Etch time factors of 0, 1.25, 2.50, 5.0, 10.0, and 15.0 were tested by photographing the same sample position under the microscope at several magnifications after each etching treatment. The first four consecutive steps are shown in figures 14(a) to (d); detailed explanations are presented with each photomicrograph. The lateral etch rate for etch time factors greater than one can be expressed by a linear curve. For the samples shown, the lateral etch rate is 0.4 um/min, but the rate depends on crack width. From the photographs and the data presented in table 2, we concluded that an etch time factor of 1.25 is best for maximum planar resolution on samples with a high density of defects. A factor of 2.5 widens the defect site considerably, making it more readily visible; also, edge defects not previously noticeable become clearly visible along the periphery of the aluminum pattern. Larger etch time factors have little merit since they decrease the detection resolution. Figures 15(a) and (b) are the samples from figures 14 (c) and (d) shown at a higher magnification. They show clearly that the glass over the large aluminum areas, but not over the line interconnects, had cracked along the edges. The series of photomicrographs presented in figures 16(a) to (d) shows the growth of demarcation areas around pinholes for etch time factors of 0, 1.25, 2.5, and 5.0. Note that the pinhole density remains constant on prolonged etching. Again, the lateral etch rate depends on the defect size. 4.1.5.4 Ultrasonic Etching The effect of ultrasonic agitation on the etching at 50°C of aluminum under cracked glass was investigated by using IC samples that had been etched under normal conditions with an etch time factor of 1.25. No new defects became visible as a result of continued etching with ultrasonic agitation. However, the protruding glass film overhang above the etched-out aluminum was removed by this treatment, leading to irregularly tapered aluminum edges. The aluminum etch rate increased by a factor of approximately three. No advantage was found by this technique. 4.1.5.5 Vacuum Impregnation - Impregnation with etchant under reduced pressure (0.3 torr) was tested to see whether removal of air from the defects prior to etching could increase the defect detection sensitivity. A pre-etched sample was used, as above. No new defects could be discovered by this technique. The aluminum etch rate increased threefold, probably due to increased speed of penetration of the etchant. 4.1.5.6 Defect Detection Sensitivity - We have clearly shown that the rate of lateral etching during metal demarcation is dependent on the size of the defect opening in the dielectric layer, regardless of whether the defect site is a pinhole or a microcrack. The smaller this opening, the slower the penetration, and hence, the lower the rates of mass transport and etching. It follows that at a certain limiting dimension the penetration should become nil. SEM studies of microcracks at their Figure 15. Detailed optical photomicrographs of demarcation-etched magnification. magnification. Figure 16. Optical photomicrograph of pinholes in a PSG layer over terminal thinning-out ends have shown that the limiting crack width for a l-um-thick PSG layer over aluminum interconnects is below 500 A. In other words, a crack width of 500 A still allows the aluminum etchant to penetrate and etch laterally at a sufficient rate to demarcate the defect under the typical conditions of selective etching. Extending the etching time to correspond to etch time factors of 5 to 15 should aid in further increasing the detection sensitivity, if this should be necessary. 4.1.5.7 Other Aluminum Etchants Several other selective etchants for aluminum were tested at 50°C including sodium hydroxide solutions containing high-pH stable and active surfactants, and aqueous hydrochloric acid solutions containing low-pH stable and active surfactants. The sodium hydroxide et chants behaved similarly to the standard phosphoric acid-based aluminum etchant but were not superior to it, whereas the hydrochloric acid etchants operated erratically and etched nonuniformly. 4.1.6 Selective Demarcation Etching of Other Metal/Insulator Structures The metal demarcation etching technique is applicable for testing other types of metallization and insulator systems as long as the metal etchant is selective in not attacking the insulator coating or the substrate. Using the principle of this method, we have successfully analyzed structural localized defects in layers of silicon dioxide, PSG, borosilicate glasses, aluminum oxide, and silicon nitride over numerous metal substrates or delineated metal film. These metals and their etchants used include the following: tungsten and molybdenum etched with an aqueous solution of potassium ferricyanide and potassium hydroxide; platinum etched with aqua regia, gold etched with an aqueous solution of potassium iodide and iodine; chromium etched in a mixture of saturated ceric sulfate solution and concentrated nitric acid; copper with ferric chloride solution; and nickel with diluted nitric acid. (It should be noted in this connection that a nickel etchant consisting of nitric acid, acetic acid, and acetone must never be used because mixing nitric acid with acetone causes a violent reaction.) These etchants, used at appropriate concentrations, temperature, and time are suitable for defect demarcation testing. A brief summary is presented in table 3. The method of demarcation by selective metal etching described in section 4.1 is capable of detecting only defects that locally expose the sublayer of metal. It would be highly desirable to have an additional method available for determining the density of buried or partial defects that extend only partially into the insulator layer, because such defects can seriously weaken the dielectric strength and protective properties |