Parallel processing

After determining the optimum parameters for producing a simple circuit with a single beam, the SLM was inserted into the optical path and used to parallel process the ITO coated glass sample. To ensure no overlapping between circuits occurred the zero order beam was blocked leaving the +/- first order beams to be used in processing.

The transmitted power of each beam is determined by the CGH, therefore before processing, the grayscale was varied in order to minimise the power distribution between the first order beams, as shown in table 19. There are two grayscale controls available for adjusting the SLM output, each of these affect the contrast of the adjacent grating periods, therein altering the path of the reflected rays. In this test, grayscale one (G1) was varied whilst grayscale two (G2) remained constant. As grayscale two was set to zero, the contrast between adjacent periods was increased with increasing values of G1. The change in contrast between the grating periods affects the interference of the reflected rays. By varying the contrast we were able to identify which parameters produced the most uniform power distribution.

Table 19: Variation of transmitted power when altering G1. Only the first orders were used in processing, however zero order data has been included for completeness.

Variation in power between diffracted first orders when G1 was varied

G1 -1 (mW) 0 (mW) +1 (mW) 210 183 228 124 200 183 236 134 180 179 255 152 160 165 288 167 140 148 337 168 130 135 319 159 120 122 390 155 100 94 465 129 80 66 554 93 60 39 625 57 40 19 627 26 20 5 650 9

Table 19 shows that varying the contrast between G1 and G2 has a significant effect on the transmitted power distribution of the diffracted orders. However, equivalent power distribution between orders was not possible. For parallel processing to be used as a micro manufacturing technique, the variation in power needs to be taken into consideration when determining the processing parameters, as variation in uniformity can affect the quality of processing.

By utilising SUSP processing it is possible to negate the effects of non-uniformity in transmitted beams. By increasing the overall fluence the user can ensure that there is sufficient power in both diffracted beams to achieve materials processing, provided that the highest fluence remains within the selective processing window.

powers. Figure 82 shows two circuits fabricated using a grayscale value of 160, a fluence of 4.24 Jcm-2 and a traverse speed of 32 mm/s, these values are below the multi-pulse ablation threshold value for glass (as determined in chapter 4). This greyscale produced highly uniform beams with a variation in intensity between orders of <2 %.

This test was repeated with a grayscale value of 100; providing a greater disparity in transmitted powers (30% between orders). The conductivity of the circuits produced using both uniform and non-uniform beams were checked with a potentiometer; no conductivity between processed and unprocessed ITO regions was detected for either circuit. Figure 83 shows the white light interferometry profile of the processed regions for the parallel circuits produced using non-uniform first order beams. Generally, disparity in beam power would lead to a difference in the amount of ITO removed by each beam; however we have shown that when using SUSP the average depth of film removal produced by each beam is in good agreement. In addition no damage to the substrate was observed despite the higher fluence of one of the diffracted beams.

Figure 83: White light interferometry profile of circuits produced using non-uniform beams. These circuits were produced using first order diffracted beams whilst the zero order beam was blocked. This prevents overlapping of the circuits.

5.2.3 Summary

Herein, we have shown that the selective processing parameters for ITO coated glass samples, as determined previously (chapter 4); can be applied for use in parallel processing of small circuits onto ITO glass samples. This was achieved by determining the

optimum pulse overlap to ensure no cross boundary conductivity occurred. An electrically addressed SLM was used in conjunction with a CGH in order to generate multiple beams. By varying the number of pixels in the grating period of the LCD we were able to successfully direct both the zero order and first order beams towards the processing region. Using multiple beams the applicability of SUSP in parallel processing with both uniform and non-uniform first order diffracted beams was demonstrated. In order to achieve this we analysed the effect of the varying the contrast between grating periods on the uniformity of the transmitted power distribution. Thereby we were able to select appropriate grayscale values for the production of both uniform and non-uniform beams.

Uniform and non-uniform beams were used in the parallel processing of a simple circuit; in the case of non-beam uniformity SUSP was utilised to ensure that the circuit was successfully manufactured, despite the disparity between the first order beams. The resulting circuit was successfully processed using beams with intensity variations of ≈1% and 30%. Examination of the processed areas shows uniform material removal from the surface using both uniform and non-uniform beams. Using a spatial light modulator and selective processing the time for circuit fabrication has been reduced by half; by increasing the number of diffracted orders this could be further increased providing a high speed micro processing solution.

This case study showed how SUSP processing can be incorporated within a different micro processing technique, in this instance parallel processing. The results obtained in this study imply that SUSP could be used in conjunction with more complex CGHs to produce a greater number of beams. This would reduce manufacturing time and costs whilst still achieving the same high quality results observed in etching and LP/CW laser direct write techniques.