**1. Introduction**

As microscale manufacturing grows in scale, the cost of production can be greatly reduced through manufacturing techniques that operate at ambient pressures and low temperatures. While electrosprays have long been a keystone of certain large-scale manufacturing industries such as paint coating and agriculture, electrospray deposition (ESD) has yet to reach broad industrial-scale application. This distinct regime of electrospray operates at low flowrates and relies on only electrostatic forces to deliver material. ESD has several unique capabilities for thin film [1–4] and nanoparticle [5–8] processing that are more relevant now than ever. However, further understanding and demonstration of the capabilities of ESD are necessary before employing it in, for example, the creation of dielectric coatings, hermetic barriers, and the deposition of active sensor materials. More specifically, this manuscript explores the potential for ESD to be combined with microfabricated templates without the need of a stencil mask through understanding and leveraging of charge dissipation mechanisms.

During ESD, a strong electric field is applied to a capillary containing a solution or dispersion. This causes the exiting fluid to undergo electrostatic breakdown and produce generations of charged monodisperse droplets that deliver their payloads to a grounded

**Citation:** Grzenda, M.J.; Atzampou, M.; Samateh, A.; Jitianu, A.; Zahn, J.D.; Singer, J.P. Microscale Templating of Materials across Electrospray Deposition Regimes. *Coatings* **2023**, *13*, 599. https:// doi.org/10.3390/coatings13030599

Academic Editor: Rainer Hippler

Received: 10 February 2023 Revised: 3 March 2023 Accepted: 6 March 2023 Published: 11 March 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

target [9]. For ESD of non-conductive soft matter, we have categorized this behavior into three regimes that depend on the ability of the deposited film to dissipate charge [10]. If the deposited film maintains a certain level of mobility and low viscosity, it will thin to increase surface area and conduct charge to uncoated regions, which is called electrowetting. While spraying liquids or solid materials above their glass transition temperature (Tg) is an obvious way to maintain mobility, it has also been observed that higher temperatures can induce greater mobility by increasing the solvent absorption, which, in turn, decreases the Tg. The charged melt regime also depends on mobile material, but in this case, the film's higher viscosity favors instabilities to dissipate charge. Due to the dependence on mobility, these electrospray regimes exist in a balance, and it has been observed that electrowetting materials can transition to charged melts as the material thickness increases. Further, if the surface affinity to the target is poor, either regime can result in dewetted films. On the completely opposite end of the spectrum is the self-limiting (SL) regime where glassy materials below their effective Tg are sprayed in a phenomenon we have named self-limiting electrospray deposition (SLED). In the SL regime, charge is unable to dissipate from the growing film, and this charge build-up eventually results in the repulsion of nearly all the incoming material. This regime also transitions to the charged melt regime as Tg is approached through either increased temperature or blending with solvent vapor or another material, as the charge is better able to dissipate through mass transport [11]. One of the main advantages of SLED is the ability to conformally coat conductive/charge mobile surfaces, and much of our prior work has focused on 3D, corona-free coatings [12]. However, the ability to target micron-scale 2D patterns is also highly valuable, but in need of further characterization.

While several studies have patterned electrosprayed materials simply by controlling local electric fields [13,14], the typical approach has been through some form of masking, with authors depositing nanomaterials (both organic and inorganic), biomaterials, and polymers [15–22]. In one of the earliest examples of patterning with ESD, Buchko et al. used a shadow mask for depositing arrays of polypeptides, with the term shadow mask implying that the mask was conductive, and the substrate and mask were equally coated like a stencil [23]. It was Morozov and Morozova who first described the use of an insulating mask for the deposition of arrays of DNA material with the observation that the electrospray quickly traps charge on the mask, leading to a repulsion that focuses the spray onto conductive regions [24]. Importantly, this focusing increases the deposition thickness relative to an unmasked substrate. Hu et al. also observed that focusing could even be achieved without a mask when spraying block copolymers onto glass using a grounded needle and a high voltage grating beneath the substrate to direct the spray [14]. Interestingly, their (likely charged melt) spray was observed to spread beyond the width of the buried grating.

While material focusing can be useful, especially when attempting to achieve small linewidths, we must also consider spray uniformity. For unpatterned, non-SLED sprays, deposition tends to lead to a peaked distribution. On the other hand, focusing depends greatly on geometry, meaning that the spray of complex 2D patterns will lead to large variations in the amount of material deposited, which can be especially problematic when manufacturing multi-layer materials. This was encountered by Morozov and Morozova, who tried to counteract the effects either by moving the spray platform or by adding a shield to the spray environment [24]. Though successful to a degree, their array features were all of equal size, so it is unclear if this effect would carry over to more complex geometries.

Further, work from our group has shown that even materials sprayed in the SLED regime can be impacted by feature size. In a recent study, focused laser-spike (FlaSk) dewetting was used to pattern circular holes of varying sizes on photoresist-coated silicon wafers as templates [22]. After spraying with polyvinylpyrrolidone in the SL regime, it was observed that larger features tended to have thicker coatings, and that all features had coatings that were thicker than an unmasked substrate. This is curious in light of our initial work that showed that thickness in the SL regime was dependent only on electric field strength, and that below the SL thickness, total sprayed mass was the only determining factor. However, at the time, it was observed that deposited materials were slightly collapsed, indicating that excessive solvent vapor may have been generated due to focusing from the template. As mentioned earlier, increased solvent absorption can lead to charge dissipation and electrowetting/charged melt behavior. These effects can, therefore, be mitigated in three ways: (1) the total patterned area can be increased, (2) the flowrate can be reduced, and (3) the strength of the dielectric layer can be reduced to decrease the effect of focusing. Regarding the third point, Zhu and Chiarot observed that thinner photoresist layers allowed for deposited nanoparticles to strike closer to recessed feature edges, indicating a reduction in the focusing effect [15].

Here, we aim to elucidate the performance of all three spray regimes by spraying substrates composed of silicon chips patterned with Ti/Pt on top of an insulating layer of Parylene. Furthermore, we aim to quantify the interaction of complex 2D geometry with ESD through the use of a metal pattern that includes multiple gratings of varying sizes and spacing. A MATLAB script is used to extract values for average deposition height, volumetric density, along with a simple figure of merit that characterizes the specificity of the spray to the template.

#### **2. Materials and Methods**

#### *2.1. Materials*

Polystyrene (35 kDa) and 2-Butanone (>99%) were purchased from Sigma Aldrich (St. Louis, MO, USA) and were used as received. Polystyrene (PS) was added to butanone at a mass loading of 1 wt.% and left on a roller overnight. A "melting gel" material was used to demonstrate a fully liquid, electrowetting spray. Melting gels (MGs) are oligomeric silsesquioxanes (synthesized via a sol–gel process described elsewhere [25]) that show thermoplastic-like behavior above their Tg, but cross-link into hybrid organic/inorganic glasses when heated past their consolidation temperature [26]. The melting gel used here was composed of 65% methyltriethoxysilane (MTES) and 35% dimethyldiethoxysilane (DMDES) (in mol%) and had a Tg below 0 ◦C. The MG was diluted using absolute ethanol to 1 wt.% MG for spray. All sprays were performed on Ti/Pt patterns on 4–5 μm thick Parylene insulation. These substrates were made in-house via photolithography and acetone lift-off using 4-inch silicon wafer substrates.

### *2.2. Feature Test Patterns*

The feature test patterns were designed to assess a series of gratings that range in both grating width and spacing (Figure 1). The gratings were attached to a 1 × 1 cm2 grid used as a grounding path connected to a 0.5 cm2 grounding pad. Every individual grating filled an approximate width of 1.5 mm and the features had a length of 1 mm. On the interior of the grounding grid, the feature width, referred to as the feature size, increased in multiples of 15 (15/30/60/120/240 μm). Around the exterior, the feature sizes increased in multiples of 20 (20/40/80/160 μm). Along each row, the feature size was constant, but the gap between each finger increased as a function of the feature size (1/3x, 1x, 2x, 3x, and 9x). For example, the 30 μm row had gaps of 10, 30, 60, 90, and 270 μm. For this work, we focused on the 30/60/120/240 μm features. The 20 μm features were included to contrast the 30 μm features due to their similar size but different spatial location on the test pattern. However, for both these smaller feature sizes, the 1/3x gratings were excluded because they were too small for the profilometer to measure.

#### *2.3. Electrospray Setup*

Strong electric fields were applied via alligator clips to a vertically suspended 30-gauge needle and a stainless-steel ring (ID 2 cm) using high-voltage power supplies (Acopian, P012HA5M, Easton, PA, USA) to achieve electrospray. The ring was placed 1 cm above the needle tip and acted to stabilize the electric field and focus the spray. The voltage on the needle was held at a constant 6 kV DC, while the voltage on the ring varied from 3 to 5.5 kV

DC to maintain stable cone jet sprays. The solution was pumped to the high-voltage vertical needle using a syringe and Teflon tubing using a syringe pump at a rate of 0.1 mL/hr. During sample fabrication, the spray was first stabilized on a grounded, blank, 4-inch Si wafer (p-type; 0–100 Ω cm), stacked on top of a steel plate and hot plate, and located 6 cm below the needle tip. Once the spray was stabilized, the test pattern was placed in the center of the spray and allowed to coat for 30 min. Separately, the grounding pad of the test pattern was taped with carbon tape to a 2-inch wafer (p-type; 0–100 Ω cm) to create a grounding path. This 2-inch wafer was able to interface directly with the grounded 4-inch wafer. While the PS sprays were sprayed in ambient atmosphere (30% humidity/22 ◦C), the MG was sprayed in a low-humidity chamber (21% humidity/25 ◦C) due to its moisture sensitivity. For the PS, two sprays were performed, one at a hotplate temperature of 30 ◦C, which is in the SL regime, and the other at 100 ◦C, which is near or above Tg due to solvent swelling. The MG was sprayed at 35 ◦C to completely ensure it was sprayed as a liquid. After spraying, the MG was consolidated overnight at 150 ◦C to fully crosslink. Although SLED sprays are highly porous, the PS was densified via solvent swelling using a brief exposure to butanone vapor to allow for profilometry scanning.

**Figure 1.** (**a**) (top) Full feature test pattern and (bottom) single row demonstrating the distal/proximal scans. (**b**) Actual uncoated test pattern profile (black) compared to calculated binary profile (shaded red). An error can be observed in the center of the grating that has been corrected in the model.
