5. Conclusions

5.1 Conclusion of this study

Plasma etching is an important technology of the microfabrication that has driven the development of semiconductor device manufacturing. In recent years, the semiconductor device has been fabricated with the molecular and atomic dimensions, and the characteristics of the materials used are utilized in various ways. In this plasma etching process, in addition to the molecular and atomic level processing control of the structural dimensions, the control of various physical properties such as electrical and mechanical properties of the materials is becoming more and more important. With this background, the author has been trying to understand the surface reaction in the plasma etching process.

Fluorocarbon plasma etching is used to process silicon oxide films used as insulating films in semiconductor devices. In this etching method, the silicon oxide film can be etched chemically selectively. In addition, it is possible to process structures such as holes and grooves in a directional manner, where only the bottom of the structure reacts with the ions, while the side walls are not etched. However, although selectivity and vertical direction-ability are realized, extremely complicated reactions are going on on the surface, such as etching of silicon oxide film and deposition of fluorocarbon (a-C:F) film, which also has etching suppression effect. It is thought that the a-C:F film deposited on the surface suppresses etching in the area where etching is not desired, and is successfully removed with etching only on the silicon oxide film of the material to be processed. In this study, the author has focused on the a-C:F film, which is thought to play an important role in this complex surface reaction, and have tried to analyze it.

The author has been investigating the surface reaction in the etching process of silicon oxide film by fluorocarbon plasma in order to control the reaction efficiently to obtain the desired etching characteristics (shape, processing speed, etc.). In order to analyze the surface reactions, the author has observed the surface during the etching process using various spectroscopic techniques such as infrared spectroscopy and electron spin resonance, and have tried to understand the surface reaction through physicochemical analysis. The following is a summary of the results obtained in this study.

(1) Surface interactions of gaseous CFx radicals (Section 3.)

Fluorocarbon radicals (CFx) have been considered to be one of the factors that affect the deposition rate of a-C:F films during etching. Based on the decay process of CFx in the plasma afterglow and the density gradient of the radicals near the surface, the interaction between CFx and the surface, such as the adsorption of CFx on the surface and the generation of CFx from the a-C:F film, has been discussed extensively. Although the surface interaction through a-C:F film has been discussed, the correlation between CFx radicals in the gas phase and a-C:F film on the surface has not been investigated. Therefore, the author decided to investigate the radicals in the gas phase by laser-induced fluorescence (LIF) and to observe the surface "in situ" by infrared spectroscopy. The density distribution near the surface in the perpendicular (z) direction is important in discussing the interaction between radicals and the surface, so the author applied the two-dimensional LIF method, in which the laser excitation is made in the form of a sheet in the z direction, and succeeded in observing the CFx distribution in the z direction and the a-C:F film deposited on the surface simultaneously.

When the plasma density ne was low (ne < 1011 cm-3), the results were consistent with the surface reaction models reported so far. On the other hand, when the plasma density is high (ne > 1011 cm-3), the CFx density is high near the surface and low in the bulk of the plasma, showing a characteristic (concave) distribution. However, it is newly revealed in this experiment that the a-C:F film on the surface is not involved in determining such a concave z-direction profile. This result is well explained by the model of destruction due to excess dissociation of CFx in the bulk of the plasma. In this high-density region, the deposition rate of the a-C:F film is no longer correlated with the CFx density. This could be due to the involvement of other chemical species in the gas phase, such as ions, or the modification of the surface to facilitate the deposition of the a-C:F film. This change in the plasma chemistry and the dominant surface reaction in the gas phase could be the reason why the findings obtained in low-density plasmas have not been applied to the high-density plasmas used in actual processing.

(2) Surface interactions of CFx ions (Section 3.)

Since the vertical process ability of plasma etching is caused by the reaction of ions, the interaction between ions and surfaces has been investigated by irradiating the surface with a beam of controlled ions. However, the author has noticed that there are areas that have not been investigated in the beam experiments performed and the data obtained. This area was of paramount importance in understanding the surface response to etching.

Although the knowledge obtained so far is important, it is inadequate in the following points. (1) The chemical composition of the fluorocarbon ion beam, such as CFx+ (x=1,2,3), is not separated, (2) The presence of radicals during ion irradiation is ignored, (3) The target surface is somewhat unrealistic, such as a quartz crystal, (4) The irradiation dose is small, probably due to the low flux of the beam, and (5) The ion beam is not used in the etching process. (5) Dose dependency was not investigated.

The chemical composition of CFx+ was changed by mass separation using an ion beam irradiation appratus, and the surface of SiO2 formed on a silicon substrate was observed "in situ" using X-ray photoelectron spectroscopy (XPS). When irradiated with CFx+ ions, which have a high degree of F oxidation, at high energies (Eion >500 eV), the results are consistent with those reported previously. However, it was newly found in this experiment that a-C:F films were deposited on SiO2 surfaces irradiated with relatively high (Eion ≤ 500 eV) energies of CF+ ions, which have a low degree of F conversion. As a result of detailed investigation of the deposition process, etching by ion sputtering occurs in the initial stage of irradiation (< 5 × 1016 cm-2) on clean SiO2, and the amount of C on the surface increases during this period. At higher doses (∼ 1017 cm-2), the etching of SiO2 stops and the surface C content appears to have reached a critical value, leading to the continuous deposition of a-C:F films. We have shown that such phenomena, from etching to polymer deposition via surface denaturation, can be achieved by well-controlled beam irradiation. In the etching process using high-density plasmas, the chemical reactions such as etching and polymer deposition can be reversed depending on the chemical composition of the ion species incident on the surface and the surface denaturation induced by the irradiation.

(3) Real-time observation of etching surface in situ during processing (Section 4)

In the etching process of silicon oxide film by fluorocarbon plasma, it has been thought that the etching proceeds while a-C:F film is deposited on the surface during etching. However, there are no examples of "in-situ" observation of the surface during the etching process. Therefore, the author decided to investigate how the surface during etching actually looks like. The optical properties of the a-C:F film were analyzed, and an optical system that can observe the surface during the etching process was successfully fabricated. The optical properties of the a-C:F film were analyzed, and an optical system was successfully fabricated to allow observation of the a-C:F film on the surface during etching.

The deposition process of the a-C:F film during etching was observed in real time with a time resolution of less than 2 s, and was analyzed through spectral peak analysis. The results show that the a-C:F film on the surface transitions to a steady-state thickness and that the steady-state thickness can be explained by the balance between the deposition rate and the removal rate. The results provide useful data to consider the transition process of a-C:F film deposition during etching, which occurs in any etching system.

(4) In vacuo electron spin resonance of the etched samples (Section 4)

In addition, the author has investigated the unpaired electrons and dangling bonds generated by chemical bond breaking in order to investigate the details of surface reactions. Since the surface reaction depends on the surface condition and the reaction aspect changes drastically, it is important to analyze the reason for the change in the surface condition. The author has focused on the role of dangling bonds formed on the inner side (subsurface) near the surface during etching. To observe the dangling bonds on the surface, it was necessary to develop an observation device because of the chemical activity of the dangling bonds. Therefore, the author has developed a method to observe dangling bonds "in-situ" using the electron spin resonance (ESR) method on a sample with an etched surface. The developed system can observe the dangling bonds generated during etching by transporting them in a high vacuum (in-vacuo), although not "in-situ" during etching, and has succeeded in observing the dangling bonds that disappear in the air or by oxygen. The dangling bonds, which are lost in air and oxygen, were successfully observed by transporting them in a high vacuum (in-vacuo).

The ESR measurement of the dangling bonds in the plasma-deposited a-C:F film revealed that they have an extremely high spin density (∼ 2 × 1021 cm-3), which disappears upon exposure to oxygen. The high density of dangling bonds is almost independent of the amount of F in the film. a-C:F films have a high density of oxygen-reactive dangling bonds in the film, and the F is present on the surface as if it is stored. This is thought to be the reason why the etching of SiO2 works well for the formation of products desorbed by SiFx and the removal of a-C:F film in response to etching. It has also been shown that plasma light irradiation plays a role in the formation of dangling bonds, which cannot be ignored in high-density plasmas due to the higher photon energy and brightness.

As described above, the author has analyzed the surface reactions of fluorocarbon plasma etching of silicon oxide films as insulating films, and have made a small contribution to the understanding of the surface reactions.

5.2 Future challenges and prospects

Based on the results of this study, the following issues can be identified.

(1) Improvement of temporal resolution for in situ real time observation of etching surface

The surface during etching has been observed "in-situ" using infrared spectroscopy, but the time resolution is limited to less than 2 s, and further improvement of the resolution is urgently needed. However, the time resolution is limited to less than 2 s, and there is an urgent need to improve the resolution. In this experiment, it was limited to observe the transition process of a-C:F film deposition on the surface, which proceeds at an etching rate of several 10 nm/min. If a temporal resolution of 100 to 10 ms could be obtained, it would be possible to analyze a realistic etching process with an etching rate of several hundred nm/min. For this purpose, it would be effective to use a laser beam or a dispersive setup to limit the observation wavenumber range, or to use an asynchronous Fourier-transform method to obtain a fast time resolution.

(2) Implement in profile simulators

Next, the interaction of individual chemical species such as fluorocarbon radicals and ions with the surface has been investigated, but in reality, they are irradiated to the surface as a group and the reaction proceeds. It is a challenge to model and verify such realistic etching reactions. Several research institutes have proposed etching reaction models, but it has not been clearly shown whether the results of this experiment can be implemented in the elementary processes of the models. For this purpose, it is necessary to construct an actual reaction model.

(3) Understanding of elementary processes of surface reactions

Is it sufficient to divide the surface reaction model into elementary processes that determine the incident species and the desorbed species from the surface state? The dose-dependence observed in the ion beam experiments may indicate the unnaturalness of modeling with such a superposition of elementary processes. Nevertheless, at least surface reactions that lead to desorbed species via elementary processes that determine the surface state should be more dominant than now. If such modeling is not realized, it will only explain a limited number of etched surface reactions and will not be a universally applicable etching reaction model. The introduction of physical quantities that can determine the surface state, such as surface dangling bonds, will be helpful in developing an overall surface reaction model. It would be important to construct an overall surface reaction model by introducing physical quantities that can determine the surface state, such as surface dangling bonds.

(4) In situ observation using in vacuo electron spin resonance technique

As for the "in-situ" observation of the surface dangling bond, the author could not observe the etching process. This is an issue that needs to be addressed. This is because ESR observation originally uses electromagnetic fields, and the formation of plasma and the movement of charged particles are also composed of electromagnetic fields. At the very least, the author has not been able to develop a means to observe these without coupling them. This may be possible if the magnitude and frequency of the electromagnetic fields involved in both can be changed significantly. For example, the magnetic field of the plasma for etching and the magnetic field for ESR observation can be uncoupled, and the observation of spin resonance absorption can be changed to laser absorption. However, this direction may be difficult to achieve. Since the author is dealing with charged particles such as ions, the influence of magnetic fields is unavoidable, and it is nearly impossible to observe a small amount of spin resonance in the presence of a plasma. If the etching reaction caused by the charged state of the charged particles themselves is negligible, then "in-situ" observation using charge-neutralized beam irradiation becomes possible. Such a development is highly desirable.

Finally, the author hopes that further understanding of the phenomenon will lead to the realization of efficient reaction control to obtain the desired etching characteristics (shape, processing speed, etc.).

Last-modified: 2021-02-06 (土) 18:52:22