1. Introduction

Femtosecond laser microfabrication is of high interest in the field of science and technology and of growing relevance for industrial applications [1,2]. The ongoing technological development in ultrashort pulse lasers leads to a cost reduction accompanied with an increase of the average pulse power at higher repetition rates. This makes ultrashort pulse lasers a promising tool for highly precise microfabrication in technological applications. Extraordinarily high peak intensities in combination with a short interaction time between laser light and matter, typically below heat dissipation rates, allow for structures of high quality, while the surrounding material remains largely unaffected. The use of ultrashort laser pulses enables a number of different processing techniques, including removing (i.e. laser micromachining), adding (i.e. laser printing), and modifying (i.e. multiphoton polymerization) a variety of materials including metals, dielectrics, polymers, and semiconductors [3].

Volume optical structures inside photovoltaic modules

In previous works of our group a direct write laser process was applied to create three-dimensional volume optical structures inside the volume of the cross-linked EVA (Ethyl-vinyl acetate) encapsulation material of solar modules leading to increased cell efficiency [4] [5]. Diffractive gratings, fabricated directly above the solar cell grid fingers (front side electrodes) reduced the optical shadowing by deflection of the incident light. This application of volumetric diffractive optical elements in the bulk of solar modules was introduced for the first time by I. Mingareev et al. [6]. Related approaches to enhance the efficiency of solar modules by reducing the optical losses at the grid fingers are described in [7–9].

In terms of optical efficiency our previous results were promising; however, the most significant drawback of the presented femtosecond laser process in [4] was the long sequential writing time. The volume optical microstructures were fabricated by 3D scanning a single tightly focused laser beam, creating a sequential “voxel-by-voxel” pattern. That is limiting the maximum processing speed and consequently prevents the industrial application of the approach. Moreover single focus scanning often requires attenuating the excessive power of the laser source and thus is not an efficient approach in terms of exploiting the available power.

Fast materials processing

Generally, high throughputs of laser processes require feasible strategies for achieving high process speeds. One approach is based on an increase of the processing speed by using high-power, high-repetition-rate lasers and very fast beam scanning techniques such as galvanometer scanners. Another strategy, as presented in this study, utilizes a parallelization of the ultrashort laser writing process. In contrast to serial “voxel-by-voxel” laser structuring, parallel techniques allow, similar as in conventional photolithography, the exposure of an entire micro/nanoscale pattern within a single exposure shot. This leads consequently to an improvement of the processing speed and addresses the large-scale fabrication of volume micro-optics, meeting the requirements of the solar module industry.

In the present work we demonstrate a spatial light modulator (SLM) based holographic laser microfabrication process [10–12] to manufacture functional volume optical structures directly in a solar module. In contrast to our earlier studies we demonstrate the approach on a photovoltaic test sample comprising identical materials and layout as used for industrial modules. The speedup of the fs laser microstructuring process, enabled by the SLM based parallelization technique, indicates a possible route towards an industrial implementation of volumetric diffractive optical elements in solar modules.

2. Spatial light modulator based holographic lithography

SLMs have proven to be powerful tools for ultrashort laser microstructuring in terms of parallelization and fabrication efficiency improvement [13–18]. In the present study, we use a LCOS-SLM (liquid-crystal-on-silicon spatial light modulator) from Hamamatsu (X13139 S) as reflection type phase spatial light modulator. The pixelated liquid crystals of the SLM change only the phase of the light, without any change in intensity. The phase shift induced by each pixel varies from 0 to 2π, corresponding to greyscale values from 0 to 255 for the related phase bitmap images. This leads to a two-dimensional tunable refractive index pattern, which is imprinted as a phase pattern on light reflected on the SLM. An SLM can therefore be considered as computer-programmable diffractive optical element. The variation of phase patterns controls the diffraction and interference of the light reflected from the SLM. As a consequence, a laser beam, being reflected from an SLM, can be shaped to arbitrary desired forms. This offers high flexibility in manipulating the light field in this kind of holographic lithography, where the information required for image formation is spread out over an entire hologram phase pattern. All these features make SLMs perfectly predestined for parallel laser processing by single shot exposures. In particular the available laser power can be exploited more effectively and the processing time is reduced proportional to the amount of parallelization.

The Hamamatsu SLM was chosen for its high laser power handling capacity (specified up to 60W with water cooling). This is an essential precondition when using SLMs for parallelization of microstructuring processes. Due to the use of multilayered dielectric mirrors instead of metal mirrors the reflectivity of the SLM is enhanced, which automatically leads to a reduced internal absorption. As a drawback, dielectric mirrors work only for a limited wavelength range. In our study this wavelength range was chosen from 750 nm to 850 nm in accordance with the laser wavelength.

Experimental laser setup

As a laser source, a commercial ultrafast Ti:Sapphire laser (Spectra Physics), consisting of a Ti:Sapphire oscillator (Maitai) and a regenerative Ti:Sapphire amplifier (Spitfire) was used. The oscillator provides a pulse train at a repetition rate of 80 MHz, seeding an amplifier, which boosts the pulse energy up to 0.1mJ (~100mW max. average power). The amplified pulse duration is approx. 150 fs, the repetition rate is 1 kHz, limited by the pump laser repetition rate of the amplifier.

The polarization of the incident laser beam is set parallel to the orientation of the liquid crystal molecules in the SLM to ensure phase-only modulation. The full area of the SLM display is exploited by using a beam expander with a 3x magnification.

The first-order of the diffracted beam from the SLM (containing the pattern information) has to be imaged from the SLM to the microscope objective aperture. The zero order beam is stopped at the focus of lens L3 by a beam block. Four lenses (L3 - L6) were selected to image the spatially modulated laser beam onto the back focal plane of the microscope objective. The focal lengths of L3 and L4 are chosen to compress the beam, that is reflected from the SLM display, by a factor of f4/f3 = 0.4 (f3 = 500mm, f4 = 200mm) to 5.12 mm, which is just a little bit larger than the microscope objective aperture of 5mm. L5 and L6 (f5 = f6 = 100mm) are used to relay the optical image to the back focal plane of the objective.

As for a lot of microstructuring processes, the fabrication of the volume optical structures relies on strong focusing. Therefore a high NA microscope objective has to be used. We used a 20 × microscope objective (Zeiss Epiplan, NA = 0.4, working distance 4.1 mm) to focus the beam. According to the objective field of view, an area of approximately 320µm × 320µm that can be treated at once. Thus, the high NA of the objective limits the area which can be processed in a single exposure. Structuring larger areas necessitates a movement of the sample between consecutive exposures using a motorized XY translation stage (Aerotech).

Individual sequences, such as hologram displaying on the SLM, translation stage motion or CMOS camera image recording, flip mount mirror switching (changing between He-Ne and Ti:Sapphire laser) and so on were controlled by a program written in LabVIEW (National Instruments). Since the Hamamatsu X10468 series can be controlled via a PC using the digital video interface (DVI), an SLM can be considered as an imaging device connected to the computer, suchlike as a second monitor.

Computer generated holograms (CGHs)

The SLM phase images generating the desired light intensity distributions, for the individual planes of the sample to be structured, are calculated by an iterative Fourier transform algorithm (IFTA) [19,20]. The IFTA is based on the Fourier transform relationship between the complex field at the back focal plane and the image at the front focal plane of a lens. The IFTA enables to calculate phase-only holograms by iteratively propagating the field between the image and the SLM plane using Fourier transforms while enforcing design constraints in the image (desired light intensity distribution) and SLM plane (uniform input laser intensity). The aim of the algorithm is to find a phase pattern at the hologram plane which gives reason for the desired pattern being transferred to the image plane.

Due to the lack of commercially available software providing the required functionalities, we implemented an IFTA variant in MATLAB. The desired intensity distribution, as input for the IFTA calculation, can be arbitrarily chosen as monochrome bitmap image, located as 256×256 pixel area in the center of a black 1024×1024 background image. This guarantees a match with the field of view of the used objective. The calculated phase patterns are loaded as 8-bit grayscale bitmap image to the 1024 × 1024 pixels in the center of the SLM, giving rise to a spatially phase modulated reflected laser beam (0-2π spatial phase shift corresponding to the gray levels from the CGH).

In spatial light modulator based holographic lithography further optical imaging functionalities can be obtained by superposing additional phase images onto the CGHs calculated by the IFTA. By combining the CGHs with superposed gratings and lenses, lateral as well as axial shifts of the images can be obtained without a mechanical movement of the sample. Furthermore, compensation phase masks can be added to correct optical aberrations. Distortions due to unevenness of the LCOS chip are compensated by a flatness calibration file delivered by Hamamatsu.

Autofocus and vision system

The fabrication and embedding of volume optical structures requires alignment of the structures in depth (Z-axis). Hence the experimental setup has to provide an autofocus and vision system. Both functions are needed: autofocusing for determining the optical interfaces in the sample, as well as the function of a vision system for locating the correct lateral positions of the electrode fingers in the sample. Both functions can be realized by a camera based system.

The basic principle of the experimentally realized autofocus system relies on a system described in [21]. The method is based on a contrast measurement of projected test illumination patterns. Special test patterns are displayed on the SLM, which is illuminated with He Ne laser light (center wavelength 633 nm, average power ~2 mW). Thereby, an image of the test pattern (regularly distributed isolated spots) is created in the focal area of the microscope objective. This light is reflected at any optical interface in the sample. This back reflected light is detected with a compact USB CMOS camera (Thorlabs DCC1545M, 1280×1024 monochrome sensor), utilizing a beam splitter, which is placed in front of the focusing objective (see Fig. 1). This system is noninvasive in a sense, that the autofocus light (He Ne laser) does not affect the optical properties of the sample. In order to determine the focal position, the microscope objective is gradually moved in the Z-direction along its optical axis and images are taken at defined positions. For the recorded images, sharpness values are calculated. The objective position giving reason for the image with the largest value for the image sharpness correlates with the position of the optical interface of interest. The whole procedure has been implemented as automated sequence in LabVIEW.

 

Fig. 1 Optical setup of the spatial light modulator based holographic laser microfabrication system. After passing the SLM, the light propagates to the back focal plane of an objective through four lenses (L3-L6). The first lens is located about one focal length away from the SLM to perform the Fourier transform.

Download Full Size | PDF

The camera system implemented in the spatial light modulator based holographic laser microfabrication system was also used as vision system to determine the spatial positions of the optical elements in the sample. In particular the positions of the grid lines on the solar cells were identified. Consequently a lateral positioning of the laser beam relative to the grid lines was possible. That enabled a positioning of the volume optical gratings directly above the grid wires of the solar cells, which deflect impinging light away from the grid fingers.

Test sample set-up

Two different types of test samples have been used for investigating the volume optical elements. The first type of test samples comprised of 6 mm thick sheets of PDMS (Sylgard 184). Those test samples have been used to conduct first structuring tests, enabling us to fabricate samples for optical microscopy as well as samples for optical characterizations of the volume optics. In particular transmission measurements by using a LAMBDA 900 Spectrophotometer from Perkin Elmer have been performed, using the transmission mode as well as an integrating sphere for the measurements.

A second type of test samples comprised of a laminated industrial solar cell made of crystalline silicon material (area = 156×156 mm²) having screen printed electrodes with a grid finger width of about 100 µm and a spacing of about 2.2 mm. That single cell was laminated behind a 3.2 mm thick glass plate made of low-iron white glass. The encapsulant used was EVA and the thickness of the EVA sheet is about 0.5 mm. On the rear side a conventional back-sheet for photovoltaic module encapsulation was applied (ICOSOLAR). A photo of the photovoltaic module test sample is shown in Fig. 2.

 

Fig. 2 Laminated industrial solar cell made of crystalline silicon material (area = 156×156 mm²) having screen printed electrodes with a grid finger width of about 100 µm and a spacing of about 2.2 mm.

Download Full Size | PDF

3. Volume optics inside solar modules

Sequential–parallel microfabrication

The volume structuring process, described more in detail in our previous study [4] relies on a threshold effect: a minimum intensity must be overcome at each focal point in order to create the desired volume optics by a local change of the optical properties of EVA. Since the femtosecond laser beam is split by the SLM into a multitude of laser spots, the laser power of the original beam is distributed over many spots and the power at each individual spot is reduced accordingly. Thus, the power of the individual spots is controlled via the extent of parallelization, i.e. the number of foci which are simultaneously created.

The aimed structure for the volume optics is, in accordance with [4] and [6] a diffraction grating. In particular dotted line gratings with a pitch in the range of a few µm or below are targeted at in the context of deflective optical elements for sunlight. Figure 3(a) shows a grayscale bitmap image (256×256 pixels) containing the desired grating pattern. The image consists of 2688 spots. For a single exposure (2688 spots simultaneously) an average laser power of several Watts (~2.7W) would be required. Since the available power of the laser system used for the experiments was only ~100mW, the illuminated area had to be reduced. Exposure tests were necessary, in order to determine the suitable number of laser spots for one parallel exposure, utilizing the full available laser power. It was found that a number of 100 laser spots per single exposure delivered good results at the available laser power. Given the available power of the laser system (~100mW) in combination with a spatially splitting of the fs beam to 100 spots, the input laser power for each focus spot could be estimated to be on the order of 0.5 mW, taking into account the limited diffraction efficiency of the Hamamstus SLM as well as optical loss along the way from the laser output to the sample surface.

 

Fig. 3 Diffraction grating as volume optical element above the grid fingers of the solar cells: (a) grayscale bitmap image (256x256 pixels) containing the aimed grating pattern, the pixels of white (2688 pixels) are the positions of the laser spots for one single exposure (b) random 100 pixel fragment taken from the grating structure in (a) as input for the IFTA, (c) 1024x1024 CGH calculated for the random 100 pixel fragment from (b)

Download Full Size | PDF

Therefore the desired grating pattern was segmented to fragments containing 100 pixels each - see one such a fragment in Fig. 3(b). This sampling is automatically performed by a MATLAB routine, which creates sub-images by randomly choosing 100 pixels from the original image. These bitmap images serve as input for the IFTA, leading to phase images, which create the structure in the focal area of the 20× objective. For each of these image fragments a corresponding phase hologram - see one such a fragment in Fig. 3(c) - is compiled into a ‘movie’ format, stored as avi-file. This phase hologram movie was displayed on the SLM in order to create the desired microscopic pattern. In our case a series of 27 phase holograms successively generates the whole grating structure of Fig. 3(a). This method represents a sequential–parallel approach, sequentially displaying arrays of diffractive beams, where each array consists of 100 beams in parallel.

Besides the laser intensity, which is predetermined by the available incident laser intensity (~100mW) in combination with the number of pixels per hologram fragment, the movie frame rate is the second important parameter affecting feature size and pattern transfer. The frame rate of the hologram movie determines the exposure time. The maximal frame rate of the used SLM is 60 Hz, corresponding to an exposure time of ~17ms.

Test gratings in PDMS

Different encapsulation materials are used in photovoltaic modules [22]. When photovoltaic panels were first developed in the 1960s and the 1970s, the dominant encapsulants were based on polydimethylsiloxane (PDMS). Polydimethylsiloxane (PDMS) belongs to a group of polymeric organosilicon compounds that are commonly referred to as silicones. The currently dominant encapsulant chosen for PV applications mainly due to economical reasons, is Ethylene-co-vinyl acetate (EVA). First structuring tests were performed using PDMS samples, as handling and especially optical characterization of structures in PDMS samples are more accessible as compared to whole solar module samples. This is feasible as it was found that the material properties regarding the formation of volume optics in EVA are comparable to the results in PDMS.

Figure 4 shows microscopic images of fabricated optical gratings in PDMS for the grating structure from Fig. 3(a). The volume gratings were inscribed in a depth of 1.5 mm below the top surface of the PDMS block. The zero order beam was blocked at L3. The shadowing caused of the beam block in the image center, was compensated by a defocusing approach that was realized by adding an additional lens function to the CGH. Figure 4(a) shows a test exposure matrix varying the hologram movie frame rate from 1 to 29 Hz (starting in the upper left corner) in steps of 2 Hz. At frame rates > 29 Hz the exposure time was too low for the laser power available and the volume optical structures were less pronounced. It can be seen, that with increasing frame rate, corresponding to decreasing exposure times, the grating structures in PDMS become less pronounced. Figure 4(b) shows a close up microscope image of the grating corresponding to a frame rate of 15 Hz. At these settings, the exposure of the whole grating takes about 1.7 seconds. The overall dimension of one field is around 106 µm x 333 µm, the grating constant is around 5 µm and the lateral sizes of the volume optical structures are around 2 µm. We did not explicitly measure the uniformity of the exposed grating structure. However, the microscopic images of fabricated optical gratings in PDMS (Fig. 4) in combination with the light beam induced photocurrent measurement results (Fig. 9) demonstrate (as an indirect evidence for the uniformity of the laser intensity in each foci) that the uniformity is sufficient for the practical use in solar modules.

 

Fig. 4 Microscopic images of fabricated optical gratings (using the bitmap in Fig. 3 (a) as input) in PDMS: (a) exposure test matrix using exposure frame rates from 1 – 29 Hz in steps of 2 Hz (starting in the upper left corner), (b) close up of the central pattern from (a) exposed with a frame rate of 15Hz

Download Full Size | PDF

For structuring areas of larger dimensions the sample must be moved between consecutive exposures. Several exposure fields were fabricated and two different types of volume gratings in PDMS were created. Namely a “single” and a “double” grating structure were made. The single grating structure is realized by stitching together the grating shown in Fig. 3(a) using a frame rate of 15 Hz. The gratings exhibit an area of 5x5 mm² and fabrication time was around 30 minutes. The double grating structure resembles the single grating structure, with the difference that here the grating layers are fabricated twice with a vertical displacement of 150 µm.

Figure 5(a) shows a PDMS block with two 5×5 mm² volume gratings, illuminated with a He Ne laser beam, and the corresponding scattering pattern on a white screen, indicating clearly the optical behavior of a quadratic diffraction grating. Measuring the transmission of just the zero order non-scattered beam - see Fig. 5(b) - delivers a first estimation of the deflection efficiency. At a wavelength of 600 nm about 68% of a direct beam is transmitted which is about three quarters of the total transmission measured to be about 89%. Thus about one quarter of the transmitted light is scattered or is diffracted into non-zero orders. For a double grating the total transmission is about 85% and the direct beam transmission is about 48%. Consequently about half of the transmitted light is deflected from the zero-order beam.

 

Fig. 5 Grating structures (5×5mm²), corresponding to Fig. 4 (b), in the volume of a PDMS block: (a) Diffraction of a He Ne laser beam on the volume structures, indicating clearly the optical behavior of a quadratic diffraction grating, (b) Transmission of the zero order beam and total integrated transmission for a single and double grating structure in PDMS.

Download Full Size | PDF

Please note that the experimental determination of the diffraction efficiency is not trivial for volume gratings. Since light diffracted at the grating has to exit the volume for being detected, internal reflection of light at the bulk-air interface significantly influence the measurement results. In particular the transmission of light deflected at an angle > the angle of total internal reflection would not be detected in a conventional transmission measurement. Ideally the diffraction efficiency of a volume grating would be measured by a detector in the same optical medium as the grating. An encapsulated solar cell can be regarded to be such a detector for volume optics in the encapsulant.

Gratings in the bulk of the encapsulant of solar modules

Following our previous works studies [4] we fabricated the optical gratings, tested in PDMS (see Figs. 2-4), directly in the bulk of solar module test samples, utilizing the described parallel spatial light modulator based holographic lithography method.

The sample consists of an industrial solar cell made of crystalline Silicon having dimensions of 156×156 mm², an encapsulant, a cover glass as well as a back sheet. The solar cell has screen-printed front side electrodes (at the side facing the sunlight) forming a grid pattern of fingers (lines) with a width of about 100 µm and a spacing of about 2 mm. Light incident on the grid fingers is either absorbed or reflected and does not contribute significantly to the photocurrent generation of the solar cell. The cell with the encapsulant is finally sealed with a cover glass comprising low iron white glass with a thickness of 3.2 mm and lateral dimensions of about 190x190 mm². As encapsulant material Ethylene-vinyl acetate (EVA), which is a transparent polymeric material used in most of today’s industrially produced solar modules, was chosen. On the rear side of the solar module test sample a back sheet is applied which is also used in industrially produced solar modules. A cross sectional sketch of the solar module set-up used for the experiments is shown in Fig. 6. The results presented in the following refer to “double” gratings, as these structures lead, in accordance to Fig. 5(b), also to more pronounced effects in solar modules.

 

Fig. 6 A cross sectional sketch of the solar cell test module used for the experiments.

Download Full Size | PDF

Figure 7 illustrates the principle of fabrication of the optical microstructures in the volume of the EVA material. The solar module was mounted on the XY translation stage, whereas the objective is mounted on the Z axis of this stage. Laser light reflected from the SLM, containing the pattern information, was focused through the cover glass into the EVA material by using the 20× microscopic objective. The working distance of 4.1 mm of this objective enables to focus the laser beam through the upper glass plane of 3.2 mm. In addition, it allows for a high resolution and minimal optical aberrations in the laser focus.

 

Fig. 7 Schematic of the optical microstructures fabrication inside the solar modules by direct femtosecond-laser writing. The volume optical elements are created approximately 120 and 270µm above the grid fingers.

Download Full Size | PDF

In the focus of the fs-laser beam the EVA is modified, giving rise to the formation of volumetric diffractive optical elements. The tight focusing of femtosecond pulses inside transparent materials, accompanied with very high intensities inside the laser beam focus, induces nonlinear optical processes like multi-photon absorption and thereby a modification of the material parameters. Details concerning the fabrication of volume optical elements in the bulk of EVA by fs-laser induced material modification are discussed in [4]. In this work the material characteristics of the volume optics were investigated by applying confocal Raman microscopic characterization, which indicates that the EVA material partially degraded upon the impact of the laser beam and is partly carbonized. Three-dimensional optical microstructures are inscribed at defined positions, determined by using the autofocus and vision system described above, inside the bulk of the EVA material. We found that for EVA a frame rate of 5 Hz (corresponding to an exposure time of 200ms or 200 laser pulses per hologram frame) delivered best results. The sizes of the volume optical structures are comparable to structures in PDMS (see Fig. 4(b)). The gratings are located approximately 120 and 270 µm above the grid fingers.

In Fig. 8 a microscope image of a grid finger in the solar cell test module is shown. At the position of the volume optics (double gratings) the grid finger appears to be “cloaked” which is a result of light being deflected away from the grid finger. That qualitatively demonstrates the feasibility of the volume optics created by parallelization.

 

Fig. 8 Microscope image (reflection mode), top view to the surface of the solar cell, showing a screen-printed grid finger (front side electrode) running in a horizontal direction. Above this grid finger we fabricated two “double” gratings in the bulk of the EVA encapsulant of the solar module. The double gratings, indicated by the two indicator lines, appears to be “cloaked” which is a result of light being deflected away from the grid finger.

Download Full Size | PDF

Photocurrent mapping

As already applied in earlier studies [5] a photocurrent mapping of the solar test module was used to determine the optical efficiency effected by the laser processed volume optical elements inside the encapsulant of the solar module. In particular we performed laser beam induced photocurrent (LBIC) measurements, where the photocurrent of the solar cell was measured when illuminating the test sample with a laser beam. By this a map of the photocurrent of the test sample could be derived. In particular mappings around grid fingers were performed in order to quantify the decrease in photocurrent at the grid finger positions and derive a figure of merit for their optical losses. When the laser beam is impinging on an active solar cell area (in between the grid lines) the photocurrent is maximal and the corresponding values are used as reference. When the laser beam hits the grid fingers the photocurrent is significantly lower and the loss in photocurrent corresponds to optical losses due to the grid fingers. The volume optical elements (diffractive gratings) reduce the optical losses at the grid fingers, which is measured as a lower relative decrease in photocurrent at the positions of the grid fingers.

The photocurrent measurement setup basically consists of the experimental setup shown in Fig. 1. However, for measuring the photocurrents we integrated a picoamperemeter (Keithley 2400) into our setup and made the appropriate modifications to the LabVIEW program, which takes care of XY-stage motion and the photocurrent data acquisition. As laser source for local solar cell illumination we used the He Ne laser (~2 mW @ 633 nm). The SLM in the setup shown in Fig. 1 is used in the case of an LBIC measurement as plane mirror, by displaying a constant phase image on it. This generates a single spot pattern in the focal plane of the objective. The laser beam was focused through the solar module onto the surface of the solar cell, using the 20× microscope objective which was also used for fabricating the grating structures. Focusing was performed by using the above described autofocus system. identifies the current reducing effects of grid fingers on the one hand and the reduction in the loss of photo current due to the “double” grating above the grid fingers. Obviously at the reference grid finger the photocurrent decreases by 0.5 compared to the photocurrent at the active solar cell area (reference value = 1). In contrast the optical losses at the grid line with a double grating are only 0.2. Consequently the “double” grating structure reduces the optical losses at the grid fingers by more than 50%. Analog experiments performed with “single” gratings fabricated 120 µm above the grid fingers in the solar cell module lead to loss reductions of around 35%. These findings roughly confirm the grating characterization measurements in Fig. 5(b). Thereby these results suggest that the fabricated optical structures in the bulk of the solar modules work quite efficiently as diffraction gratings: A significant amount of the incident light is obviously deflected onto the active area of the solar cell, where it is absorbed.

Figure 9 shows the result of a light beam induced photocurrent measurement. The measured 2D photocurrent map (2.5 µm measurement step size) clearly show that the photocurrent at grid finger positions is significantly increased when volume optics are located above.

 

Fig. 9 Results of a LBIC measurement: (a) two-dimensional photocurrent map around a solar cell grid finger. Photocurrent measurements corresponding to active solar cell areas (no grid finger) were normalized to 1. The grid finger is found at Y values between 100 and 200µm. The left part of the grid finger (X Distance ~0-120 µm) is covered with a “double” grating structure, the remaining part of the grid finger was used as reference. The measured photocurrent is coded in terms of color (see color bar), (b) photo current line scans across the grid finger, comparing the positions at X = 50µm (grid finger covered with a “double” grating) and X = 167 µm (grid finger without a “double” grating)

Download Full Size | PDF

4. Conclusions

In the present study we applied an SLM for parallelizing the laser based fabrication of volumetric diffractive optical elements in the bulk of solar modules. The structuring method enables the full utilization of the available laser power. In contrast to a relatively time-consuming sequential laser structuring approach of a scanning laser beam, this approach increases the processing speed of the microstructuring through parallel processing. This is achieved in the present study by parallelizing the microstructuring process, using simultaneously 100 spots. Using the SLM technique an optical grating (5x5 mm²) in the volume of a transparent polymer (PDMS) could be fabricated in 30 min. Applying sequential fabrication using single laser pulses with 1 kHz repetition rate, structuring the same area, would take at least the threefold processing time. This findings corresponds to an acceleration factor of about 3. That acceleration factor is enabled by a more efficient exploitation of the maximal available laser power (about 100 mW for the laser source used). In particular in a sequential laser writing process the pulse energy needs to be adjusted (reduced) for meeting the desired quality of material processing. When applying a parallelization by using an SLM the maximum available laser power can be applied and “excess” energy can be used for parallelization of the process.

The feasibility of the SLM based lithography method was demonstrated by the fabrication of functional diffractive optical elements directly in the volume of solar modules. Laser beam induced photocurrent measurement indicated that the fabricated structures decrease efficiently the shadowing losses induced by the front side contacts on the solar cells. When taking into account that about 5% of the solar cell area is covered with grid fingers the application of volume optics offers a potential increase in total photocurrent of about 2-3% (taking into account an angle of incidence of 0°). For further evaluation of the feasibility of the method, volume optic needs to applied on larger areas in order to characterize potential total current improvements of solar cells. Moreover, the influence of the angle of incidence needs to be investigated.

Considering an industrial fs-laser source with an average laser power of 500 W the process speed achieved in the presented results can be projected to be improvable by a factor of 5000. Consequently a volume grating of 5x5 mm² could be created in about 0.36 s and the processing speed would correspond to 69.5 mm²/s. Thus gratings with a total width of 100 µm (corresponding to the width of solar cell grid fingers) could be made at a processing speed of 695 mm/s. An industrially fabricated solar module typically consists of 60 solar cells with about 78 grid fingers having a length of 156 mm each. The corresponding processing time would be 17.5 min for a standard module when extrapolating the results shown above to a high power laser source of 500 W. Such processing time would be compatible to the processing time required for laminating the module. In addition to the high laser power SLMs are required which are able to deal with those high powers. Recently Hammamtsu announced an SLM with a high laser power handling capacity specified up to 100W (with water cooling). Utilizing an industrial fs-laser with an average laser power of 500 W requires a splitting of the laser power to 5 independent SLMs. That allows the conclusion, that laser fabricated volume optics for improving the photocurrent of solar cell modules can be a feasible approach even on industrial scale when applying high power laser sources and parallelization techniques such as the SLM based approach presented here.