III-I. Etching of ONONstack using C4F8, C4F6, and C4H2F6-based gases
Figure 2 (a) shows etch rates of blank SiO2, SiNx, and ONON stack, and Figure 2 (b) shows the etch selectivity between SiNx and SiO2 using C4H2F6, C4F6, and C4F8-based gases. As the process conditions, 2000 W of source power, -60 V of DC bias voltage, and additive gases composed of CF4/O2/Ar were used. The process conditions were selected through the maskless ONON stack feature etching with C4H2F6-based gas (see supplementary information Figure S3). For C4F8 and C4F6-based gases, which are the fluorocarbon gases generally used for staircase etching, the optimized etch conditions were similar to C4H2F6-based gas as functions of process variables except for C4F8 and C4F6 gas flow rates. As shown in Fig. 2(a), for blank sample etching with this process conditions, all three gases showed similar etch rates of 72 ~ 81 nm/min for ONON stack even though the ONON stack was the highest for C4H2F6-based gas. On the contrary, the SiO2 etch rates were higher in the sequence of C4H2F6, C4F6, and C4F8-based gases by showing 72, 87, and 92 nm/min, respectively, and SiNx etch rates were higher in the reverse sequence of C4F8, C4F6, and C4H2F6-based gases, respectively. The highest SiNx etch rates and the lowest SiO2 etch rates observed for C4H2F6-based gas compared to C4F8 and C4F6-based gases are believed to be related to existence of hydrogen in the gas chemical formula which assists SiNx etching by forming NHx but decreases SiO2 etching by forming a thicker fluorocarbon layer on the SiO2 surface. The etch selectivities of SiO2/SiNx estimated by the blank sample etching of SiO2 and SiNx were 1.25, 0.62, and 0.56 for C4H2F6, C4F6, and C4F8, respectively, as shown in Fig. 2(b), therefore, the higher etch rate of ONON stack observed for C4H2F6-based gas compared to C4F6-based and C4F8-based gases in Figure 2(a) was related to the similar etch rates between SiO2 and SiNx.
The ONON stack having maskless ~510 nm thick ONON pre-etched feature shown in Figure 1 (c) was etched using the C4H2F6, C4F6, and C4F8-based gases with the process conditions in Figure 2. Figure 3 shows cross-sectional SEM images of maskless ONON stack feature etched about 300 nm, 600 nm, and 1000 nm with (a) C4H2F6, (b) C4F6, and (c) C4F8-based gases. The other etch conditions are the same as those in Figure 2. When comparing the shape differences between the reference (pre-etch) feature profile and etched ONON stack feature, the ONON stack feature etched with C4H2F6-based gas showed a square shaped feature without changing the feature shape significantly during the etching of 1000 nm thick ONON stack. On the contrary, C4F8-based gas exhibited gradual change in feature size with etch depth and finally showed a thin trapezoidal shaped feature after etching 1000 nm thick ONON stack while C4F6-based gas showed intermediate feature shape change. Additionally, the most significant trenching phenomenon was observed for C4F6-based gas.
Using SEM images, the top/bottom CDs and sidewall angles of maskless ONON stack feature etched ~1000 nm with C4H2F6, C4F6, and C4F8-based gases were measured as shown in Fig.4 (a), and the results on sidewall angles, CD differences (ref top CD – top CD of etched structure), and (bottom CD – top CD)/2 of the etched features for C4H2F6, C4F6, and C4F8-based gases are shown in Figure 4(b). As shown in Fig. 4 (b), C4H2F6-based gas exhibited the smallest CD differences, with sidewall angles close to 90 degrees while C4F8-based gas showed the largest CD difference (reduction in top and bottom CDs) compared to the reference, and its sidewall angle was also observed to be the least favorable. For C4F6-based gas, intermediate CD differences and sidewall angles were observed compared to C4H2F6-based and C4F8-based gases.
III-2.Plasma analysis for C4F8, C4F6, and C4H2F6-based gases
For the etching conditions in Figure 2, dissociated gas species were investigated using QMS and the mass spectra of the positive ions directly extracted from the plasma for C4H2F6, C4F6, and C4F8-based gases are shown in Figure 5 (a). (Dissociated radicals in the plasmas observed using QMS for these gases are shown in supplementary information Figure S4.) As shown in Fig. 5 (a), various ions dissociated and recombined from the reactive gases can be seen. Among those reactive ions, ions such as F+, CF+, CF2+, CF3+, CHF+, and CHF2+ can be more related to passivation or etching of SiO2 and SiNx. Mass amount (intensity) of these ions detected by the QMS are shown in Figure 5 (b). Among these ions, CF+, CF2+, CHF+, and CHF2+ are more related to the passivation on the materials surface while CF3+ and F+ are more related to etching. Therefore, the ratio of (CF+ + CHF+ + CF2+ + CHF2+)/(CF3++ F+) was taken to estimate the reactive ion flux ratios from plasma to the ONON stack sample surface between passivation flux and etchant flux, and the result is shown in Figure 5 (c). As shown in Figure 5 (c), the ratio was higher for C4H2F6-based gas compared to C4F6 and C4F8-based gases. This indicates that among the three different fluorocarbon gases, C4H2F6 provides the most polymeric radicals to the ONON stack sample surface during the etching, and which can provide the strongest sidewall protection condition. And, it is believed that the sidewall protection during the maskless ONON stack feature etching with C4H2F6-based gas is the related to maintaining a square shaped ONON stack feature until 1000 nm ONON thickness is etched.
Using OES, the radical species formed in the plasma were also observed with C4H2F6, C4F6, and C4F8-based gases mixed with CF4/O2/Ar. The process conditions are the same as those in Fig. 2. From OES, species such as CF2 at 251.9 nm, CH at 390 nm, F at 704 nm, Ar at 750 nm, O at 844.7 nm, etc. could be observed as shown in Figure 6 (a) [19–21]. The radical peak intensities such as CF2, CH, and F which are related to passivation and etching were normalized by Ar peak intensity to estimate the radical density in the plasma and the results are shown in Figure 6 (b). As shown in Fig. 6 (b), the F/Ar which is related to the etching was the highest for C4F8-based gas and the CF2/Ar + CH/Ar which is related to the passivation was the highest for C4H2F6-based gas. Figure 6 (c) shows the ratio of (CF2 + CH)/F for three gases which could show the degree of sidewall protection or sidewall etching during the maskless ONON feature etching and, among the three gas compositions, the C4H2F6-based gas showed the highest while it is lowest for C4F8-based gas. The results were similar to the QMS results in Fig. 5 (c) but, in the case of QMS with the positive ion measurement mode, even though it can measure all the positive ions in the plasma as-is for the estimation of the radicals in the plasma, it is difficult to estimate the F density due to the difficulty in positive ionization of F in the plasma. Therefore, for the estimation of F radical density, OES shows more reliable data. Therefore, based on the results of OES and QMS shown in Figure 5 and 6, it can be confirmed that C4H2F6-based gas provides an environment with the highest abundance of hydrofluorocarbon polymer, which protects the sidewalls of the profile the most. Additionally, it was noted that C4F8 provides an environment with the least amount of polymer. Under maskless conditions, it was observed that, to maintain the horizontal CD of the maskless ONON stack feature, the polymer layer on the sidewalls needs to be sufficiently thick, and a lesser amount of polymer layer can result in a reduction of the final CDs of the maskless ONON stack feature.
To understand the differences in trenching of the etched maskless ONON stack feature for the gases used in the experiment, the total positive ions, the sum of light positive ions (< 40 amu, that is, lighter than Ar mass), and the sum of heavy positive ions (≥ 40 amu), and in the plasma were calculated from the QMS data in Figure 5 (a) and the results are shown in Figure 6. As shown in Fig. 7, not only the sum of total positive ions but also the sum of heavy positive ions was the highest for C4F6-based gas and C4H2F6-based gas showed the lowest total positive ions for both total positive ions and heavy positive ions. The positive ions incident to the sidewall of the maskless ONON stack feature can be reflected at the sidewall and can lead to trenching due to increased ion flux at the edge of the feature. The most significant trenching observed for the maskless ONON stack feature etched with C4F6-based gas and the least significant trenching for C4H2F6-based gas are believed to be related to the differences in the positive ion flux, especially in the heavy positive ion flux.
III-3.ONON stack etch mechanism for C4F8, C4F6, and C4H2F6-based gases
XPS surface analysis was conducted to investigate the residue remaining on the sidewall of the etched maskless ONON stack features. To observe the residues at the sidewall of the etched maskless ONON stack features, XPS analysis was performed after tilting the sample 50° as shown in Figure 8 (a). (XPS widescan data measured for the sidewall residues remaining on the etched maskless ONON stack features for C4H2F6, C4F6, and C4F8-based gases are shown in supplementary information Figure S5.) The atomic percentages of the elements etched with C4H2F6, C4F6, and C4F8-based gases were observed by the XPS and the results are shown in Figure 8 (b) and the ratios of C/(Si+O+N) and F/(Si+O+N), which are the ratio of fluorocarbon residue component/substrate components, are shown in Figure 8 (c). As shown in Figure 8 (b) and (c), the carbon and fluorine forming fluorocarbon residue at the sidewall of the ONON stack feature were the highest for C4H2F6-based gas and the lowest for C4F8-based gas. To investigate the bonding states of the fluorocarbon residue at the sidewall of the ONON stack features, XPS narrow data of C1s were also measured and Figure 8 (d), (e), and (f) are C1s narrow scan XPS data for C4H2F6, C4F6, and C4F8-based gases, respectively. As shown in Figure 8 (d), (e), and (f), carbon bonding related to C–C (~285 eV), C-CF (~287.5 eV), C-F (~289.5 eV), and C-CF2 (~291.8 eV) were observed.[22–25] Among the bonding peaks, C-CF, C-F, and C-CF2 are related to the fluorocarbon layer [26,27] and C4H2F6-based gas showed also highest intensities of these bonding peaks and the C4F8-based gas showed the lowest intensities. Through XPS surface analysis, it can be understood that C4H2F6-based gas provided the most carbon-rich polymer at the sidewall of the etched ONON feature while C4F8-based gas formed the least polymer.
From the results of plasma analysis in Fig. 5~7 and surface analysis in Fig. 8 in addition to cross-sectional images of etched maskless ONON stack features in Fig. 3~4, the etch mechanism for the etching of maskless ONON stack features can be shown schematically as Figure 9. Etch profiles and CDs of the maskless ONON stack feature are influenced by the thickness of the CxHyFz hydrofluorocarbon or fluorocarbon polymer layer that provides protection of the sidewall, preventing the reduction of the subsequent pattern's mask line. In addition, heavy ion bombardment from the plasma to the substrate can form trenching at the edge of the maskless ONON stack feature by reflecting heavy ions at the sidewall of the maskless ONON stack feature during the etching. In the case of C4H2F6-based gas, despite ion bombardment, it is evident that the vertical sidewall is maintained and trenching phenomenon is protected by a thick polymer layer formed on the sidewall area as shown in Fig. 8(a). However, in the case of C4F6-based gas, due to a thinner polymer layer compared to C4H2F6-based gas, the CD and etch profile of the maskless ONON stack feature were slightly degraded and trenching was prominent due to heavy ion bombardment effect as shown in Fig. 8(b). Especially for C4F8-based gas, due to the thinnest polymer layer at the sidewall of the maskless ONON stack feature, sidewall etching was dominant during the etching even though the trenching was not significant due to the lower ion bombardment.
For the staircase etching for 3D NAND device, it is important to maintain pattern CD width and to keep vertical etch profile without trenching during etching maskless ONON stack features because, if the CD is decreased and slanted etch profile is formed, during the following additional ON layer-by-layer etching with PR trimming, the metal contact area for each ON layer can be significantly reduced. (for more details, see supplementary information Figure S1.) Therefore, under maskless etching conditions, maintaining the etched CD is crucial because the current feature pattern serves as a mask for the etching of the next layer, necessitating the formation of a sufficiently thick passivation layer to prevent etching of the sidewall. Furthermore, it can be observed that adequate polymer layer on the sidewall in addition to low ion bombardment is also required to suppress trenching phenomena in addition to preventing the reduction of the CD.