1 Flat-top shaped laser beams: reliability of standard parameters P. Di Lazzaro 1, S. Bollanti 1, D. Murra 1, E. Tefouet Kana 2, G. Felici 3 ENEA, Dept. FIS-ACC, P.O. Box 65, 44 Frascati (Italy). Fax: +396945334, E-mail: dilazzaro@frascati.enea.it 2 ENEA fellow 3 Info&Tech, via Teognide 24, 124 Rome (Italy) ABSTRACT We present experimental results of reshaping and making uniform the spatial energy distribution of raw beams respectively emitted by a low-coherence excimer laser and by a highly coherent diode pumped Nd-YAG laser. We used an optical system which is able to homogenize bad beams having strong local intensity spikes, and to modulate almost continuously the spot size of the homogenized beam along one or both axes in a fixed target plane. We have evaluated the results using the standard parameters described in the document of the International Organization for Standardization ISO 13694. We found that the reliability of the results is dependent both on the experimental setup and on the definition of the edge steepness and plateau uniformity of the quoted ISO document. Here we propose an amendment to the definition of these standard parameters that could improve their reliability. Keywords: ISO standard, edge steepness, homogenizer, excimer laser Introduction Most beams emitted by industrial lasers are multi-mode, each mode having a different shape and width, and show a non-homogeneous energy distribution across the spot size. In a word, they are raw beams. On the other hand, many industrial laser applications require photon energy uniformly distributed across target areas. In most cases it is necessary to use flattopped spatial energy distributions with steep-edges and high-uniformity in the plateau region. Examples include laser material processing (e.g., ablation, surface cleaning, marking, drilling, metal-hardening), flat-panel display fabrication, medical applications (e.g., corneal reshaping, UV-curing dermatological diseases), laser injection in fibers. For these reasons, different beam shaping techniques are employed to smooth out the intensity fluctuations, which all mix fractions of the original beam [1]. Reliable standard methods for evaluation of nearly flat-top beam parameters are of great importance for a correct characterization of laser beams, as well as for proper comparisons between different laser systems performance. The International Organization for Standardization (ISO) prepared a document, named ISO 13694, containing standard procedures and definitions to evaluate the goodness of nearly flattopped beams [2]. In this paper, we present a novel optical system able to reshape and homogenize raw laser beams, and the experimental results of the characterization of different laser beams done in accordance with the ISO 13694. We will show that the reliability of the results is dependent on the experimental setup and on some ISO standard definitions. 1. Beam shaping The most popular beam shaping technique is based on splitting the beam in many beamlets that overlap in a focal plane. The optical element which divides the incident beam (the divider ) is made by a spherical fly-eye or by a couple of crossed cylindrical-lenses arrays, while the element which overlaps the splitted beams on a focal plane (the condenser ) is a spherical lens or a couple of cylindrical lenses. The integral of different portions of the input raw-beam provides a smoothing of the intensity profile proportional to the number of beamlets. That is, the larger the number of the divider
lenses, the more uniform the beam in the focal plane. Unfortunately, this number cannot be increased too much because the size of the divider lenses has an intrinsic lower limit. In fact, a divider-lens size smaller than the transverse coherence length of the laser creates unwanted interference effects, thus frustrating the homogenizing effect. As a consequence, designers usually make a trade-off yielding homogenizers able to integrate only the average fluctuations of the input irradiance. Then, relatively poor results are expected in the case of light beams having local intensity fluctuations much stronger than the average fluctuations (e.g., beams with irregular irradiance shapes emitted by dischargepumped lasers, like excimer, copper vapor, TEA CO 2, and beams emitted by diode-laser arrays). A further inconvenience of conventional homogenizers is that once the geometry and lens power are fixed, the energy density on the focal plane can be modulated only by changing the energy of the input beam. Sometimes it is not convenient changing the output energy of a laser, either because this may reduce the efficiency and stability of the laser emission, or because this may be hardly done continuously. As a consequence, processes requiring different energy density values and/or different spot sizes usually need distinct homogenizer systems. A possibility to overcome the above limitations is given by a recently proposed homogenizer with zoom and asymmetric array dividers [3]. In short, this patented [4] homogenizer shows two advantages with respect to existing homogenizer technology: it allows to make homogeneous any laser beam, including those having asymmetric intensity spikes, and to vary the spot size of the homogenized beam (which can be either a square or a rectangle with a fixed or a variable aspect ratio) without changing any optical element and keeping constant the path length from the laser to the target plane. In order to fully exploit this homogenizer technology, ENEA developed proprietary software to design an optical system optimized to achieve the wished output beam performance (once the input beam characteristics are known) [5]. In addition, we recently developed a more general theoretical modeling of the homogenization of a laser beam (supposed spatially coherent) by the integration method [6] The principle of the ENEA homogenizer, with the asymmetric divider and the zoom lens added after the condenser lens, is disclosed in Fig. 1. Beam intensity profile Focal plane Figure 1. 2-D scheme of the ENEA transfocal homogenizer. When the beam has an asymmetric profile, the divider array is made by different lenses, such that the smaller lenses intercept the less homogeneous part of the beam, and each lens size is much larger than the local coherence length of the corresponding portion of the beam. 2. Experimental The Hercules [7] laser pulses (λ=38 nm, output energy 8 J/pulse, near-field beam size 1cm 5cm), properly attenuated, are measured both before and after passing through the transfocal homogenizer using a CCD camera (Andor, mod. DV-43UV, 77 1152 pixels). To obtain the shape of the whole spot, we captured consecutive shots by rastering the CCD camera across-the beam with the help of a step-motor-driven linear positioner (P.I., mod. M521-DD).
Figure 2 shows the typical near field profiles of the laser beam emitted by Hercules. The horizontal profile (along the discharge electric field) shows a marked asymmetry caused by characteristic discharge constrictions (the so-called streamers [8]) next to the grounded electrode. Here we have used a transfocal homogenizer made by two crossed arrays of 4+4 cylindrical lenses as dividers, by a cylindrical condenser along the horizontal direction and by cylindrical condenser and zoom along the vertical one. According to the design software, this homogenizer reshapes the 1cm 5cm Hercules beam into a beam having a fixed 9cm-size along the horizontal direction, while, along the vertical one, the homogenized beam size is variable from.8 cm up to 4.5 cm, with a zoom factor larger than 5. The distance between the first lens array and the focal plane is fixed to 2.4 m. As an example, Fig. 3 shows the 3-D profile of the homogenized beam with an intermediate zoom-factor. A comparison of Fig. 3 with Fig. 2 shows that the homogenizer made the beam shape smoother, symmetric and with steeper edges. Also, the substantial reduction of the high-spatial-frequency intensity fluctuations on the plateau is noticeable. Hercules Near Field - Horizontal Hercules Near Field - Vertical CCD Counts 8 7 6 5 4 3 2 1 2 4 6 8 1 12 mm 14 RX side H.V. side CCD Counts 7 6 5 4 3 2 1 2 4 6 8 mm 1 Bottom Top Figure 2. Typical cross-sections of the near-field profile of the XeCl laser Hercules, measured.5m far from the output mirror. Figure 3. 3-D spatial profile of the homogenized beam with an intermediate zoom factor. Vertical scale: CCD counts 3. Results evaluation A correct evaluation of the experimental results requires standardized methods and reliable parameters. Therefore, we have used the ISO document 13694 [2], calculating the plateau uniformity and the edge steepness of the beam measured after passing through the homogenizer. Let us recall that the edge steepness and plateau uniformity are the most meaningful parameters of beam distributions having a nearly flat-top profile [9]. The ISO 13694 defines the edge steepness as
S 1%, 9% = (A 1% - A 9% ) / A 1%, (1) where A 1% and A 9% are the effective irradiation areas with energy density values respectively above 1% and 9% of the maximum energy density value H max, and the plateau uniformity as U P = H / H max, (2) where H is the full width at half maximum of the peak near H max in the energy density histogram curve N(H i ). Here N(H i ) is the number of data points corresponding to a given energy density H i. The width H approaches zero for ideally flat distributions. Figure 4 shows the energy density histogram curve of the beam in Fig. 3. As expected, our results show that the reliability of these parameters is influenced by a number of factors dependent on the set-up, including the quality of bulk neutral-density filters used, the beampath through each attenuator, the dynamic range of the CCD camera, the shot-to-shot stability of the laser pulse. All these factors can be managed to minimize their impact on the final results. Unexpectedly, we also found a reliability problem related to the above standard parameter definitions. In fact, when applying Eqs (1) and (2) to the data of the beam of Figs. 3 and 5, we have S 1%, 9% = 66.8% and U P = 8.8%. Although the beam shown in Fig. 3 is not a perfect top-hat beam, an edgesteepness value in excess of 66% is not consistent with the measured shape of the homogenized beam, as demonstrated in [5]. 11 1 3 Number of pixels 5 1 3 1 Energy density (%H max ) H peak Figure 4. Energy density histogram of the homogenized laser beam shown in Fig. 3. Vertical: number of pixels. Horizontal: energy density (% of H max ). H and H peak are, respectively, the full width at half maximum and the energy density value corresponding to the vertex, of the peak just below the maximum energy density H max. Then, to see if this is a general problem or not, we tested a different laser system, namely the highly coherent second harmonic beam emitted by a diode-pumped Nd-YAG laser. As in the case of the excimer laser previously presented, we measured the Nd-YAG beam shape before and after passing through a transfocal homogenizer by rastering a CCD across the beam. The typical beam shape after the homogenizer is shown in Fig. 5: note the irregular uniformity on the plateau, which is much larger than that of the excimer beam (see Fig. 3). This is because in the Nd-YAG case the beamlets emerging from the divider are mutually coherent, and their overlap on the focal plane produces a speckle pattern, due to interference [5]. Most important, the calculation of the edge steepness by Eq. (1) gives an amazing value of 95%! This happens because there are few pixels giving unreliable high values of intensity (on the right part of the beam in Fig. 5), such that the area corresponding to 9% of the maximum intensity is very small, and then the normalized difference of areas in Eq. (1) is very large.
4 35 3 25 2 15 1 5 2 4 6 8 1 12 Figure 5. Typical cross-section of the homogenized Nd-YAG beam along the horizontal direction. Vertical scale: CCD counts. Horizontal scale: mm. In our opinion, the main reason for this inconsistency is the definition of edge steepness itself [1]. In fact, when referring the effective irradiation areas to energy density values larger than fractions of the absolute maximum value H max, it may happen that a misrepresentative single-pixel value of H max (that is, a wrong value of H max ) can seriously distort the result of Eq. (1). To overcome this problem, when calculating the effective irradiation areas in Eq. (1) we suggest using, in place of H max, the energy density value corresponding to the peak in the histogram curve of the plateau uniformity (see Fig. 4), which we call H peak. This new edge-steepness definition unambiguously gives a value independent of single-pixel malfunctioning, as it refers the effective areas to fractions of a welldefined H value (note that the peak in the histogram curve of the plateau uniformity is very sharp for nearly flat-top beams), which is representative of a large number of photons. Applying the above edge-steepness definition to the beams of Fig. 3 and 5 we obtain, respectively, S 1%, 9% = 43.6%, and S 1%, 9% = 2%, which are more reasonable values, in our opinion, to fit the measured steepness of the examined beams. Clearly, if a malfunctioning pixel gives a wrong value of H max, it cannot be used for the calculation of the plateau uniformity (see Eq. (2)). Therefore, it makes sense normalizing the width H in Eq. (2) to H peak curve instead of H max. Being H peak H max, this new definition systematically increases the plateau uniformity value with respect to the ISO definition. In the excimer beam case, for example, this new definition gives U P = 9.7%, a value 1% larger than that obtained from Eq. (2). 4. Conclusions Reliable standard methods for evaluation of beam parameters are of great importance for a correct characterization of laser beams and for proper comparison between different laser systems performance. Using a CCD camera to monitor the energy density distribution of nearly flat-top shaped excimer and Nd-YAG beams, we found that the standard definition of both edge steepness and plateau uniformity in the document ISO 13694 [2] may overestimate the importance of data coming from few pixels which can alter the value of H max, thus giving unreliable values of these parameters. We have proposed a simple change in the normalizing H value that unambiguously allows overcoming this problem. Acknowledgments One of us (E.T.K.) gratefully acknowledges the grant awarded by the TRIL Program in the frame of the ENEA/ICTP agreement. REFERENCES 1. F. M. Dickey and S. C. Holswade, Laser beam shaping: theory and techniques (M. Dekker, Inc. New York Basel, 2). 2. EN ISO 13694 Test methods for laser beam power (energy) density distribution (April 2).
3. P. Di Lazzaro, S. Bollanti, G. Felici, D. Murra: A novel light beam homogenizer Proc. SPIE 512, 15-155 (23). 4. ENEA patent n IT 1316395 (UIBM, 2); Optical system for the homogenization of light beams, with variable cross-section output : United States patent No.: US 6,639,728 B2 5. S. Bollanti, P. Di Lazzaro, D. Murra: More about the light beam shaping by the integration method Eur. Phys. J. D, in press (24). Published on line DOI: 1.151/epjap:24164T. 6. E. Tefouet Kana, S. Bollanti, P. Di Lazzaro, D. Murra: Beam homogenisation : theory, modelling and application to an excimer laser beam, this Proceeding volume. 7. Letardi, S. Bollanti, P. Di Lazzaro, F. Flora, N. Lisi, C.E. Zheng: Il Nuovo Cimento D 14, 495-57 (1992). See also P. Di Lazzaro: Hercules, an XeCl laser facility for high-intensity irradiation experiments Proc. SPIE 3423, 35-43 (1998). 8. S. Bollanti, P. Di Lazzaro, F. Flora, T. Letardi, N. Lisi, C.E. Zheng: Appl. Phys. B 55, 84-91 (1992). 9. K. Mann, J. Ohlenbusch, V. Westphal: Characterization of excimer laser beam parameters, Proc. SPIE 287, pp. 367-377 (1996). 1. S. Bollanti, P. Di Lazzaro, D. Murra, E. Tefouet Kana, G. Felici: Edge steepness and plateau uniformity of a nearly flat-top-shaped laser beam, Appl. Phys. B 78, 195-198 (24).