Abstract
The characteristics of the dry etching of :H thin films for display devices using SF6/O2 and NF3/O2 were investigated using a dual-frequency capacitively coupled plasma reactive ion etching (CCP-RIE) system. The investigation was carried out by varying the RF power ratio (13.56 MHz/2 MHz), pressure, and gas flow ratio. For the :H film, the etch rates obtained using NF3/O2 were higher than those obtained using SF6/O2 under various process conditions. The relationships between the etch rates and the usual monitoring parameters—the optical emission spectroscopy (OES) intensity of atomic fluorine (685.1 nm and 702.89 nm) and the voltages and —were investigated. The OES intensity data indicated a correlation between the bulk plasma density and the atomic fluorine density. The etch rate was proportional to the product of the OES intensity of atomic fluorine and the square root of the voltages on the assumption that the velocity of the reactive fluorine was proportional to the square root of the voltages.
1. Introduction
In the fabrication of flat panel displays, SF6 is used to etch various thin films. NF3, on the other hand, is used for two reasons in particular. First, it is used to etch Si. Focus is put on films such as SiO2 and Si3N4, which are used for semiconductor devices [1–7]. Second, NF3 is used to clean the chamber in the plasma enhanced chemical vapor deposition (PECVD) process [8]. Previous studies have looked at the etching characteristics and mechanisms of films with NF3. In these studies, a number of analysis tools, including a quadrupole mass spectrometer (QMS), ellipsometer, X-ray photoelectron spectrometer (XPS), and optical emission spectroscopy (OES), were used. With these studies in mind, this current paper postulates that NF3 could be used as an alternative to SF6 in the dry etching process for flat panel display manufacturing.
The replacement of SF6 with NF3 would offer certain advantages. Manufacturers would be able to change the etching gas on demand and would still be able to use existing facilities. Moreover, because of increasing concerns regarding the greenhouse effect, many research groups have looked for gases that could be used to replace SF6 to reduce the global warming potential (GWP). Low GWP gases such as F2 [9, 10], C4F8 [11], CHF3O2 [12], C3F8 [7], and C3F6O [13] in the dry etch process have been studied. These investigations indicated that it would be possible to use low GWP gases to replace SF6 in the etching processes. Of importance to this current study, it has also been shown that the use of NF3 would decrease the GWP of the flat panel manufacturing process [14–17].
To evaluate the possibility of replacing SF6 with NF3, the SiNx:H dry etching characteristics using SF6 and NF3 were compared. The similarities and dissimilarities between the processes using the two gases were studied. Because simple parameters such as the voltage of the input power and optical emission lines are commonly monitored in manufacturing equipment, the relationships between these parameters and the etch characteristics could be investigated in this work.
In this paper, Section 2 provides details regarding the experimental setup and process conditions. In Section 3, the results and a description of the experiment are presented. Section 4 covers the interpretation and analysis of the results.
2. Materials and Methods
The SiNx:H film samples were deposited on a glass substrate using PECVD and etched using a dual-frequency CCP-RIE system with SF6/O2 and NF3/O2 gas mixtures. The reference process conditions consisted of a gas flow ratio (gas/O2) between the fluorine gas (SF6 or NF3) and O2 of 30 sccm/30 sccm, a pressure of 200 mTorr, and a dual-frequency power ratio (13.56 MHz/2 MHz) of 25 W/75 W.
The gas flow ratio was varied by varying the O2 flow rate from 15 sccm to 45 sccm. The pressure was varied from 100 mTorr to 200 mTorr at an interval of 50 mTorr, and the power ratio was changed between values of 25 W/75 W and 75 W/25 W, where the total power was fixed at 100 W.
The etch rate was measured with a general procedure as follows.
First, the photoresist was deposited on the SiNx:H film using the photolithography process, after which the thickness of the deposited photoresist was measured using a surface profiler (-step). Second, the thickness of the etched film and the oxidized photoresist was measured using the surface profiler after dry etching using a CCP-RIE. Third, the thickness of the etched film was measured using the surface profiler after stripping the oxidized photoresist.
A schematic diagram of the dual-frequency CCP-RIE system for the dry etching process is shown in Figure 1. The optical emissions from the plasma were collected through a view port on the chamber side wall. A V-I probe (V-I probe 4100 by MKS, ENI Products) measured the loads V and I for the RF plasma discharge. The optical emissions were analyzed with a spectrometer. The spectrometer (HR2000+ by Ocean Optics Inc.) with optical resolution of 0.035 nm and range of 200–1100 nm was used.
Typically, various radicals such as NF2 and SF5 participate in the dry etching processes. However, the optical emission intensity of atomic fluorine was measured to observe the common element between the SF6 and NF3 processes. The emission peaks of atomic fluorine at 658.6 nm and 703.7 nm were used [18–20] and the sum of them referred to the optical emission intensities of atomic fluorine in this experiment.
3. Results and Discussion
3.1. Results
The dry etching conditions and results are given in Table 1. Figure 2 shows the etching characteristics and optical emission intensities of atomic fluorine obtained using the two gases for different gas flow ratios (gas/O2) at a fixed RF power ratio (13.56 MHz/2 MHz) of 25 W/75 W and 200 mTorr. A small increase in O2 flow increased the etch rate, while further increases in the O2 flow rate decreased the etch rate, as is usual [21]. In Figure 2(b), the behavior of for SF6/O2 follows the etch rate behavior.
(a)
(b)
Figure 3 shows the etching characteristics and for the two gases obtained using different pressures with a power ratio (13.56 MHz/2 MHz) of 25 W/75 W and a gas flow ratio of 30 sccm/30 sccm. The etch rate and tended to increase as the pressure increased.
(a)
(b)
Figure 4 shows the results obtained when the power ratio of high and low RF frequencies was varied with a fixed total power of 100 W at 200 mTorr and a gas/O2 flow ratio of 30 sccm/30 sccm. Although increased as the power of the 13.56 MHz source increased, the etch rate did not directly depend on .
(a)
(b)
In Figure 5, the voltages obtained under the same conditions are shown. As was the case in Figure 4(a), the etch rate did not directly depend on the voltages.
(a)
(b)
(c)
3.2. Discussion
When SiNx:H film is etched using fluorine-based gases such as SF6 and NF3, the film is eliminated in the form of SiFx. In this mechanism, the gas-phase reactions that create the reactive species (F) are as follows [22, 23]:
“Reactions of ” “Reactions of ”
Because the different reactions that create the reactive F cause differences in the etch rate and lead to different characteristics of the SF6 and NF3 processes, the optical emission intensity was observed as an alternative to a quantitative analysis of the reactive F.
In dual-frequency RF discharges, the following relationship between the density of the bulk plasma and the voltage of the RF sources exists [24]: where is the density of the bulk plasma and is the collisional energy loss per electron.
In the etching of Si containing films in fluorine-based plasma, the F-atom density is directly involved in the etching process [25]. Although the plasma density would not be directly proportional to the atomic fluorine density, there would be a strong relation between them. As shown in Figure 3, increased as the pressure increased. It seems that the increase in pressure led to a higher bulk plasma density, and this higher density caused to increase. To confirm this reasoning, the relationship was investigated in Figure 6. In this work, the value of was dominant, so that . When the linear fitting model (least square fitting method) was applied to Figure 6, different slopes were obtained for SF6/O2 and NF3/O2. It is assumed that the different values and different reactions for the two gases led to the different slopes. In conclusion, was proportional to , and the density of atomic fluorine () could be described as directly dependent on . Figure 7(a) shows the proportional relation between the etch rate and . If we assume that the atomic fluorine density directly depends on , this shows that a relationship exists between the density of the atomic fluorine and the SiNx:H etch rate. In fact, a number of recent studies have produced similar results to utilize a relation between the OES intensity and plasma density [26–28].
(a)
(b)
(c)
In Figures 4(a) and 5, the etch rate did not depend on the voltage (, or ) directly. In the bulk plasma, the velocity of the fluorine ion was proportional to . This increased the velocity of the reactant atom () via an ion-neutral momentum transfer collision or charge exchange. However, the etch rate was not dependent on alone. Generally speaking, an etch rate is influenced by the flux of the reactant transported to the substrate. If is related to and is proportional to , the etch rate will be proportional to . Under the same linear fitting model, the relation of the etch rate to is shown in Figure 7(b). It was found that the standard fitting errors were reduced when the voltage was considered. This shows that the etch rate had a closer relationship with than it did solely for . In the dual-frequency CCP, it is believed that the bias (lower frequency) power contributed more to the ion energy than the source (higher frequency) power. Therefore, it is believed that the bias power had a greater effect on the etch rate. In Figure 7(c), only the bias power voltage is considered to see the bias power effect. The fitting results were similar to those of Figure 7(b). If the assumption that the velocity of the atomic fluorine was affected by voltage is invoked, this result shows that the major contribution on the flux of the atomic fluorine comes from bias power.
4. Conclusion
In this paper, the etch characteristics of SF6/O2 and NF3/O2 were compared for different process conditions using a CCP-RIE system, and the similarities and dissimilarities between the processes using the two gases were analyzed. The etch rate of SiNx:H using NF3/O2 rather than SF6/O2 increased by 1.4–2.0 times in the same process conditions. These results indicate that the etch time can be reduced to obtain similar results by replacing gases. Because the GWP of SF6 is 3.1 times higher than that of NF3, the gas replacement can reduce the GWP by factor of 1/4.3–1/6.2, while obtaining the same process results.
An optical analysis using an OES showed that the density of the atomic fluorine was dependent on the intensity of the atomic fluorine. By inspecting the relationships between the etch rate of SiNx:H and , the effects of atomic fluorine density on the etch rate and its proportionality could be monitored indirectly. A larger (more abundant atomic fluorine) enabled a capacity to contribute to the etch rate. An electrical analysis using a V-I probe indicated that the velocity of the reactive F was proportional to . The flux of the atomic fluorine was interpreted as , and it was proportional to the etch rate. The importance of the bias power on the etch rate was examined indirectly, and it was concluded that the greatest contribution to the etch rate was from the bias power. This suggests the possibility of utilizing the flux of the atomic fluorine for the simple monitoring parameters (OES and voltages).
Conflict of Interests
The authors declare that there is no conflict of interests regarding the publication of this paper.
Acknowledgments
This work was supported in part by the International collaborative R&D program (N0000678) and by the Industrial Strategic Technology Development Program (10041681) funded by the Ministry of Trade, Industry and Energy of Korea.