Effects of O 2 Plasma Treatments on the Photolithographic Patterning of PEDOT:PSS

: Poly (3,4-ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS) is known for its potential to replace indium–tin oxide in various devices. Herein, when fabricating ﬁnger-type PE-DOT:PSS electrodes using conventional photolithography, the cross-sectional proﬁles of the patterns are U-shaped instead of rectangular. The ﬁlms initially suffer from non-uniformity and fragility as well as defects owing to undesirable patterns. Adding a small amount of hydrolyzed silane crosslinker to PEDOT:PSS suspensions increases the mechanical durability of PEDOT:PSS patterns while lifting off the photoresist. To further improve their microfabrication, we observe the effects of two additional oxygen (O 2 ) plasma treatments on conventional photolithography processes for patterning PEDOT:PSS, expecting to observe how O 2 plasma increases the uniformity of the patterns and changes the thickness and U-shaped cross-sectional proﬁles of the patterns. Appropriately exposing the patterned photoresist to O 2 plasma before spin-coating PEDOT:PSS improves the wettability of its surface, including its sidewalls, and a similar treatment before lifting off the photoresist helps partially remove the spin-coated PEDOT:PSS that impedes the lift-off process. These two additional processes enable fabricating more uniform, defect-free PEDOT:PSS patterns. Both increasing the wettability of the photoresist patters before spin-coating PEDOT:PSS and reducing its conformal coverage are key to improving the photolithographic microfabrication of PEDOT:PSS. observations suggest that removing the PEDOT:PSS from the top of the photoresist may help improve the quality and uniformity of the ﬁnal PEDOT:PSS patterns. O 2 plasma increases the hydrophilicity of surfaces. Contact angle measurements revealed that the surface of the patterned photoresist was rather hydrophobic, which prompted us to add an O 2 plasma treatment before spin-coating PEDOT:PSS, which would also remove the PEDOT:PSS from the top of the patterned photoresist. Figure 5 shows how PEDOT:PSS suspensions spread on the surface of the photoresist with and without an O 2 plasma treatment; clearly, increasing the O 2 plasma operation time further increased the wettability of the patterned photoresist surface.

To improve transmittance characteristics, on one hand, finger-type ITO electrodes [12], which mitigate the absorption of light via ITO, have been fabricated by patterning flat ITO electrodes. On the other hand, materials with higher transmittance in the THz region such as graphene [13] are actively being attempted to replace ITO. For this purpose, poly (3,4-ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS) [14][15][16][17][18][19] is a promising candidate, and patterning PEDOT:PSS as a microfabrication strategy has been well studied. PEDOT:PSS exhibits high transmittance in the THz region, and because of its solution-process capability, it can be patterned using photolithography. Under these circumstances, replacing the finger-type ITO electrodes with patterned PEDOT:PSS ones would attain even higher transmittance in the THz region. Therefore, we attempted to fabricate finger-type electrodes using PEDOT:PSS; however, the films suffered from non-uniformity and fragility as well as defects owing to undesirable patterns. Figure 1 illustrates the designed mask patterns of a unit cell together with an enlarged diagram of a corner of the electrode area showing a part of the finger-type electrodes near the edge. The total electrode dimensions are 14 × 13 mm, and each finger-type electrode is 20 µm wide with a 20 µm gap between neighboring electrodes. In practice, the four designed unit cells were laid out in a soda-lime mask with dimensions of 5 × 5 × 0.09 in (12.7 × 12.7 × 0.229 cm). for 7 min. The photoresist films were exposed to UV light for 7 s using an EVG ® 6 aligner. The photoresist films were developed in a developer solution AD-10 KGaA, Darmstadt, Germany) for 3 min. Figure 1 illustrates the designed mask patterns of a unit cell together with an diagram of a corner of the electrode area showing a part of the finger-type electro the edge. The total electrode dimensions are 14 × 13 mm, and each finger-type ele 20 µ m wide with a 20 µ m gap between neighboring electrodes. In practice, the signed unit cells were laid out in a soda-lime mask with dimensions of 5 × 5 × 0.0 × 12.7 × 0.229 cm). A PEDOT:PSS dispersion (PH1000, Heraeus Epurio Clevios™, Leverkus many) was filtered through a 0.45 μm syringe filter and then mixed with 0.2 glycidyloxypropyl)trimethoxysilane (GOPS) [28,29], a hydrolyzed silane cro (Sigma-Aldrich, St. Louis, MO, USA). The suspension was sonicated for 30 min to enize the PEDOT:PSS dispersion, which was then spin-coated at 3000 rpm for 6 the patterned photoresist. The spin-coated substrates were baked at 130 °C for 15 finally sonicated in acetone to lift off the photoresist.
In our modified process, two O2 plasma treatments were independently ad before spin-coating PEDOT:PSS and one before lifting off the photoresist. The O power and the O2 flow rate were set to 100 W and 15 sccm, respectively, for both ion etching (RIE) and plasma etching (PE) modes, and each operation time wa The latter mode is more isotropic and gentler, whereas in the former mode, the O are generated more vertically and hit the substrates more vigorously because o tential between the electrodes in the chamber.

Measurements
A Dimension ICON scanning probe microscope system (Bruker, Billerica, M which offers precisions of <0.15 Å nm for XY noise and <0.35 Å Z sensor noise, w to measure the film thickness and probe its height profiles on the surface of the su We prepared three samples for each set of conditions. For each sample, we mea least three points in the upper, middle, and lower parts of the electrode area s Figure 1, and we measured each point three times. We also used a Quatek (Sheu Hong Kong) four-point probe test system, composed of a 5601TSR surface resistan and a QT-50 manual test console, to measure the surface resistance Rs(Ω) of film which the conductivity σ (S/cm) = 1/(Rs· d) was deduced using the film thickness

Transferring an Image to the Photoresist and the Fragility of PEDOT:PSS Patterns
To draw a finger-type (stripe-type) image on the photoresist, a mask pattern A PEDOT:PSS dispersion (PH1000, Heraeus Epurio Clevios™, Leverkusen, Germany) was filtered through a 0.45 µm syringe filter and then mixed with 0.2 wt % (3-glycidyloxypropyl)trimethoxysilane (GOPS) [28,29], a hydrolyzed silane crosslinker (Sigma-Aldrich, St. Louis, MO, USA). The suspension was sonicated for 30 min to homogenize the PEDOT:PSS dispersion, which was then spin-coated at 3000 rpm for 60 s onto the patterned photoresist. The spin-coated substrates were baked at 130 • C for 15 min and finally sonicated in acetone to lift off the photoresist.
In our modified process, two O 2 plasma treatments were independently added, one before spin-coating PEDOT:PSS and one before lifting off the photoresist. The O 2 plasma power and the O 2 flow rate were set to 100 W and 15 sccm, respectively, for both reactive-ion etching (RIE) and plasma etching (PE) modes, and each operation time was varied. The latter mode is more isotropic and gentler, whereas in the former mode, the O 2 species are generated more vertically and hit the substrates more vigorously because of the potential between the electrodes in the chamber.

Measurements
A Dimension ICON scanning probe microscope system (Bruker, Billerica, MA, USA), which offers precisions of <0.15 Å nm for XY noise and <0.35 Å Z sensor noise, was used to measure the film thickness and probe its height profiles on the surface of the substrates. We prepared three samples for each set of conditions. For each sample, we measured at least three points in the upper, middle, and lower parts of the electrode area shown in Figure 1, and we measured each point three times. We also used a Quatek (Sheung Wan, Hong Kong) four-point probe test system, composed of a 5601TSR surface resistance tester and a QT-50 manual test console, to measure the surface resistance R s (Ω) of films, from which the conductivity σ (S/cm) = 1/(R s ·d) was deduced using the film thickness d (cm).

Transferring an Image to the Photoresist and the Fragility of PEDOT:PSS Patterns
To draw a finger-type (stripe-type) image on the photoresist, a mask pattern (design drawing or pattern) is first transferred. The stripe-type-patterned photoresist works as a template or mold to create PEDOT:PSS patterns [16,30]. To this end, some process conditions are optimized using various quality engineering methods [31], i.e., by controlling the spin-coating rotational speed, baking temperature, UV-light-exposure time, and develop- ing time, as described in the experimental section. Using these optimized conditions, we attain a high-quality patterned photoresist, and the patterns are transferred from the mask patterns with a gap of 20 µm between neighboring patterns. By using the image (pattern) of the photoresist, PEDOT:PSS is then patterned via microprocessing.
Initially, the PEDOT:PSS patterns were very mechanically fragile, especially in the lift-off process, during which the substrates were shaken in an acetone bath. To remove the photoresist patterns from the substrates, they were inevitably shaken, which presumably applied a frictional force to the PEDOT:PSS patterns on the substrates. In fact, the PEDOT:PSS patterns were partially peeled off and bent out of the substrates, as shown in Figure 2. Simply immersing substrates in acetone, however, does not remove photoresist patterns from substrates; therefore, to avoid damaging the patterns, the films had to be strengthened. This fragility can be attributed to the weak anchoring of the PEDOT:PSS films to the surface of the substrates, although this technique is well known [28,29,32]. However, exposing the photoresist patterns to O 2 plasma did not mitigate this problem, so it had to be resolved using surfactants.
template or mold to create PEDOT:PSS patterns [16,30]. To this end, some proces tions are optimized using various quality engineering methods [31], i.e., by con the spin-coating rotational speed, baking temperature, UV-light-exposure time, veloping time, as described in the experimental section. Using these optimized con we attain a high-quality patterned photoresist, and the patterns are transferred f mask patterns with a gap of 20 µ m between neighboring patterns. By using th (pattern) of the photoresist, PEDOT:PSS is then patterned via microprocessing.
Initially, the PEDOT:PSS patterns were very mechanically fragile, especial lift-off process, during which the substrates were shaken in an acetone bath. To the photoresist patterns from the substrates, they were inevitably shaken, which ably applied a frictional force to the PEDOT:PSS patterns on the substrates. In PEDOT:PSS patterns were partially peeled off and bent out of the substrates, as s Figure 2. Simply immersing substrates in acetone, however, does not remove ph patterns from substrates; therefore, to avoid damaging the patterns, the films h strengthened. This fragility can be attributed to the weak anchoring of the PED films to the surface of the substrates, although this technique is well known [2 However, exposing the photoresist patterns to O2 plasma did not mitigate this p so it had to be resolved using surfactants. Specifically, GOPS, a hydrolyzed silane crosslinker, was used to improve the ing, as its epoxy group can react open under acidic conditions, yielding a hydroxy When GOPS is added to PEDOT:PSS suspensions, the hydroxyl groups strongly with the sulfonic acid groups of PEDOT:PSS via hydrogen bonds. The trimetho groups of GOPS are hydrolyzed in acidic suspensions and become Si-OH group form stable Si-O-Si bonds on the glass surface [32]. To reveal the effects of GOP PEDOT:PSS suspensions, its concentration was varied, and the durability of the r PEDOT:PSS patterns was evaluated. As the GOPS concentration increased, fewe ended to appear in the PEDOT:PSS patterns. However, 3.0 wt % GOPS caused PED to agglomerate, which was previously reported to hinder spin-coating [29]. Fur confirmed that adding GOPS decreased the conductivity of the PEDOT:PSS film gesting a trade-off relationship between improving the durability and maintai conductivity of the PEDOT:PSS films. Figure 3 shows the relationship between ductivity of PEDOT:PSS films and the GOPS concentration, which shows almost t trend as that observed in previous work [28], although the initial conductivity GOPS was slightly lower here. Based on these findings, we used 0.2 wt % GOPS i quent experiments because this concentration yielded PEDOT:PSS films that we ciently durable during the lift-off process. Specifically, GOPS, a hydrolyzed silane crosslinker, was used to improve the anchoring, as its epoxy group can react open under acidic conditions, yielding a hydroxyl group. When GOPS is added to PEDOT:PSS suspensions, the hydroxyl groups strongly interact with the sulfonic acid groups of PEDOT:PSS via hydrogen bonds. The trimethoxysilane groups of GOPS are hydrolyzed in acidic suspensions and become Si-OH groups, which form stable Si-O-Si bonds on the glass surface [32]. To reveal the effects of GOPS in the PEDOT:PSS suspensions, its concentration was varied, and the durability of the resulting PEDOT:PSS patterns was evaluated. As the GOPS concentration increased, fewer defects ended to appear in the PEDOT:PSS patterns. However, 3.0 wt % GOPS caused PEDOT:PSS to agglomerate, which was previously reported to hinder spin-coating [29]. Further, we confirmed that adding GOPS decreased the conductivity of the PEDOT:PSS films, suggesting a trade-off relationship between improving the durability and maintaining the conductivity of the PEDOT:PSS films. Figure 3 shows the relationship between the conductivity of PEDOT:PSS films and the GOPS concentration, which shows almost the same trend as that observed in previous work [28], although the initial conductivity without GOPS was slightly lower here. Based on these findings, we used 0.2 wt % GOPS in subsequent experiments because this concentration yielded PEDOT:PSS films that were sufficiently durable during the lift-off process. Figure 4 shows the two-dimensional height profiles of the PEDOT:PSS patterns obtained using the traditional process. Each sample was measured at several points in the sample, and an average was taken at each position. Considering the photoresist patterns, we expect to observe a rectangular cross-section with a line width of 20 µm. Surprisingly, however, the actual profiles are U-shaped [33,34], with large variations in thickness at the center of the lines. The U-shaped profiles could be attributed to the evaporation of solvent and its corollary mass transfer [35]. If the surface of the photoresist has good wettability, the suspension spreads not only in the planar direction but also upward along the sidewalls of the photoresist. In addition, as the solvent evaporates, mass transfer occurs at the edge of the suspension on the sidewall of the photoresist, thereby leading to the U-shaped profiles at the edge. According to this mechanism, a higher concentration with higher viscosity would reduce the peak height of the U-shaped profiles, whereas better wettability, including on the sidewalls of the photoresist, would develop U-shaped profiles.  Figure 4 shows the two-dimensional height profiles of the PEDOT:PSS patterns obtained using the traditional process. Each sample was measured at several points in the sample, and an average was taken at each position. Considering the photoresist patterns, we expect to observe a rectangular cross-section with a line width of 20 μm. Surprisingly, however, the actual profiles are U-shaped [33,34], with large variations in thickness at the center of the lines. The U-shaped profiles could be attributed to the evaporation of solvent and its corollary mass transfer [35]. If the surface of the photoresist has good wettability, the suspension spreads not only in the planar direction but also upward along the sidewalls of the photoresist. In addition, as the solvent evaporates, mass transfer occurs at the edge of the suspension on the sidewall of the photoresist, thereby leading to the U-shaped profiles at the edge. According to this mechanism, a higher concentration with higher viscosity would reduce the peak height of the U-shaped profiles, whereas better wettability, including on the sidewalls of the photoresist, would develop U-shaped profiles.    Figure 4 shows the two-dimensional height profiles of the PEDOT:PSS patterns obtained using the traditional process. Each sample was measured at several points in the sample, and an average was taken at each position. Considering the photoresist patterns, we expect to observe a rectangular cross-section with a line width of 20 μm. Surprisingly, however, the actual profiles are U-shaped [33,34], with large variations in thickness at the center of the lines. The U-shaped profiles could be attributed to the evaporation of solvent and its corollary mass transfer [35]. If the surface of the photoresist has good wettability, the suspension spreads not only in the planar direction but also upward along the sidewalls of the photoresist. In addition, as the solvent evaporates, mass transfer occurs at the edge of the suspension on the sidewall of the photoresist, thereby leading to the U-shaped profiles at the edge. According to this mechanism, a higher concentration with higher viscosity would reduce the peak height of the U-shaped profiles, whereas better wettability, including on the sidewalls of the photoresist, would develop U-shaped profiles.   From another perspective, the bottom of the lines is approximately 30 µm, which is 1.5 times larger than the designed value. In contrast, the peak-to-peak width is almost 20 µm, suggesting that each peak must be determined by the sidewall of the patterned photoresist. Presumably, when PEDOT:PSS is spin-coated onto the patterned photoresist, PEDOT:PSS not only fills the spaces between the lines of the patterned photoresist but also covers the top of the photoresist. The PEDOT:PSS initially present on top of the photoresist can be connected to the PEDOT:PSS between the lines via conformal coverage. Therefore, even after the photoresist patterns are removed during the lift-off process, partially connected PEDOT:PSS would remain at the bottom of the patterns [30]. These

Adding an O 2 Plasma Treatment before Spin-Coating
O 2 plasma increases the hydrophilicity of surfaces. Contact angle measurements revealed that the surface of the patterned photoresist was rather hydrophobic, which prompted us to add an O 2 plasma treatment before spin-coating PEDOT:PSS, which would also remove the PEDOT:PSS from the top of the patterned photoresist. Figure 5 shows how PEDOT:PSS suspensions spread on the surface of the photoresist with and without an O 2 plasma treatment; clearly, increasing the O 2 plasma operation time further increased the wettability of the patterned photoresist surface.
covers the top of the photoresist. The PEDOT:PSS initially present on top of the p sist can be connected to the PEDOT:PSS between the lines via conformal coverage fore, even after the photoresist patterns are removed during the lift-off process, p connected PEDOT:PSS would remain at the bottom of the patterns [30]. These o tions suggest that removing the PEDOT:PSS from the top of the photoresist may h prove the quality and uniformity of the final PEDOT:PSS patterns.

Adding an O2 Plasma Treatment before Spin-Coating
O2 plasma increases the hydrophilicity of surfaces. Contact angle measurem vealed that the surface of the patterned photoresist was rather hydrophobic prompted us to add an O2 plasma treatment before spin-coating PEDOT:PSS would also remove the PEDOT:PSS from the top of the patterned photoresist. F shows how PEDOT:PSS suspensions spread on the surface of the photoresist w without an O2 plasma treatment; clearly, increasing the O2 plasma operation time increased the wettability of the patterned photoresist surface. We found that adding the O2 plasma treatment at this point during processing improves the uniformity of the thickness. Figure 6 shows how the height profile cross-sectional PEDOT:PSS patterns change when the patterned photoresist is treated in RIE mode before spin-coating it with PEDOT:PSS. As the operating tim O2 plasma treatment increases, the center of the lines becomes more uniform. I ingly, however, the peaks that appear at the edge of each line become more pron with the increasing operating time, which suggests that the plasma-treated surfac photoresist-patterned substrates indeed provides better wettability for PEDOT:P hence better uniformity at the center of the lines. Specifically, the thickness varia the center of the lines are improved, decreasing from ±33.9 to ±5.7 nm. Furtherm plasma increases the wettability of the sidewalls of the photoresist patterns, and th the peaks become more pronounced with the increasing operating time owing to called coffee ring effect [35,36]. This finding can be verified by using a gentler O2 treatment in PE mode. We found that adding the O 2 plasma treatment at this point during processing indeed improves the uniformity of the thickness. Figure 6 shows how the height profiles of the cross-sectional PEDOT:PSS patterns change when the patterned photoresist is plasmatreated in RIE mode before spin-coating it with PEDOT:PSS. As the operating time of the O 2 plasma treatment increases, the center of the lines becomes more uniform. Interestingly, however, the peaks that appear at the edge of each line become more pronounced with the increasing operating time, which suggests that the plasma-treated surface of the photoresistpatterned substrates indeed provides better wettability for PEDOT:PSS and hence better uniformity at the center of the lines. Specifically, the thickness variations at the center of the lines are improved, decreasing from ±33.9 to ±5.7 nm. Furthermore, O 2 plasma increases the wettability of the sidewalls of the photoresist patterns, and therefore, the peaks become more pronounced with the increasing operating time owing to the so-called coffee ring effect [35,36]. This finding can be verified by using a gentler O 2 plasma treatment in PE mode. Figure 7 shows how the height profiles of the cross-sectional PEDOT:PSS patterns change when the photoresist is plasma-treated in PE mode before spin-coating PEDOT:PSS. In contrast to the results of the RIE O 2 plasma treatment, the thickness of PEDOT:PSS changes more mildly and gently, with little impact on the sidewalls of the photoresist pattern. Figure 8 compares the changes in thickness and its variability at the center of the PEDOT:PSS patterns when the patterned photoresist is plasma-treated in RIE or PE mode before spin-coating. In RIE mode, the O 2 plasma treatment is indeed vigorous and more rapidly reduces the film thickness. Presumably, this mode makes the surface of the photoresist more hydrophilic than PE mode, which is consistent with the results shown in Figure 5. Although in principle, the RIE plasma mode bombards the substrate surface more vertically, it also more severely affects the sidewalls. Therefore, the hydrophilicity increases on the entire surface of the patterned photoresist, including the sidewalls, so the U-shaped profiles become more pronounced. Meanwhile, the films become thinner, with less variability at the center of each patterned line and a higher peak at the edge.  Figure 7 shows how the height profiles of the cross-sectional PEDOT:PSS patterns change when the photoresist is plasma-treated in PE mode before spin-coating PE-DOT:PSS. In contrast to the results of the RIE O2 plasma treatment, the thickness of PE-DOT:PSS changes more mildly and gently, with little impact on the sidewalls of the photoresist pattern. Figure 8 compares the changes in thickness and its variability at the center of the PEDOT:PSS patterns when the patterned photoresist is plasma-treated in RIE or PE mode before spin-coating. In RIE mode, the O2 plasma treatment is indeed vigorous and more rapidly reduces the film thickness. Presumably, this mode makes the surface of the photoresist more hydrophilic than PE mode, which is consistent with the results shown in Figure 5. Although in principle, the RIE plasma mode bombards the substrate surface more vertically, it also more severely affects the sidewalls.
Therefore, the hydrophilicity increases on the entire surface of the patterned photoresist, including the sidewalls, so the U-shaped profiles become more pronounced. Meanwhile, the films become thinner, with less variability at the center of each patterned line and a higher peak at the edge.  In subsequent experiments, to avoid developing U-shaped cross-sectional profiles in the PEDOT:PSS patterns, we performed the O2 plasma treatment for 3 min in PE mode before spin-coating PEDOT:PSS for the following step, i.e., lifting off the photoresist.  In subsequent experiments, to avoid developing U-shaped cross-sectional pro the PEDOT:PSS patterns, we performed the O2 plasma treatment for 3 min in PE before spin-coating PEDOT:PSS for the following step, i.e., lifting off the photoresi  In subsequent experiments, to avoid developing U-shaped cross-sectional profiles in the PEDOT:PSS patterns, we performed the O 2 plasma treatment for 3 min in PE mode before spin-coating PEDOT:PSS for the following step, i.e., lifting off the photoresist.

Adding an O 2 Plasma Treatment before Lifting Off the Photoresist
The next step introduces another difficulty that disturbs the uniformity of the patterns. Undesirable patterns, a type of defect, are often observed when lifting off the photoresist. Figure 9 shows an image of typical undesirable patterns remaining on the surface of the patterned PEDOT:PSS, which we assume are related to the PEDOT:PSS covering the top of the photoresist via conformal coverage [30]. Presumably, when the photoresist patterns are lifted off, the PEDOT:PSS anchored to the surface of the photoresist mechanically causes this problem. Further, brush-like wrinkles often appear around the boundaries after spin-coating PEDOT:PSS onto the patterned photoresist, as shown in Figure 10b, which prompts us to consider the effects of conformal coverage. During the spin-coating process, PEDOT:PSS spreads over the entire patterned photoresist surface, meaning that PEDOT:PSS can spread not only in the gaps between the strip patterns but also on top of the patterned lines. Therefore, removing such PEDOT:PSS covering the photoresist before lifting it off would help reduce these defects.
Coatings 2021, 11, x FOR PEER REVIEW

Adding an O2 Plasma Treatment before Lifting off the Photoresist
The next step introduces another difficulty that disturbs the uniformity of terns. Undesirable patterns, a type of defect, are often observed when lifting off toresist. Figure 9 shows an image of typical undesirable patterns remaining on th of the patterned PEDOT:PSS, which we assume are related to the PEDOT:PSS the top of the photoresist via conformal coverage [30]. Presumably, when the ph patterns are lifted off, the PEDOT:PSS anchored to the surface of the photoresist ically causes this problem. Further, brush-like wrinkles often appear around the ries after spin-coating PEDOT:PSS onto the patterned photoresist, as shown in Fig  which prompts us to consider the effects of conformal coverage. During the spin process, PEDOT:PSS spreads over the entire patterned photoresist surface, mean PEDOT:PSS can spread not only in the gaps between the strip patterns but also o the patterned lines. Therefore, removing such PEDOT:PSS covering the photoresi lifting it off would help reduce these defects.    Thus, to remove the PEDOT:PSS from the top of the photoresist caused by conformal coverage, we introduce another O2 plasma process before lifting off the photoresist patterns. As expected, applying this additional O2 plasma treatment at this stage reduces the undesirable patterns and defects in the resulting PEDOT:PSS when the photoresist is finally removed with acetone, as shown in Figure 11. Obviously, however, a longer plasma treatment at this stage also reduces the thickness of PEDOT:PSS itself or even damages the patterns. As an extreme case in which we plasma-treated the PEDOT:PSS pattern for 120 s, its thickness decreased to one-fifth of its designed value, and its linewidths became two-thirds narrower. Further, we observed jagged line patterns, which are side effects of O2 plasma exposure at this stage. Therefore, the operating time of O2 plasma has a tradeoff relationship between mitigating these undesirable patterns and maintaining the quality of the PEDOT:PSS patterns. Thus, to remove the PEDOT:PSS from the top of the photoresist caused by conformal coverage, we introduce another O 2 plasma process before lifting off the photoresist patterns. As expected, applying this additional O 2 plasma treatment at this stage reduces the undesirable patterns and defects in the resulting PEDOT:PSS when the photoresist is finally removed with acetone, as shown in Figure 11. Obviously, however, a longer plasma treatment at this stage also reduces the thickness of PEDOT:PSS itself or even damages the patterns. As an extreme case in which we plasma-treated the PEDOT:PSS pattern for 120 s, its thickness decreased to one-fifth of its designed value, and its linewidths became two-thirds narrower. Further, we observed jagged line patterns, which are side effects of O 2 plasma exposure at this stage. Therefore, the operating time of O 2 plasma has a trade-off relationship between mitigating these undesirable patterns and maintaining the quality of the PEDOT:PSS patterns. Nevertheless, adding an O2 plasma treatment at this stage enables the removal of the unfavorable PEDOT:PSS covering the photoresist, suggesting that it is crucial to remove such PEDOT:PSS from the top of the photoresist or avoid conformal coverage. In order to remove this PEDOT:PSS without affecting the film thickness or damaging the film, O2 plasma etching with masking would be more reliable.

Combination of the Added O2 Plasma Treatments
Naturally, we are also interested in how combining the two added O2 plasma processes eventually improves the quality of the PEDOT:PSS patterns. We thus optimized the plasma treatment operation time in PE mode to be 3 min before spin-coating PEDOT:PSS and that before lifting off the photoresist in RIE mode to be 1 s. Our investigation reveals that even 1 s of O2 plasma in the RIE mode assists with removing undesirable PEDOT:PSS, thereby promoting the lift-off process while avoiding defects. Figure 12 shows microscopic images and height profiles of the PEDOT:PSS patterns when both O2 plasma treatments are added to the process. Applying both treatments enables fabricating more uniform, defect-free PEDOT:PSS finger-type patterns with a deviation in the thickness of ±10.6 nm at the center of the cross-section. The bottom of the lines is still beyond the designed value, thus requiring further optimization; however, avoiding conformal coverage would be crucial to improving microfabrication using photolithography. Nevertheless, adding an O 2 plasma treatment at this stage enables the removal of the unfavorable PEDOT:PSS covering the photoresist, suggesting that it is crucial to remove such PEDOT:PSS from the top of the photoresist or avoid conformal coverage. In order to remove this PEDOT:PSS without affecting the film thickness or damaging the film, O 2 plasma etching with masking would be more reliable.

Combination of the Added O 2 Plasma Treatments
Naturally, we are also interested in how combining the two added O 2 plasma processes eventually improves the quality of the PEDOT:PSS patterns. We thus optimized the plasma treatment operation time in PE mode to be 3 min before spin-coating PEDOT:PSS and that before lifting off the photoresist in RIE mode to be 1 s. Our investigation reveals that even 1 s of O 2 plasma in the RIE mode assists with removing undesirable PEDOT:PSS, thereby promoting the lift-off process while avoiding defects. Figure 12 shows microscopic images and height profiles of the PEDOT:PSS patterns when both O 2 plasma treatments are added to the process. Applying both treatments enables fabricating more uniform, defect-free PEDOT:PSS finger-type patterns with a deviation in the thickness of ±10.6 nm at the center of the cross-section. The bottom of the lines is still beyond the designed value, thus requiring further optimization; however, avoiding conformal coverage would be crucial to improving microfabrication using photolithography. thereby promoting the lift-off process while avoiding defects. Figure 12 shows microscopic images and height profiles of the PEDOT:PSS patterns when both O2 plasma treatments are added to the process. Applying both treatments enables fabricating more uniform, defect-free PEDOT:PSS finger-type patterns with a deviation in the thickness of ±10.6 nm at the center of the cross-section. The bottom of the lines is still beyond the designed value, thus requiring further optimization; however, avoiding conformal coverage would be crucial to improving microfabrication using photolithography.

Conclusions
We fabricated finger-type PEDOT:PSS electrodes using conventional photolithographic patterning techniques, during which the cross-sectional U-shaped profiles of the PEDOT:PSS patterns were found instead of rectangular cross-sections. The fragility of the PEDOT:PSS patterns was improved by adding 0.2 wt % GOPS to the PEDOT:PSS suspensions. Using the strengthened PEDOT:PSS patterns, we investigated the effects of adding O2 plasma treatments to the traditional photolithography process for patterning PE-

Conclusions
We fabricated finger-type PEDOT:PSS electrodes using conventional photolithographic patterning techniques, during which the cross-sectional U-shaped profiles of the PE-DOT:PSS patterns were found instead of rectangular cross-sections. The fragility of the PEDOT:PSS patterns was improved by adding 0.2 wt % GOPS to the PEDOT:PSS suspensions. Using the strengthened PEDOT:PSS patterns, we investigated the effects of adding O 2 plasma treatments to the traditional photolithography process for patterning PE-DOT:PSS, expecting to observe how this treatment improved the uniformity of the patterns and how it changed the thickness and U-shaped profiles of the patterns. One O 2 plasma treatment was introduced before spin-coating PEDOT:PSS onto the patterned photoresist layer, and the other was employed before the lift-off process for removing the photoresist layer. The former improved the wettability of the patterned photoresist surface, including that of its sidewalls, and the latter helped partially remove the spin-coated PEDOT:PSS that impeded the lift-off process. In the former O 2 plasma treatment, the uniformity of the thickness at the center of the lines became more uniform, with an improved deviation from several tens of nanometers to several nanometers; however, the U-shaped patterns became more pronounced when the sidewalls of the patterned photoresist were exposed to the plasma in RIE mode, which could be avoided using more gentle plasma in PE mode. The latter O 2 plasma treatment eventually facilitated the lift-off process while preventing defects, uniformly reducing the entire thickness of the PEDOT:PSS patterns while removing the PEDOT:PSS covering the photoresist. Our findings suggest that the most important factor is how to remove PEDOT:PSS from the top of the photoresist or avoid its conformal coverage. Applying these two additional processes enabled fabricating more uniform, defect-free PEDOT:PSS patterns. Realizing a real THz device using the investigated techniques requires considerable effort to optimize the entire procedure, which would involve many factors, such as the cross-sectional shape of the patterned lines, defects, conductivity, and anchoring strength.