May_EDFA_Digital

edfas.org ELECTRONIC DEVICE FAILURE ANALYSIS | VOLUME 21 NO. 2 6 In addition to the stronger absorption for 532-nmand 1064-nm lasers, the spot size/diameter of the laser spot (d) is also smaller for both lasers compared to a 1340-nm laser because the spot size is directly proportional to the wavelength. Power density (4*power/[π*d 2 ]) is, in turn, inversely proportional to the square of the wavelength. For the same laser power, the power density of a 532-nm laser is approximately 6.3 times larger than that of a 1340-nm laser. Similarly, the power density of a 1064-nm laser is approximately 1.59 times larger than that of a 1340-nm laser. Larger power density means that more heat is generated at the defect site, resulting in a higher temperature rise and a stronger TIVA signal for both 532-nm and 1064-nm lasers. Because the spatial resolution is directly proportional to the wavelength of a laser, the spatial resolution of a 532-nm and a 1064-nm laser is also better than that of a 1340-nm laser due to smaller wavelengths. In summary, visible and 1064-nm lasers can provide stronger signals and better spatial resolution in TIVA measurements. EXAMPLE 1: TIVA MEASUREMENTS ON A METAL COMB TEST STRUCTURE [4] Figure 4 shows the comparison of a TIVA signal obtained from532-nmand 1340-nm lasers. The TIVAmeasurements were performed using a 20X objective and a laser power of 2 mW for both 532-nm and 1340-nm lasers. The TIVA images from both lasers look very similar and the signals are dominated primarily by the localized heating. There are little or no LIVA effects (photocurrent generation) to obscure the TIVA contrast. Photocurrent generation does not dominate in this structure because the circuit path does not involve any underlying transistors. The 532-nm laser, however, provides stronger TIVA signals with a stronger bright and dark contrast. The defect sitewith the 532-nmTIVA image is also more resolved than the 1340-nmTIVA image, particularly at the defect site (denotedby the yellow arrow) where a distinct boundary is observed in the 532-nm image. In compari- son, the boundary at the defect site is not as well defined for the 1340-nm TIVA image. EXAMPLE 2: TIVA MEASUREMENTS ON AN IC WITH DENSE METALLIZATION Figure 5 shows a comparison of TIVA images taken on an IC with several layers of dense metallization. All scans were per- formed using a 5X objective at 31.2 seconds per framewith a four-frame average. Figures 5a and 5b compare scans with 532-nm and 1340-nm lasers, respectively, at the same power (25 mW). It is apparent that signals from the defect (indicated by the yellow arrow) are stronger for the 532-nm laser with significantly less noise. While it is useful to know that the signal generated by a 532-nm laser is stronger than that of a 1340-nm laser at the same power Fig. 4 Medium-magnification (20X objective) reflected- light image (top left) and corresponding 532-nm(top right), and 1340-nm (bottom left) TIVA images. Fig. 5 TIVA images collectedwith (a) 532-nm laser at 25mW, (b) 1340-nm laser at 25mW, and (c) 1340-nm laser at 100mW. Reflected-light image of the device showing the dense metallization is shown in (d). (a) (b) (c) (d)