3.1.

PMMA Etching in Argon Plasma or Oxygen Plasma

The etching rates and the PMMA/PS etching selectivity in argon or oxygen plasmas are listed in Table 1. In the case of both plasmas, etching rates of PMMA are higher than those of PS. In the case of argon plasma, selectivity is 3.9, which is higher than that in the case of oxygen plasma, which is 1.7. This result, namely that PMMA/PS etching selectivity in argon plasma is higher than that in oxygen plasmas, is consistent with recent experimental measurements reported by Liu et al.¹⁷ and Ting et al.¹³

Table 1

Etching selectivity of PMMA/PS in argon and oxygen plasmas (100-W source power, 50-V FSV, 10-W bias power, 2.0-Pa gas pressure, and 120-s etching time).

In argon plasmas, physical sputtering by argon ions is the main mechanism of etching. The high-etching selectivity in argon plasmas is probably due to the difference between the molecular structures of PS and PMMA. The schematics of the molecular structures of PS and PMMA are illustrated in Fig. 1. The molecular structure of PS includes benzene rings, which enhance the durability of polymers against etching. The enhancement in the durability is due to $\pi$ electrons in these benzene rings.²¹ In contrast, the molecular structure of PMMA includes oxygen atoms instead of benzene rings. The oxygen atoms enhance the etching rate, because carbon atoms in polymers are easily volatized when oxygen atoms are present. This difference in the molecular structures of PS and PMMA is the cause of their different etching rates.

Fig. 1

Molecular structures of polystyrene (PS) and poly(methyl methacrylate) (PMMA).

To investigate the argon plasma process in detail, time dependence of etching depth was evaluated (Fig. 2). The PMMA etching rate is reduced by more than one half after the etching depth exceeds 15 nm. If PS etching rate was constant regardless of etching depth, this result (i.e., over 50% reduction in etching rate) explains the low-etching selectivity and pattern degradation in formation of a PS mask with thickness of more than 20 nm.

Fig. 2

Time dependence of PMMA etching depth in argon plasma at 200-W source power, 50-V FSV, 5-W bias power, and 2.0-Pa gas pressure. PMMA etching rate decreases from 54.4 to $16.8\mathrm{nm}/\min$ when etching depth exceeds 15.4 nm.

3.2.

Effect of Oxygen Impurities on PMMA Etching

One possible cause for the decrease in PMMA etching rate in argon plasma is the change in the amount of oxygen from the ${\mathrm{Al}}_2{\mathrm{O}}_3$ dome or the change in the amount of residual oxygen impurities, such as water or other adsorbates, on the chamber wall or on the surface of the wafer. The amount of oxygen incorporation is expected to be especially high immediately after plasma ignition. The time dependence of oxygen-emission intensity was determined by using a bare silicon wafer under argon plasma at high-source power (1000 W) and high-FSV (400 V) etching conditions (Fig. 3). According to the figure, the intensity of oxygen emission decreases with increasing time after plasma ignition. The high-oxygen intensity immediately after plasma ignition is probably due to the oxygen impurities that originate from the ${\mathrm{Al}}_2{\mathrm{O}}_3$ dome or residual oxygen impurities. It should be noted that oxygen emission did not change under low-source power (100 or 200 W) etching conditions, as shown in Table 1 and Fig. 2. However, a small amount of oxygen impurities, which was below the detection limit, was supposed to exist in the plasma under the low-source power etching condition.

Fig. 3

Time dependence of oxygen-emission intensity in argon discharge at 1000-W source power, 400-V FSV, 0-W bias power, and 0.5-Pa gas pressure.

Time dependence of etching depth with IBE was evaluated next. By using IBE, it was possible to avoid the etching-rate change (namely avoid the PMMA-etching-rate enhancement by oxygen impurities), because PMMA is etched only by argon ions when the PMMA and plasma (i.e., ion source) are separated. Furthermore, the etching-rate change due to variations in the ion-source condition can be suppressed by using IBE, because the shutter used in IBE does not open until the ion current reaches a stable value.

Time dependences of PMMA etching depth for IBE are shown in Fig. 4. According to the figure, etching depth increases linearly with increasing etching time until it reaches 26.7 nm, and the etching rate is reduced to less than one half of its initial value after the etching depth exceeds 26.7 nm. This result indicates that the change in concentration of oxygen impurities is not the cause of the reduction in etching rate.

Fig. 4

Time dependence of PMMA etching depth with ion-beam etching (IBE) at 500-V acceleration voltage and $0.4\text{-}\mathrm{mA}/{\mathrm{cm}}^2$ ion current. PMMA etching rate decreases from 106.9 to $20.4\mathrm{nm}/\min$ when etching depth exceeds 26.7 nm.

3.3.

Effect of Surface Oxygen on PMMA Etching

To identify the cause of the reduction in PMMA etching rate, the surface composition of PMMA was examined by XPS with an argon-ion gun. Oxygen-to-carbon (O/C) ratios are plotted in Fig. 5 as a function of argon-ion irradiation time. Acceleration voltage of the ion gun was either 500 V or 3 kV, while ion current was kept at $12.5\mu \text{A}/{\mathrm{cm}}^2$. The O/C ratios were calculated from the narrow-scan spectra of ${\mathrm{O}}_{1s}$ and ${\mathrm{C}}_{1s}$. According to the figure, O/C ratio decreases with increasing ion-irradiation time. Moreover, the reduction of O/C ratio is more rapid under high-ion energy conditions (i.e., 3 kV). These results provide strong evidence that the reductions in etching rate for both argon plasmas and IBE are caused by the depletion of oxygen concentration by ion bombardment. This finding is consistent with a previous study, which showed that etching rate of organic materials decreases with decreasing O/C ratio.¹⁴

Fig. 5

Time dependence of O/C ratio in the PMMA film measured by x-ray photoelectron spectroscopy (XPS) with argon-ion gun. Acceleration voltage of the ion gun was 500 V or 3 kV with constant ion current of $12.5\mu \text{A}/{\mathrm{cm}}^2$.

To evaluate the PMMA/PS etching selectivity, the changes in O/C ratio due to argon-ion irradiation in the cases of PMMA and PS were compared (Fig. 6). Acceleration voltage of ion irradiation was 3 kV, ion current was $12.5\mu \text{A}/{\mathrm{cm}}^2$, and irradiation time was 60 s. The O/C ratio of PMMA decreases from 0.35 to 0.026 after ion irradiation. On the other hand, the O/C ratio of PS is almost unchanged (i.e., decreases from 0.059 to 0.057). A small change in PS etching rate is therefore expected, because the change of O/C ratio of PS is much smaller than that of PMMA. The O/C ratio of PS is considered to be zero, because there is no oxygen in PS according to the PS molecular structure in Fig. 1(a). However, the peak of ${\mathrm{O}}_{1s}$ in PS is observed by our XPS analysis, and the O/C ratio of PS does not become zero. The origin of oxygen is tentatively identified as oxygen absorbed during air exposure or oxygen existing as production impurities, such as lubricant or surfactant molecules, in the initial films. The PS film was etched in argon plasma for 300 s at 200-W source power, 5-W bias power, and 2.0-Pa gas pressure. PS etching depth was estimated to be 42.5 nm. On the assumption that PS etching rate is constant regardless of etching depth, high PMMA/PS etching selectivity, namely 6.9, is expected when the depth is below 15 nm. The lower etching rate of PS compared with PMMA is due to the difficulty in breaking the stable benzene ring to form volatile species, as described in Sec. 3.1.

Fig. 6

Comparison of O/C ratio change by argon-ion irradiation of PMMA and PS. Irradiation conditions: 3-kV acceleration voltage, $12.5\text{-}\mu \text{A}/{\mathrm{cm}}^2$ ion current, and 30-s irradiation time.

The change in O/C ratio of PMMA is probably due to the change in oxygen composition that comes from the following two origins: One is the oxygen originally contained in the film, as shown in Fig. 1(b) and the other is the oxygen absorbed to the film surface during air exposure. To investigate the oxygen-composition change due to air exposure, the difference in the surface compositions before and after air exposure was evaluated. Two wide-scan spectra taken by XPS are shown in Fig. 7: one was taken immediately after 30-s argon-ion irradiation under 3-kV acceleration voltage and the other was taken after 120-s air exposure after argon-ion irradiation. Clearly, the peak height of ${\mathrm{O}}_{1s}$ increases after air exposure. Furthermore, O/C ratio, which is calculated from the narrow-scan spectra, increases from 0.026 to 0.098 after air exposure. According to these results, it is expected that PMMA etching rate will be increased by air exposure, because the oxygen composition is increased by air exposure.

Fig. 7

Comparison of wide-scan spectra of PMMA measured by XPS before and after air exposure: (a) wide-scan spectrum after 30-s argon-ion irradiation under 3-kV acceleration voltage and $12.5\text{-}\mu \text{A}/{\mathrm{cm}}^2$ ion current and (b) wide-scan spectrum after 120-s air exposure.

To clarify the effect of air exposure on etching rate of PMMA, the time dependence of etching depth was evaluated. Measured etching depth in an ion-beam etched sample without any treatment before IBE is plotted in Fig. 8. Measured etching depth of an ion-beam etched sample that was etched by argon plasma before IBE is also plotted in the graph. The etching time in the case of the argon plasma was 32 s under 200-W source power, 50-V FSV, 5-W bias power, and 2.0-Pa gas pressure. These two samples were exposed to air for more than 1 day before IBE. Both etching depths show a similar dependence on IBE time. In particular, the slopes of etching depth up to 15 s are much steeper than those after 15 s. This finding suggests that surface-oxygen composition (i.e., O/C ratio) of the sample after 32-s plasma etching is depleted. However, etching depth rapidly increased during the early etching stage, even in the case of the sample with argon-plasma pretreatment. This result shows that the primary cause of etching-rate reduction is attributed not to the reduction in oxygen composition in the film, but to the reduction in oxygen absorbed by the film surface during air exposure. It should be noted that high PMMA/PS etching selectivity when the depth is below 15 nm may be unstable, because the amount of attached oxygen easily changes according to the storage condition.

Fig. 8

Comparison of time dependence for two PMMA samples: (a) untreated sample before IBE and (b) etched sample by argon plasma before IBE. Etching time of argon plasma was 32 s (which is one of the points plotted in Fig. 2). Sample (b) was exposed to air for about 1 day between plasma etching and IBE.

3.4.

Effect of Oxygen Addition to Argon Plasma

To avoid the change in PMMA/PS etching selectivity due to etching-rate change and variable storage conditions, a small amount of oxygen was added to the argon plasma. The time dependences of etching depth for a pure-argon plasma, argon plasma with 1% oxygen addition, and pure oxygen plasma are plotted in Fig. 9. Bias power was 5 W, and gas pressure was 2.0 Pa. In the case of pure oxygen plasma, plasma-source power was adjusted to ensure that the ion energy was the same as that in the case of the argon-based plasma. The samples etched in mixed argon-oxygen plasma and oxygen plasma only show linear dependences of etching depth on etching time, even when the etching depth exceeded 50 nm. This result indicates that the addition of a slight amount of oxygen to the argon plasma suppresses the variation in etching rate due to the depletion of oxygen concentration during plasma etching. The value of PMMA/PS etching selectivity in the case of the mixed argon-oxygen plasma was estimated to be 4.2 at 60-s etching time.

Fig. 9

Time dependence of PMMA etching depth with the addition of oxygen to argon plasma: (a) argon plasma with 200-W source power, (b) addition of 1% oxygen gas to argon plasma with 200-W source power, and (c) oxygen plasma with 600-W source power. Plasma-etching conditions were 50-V FSV, 5-W bias power, and 2.0-Pa gas pressure in all lines. Source power was controlled so that the time dependences could be compared under the same ion energy.

Etched fingerprint patterns of PS-b-PMMA, which have three kinds of patterns with full pitches, are shown in Fig. 10. The mixed argon-oxygen plasma was used to fabricate these patterns. It is clear from these images that fine PS mask patterns with a full pitch in the range of 25.5 to 77 nm were successfully formed with the mixed argon-oxygen plasma.

Fig. 10

Top- and cross-sectional view of three PS-mask structures: (a) 25.5-nm full pitch, (b) 41.0-nm full pitch, and (c) 77.0-nm full pitch. All the patterns were made from the PS-b-PMMA fingerprint pattern by using mixed argon/oxygen plasma.