Effect of water vapor on plasma processing at atmospheric pressure: Polymer etching and surface modification by an Ar/H₂O plasma jet

Atmospheric pressure plasma (APP) reactors can be operated as low temperature plasma (LTP) sources in ambient air environments which usually contain a certain amount of water vapor. Due to the nonequilibrium nature and the highly energetic state of LTP, water vapor in APP is often dissociated, which initiates a wide variety of reaction pathways that lead to the generation of highly reactive species such as OH, H, HO₂, H₂O₂, HNO₂, and HNO₃.^1,2 The effect of adding water vapor to the feed gas and/or the surroundings of the APP reactors has been extensively studied through both experiments^3–6 and computer assisted modeling.^7–11 The characterization of gas phase species generated by water-containing APP, e.g., their space- and time-resolved density profiles, has been reported in several articles.^{3–6,10,12–18}

Despite the increasing attention in characterizing water-containing APP, less is known about the effect of water-containing APP on material surfaces—especially an understanding on the molecular level. Previous publications on the treatment of materials using water-containing APP mainly focused on polymer processing by corona discharge^19–21 and antimicrobial effects and mechanisms.^6,22–24 Due to the complexity of microbes and living cells, extracting information on plasma etching, surface modification, and material deposition from these studies is difficult, let alone identifying the reactive species responsible for these surface phenomena. The effect of H₂O in low pressure plasma (LPP) processing of polymers has been reported previously²⁵ where the etching and surface oxidation of polycarbonate were observed. Hydroxyl radicals (OH) were recognized by the authors as the effective etchant species in H₂O containing LPP.²⁵ However, due to the possible large difference in chemical kinetics between LPP and APP,²⁶ the role of H₂O in APP needs yet to be determined.

Compared to LPP, the difficulty of studying short-lived reactive species in water-containing APP comes from the short mean-free path of particles (<100 nm) at atmospheric pressure.²⁷ After generation in the discharge region, reactive species are transported through convection and diffusion from the nozzle to the target surfaces that are typically located at distances ranging from one to hundreds of millimeters. The interaction of reactive species with each other and with the gaseous environment wherein plasma-surface interaction (PSI) takes place may also cause the destruction and conversion of these species.²⁸ Hence, controlling the gaseous composition of the PSI vicinity is essential for being able to interpret plasma induced surface phenomena.

The correlation between gas phase reactive species and the responses of the material surface can help identify the dominant reactants from water APP. Besides the composition of the controlled gaseous environment, a model plasma-material system that consists of a well-characterized APP reactor and target materials with defined properties is desired for establishing such correlations.²⁹ Previously, we studied the etching and surface modification of polymers by Ar/O₂ plasma using a model system consisting of a radio-frequency (RF) plasma jet and vinyl polymers.^30–33 We found that O atoms were the dominant etchant generated by atmospheric Ar/O₂ plasma, and a linear response between the estimated incident O flux from the gas phase and the measured etched C flux on the material surface was observed. We estimated the etching reaction coefficient³⁴ of O atoms from these measurements as on the order of 10⁻⁴.³⁵ 

In this work, we use the same model system to study the effect of Ar/H₂O plasma on the polymer surface. By varying a number of plasma processing parameters such as treatment distance, environmental gas composition, feed gas composition, and substrate temperature, we explored the change of the polymer surface properties using in situ ellipsometry and x-ray photoelectron spectroscopy (XPS). Correspondingly, the OH density in the gas phase was measured by laser induced fluorescence (LIF). The role of OH radicals, especially on polymer etching, was then evaluated. We also indirectly assessed the effect of H atoms on polymers by comparing etching with Ar/H₂ and Ar/H₂O plasmas.

Polystyrene (PS) beads with an average molecular weight of 35 000 were purchased from Sigma-Aldrich (St. Louis, MO, USA) and used without further purification. Thin PS films were spin-coated on an Si substrate using 5 wt. % of PS in propylene glycol methyl ether acetate (Sigma-Aldrich, ReagentPlus®, ≥99.5%) solution. With 2000 rpm spin-speed and 3 s ramping time, the resulted PS films have a thickness of ∼180 nm. Our atomic force microscopy (AFM) results show that these PS films have an RMS surface roughness of 0.35 nm.

The model APP reactor selected for studying water-containing plasma at atmospheric pressure is the time-modulated RF plasma jet whose detailed characteristics can be found elsewhere.^30,36,37 Briefly, the RF jet is a capacitively coupled plasma reactor with a 1 mm diameter tungsten needle as the driven electrode, a quartz tube (3 mm outer and 1.5 mm inner diameter, ID) as the dielectric, and a copper ring as the grounded electrode. The 20 kHz modulated 14.0 MHz RF signal with 20% duty cycle³⁸ was generated through a function generator (Tektronix AFG3021B) and then amplified by a power amplifier (ENI A500). The feed gas of the RF jet studied in this work includes 1.5 standard liters per minute (slm) of Ar plus 1% O₂, 1% H₂O, 1% (O₂ + H₂O) mixture, and 1% H₂. All gases were purchased from Airgas (Hyattsville, Maryland) with purity grade as follows: Ar and N₂: High Purity, 99.998%; O₂: Ultra High Purity, 99.994%; H₂: Ultra High Purity, 99.999%. Since all feed gas compositions were composed of Ar plus 1% molecular gas admixture, we use the notation of “Ar/molecular gas” (e.g., Ar/H₂O) to represent “Ar + 1% molecular gas” in the rest of this work. We kept the visible plume length of the RF jet at a constant length of 3 mm for all feed gas conditions. The average plasma power dissipation of the RF jet was measured as described previously³⁰ and is approximately 2 W for Ar/O₂, Ar/H₂O, Ar/(O₂+ H₂O), and 1.26 W for Ar/H₂ unless noted otherwise. At the same plasma power, the following feed gas compositions yield comparable plume length: Ar/O₂, Ar/H₂O, Ar/(O₂+ H₂O), which is shorter than the plumes from Ar/H₂ and Ar plasma.

Due to the small cross section area of the RF jet nozzle (1.77 mm²), a homemade two-dimensional (2D) scan-processing stage was used to achieve uniformly treated target surfaces for further characterizations.^29,35 During treatment, the RF jet scans over the sample surface with a line-by-line processing pattern at a speed of either 1.2 or 2.4 mm/s. After treating the material surface back-and-forth along the same straight line, the RF jet moves to another processing line located 0.8 mm away and repeats the same line treatment. As illustrated in Fig. 1(a), the treatment angle φ (either 30° or 90°) and treatment distance d (4, 8, 12, 16, and 20 mm) relative to the material surface can be adjusted. The treatment distance d is defined as the length from the nozzle of the plasma jet to the polymer surface along the axis of the RF jet. For the tilted treatment configuration (φ = 30°), based on the law of sines, we set up the vertical distance from the nozzle of the RF jet to the polymer surface as d/2 instead, which avoids the difficulty in directly measuring the distance d.

The use of tilted treatment configuration is due to the spatial limitation of one of our environment control chambers where the treated polymer samples were vacuum transferred to XPS for surface characterization. Therefore, all experiments related to XPS were performed in the titled configuration. Otherwise, vertical treatment configuration was preferred, especially for the correlation between etching and gas phase reactive species. The scan speed, either 1.2 or 2.4 mm/s, was chosen carefully to avoid excessive etching (e.g., near total removal of the entire polymer film) in a series of experiments performed in the same figure. To avoid confusion, here we list all treatment configurations used in this work—Fig. 2: 30°, 2.4 mm/s; Figs. 3 and 9: 90°, 1.2 mm/s; Figs. 4–6: 30°, 1.2 mm/s; Figs. 7 and 8: 90°, 2.4 mm/s. We will further describe the treatment angle and scan-processing speed in the caption of each figure.

The treatment of polymers by the RF jet was conducted in 50 l sealed chambers. Controlled environments wherein PSI takes place were achieved by refilling these chambers (pre-evacuated to below 10 mTorr) with N₂/O₂ gas mixture at 4 slm constant flow rate.

The water vapor admixture in the feed gas of the RF jet was generated by a homemade gas delivery system consisting of a MKS M330AH mass flow controller (MFC, heated to 135 °C), a stainless steel water container (110 °C), and gas delivery lines (110 °C), as shown in Fig. 1(b). Compared to the heated gas lines, the temperature of the MFC is elevated by 25 °C for creating a pressure gradient which ensures the effective delivery of water vapor into the dry gas lines. The temperature of the water container was maintained at 110 °C due to safety concerns. The temperatures of the water vapor generation system were monitored by thermocouples and controlled by proportional–integral–derivative (PID) controllers. For generating 1% volume ratio of water vapor in Ar gas, 15 sccm water vapor (at atmospheric pressure) was delivered through the MFC to the heated Ar carrier gas. There is no condensation during or after mixing with dry gas, because (1) the 1% H₂O mixing ratio is well below the saturation concentration and (2) the gas lines near the mixing point were heated to 110 °C. The stability and humidity of the Ar/H₂O mixture in the feed gas were further confirmed by a hygrometer (TPI 597C1).

To evaluate the effect of substrate temperature on polymer etching by the water-containing APP, PID-controlled heating (above 20 °C) and thermoelectric cooling (below 20 °C) stages were integrated into the plasma processing setup, as shown in Fig. 1(a). Prior to the RF jet treatment, the polymer samples were heated or cooled to the desired temperature (0–90 °C). To avoid excessive polymer etching, the scan-processing speed of 2.4 mm/s was used for all temperature dependent experiments. We evaluated the potential heating of the target surfaces by the RF jet through measuring the thermal expansion of 745 nm SiO₂ during treatment at d = 4 mm. Our ellipsometry results show that the substrate temperature variation caused by plasma processing is less than 20 °C. We also measured the thickness change of PS films due to heating, and no observable thickness loss was measured at 40 °C.

The detailed characterization methods of the RF jet-treated polymer films have been described in our previous publications.^29,35 Briefly, we used in situ ellipsometry and XPS to characterize the thickness/refractive index and the surface chemical composition of the plasma-treated polymer films, respectively. To extract the real-time polymer thickness and refractive index information, an optical model was applied to the in situ ellipsometric data. The XPS system (Vacuum Generators ESCALAB MK II) is equipped with Al K_α x-ray source (1486.3 eV), and high-resolution C 1s, N 1s, and O 1s spectra were obtained at an electron take-off angle of 20° (probing depth ≈ 2 nm). Using CasaXPS software,³⁹ we fitted the C 1s spectra with peaks corresponding to C—C/H (285 eV), C—O (286.5 eV), O—C—O/C=O (288 eV), O—C=O (289 eV), O—CO—O (290.2 eV), and π–π* shake-up (291.6 eV). Due to the overlapping of peaks in the O 1s spectra, two peaks with binding energy at 532.7 eV and 533.9 eV were fitted. For the N 1s spectra, nitroso (NO, 401.8 eV) and nitrate (ONO₂, 407.5 eV) peaks were fitted. All spectra were calibrated with the C—C/H peak at 285 eV. After Shirley background subtraction, the elemental composition was calculated with sensitivity factors of C 1s = 1, N 1s = 1.77, and O 1s = 2.85, respectively.⁴⁰ The surface morphology of the pristine and RF jet-treated PS films was measured using atomic force microscopy (Bruker MultiMode AFM) in the tapping mode.

The absolute density of OH radicals generated by Ar/H₂O plasma was measured by LIF in open-air environment as shown in Fig. 1(c). An alumina plate (2.5 × 2.5 cm²) was placed underneath the plasma jet to mimic the polymer treatment experiments but without the significant etching of the polymer films. To reduce the influence of environmental O₂ and H₂O, we applied 10 slm pure N₂ shielding gas in the coaxial direction as the Ar/H₂O feed gas. We did not use a sealed chamber (as the one used for polymer etching) for LIF measurement due to the long duration of laser measurements which requires a steady-state gas composition. To have accurate quenching measurements in addition to the actual LIF measurement—which requires many LIF measurements at different time delays compared to the laser pulse—the gas composition needs to be stable over these subsequent measurements, which typically take more than 30 min. Our computational fluid dynamics model showed that such steady-state gas composition cannot be obtained on a time scale of 30 min in the sealed chamber and thus a N₂ shielding gas flow was applied to avoid O₂ diffusion into the jet effluent as in the chamber. For LIF measurement, the fluorescence of OH is induced with a frequency doubled dye laser (Sirah, Precision Scan), with Rhodamine 6G as the dye, pumped by a frequency doubled Nd:YAG laser (Spectra-Physics LAB-170-10H) at 532 nm. The P₁(2) transition of the OH[(A; ν′ = 1) ← (X; ν″ = 0)] system at 282.6 nm is used and in particular the f₁(2) transition. The laser pulse had a repetition frequency of 10 Hz and a pulse width of about 6 ns full width at half maximum (FWHM). A combination of a spherical (f = 25 cm) and a cylindrical lens (f = 50 cm) was used to shape the laser beam at the position of the plasma into a sheath with a thickness of about 206 μm (FWHM) and a height of approximately 4, 8, or 12 mm. The laser energy was ≈6 μJ for 4 mm distance and 9 μJ for 8/12 mm distances, resulting in the measurement being performed in the linear LIF region.

Time and spatially resolved images of the fluorescence were taken with an iCCD camera (Andor IStar DH340T) equipped with a Nikkor 105 mm f/4.5 UV lens. The OH density does not vary more than 10% near the substrate during one modulation cycle of the RF power. Hence, to avoid contributions of plasma emission to the collected fluorescence, the LIF measurements were performed in the plasma off period of the modulation cycle. If the modulation time is faster than the gas residence time, the variation in the species density is negligible as is the case here. Each measurement is an accumulation of 1200 pulses. The camera gate width was kept at 50 ns to ensure that all fluorescence was collected. The gas temperature was obtained by Rayleigh scattering.⁴¹ 

The absolute calibration of the OH LIF was performed using Rayleigh scattering of air. We used a four-level LIF model to obtain the absolute OH density from the calibration measurements as described in detail by Verreycken et al.⁴² A key unknown in the present experiment is the gas composition at where the OH LIF is measured. The total fluorescence lifetime of the OH(A) state was measured with a camera gate of 5 ns. However, this did not fully allow us to determine the gas composition as both H₂O and N₂ concentrations may vary in the jet effluent. We calculated the OH density from the fluorescence for two extreme cases: assuming that the quenching is due to N₂ and due to H₂O only. This approach leads to a range of possible OH densities. We report the average OH density obtained by this procedure from LIF measurements in the core of the jet 0.5 mm above the substrate and use the extreme values as an estimate of the uncertainty on the obtained OH density. The collisional quenching constants used for this calculation were taken from a previous publication.⁴³ 

When treating polymer films with Ar/H₂O plasma jet, we observed fast material removal similar to that found with Ar/O₂ plasma generated by the same reactor.²⁹ As shown in Fig. 2(a), the real-time thickness change of PS films during scan-processing was obtained through modeling of the in situ ellipsometry data.³⁹ The plasma jet was held at a distance d = 4 mm away from the material surface using titled configuration (φ = 30°). It can be seen that the polymer thickness decreases with scan-processing time and the curves in Fig. 2(a) show “staircase” profiles due to the line-by-line scan-processing pattern. Since the ellipsometry probing laser spot (diameter 3–4 mm) is located at the center of the 9.6 × 9.6 mm² etched surface area, at the beginning or the near-end of the scan-processing, the RF jet is located further away from the ellipsometry probing spot which corresponds to the observed slower etch rate (ER). The highest instant ER [absolute value of the slope in Fig. 2(a)] can be observed when the plasma plume hovers directly over the ellipsometry probing spot. For the Ar/H₂O plasma etching of PS in N₂ environment [black line in Fig. 2(a)], the highest ER is 64 nm/min. Because the instant ER changes with the position of the RF jet during scan-processing, it is not accurate to compare the instant ER for different plasma processing conditions. Since we only compare PS treatments with the same scan-processing parameters, the total etching depth (Δ_t), which reflects the average ER, is a more suitable parameter for evaluating the etching efficiency of the plasma treatment and will be discussed in the rest of this work.

We observed that the composition of the environment gas, wherein Ar/H₂O plasma treatment of polymers takes place, has a significant influence on the polymer etching efficiency. As shown in Fig. 2(a), the Δ_t of PS in a N₂ environment is 41.9 nm, whereas the corresponding Δ_t in a 95% N₂ + 5% O₂ environment is only 17.1 nm. We further tested other gas compositions with higher O₂ content as well as another polymer, i.e., poly(methyl methacrylate) (PMMA) to examine how universal the observed behavior is. As illustrated in Fig. 2(b), the total etching depth of both PS and PMMA treated by Ar/H₂O plasma quickly drops with small amount of O₂ added in the environment. Due to the large difference between artificial air (20% O₂) and N₂ environment (0% O₂) in Fig. 2(b), in the rest of this work, we evaluate the effect of feed gas composition on polymer processing in both of these two environments.

We find that the polymer etching efficiency of Ar/H₂O plasma decreases exponentially with treatment distance d, which is similar to that of Ar/O₂ plasma observed in our previous work.²⁹ As shown in Fig. 3, the exponential decay of the etching depth with the treatment distance was seen for both N₂ and artificial air environments, though etching in N₂ is more effective than in air environment at all treatment distances. By defining the exponential decay constant λ as the distance at which the etching depth is reduced to 1/e = 0.368 times of its initial value, we can fit the etching depth (Δ_t) data in Fig. 3 using the following formula with the least-square method:

Here, A is a fitting constant. For the PS film treated by Ar/H₂O plasma with vertical configuration (φ = 90°), the fitted exponential decay constant for the N₂ environment is $λH2O,N2,90°=6.07mm,$ whereas that for the air environment is $λH2O,air,90°=3.40mm$ (both adjusted R²> 0.998). Therefore, with increasing treatment distance d, the etching depth and presumably the density of the etchant species in the plasma jet effluent drop off more rapidly in air than in N₂ environment. This also suggests that the entrainment of ambient O₂ into the plasma jet effluent impacts the density of etchant species in Ar/H₂O plasma.

Besides etching, Ar/H₂O plasma also modifies the chemical composition of the etched polymer surfaces, which is illustrated by the high-resolution C 1s, N 1s, and O 1s spectra of the pristine and treated PS surfaces shown in Fig. 4. The electron take-off angle of the XPS measurements is 20°, which corresponds to chemical information from the top ∼2 nm of the PS films. The pristine PS film (black line in Fig. 4) shows no surface N and minimal amount of surface O. The pre-existing O might come from weak surface oxidation and/or the formation of a hydrocarbon contamination layer. The most abundant moiety on the pristine PS surface is the C—C/H bond (91.1% of the C 1s electrons, same below), and it can be found on both the main chain and the side ring of the PS structure. Additionally, π–π* shake-up (5.4%, comes from phenyl rings) and C—O bonds (2.54%, due to surface oxidation or hydrocarbon contamination) were also observed in the C 1s spectrum of the pristine PS.

After Ar/H₂O plasma treatment in either N₂ or artificial air environment, the PS surfaces showed the destruction of both C—C/H bonds and aromatic rings, accompanied by moderate surface oxidation and the formation of NO, which is similar to changes seen on PS surface treated by Ar/O₂ plasma jet.²⁹ The C 1s spectra of Fig. 4(a) show a decrease of C—C/H and π–π* shake-up peaks after treatment in addition to the formation of C—O, O—C—O/C=O, O—C=O, and O—CO—O groups. These effects are more pronounced for the PS films treated in air environment. The O 1s spectra in Fig. 4(c) show peaks that belong to the aforementioned O containing moieties with carbonate ester (O—CO—O) being the most abundant. Previously, we also observed the formation of this carbonate ester group for the Ar/O₂ plasma-treated PS surfaces.²⁹ Interestingly, we did not find the enrichment of the O—CO—O group on polymer surfaces treated by other types of APP sources including the surface microdischarge³⁹ and the double-ring kHz atmospheric pressure plasma jet.^28,44 This indicates that the O—CO—O formation might be characteristic to the polymer surfaces etched by the RF jet. Compared to PS surfaces treated in N₂, we find that the PS surface treated in artificial air is more oxidized. For example, in Fig. 4, at d = 12 mm, the O elemental composition of the PS surface treated in air is 29.2%, whereas that treated in N₂ is 19.0%.

The Ar/H₂O plasma treatment also modifies the polymer surface morphology. AFM measurements show that the RMS roughness of PS surface increases from 0.35 to 4.03 nm after Ar/H₂O treatment in a N₂ environment at d = 8 mm.

To evaluate how the etching depth and the surface chemistry of the polymer films change with treatment distance d, we performed both the ellipsometry and XPS measurements of PS films treated at various distances (d = 4, 8, 12, 16, and 20 mm). Due to the similarity of the real-time thickness profile and the high-resolution XPS data to that shown in Figs. 2 and 4, respectively, we only present the total etching depth and the O elemental composition abstracted from the raw ellipsometry and XPS data, which simplifies the description and eases the comparison of different operating conditions.

As shown in Fig. 5, with tilted configuration, we also observe the exponential decay of total etching depth as a function of treatment distance for both N₂ and air environments, which is similar to the vertical jet configuration illustrated in Fig. 3. However, the surface O elemental composition of the Ar/H₂O plasma-treated PS shows a maximum at 12 mm. This difference in the trends between etching and surface modification indicates that the etching and modification processes might be controlled by different surface reaction pathways, as will be discussed in Sec. IV C.

When comparing N₂ and artificial air environments in Fig. 5, we found that the additional O₂ from air environment has significant influence on both the etching efficiency and the surface modification capability of Ar/H₂O plasma. First, polymer etching in a N₂ environment is more effective than that in an air environment, especially for shorter plasma treatment distances. For example, at 4 mm, the etching depth in a N₂ environment is 106.9 nm compared to 40.1 nm in an air environment. Furthermore, the etching depth in Fig. 5 also decreases much faster with treatment distance in an air environment than in a N₂ environment, with fitted decay constants of $λH2O,air,30°=2.13mm$ and $λH2O,N2,30°=5.47mm$ (both adjusted R² > 0.978), respectively. These facts suggest that the O₂ molecules from the environment are able to impact the amount of etchant species bombarding the polymer surface. Second, the polymers treated by Ar/H₂O plasma in a N₂ environment is less oxidized than those treated in an air environment for all treatment distances. For treatments performed in the N₂ environment, the surface O originates mostly from the Ar/H₂O plasma plume, whereas in the air environment the environmental O₂ molecules may also participate in the surface oxidation processes. Summarizing the differences between N₂ and air environment as shown in Fig. 5, we conclude that the presence of O₂ reduces polymer etching but enhances polymer surface oxidation.

In Fig. 6, a detailed comparison of the etching and the surface modification of PS films treated by three different feed gas compositions, i.e., Ar + 1% H₂O, Ar + 1% O₂, and Ar + 0.333% H₂O + 0.667% O₂, is shown. All experiments in Fig. 6 were performed with the same tilted jet configuration (φ = 30°) at a fixed plasma power of 2 W. We selected N₂ environment instead of air to prevent any effect of additional O₂ from the ambient gas. The Ar + 1% H₂O plasma data in Fig. 6 are the same as that presented in Fig. 5, and to link the present work with our previous publications, the Ar + 1% O₂ plasma data are the same as that presented in our prior work.²⁹ 

As discussed in Secs. III B and III D, the polymer etching depth of both Ar/H₂O and Ar/O₂ plasma dropped exponentially with treatment distance, and the exponential decay constants for Ar/H₂O and Ar/O₂ plasma can be fitted as $λO2,N2,30°=4.15mm$ (adjusted R² = 0.995) and $λH2O,N2,30°=5.47mm$ (adjusted R² = 0.978), respectively.²⁹ In contrast, the surface O elemental composition of the PS films treated by Ar/H₂O and Ar/O₂ plasma show a maximum at d = 8 and d = 12 mm, respectively. When comparing the etching efficiency between Ar/H₂O and Ar/O₂ plasma, we find that Ar/H₂O plasma is more effective. The surface O elemental composition is also dramatically different between PS films treated by Ar/O₂ and Ar/H₂O plasma. As indicated in the right-axis of Fig. 6, the surface O composition ranges from 24.1% to 28.1% for the Ar/O₂ treated films, whereas that of Ar/H₂O plasma-treated films ranges from 5.5% to 12.0%. This difference in surface O can be attributed to two possible causes: (1) Ar/O₂ plasma contains more reactive species that lead to the surface oxidation or (2) Ar/H₂O plasma etches more material than Ar/O₂ plasma which hinders the accumulation of O containing surface moieties.

When simultaneously adding both H₂O and O₂ admixtures to the Ar feed gas, we observed much reduced etching depths but similar amount of surface O as those treated by Ar + 1% O₂ plasma. As shown in Fig. 6, the etching depth of PS films treated by Ar/H₂O/O₂ plasma is too small to show a clear trend. Interestingly, when comparing directly adding O₂ to the feed gas of Ar/H₂O plasma (shown in Fig. 6) with adding O₂ to the environment where the Ar/H₂O plasma operates (shown in Fig. 5), we observed similar changes on both the etching efficiency and the surface O composition of the plasma-treated polymers. The effect of O₂ from the environment gas shown in Figs. 2 and 5 might be due to its entrainment in the plasma effluent. The percentage of ambient gas entrainment on the axial direction of the RF jet can rise from a few percentages at 4 mm to near 40% at 16 mm as measured by molecular beam mass spectrometry,⁴⁵ which may have a large impact on the type and density of the reactive species generated in the jet effluent. However, we do not have enough evidence to determine whether the O₂ entrainment from environmental gas has a similar effect on the plasma property as directly adding O₂ to the feed gas.

We further studied the effect of the substrate temperature on polymer etching by both Ar/H₂O and Ar/O₂ plasmas. Both types of plasma treatments were performed at 4 mm with the vertical jet configuration (φ = 90°) and the same plasma power (2 W). The substrate temperature studied ranges from 0 to 90 °C. It is worth mentioning that the glass transition temperature of PS is near 100 °C, and the dew point of Ar + 1%H₂O mixture at atmospheric pressure is ∼4.8 °C.

As shown in Fig. 7, the etching efficiency of Ar/O₂ plasma increases with higher substrate temperature, and we estimated an apparent activation energy for Ar/O₂ etching reaction as ∼0.1 eV, which is similar to the previously identified value.²⁹ However, polymer etching by Ar/H₂O plasma has a more complex relationship with substrate temperature. At relatively high substrate temperatures (>70 °C), the PS etching depth increases with substrate temperature similar to that seen for Ar/O₂ plasma. In the temperature range from 4 to 70 °C, the polymer etching depth for Ar/H₂O plasma drops as the substrate temperature increases. For substrate temperatures below 4 °C, we observed liquid water condensation on the polymer surface. As shown in Fig. 7, the polymer etching depth falls rapidly at 0 and 2 °C.

The complex dependence of Ar/H₂O plasma etching efficiency on substrate temperature might be due to the change in surface adsorption of various gas phase species. For substrate temperatures lower than 4 °C, a condensed liquid water layer may form on the polymer surface which serves as a protection layer that shields off the reactive species and prevent the polymers from being removed. For the intermediate temperature from 4 to 70 °C, there are two possible channels that might result in the enhanced etching with lower substrate temperature. On one hand, the sticking coefficient of etchant species generated in the Ar/H₂O plasma, presumably OH radicals, might become larger with lower substrate temperature, which directly leads to the higher etching efficiency. On the other hand, the physisorption rate of water molecules on the surface might increase with lower substrate temperature as can be deduced from the isobar plot of water adsorption.⁴⁶ These adsorbed water molecules may react with the plasma generated reactive species and form additional etchants (e.g., OH). In addition, a change in gas temperature near the substrate boundary layer could also impact recombination reactions, although the anticipated effect in the investigated substrate temperature range is small.

Recent modeling results of Ar/H₂O plasma generated by APPJs show that H atoms can be produced at similar or even larger concentrations than OH radicals through various reaction pathways.^10,16 Since H atoms can also participate in the polymer etching process,⁴⁷ it is important to compare the etching efficiency between H atoms and OH radicals.

Figure 8(a) shows polymer etching depths by Ar/H₂ plasma with various treatment distances from 4 to 20 mm. A vertical jet configuration (φ = 90°), N₂ environment, and 2.4 mm/s scan-processing speed were used for these experiments. Compared to the data shown in Fig. 3, the power consumption of Ar/H₂ plasma (1.26 W) is smaller. However, this is the largest possible power for the stable operation of Ar/H₂ plasma at 4 mm treatment distance without sparking/arcing. Recent measurements of H densities by two-photon absorption LIF in the effluent of the same RF jet (without target materials) show that the RF jet can create H densities in excess of 10²² m⁻³, which are larger than the measured OH densities in Ar/H₂O plasma.⁴⁸ Since the dominant reactive species generated by Ar/H₂ plasma in nitrogen environment is the H atom, the data in Fig. 8(a) reflect the polymer etching behavior by H atoms. Only ∼20 nm of PS was etched after Ar/H₂ plasma treatment compared to 43.7 nm by Ar/H₂O plasma shown in Fig. 2. Moreover, for Ar/H₂ plasma, we did not find the exponential decay of etching depth with treatment distance as that seen by Ar/H₂O and Ar/O₂ plasma. As shown in Fig. 8(a), the PS etching depth did not change much as a function of distance from 4 to 20 mm, which suggests that the dominant etchant from Ar/H₂ plasma, i.e., H atoms, behaves differently from the etchant species produced in Ar/H₂O and Ar/O₂ plasma.

We also studied polymer etching by Ar/H₂ plasma in various N₂ + O₂ environment gas mixtures. As shown in Fig. 8(b), we observe a significant increase of polymer etching when operating the Ar/H₂ plasma jet in O₂ containing environments. The inset of Fig. 8(b) indicates a maximum of polymer etching depth for Ar/H₂ plasma operated in 99% N₂ + 1% O₂ environment. This increase in polymer etching efficiency with the presence of environmental O₂ might be due to the formation of OH radicals, or the occurrence of synergetic surface reactions for H atoms with other reactive species related to O₂. One possible pathway is the generation of OH radicals from the fast formation of HO₂ through reaction

In the condition of an excess in H atoms, HO₂ can be efficiently converted into OH radicals through reactions with H atoms

As reported previously,¹⁹ the reactivity for polymer H-abstraction by H atoms are 2–4 orders of magnitude smaller than those by O or OH, which also indicates that H atoms are less likely to be a significant etchant when O or OH is present. The decrease of etching depth by Ar/H₂ plasma with higher environmental O₂ concentration (>1%) might have a similar reason as that of Ar/H₂O plasma seen in Fig. 2(b), which will be discussed in Sec. IV B. Nonetheless, at larger treatment distances, the contribution of H radicals to the polymer etching by Ar/H₂ and Ar/H₂O plasma should be considered due to the low density of other short-lived species.

Polymer etching by low temperature plasma is closely related to the polymer chain scission reactions, since dry etching relies on the formation of small molecular weight volatile products, such as CO and CO₂, at room conditions. Based on the free-radical reaction mechanism of PSI,^19,20,29,49 the chain scission of polymers starts from the creation of radical sites, which could be through the bombardment of energetic species (ions, electrons, UV photons, etc.) and/or the chemical reaction with neutral gas phase radicals. An efficient pathway of creating surface radical sites by H₂O and O₂ containing plasma is the H-abstraction by atomic O and OH radicals, which has been widely reported for both low and atmospheric pressure plasmas:^19,20

where p is the estimated reaction probability of O atoms and OH radicals for the tertiary, secondary, and primary carbon sites provided by Bhoj and Kushner.²⁰ It can be seen that OH radicals have a much higher reaction probability than O atoms. In addition, the reaction rate of OH radicals is also faster than atomic O with gas phase organic molecules. For example, at room temperature (298 K), the rate constants of H-abstraction from methane by O atoms and OH radicals are^50,51

Reactions (6) and (7) indicate that OH radicals are much more reactive than O atoms. Previously, we evaluated the etching reaction coefficient of atomic O by correlating the gas phase atomic O density to the polymer etch rate, and we found that it was on the order of 10⁻⁴ which indicates that for removing one C atom from the polymer surface, ∼10⁴ O atoms from the RF jet effluent are needed.²⁹ 

Since our measurements suggest that the dominant etchant species in Ar/H₂O plasma are OH radicals, as discussed in Secs. III F and III G, we can estimate the etching reaction coefficient of OH radicals by correlating the etching depth shown in Figs. 3 and 5 with the amount of OH radicals generated by the Ar/H₂O plasma. Using LIF, we measured the OH density at 0.5 mm above the substrate positioned at 4, 8, and 12 mm away from the plasma jet nozzle. We find that the OH density decreases exponentially with the treatment distance, which is similar to the polymer etching depth. The average OH flux leaving the bulk of the gas phase $(Γ¯OH)$ can be estimated using the measured gas temperature (T_gas) and the density of OH radicals ([OH]) by

where $v¯$ is the average kinetic speed of the OH radicals, R is the gas constant, and M is the molecular weight of OH. If the resistance of mass transfer across the stagnant gaseous boundary layer near the target surface is neglected, Eq. (8) can be used to estimate the incident OH flux bombarding the material surface. Similar to the approach provided in our prior work,²⁹ we estimated the average carbon atom flux etched from the PS surface using the following equation:

where Δ_t is the etching depth in Figs. 3 and 5, A_treated is the scan-processed area (0.922 cm²), ρ is the density of PS (1.04 g/cm³), A_C = 12 and A_H = 1 are the atomic mass of carbon and hydrogen atoms, N_A is the Avogadro's number, ID is the inner diameter of the RF jet quartz tube (1.5 mm), and t is the total scan-processing time (200 s). This simplified estimation of etched C flux assumes that material etching only happens in the instantaneously treated area of π/4 ⋅ ID², where the RF jet directly hovers over, and this location is assumed to correspond to the area of which the OH flux is applied.

In Fig. 9, we show the correlation between the incident OH flux and the etched C flux both estimated at treatment distances d = 4, 8, 12 mm. The fitted slope of data yields the etching reaction coefficient³⁴ of OH radicals which is 1.95 ± 0.50 × 10⁻². This indicates that for removing one C atom, ∼10² OH radicals from the gas phase are required. Compared to the etching reaction coefficient of O atoms, this value is 2 orders of magnitude higher, which is consistent with the work of Dorai and Kushner¹⁹ who proposed reaction probabilities based on the work by Gomez et al. for O (Ref. 52) and rate constants for reactions of OH with long chain saturated hydrocarbons.⁵³ 

Previously, we studied the effect of various 1% (O₂ + H₂O) admixtures in the Ar feed gas with a wide range of mixing ratios between O₂ and H₂O. We found that, compared to Ar/O₂ or Ar/H₂O plasma, the polymer etching efficiency of Ar/(O₂ + H₂O) plasma at 4 mm treatment distance dropped dramatically.³⁵ In Fig. 6, we further examined the effect of (O₂ + H₂O) admixtures for longer treatment distances up to 20 mm, and we observed similar results. Interestingly, the direct addition of O₂ in the Ar/H₂O feed gas has a similar impact on the etching rate as the O₂ entrainment from the environment, despite the anticipated difference in chemical kinetics between the plasma core and the effluent.

In fact, this reduction of polymer etching efficiency when introducing O₂ in Ar/H₂O plasma or introducing H₂O in Ar/O₂ plasma is consistent with the difference in dominant etchant species from Ar/O₂ and Ar/H₂O plasma–presumably O atoms and OH radicals, respectively. Models of Liu et al.⁵⁴ for plasmas with He + H₂O/O₂ mixtures show that the O density monotonically decreases with the increase of H₂O:O₂ ratio while the OH density monotonically increases. This indicates that there might be a change in the dominant etchant species from OH to O when the H₂O:O₂ ratio varies from 1:0 to 0:1. Experimentally, we observed a minimum of the polymer etch rate when having the H₂O:O₂ = 1:2 in Ar + 1% (H₂O + O₂) plasma.³⁵ 

A plug flow model of the same Ar/H₂O plasma jet operating in air environment shows an increase in the O density at further distances off the nozzle, which is due to the higher O₂ concentration in the jet plume at these more remote locations.⁵⁵ As the OH density monotonically decreases with further distances, the O density becomes significantly larger than the OH density. As the estimated etching probability of O is 2 orders lower than that of OH, this increase in O density, which does not exceed the original OH density,⁵⁵ is not expected to compensate for the loss of OH. In addition, the expected increase in O density in air environment compared to N₂ environment will lead to a reduction in the OH density generated by Ar/H₂O plasma, because atomic O effectively reacts with OH to form HO₂ in a 3-body reaction [reaction (2)].¹³ This could explain the reduced etch rate of Ar/H₂O plasma in air environment compared to in N₂ environment.

The PSI involves the generation of reactive species in the plasma source, the transport of these reactive species to the target surface, and the chemical reactions of these species with the polymer surface.⁵⁶ When we extend the treatment distance from the RF jet nozzle to the target, the type and density of etchant species can be greatly altered due to the high concentration of reactive species and the short mean-free path of particles at atmospheric pressure.²⁷ 

As shown in Figs. 3–6, the etching efficiency of Ar/H₂O plasma drops exponentially with treatment distance regardless of treatment configuration (φ = 30° or 90°) and environmental gas composition (air or N₂). This indicates that the flux of etchant species bombarding the target surface, presumably OH, also drops exponentially with the treatment distance. We can define the apparent lifetime of OH radicals $τOH¯$ using the exponential decay constant of OH radicals λ_OH, which can be assumed as the same as the exponential decay constant of etching depth in Figs. 3 and 5:

where $vflow¯=14.15m/s$ is the average feed gas velocity. Therefore, we can estimate the apparent lifetime of OH in N₂ and air environment as $τOH,N2=0.41±0.02ms$ and τ_{OH, air} = 0.20 ± 0.05 ms, respectively. We can further estimate the density of OH radicals corresponding to this apparent lifetime following the approach described by Verreycken et al.⁵⁷ by assuming that the OH density is equal to the H density and the O density is much smaller than both the OH and H density. This leads to an estimated OH density of 9.8 × 10¹⁹ m⁻³, which is close to the OH density measured experimentally at 3.5 mm away from the nozzle.

Compared to $τOH,N2$⁠, the reduction of τ_{OH, air} by a factor of 2 is due to the change in the OH, H, and O density. We ruled out the possibility of OH density change as the only cause, because if it were true the OH density would be doubled, which is inconsistent with the change in polymer etch rate as observed in Figs. 3 and 5. If we consider an increase in O density, it would lead to an O density between 1 and 3 × 10²⁰ m⁻³ depending on the reduction in OH and H density. An O density that is a factor 5 times larger than the OH density will not lead to a switch of the dominant etchant from OH to O in view of the 2 orders of magnitude higher etching reaction coefficient of OH compared to O. Therefore, the change in O density is more likely the cause for the observed differences in the etch efficiency between air and N₂ environments.

When it comes to surface modification by Ar/H₂O plasma, we found that the O elemental composition of the etched polymer surface shows a maximum at intermediate treatment distance (∼12 mm) rather than dropping exponentially, as shown in Fig. 5. This suggests that etching and surface modification are controlled by different reactive species or the same species through different surface reaction processes. However, etching and surface modification might not be completely independent—the low surface O elemental composition at short treatment distance (<8 mm) can be due to the fast material removal, which prohibits the accumulation of oxidized surface sites.²⁹ The decrease of both etching depth and surface O composition at long treatment distances (>12 mm) in Fig. 5 is due to the reduction of reactive species arriving at the polymer surface. It is worth mentioning that the same relation between etching and surface modification was also observed for Ar/O₂ plasma-treated polymers,²⁹ as shown in Fig. 6.

We studied the interaction of Ar/H₂O plasma with polymers using PS and the well-characterized RF jet as a model system. We found that the Ar/H₂O plasma is able to etch polymers at a relatively high etch rate and the resulting etched polymer surface is mildly oxidized. We identified the dominant etchant species generated in Ar/H₂O plasma as OH radicals, and by correlating the amount of OH radicals arriving at the polymer surface with the etched C atoms leaving the polymer surface, we estimated the etching reaction coefficient of OH radicals to be of the order of 10⁻² which is 2 orders of magnitude higher than the etching reaction coefficient of O atoms. This indicates that OH radicals are much more reactive than O atoms in terms of etching polymers.

When changing the treatment distance of Ar/H₂O plasma, we find that the polymer etch rate drops exponentially, whereas the surface O composition shows a maximum at an intermediate distance. This difference between etching and surface modification indicates the complex reaction processes during plasma-surface interaction, as the etching and surface modification appear to be controlled by different surface processes.

We found that mixing O₂ into Ar/H₂O plasma, by either directly injecting it in the feed gas or through O₂ entrainment from the environment gas, reduces the etching efficiency of the Ar/H₂O plasma while resulting in higher O elemental composition on the etched surface. This suggests that the presence of O₂ in the Ar/H₂O plasma leads to a reduction of OH radicals while potentially enhancing the generation of weakly oxidative species that modify the polymer surface without etching it. In addition, we found that adding H₂O into Ar/O₂ plasma results in the reduction of its etching efficiency while leaving the polymer surface O composition relatively stable, which indicates that the presence of H₂O might reduce the atomic O density.

Finally, we tested the effect of substrate temperature on the polymer etch rate of Ar/H₂O plasma and observed a complex temperature dependence. The polymer etch rate decreases with increasing substrate temperature from 4 to 70 °C while an opposite trend was found for Ar/O₂ plasma. A possible explanation is the increase of the etchant (OH) sticking coefficient or the enhancement of H₂O adsorption on polymer surfaces at lower substrate temperature.

The authors gratefully acknowledge financial support by the National Science Foundation (NSF) (No. PHY-1415353) and the U.S. Department of Energy (No. DE-SC0001939). They thank H. Wang for his contribution on preparing the XPS data. They also thank C. Anderson and D. B. Graves of UC Berkeley for helpful discussions on this collaborative project. They are grateful to E. A. J. Bartis, D. Metzler, C. Li, and A. Pranda for helpful discussions and collaborations.