Interaction of long-lived reactive species from cold atmospheric pressure plasma with polymers: Chemical modification by ozone and reactive oxygen-nitrogen species

Cold temperature atmospheric pressure plasma (APP) has been used as a material processing technique for improving the wettability and adhesion of plastics,^1,2 sanitizing microbes,^3,4 promoting wound healing,^5,6 cancer treatment,^7,8 and promoting catalysis.^9,10 The effectiveness of the APP treatment comes from the interaction between the target materials and the plasma generated reactive species which include charged particles, photons, and neutral reactive atoms/molecules.¹¹ Based on the configuration of the APP source and its operating parameters,¹² the type and amount of reactive species interacting with the target material can be drastically different,^13,14 which leads to diverse impacts on the surface, subsurface, and for certain conditions even the bulk of the target materials.^15–17 In order to achieve controlled APP treatment that produces desired effects, it is essential to understand the role of individual reactive species in the plasma-material interaction processes.

To this end, previously we studied the effect of short-lived neutral species, i.e., atomic O^16,18 and OH,¹⁹ on material surfaces using an atmospheric pressure plasma jet and vinyl polymers. We found that both O atoms and OH radicals could lead to fast polymer etching at atmospheric pressure due to their high reactivity. We also evaluated the effect of ultraviolet (UV) photons²⁰ and charged species²¹ at atmospheric pressure using the poly-methyl methacrylate-based 193 nm photoresists (PR193), and we found that these energetic species can cause directional etching and material modifications.

Long-lived neutral species, although less reactive, are expected to be the dominant gas phase reactants for target materials placed in atmospheric pressure at a remote location¹⁶ since the incident fluxes of short-lived reactive species are greatly reduced for such a condition.^22,23 Recently, we studied the treatment of polystyrene (PS) using a prototypical remote plasma reactor named surface microdischarge (SMD), and we observed three polymer transformation stages during prolonged plasma exposure, i.e., (1) the oxidation and nitritation of the polymer surface, (2) the oxidation of the polymer bulk, and (3) etching.¹⁶ Despite the evident effects on materials, it is still unclear which plasma generated reactive species are responsible for these transformation stages, especially the bulk oxidation stage where the permeation of reactive species below surface might be involved. Besides, the depth of the plasma modified layer and the chemical reaction pathways are also unknown.

For understanding the effect of long-lived reactive species on materials, we characterize the type and density of long-lived reactive neutrals generated by the SMD using infrared (IR) and UV absorption techniques. Ultrathin (∼10 nm) PS film is used due to its simplicity. We characterize the changes of PS film thickness, elemental composition, and chemical bonds using ellipsometry, x-ray photoelectron spectroscopy (XPS), and attenuated total reflection (ATR)-Fourier-transform infrared spectroscopy (FTIR). By correlating the SMD-induced effects on the PS films with the gas phase characterization results, key reactive species that cause the PS film expansion, oxidation, and nitritation are identified. In another article published separately,¹⁷ we discuss the role of macromolecular structure, substrate temperature, and the diffusion of reactive species below the polymer surface. Together, the diffusion of O₃ into the polymer bulk and the chemical modification of the styrene-based polymers by O₃ are discussed.

We chose polystyrene (PS) for correlating the APP generated long-lived reactive species with material surface responses. Ultrathin (∼10 nm) PS films were prepared using the spin coating of 0.5 wt. % PS (Sigma-Aldrich) solution with propylene glycol methyl ether acetate (Sigma-Aldrich, ReagentPlus, ≥99.5%) as solvent. To achieve ∼10 nm film thickness, 6000 rpm spin speed with 3 s ramping time was used. Based on the intended use of the polymer films, we coated PS on three different types of substrates: (1) SiO₂/Si substrate with 100 nm-thick SiO₂ surface layer on Si, (2) SiO₂/Si substrate with 1.7 nm-thick native oxide (SiO₂) layer on Si, and (3) Au/Cr/SiO₂/Si substrate with 100 nm Au, 20 nm Cr, and 1.7 nm native SiO₂ stacked on Si. The SiO₂/Si substrates were used as they are, and the metallic layers (Au and Cr) were sputter-coated by an ATC 1800 Sputtering system (AJA International). PS films coated on the first type of substrate were used for ellipsometry studies, and films with the latter two types of substrates were used for material chemical composition analysis by XPS and ATR-FTIR. The thickness of the PS films and the SiO₂ layers was measured by both 633 nm single wavelength²⁴ and spectroscopic ellipsometer (Sophie STE70 and J.A. Woollam alpha-SE, respectively).

The APP reactor selected in this work is the surface microdischarge tube array (SMD-TA), which is a variation of the planar configuration SMD.^11,25 Details about this reactor have been described previously.^16,26 Briefly, the SMD-TA reactor is a type of surface dielectric barrier discharge (DBD) and it is composed of many individual discharge tubes as shown in Fig. 1(a). Each of these discharge tubes features a curved stainless steel (SS) woven mesh as the powered electrode, a 10 mm inner diameter quartz tube as the dielectric barrier, and a perforated center rod as the working gas distributor. Outside of the discharge tubes is the grounded metallic chassis that conducts electrical current and simultaneously serves as a heat sink. Cold temperature APP is generated around the SS mesh. The SMD-TA reactor used in this work contains seven discharge tubes with hexagonal packing arrangement for increasing the density of the reactive species near its nozzle. The reactor is powered by a sinusoidal power supply with 6 kV peak-to-peak (kV_pp) voltage at 23 kHz. The calculated plasma power density,¹¹ i.e., the power consumed by plasma per glow area, is less than 0.1 W/cm².

To tune the type and density of reactive species generated by the SMD-TA, 2 standard liters per minute (slm) of working gas with various N₂ (ultrahigh purity, 99.994%, H₂O <2 ppm, Airgas) and O₂ (ultrahigh purity grade, 99.999%, H₂O <1 ppm, Airgas) mixtures was fed through the discharge tubes. The working gas mixtures used in this work include 0% O₂, 20% O₂, 50% O₂, 70% O₂, 80% O₂, 90% O₂, 95% O₂, and 100% O₂ with the rest of the percentage as N₂ (same notation for working gas composition is applied throughout this report). A sealed 50 l chamber was used to control the plasma processing environment—the chamber was pumped to 20 mTorr before every experiment and then refilled with 4 slm working gas to atmospheric pressure. The SMD-TA reactor was mounted on a 1-dimensional (1D) scanning stage^16,18 that was programmed to transport the plasma source between the following two positions with a speed of 9.6 mm/s: (1) the treatment position where SMD-TA stays on top of the PS film for 10 s, (2) the ellipsometry data acquisition position where SMD-TA is 4.5 cm away from the treatment position and stays stationary for 2 s. Therefore, each processing cycle takes 21.375 s.

Since the electrical configuration and the discharge behavior of the SMD-TA are very similar to the planar configuration SMD reactor,^16,26 we use the SMD-TA to study the generic effect of SMD on polymers with the benefit of controlled working gas composition and refreshing rate. Qualitatively comparable results between the SMD-TA and the SMD were obtained for both the plasma generated long-lived species and the plasma induced effects on polymers, although the planar SMD is slightly more effective for oxidizing polymers. In the rest of this work, we use the term SMD to represent the SMD-TA reactor.

The gas phase reactive species generated by the SMD were characterized by both IR and UV absorption. As shown in Fig. 1(b), we measured the IR spectrum of the SMD exhaust using an FTIR (Shimadzu IRtracer 100) equipped with a liquid nitrogen-cooled HgCdTe (MCT) detector. To quantify various types of reactive neutrals with strongly differing densities, a variable path gas detection cell (PIKE Technologies, 1–16 m variable, Part # 163-1618) with optical path length set at 1.33 and 16 m was used. The gas cell has a volume of 3.5 l, and was flushed with N₂/O₂ working gas mixture for 20 min before spectrum acquisition. The IR absorption background was taken prior to the onset of plasma reactor. To guide the flow of reactive species into the detection cell, the outlet of the gas cell is connected to a mass flow controller (set at 2 slm) and a vacuum pump. To reduce noise signal from ambient humidity and CO₂, the optical path of the IR light outside of the gas detection cell was constantly flushed by pure N₂, as shown in Fig. 1(b). The IR spectra of the SMD effluent were acquired in the range of 4000 –600 cm⁻¹ with 0.5 cm⁻¹ resolution and 20-scan averages.

Based on Beer's law, a homemade UV (254 nm) absorption system^27,28 with 1 cm optical path length was used to quantify the density of ozone generated by the SMD reactor. The UV absorption setup is composed of a Hg-Ar lamp, a 10 nm band pass filter centered at 254 nm (Edmund), a 34 mm² effective area deep UV photodiode (Edmund), a high gain operational amplifier circuit (LMP7721, Texas Instruments), and optical fibers. The optical path of the UV absorption measurement was located ∼3 mm away from the nozzle of the SMD reactor. For O₃ density calculation, cross section value $σO3=1.15×10−17cm2$ was used without further calibration.²⁷ 

We used in situ ellipsometry to monitor the changes of optical properties of the PS films during plasma treatment. The ellipsometer (Sophie STE70) has an He-Ne laser (λ = 632.8 nm) with ∼72° incident angle and an automated rotating compensator in the polarizer-compensator-sample-analyzer configuration.^15,16 Upon the reflection of polarized laser beam on polymer sample, the ellipsometer records the change in the phase difference (Δ) and the angle (Ψ) whose tangent is the ratio of the magnitudes of the total reflection coefficient.²⁹ Due to the scan-processing by the SMD, ellipsometry data were taken for 2 s after every 19.375 s. Polymer thickness and refractive index values were fitted from the measured Ψ and Δ data using optical models described previously.¹⁶ 

To characterize the chemical composition of PS surface, we used an XPS system (Vacuum Generators ESCALAB MK II) with nonmonochromatic Al K_α radiation (hν = 1486.6 eV) as the light source. The survey and high-resolution C 1s, N 1s, O 1s XPS were obtained at electron take-off angles of 20° and 90° (probing depth ∼2 and ∼8 nm, respectively).³⁰ The peak fitting and elemental composition analysis were performed using casaxps software with parameters described as following.^15,16,31 First, we fitted the C 1s spectrum with peaks corresponding to C‒C/H (285 eV), C‒O (286.5 eV), O‒C‒O/C˭O (287.9 eV), O‒C˭O (289.1 eV), O‒CO‒O (290.2 eV) and π–π* shake-up (291.5 eV). Second, only organic nitrate (ONO₂, 408.2 eV) peak was fitted in the N 1s spectrum due to the absence of other nitrogen peaks.^32,33 Last, considering the small chemical shift of O 1s core electrons, two peaks at 532.7 and 533.9 eV with each representing oxygen moieties with binding energy from 532.2 to 533.1 eV and from 533.6 to 535.3 eV, respectively, were fitted.¹⁵ After calibrating the C‒C/H peak to 285 eV, we performed Shirley background subtraction to all spectra. For quantifying surface elemental composition, Scofield cross section values³⁴ of C 1s, N 1s, and O 1s spectra, i.e., 1, 1.77, and 2.85, respectively, were used as the relative sensitivity factors for quantifying the measured peak areas.

The enhanced ATR-FTIR spectra²⁵ of the ∼10 nm PS films were obtained using the same FTIR equipped with MCT detector as the gas phase IR characterizations. A variable angle single reflection ATR accessory (VeeMaxIII, PIKE Technologies) along with a ZnSe polarizer was used to achieve p-polarized ATR-FTIR at 60° incident angle.²⁵ The internal reflection element of the ATR accessory is a Ge crystal. A constant torque pressure clamp with 7.8 mm diameter tip (PIKE Technologies) was applied to maintain consistent contact between the Ge crystal and the polymer samples. The optical path of the IR light was constantly purged by pure N₂ gas. The reference backgrounds were obtained with no sample coupled to the Ge crystal. The IR spectra were collected in the range of 3400–600 cm⁻¹ with 4 cm⁻¹ resolution averaged over 20 scans. Multiple-point baseline correction was consistently applied to all the acquired ATR-FTIR spectra as described previously.²⁵ 

Based on the position and shape of the IR absorption peaks, we are able to identify a number of long-lived reactive neutrals produced by the SMD reactor. As illustrated in Fig. 2, the IR absorption spectrum of the SMD effluent with 20% O₂ (artificial air) working gas showed peaks from four reactive neutrals, namely, O₃, N₂O₅, N₂O, and HNO₃. Based on Beer's law,¹⁷ the density of these species can be determined using their corresponding IR peak intensity and cross section values reported previously. For data shown in Fig. 2, we can estimate the density of N₂O₅,³⁵ N₂O,^36,37 and HNO₃ (Ref. 38) as $1.63±0.09×1015cm−3$⁠, $7.90±0.55×1014cm−3$⁠, and $6.23±0.87×1014cm−3$⁠, respectively. The density of O₃ was estimated based on UV absorption measurement at 254 nm with cross section $σO3=1.15×10−17cm2$⁠,²⁷ and it is $2.80×1016cm−3$ at 3 mm away from the nozzle of the SMD.

When changing the mixing ratio between N₂ and O₂ in the working gas, we found that the densities of the long-lived reactive neutrals varied. As shown in Figs. 3(a) and 3(b), the position and shape of the IR peaks from N₂O, O₃, N₂O₅, and HNO₃ remain the same, but the absorbance of these IR peaks changes with working gas composition. We also examined the absorption spectrum of the SMD effluent with 16 m optical path length, but no observable IR peaks corresponding to NO, NO₂, and NO₃ species were found for all working gas compositions. This suggests that the density of these NO_x species is below the detection limit of our FTIR setup $(1×1014cm−3)$⁠.

The low density of NO_x species from SMD is consistent with the previous reports studying DBD-based ozonizers in which only oxides N₂O and N₂O₅ were detected besides ozone.³⁹ Eliasson and Kogelschatz reported that lower oxides NO and NO₂ can be generated in high power mode where ozone density is significantly reduced (discharge poisoning or ozoneless mode).⁴⁰ Similarly, Shimizu et al. reported a mode transition of SMD reactors from “ozone mode” to “nitrogen oxides mode” at power density higher than 0.1 W/cm².⁴¹ Considering the low power dissipation of our reactor (<0.1 W/cm²) and the high density of O₃, it is expected that the NO and NO₂ are not present in the IR spectra. For the HNO₃, we suspect that the additional source of hydrogen might come from the water molecules adsorbed on the surface of the reactor, gas lines, and the variable path gas detection cell. Another possibility is the reaction between the plasma generated reactive species and the gas lines made of polyethylene.

In Fig. 4, we show the calculated densities of O₃, N₂O₅, N₂O, and HNO₃ as a function of the O₂ percentage in the working gas mixture. Since we did not calibrate the IR and UV absorption measurements with known density of the corresponding reactive species, the density values of estimated in this work may only be accurate to the order of magnitude. However, the density relationship of the same species measured in different working gases should still hold. We find that O₃ is the most abundant reactive species generated by SMD with a maximum density of $1.38×1017cm−3$⁠. Similar to the results shown in Fig. 3, the highest densities of O₃ and reactive nitrogen species (RNS) appear at different working gas compositions, namely, 95% O₂ and 50% O₂, respectively. For all working gas compositions besides pure N₂, the density of O₃ is 1–2 orders of magnitude higher than that of RNS. When comparing different RNS in Fig. 4(b), we find that the density of N₂O₅ is always higher than that of N₂O and HNO₃. Therefore, O₃ and N₂O₅ are the two dominant long-lived reactive neutrals generated by the SMD. Although it is well known that small amounts of nitrogen additives can enhance ozone generation in DBDs,⁴² it is unclear why the RNS reached peak density when the O₂/N₂ working gas ratio is close to 1:1. More detailed discussions of gaseous reaction pathways of the SMD reactor using N₂/O₂ working gas can be found in the article by Sakiyama et al.¹⁴ 

Although charged species and short-lived reactive neutrals such as N, O, OH, O₂(a¹Δ_g), and HO₂ can be generated in the discharge region of the SMD reactor, as demonstrated by Sakiyama et al.,¹⁴ minimal amount of these species was found atop the sample surface located 3 mm away from the nozzle. Due to the electrode configuration of the SMD reactor and the small mean free path (<100 nm) of gas molecules at atmospheric pressure,⁴³ the charged species are mostly confined in the discharge area around the metal mesh, which is positioned inside the quartz tubes as shown in Fig. 1(a). Based on the reactionless fluid model of the transport of neutral species,¹⁵ the average gas velocity at the nozzle of the discharge tubes can be estimated as 0.21 m/s, which indicates that it takes more than 14 ms for the gas molecules to be delivered to the position located 3 mm away from the nozzle. This is much longer than the lifetime of the short-lived neutrals at atmospheric pressure, which is typically less than 1 ms, although we did not consider the influence of the stationary boundary layer at the polymer surface where the transport of species might mostly rely on diffusion.

We also performed the IR emission $(a1Δg→X3Σg−)$ measurement of O₂(a¹Δ_g) with an identical setup as used by Sousa et al.,⁴⁴ and we found that the density of O₂(a¹Δ_g) is below the detection limit of $4×1013cm−3$ for all tested working gas compositions. Therefore, the dominant reactive species at the polymer surface are the long-lived reactive neutrals as characterized by the FTIR.

To understand how polymers respond under the exposure of the reactive plasma species generated by the SMD reactor, we performed in situ ellipsometry measurement of the ultrathin PS films (11–12 nm) during treatment with different working gas compositions. In Fig. 5, we show the in situ Δ-Ψ plot from the ellipsometry measurement along with the fitted thickness and refractive index values obtained by applying a single layer optical model.¹⁶ Although we have investigated all working gas compositions in Fig. 4, here only the PS films treated by three representative working gas compositions, i.e., 95% O₂, 80% O₂, and 20% O₂, are shown as examples.

First, for the PS film treated with 95% O₂ working gas shown in Figs. 5(a) and 5(b), we observe three distinct polymer transformation stages.¹⁶ Prior to the SMD treatment, the Δ-Ψ plot of the PS film [Fig. 5(a)] begins at point A which corresponds to a pristine film thickness of 11.67 nm. When we start treating the PS film, the Δ-Ψ trajectory moves from point A to two consecutive turning points labeled as B and C, and then to point D, which is near the end of the treatment. Previously, we examined the behavior of PS films under similar remote treatment conditions by the SMD, and we attributed the three sections AB, BC, and CD in the Δ-Ψ plot to three polymer transformation stages,¹⁶ i.e., (1) AB: surface oxidation and nitritation stage; (2) BC: thickness expansion stage resulted from the oxidation of the PS film; and (3) CD: thickness reduction stage due to etching. In Fig. 5(b), we show the fitted thickness and refractive index of the PS film as a function of time. We also labeled the three transformation stages corresponding to those in Fig. 5(a), and we find that (1) the surface oxidation and nitritation stage takes ∼85 s and shows a thickness increase of 0.26 nm, (2) the thickness expansion stage shows a thickness increase rate of 2.27 nm/h, and (3) the etching stage shows an etch rate of 0.64 nm/h. The continuous drop of refractive index in Fig. 5(b) indicates the oxidation of the polymer film throughout all stages.

Similarly, for the PS film treated with 80% O₂ working gas, we also observed the two turning points B and C in Fig. 5(c). For the AB stage, the maximal thickness gain due to surface oxidation and nitritation is about 0.40 nm, which is achieved after ∼85 s of treatment as well. However, as shown in Fig. 5(d), the thickness expansion rate in the BC stage drops to 2.16 nm/h. We did not observe an evident etching stage in Fig. 5(d), which is because that the thickness expansion from bulk oxidation is larger than the thickness reduction from etching.¹⁶ These evidences indicate that the reactive species generated by 80% O₂ working gas are less reactive than those produced by 95% O₂ working gas.

For the PS film treated with 20% O₂ working gas, no turning points in the Δ-Ψ plot [Fig. 5(e)] can be found. As shown in Fig. 5(f), the plot of fitted PS film thickness and refractive index indicates that during 4 h of treatment, the PS film only experiences a slow thickness increase stage with an expansion rate of 0.37 nm/h. Compared to the other two working gas conditions discussed above, the surface oxidation/nitritation and the etching stages are absent. This indicates that the effluent of SMD with 20% O₂ working gas is even less reactive.

In Fig. 6, we correlate the polymer thickness change in the three transformations stages with the working gas composition. First, as shown in Fig. 6(a), for the thickness gain in the surface oxidation and nitritation stage, we find that the treatment time required for completing this stage is about 85 s for all working gas compositions except 100% and 20% O₂. Second, in Fig. 6(b), we plotted the PS film expansion rate during the bulk oxidation stage as a function of working gas composition. We find that the expansion rate shows a maximum at 95% O₂ working gas and then drops quickly with less O₂ in the working gas. The shape of this curve resembles that of the O₃ density, as shown in Fig. 4(a). Last, in Fig. 6(c), we show the correlation between the polymer etch rate in the etching stage and the working gas composition. Since we did not observe evident film thickness reduction when the working gas has less than 80% O₂, zero etch rate was shown for these working gas mixtures. We find that the polymer etch rate has a maximum at 95% O₂ working gas; however, the etch rate curve in Fig. 6(c) does not resemble that of the O₃ density curve. This indicates that material etching might be related to O₃ but not controlled by it.

Besides polymer thickness and refractive index, we also characterized the chemical composition of PS films using XPS. In Fig. 7, we show the high-resolution C 1s, N 1s, and O 1s spectra of the pristine and the 0.5 h treated PS films using working gas with 20% O₂ and 95% O₂ compositions. Only spectra with 20° electron take-off angle are shown, which contain the chemical composition from the top 2 nm of the PS films. Since there is neither O nor N element in the pristine PS structure, the untreated film shows only two C 1s peaks with 94% of the core electrons belonging to the C‒C/H bond, whereas the rest coming from π–π* shake-up satellite.

After 0.5 h of treatment with both working gas compositions, we observe the decrease of the C‒C/H and π–π* shake-up peaks and the rise of the oxygen containing functional groups such as C‒O, O‒C‒O, C˭O, O‒C˭O, and O‒CO‒O in Fig. 7(a), which indicates the oxidation of the aromatic ring and possibly the polymer main chain. Moreover, the N 1s spectra of the treated PS films show the emergence of organic nitrate (ONO₂) peaks at 408.2 eV. Correspondingly, the oxidation and nitritation of PS films are confirmed by the O 1s spectra shown in Fig. 7(c): the broad O 1s peak formed after SMD treatment can be attributed to both the organic nitrate and the O containing functional groups seen in the C 1s spectra. When comparing the spectra from 20% O₂ and 95% O₂ treated PS films, we find similar amount of ONO₂ but more surface oxidation for the 95% O₂ treated film. This difference in surface oxidation is consistent with that of the thickness expansion rate shown in Fig. 5(b), where 95% O₂ treated film shows higher expansion rate, which indicates more oxidation to the polymer films.

Besides the 20% O₂ and 95% O₂ treated PS films, we further characterized the XPS of PS films treated by other working gas compositions studied in Fig. 4. For each gas condition, two treatment times, namely, 0.5  and 2 h, were performed on PS films for investigating the surface chemical composition at different polymer transformation stages. Since all high-resolution spectra of the treated PS films share the same peak features as those shown in Fig. 7, we present the calculated elemental composition of C, O, and N as a function of working gas composition instead of the high-resolution spectra. As illustrated in Fig. 8(a), the surface O compositions show maxima when the PS films are treated with large amount of O₂ in the working gas mixture (95% and 80% for 0.5 and 2 h treated PS films, respectively). Correspondingly, the surface C compositions show minima at the same working gas composition. We also find that longer treatment time generally leads to higher surface O composition, which indicates that the polymer surface oxidation depends on the time integrated flux, or dose,^16,26 of the reactive species produced by the plasma treatment. Interestingly, the surface O composition curve in Fig. 8(a) strongly resembles the O₃ density curve and the thickness expansion rate curve in Figs. 4(a) and 6(b), respectively. We will further discuss their correlations in Sec. IV A.

Furthermore, we find that the surface N composition drops dramatically when the working gas contains more than 80% O₂, as shown in Fig. 8(b). Since all surface N comes from the ONO₂ group as indicated in Fig. 7(b), the trend in Fig. 8(b) also represents that of the ONO₂ group. The small amount of surface N on PS films treated with 100% O₂ working gas might come from the residual N₂ absorbed on the chamber wall and/or leaks during the refilling of the chamber to atmospheric pressure. When comparing the two treatment times, i.e., 0.5 and 2 h, we find that longer treatment time leads to slightly larger amount of surface N composition, especially with less O₂ in the working gas. The surface N composition of the SMD-treated PS films is relatively high (>3%) compared to the typical value of less than 1% for polymers treated by other APP sources such as the plasma jets, corona discharge, and direct DBDs. This might have resulted from (1) the prolonged treatment time which accumulates the surface N composition and (2) the negligible amount of material removal by etchants, which may cause the desorption of N containing surface moieties.

In Fig. 9, we show the C 1s decomposition of PS films treated by the SMD for 0.5 h as a function of working gas composition. Each line in Fig. 9 represents the percentage of C 1s electrons coming from one surface moiety observed in the high-resolution C 1s spectra shown in Fig. 7(a). The relatively large chemical shift from different functional groups in the C 1s spectrum makes it suitable for identifying and comparing the relative concentration of these surface moieties. From Fig. 9(a), we find that higher O₂ concentration in the working gas results in smaller amount of C‒C/H and π–π* shake-up groups on the treated PS surface, which indicates that more aromatic ring structures and possibly the polymer main chains are destroyed with higher O₂ concentration in the working gas. A minimum of the C‒C/H bond can be found on the 95% O₂ treated PS surface. Correspondingly, in Fig. 9(b), we find that higher O₂ concentration in the working gas could lead to larger amount of O containing functional groups forming on the treated PS surface. Similarly, maxima of these O containing groups are seen with the 95% O₂ treated surfaces. This indicates that the C‒C/H bonds and the aromatic rings have been converted into various O containing moieties by the SMD treatment. As indicated in Fig. 9(b), when comparing the relative concentration of the O containing functional groups for the 0.5 h treated PS films, we find the following relation: C‒O > O‒C‒O/C˭O > O‒C˭O > O‒CO‒O.

In order to show the effect of extended treatment time on the PS surface, we calculated the difference of C 1s composition between the 2 h and the 0.5 h treated PS films, as shown in Fig. 10. The positive values in Fig. 10 indicate that 2 h treatment leads to higher concentrations of the specified surface moieties than 0.5 h treatment. It can be seen that although longer treatment time with 20% O₂ (artificial air) working gas results in the universal increase of all O containing moieties, generally longer treatment time with higher O₂ percentage in the working gas tends to cause the accumulation of O‒C˭O groups. As discussed previously,¹⁶ such accumulation of O‒C˭O bond might be due to the unique etching mechanism of SMD. After 2 h of treatment with high O₂ concentration (≥80%) working gas, the PS films have proceeded to the etching stage as indicated in Fig. 5(c), whereas the PS films treated for 0.5 h is still in the expansion stage. In the etching stage, other functional groups are converted to O‒C˭O, which might be the precursor for the formation of etching product CO₂. The concentration of surface O‒C˭O will eventually saturate when a dynamic balance between etching and oxidation is established on the polymer surface. More detailed discussion can be found elsewhere.¹⁶ 

Because the penetration depth of IR light in ATR configuration (hundreds of nanometers)⁴⁵ is much deeper than the thickness of our PS films (∼10 nm), the ATR-FTIR spectra reflect the average chemical composition of the entire ultrathin PS film.^26,45 In contrast, the XPS with 20° take-off angle provides chemical information from the top 2 nm of the polymer film.

In Fig. 11, we show the ATR-FTIR spectra of the pristine and the 0.5 h treated PS films coated on Au surface, which serves as the enhancing substrate.^26,46 Due to the lack of a universal reference peak in this sampling configuration, all spectra are normalized to their own highest IR peak intensity.²⁵ For the pristine ultrathin PS film, characteristic vibrational modes of PS are observed and labeled as A through H in Fig. 11. Besides peak G (aliphatic chain C‒H stretch), all labeled IR peaks are from the aromatic ring structure. The two broad IR bands marked as area I and area IV can be attributed to carbonyl and ether groups respectively, and these features indicate that either the pristine PS surface or the PS/Au interface is weakly contaminated by hydrocarbon or surface oxidation.

After 0.5 h of plasma treatment, we observe the significant growth of four IR bands marked as I through IV in the shaded areas of Fig. 11. These new bands have a broad width, which suggests that the product of plasma modification does not have a homogeneous structure. Similar to our previous observation,^17,26 the IR band in area I can be assigned to the general carbonyl (C˭O) stretching modes (1600–1800 cm⁻¹) and asymmetric NO₂ stretching mode (1615–1660 cm⁻¹) from organic nitrate.^47,48 The two peaks labeled within area I (dashed lines) may come from saturated carbonyl and aromatic carbonyl, respectively. The aromatic carbonyl has lower wavenumber because the conjugated system of aromatic ring and carbonyl carbon reduces the force constant of the C˭O bond.⁴⁹ Furthermore, we can assign the IR peak in area II (dashed line, 1290 cm⁻¹) to the symmetric NO₂ stretch of organic nitrate group⁵⁰ and possibly aromatic ketone and aromatic ester. The IR bands in areas III and IV are mostly from the C‒O stretch of ether and ester groups. Due to the absence of OH stretch band (3500–2500 cm⁻¹), aldehyde C‒H stretch (2850–2700 cm⁻¹) and aldehyde C‒H bend (near 1390 ± 10 cm⁻¹) peaks in Fig. 11, we can eliminate the possibility of having alcohol (-OH), carboxyl (-COOH) and aldehyde (O˭C‒H) groups in the plasma treated films.

Since the spectra in Fig. 11 are not normalized to a generic reference peak with constant intensity, the spectra from different samples cannot be directly compared. In order to compare the spectrum of PS films treated by different working gas compositions, we performed the ATR-FTIR characterization of ultrathin PS films coated on SiO₂/Si substrates, which have a 1.7 nm constant thickness SiO₂ layer. The Si–O‒Si peak at 1225 cm⁻¹ from the native SiO₂ can be used as the reference peak for unit intensity.²⁵ 

As shown in Fig. 12, the IR spectra of PS films coated on SiO₂/Si substrate have similar IR features as those labeled in Fig. 11 except an extra band in the region of 1280–1010 cm⁻¹, which can be assigned to Si–O‒Si stretch. Since the peak intensity ratio to the Si–O‒Si stretch at 1225 cm⁻¹ is calculated and presented in Fig. 12, the relative concentration of the same functional group from different samples can be directly compared.²⁵ 

As illustrated by peaks A through F in Fig. 12, we notice that the IR peak intensity of aromatic ring diminishes for films treated by all working gas compositions. This is especially evident for the 95% and 100% O₂ treated PS films, where the IR peaks from aromatic ring are completely eliminated. Moreover, the plasma treated PS films also show the formation of ether (shaded areas III and IV), organic nitrate (area II), and carbonyl (area I) groups. This evidence confirms that the SMD treatment can destroy aromatic ring structures and convert them into O containing functional groups, such as ether, ester, ketone, and organic nitrate.

When comparing different ATR-FTIR spectra in Fig. 12, we find that the chemical composition of the SMD-treated PS films changes with working gas compositions. In Fig. 13(a), we summarize the relative amount of aromatic ring in PS films as a function of working gas composition using the peak A (aromatic ring out-of-plane deformation) intensity ratio of the treated PS films to the pristine PS film. We can see that there are fewer aromatic ring structures left in the PS films after treatment with higher O₂ concentration working gas. The shape of the curve in Fig. 13(a) strongly resembles that of the C‒C/H curve measured by XPS in Fig. 9(a). Furthermore, we can also calculate the relative amount of carbonyl group using the integrated area intensity of IR band I in Fig. 12. As shown in Fig. 13(b), we find that the carbonyl group in the treated PS films shows a maximum at 80%–95% O₂ working gas composition, and the curves have roughly the similar trend as that of the O containing surface moieties in Fig. 9(b). In addition, we also find that longer treatment time could lead to more aromatic ring being destroyed and more carbonyl being formed.

Besides, when comparing the two carbonyl peaks in the shaded area I of Fig. 12, we find that SMD treatments with lower O₂ concentration working gas might favor the formation of aromatic carbonyl (∼1672 cm⁻¹), whereas working gases with higher O₂ concentration might favor the formation of saturated/aliphatic carbonyl (∼1740 cm⁻¹). We think that this shift from aromatic carbonyl to aliphatic carbonyl comes from the further oxidation of the aromatic ring, which indicates that the plasma effluent is more oxidative when larger amount of O₂ is in the working gas. Evidence can be found in Fig. 13(a), where more aromatic rings are destroyed with more O₂ in the working gas.

In order to determine the role of the long-lived reactive species generated by SMD on polymers, we correlate the thickness and chemical composition of the treated PS films with the incident flux and dose of the gas phase reactive species.

The transport of reactive species from the bulk of the gas phase to the polymer film is an interphase mass transfer problem.⁵¹ Different from low pressure plasma processing of materials where reactive species are typically in the molecular flow regime, at atmospheric pressure, a stagnant gas layer is expected at the interface to the solid substrate. This biphasic transport of gaseous reactants into thin films has been discussed in prior works such as the well-known Deal–Grove model of the thermal oxidation of silicon.^52,53 To establish a model that is suitable for describing the transport of gaseous reactive species onto polymer surface, we followed a similar approach as the Deal–Grove model, with the assumption that the transport of the reactive species in both the gas phase and the polymer film is in steady-state.

While the plasma induced oxidation of polymers and the thermal oxidation of silicon share similar aspects, the actual transport and reaction processes in the polymer are much more complex. For polymers like PS, a variety of bonds exist with different reactivity, and many oxidation reactions can result in the formation of gaseous products. Therefore, while the Deal–Grove treatment of the interphase mass transfer problem may be applicable to treat the initial incomplete oxidation of the polymer, i.e., the first and second stage of polymer transformation, as the polymer becomes more oxidized and volatile products form in the later stages of the interaction, a different treatment is required.

As shown in Fig. 14, the interfacial transport process can be visualized as reactive species passing through a stagnant gas film $(δg)$⁠, a gas–solid interface, an oxidized solid film $(δs)$⁠, and an interface between the oxidized and nonoxidized polymer sections. The flux of reactive species coming from the bulk of the gas phase, on the interface, in the oxidized section of the polymer, and at the oxidized/nonoxidized polymer interface is denoted by $F0$⁠,$F1$⁠, $F2$⁠, and $F3$⁠, respectively.

Based on the kinetic theory of ideal gases, the average flux of species from the bulk gas phase $F0$ (arriving at the beginning of the stagnant gas layer) can be estimated as

where $Cgb$ is the concentration (number density) of reactive species in the bulk of the gas phase, $v¯$ is the average gas speed, R is the gas constant, T_gas is the gas temperature near material surface, and M is the molecular weight of the reactive species. This impingement rate $F0$ is typically used for low pressure conditions where the mean free path of particle is comparable or larger than the treatment distance, but at atmospheric pressure, the actual flux interacting with the material surface will be smaller than this value.

In Fig. 14, we denote $Cgb$ as the concentration of reactive species at beginning of the stagnant layer, $Cgs$ as the concentration of reactive species right next to polymer surface (in gas phase), $Cps$ as the concentration of reactive species at the surface of polymer (within the solid phase), $Cpi$ as the concentration of reactive species at the oxidized/nonoxidized polymer interface. Based on the general gas equation, the concentration of the reactive species $(Cgj)$ in the gas phase can be related to partial pressure $(pgj)$⁠,

where $kB$ is the Boltzmann constant and T is the temperature. If the resistance of mass transfer in the gas phase solely comes from diffusion and the concentration profile is linear, Fick's first law can be used to express the flux $F1$⁠,

where $Dox,g$ is the effective diffusion coefficient in the gas phase. In general, Eq. (3) is expressed in terms of mass transport coefficient $hg$ of reactive species in the gas phase⁵² 

and in the simplest case, we have $hg=Dox,g/δg$⁠. According to Eq. (2), the concentrations in Eq. (4) can be replaced by partial pressures

The concentration of reactive species at the surface of the polymer $Cps$ can be linked to the partial pressure of the reactive species in the gas phase by Henry's law constant H,

The partial pressure of the reactive species in the bulk gas phase $pgb$ can be related to its equilibrium concentration in the oxidized section of the polymer film $Cp∗$ through Henry's law⁵³ 

Replacing the $pgs$ and $pgb$ in Eq. (5) by that in Eqs. (6) and (7), we have

The dose of the reactive species interacting with a unit area of material surface can be calculated as the time integrated infusing flux $F1$⁠. If we further assume that for the same plasma treatment $Cgb$ and $Cps$ do not change over time

Following the Deal–Grove model and similar to Eq. (3), we can express the flux of reactive species in the oxidized section of the polymer film as

where $Dox,p$ is the effective diffusion coefficient of reactive species in the polymer.

The flux of reactive species at the interface between the oxidized/nonoxidized polymer $F3$ can be approximately expressed by a first order reaction with rate constant $kr$

Because of the differences between the thermal oxidation of silicon and the plasma oxidation of polymers, we will not further investigate the actual oxidation processes and their sequential reactions within the polymer.

With regard to the initial oxygen uptake by the polymer, the flux and dose through the gas–polymer interface can be approximately described by Eqs. (8) and (9) if the formation of volatile product can be ignored, which is the case initially. Replacing $Cp∗$ in Eqs. (8) and (9) with the measurable concentration of reactive species in the bulk gas phase $Cgb$⁠, we have

Since $Cps$ is controlled by the interfacial mass transfer, its value should not change much for the various plasma operating conditions discussed in the Sec. III, whereas $Cgb$ varies dramatically as plasma operating conditions are changed. Therefore, according to Eqs. (1), (12), and (13), the incident flux $F1$ and dose $D1$ of reactive species are approximately proportional to $Cgb$ and $F0$⁠. Because $Cgb$ was experimentally estimated in Sec. III A, in the rest of the discussion, we correlate the plasma modification of polymers with $Cgb$⁠.

Since O₃ is the most abundant product in the SMD effluent and its density is one to two orders of magnitude higher than that of the RNS, we first correlate the observed expansion of polymers with the measured O₃ density. Combining data from Figs. 4(a) and 6(b), we find that the thickness expansion rate during the bulk oxidation stage is proportional to the flux $(F0)$ or the density of $cgb(O3)$⁠, as shown in Fig. 15. Since the thickness expansion rate indicates the reaction rate between gas phase species and polymers, Fig. 15 suggests that O₃ may play a key role for the thickness expansion of the PS film bulk, which is related to oxygen uptake.

In our other publications,^16,17 we show further evidence that the thickness expansion is due to the cleavage and oxidation of aromatic rings caused by the diffusion-reaction of the plasma species in the bulk of the polymer films. The O₃ molecules appear to be the dominant reactant for such diffusion-reaction process due to its moderate reactivity larger than ground state O₂ but smaller than atomic O which could cause fast polymer etching from the surface.¹⁵ The cleavage and oxidation of aromatic ring by O₃ have also been observed previously,^54,55 and it is known that O₃ can behave as an electrophilic reagent when reacting with aromatic compounds.⁵⁴ Another effective reaction pathway might be through the atomic O produced by the dissociation of O₃

The O atoms are able to initiate polymer reactions by abstracting H from the tertiary carbon sites on the PS structure. The activated carbon site after H-abstraction may lead to further reactions including ring cleavage and oxidation.⁵⁶ 

Furthermore, we find that the surface O elemental composition of the SMD-treated PS films, measured by XPS, increases rapidly with the dose of O₃ species and then saturates at a upper limit as shown in Fig. 16. The curve of O elemental composition as a function of O₃ dose can be fitted with the exponential function

where y and $D1(O3)$ stand for the surface O elemental composition and dose of O₃, respectively. $C1,C2,andC3$ are constants. The exponential decay of the surface O accumulation can be understood as the depletion of the polymeric reactants as the treatment continues. At the beginning of the SMD treatment, there is a sufficient amount of polymeric reactants on surface, and thus the oxidation occurs with a higher reaction rate. As the treatment continues, the polymeric reactants within the top 2 nm of the film (probing depth of 20° XPS) are gradually depleted which explains the decrease of surface oxidation speed with the O₃ dose. When the treatment reaches the etching stage shown in Fig. 5, the surface O composition saturates because more oxidation would eventually lead to the formation of volatile products, e.g., CO₂ and H₂O, as discussed in Sec. III C.

Similarly, we can also correlate the average chemical composition of the ∼10 nm PS films measured by ATR-FTIR with the dose of O₃. As shown in Fig. 17(a), the percentage of aromatic rings left in the treated PS films compared to the pristine PS film, estimated by the peak intensity ratio at 702 cm⁻¹ between the treated and the pristine film, was plotted as a function of the O₃ dose. We find that less amount of aromatic ring is left in the ∼10 nm PS films when they are treated with a higher O₃ dose. In Fig. 17(b), we show the amount of carbonyl groups accumulated in the SMD-treated PS films, estimated by the areal peak intensity of the shaded area I in Fig. 12, as a function of the O₃ dose. It can be seen that the amount of carbonyls increases with the dose of O₃. Using Eq. (15), we can fit the experimental data in both Figs. 17(a) and 17(b) with the decreasing and increasing exponential functions, respectively. The fitted exponential decay constants (C₃) in Figs. 17(a) and 17(b) are comparable but smaller than that from Fig. 16. This indicates that (1) the destruction of aromatic rings and the formation of carbonyl groups may be related to the same reactive species generated by SMD and (2) the reactions inside the ∼10 nm PS films take longer time to complete than that in the top 2 nm of the PS film, which might be a result of the slow diffusion process of reactive species, presumably O₃, into the polymer bulk. This also supports our speculation that the aromatic rings are cleaved and partially converted into carbonyl groups by reactions involving O₃.

We find that the structure of polymers also affects how reactive species interacts with the material. Although here we only showed results obtained with PS, in another article,¹⁷ we study various polymers with different structures, including other styrene-based polymers and nonaromatic polymers. We find that only styrene-based polymers show a thickness increase stage when interacting with SMD generated reactive species, mainly O₃. This provides further support for the diffusion-reaction of O₃ in the polymer film and its selective interaction toward aromatic rings compared to the polymer main chain. Besides, we also find that styrene-based polymers show better etching resistivity—the aromatic rings might serve as a sink for the plasma generated reactive species, which reduces reactions on the main chain that cause chain scission and eventually etching.¹⁷ 

Since pristine PS does not contain N element, the formation of organic nitrate group confirmed by XPS in Sec. III C comes from the SMD treatment. However, we did not find evident linear or exponential decay relationships between the N elemental composition of the treated PS films and the dose of RNS, namely, N₂O₅, HNO₃, and N₂O. For example, in Fig. 18, we plotted the surface N composition of the treated PS films against the dose of N₂O₅, but no direct correlation between the two is seen. We also did not find correlation between the surface N composition and the dose of O₃. We suspect that the formation of surface organic nitrate might be a result of the synergistic effect between reactive oxygen species (ROS) and RNS. Due to their low reactivity, the long-lived RNS generated by the SMD, i.e., N₂O₅, HNO₃, and N₂O, might not be reactive enough to cause surface nitritation directly. Instead, an alternative reaction pathway is that the ROS first react with the polymer surface, which leads to the formation of surface radical sites. These radical sites on surface are highly reactive and may further react with the RNS from the gas phase. Therefore, the surface N composition may be determined by both the doses of ROS and RNS.

We characterized the reactive species generated in the effluent of the SMD reactor using IR and UV absorption. With N₂ and O₂ mixtures as working gas, the dominant reactive species in the effluent of SMD are identified as O₃, N₂O₅, N₂O, and HNO₃. We also quantified the density of these long-lived reactive species, and we found important changes in their densities as the ratio of N₂ and O₂ in the working gas is varied. To study the effect of these long-lived reactive species on polymers, we exposed ultrathin (∼10 nm) PS films to the effluent of the SMD with various working gas compositions. Under such a treatment, we observed and quantified the thickness change, the oxidation, and the nitritation of the polymer films as a function of the working gas composition. By correlating the thickness expansion rate of the PS film with the flux of the plasma generated reactive species, we found a linear dependence of expansion rate on the flux of O₃, which suggests that O₃ might be an essential reactant for causing the bulk oxidation of the PS film. When correlating the quantified chemical composition of the treated PS films with the dose of the long-lived reactive species, we found that (1) the O elemental composition of the PS surface (top ∼2 nm), (2) the aromatic ring content inside the ∼10 nm film, and (3) the carbonyl groups inside the film all show either increasing or decreasing exponential behavior versus the dose of O₃ delivered onto the PS films. This indicates that O₃ might participate in the chemical modification of the PS structure, such as cleaving the aromatic rings and oxidizing them into carbonyls, both on surface and in the polymer bulk. At last, we also correlated the surface organic nitrate of the treated PS films with the dose of gas species, but no evident relationship between the surface nitrate concentration and the dose of individual species was discovered. This suggests that the formation of surface organic nitrate might be a result of the synergistic interaction of ROS and RNS on polymers.

The authors gratefully acknowledge the financial support by the National Science Foundation (NSF) (No. PHY-1415353) and the U.S. Department of Energy (No. DE-SC0001939). They thank J. Santous Sousa for the help with SDO detection of the SMD reactor. They also thank A. J. Knoll, C. Li, A. Pranda, K. Lin, and M. Lai for helpful discussions and collaborations.