To study the effect of temperature gradient on solidification structures, different furnace translation velocities were applied. A banding structure is observed and dendritic/spear-like symmetric dendritic features are also observed throughout the microstructure. Primary dendrite arm measurements of naphthalene were obtained from stitched cross-section images taken perpendicular to the freezing direction. Dendrite thickness is expected to increase with decreasing furnace translation velocity; this relationship is observed for all regions. For all velocities, dendrite widths are largest in the outer region of samples and decrease with decreasing distance from the center (e.g., for samples solidified at V = 6.5 µm·s^(-1), dendrite width at the outer and inner regions are 87±11 and 40±9 µm, respectively; for V = 80 µm·s^(-1), these values decrease to 60±14 and 17±4 µm, respectively). The relationship we observe here, with increasing dendrite width at outer regions of samples relative to inner regions, indicates that the local solidification velocity is higher in the central region of the sample relative to the outer region. The corresponding macroscopic convection pattern is likely characteristic of a convex macroscopic interface. Interdendritic convective fluid flow is also likely present, given the observation of asymmetric dendritic features.

2. Freeze-thaw studies

As flight ampoules will be filled on the ground and transferred to the international Space Station (ISS) in a solidified state, samples will need to be melted back immediately prior to experimental tests, and suspension stability cannot degrade appreciably due to that freeze-thaw cycle. We conducted freeze-thaw studies using four naphthalene/Cu/surfactant systems to assess suspension destabilization after freeze/thaw. Within the four surfactant types tested, two stabilization mechanisms were studied: steric stabilization and electrosteric stabilization.

For suspensions employing low dielectric constant fluids (e.g., naphthalene), steric stabilization is the most commonly employed stabilization method; polymers adsorbed on particles create a hindrance for particle aggregation. Aggregation that does occur when particles overcome steric repulsion is thought to be reversible. Three of the surfactants we tested offered steric stabilization; two of the three (Triton X-100 and Pluronic F-68) produced suspensions that were inconsistently stable even prior to freeze-thaw testing, likely due to thermal degradation of the polymer due to the high melting temperature of naphthalene. The third steric stabilization surfactant, Hypermer KD-13, produced stable suspensions that, however, destabilized after freeze-thaw.

AOT is an ionic surfactant that is used for electrosteric suspension stabilization. Electrosteric stabilization in low-dielectric constant fluids offers increased stabilization relative to steric stabilization alone. In this case, we have a non-polar fluid, so the polar head of the surfactant molecule adsorbs to the particle surface (with help from the counterion, which creates a bridge). At the critical micelle concentration, reverse micelles that are charge-stabilized are formed. With this system, we have been able to reliably create stable naphthalene/particle suspensions which, as shown previously, produce anisotropic, directional microstructures upon solidification. Moreover, we did not observe any evidence of suspension destabilization after freeze-thaw. However, tests were conducted after subjecting suspensions to a single freeze-thaw cycle; additional cycles may cause degradation not described in the following. The suspension must remain in the liquid state during various stages of ampoule filling. Also, science requirement 19 specifies a storage transport requirement for suspensions prior to flight testing at <80°C to prevent an uncontrolled freeze-thaw cycle.

To assess their freeze-thaw stability, suspensions with 5 vol.% Cu and 1 wt.% AOT (with respect to Cu) were prepared as described previously. Prepared suspensions were subjected to an initial freeze in an ultrasonic bath using bath temperatures of either ~10 or ~50°C, representing slow (FTS) and fast (FTF) freeze-thaw testing, respectively. Solidified suspensions were melted immediately prior to directional solidification. In an attempt to mitigate the risk of particle aggregation during freezing, the sonication step was utilized to promote disordered growth during the initial freeze step, which should reduce the propensity for particle aggregation (Science Requirement 10).

Primary dendrite arm measurements of naphthalene were obtained from stitched cross-section images taken perpendicular to the freezing direction, as described previously. As observed previously, dendrite width increases with increasing radial distance from the center. Within a given region, values of mean dendrite width are in relatively good agreement among the freeze-thaw conditions. For the outer region, these values are 60±14, 54±8, and 60±11 for the no-freeze-thaw (STD), freeze-thaw-fast (FTF), and freeze-thaw-slow (FTS) conditions, respectively. For the inner regions, values for STD, FTF, and FTS are 17±4, 21±6, and 20±5, respectively.

3. Pre-Science Requirements Review (Pre-SRR) PFMI furnace testing

3.1 Sample Ampoule Assembly Microgravity solidification experiments are intended for the Pore Formation Mobility Investigation (PFMI). The PFMI furnace allows for a high gradient of 50 °C/cm, low gradient of 10°C/cm, and nominal melt-back and growth velocities of 100 and 1 µm/s, respectively. The Sample Ampoule Assembly (SSA), as designed for the previous PFMI work, was used. Techshot is designing the SSA for this work to be similar in construction to those designed previously, components of which include: the main tube body (composed of Schott 8250 borosilicate glass), stainless steel (SS) spring, Kovar piston assembly, cartridge mount head, and six in-situ thermocouples. The inside and outside diameter of the ampoule is ~10.9 and 12.75 mm, respectively. To ensure similar thermal behavior with respect to the glass, Science Requirement 3 specifies an internal diameter tolerance of ±0.2 mm. Based on toxicology assessments conducted by Techshot, the maximum volume of naphthalene that can be used in an individual ampoule is ~6.8 mL. Accordingly, ampoules will be filled with ~5.6 mL of science material, corresponding to an effective science material length of ~6 cm which is shorter than the overall length of the ampoules (~28 cm, not including the piston assembly or insertion of the Kovar head). The first assembled ampoule #1 had an issue with the Kovar piston motion during pumping, which bent thermocouples from designed position which would have affected measurements of temperature gradients (it was thus not used). Ampoules #2 and #3 were assembled with stronger springs to retard the movement of the piston. In addition, a magnetic frame equipped with 4 high strength magnets is also utilized to constrain the piston during naphthalene filling. It was successful, as the piston only moved less than 2 mm under high vacuum after pumping down.

3.2 Naphthalene remelting testing To eliminate bubble formation in solidified naphthalene, induced by air leakage during ampoule filling, the gas lines of the filling station were upgraded using stainless steel pipes, Pirani pressure gauge and Swagelok connectors. A vacuum of 7x10-2 Torr was reached in gas lines after pumping for 40 mins with a roughing pump, indicating good resistance to leaks. As-received naphthalene was distilled first with this upgraded setup.

To test performance and stability of naphthalene melting with the PFMI furnace, sample ampoule #3 with embedded thermocouples was filled with purified, distilled naphthalene at Northwestern University. Filled ampoules were remelted and directionally solidified with different processing parameters in the PFMI furnace at TechShot. Naphthalene solidification was performed with two batches of processing parameters to induce different solidification morphologies, i.e. dendrites and bands, at solid-liquid interface. High furnace translation velocity, V = 100 µm ·s^(-1), and shallow temperature gradient, G = 10 K ·cm^(-1), result in dendritic morphologies. In contrast, the low V =1 µm ·s^(-1) and steep G = 50 K ·cm^(-1) expect to induce bands morphologies during solidification.

The filled ampoule was placed vertically in the PFMI furnace (at Techshot) and heated up to 87 ºC to melt the naphthalene prior to solidification. The cold zone was moved to solidify liquid naphthalene in the molten zone with a defined translation velocity. The temperature as a function of time was measured by 6 thermocouples during solidification with translation velocities of 100 µm/s and 1 µm/s. The solidification of naphthalene was recorded as a plateau (from latent heat release) in temperature profiles during furnace translation.

Temperature gradients in liquid naphthalene during solidification are also calculated from each temperature profile measured by six thermocouples. Cooling rate (k, K/s) was first calculated from the slope of temperature profile, and then divided by furnace translation velocity (V, cm/s) to achieve the temperature gradient (G, K/cm). With a translation velocity of V=100 µm/s, calculated temperature gradient has an average value of 8.4 K/cm, close to the designed value of 10 K/cm. In the case of translation velocity of V=1 µm/s, measured temperature gradients of 26 K/cm are smaller than 50 K/cm. This could be solved by adjusting the distance between electrode 1 and the cool zone.

Optical micrographs showing the microstructure of copper freeze-cast materials obtained by directionally solidifying 5 vol.% Cu particles suspended in naphthalene that contained 0.4 wt.% dissolved, AOT surfactant (1 wt.% AOT with respect to Cu) were provided in our annual technical report to NASA; structures were solidified using a furnace translation velocity, V = 80 µm·s^(-1) and thermal gradient, G = 35°C·cm^(-1). Plate-like microstructures are observed for all samples. The microstructure predicted by You et al. for the solidification condition is “spears” (i.e., presence of significant dendritic features that extend into neighboring dendrites, producing a cellular-type structure after sublimation of the solidified fluid). [Ed. Note: See Reference.] The images show that, while significant dendritic features are present, a wall-type structure is retained.

Reference: You J, Wang Z, Worster MG. Controls on microstructural features during solidification of colloidal suspensions. Acta Materialia. 2018 Sep 15;157:288-97. https://doi.org/10.1016/j.actamat.2018.05.049

Succinonitrile (NC(CH2)2CN; SCN) was chosen as the suspending fluid for freeze-casting test suspensions due to: (i) known compatibility with the PFMI apparatus [Grugel et al., 2012], (ii) ease of sample transport (the melting point of SCN is ~58°C; thus, transport of test suspensions and solidified samples requires minimal environment control), and (iii) system simplification. It was determined that simplifying the system to the largest possible extent would offer the greatest degree of fundamental knowledge necessary to improve the understanding of microstructural formation and would also offer the opportunity to validate and improve existing freeze-casting models. This fundamental basis shall provide a basis upon which future microgravity work can build. Unlike water, which is the most-often utilized fluid in freeze-casting studies, SCN exhibits a linear temperature-density relationship within the temperature range of interest; thus, a density inversion during solidification is avoided.

SCN has not been reported as a suspending fluid for use in freeze-casting suspensions systems. Previous research has shown that anisotropic solidification behavior of suspending fluids is a necessary, but insufficient criterion, for attaining directional pore structures for particle-based suspension systems [Naviroj et al., 2017]. We conducted preliminary tests to verify the feasibility of attaining directional microstructures using SCN-based particle suspensions. Directional microstructures were confirmed via scanning electron microscopy investigation of the fractured surface of a titanium/SCN freeze-cast structure where 20 vol.% titanium particles (20 µm size) was suspended in molten SCN and solidified under the presence of a thermal gradient.

There are two main limitations of our preliminary demonstration, including: (i) constant cooling was utilized and the solidified SCN was not sublimated from the sample and sintering steps were not carried out. For the former issue, one side of the molten SCN suspension was cooled using a constant cold plate temperature (~20°C) while the other side was held at a constant warmer temperature (~65°C), whereas controlled cooling will be utilized during experimental operations as it provides greater control over microstructures templated. With regard to the second issue, our previous freeze-casting projects have mainly utilized water as the suspending fluid. In such cases, frozen samples are sublimated using a conventional freeze-dryer. The physical properties and toxicity of SCN necessitate the development of a new sublimation procedure. Progress toward completing these tasks is described below.

As SCN is solid at room temperature, controlled cooling requires temperature control of both the top and the bottom of the suspension. We designed a sample container system to conduct these initial controlled cooling tests. The container consists of a Teflon tube (low thermal conductivity) that is covered using top and bottom aluminum plates (high thermal conductivity). Temperature sensors are placed perpendicular to the heat flow inside each of the aluminum plates to measure and control (through feedback loop) the temperature gradient during solidification. Resistance heaters, consisting of nichrome wire wrapped in Kapton (polyimide) tape, were placed on each aluminum plate and are operated using a microcontroller.

A basic solidification program has been developed; the loop function involves reading the voltages of the temperature sensors, performing temperature conversions, comparing actual temperature values for each side of the suspension to target temperature values, and independently controlling each heater depending on the aforementioned comparison. Target temperatures are calculated as a function of time using an exponential cooling profile in an attempt to produce a constant solidification velocity during freezing experiments. The temperature of the solidification interface is assumed to be equal to the melting point of pure succinonitrile and an exponential cooling function determine the target temperature of the cold plate as a function of time. The target temperature of the hot side is determined based on a pre-defined temperature gradient which is held constant throughout solidification.

The first stage of the program involves heating the aluminum plates to a starting temperature. During such time, the SCN/particle suspension is kept several degrees warmer than the melting temperature using a hot plate and magnetic stirrer. After the initial temperature is reached for both the hot and cold plates, the plates are kept at this temperature for five minutes during which time, the suspension is loaded into the Teflon mold. Thereafter, the hot plate is kept at a constant temperature and the temperature of the cold plate is brought down to the melting temperature of SCN using a linear cooling rate (here, the rate is set at 5°C/min). Once the cold plate reaches the starting temperature, the exponential cooling function is used to determine the target temperature for the cold plate; during this time, the target temperature of the hot plate is determined by the pre-determined temperature gradient (the temperature difference between the solidification interface and the hot-plate temperature divided by the height of the molten region). The lowest attainable temperature using this apparatus is room temperature; thus, the solidification experiment is completed after the target cold plate temperature reaches room temperature. The last stage of the program is “thermal hold” wherein the temperatures of the plates are held at a predetermined temperature for a predetermined period of time before demolding; this stage ensures consistency for post-solidification sample handling.

We previously reported the use of a plastic vacuum desiccator connected to an Edwards RV8 6.9 CFM dual-stage vacuum pump as a vacuum chamber. This design allowed for very heating of the chamber environment (~5°C above ambient) and resulted in extremely slow sublimation times. We re-designed the sublimation apparatus to provide better temperature control of the vacuum chamber. Samples are placed inside a metal vacuum chamber and the temperature of the chamber is controlled via a heating pad and PID temperature controller. A condenser, responsible for collecting and condensing SCN vapor, is connected to the vacuum chamber and cooled using an ice water and pump system. Cooling the condenser with ice water required numerous ice changes during sublimation; thus, we further redesigned this portion of the apparatus by adding a copper chiller that utilizes a cold-water supply source in the lab. The copper chiller is housed in a Styrofoam cooler. The sublimation time for a sample containing ~1.5 g is ~50 h; the time can be reduced by increasing the temperature of the vacuum chamber, however, doing so will increase the propensity for drying cracks to develop; thus, this is a parameter that requires further optimization. After ~80% mass loss, the SCN/CuO sample collapsed during sublimation. Similar tests were repeated and resulted in the same issue. In these cases, some portions of the specimen, especially those corresponding to peripheral regions, can be recovered. Our initial tests were conducted without the use of a binder in an effort to minimize the influence of suspension additives during solidification. We are in the process of testing samples with binder and higher particle volume fractions to mitigate this issue (while still attempting to minimize the effect of suspension additives).

To determine the safe storage requirements of post-experiment (solidified) samples, a series of thin sample studies were performed. This component of the testing was essential because the samples could be exposed to a variety of temperatures and conditions while traveling back from the station and will be stored for an indeterminate amount of time before being sent back for analysis. During that time, if any structural changes occur, microstructures of the samples would no longer be an accurate representation of that which was templated during solidification. Thin samples were used in the place of bulk samples for these tests because the microstructure of thin samples is immediately visible without carrying out post-solidification processing steps (i.e., sublimation and sintering). This distinction allows for the samples to be imaged immediately after solidification so that those images can be compared with ones taken of the same areas of the sample samples at designated time intervals.

A first round of testing was performed using a heating system involving the transmission of heat through nichrome wire wrapped around two aluminum blocks. Two glass slides held apart with a spacer (tape, 2 mm in width) are loaded between the blocks and filled with suspensions containing 5 vol.% CuO particles dispersed in liquid SCN. Samples were solidified using the same temperature profile and solidification program that was used for the bulk samples (described above).

After solidification, thin samples were imaged using a stereoscope once every day for one month. From these images, it was concluded that the microstructure became slightly hazy over time not because of actual structural changes, but because the specimen pulled away from the glass slide (likely due to post-solidification SCN shrinkage). This hypothesis is further supported by the presence of air bubbles in the top corned of the image taken at the one-month mark. These are air bubbles were not present in the image taken immediately after solidification, meaning that there was more space between the glass and the sample for the bubble to insert itself into after the waiting period.

The stereoscope that was used for imaging also provides a high depth of field, which makes focusing on the same area of the sample more difficult when that area begins pulling away from the glass slide (thus, increasing the object distance between imaging days). Thus, the distortion of the features is not significant enough to indicate that they themselves have actually changed, but rather that the methods used to distinguish them are less effective due to a poorer ability to fully resolve all of the same features over time. When comparing the larger highlighted dendritic section immediately after solidification with that same section after one month of storage at room temperature, a significant difference in dendrite width or spacing was not detected.

A second set of tests was used to produce finer microstructures (which should exhibit a higher propensity for microstructural rearrangement after solidification). For these tests, the cold-side aluminum block was refrigerated until it reached 0°C while the slides were heated to 100°C. The length of the slide was decreased such that the sample could be loaded without solidifying prematurely. As expected, the dendrite structure was finer than those observed for the slower solidification tests reported above. Although these finer structures are more difficult to compare due to magnification limitations of the stereoscope, we were not able to observe significant post-solidification processing changes to the microstructures. The same cloudiness and distortion of the image can be observed that was hypothesized earlier to be caused by separation of the sample from the glass but, otherwise, the dendritic structure itself appears to remain unchanged. As we are currently working on modifying the suspension characteristics, these tests will need to be repeated after the suspension preparation process is finalized.

References

Scotti KL, Dunand DC. "Freeze casting – A review of processing, microstructure and properties via the open data repository, FreezeCasting.net." Progress in Materials Science. 2018 May;94:243-305. [See also Cumulative Bibliography listing below]

R.N. Grugel, L.N. Brush, A.V. Anilkumar, Disruption of an aligned dendritic network by bubbles during re-melting in a microgravity environment, Microgravity Sci. Technol. 24(2) (2012) 93-101.

M. Naviroj, P. Voorhees, K. Faber, Suspension-and solution-based freeze casting for porous ceramics, Journal of Materials Research (2017) 1-11.

This research represents the first microgravity study of quasi-steady state solidification behavior in the freeze-casting process. Given the wide range of typical processing parameters and great number of research-worthy questions that remain unanswered about the technique, an exhaustive literature review was conducted to aid in experiment design. Data linking processing conditions to microstructural characteristics and mechanical properties were extracted from ~900 freeze-casting papers and a systematic analysis of these data was conducted. In accordance with the aim of this program, we created a public freeze-casting data repository ( http://www.freezecasting.net/ ) in an effort to facilitate broad dissemination of relevant data to freeze-casting researchers, promote better informed experimental design, and encourage modeling efforts that relate processing conditions to microstructure formation and material properties. A description of the resulting SQL database/website and results of our analysis were published in a review article in Progress in Materials Science [see Bibliography section]. Typical processing parameters that have been identified will be utilized during experiment design to ensure maximum generalizability of these results. Experimental data from the database will also be utilized to test models developed during this project.

Succinonitrile (NC(CH2)2CN; SCN) was chosen as the suspending fluid for freeze-casting test suspensions due to: (i) known compatibility with the Pore Formation and Mobility Investigation (PFMI) apparatus [Grugel et al., 2012], (ii) ease of sample transport (the melting point of SCN is ~58°C; thus, transport of test suspensions and solidified samples requires minimal environment control), and (iii) system simplification. It was determined that simplifying the system to the largest possible extent would offer the greatest degree of fundamental knowledge necessary to improve the understanding of microstructural formation and would also offer the opportunity to validate and improve existing freeze-casting models. This fundamental basis shall provide a basis upon which future microgravity work can build. Unlike water, which is the most-often utilized fluid in freeze-casting studies, SCN exhibits a linear temperature-density relationship within the temperature range of interest; thus, a density inversion during solidification is avoided.

SCN has not been reported as a suspending fluid for use in freeze-casting suspensions systems. Previous research has shown that anisotropic solidification behavior of suspending fluids is a necessary, but insufficient criterion, for attaining directional pore structures for particle-based suspension systems [Naviroj et al., 2017]. We conducted preliminary tests to verify the feasibility of attaining directional microstructures using SCN-based particle suspensions. Directional microstructures were confirmed via scanning electron microscopy investigation of the fractured surface of a titanium/SCN freeze-cast structure where 20 vol.% titanium particles (20 µm size) was suspended in molten SCN and solidified under the presence of a thermal gradient.

There are two main limitations of our preliminary demonstration, including: (i) constant cooling was utilized and the solidified SCN was not sublimated from the sample and sintering steps were not carried out. For the former issue, one side of the molten SCN suspension was cooled using a constant cold plate temperature (~20°C) while the other side was held at a constant warmer temperature (~65°C), whereas controlled cooling will be utilized during experimental operations as it provides greater control over microstructures templated. With regard to the second issue, our previous freeze-casting projects have mainly utilized water as the suspending fluid. In such cases, frozen samples are sublimated using a conventional freeze-dryer. The physical properties and toxicity of SCN necessitate the development of a new sublimation procedure. Progress toward completing these tasks is described below.

We developed an apparatus that will enable controlled solidification testing and a sublimation apparatus suitable for sublimating SCN from solidified samples. Testing of these systems will be conducted over the next two months.

References

R.N. Grugel, L.N. Brush, A.V. Anilkumar, Disruption of an aligned dendritic network by bubbles during re-melting in a microgravity environment, Microgravity Sci. Technol. 24(2) (2012) 93-101.

M. Naviroj, P. Voorhees, K. Faber, Suspension-and solution-based freeze casting for porous ceramics, Journal of Materials Research (2017) 1-11.