Electrospray deposition tool: Creating compositionally gradient libraries of nanomaterials

A fundamental challenge of material science is understanding the interplay between a desired material property and its structure, particularly when the structure is a function of individual constituents and the assembly process. Combinatorial and high-throughput methodologies have greatly accelerated research by addressing multiple experimental parameters in parallel.¹ Pioneering efforts in the creation of single-sample libraries of components date back to the mid-1960s for metal catalysts, but this “multisample concept” only became practical after innovations in library preparation, rapid testing, noncontact or nondestructive screening, automated analysis, rapid interpretation, and data management.^2–4 For soft matter research, in particular, the creation of a gradient library over a range of compositions, processing conditions, and other parameters is desirable for exploring an entire parameter space and finding precise phase boundaries or other regions of interest. Furthermore, unlike discrete samples, continuous gradient libraries can illuminate trends and key results without extensive analysis and reduce sample to sample errors and inconsistencies without the fear of missing data points.⁵ These combinatorial libraries can be combined with high-throughput measurement tools to enable rapid screening of parameter spaces underlying complex multicomponent materials. X-ray scattering is particularly well-suited to such studies, owing to the fast data collection without requiring sample staining or other preparation. Moreover, recent advances in autonomous experimentation—which allow efficient machine-guided explorations of parameter spaces—can be combined very naturally with electrospray deposition (ESD)-prepared combinatorial libraries to enable mapping of multicomponent material composition spaces.^6,7

Conventionally, thin films are prepared through spin casting, but various techniques have been recently developed to create gradients of thickness, temperature, surface energy, and composition.⁸ These gradient preparation methods include, but are not limited to, blade-casting and flow coating techniques,⁹ gradient hot stages,¹⁰ gradient UV-Ozone devices,¹¹ and dipping techniques.¹² While flow coating techniques may start out with a gradient line of solution, thus making a dual gradient of both composition and thickness,¹³ the composition is limited to two components. The sequential dipping technique has synthesized two axial composition gradients of surface-tethered diblock and triblock copolymers, but their application is limited to surface-anchored polymer systems and requires precise precursors.¹⁴ We present a general user tool for creating binary or ternary composition gradients with independent control over film thickness. Our work builds on the utilization of electrospray atomization techniques to slowly deliver small charged quantities of polymer in solution to a heated substrate as shown by the Osuji group for block copolymer systems.^15,16 Electrospray deposition (ESD) of polymer thin films allows for the creation of vertically persistent and ordered domains,¹⁷ continuous and controlled deposition,¹⁸ as well as sequential deposition of multiple polymer systems without the need for orthogonal solvents or extensive cross-linking.¹⁹ By combining gradient flowrates of solutions with a movable x-y-z stage system, compositional gradients of thin film nanomaterial libraries may be created using our ESD tool. Whereas prior gradient methods could only create up to two component systems, this ESD tool is capable of ternary composition gradient thin films with independent thickness control. The ESD tool is available for users at the Center for Functional Nanomaterials (CFN) at Brookhaven National Laboratory with an easy to use LabVIEW interface which offers automated operation as well as real-time stability control of the spray mode.

The use of electrostatic force induced by high voltage application for droplet generation has been used extensively in the last century. The early studies on the mechanism of electrospray date back to Zeleny^20,21 with the theory later explained by Taylor.²² The applications of electrospray span an extensive range from ink-jet printing, paint and surface coatings, fuel atomization and propulsion, drug delivery, and ion source generation in various mass spectrometry techniques.^23–25 In recent years, it has become an attractive technique for the creation of polymer thin films because of the monodispersed, submicron sized droplets created from the break-up of solution under electric stress and the further droplet breakdown from electrical instability conditions. Studies have established scaling laws that describe the droplet diameter, d, as a function of process parameters,²⁶ 

where α is a constant depending on the fluid’s dielectric permittivity, Q is the flowrate, ε is the dielectric constant, ρ is the density, σ is the surface tension, and γ is the conductivity of the sprayed solution. As voltage is increased between a nozzle and a grounded substrate, the meniscus undergoes various spray modes. The cone-jet mode (often referred to as a Taylor cone) gives the most steady and robust performance yielding monodisperse droplets with standard deviations of 5%–15%.²⁷ For uniform film deposition, the spray mode must be monitored as meniscus fluctuations may occur even with constant applied voltage, causing unstable jetting. To address this difficulty, various feedback controls have been presented in the literature. Valaskovic et al. employed photodetectors to detect light patterns and generate photoelectronic signals from a light source which intersects the spray jet, but this process requires extensive calibration for each new solution.²⁸ Other groups have utilized a feedback control by monitoring the electrical current between the nozzle and substrate, but this method needs precise analysis of any leakage current or deposition occurring elsewhere on the setup.²⁹ Commercially, the most common feedback control uses visual image analysis of the shape of the jet to determine the spray mode, but many algorithms only detect changes from one spray mode to another, outputting a constant voltage once a stable cone-jet mode is detected.³⁰ However, over time, tiny differences in experimental details such as shape or wettability of the nozzle, pressure drops, or liquid vaporization may lead to the initiation of one mode rather than another and considerably change the characteristics of the spray.³¹ This is especially significant when spraying from a compositionally gradient solution as gradual changes in solution conductivity and surface tension may cause the spray mode to change even at a constant voltage. The conductivity of the solution also plays an important role in influencing the length of the jet and angle formed by the plume of droplets after jet break up.³² In turn, the length of the jet and angle of the plume influences the uniformity of the film. We base our control algorithm on the analysis of the meniscus morphology, specifically the height of the meniscus in the cone-jet mode, to control the output voltage. By specifying a meniscus height set-point, the applied voltage is automatically adjusted to yield a steady cone-jet mode.³³ 

The basic hardware components of the ESD Tool consist of the substrate positioning system, the gradient syringe pump system, the spray nozzle and monitoring system, high voltage modules, and the substrate temperature controller. The various components and their specifications are identified in Fig. 1 and Table I.

As can be seen from the zoomed in system diagram in Fig. 2, the substrate positioning system is composed of X-Y linear stages (Newport UTS100PP) and a Z vertical stage (Newport GTS30V). The motors give the substrate 100 mm travel distance in the x-y direction and 30 mm vertical travel displacement with a minimum incremental motion of 0.30 µm and an accuracy of 1.7 µm. While many electrospray systems utilize a pressure controller to supply the solution at constant pressure, the Harvard Apparatus gradient pump system is able to achieve a pulse-free solvent delivery due to their improved mechanics, higher linear force, and advanced microprocessor control. Syringe pumps also offer the advantage of precise control of the amount of material deposited as compared with pressure driven systems which would require additional flow control. The gradient syringe pump system includes one main and two satellite units (Harvard Apparatus PHD Ultra) to make gradient solutions of up to three components. The minimum flowrate is 1.56 pl/min using a 0.5 µl syringe, but typical ESD operations only warrant a 1–20 µl/min flowrate using a 1–10 ml syringe. The potential difference is delivered by two high voltage power supplies (Keithley 2290-10) with single digit voltage control up to 10 kV. There is an applied voltage potential to the nozzle and an extractor located a few millimeter below the nozzle. The extractor improves stability as the grounded substrate moves, and it provides a focusing effect.³⁴ For this user tool, the current configuration offers only direct current (DC) capabilities. Use of alternating current (AC) requires the optimization of an additional parameter, the frequency, which has significant impact on the final outcome of the electrospraying process but may offer alternative benefits, such as ionizing large negatively charged molecules.^35,36 The spray adapter (New Objective ADPC-IMS) holds a connecting junction (IDEX MicroCross and MicroTees PEEK) which connects the tubing from the syringe pumps with a metal tapered nozzle (New Objective TaperTips MT32010055). The nozzle end has an opening of 100 µm. Hamilton Gas Tight Glass syringes ranging from 1 to 10 ml were used. The high voltage is connected to the metal needle through the spray adapter. The visualization of the spray is provided by a 1.3 Megapixel resolution CCD camera (Dino-Lite AM4515ZT4). The image resolution is 1280 × 1024 pixels with a 430×–470× optical magnification at 30 frames per second (FPS). Light-emitting diode (LED) illumination is used in the background to help with the image contrast and edge detection. To avoid depositing any materials while the spray mode is being calibrated, a shutter system was implemented (Thorlabs Filter Flipper MFF102) to cover the substrate from the spray plume. The ESD Tool is located in a HEPA-filtered fume hood, and the system contains a safety interlock tab which shuts off the high voltage if the glass pane to the hood is opened during the spray process.

The software to hardware interface and the front-end graphical user interface (GUI) are handled by LabVIEW. Figure 3 shows a schematic of the software-hardware interfacing. LabVIEW was chosen due to its ease in making and using the graphical interface and accessibility to modular capabilities in the back-end hardware interfacing through either universal serial bus (USB) or general purpose interface bus (GPIB). Figure 4 shows the main Virtual Instrument (VI) in LabVIEW which allows the user to control and monitor each step in the electrospray process. On the left side of the screen, the user has control over the two high voltage power supplies, each syringe pump, and the substrate temperature proportional-integral-derivative (PID) controller. The syringe pump controls allow the user to specify the syringe type used, volume, flowrate, and force desired. It also shows the total volume infused and the accumulated time the syringe has spent infusing. A reset option allows the user to clear the infused volume, time, or both as necessary. In the gradient step feature, the user can input a single gradient step command to the pumps, specifying the start and stop percentages of a total flowrate. In the middle of the GUI, the CCD camera offer visualization of the tip of the nozzle. Video format and video setting options can be used to improve the quality through brightness, contrast, saturation, hue, gamma, etc. Adjusting the focus of the camera has to be done manually on the body of the camera. The image analysis thresholds the incoming camera image and has a feature to set the baseline for the tip of the nozzle at no flowrate. When infusion begins, the difference in the baseline and the thresholded image shows up in a different color to denote the solution. The software is used for the automated feedback control loop by determining the meniscus height. This is illustrated in a process flow diagram in Fig. 5. The image analysis finds the intersection of two parabolas created from three points on each side of the cone to find the vertex of the cone. A sharp-angle coefficient is also calculated to determine the concavity or convexity of the shape of the cone. The meniscus vertex point is then used to calculate a meniscus height as simply the distance to the base of the nozzle, and this height can be automatically controlled by a feedback loop on the voltage applied to the needle for added stability. Raising the voltage deceases the meniscus height, while lowering the voltage increases the meniscus height for a cone-jet mode operated spray. Complex programs which give directives to the motors and the pumps simultaneously are imported through recipes in the “Run Program from CSV File” part of the GUI. Single lines in a simple CSV or TXT file indicate the starting and stopping flowrates of the syringe pumps, the x, y, z coordinate motion for the pumps and the speed at the start and end of the motion of the motors. For constant speed motions, the start and end speeds are the same value, while for acceleration/deceleration, the start and end speeds differ. On the GUI, the user is able to assess the progress of the system by monitoring parameters the software records including current and target positions of the motors, instantaneous speed and acceleration measurements, a position map of the substrate, pump percentage fill bars, and a time remaining counter. The software also allows for resetting the motors to the origin, an emergency stop button, and a logfile button. The logfile saves a Technical Data Management Streaming (TDMS) which is compatible with Excel. The logfile records all the hardware states and indicator values for every experiment for a particular user. Finally, an automatic shutter system is offered which opens only when a stable cone-jet mode is determined. While operating the system from a CSV file of list of steps, the time is precisely calculated to the operation, but the system can also be used for manual spray modes in which case a regular stopwatch feature is included to help the user determine the length of spray time.

Substrates for electospray deposition presented here were boron doped Si test wafers, 500 µm thick, ⟨100⟩ orientation, with a resistivity range of 0–100 Ω-cm (University Wafers). Other types of substrates, including insulated substrates, may also be possible as previously demonstrated in the literature with minimal changes to the tool.³⁷ However, insulated substrates require a counter electrode (such as the aluminum heating block on the ESD tool) and the resulting thin films may be thickness limited when sufficient charging occurs on the substrate surface. As with other deposition techniques, care should be taken to clean the surface of the substrate of any unwanted organic contaminants. ESD should be performed immediately after cleaning/drying to minimize adsorption of contaminants. Future versions of the setup could have an inert gas environment instead of air. For the calibration studies, dilute polymer solutions were prepared by dissolving polymer in an appropriate volatile organic solvent, i.e., acetone or propylene glycol methyl ether acetate (PGMEA). Typical concentrations are in the range of 0.1–1 weight % polymer. All polymer solutions were filtered (2 µm) prior to use coating to prevent any particulates or undissolved matter clogging the nozzle or affecting the subsequent film. The conductivity of PGMEA is on the order of 0.2 µ/m from available industry standards using ASTM International (ASTM) Standard Test Method D4308. The tip of the nozzle was also sonicated for 30 min before use. Various extractor ring designs can be used to shape the deposition area (Fig. 6); for the calibration measurements, a cylinder tube extractor was used to produce constant velocity and constant acceleration lines.

In the recipes that users enter, each line contains information for the stage movements and syringe pump gradient operations. The user enters the next position coordinate (x, y, z position in millimeter), the constant speed or acceleration desired, the total flowrate, and the three syringe pump percentage changes during this motion. The stages can operate in a constant velocity or constant acceleration mode. The overall thickness of the deposited film depends on the deposition parameters. In particular, the speed of the stage, solution concentration, and flowrate are the main variables for determining film thickness. The exact geometry of the setup (the z height direction of the extractor and needle from the substrate) also affects the thickness by changing the area of deposition. The setup geometry, however, is usually optimized for a specific purpose, and other parameters are more easily manipulated. Figure 7 shows the calibration results for the two modes of stage operation, constant velocity, and acceleration, respectively. The film thickness is linearly a function of the flowrate, solution concentration, and inversely dependent on motor speed. The polymer used was 1 kg/mol molecular (Polymer Source) weight polystyrene in PGMEA solution. A UV-visible reflectometer (Model F20, Filmetrics, Inc.) was used for thickness analysis across 5 mm linear samples. Spot size was roughly 0.5 mm with acquisition time around a second.

The tool is also capable to create composition gradients which are shown in Fig. 8 where a triangle was created with vertices of single component dyes of 1% by weight in PGMEA: Eosin Y, Rose Bengal, and Oil Red O (Sigma-Aldrich). The three sides of the triangle were each a linear gradient of two of the three dye solutions. The specular reflectance was measured by a UV-visible reflectometer (Model F20, Filmetrics, Inc.), and the absorptance was calculated assuming negligible transmission as A = 100 − R. Since the three dyes have some overlapping wavelengths along the visible light range, a ratio of absorptances at chosen characteristic wavelengths was plotted against the sample position along the side of the triangle (Fig. 8).

A powerful use of this ESD tool is in combination with X-ray scattering measurements in which a created combinatorial “library” on a single wafer is systematically measured. One demonstration of this was creating a gradient of block copolymer to the homopolymer sample. Lamellae forming polystyrene-block-poly(methyl methacrylate) (PS-b-PMMA) of 75 kg/mol (Polymer Source) solution was prepared in one syringe, while a 1:1 ratio of homopolymers PS and PMMA of 22 kg/mol was prepared in the second syringe. A linear gradient of pure block copolymer to homopolymer at 0.1 wt % in PGMEA was deposited for grazing incidence X-ray scattering measurements. The flowrate was 5 µl/min at 190 °C substrate temperature. Prior to film deposition by ESD, the substrate was functionalized by grafting a hydroxyl-terminated “neutral” PS-r-PMMA random copolymer (60% PS, from Dow Chemical) as described previously.³⁸ The Complex Materials Scattering (CMS) beamline 11-BM at the National Synchrotron Light Source II was used to acquire grazing incidence small angle X-ray (GISAXS) patterns at 0.1 mm steps along the 5 mm gradient of increasing homopolymer fraction. For analysis, cross-sectional linecuts in the q_r direction were taken at q_z = 0.026 with d_q = 0.006 [Fig. 9(a)]. Linecuts were also compiled in a logarithmic plot of intensity vs q_r along the 5 mm samples [Fig. 9(b)]. The continuous decrease in the position and eventual loss of the scattering peak are indicative of a smooth increase in the fraction of homopolymer along the 1D gradient. The decrease in the peak position marks an increase in the characteristic spacing of microphase separated domains in the blend. Thus, the ESD tool combined with GISAXS illustrates the precise compositional dependence of the domain spacing in the polymer blend.

We report the design, fabrication, and commission of a new user tool at the CFN that combines electrospray deposition with precise motor movements and gradient pump control. These capabilities make it possible to construct thin film multicomponent “libraries” on a single substrate to rapidly and systematically characterize composition-dependent properties. As an example, we fabricated and characterized thin films involving homopolymer and block copolymer blends. This tool forms an integral part of a new platform for high-throughput, autonomous characterization and design of nanomaterials when combined with X-ray scattering techniques.

This research used resources of the Center for Functional Nanomaterials and the National Synchrotron Light Source II, which are U.S. DOE Office of Science Facilities, at Brookhaven National Laboratory under Contract No. DE-SC0012704. We thank Masafumi Fukuto and Ruipeng Li for supporting experiments at the CMS beamline. K.T. acknowledges NSF support through Graduate Research Fellowship Grant No. DGE-1122492 and an INTERN supplement to Grant No. CBET-1703494. C.O.O. acknowledges NSF support through Grant No. DMR-1410568.