Two-component additive manufacturing of nanothermite structures via reactive inkjet printing

In the growing field of micromechanical devices, there is a subset of systems that require the microscale integration of energetic materials. These applications range from micropropulsion systems^1,2 to microinitiators.^3,4 However, there is an inherent risk associated with handling energetic materials, which limits the diversity of materials deposited using traditional methods such as electrophoresis,⁵ doctor blade casting,⁶ and conventional inkjet printing.^7–9 In addition, methods such as electrophoresis and doctor blade casting lack the spatial and volumetric control of the deposited energetic material, which is advantageous in the aforementioned microscale applications.

In many industries, reactive inkjet printing, the process of combining one droplet of ink A with a droplet of ink B to produce a new material, has been used to overcome material incompatibilities with inkjet printers.¹⁰ Full reactive inkjet printing has been used for the deposition of polyurethanes,¹¹ self-assembled gold nanoparticles,¹² and conductive silver films.¹³ Due to the separate nature of the inks, reactive inkjet printing opens the door for the fabrication of devices using a wider range of materials. The individual inks can be fine tuned with an appropriate solvent or surfactant for only one key element rather than trying to optimize an ink formulation for many constituents, as in traditional inkjet printing. In addition, materials that are volatile when mixed together can remain separate until the fabrication process to improve material safety and handling. The most significant challenge with this process lies in the significant coordination of the printer systems to achieve reliable droplet placement of the inks respective to one another. However, this can be mitigated with a well-controlled fabrication system. Ultimately, this technique has been shown to produce picoliter volumetric control and microscale spatial precision in the deposition of new materials previously unobtainable with inkjet printing due to material properties and safety concerns.

This work utilized a piezoelectric inkjet printer for the side-by-side deposition of colloidal suspensions of nanoaluminum (nAl) and nanocopper (II) oxide (nCuO). These suspensions were printed sequentially such that their droplets were adjacent but overlapping. This method resulted in the localized mixing of two inert components that produced a nanothermite. This nanothermite was analyzed by high speed thermal imaging and scanning transmission electron microscopy (STEM) to compare its reaction performance and constituents to those of nanothermite printed with a single nozzle piezoelectric inkjet printer.

As a means to test reactive inkjet printing as a method for the manufacturing of nanothermite samples, two ink suspensions, a fuel and an oxidizer, were prepared and printed, and the resulting samples were tested for energetic performance. These suspensions were created to achieve a stoichiometric ratio of solids when the inks were mixed volumetrically 1:1. A nAl and a nCuO suspension, of 3.5% and 6% volumetric solid loading respectively, were prepared in dimethylformamide (DMF) with 0.5% polyvinylpyrrolidone (PVP) immediately prior to printing.¹⁴ nCuO (Sigma Aldrich, 50 nm) and nAl (NovaCentrix, 80 nm, 82% active aluminum) were deposited in 1.5 ml microcentrifuge tubes. Subsequently, a solution of DMF with 0.5% PVP was added to the vials. These vials were then suspended in a sonicating bath (Branson Ultrasonics) for 30 min. The solution was loaded into the inkjet printer 5 min after sonication.

As a comparison, a pre-mixed aluminum copper (II) oxide suspension was mixed at a 6% volumetric solid loading.¹⁵ To prepare the material, nCuO was mixed with nAl and suspended in a solution of DMF and PVP and placed in a 10 ml syringe (BD, slip tip). The syringe was loaded into a custom polytetrafluoroethylene (PTFE) holder and secured on a LabRAM resonant mixer (Resodyn Acoustic Mixer, Inc.). The syringe was mixed at 80% intensity for 16 min and inverted after 8 min.

The prepared inks were loaded into 70 μm piezoelectric inkjet nozzles (MicroDrop, MD-K-130-022) and secured above a dual-axis linear positioning stage (Aerotech Planar DL 200-XY, 200 mm travel, 0.5 μm accuracy) that was controlled by an in-house LabView program [Fig. 1].

For the samples printed with dual nozzles, the nAl ink, ink α herein, was printed with a 143 V trigger pulse, 27 μs pulse width, 75 Hz firing frequency, and −8 mbar back pressure. The nCuO ink, ink β herein, was printed with a 93 V trigger pulse, 25 μs pulse width, 230 Hz firing frequency, and −8 mbar back pressure. These settings were optimized for quality droplet formation as observed by side-view imaging with a 3 × 110 mm telecentric lens and a color Universal Serial Bus (USB) camera (Edmund Optics EO-1312) backlit by a light emitting diode (LED). Acceptable droplet formation consisted of a droplet head forming at the nozzle orifice, followed by the necking of the droplet to produce a tail. This droplet then pinched away from the orifice as it fell to the substrate. The majority of the tail morphed back into the main droplet, while the remainder produced a small satellite drop. The satellite drops were deemed acceptable in this study if they fell within the spatial bounds of the main drop. Representative drop formation images can be found in Fig. 2.

Samples were prepared using a four pass printing method in which a 10 pixel × 10 pixel square bitmap was parsed into four sublayers, described herein as A, B, C, and D. Each pixel was assigned to one of the sublayers using the pattern shown in Fig. 3. Sample preparation was achieved by printing sublayers A and D using ink β immediately followed by the printing of sublayers B and C with ink α. This pattern was repeated to achieve the desired number of layers; however, the sublayer associated with a particular ink alternated with each layer. For example, ink β was printed from sublayers A and D on layer 1 but sublayers B and C on layer 2. Samples of 3, 5, and 7 layers were printed on Novele, a mesoporous media which promotes strong adhesion and uniform deposition (NovaCentrix, IJ-220).¹⁶ 

A pre-mixed nAl/nCuO ink was printed using the same 10 pixel × 10 pixel square bitmap and a 70 μm piezoelectric inkjet nozzle (MicroDrop, MD-K-130-022) on Novele to offer a comparison to the dual nozzle system.

After sufficient drying, printed samples were ignited using a capacitive discharge unit (Information Unlimited). The ignition was observed with a Black and White (BW) Phantom Camera V 7.3 (Vision Research, Inc.) at 20 000 fps with an exposure time of 48 μs and an SC2500 thermography camera (FLIR Systems, Inc.) at 2000 fps with an exposure time of 6 μs.

Samples prepared with the two different printing methods were examined with a scanning transmission electron microscope (STEM, Titan 80–300 kV Environmental Transmission Electron Microscope) equipped with an X-ray energy dispersive spectroscopy (EDX) detector to compare localized mixing and the spatial distribution of the aluminum and copper (II) oxide particles. The inks were directly deposited on copper mesh carbon coated TEM grids and dried in air.

Representative images of the printed samples before and after ignition are shown in Fig. 4.

The ignition of the dual and single nozzle samples was qualitatively compared using a high speed camera as an initial proof of concept for the dual nozzle manufacturing technique. The ignition progression of a 5 layer sample prepared by dual nozzle printing is shown in Fig. 5(a). The corresponding ignition of a 5 layer sample prepared with pre-mixed ink using a single nozzle system is found in Fig. 5(b). These images suggested comparable reaction performance due to the brightness and time scale of the reaction. As such, the samples were tested further with infrared thermography and scanning transmission electron microscopy (STEM).

The maximum temperature reached by the nanothermite samples, as determined by high speed infrared thermography, is shown in Fig. 6. These measurements were taken assuming an emissivity of 1, and thus, the reported temperature is a lower bound. The data suggest that there is approximately a 200 K difference in maximum reaction temperature between the dual nozzle and single nozzle printed samples. Both types of samples reached peak temperatures above 1000 K. Samples with 5 and 7 layers had overlapping bounds of peak reaction temperature. This result confirms that there is reasonable reaction performance of the samples created with the dual nozzle system when compared to that of the samples printed using a single nozzle. When sufficient material is deposited, the two methods produce samples that are qualitatively similar. The dual nozzle technique results in samples that are slightly inferior, quantitatively, but still demonstrate acceptable relative performance for many applications. As such, this indicates that the dual nozzle system is a viable deposition technique that improves upon the safety of the single nozzle system without notably compromising reaction performance.

Representative STEM images of single nozzle samples [Fig. 7(a)] and sequentially deposited dual nozzle samples [Fig. 7(b)] and the corresponding EDX mapping from the insets [Figs. 7(c) and 7(d)] clearly show the contrast between the nCuO and nAl areas and that both methods are consistent with each other with respect to the particle distributions; these results are indicative of several identical samples tested at multiple locations on the sample. The single nozzle sample shows a mixture of agglomerates of nCuO and nAl, with length scales for both agglomerate sizes and mixing on the order of 100s–1000s of nanometers. It is probable that the resonant mixing method used for the pre-mixed ink is not completely effective in breaking up the agglomerates, which might limit the degree of mixing. On the other hand, the dual nozzle samples show a more layered structure. Interestingly, uniform layers of nCuO were observed to cover only the layer of nAl, possibly due to the preferential wetting of the nAl layer and subsequent drying. The thickness of the nCuO layer was about the size of the agglomerates, so the degree of mixing is quite similar to the single nozzle case. This is due to the minute volume of the individual droplets that only deposit a very small amount of nanoparticles. This indicates that a dual nozzle system is a viable safer alternative to a single nozzle system containing a premixed nanothermite solution. Uniquely, it can also be used to produce multilayered structures with tailorable characteristic thicknesses that can have different reaction rates by only modifying inkjet printing rates.

This work demonstrates the ability to perform in situ mixing of a fuel and an oxidizer to produce a nanothermite with notable energetic performance. Suitable inkjet printing of two colloidal suspension inks was achieved and used for the precise deposition of materials with picoliter-scale volumetric control. When sufficient material was deposited, there was no qualitative difference between the combinatorial printing method and traditional inkjet printing. Quantitatively, the dual nozzle samples produced slightly less heat than the single nozzle samples but still operated within acceptable ranges. By utilizing two part reactive printing, materials can be mixed and stored as inert components, which improves safety and shelf stability. In addition, the separation allows for ink tuning based on one key component rather than a mixture of constituents, as required with conventional inkjet printing. This method opens the door for integrated deposition and mixing techniques for delivering precise quantities of energetic materials without the added safety concern of handling energetic materials prior to deposition. It can also be easily translated to other multi-component materials that are incompatible with traditional inkjet printing.

The work presented herein provides a proof of concept for the use of reactive inkjet printing as a means to fabricate nanothermite samples. However, there are many avenues to still explore in this area of interest. For instance, there is a need to develop a technique for the bulk characterization of the samples in future work. It is acknowledged by the authors that the SEM imaging of a small component of a sample is not sufficient to definitively represent the bulk properties of the printed materials due to the significant size discrepancy. In addition, the deposition feature size is an avenue of future exploration, particularly as to how it relates to the overall energetic performance of the material. This initial proof of concept and testing provide ground work for the future development of this technique as it applies to energetic materials.

The authors would like to thank Christopher J. Morris for his helpful and constructive input. This research was supported in part by the U.S. Department of Defense, Defense Threat Reduction Agency, through Grant No. HDTRA1-15-1-0010, which is managed by Dr. Allen Dalton. Additional support (relating to the TEM experiments) was provided by the Young Investigator Program of the U.S. Department of Defense, Office of Naval Research, through Grant No. 108305. The content of the information does not necessarily reflect the position or the policy of the U.S. federal government, and no official endorsement should be inferred.