ripples was less than those found on GaAs. It was noted that on Ge, the surface became mottled before even transient heating patterns were observed whereas, on GaAs it went directly from smooth to rippled. The consequence of the mottling on Ge was that it reduced the quality of the images, particularly when separating transient heating patterns from the surface mottling. Typical of the results is figure 5 which shows the transient heating pattern of spacing 6 um induced on Ge by S polarised light incident at 60°. The spacing of this pattern, which appears anomalously large, will be discussed in the next section. Attempts to quantify the transient temperature profile across test surfaces by computer analysis using routines originally written for laser beam profiling were made [5]. Due to the small ripple spacing, these measurements were restricted on GaAs but indicated that any surface heating was localised to the vicinity of the ripple. As a result of the larger ripple spacing on Ge it was possible to further pursue these measurements. Figure 6b shows the temperature profile through a transient heating pattern on a surface of Ge along the vertical line in figure 6a, and it can be seen that between the ripples there is essentially no surface heating. It was also noted that on a well rippled surface it required a fluence only a fraction e.g. 10% of that used to induce the ripples to cause further permanent changes on the surface 4.Discussion The results that we have obtained not only validate the basic assumption that the surface must have a periodic temperature profile but also, for the excimer fluences used, the idea of localised melting in the process of ripple formation. Young et al [11] by measurement of specular reflectivity and first order transient diffraction as opposed to direct surface imaging, showed that there are 4 regimes of LIRS formation dependent upon the incident laser fluence. At low fluence the surface undergoes At localised melting i.e. melts only in the vicinity of the E field maximum such that the surface consists of a periodic array of molten strips. Under these conditions, good agreement of the ripple spacing and that predicted by the theory in reference [9] is obtained by use of optical constants characteristic of the solid state of the material. high fluences the surface melts uniformly i.e. a continuous layer of liquid is formed with the consequence that both the morphology and spacing of the ripples is significantly different to that observed in the low fluence regime. It is worthwhile to note that in nearly all other theories of LIRS, notably that by Guosheng et al [13] and Erhlich et al [14] there is a requirement for the surface to melt uniformly before any periodic structure can develop. In the uniformly molten regime significant heating between the induced ripples would be expected as all the surface has been liquified whereas in the locally melted regime little if any heating between the ripples would be expected. The results that we have obtained by directly imaging the surface, are consistent with the above and clearly show that at the low fluences used herein to form LIRS, the surface melts only locally as there is essentially no heating between the ripples. Early imaging experiments showed that the dye laser had to be strongly pumped in order to produce detectable light on the cameras at the high magnification used. By using HeNe laser illumination, the incidence on a test surface of such a dye beam was found to produce no permanent effects unless the surface was already well rippled, in which case the ripples were very slightly enhanced. Unquestionably such a probe beam incident on a nominally smooth surface would cause some transient surface heating and ideally the dye beam fluence should be much less than that of the excimer to ensure that essentially all the observed effects are from the excimer and not the dye. Given that the absorbed excimer fluence was more than 10 times that of the dye, and as the dye wavelength was almost exactly double that of the excimer we feel that it is not unreasonable to conclude that as a result of the ratio of their wavelengths, the heating from the dye beam will couple directly into i.e. enhance that induced by the excimer beam rather than create its own pattern. Further more, the heating effect of the dye beam will be small compared to that of the excimer beam and will not have effected the results obtained. The ripple spacing on Ge samples whilst anomalously large is explained on the basis of an intensity interference mechanism whereby the surface melting from 2 induced wavevectors periodically overlaps and produces the large ripple spacings. Further details of this can be found in references [5,6 and 15]. The observation that once rippled only small fluences are required to further change the surface, is indicative of a positive feedback mechanism that preferentially couples the incident light into existing ripples. 5. Conclusions We have for the first time, shown directly, the requirement for a surface to have a periodic temperature profile in order to form LIRS and that at low fluences the surface melts locally i.e. forms periodic bands of molten material. We have also noted the reduced fluences required to change a surface once it has become rippled. N.C.K would like to acknowledge the financial support of S.E.R.C and B.D.H Ltd of Poole, Dorset. References [1] Birnbaum M., Semiconductor surface damage produced by ruby lasers, J. Appl. Phys. vol. 36, pp. 3688-3689, 1965 [2] Emmony D.C., Howson R.P. and Willis L.J., Laser mirror damage in germanium, Appl. Phys. Lett vol. 23, pp 598-600, 1973 [3] Marcus G.N., Harris G.L., Lo C.A. and McFarlane R.A., On the origin of periodic surface structures of laser-annealed semiconductors, Appl. Phys. Lett. vol. 33, pp. 453-455, 1978 [4] Mansour N., Reali G., Aiello P. and Soileau M.J., Laser generated ripple patterns on dielectrics and intermediate band gap semiconductors, National Bureau of Standards (USA) Spec. Pub. 727, 1984, pp137-146 and references therein [5] Clark S.E., Excimer laser induced modifications of optical surfaces, Ph.D. Thesis, 1988 Loughborough University [6] Clark S. E. and Emmony D.C., UV Laser Induced Periodic Surface Structures, submitted to Phys. Rev. B [7] Emel'yanov V.I., Zemskov E.M. and Seminogov V.N., Theory of the formation of 'normal' and 'anomalous' gratings on the surfaces of absorbing condensed media exposed to laser radiation, Sov. J. Quant Electron, vol. 14, pp. 1515-1521, 1984 [8] Temple P.A. and Soileau M.J., Polarization charge model for laser induced ripple patterns in dielectric materials, IEEE J. Quantum. Electron, vol. QE-17, pp. 2067-2072, 1981 [9] Sipe J.E., Young J.F., Preston J.S. and van Driel H.M., Laser induced periodic surface structures I. Theory, Phys. Rev. B, vol. 27, pp. 1141-1154, 1983 [10] Young J.F., Preston J.S., van Driel H.M. and Sipe J.E., Laser induced periodic surface structures II. Experiments on Ge, Si, Al and brass, Phys. Rev. B, vol. 27, pp. 1155-1172, 1983 [11] Young J.F., Sipe J.E. and van Driel H.M., Laser induced periodic surface structures III. Fluence regimes, the role of feedback and details of induced topography in germanium, Phys. Rev B, vol.30, pp. 2001-2015, 1984 [12] Holder D.W. and North R.J., Schlieren Methods, National Physical Laboratory, England, 1963 [13] Guosheng Z., Fauchet P.M. and Sigman A.E., Growth of spontaneous periodic surface structures on solids during laser illumination, Phys. Rev. B vol. 26, pp. 5366-5381, 1982 |