Fig. 1a shows a photo of the white spot-stained silicon wafer. Visible white spot stains are shown in the three yellow boxes in the photo. The white spot stain’s irregular morphology was observed by SEM, as shown in Fig. 1b. The spot was about 144 μm in length and 155 μm in width, The surface of the material was distributed with different-sized highlights. EDS element maps are shown in Fig. 1c-f, which show that the white spot stain contained C, Ca, O, and Si elements. C was present in the highest amount, so it is speculated that the white spot stains may be organic matter.
The XPS spectrum was used to further analyze the valence states and composition of the white spot stains. The XPS spectrum in Fig. 2a shows that the white spot stains mainly contained C, Ca, and O, and a small amount of Si. Based on the XPS spectrum in 2a, Ca 2p was deconvoluted into two peaks. The peaks of 347.07 eV and 350.6 eV correspond to Ca 2p3/2 and Ca 2p1/2, respectively (Fig. 2b) [25]. The peaks at 531 eV in the O 1s spectrum and 347.07 eV and 350.6 eV in the Ca 2p spectrum show that the white spot stains contained CaCO3 particles (Fig. 2d). The C 1s spectrum of the white spot stains was decomposed into three components: the peak of C-C was located at 284.7 eV, the peak of C-O-C was at 286.4 eV, and the peak of O-C=O (carbon in carbonate) was located at 289 eV (Fig. 2c) [26]. The Si 2p spectrum presented two characteristic peaks at 99 eV and 103.3 eV, which corresponded to monatomic silicon and SiO2, respectively (Fig. 2e) [27]. During the production of some polymer materials, to ensure their mechanical properties and flame retardancy, CaCO3, SiO2, and other inorganic fillers are often added to the glues, rubbers, and other polymeric materials. The results further confirmed that the white spot stains were an organic polymer [28, 29].
During the slicing of the silicon wafer, the surface of the silicon wafer will only contact the diamond wire. Therefore, the diamond wire was characterized using EDS, to determine the reason for the stains. Fig. 3a shows that the diamond along the EDS line scan changed from a sharp structure to a flat structure, with very serious wear. This indicates it is a used diamond wire. There was a circular substance with a diameter of about 20 μm on the surface of the diamond wire. The EDS analysis of the substance is shown in Fig. 3b, which shows it contained C, Ca, O, and Si. However, the diamond wire was nickel plated, and the main elements were Fe, Ni, and C. This shows that the substance was not derived from the diamond wire. The elements in this substance are consistent with the elements in the white spot stain, showing that it is also a polymeric material.
To determine where the substance on the diamond wire was introduced, the entire slicing process was analyzed. When slicing silicon wafers, sticky stick glue is used to hold the silicon ingots and plastic boards together. The only polymeric materials that can contact the diamond wire are sticky stick glue and plastic board. These two substances were analyzed by EDS. Fig. 4a shows an SEM image of the plastic sheet covered with slice marks caused by diamond wire slicing. The element maps in Figs. 4b and 4c show that it only contained C and O, indicating that the substance was not introduced by the plastic board. Fig. 5a shows the SEM image of sticky stick glue, in which several areas in the figure were brighter than others. The element map in Fig. 5b-e shows that it contained C, Ca, O, and Si. The bright spots in the SEM image are Ca. Sticky stick glue contains the same elements as the white spot stain and diamond wire, indicating that the substance is sticky stick glue.
The process of restoring the white spot stains is shown in Fig 6. Yellow is a salver, red is sticky stick glue, blue is a plastic plate, and gray is a silicon wafer. During the slicing process, the diamond wire slices the silicon ingot from the bottom to the top. Due to the uneven stress at each point of the wire, a line bow phenomenon occurs [30] in which the outer slicing depth is greater than the inner slicing depth. Therefore, the outer side of the silicon ingot is sliced before the inside. To ensure that the diamond wire has enough slicing depth, the silicon ingot can be fully sliced. A plastic plate was added in the middle of the silicon ingot and the crystal bead. When the diamond wire slice the plastic plate, the silicon ingot slice through and contacts sticky stick glue. As a result, a large amount of sticky stick glue adhered to the diamond wire, so when the diamond wire slice the next batch of silicon ingots, sticky stick glue adhered to the silicon wafer and formed stains.
Fig. 7a and 7b show the SEM diagram before and after the deposition of the silicon wafer in the HF/Cu(NO3)2 solution. Fig. 7b shows that stain edges near the red dotted edges were darker than the area outside the red dotted line. Compared with before deposition, the area outside the red dotted line was much brighter, showing that this area was deposited with copper. There were no significant differences in the stain morphology before and after deposition. The stain did not dissolve in the deposition solution. The EDS results show that the distribution of copper outside the red dotted lines was denser than within the dotted line. There were almost no copper deposits on the surface of the stain. At the edges of the red dotted area, only a very small amount of copper was deposited because CaCO3 in the stain reacted with HF in the deposition solution via the equation 2HF + CaCO3 à CaF2↓ + CO2↑ + H2O. The reaction generated CaF2 precipitate, resulting in the rapid reduction of F- around the stain. The pH increased, and the conductivity of the solution decreased. Therefore, F - in the area outside the red dotted line was not added to the reaction interface outside the red solid line. Thus, concentration polarization occurred [31]. F– in the HF solution easily reacted with Si atoms to form Si-F bonds, which greatly reduced the Si-Si bond strength. This made it easier for Cu2+ ions to obtain electrons and be reduced to elemental copper. The F- concentration around the stain decreased, which decreased the rate of Cu2+ deposition, which resulted in uneven copper deposits [32].
Fig. 8 shows an SEM image of a white spot-stained silicon wafer etched in HF/Cu(NO3)2/H2O2 solution at 40 oC for 7 min. There was oval-shaped material at the center of the picture, and the EDS results showed that the material was a stain adhered to the surface of the silicon wafer. After reacting for 7 min in a strongly-oxidizing solution, the stain did not dissolve, and many nanopores were formed by the etching solution in areas not covered by the stain.
Sticky stick glue is an epoxy resin, whose hydrolytic degradation was accelerated by OH- in the KOH solution. This reduced the bonding performance. To understand the effect of KOH solution on stain removal, a metallographic microscope (MO) was used to observe changes in the stain morphology at different cleaning times. Figure 9a shows the MO image of the uncleaned silicon wafer. A large amount of black material stained the surface of the silicon wafer. Fig. 9b shows the stain morphology after cleaning for 1 min. The amount of stain shedding was very small, which exposed the silicon substrate. When cleaned for 3 min, the stain shedding was accelerated, and the stained area gradually shrank. Stains in the upper-left and lower-right corners were removed, and those in the middle part became sparser. After 4 min of cleaning, only a few spots remained on the wafer. At 5 min, no stains were visible under the microscope, and the stains were completely removed.