Atomic Oxygen Sensing Using a Quartz Crystal Microbalance with a Polymer Thin Film Prepared by RF Sputtering

Polymer thin fi lms, sputtered using a polyimide target on a quartz crystal, were investigated to determine their suitability for application in active oxygen sensors. Active oxygen generated under an inductively coupled plasma (ICP), particularly atomic oxygen radicals with a fl ux of 6.93×1013 atoms/cm2/s, was successfully detected from the frequency shift using a polymer-coated quartz crystal microbalance (QCM). To investigate the reaction mechanism between the polymer surface and atomic oxygen, chemical bonding was evaluated by electron spectroscopy for chemical analysis (ESCA) and surface morphology by atomic force microscopy (AFM). We also compared sputtercoated polymers with spin-coated polymers in terms of their sensing characteristics. We found that a sputter-coated polymer QCM has a great potential for atomic oxygen sensing.


Introduction
Active oxygen, generated by ultraviolet lamps, ozonizers, and discharge plasma, plays an important role in several industrial processes, including surface cleaning, surface modifi cation, sterilization, and oxidation. With the current need to reduce greenhouse gas emissions (e.g., fl uorocarbons used in semiconductor manufacturing processes), the use of active oxygen is attracting more attention. In particular, atomic oxygen has an extremely strong oxidative ability and could therefore be applied to the above processes. These processes, however, demand precise reproducibility from the viewpoint of attaining a high production yield. Therefore, being able to accurately measure the amount of atomic oxygen is critical to the precise control of these processes.
Such measurement requires costly specialized equipment incorporating a laser optical system (1) or a vacuum ultraviolet light source, (2,3) such that a reasonably priced and simple method of real-time sensing is not currently available to the best of our knowledge.
A quartz crystal microbalance (QCM) is a sensor device that is capable of the nanogram-order measurement of changes in mass on the surface of a quartz crystal by observing shifts in resonant frequency. (4) Hence, this method can have different gassensing applications by selecting the appropriate electrode material to be formed on the quartz crystal as the detection layer. (5) In our previous study, we verifi ed the atomic oxygen sensing and accuracy of carbon-coated and silver-coated QCMs. (6) Although the silver-coated QCM is highly sensitive, frequency changes rapidly over several seconds and without any linearity as a result of the marked surface oxidation by atomic oxygen. Therefore, a silver-coated QCM cannot provide stable sensing in industrial processes. On the other hand, the carbon-coated QCM showed stable detection properties. Carbon thin fi lms, however, generally applied by a fl ash deposition method based on the resistive heating of a pair of carbon rods, suffer from a serious drawback; it is diffi cult to obtain thick fi lms and reproduce fi lm thickness owing to restrictions imposed by the deposition principle. Clearly, a reproducible mass is indispensable from the viewpoint of practical use. This is disadvantageous when carbon thin fi lms are applied to processes that require atomic oxygen sensing over a longer duration.
Polymer fi lms such as polyimide (PI), polyethylene terephthalate (PET), polyvinyl fluoride (PVF), and polytetrafluoroethylene (PTFE) exhibit relatively low reaction effi ciencies for atomic oxygen over an extended duration. (7) To fabricate a durable atomic oxygen sensor, we fi rst developed a polymer-coated QCM that was prepared by RF sputtering with a polyimide target, because this method can readily produce thin fi lms with excellent thickness controllability and reproducibility. In this paper, we investigate the atomic oxygen sensing properties of this QCM. In particular, the long-term (over 30 min) sensing property was verifi ed to determine whether the QCM could be used for practical applications. Moreover, we considered the effect of the atomic oxygen sensing mechanism on a polymer-coated QCM by evaluating surface chemical bonding states and morphologies.

Preparation of polymer-coated QCM
The QCM is a sensor device that supports the nanogram-order measurement of changes in mass on the surface of a quartz crystal by observing frequency shifts. In this study, we investigated the use of the QCM technique for sensing active oxygen species, particularly atomic oxygen. A polymer thin fi lm was formed as the atomic oxygen sensing layer on a commercially available AT-cut quartz crystal (with a resonant frequency of 6 MHz, a diameter of 14 mm, and gold electrodes on both sides) by RF sputtering with a polyimide target (Upilex-S, Ubekousan).
The coating was applied as follows. A quartz crystal plate was mounted in the sputtering equipment (SBR-1104E, ULVAC). After evacuating the equipment to less than 6.7×10 −3 Pa, argon (Ar) gas was introduced at 5.3×10 −1 Pa. An RF (13.56 MHz) power of 150 W was applied between the target and the stainless steel substrate holder, and then polymer coating onto the quartz plate was carried out. (The sputter-coated polymer fi lms are described in greater detail in ref. 8.) The thin fi lm was formed on one side, the area being ca. 8 mm in diameter and ca. 75 nm thick, which was determined by measuring the heights between the fi lm and the noncoated area with a contact-type roughness gauge. We can assume that the fi lm has no complete polyimide structure because of the decomposition caused by sputtering with argon ions.
For comparison, a polyimide amide acid (Semicofi ne SP-341, Toray Industries, Inc.) was prepared and coated on the above-mentioned quartz crystal using a spin coater (K-359 S-1, Kyowariken). The amide acid was diluted to 1/4 of the original concentration in N-methyl-2-pyrrolidine, and then 10 μl of the obtained solution was dropped onto the quartz crystal mounted in the spin coater. The spin-coating speed was set to 3000 rpm for 100 s, and the fi lm was cured to dehydrate and form the polyimide structure at 140°C for 30 min, 200°C for 30 min, and then 300°C for 1 h. The fi lm coating was formed over the entire quartz crystal, and the obtained fi lm thickness was ca. 150 nm. Figure 1 shows the setup of the atomic oxygen irradiation chamber used in this study. The equipment features an inductively coupled plasma (ICP) source in the upper part of the vacuum chamber. This acts as the atomic oxygen source. The vacuum chamber was evacuated to less than 7×10 −1 Pa using a rotary pumping system, and then oxygen gas (5N purity) was introduced at a constant fl ow rate of 0.05 SLM (standard liters/min) by mass fl ow (MODEL 3660, KOFLOC) at a pressure of 25 Pa. After a 30 s pause, it was confi rmed that the pressure was stable, and then an RF (13.56 MHz) power of 200 W was supplied to the antenna through the matching network such that inductively coupled oxygen plasma was generated within the ICP source. Early in this study, we assumed that there are many active oxygen species (e.g., oxygen ions, excited molecules, and ozone) in the plasma and vacuum chamber regions, although a subsequent spectroscopic study showed that atomic oxygen radicals (O * : 1 D→ 3 P state) are the predominant form of active oxygen in the irradiation area. These atomic oxygen radicals are assumed to fl ow downwards, drawn by the fl ow of evacuation. The QCM sensor was inserted into the sensor head and then connected to the oscillation circuit at a position 35 mm below the ICP source and on the central axis of the equipment. Frequency shifts during plasma generation were monitored every 30 s by a frequency counter (SC-7205, Iwatsu), and the results were recorded.

Results and Discussion
To verify the possibility of atomic oxygen sensing using a polymer-coated QCM, we fi rst investigated the frequency shift of sputter-coated and spin-coated polymer QCMs during irradiation with atomic oxygen. The results are shown in Fig. 2. The total sensing time was set to 10 min, including 30 s to confi rm the sensor stability with only the oxygen gas fl ow (from 0 to 0.5 min, in Fig. 2). For the sputter-coated QCM, the frequency increased linearly to 86 Hz for over 9 min. This frequency shift value is ca. four times larger than the result obtained with the carbon-coated QCM (22 Hz) under the same irradiation conditions that correspond to an atomic oxygen fl ux of 6.93×10 13 atoms/ cm 2 /s. (9) This fl ux number, which is estimated assuming that the carbon is removed by atomic oxygen and then volatile fl ux (CO) is formed, indicates the lower limit of detection by this QCM, due to variations in frequency shift behavior. As mentioned above, we also evaluated optical emission spectroscopy as a means of qualitatively analyzing atomic oxygen. However, the emission intensity decreases with the distance from the ICP source and is very weak near the QCM sensor. Under these conditions, surface treatments often take a long time in practical industrial applications, i.e., this indicates that an atomic oxygen fl ux on the order of 10 13 is relatively ineffi cient for surface treatments. This fact suggests that the QCM enables the detection of small quantities of atomic oxygen with a high resolution.
On the other hand, the spin-coated QCM exhibited a phased frequency shift for 6 min from the beginning of the atomic oxygen irradiation. Then, the frequency increased up to 120 Hz. The total frequency shift of the spin-coated QCM was larger than that of the sputter-coated QCM, and the difference was slightly less than 40 Hz, as shown in Fig. 2.
Iwamori et al. noted that the sputter-coated polymer has a broad FT-IR spectrum, which differs from the bulk polyimide. This suggests that new functional groups were generated in the polymer as a result of RF sputter coating with a polyimide target. (10) Therefore, the reason for this difference in frequency shift behavior between the spinand sputter-coated QCMs is assumed to be the difference in the moieties of the polymer thin fi lms. Namely, this result indicates that the difference in reaction effi ciency with atomic oxygen is due to the difference in the molecular structures of the thin fi lms.
As mentioned below, the sputter-coated QCM exhibits stable detection properties over a long duration despite a reduction in its weight. In other words, the difference in the initial fi lm thickness, spin-coated to 150 nm and sputter-coated to 75 nm, is thought to have no substantial relationship with this result. We also confi rmed that the initial thickness of the sputter-coated polymer has no infl uence on the detection properties.
For the sputter-coated QCM, it is thought that the reaction with atomic oxygen was slightly suppressed owing to the formation of functional groups, such that the frequency shift is reduced. These results indicate that the sputter-coated polymer QCM enables the highly sensitive quantifi cation of atomic oxygen from the linear slope of the frequency shift, and is superior to the conventional carbon-coated QCM.
Next, to clarify the mechanism of atomic oxygen sensing on the polymer-coated QCM, we evaluated the surface chemical bonding state of the sputter-coated polymer QCM using electron spectroscopy for chemical analysis (ESCA) (AXIS Ultra, KRATOS). ESCA measurement was performed in a high vacuum of less than 6×10 −6 Pa, with a monochromatic X-ray source (Al-Kα, 10 kV, 10 mA). First, a wide scan was performed in the energy range from 1350 to −5 eV in 1 eV steps.
For the ESCA wide-scan spectra, prominent peaks corresponding to N1s, O1s, and C1s were observed. The peak intensity ratio of O1s/C1s increases from 0.41 (before irradiation) to 0.86 (after 10 min irradiation), as shown in Table 1, which might be caused by carbon removal and surface oxidation under the infl uence of atomic oxygen.
To analyze the sensing reaction in greater detail, particularly the removal of organic compounds, which leads to polymer weight loss, a narrow scan in 0.1 eV steps with regard to the C1s spectra was evaluated. Gaussian curve fi tting and peak synthesis were performed for the obtained C1s spectra at each binding energy. The results are shown in Table 2 and Fig. 3. Then, the full width at half-maximum (FWHM) of the fi tting curve was fi xed to 1.4 eV using the method reported by Kinoshita et al. (11) The atomic compositions listed in Table 2 were calculated from the ratio of the area under each curve. A decrease in the BPDA and PDA contents and an increase in those of all the oxidative functional groups -C-O-C-, -C=O-, -C-O-N-, and -CO-OH-were clearly observed. The trends appearing in these results are in good agreement with the report published by Yokota et al., in which the durability of the spin-coated polyimide in the presence of atomic oxygen was clarifi ed. (12) The polymer surface morphologies, both before and after the atomic oxygen sensing, were evaluated using atomic force microscopy (AFM; SPA-300, Seiko Instruments). The scan mode was set to the dynamic force mode (DFM). The scan frequency was 2.0 Hz and the area was 1000 nm. Figure 4 shows an arbitrary cross-sectional profi le of the polymer surface. Note that a polymer thin-fi lm coating was formed on a polished silicon wafer having a nano-order fl at surface to eliminate the effect of roughness caused by the substrate surface. We confi rmed that the pristine polymer surface has a smooth surface morphology with a root-mean-square (RMS) roughness of ca. 0.17 nm. In contrast, we observed the formation of fi ne protuberances with a height of 1.4 nm and an RMS roughness of 0.36 nm on the irradiated surface.
From a series of the above results, the mechanisms of the reaction between atomic oxygen and polyimide are assumed to be as follows. Functional groups, e.g., phenol compounds, are produced as a result of the absorption of atomic oxygen onto the imide or benzene rings of the polyimide, after which the cleavage of the rings advances as a result of a cycloaddition and/or other reactions (scheme (1), in Fig. 5). After the cleavage of the benzene and imide rings, olefi n or aldehyde compounds containing unsaturated carbon are produced (scheme (2)). Carbonyl moieties of these compounds are fi nally desorbed as volatile fl uxes such as CO or CO 2 owing to further reaction with atomic oxygen (scheme (3)). (13) This phenomenon leads to the weight loss of the polymer and consequently to the frequency increase of the QCM, which enables atomic oxygen sensing.
The above assumptions relate to the "bulk-polyimide surface." It is thought, however, that a similar reaction based on these assumptions occurs not only on the bulk polyimide Table 1 O1s/C1s ratio of sputter-coated polymer before and after irradiation. but also on the sputter-coated polymer surface, because these effects are consistent with the ESCA results, i.e., an increase in the the oxidative functional group content and a decrease in the BPDA/PDA structure content after the atomic oxygen irradiation of the sputter-coated polymer surface. Besides, there is a possibility that new functional groups are formed on the polymer surface, which do not appear on the bulk polyimide, leading to a difference in the reaction effi ciency with atomic oxygen, as shown in Fig. 2.
The fi ne protuberance formation on the irradiated polymer surface, as verifi ed in the AFM image where no fl at-etching surface appears, indicates the formation of residual moieties as a result of atomic oxygen reactions in the above-mentioned schemes.
To investigate long-term atomic oxygen sensing, which is the goal of this study, the frequency shift characteristic of the sputter-coated QCM, as measured for more than 30 min, was determined. The experimental conditions were the same as those described above, with the exception of sensing time. Figure 6 shows the frequency shift result. From these results, we fi nd that frequency increases up to 255 Hz in 35 min. This suggests that the sputter-coated QCM has great potential for atomic oxygen sensing over Table 2 Synthesis results. extended periods, compared with the conventional QCM.
In this study, real-time monitoring of atomic oxygen and its quantifi cation were verifi ed. It is known that atomic oxygen is the predominant species not only in the plasma process but also in the low-pressure mercury ultraviolet lamp and/or xenon excimer lamp processes under atmospheric pressure. (14) Because atomic oxygen has a rate constant that is three orders of magnitude higher than that of ozone, the effect on the surface is noticeable. (15) Therefore, the QCM-based sensor has a great potential for the detection of atomic oxygen generated under UV lamps.