Real time characterization of polymer surface modifications by an atmospheric-pressure plasma jet: Electrically coupled versus remote mode

Atmospheric-pressure plasma sources have been increasingly investigated for a wide range of applications, from biological^1–3 to industrial surface treatments.^4,5 Here, we distinguish two modes of APPJ operation depending on if the plume is coupled to the surface and charged species dominate versus remote treatments where reactive neutrals dominate. Previous work on APPJ surface interaction has not systematically distinguished between the two interaction modes described here, despite substantial difference in the physical mechanisms, and have primarily been investigated post-treatment through methods such as biological deactivation,⁶ water contact angle,⁷ atomic force microscopy,⁸ Fourier transform infrared spectroscopy,⁹ and profilometry.¹⁰ In this work, we use a combination of in situ techniques including real time ellipsometry and high speed photography with post characterization techniques. We find that APP source/sample geometry, feed gas, environmental chemistry, and APP source temperature play important roles in determining the consequences of the APP-surface interactions.

The plasma source used in this work has been described previously¹ and is based on the design by Teschke et al. called an atmospheric-pressure plasma jet (APPJ).¹¹ A diagram of the experimental setup is shown in Fig. 1(a). The plasma source consists of two 20 mm long cylindrical copper electrodes separated by a 12 mm thick Teflon block around a 6.35 mm outer diameter, 3.7 mm inner diameter alumina tube through which 2 lpm Ar gas was flowed with or without 1% O₂ or N₂ admixtures. The plasma was generated using a high voltage power generator driven at 25 kHz. Applied voltage and voltage across measurement capacitor was monitored by using a Tektronix P6015A high voltage probe connected to a Hewlett-Packard 400 MHz oscilloscope. Power dissipated by the plasma was calculated using the Lissajous method using a 1 nF measurement capacitor (C_m).^12,13 High speed photography of the APPJ was taken using a Phantom v7.3 camera. An ellipsometer utilizing a 1.5 mW HeNe laser at 632.8 nm was used to non-invasively investigate thickness and index changes extracted through optical modeling of the ellipsometric data. Surface chemistry analysis by XPS was performed by a Vacuum Generators ESCALAB MK II surface analysis system post treatment. Surface morphology was imaged post treatment using a Hitachi SU-70 Analytical scanning electron microscopy (SEM) and through differential interference contrast microscopy. The surface studied was a 300 nm thick PMMA-based 193 nm photoresist polymer (PR193) film described previously.¹⁷ 

Figure 1(a) displays the two APPJ modes of operation. For the remote treatment mode, the plasma is generated with a source to nozzle distance (s_n) of 15 cm and a nozzle to sample distance (s_s) of 1 cm to ensure no direct electrical interaction. Plasma coupling mode is evident by filaments visually terminating at the film surface and occurs at combined distances closer than 15 cm, depending on gas chemistry and whether the plasma is confined within the tube (small s_n) or exposed to the ambient (small s_s). The setup also allows for the addition of a removable, electrically insulating shield between the plasma and sample (not shown).

High speed photography of the APPJ discharge with no surface interaction is shown in Fig. 1(b). The APPJ was flowed into either open air (top row) or a controlled environment matching the feed gas chemistry (middle row). Applied voltage for all gas chemistries was held constant at 14 kV peak to peak, and power dissipation was measured to be 6.2 W for Ar, 9.2 W for 1% N₂ in Ar, and 11.6 W for 1% O₂ in Ar. The pure Ar plume is concentrated in one central channel and increases in length as the environment changes from air to Ar. This central channel nature has been attributed previously to the metastable state Ar (4³P₂⁰) with 11.6 eV, which does not have enough energy to ionize Ar (E_i = 15.8 eV), so the plasma is only formed by streamer process.¹⁴ N₂ addition creates a comparatively diffuse plume, with discharges branching away from the central filament and extending outward when the environment matches the feed gas. This is likely due to the introduction of Penning ionization that occurs with N₂ metastable species.¹⁵ O₂ addition to the feed gas, regardless of environment, greatly reduces the plume length and intensity. This observation suggests reduced plasma density for the addition of O₂, likely resulting from the loss of electrons by creation of negative oxygen ions¹⁶ and is predicted by modeling O₂ addition to Ar plasma.¹⁷ The confined geometry of the APPJ controls the gaseous environment around the plasma similar to the controlled environment shown here.

Figure 1(b) (bottom row) shows high speed photography with 30 μs exposure time of the plasma coupling to the surface. Images show filaments that extend outward from the source tube, sometimes several in one image. At close distances, the etch region is annular following the path of these filaments, as shown in Fig. 2(a). However, as the source is moved away from the sample, the diameter of the etched area shrinks from ∼7 mm to 3.7 mm, matching the tube inner diameter. This agrees with previous work that shows that plasma bullet emission occurs in a ring shape.¹⁸ Typical etch rates are locally >100 nm/min where the filaments terminate at the surface for coupled interactions compared to the remote interaction, where etch rates are below 3 nm/min. The thickness removal is lower than similar treatments by Fricke et al.,¹⁰ who see etch rates above 1000 nm/min at distances of 5 mm from the sample, likely due to differences in the APPJ design.

The electrically coupled surface interaction mode visibly damages the photoresist in the form of spots due to intense, localized energy deposition, leading to fast film removal. Figure 2(b) shows a SEM image of a damage spot which has high roughness throughout the affected area. High speed photography was used during a 0.5 second treatment and the position of each filament impact was extracted. The position of these strikes is shown in Fig. 2(c) and compared to the position of the significant spot damage identified by microscopy. Areas with higher densities of strikes to the surface positively correlate with visible spots. This indicates that the plasma will more likely electrically couple to previously damaged spots. The small number of spots compared to the number of filament interactions with the surface suggests that either not all filaments cause damage or that it takes many strikes before a large spot is formed. The surface area where more than 50 nm of photoresist was removed during the 30 s treatments was determined for various source to sample distances (s_n + s_s) and is shown in Fig. 2(d). The amount of photoresist removed decreases as distance increases and decreases more quickly when the plasma is exposed to the ambient environment. This agrees with previous work demonstrating the importance of the ambient chemistry around the APPJ.¹ This result also agrees with the images of Fig. 1(b) which show reduced plume length in open air versus controlled environment matching feed gas, similar to how the plasma is confined in the APPJ tube.

In the following, we discuss remote plasma processing where direct electrical interaction with the sample is avoided. For pure Ar flowing into air, the etch rate of the PR193 film is initially low, called etch lag, and subsequently increases. The duration of this etch lag increases with the amount of time the plasma has been off, as shown in Fig. 3(a). We found that the increase of the polymer etching rate correlated with heating of the dielectric tube. The temperature of the APPJ source was measured by a FLIR i7 thermal imaging camera and plotted versus etch rate over several experiments to make an Arrhenius plot and estimate an apparent activation energy of ∼0.9 eV. Cooling the source externally with compressed air from approximately 80 °C to 50 °C showed a reduction of 0.05 nm/min in etch rate that then increased again once cooling was stopped. This slight reduction further supports the dependence of the plasma surface interaction on source temperature.

The sample was also shielded from the APPJ (using the electrically insulating shield mentioned above), preventing etching while leaving the plasma on to prevent the dielectric tube from cooling. As shown in Fig. 3(b), etch lag does not occur and the etch rate recovers immediately after the shield is removed. This suggests that etching is not related to sample heating during the treatment since the sample would have ample time to cool while shielded. The sample temperature was estimated to increase by less than 30 °C during a 30 min argon treatment by measuring the refractive index changes of bulk Si.¹⁹ 

Feed gas chemistry at constant applied voltage plays a large role in determining plasma properties^15,20 as well as etching and surface modifications. Figures 3(c) and 3(d) show a comparison of a remote treatment with alternating pure Ar and 1% N₂ or O₂ admixture to Ar under same electrical characteristics for high speed photography work. Etch rate for N₂/Ar increases to ∼2.0 nm/min, which is several times greater than for pure Ar. N₂/Ar plasma tends to extend further down the APPJ tube at similar conditions than other gas chemistries and shows increased power dissipation from pure Ar, which likely explains the increased etch rate. O₂/Ar has reduced etch rates of ∼0.05 nm/min compared to pure Ar, which saturates at ∼0.5 nm/min, despite having increased dissipated power. This decrease mirrors the drop in emission intensity seen in the high speed photography.

While oxygen addition to the feed gas reduces the etch rate, increased oxygen uptake at the surface is seen in XPS surface analysis (Fig. 4). The oxygenation of the surface for the pure Ar and N₂/Ar treatments likely comes from plasma interactions with ambient air. All three gas chemistries also show the formation of NO_x species at the surface. The O₂ treatment shows increased species with greater oxygen content such as NO₃, whereas the N₂ treatment shows much less NO₃ and increased species that have greater N:O ratios.

In conclusion, the consequences of APP surface interactions are highly dependent on the interaction modes described here due to the distinct physics governing charged versus neutral species. Understanding what types of species dominate the interaction with the surface is critical to the operation of APP sources and control of surface functionalization. Average polymer etch rates are significantly higher when the plasma couples directly to the surface and produces localized damage as compared to the remote plasma mode interaction, which induces slow uniform etching and subtle surface chemical modifications. In the remote case, etch rate and surface modifications are highly dependent upon feed gas chemistry, APPJ source temperature, and plasma interaction with the environment.

The authors gratefully acknowledge financial support by the US Department of Energy (DE-SC0001939 and DE-SC0005105) and National Science Foundation (PHY-1004256). We also thank D. Metzler, C. Li, and E. Merino for their helpful discussion and collaboration on this project.