A Systematic Study of Plasma Activation of Silicon ... - ACS Publications

Research Article www.acsami.org

A Systematic Study of Plasma Activation of Silicon Surfaces for Self Assembly Savas Kaya,*,† Parthiban Rajan,† Harshita Dasari,† David C. Ingram,‡ Wojciech Jadwisienczak,† and Faiz Rahman† †

School of Electrical Engineering and Computer Science and ‡Department of Physics and Astronomy, Ohio University, Athens, Ohio 45701, United States S Supporting Information *

ABSTRACT: We study the plasma activation systematically in an attempt to simplify and optimize the formation of hydrophilic silicon (Si) surface critical for self-assembly of nanostructures that typically uses piranha solution, a high molarity cocktail of sulfuric acid and hydrogen peroxide at elevated temperatures. In the proposed safer and simpler approach, O2 plasma is used under optimized process conditions in a capacitively coupled parallel-plate chamber to induce strong hydrophilic behavior on silicon surfaces associated with the formation of suboxide groups. Surface activation is validated and studied via contact angle measurements as well as XPS spectra and consequently optimized using a novel atomic force spectroscopy approach, which can streamline characterization. It is found that plasma power around 100 W and exposure duration of ∼65 s are the most effective parameters to enhance surface activation for the reactive ion etcher system used. Other optimum plasma process conditions for pressure and flow-rate are also reported along with temporal development of activation, which peaks within 1 h and wears off in 24 h scale in air. The applicability of the plasma approach to nanoassembly process was demonstrated using simple drop coating and spinning of polystyrene (d < 500 nm, 2.5−4.5% w/v) and inkjet printing on polydimethylsiloxane. KEYWORDS: plasma activation, hydrophilic silicon surface, nanosphere lithography, force spectroscopy, oxygen plasma, wafer bonding, self assembly

■

INTRODUCTION Conventional deep-UV photolithography with its soaring costs and complexity is neither practical nor necessary for formation of low-cost and simpler arrays of devices such as plasmonic, magnetic, and photonic sensors with sub-100 nm features.1 These simpler-to-fabricate large-area devices demand a more affordable and less-complex technique to define the minimum size of typically between 500 and 10 nm even if the tolerances may be high and the yield is low.2 One popular alternative for making large arrays of nanostructures in this range is selfassembly that relies on the deposition of monolayers of molecules, proteins, or colloids via activated or functionalized surfaces on a given substrate. Being a maskless process and requiring no radiation source for exposure, self-assembly is probably the cheapest and simplest approach to form ordered nanostructures down to 10 nm as long as the intended pattern is a simple 2D hexagonal-closed-packed (HCP) or square lattice. In particular, nanosphere lithography (NL), where nanoscale (d ≪ 1 μm) polymer or glass spheres form a HCP 2D lattice on an “activated” surface such as glass or silicon, offers a scalable and low-cost fabrication alternative. NL is especially useful for building artificial 2D-crystals for magnetic, plasmonic, and photonic sensors as well as nanowire/nanotube © XXXX American Chemical Society

growth templates that can be defined with a single lift-off or etching step, as found in many previous published works.3−5 The most critical step in self-assembly is the surface treatment that subsequently allows formation of the nanostructures of interest. Existing methods for this step either involve purely chemical functionalization or make use of so-called piranha treatment.6,7 The former relies on coating of the surface with unique functional molecules with highly specialized bonding groups with high affinity for the nanostructure or molecular groups to be assembled, while the latter requires prolonged immersion of substrates into a high-molarity mixture of sulfuric acid (H2SO4) and oxidizing agent (H2O2) in elevated temperatures (≥80 °C) to form a hydrophilic surface with a high density of siloxane (≡Si−O−Si≡) groups. Another alternative activation technique may use UV/ozone (UVO) cleaning technique that exposes target surfaces to a high intensity UV source or high density O3 sources or a combination of both.8 The UVO process requires high-power lamps, high (≥200 °C) temperatures, and hazardous levels of Received: December 29, 2014 Accepted: October 28, 2015

A

DOI: 10.1021/acsami.5b08358 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Research Article

ACS Applied Materials & Interfaces

Table 1. Qualitative Comparison of Significant Design Concerns among the Main Technologies Used for Activation of Si Surfaces for Nanosphere Lithography Applications activation technology 6

functionalization piranha solution7 UVO8 O2 Plasma

energy usage

user safety

waste products

corrosive impact

process variables

thermal load

process time

substrate choice

low medium high medium

high low low high

low high low none

low high high low

none few few many

low medium high low

high high high low

none many few many

sonicated for another hour at 80 °C. Base piranha treatment increases surface hydrophilicity by added OH groups. After an hour, these substrates are thoroughly washed again with DI water and stored in water for future use. Surface Characterization. All processed samples were characterized within the first 5 min of plasma activation using an Agilent 5500LS AFM utilizing either sharp (r ≤ 10 nm, f = ∼200 kHz) silicon tips for imaging or large (r ≤ 2.5 μm) SiO2 colloidal tips for spectroscopy mounted on fairly stiff ( f = 120 kHz, k = 14 N/m) cantilevers. Colloidal tips provide much larger area than standard tips, resulting in better representation of surface-to-sphere interactions and larger signal-to-noise ratio during the spectroscopy measurements. For optimum plasma conditions, where Si surface becomes extremely active, standard Si (radius r ≈ 10 nm Si tips with f = ∼15 kHz) were used. Each data point in the spectroscopy results represents the statistical average of 16 measurements on the same sample. The same cantilever was used throughout the methodical study to eliminate possible errors associated with the uncertainty in cantilever elastic properties. Contact angle measurements have been obtained using a home-built micro droplet delivery (50 μL) and side-camera imaging system. X-ray photoelectron spectroscopy (XPS) spectra for the Si surfaces were collected using a Kratos XSAM 800 instrument with VISION2 control and a stainless steel sample holder. A monochromatic Al−Kα X-ray (hν = 1486.6 eV) was used as the source, and Si 2p peaks were used as reference. Deposition and Assembly. Different nanosphere suspensions and two different (drop-coating and spinning) methods for nanosphere deposition were tested to demonstrate the usefulness of plasma activation process by assembling a uniform monolayer over large sections of the sample. This large monolayer is an utmost necessity to promote this process from a mere laboratory test to practical device manufacturing process. Polystyrene latex ([CH2CH(C6H5)-]n) nanospheres of sizes 0.35 μm (2.5 wt % in water), 0.39 μm (4 wt % in water), and 0.5 μm (2.5 wt % in water) were used for testing. To further dilute the nanosphere stock solutions and reduce their viscosity and surface tension, a 1:1 emulsification was done in ethanol (C2H5OH). During coating, typically 8 μL of solution is used, which is chosen to have a sufficient number of nanospheres to cover the entire sample surface using the w/v density. This helped in adjusting the relative concentration (g/mL) of the nanospheres in the solution so as to facilitate formation of monolayers without extensive defects and clutters especially around the edges. For the inkjet printed example, a standard Epson Workforce-30 printer utilizing Novacentrix silver conductive nanoink is used.

ozone with safety and environmental implications. Hence, the existing methods either demand a highly specialized chemistry limited to a unique set of materials or copious amount of hazardous solutions and environmentally unfriendly byproducts, as summarized in Table 1. Therefore, a safer and easier approach to activate Si surfaces with high degree of control and efficiency would be very desirable. In this paper, the use of O2 plasma activation9 is proposed as a safer, simpler, cost-effective, scalable, and easy-to-control alternative for self-assembly. More specifically, the use of O2 plasma treatment to improve silicon surface hydrophilicity is demonstrated via contact angle measurements and methodically studied by a novel approach utilizing atomic force microscopy (AFM) force-spectroscopy measurements. As outlined in Table 1, O2 plasma has many significant advantages10 over techniques used for enhancing surface hydrophilicity or specificity in the context of NL process. Its main drawback is the fact that it is a vacuum-based process that requires specific instrumentation. However, this is hardly an issue in modern nanofabrication environments since O2 plasma generation is possible using relatively compact and affordable systems that are used also for other fundamental processes such as cleaning and etching steps in device fabrication. Although the plasma activation process is routinely used for wafer-to-wafer bonding process11 and microfluidics12 technology, surprisingly, it has not been applied to the self-assembly of nanospheres in the NL process previously. Moreover, to our best knowledge, plasma activation has not been studied methodically using the force-spectroscopy approach proposed in this work either. Therefore, the work presented here has both practical and fundamental implications for the use of O2 plasma process for self-assembly, its characterization, and its application.

■

EXPERIMENTAL METHODS

In the following we provide a brief description of materials and experimental techniques used in this study. All experiments were completed under room temperature. Surface Activation. Boron doped p-type Si wafers are used as the target substrate in this work by cleaving the original 3-in wafers into roughly 2 × 2 cm2 samples that were cleaned with acetone, isoproponal alcohol, rinsed with deionized (DI) water, and blow dried in N2. Oxygen plasma used for surface treatment is generated using March Instruments CS-1701 reactive ion etcher (RIE) system with a parallel-plate chamber capacitively coupled to an RF source at 13.56 MHz and chilled at ∼15 °C with facility water recirculation. Any given O2 plasma treatment in all cases ends with the chamber pumped to the base pressure ∼40 mTorr followed with high purity N2 purge for 10 s. Several control samples were also prepared using the piranha solution treatment based on a 3:1 mixture of 96% sulfuric acid (H2SO4) and 30% hydrogen peroxide (H2O2). Extreme care must be exercised in handling this volatile mixture heated to 80 °C while boiling the samples for 1 h. Following piranha treatment, the samples were washed in DI water and transferred into a 5:1:1 solution of water, ammonium hydroxide (NH4OH), and hydrogen peroxide (H2O2), also called base piranha solution. Then substrates in this solution are

■

RESULTS Verification of Plasma Activation. First, a preliminary study that compares plasma activation with a more conventional piranha solution treatment and verifies its hydrophilic response on silicon surface is presented. Figure 1 shows several examples from the contact angle measurements performed on O2 plasma processed (P = 75 W and 250 W, p = 300 mTorr, t = 50 s, flow = 10 sscm) silicon samples and two control samples (no plasma and N2 plasma). Obviously, the oxygen plasma leads to strong hydrophilic behavior with water droplet spreading thinly with a substantial (almost ×6) reduction in the contact angle as compared to the Si (unprocessed) control sample (see Table 2). Moreover, a second control sample B


Research Article


(control) and activated Si surfaces are remarkably different. Furthermore, the technique also discloses that energy is slightly higher for the plasma activated surface than the piranha treated surface, which implies that the former process is likely to be more effective. Since these curves were obtained using sharp Si tips used normally for AFM imaging, the contact area (hence the resulting stiction energy) is rather small. Thus, it is decided that colloidal tips, while not capable of proving a highresolution image, are the preferred choice for force spectroscopy on activated surfaces. The effectiveness of plasma activation and the colloidal-tip force spectroscopy approach is illustrated in Figure 3, which

Figure 1. Pictures captured during contact angle measurements of (a) Si control, (b) O2 plasma with P = 75 W, (c) O2 plasma with P = 250 W, and (d) N2 plasma with P = 150 W. Summary of measured contact angles is provided in Table 2.

Table 2. Contact Angle Measurements on Si Samples Processed under Different Power Levels and Plasma Gases. DI Water (18 kΩ) Is Used for These Tests power

contact

plasma gas

comments

0W 75 W 150 W 150 W 250 W

37.91° 14.43° 7.40° 41.98° 19.83°

none O2 O2 N2 O2

control

control

processed in nitrogen plasma under identical conditions as the oxygen case failed to produce hydrophilic behavior, as also shown in Figure 1. In fact, there is a slight increase in the angle, that is, hydrophobic behavior. Although similar results exist in the literature,13,14 the above contact angles study confirms the viability of our toolset and effectiveness of O2 plasma conditions to induce hydrophilic behavior. It also indicates that RF power is a suitable parameter of process optimization as the contact angle is lowest at 150 W. Next, the feasibility of AFM force spectroscopy in place of contact angle measurements is studied. Force spectroscopy is a powerful technique. Not only does it allow a direct record of the surface tip interactions, but also it provides a quantitative platform to measure the stiction energy associated with negative deflection as the cantilever retracts. Figure 2 illustrates how force spectroscopy technique is useful to differentiate between untreated and active silicon surfaces prepared using both O2 plasma or piranha treatments. The force−distance (F−d) curves obtained from this spectroscopic comparison clearly show that the stiction energies recorded for an untreated

Figure 3. Example F−d retraction curves measured using colloidal tip (inset) at two different plasma RF power for surface activation. For comparison Si control sample is also shown.

compares the retraction curve obtained from an untreated silicon surface with two other samples exposed to 100 and 200 W plasma. Clearly, the higher the plasma power, the larger the stiction energy associated with the triangular area under the free cantilever position. XPS Study of Surface Bonds. To provide further verification for and insight into the O2 plasma-mediated surface activation process, we present in Figure 4 the results of a mini XPS study that compares spectra for as-received Si samples versus three samples activated with RF power of 45 W, 100 W, and 300 W for 65 s. For brevity and clarity, we only show O 1s and Si 2p peaks that are known to dominate the response of O2 plasma activeated surfaces.9 We observe that the O 1s peak does shift to higher energies and also grows in fwhm, which is indicative of formation of Si−O2 and Si−O4 bonds during plasma-driven oxidation of Si surface. The shift to higher energy is less for highest power (300 W) sample, which also has substantial broadening. This indicates formation of Si−OH bonds9,15 that are likely to be supplied by residual water molecules from chamber walls bombarded by more energetic ions. In the case of Si 2p spectra, we can clearly establish the weakening of the dominant Si−Si bonds and formation of Si− O2 bonds as the power increased. Notably, the highest Si−O2 peak in this case belongs to 100 W sample. The corresponding peak for the 300 W sample is not only reduced in fwhm, but also it is slightly shifted to lower energies, which may indicate the breaking of some of the Si−O bonds by high-energy collisions. Given a similar shift in the O 1s peak for the 300 W case, the same residual OH groups are likely to play a part in this shift and reduction in Si−O bond density.

Figure 2. Force−distance (F−d) curves measured using standard AFM tips on (plasma and piranha) activated surfaces and untreated samples. C


Research Article


Figure 4. XPS spectra recorded for three Si samples activated at different RF power levels for 65 s and control sample. (a) O 1s and (b) Si 2p peaks. All spectra are aligned to Si 2p reference.

Figure 5. Surface energy plots obtained from statistical F−d curve assemblies versus plasma activation process parameters: time, pressure, and flow rate. Horizontal gray band signifies the average value and standard deviation of the same measurement on four unprocessed Si samples.

Optimization of Plasma Activation. Since we established the effectiveness of plasma activation and F−d curves, optimum processing conditions that can maximize the hydrophilic

behavior can be investigated in a methodical manner. This is best accomplished with a statistical approach where multiple F−d curves are collected on a given sample in an effort to D


Research Article


Figure 6. Correlation between contact angle and surface energy plots versus plasma RF power after process is optimized. For comparison, surface energy plot from initial process conditions is also shown.

perturb established suboxides and siloxane groups responsible for surface activation. This is consistent with the shift of O 1s and Si 2p peaks to lower energies in XPS observations for P = 300 W sample. Accordingly, the secondary broader peak in Figure 6 centered around 225 W is not recommended for optimal conditions. Third, extending plasma exposure time beyond 60 s does not lead to linear improvements in activation, analogous to higher power behavior. In other words, prolonged exposures may not necessarily lead to much more active surfaces. We also observe an anomalous dip in activation for flow rates between 25 and 35 sccm. We believe this is an artifact of the flow pattern of O2 into the chamber, as it is introduced from one side, and not the top, of the chamber. Moreover, such sensitivity to O2 flow-rate suggests that the process is masstransport limited rather than reaction limited. Likewise, pressure does not lead to monotonic increase in activation either. In fact, it dips around 200 mTorr level, followed by an almost linear recovery. Similar to the flow-rate behavior, O2 pressure can impact the flow pattern into the chamber, which can also explain this anomalous dip. In any case, limited pressure control in the present RIE system was the factor preventing a wider pressure window below 100 mTorr, where O2 ions would be more energetic. It would be useful to explore the impact of lower pressure and flow rates, which may provide additional flexibility in process design. Temporal Development of Activation. To probe the temporal changes on plasma activated samples upon exposure to the room temperature conditions, we studied two sets of samples (control versus 100 W and 250 W activation) over 4 days using the AFM force spectroscopy. The first sample has data 2−96 h after activation, while the second sample includes detailed data from the first 2 h of activation as well. Shown in Figure 7, these data disclose that activation peaks after the first 60 min by two orders of magnitude and remains strong for 12− 24 h, after which it rapidly wears off in a scale of ∼24 h. Activated using the optimum conditions introduced earlier, these unique data clearly show what the XPS data implied: humidity and O2 in the air plays a crucial role in the activation. The plasma activation is ultimately lost, presumably due to minimization of surface energy with all volatile species lost and

deduce an average behavior over different sections of the active substrate. In fact, since plasma activation itself is a stochastic process and F−d curves are prone to systematic experimental errors and variations in ambient conditions such as humidity and temperature, statistical averaging is the only accurate approach to pick up minute changes in the process conditions and eliminate these errors. Consequently, 16 sets of data were taken on each sample processed in a given set of process parameters. To optimize plasma activation, major process parameters including gas pressure, flow rate, RF power level, and duration of exposure were adjusted systematically. Without a priori knowledge of best conditions for processing, initially we fixed process parameters at P = 125 W, p = 300 mTorr, FR = 10 sccm, and t = 50 s, unless otherwise noted. A consolidated plot of the dependence of stiction energy on each of these parameters while others were kept constant is shown in Figures 5 and 6. It can be seen that the most significant change in the surface stiction energy is found by altering power and time of exposure, while the flow rate and pressure have a weaker impact on the surface activation. After these initial tests, a secondary power optimization was run in which all other process parameters were fixed at optimum values from the initial run (p = 250 mTorr, FR = 20 sccm and t = 65s), as indicated in Figure 6. For the “optimized” process, we also recorded contact angles 24 h after AFM spectroscopy since surface exhibited extreme hydrophilic behavior immediately following plasma exposure. This second run yielded optimum RF power level for plasma activation around 100 or 225 W. On the basis of the extensive study of process variables presented in Figures 5 and 6, we can make a number of important observations. First, it appears that the contact angle data almost perfectly correlate with the AFM stiction energy, providing further confidence in this technique to measure surface activation. Since AFM imaging can be done on the same platform, this may be a desired approach. Second, large power levels do not guarantee best activation, which is visible in both initial and optimized runs. In fact, it leads to higher standard deviations in stiction forces across the sample. High RF power results in higher DC bias across the plasma, increasing the kinetic energy of oxygen ions, which can create defects and E


Research Article


the initial and optimized runs. Admittedly, any given process optimization study is specific to the geometry of the chamber and the capabilities of the instrument used. This factor should be taken into account while using O2 plasma activation and optimizing the process. However, the general trends observed here are likely to be valid for other systems, insofar as specific values for ideal processing windows are subject to change. To check that plasma process does not alter surface morphology, known to affect the hydrophobic/philic behavior, AFM scans were also recorded on 3 × 3 μm2 areas of samples at different power levels, as shown in Figure 8. The average (Ra) and RMS (Rrms) roughness calculated from multiple (≥3) scans reveal that micro level roughness remains within the specifications of the original wafer and cannot be behind the observed changes in surface activation. However, this observation is at odds with the roughness argument used by Alam and co-workers9 to explain minor changes in activation with nm level roughness. Roughness of the samples studied in this work remained the same in both initial and optimized runs, despite large jumps in activation levels, lending no support for this claim. The XPS data reported here show that chemical interaction with oxygen and the formation of different suboxides in the form of Si−O2 or Si−O4 bonds plays important role in Si activation. Both O 1s and S 2p XPS peaks shifted to higher energies, which is consistent with oxidation, as discussed by

Figure 7. Surface stiction as a function of time shows that, upon exposure to air, activation reaches a maximum within an hour and fades away after 12−24 h depending on the room humidity conditions (uncontrolled) for two samples studied. Inset: the same data in linear scale.

open bonds consumed. This is a very important observation because not only the exposure times, but also the storage time and conditions are clearly pivotal in optimizing the activation.

■

DISCUSSION The data presented above show that O2 plasma activation can and should be optimized. An almost hundred-fold increase in the stiction energy was possible at 100 W power level between

Figure 8. AFM scans of (a) Si control and plasma processed (b) 45 W, (c) 100 W, and (d) 250 W samples reveal that average (Ra) and RMS (Rrms) roughness remain the same and are not correlated with hydrophilic behavior. F


Research Article

ACS Applied Materials & Interfaces Alam.9 Generally, the higher the power, the higher the oxidation levels (shifts in XPS peaks), but activation does not increase monotonously, as was also the case in the AFM stiction curves and contact angle data. Hence, this is a complex story of breaking of Si surface bonds, attachment of O2, and residual OH− groups in the chamber at higher energies that must be systemically studied before a firm conclusion on the exact contributions from each group to surface activation is understood.9,15 The limited resolution of XPS data in the present work prohibits a systematic fitting and detailed analysis. Temporal data in Figure 7 are significant at two levels. First, it shows that using AFM force curves, we can collect series of data using the same sample and tip, whereas in the contact angle technique, a separate sample must be prepared for each time interval studied. This is because water contact can irreversibly change development of activation, while this is not the case for “dry” AFM tips. Second, these data clearly disclose the time scale for maturation and wearing off of activation that can be important in applications that require maximum adhesion or assembly conditions. We believe it is the first such detailed temporal statistical study of RIE activated Si surfaces. These results also provide additional insight into understand the importance of physical and chemical components of plasma activation process. For instance, the absence of activation in N2 plasma (Figure 1 and Table 2) suggests that pure physical bombardment is not enough for surface activation since oxygen and nitrogen atomic masses are very close. Thus, the chemical effect is the key element. This is further supported by weak dependence of plasma activation on excitation power and prolonged exposure times. However, the interplay between the physical and chemical processes must be further studied to eliminate the possibility that they combine at a “special” set of conditions to maximize or minimize surface activation.

Figure 9. SEM (black and white) or optical microscope (×200 color) images of samples on which 500 nm PS nanospheres (2.5% w/v) were deposited after (a) piranha treatment (control) via drop coating, (b and c) plasma activation via drop coating, and (d) plasma activation via spinning (4% w/v PS solution). Inset in panel c shows the same sample in higher magnification. Plasma conditions correspond to the initial process (150 W, 50 s, 300 mTorr, 10 sccm).

■

APPLICATION EXAMPLES As a means to further verify the plasma activation behavior and demonstrate the use of the proposed technique for selfassembly, several different kinds of nanospheres (PS and SiO2) were deposited on plasma treated samples. Optical and scanning electron microscopy (SEM) images of PS deposition on plasma treated samples can be found in Figure 9. First, to eliminate false negatives, the same spheres were deposited also on piranha treated surfaces (Figure 9a), confirming the selfassembly process. Subsequently, both drop (Figure 9a−c) and spin (Figure 9d) coating on plasma treated Si have been tested. Similar results were also obtained using SiO2 nanospheres samples, which are omitted in the picture for brevity. Clearly, the plasma activation results in excellent monolayer assemblies of the nanospheres, with sporadic point and line defects produced by these primitive deposition techniques. Moreover, the difference between a highly and under-activated surface becomes visible upon assembly of nanostructures, as shown in Figure 10, panels a and b, where we deposit 350 nm PS (4% w/ v) nanospheres on Si. While the assembly is not perfect, using larger samples, hands-free delivery apparatus, or with the aid of more accurate assembly techniques such as meniscus coating4 and surfactant-mediated5 deposition, the NL yield can be further improved. However, this is outside the scope of the present work and may be attempted in another publication along with other plasma configurations such as inductively coupled plasma chambers where the chemical effect is typically stronger than the physical bombardment found in RIE systems.

Figure 10. Additional examples to the use of nanoassembly via optimum plasma activation. The 350 nm PS spheres (4% w/v) deposited on Si at (a) low (75 W) and (b) high (125 W) activation in the initial process (50 s, 300 mTorr, 10 sccm). Commercial silver nanoink (65 nm Ag nanoparticles) deposited via inkjet printing on PDMS sheets (c) before and (d) after a mild (50 W, 10 s, 300 mTorr, 20 sccm) plasma activation confirms the broad use of plasma activation on any silicon-rich surface.

In Figure 10, panels c and d, we provide an additional example to the use plasma treatment on a silicon-rich polymer (PDMS, polydimethylsiloxane) surface using a commercial nanoink containing 65 nm Ag nanoparticles. In this case, the delivery was facilitated via an inkjet printing system that utilizes piezoelectric drive to form ∼10 pL droplets. Evidently, the original PDMS surface is highly hydrophobic, and the printing G



■

ACKNOWLEDGMENTS The authors are thankful to Ning Jing of Institute for Corrosion and Multiphase Technology (ICMT) at Ohio University for the assistance during the contact angle measurements. Dr. Munir Nazzal is also thanked for useful discussions on force spectroscopy technique. This work is partially supported by Stocker Endowment of Russ College of Engineering at Ohio University.

directly on fresh PDMS surfaces fails. Upon a mild O2 plasma activation, however, it is possible to obtain rather uniform and continuous printed lines of silver. Plasma power (50 W) and time (10 s) were reduced in this case to avoid damage to C-rich surface and loss of resolution in the inkjet printing process. Although plasma activation on a PDMS surface was studied in detail before,15 its use for nanoassembly or printing is attempted for the first time. Finally, it is worth reminding that plasma activation is by no means limited to NL or printed electronics. It is already used in many other micro- or nanofabrication processes including wafer-to-wafer bonding,16 packaging, microfluidics, wettinglayer formation, and membrane deposition. Thus, the practical implication of the presented results is well beyond the NL, which serves as an example here and happened to be the initial motivation behind this work while an alternative to hazardous piranha treatment was sought.

■

CONCLUSIONS In summary, oxygen plasma activation of silicon surfaces has been systematically studied as a viable alternative for simplifying and optimizing the activation process critical to the selfassembly processes. The plasma-mediated enhancement of hydrophilic behavior on silicon has been verified using both contact angle and force spectroscopy measurements, the latter of which has not been reported before. By using this novel approach, optimum conditions for O2 plasma processing have been identified for the capacitively coupled RF plasma. Plasma power of 100 W, pressure at 250 mTorr, and 65 ∓ 5 s of plasma exposure were found to be most effective, while the flow rate is found to have a weaker impact on the surface activation within the range of parameters available to this investigation. XPS spectra of the activated Si surfaces confirm the presence of suboxides (Si−Ox) via shifting of the O 1s signal to higher energy, and emergence of strong Si−O2 peak in the Si 2p spectra, as plasma power is varied. AFM scans indicate that surface roughness remains

A Systematic Study of Plasma Activation of Silicon ... - ACS Publications

Recommend Documents