Surface properties of molds for powder injection molding and their effect on feedstock moldability and mold adhesion

The surface energy of various mold materials for low-pressure powder injection molding was evaluated using values of contact angles (Owens–Wendt method), and correlated with the feedstock moldability and mold adhesion. The surface tension of the binder used to formulate a metallic-based feedstock was also measured in the molten state at a typical injection temperature using the pendant drop technique. Real-scale injection tests were performed into metallic and polymeric mold cavities to assess the feedstock moldability and its adhesion with the mold surfaces that were compared with theoretical predictions obtained from the surface energies values. The results confirmed that the adhesion was significantly affected by the interfacial energy between the mold and the binder—in this case, the metallic mold exhibited low adhesion as compared to the polymeric mold. It was finally demonstrated that the adhesion phenomenon is only related to the surface properties of the mold (i.e., they are not related to the solidification rate)—in this case, a gold-coated polymeric mold produced the moldability of a polymeric mold and the adhesion properties of a metallic mold, which translated into a high moldability potential, with no resulting adhesion of the feedstock with the mold.


Introduction
Metal injection molding (MIM) is an advanced manufacturing process in which fine metallic particles are mixed with a polymeric binder to form a powder-binder mixture called feedstock, which is injected into a mold cavity to form a part (also called green part) with the required shape after solidification of the binder. This molded part then goes through the debinding and sintering processes to completely remove the binder and finally obtain an intricate near-net shape dense metallic part at low cost and with high service properties [1][2][3]. The MIM process can be divided into two branches according to the viscosity of the feedstock. High-pressure metal injection molding (HPIM) uses high-viscosity feedstocks (i.e., feedstocks falling within the 100-1000 Pa⋅s range) at temperatures up to 220 °C and pressures ranging from 50-150 MPa, while low-pressure powder injection molding (LPIM) was recently developed to inject low-viscosity feedstocks (i.e., feedstocks falling within the 0.1-20 Pa⋅s range) at temperatures below 120 °C and pressure below 1 MPa. Due to the low injection pressure used in the LPIM process, the size of the injection machines and the overall size of the tooling are significantly reduced compared with the HPIM process. The lower costs associated with this size reduction render the fabrication of intricate parts costeffective, either in low or in high production volumes [4,5]. Despite all these advantages, the LPIM process is still in its infancy, especially with regard to metallic feedstocks.
Efforts to optimize the LPIM process are typically focused on the debinding and sintering steps and how they can affect the mechanical properties of LPIM metallic materials [6][7][8][9], while studies aiming to optimize the moldability or demoldability of feedstocks are much less common. A better understanding of the fundamental mechanisms underlying LPIM including feedstock rheological and surface properties could help realize the full potential of the process. As an example of recent developments, the studies of Demers et. al. (2018) and Thavanayagam et al. (2015) on the rheological behavior of metallic-based feedstocks found that binder composition, solid loading, feedstock temperature, powder characteristics, and the shear rate applied on the feedstock are all key parameters driving the feedstock viscosity, and in turn its moldability [10,11]. Another fundamental area, which has been mostly overlooked, is the interaction between the feedstock and the mold surface, either in the molten state during injection or in the solid state after injection and solidification. Two main phenomena happening at the mold/feedstock interface influence the quality of parts in production: the mold filling during injection and the adhesion of feedstock with the mold surface after solidification. The mold filling is driven by the feedstock moldability, which in turn is controlled by the aforementioned rheological properties and the wetting ability of the molten feedstock, leading to feedstock spreading, sticking, solidifying, etc. on the mold cavity surfaces. Although the ability of a feedstock to be injected into the mold cavity is often quantified using the injected length of a given feedstock at different injection conditions, discretization of the mechanisms related only to the feedstock, only to the mold surface, or to the interaction between the feedstock and the mold have not received specific attention in the literature [12][13][14]. The adhesion of feedstock with the mold surface after solidification, which is a measure of the demoldability of the feedstock, is another practical parameter that can significantly affect part surface quality and production. In the HPIM process, adhesion magnitude and demoldability are typically assessed by measuring the amount of feedstock stuck on the mold surface after feedstock solidification [15], but even this method has not been used in LPIM studies.
Adhesion is related to wettability, which depends on mold surface chemical composition and roughness [16] that can be altered or managed by different surface modification techniques [17,18]. The range of applications that derive from these modifications is broad and most of the examples are not related to LPIM. Laser treatment, for instance, can be used on metallic surfaces to change roughness and control wettability and adhesion between substrate and water or ice [19][20][21][22], while plasma treatment can be employed to tailor the surface chemistry of composite tapes to increase adhesion with conductive inks [23]. In MIM, injection parameters and material properties such as mold temperature, injection temperature, feedstock coefficient of thermal expansion of feedstocks, and thermal conductivity of the mold may also alter the surface properties to ultimately increase or reduce the moldability/demoldability. Hemrick et al. [24] showed that when high injection temperatures were used, large feedstock shrinkage led to higher demoldability but more surface defects on the green parts, while lower injection temperatures led to higher surface quality but strong adhesion and low demoldability. There can be a trade-off given that in some cases the materials and parameters that lead to high moldability can also lead to strong adhesion and low demoldability [3]. In summary, during an injection, the feedstock performance depends on the raw material properties (mainly the feedstock viscosity), the mold characteristics (chemical nature of the material driving the thermal transfer and interface properties), and external conditions (mainly from the injection press controlling the injection pressure, the volumetric flow, the injection time, and the temperature). However, these parameters, especially the raw material properties and the mold characteristics, are rarely studied separately. The surface and interfacial energy of the different materials involved during an injection are properties that are rarely used to predict the equilibrium behavior of such feedstock/mold systems. On a flat solid surface, the surface energy is generally defined by the surface chemical composition, which is a function of the material and possible surface modification treatments [17,18,25]. Hausnerova et al. [26] presented one of the few studies that calculated the surface energy of feedstocks and molds with different surface treatments. Using a ceramic-based feedstock, they assessed and then maximized the adhesion between the mold and the feedstock in order to avoid wall slip during injection. However, this research group did not consider demoldability of the injected parts. The surface energy of a given solid surface can be calculated from contact angle measurements of different standard liquids with known surface tension on the solid substrate surfaces [26][27][28][29]. The surface tension of a liquid or molten polymer-based materials such as the LPIM binders can be obtained by analyzing the shape of a pendant drop [30]. Surface tension/energy values can then be used to calculate the work of adhesion between two materials [31]. It is still unclear whether the equilibrium conditions derived from surface energy measurements can be used to predict the moldability and adhesion behavior during or after the injection stage of the LPIM process. Therefore, the objective of this work is to study the role of surface properties on feedstock moldability and mold adhesion by discretizing the impact of mold characteristics. Optimizing these phenomena could significantly impact part production by increasing speed and reducing defects on molded parts.

Materials for the feedstock and molds
Water-atomized stainless steel 17-4PH powder (Epson Atmix Corporation, Japan), with a typical near-spherical or ligament shape and nominal particle size of 12 µm were used for the formulation of feedstock. This precipitationhardening stainless steel is widely used in the aerospace, chemical, petrochemical, and many other sectors for its high strength and good corrosion resistance. The dry powder was combined with molten binder (90 °C) in a laboratory mixer and blended for 1 h under vacuum. The solid loading was set at 60 vol. % of powder to prepare a feedstock from a low melting point binder system formulated from 34 vol. % of paraffin wax, 1 vol. % of stearic acid, and 5 vol. % of ethylene vinyl acetate. These binder constituents were selected due to their extensive use in LPIM, to help with the mold filling, to promote the surfactant effect enhancing chemical links between the powder and binder, and to produce the thickening effect needed to control the segregation of powder [32][33][34].
6.35 mm thick aluminum, copper, Inconel 625, polycarbonate (PC), polytetrafluoroethylene (PTFE) and steel plates were first used for the surface energy measurements as they represent potential candidates for LPIM mold materials. The extruded polycarbonate plates (100% Polycarbonate; ρ = 1.19 g/cm 3 ) were manufactured by Plaskolite, LLC. and purchased via McMaster-Carr. The molded polytetrafluoroethylene plates (ρ = 1.19 g/cm 3 ) were manufactured by Fluoro-Plastics, Inc and purchased via McMaster-Carr. The PTFE and steel plates were then selected for the moldability and adhesion tests. These interchangeable plates were prepared using two different surface finish conditions to investigate the effect of roughness on moldability and adhesion during injections. The first condition (the polished surface) was prepared by manual high-speed polishing using a 0.05 μm silica solution, while the second condition (rough surface) was obtained by shot peening. The arithmetical mean roughness values (Ra) resulting from polishing or shot peening operations were obtained from three measurements taken on each specimen at random positions using a profilometer (Mitutoyo Surftest SJ-400) with a 0.0001 μm minimum resolution over an 800 μm length. According to ASME B46.1 [35], the measuring speed, pin diameter, and pin top angle of the tool were 0.5 mm/sec, 2 μm, and 90°, respectively. The arithmetic average surface roughness for the PC, PTFE, gold-plated PTFE, and steel plates were Ra = 0.013, 0.98, 2.62, 0.67 μm, respectively. The moldability and adhesion tests were performed using a rectangular mold cavity formed by one support plate (steel), two thick base plates (steel), and two thin interchangeable plates (PTFE or steel), as illustrated in Fig. 1a. The rectangular cavity represented in Fig. 1b was used to assess the injected length values (as seen in Fig. 1c) for different feedstock temperatures and mold conditions. The feedstocks were injected at 70, 75, and 80 °C using a controlled constant volumetric flow of 1.15 cm 3 /s, leading to an injection pressure varying from 0.1 to 0.5 MPa until a sudden increase in pressure, indicating excessive friction with the mold walls or complete solidification. The mold temperatures were set at 40, 45, 50, and, 70 °C, and monitored using three thermocouples inserted into the mold as illustrated in Fig. 1a.

Measurements of the surface energy of the mold, surface tension of binder, and assessment of the feedstock adhesion
The surface energy of the six flat molds was calculated by the Owens and Wendt method [27][28][29]. Contact angles values measured using a goniometer (VCA Optima, AST Products, Inc.) formed by liquid droplets of water or diiodomethane deposited on each surface were used for that purpose. The measurement of the contact angle was repeated five times on each plate to calculate the surface energy of these six different materials, and was then used to identify two materials (i.e., steel and PTFE) exhibiting different surface properties to be used as interchangeable plates for the injection and adhesion tests. The surface tensions of the binder and its single constituents, namely, paraffin wax, stearic acid, and ethylene vinyl acetate alone, or combined as a binder, were calculated according to Daerr et al. [30], using the same goniometer and the molten pendant drop technique.
Since the binder constituents are solid at room temperature, a nozzle band heater combined with a plastic syringe was set up to produce a liquid pendant droplet at 70-75 °C, as illustrated in Fig. 2a-b. The mold and binder surface energies were then used to calculate the work of adhesion and interfacial energy. Finally, the measurement of the contact angle formed by direct binder deposition on hot mold was also investigated. The surface of the mold was heated up to  (Fig. 2c). After injection into steel or PTFE mold cavities, the adhesion or demoldability of feedstock was estimated using a sticking ratio calculated as the ratio of the feedstock remaining on the mold surface over the total injected length. This metric was categorized into four groups with the following distributions: 0% (no adhesion), 0-10% (light adhesion), 10-30% (moderate adhesion), and > 30% (high adhesion).

Surface properties of the molds
The surface energies and contact angle formed by drops of water on the different materials used for the molds are presented in Fig. 3. All metallic materials and one polymeric material (the PC) exhibited a hydrophilic behavior, with water contact angles less than 90°, leading to higher surface energy values. With a contact angle of around 105°, the PTFE plate could be considered as a hydrophobic material, resulting in the lowest surface energy obtained in this study. Due to their high and low surface energies, steel and PTFE were identified as good candidates for further realscale injection tests. Since high-strength steel or tool steel are often used for the fabrication of LPIM molds, tool steel was preferred over aluminum or Inconel even though it presented higher surface energy values.
The surface tensions of the binder and its single constituents calculated from the pendant drop technique are presented in Fig. 4. The surface tension of the binder is similar to that of pure paraffin wax, indicating that this constituent is more concentrated at the surface of the binder molten droplet. Also, paraffin wax, stearic acid, and ethylene vinyl acetate represent 85, 3, and 12 vol. % of the binder, respectively. Therefore, paraffin wax (i.e., the low surface tension component) is expected to migrate to the surface and reduce the overall surface tension of the system [36]. A comparison of the values of the surface energies obtained for the mold (Fig. 3) and the surface tension of the binder (Fig. 4) shows that the surface tension of the binder is about two-fold lower than the surface energy of the steel plate, but only slightly lower than the surface energy of the PTFE plate. This indicates that the binder will tend to wet both surfaces, but the wetting will be more favorable on steel since the difference between the surface tension of the binder and the surface energy of steel is greater than that between the surface tension of the binder and the surface energy of PTFE.
The analysis above considered the values of the surface tension of the binder and the surface energy of the mold separately, and did not take into account other mold surface characteristics such as chemical heterogeneity. In order to observe the binder spreading directly on the surface of the molds, contact angles formed by drops of molten binder on steel and PTFE interchangeable plates were evaluated, as illustrated in Fig. 5a-f. When the plates were at room temperature, the molten binder dropped on the metallic surface quickly formed a solidified droplet, resulting in a high contact angle value (Fig. 5a), while a liquid binder dropped over the polymeric surface takes a longer time to solidify, producing a lower contact angle (Fig. 5b). The solidification of the binder stops the liquid spreading regardless of the thermodynamic equilibrium state estimated by surface energy analysis. The difference in solidification time between the metallic and plastic molds can be explained by their differences in thermal conductivity, where k steel ≈ 10-50 W.m −1 [37,38]. To discretize the effect of solidification on the contact angle, the molds were heated at 70 °C (i.e., 10 °C higher than the melting point of the binder) to maintain the molten binder droplet in liquid state and evaluate its spreading over the mold surfaces after several time periods (Fig. 5c-f). Using this near-adiabatic condition (i.e., mold artificially maintained at 70 °C), the heat transfer between the binder and mold is hindered and the difference in thermal conductivity of the plates does not play a significant role on the final contact angle. Therefore, the difference in surface energies reported above may explain why the binder forms a low contact angle on the steel surface as compared to the PTFE surface. On the one hand, the thermal conductivity plays a predominant role when the difference in temperature between the binder and the mold is high (where the influence of the surface energy is low or simply absent, as represented by grey bars in Fig. 5g). On the other hand, the surface energy of the mold is the predominant mechanism for spreading when the difference in temperature between the binder and the mold is low (where the influence of thermal conductivity is low or simply absent, as represented by black bars in Fig. 5g). From a practical perspective, a slight heating of steel molds, usually at temperatures varying from 35 to 50 °C, is often used  to promote feedstock spreading and mold filling, while preventing feedstock adhesion. In summary, the results obtained using the sessile drop approach (illustrated by the black bars in Fig. 5g, where an average binder contact angle of 11 and 32° was obtained with steel and PTFE plates, respectively) corroborate the theoretical pendant drop approach (Figs. 3 and 4, with a significant difference in surface energy between the mold and the binder), finally predicting a higher spreading with the steel surface.

Moldability and adhesion of the feedstock
Moldability and adhesion tests were performed to validate whether the interfacial energy between the mold and binder can be used to predict the behavior of feedstock after real-scale injections. The moldability was quantified using the injected length over a rectangular mold cavity, and is reported in Fig. 6 as a function of mold temperature (40,45,50, and 70 °C), mold material (steel or PTFE), surface finish (polished or rough), and injection temperature (70, 75, and 80 °C). Note that a standard deviation of the experimental moldability measurements varying from values as low as 1 to 2.8 mm was calculated from the real-scale injections (not reported in Fig. 7 because the error bars are lower than the size of the marks). The results obtained with the PTFE plates show high moldability regardless of the surface finish, mold temperature, and feedstock temperature since the injected length reached the experimental limit (i.e., complete mold filling) for all injection conditions. For the steel plates, the results reported in Fig. 6 confirmed that an increase in injection temperature and/or mold temperature produces an increase in moldability generally explained by a decrease in viscosity associated with any increase in binder temperature and by a delay in feedstock solidification during injection.
It is interesting to note that the moldability obtained with a high feedstock temperature and steel mold temperature (i.e., T mold = 70 °C and T feedstock > 75 °C) was similar to that obtained with the PTFE plates. This result confirms that the solidification of the binder determines the final moldability of the feedstock, where a faster cooling rate leads to a decrease in the injected length. For similar mold cavity geometries, the rate of heat transfer for thermal conduction (Q/t) depends on the material thermal conductivity (k) and the difference in temperature between the feedstock and the mold (ΔT). Under the same injection condition (i.e., same cavity, injection temperature, mold temperature, volumetric flow, etc.), the high thermal conductivity of steel plates will produce a lower moldability as compared to the PTFE plates simply due to the fact that the steel plates have a faster binder solidification rate. Since PTFE is a poor conductor of heat, an injection using this material maintains the binder in the molten state for a longer time period as compared to steel plates. However, it is demonstrated below in this work that this high moldability potential is counterbalanced by a high adhesion of the feedstock (see Fig. 8). From a practical perspective, these two parameters must be evaluated simultaneously to determine an adequate combination of mold/ feedstock/injection parameters producing the highest moldability and the lowest adhesion of the feedstock. Figure 6 also shows a counterintuitive result, where an increase in mold surface roughness results in higher moldability (dashed vs. continuous lines), regardless of the mold temperature or feedstock temperature. According to the Wenzel wettability model, an increase in surface roughness may lead to higher binder spreading (lower binder contact angle) and a possible increase in moldability [39,40]. However, the increase in surface area associated with a rougher surface would also result in a higher contact surface between the mold and feedstock, leading to a higher solidification rate and lower moldability. Given the important role played by heat transfer, the most plausible explanation is that the dynamic occurring during an injection prevents the penetration of the feedstock into the grooves of the surface, which are rather filled with trapped air, producing an insulating layer which delays the feedstock solidification, and ultimately leads to an increase in moldability. Although the surface energy of the mold can be used to predict the spreading of feedstock at the mold interface, it does not seem to be a critical parameter for describing the moldability during the LPIM injections. In fact, the surface energy analysis and binder contact angle reported above in Figs. 2, 3, 4 and 5 predicted a higher moldability for steel (i.e., opposite to real-scale injections) without considering the solidification rate of the feedstocks related to the mold material.
The adhesion of the feedstock with the mold is also an important parameter describing the behavior of the injected part after its solidification. This phenomenon affects the demoldability (i.e., the capability to demold the part from the mold cavity), and consequently, the production of parts. The adhesion was quantified by the amount of feedstock stuck on the mold surface following the injection of molten feedstock, the solidification, and the demolding sequence. The injections were performed in the same rectangular mold cavity used previously for the moldability tests and the adhesion results are reported for different conditions as a map in Fig. 7a. In general, the feedstocks did not adhere to the steel plates in most mold and feedstock temperature conditions, while only a few combinations of high feedstock temperature (75 or 80 °C) with high mold temperature (45 °C) produced adhesion between the mold and the feedstock. Conversely, the feedstock fully adhered to the PTFE mold under all conditions, as illustrated by red color code in Fig. 7a. From a practical perspective, no adhesion is generally tolerated for the LPIM process, meaning that although they presented very high moldability, the PTFE plates could not really be used as is for the fabrication of a mold cavity. Given the importance of adhesion during the LPIM injection stage, the work of adhesion and interfacial energy at the mold/binder interface were used to better understand the mechanism of adhesion. The work of adhesion W, or, in other words, the work required to separate two adhered surfaces, is a measurement of the contact strength between these two phases -the binder and plate -and can be estimated by Eq. (1): where γ 12 represents the interfacial energy between phases 1 and 2 calculated with the harmonic mean equation (Eq. 2) based on the surface energy values γ 1 and γ 2 calculated above in Figs [27,40].
The work of adhesion and interfacial energy for both the binder/steel and binder/PTFE contacts are reported in Fig. 8. Similar work of adhesion calculated for both the binder/steel and binder/PTFE contacts confirms that this metric cannot be used directly to correlate the experimental adhesion results obtained in Fig. 7a. However, the interfacial energies calculated for the binder/PTFE and binder/ steel interfaces were significantly different at 1 and 6 mJ/m 2 , respectively. The interfacial energy indicates the compatibility between the two materials composing the new interface, and the adhesion between the two phases increases when this parameter decreases. In many industrial sectors involving adhesives and bonding, a low value of the interfacial energy is often sought to achieve higher adhesion strength and durability of the assemblies [41][42][43]. In this respect, the high and low values of the interfacial energy measured respectively for the binder/steel and binder/PTFE interfaces can be used to predict the level of incompatibility or compatibility, which will contribute to an easier or difficult separation of the feedstock from the mold. In other words, a high interfacial energy between the feedstock and the mold is expected for the LPIM process in order to promote the demolding/ ejection of the injected parts. From a practical perspective, this experimental tool could be used to predict the adhesion behavior of new feedstocks (i.e., new binder recipes) and new mold materials.
To confirm that the adhesion of feedstock is mainly related to the interfacial energy, and not to the solidification rate, PTFE plates were gold-coated using sputtering coating, as illustrated in Fig. 7b-c. Since the gold layer is very thin as compared to the PTFE plates (t PTFE = 6.35 mm vs. t gold ≈ 5 × 10 −5 mm), these coated plates should have similar thermal behaviors as the uncoated PTFE plates. The visual result presented in Fig. 7c, as well as the adhesion map reported in Fig. 7a for different conditions, show that the injected parts were easily removed from the goldcoated section of the mold, but not from the uncoated section of the PTFE mold (as previously seen from the red color code in the adhesion map in Fig. 7a). In other words, the significant increase in interfacial energy achieved by the gold coating on PTFE plates (Fig. 8) drastically changes the adhesion behavior of the feedstock, even for a bulk material producing a low solidification rate. Since the uncoated PTFE produces high moldability and high adhesion, and steel produces low adhesion but only moderate moldability, this bi-material mold could be envisaged to fabricate steel molds with some gold-coated PTFE inserts placed in critical sections of the cavity to locally improve the moldability without a negative impact on the mold adhesion. This new tooling potential for the LPIM process involving coated PTFE was recently confirmed using two spiral shaped molds one made of steel and another one made of PTFE coated with a chromium layer (PVD process producing a coating of about 150 nm). It was shown that the chromium-plated PTFE mold produces an injected length three times greater than that produced with a steel mold [44].

Conclusion
In this work, the surface properties of different mold materials (steel and PTFE) and the interfacial energy between binder and mold used in the LPIM process were evaluated. Moldability and adhesion tests were performed in different conditions (mold surface finishes, mold temperatures, binder temperatures) using real-scale injections and correlated with the theoretical predictions based on the surface energies of the molds and interfacial energies of the binder/mold interfaces. The results are summarized as follows: • The moldability of a feedstock was not affected by the surface energy of the mold, but mainly driven by the solidification rate of the feedstock. Since the PTFE material is a poor heat conductor, this mold produced high moldability regardless of the surface finish, the mold temperature, and the feedstock temperature. For the steel mold, an increase in the injection temperature and/or in the mold temperature produced an increase in moldability due to a decrease in feedstock viscosity and a delay in feedstock solidification associated with any increase in the binder temperature. • An increase in the mold surface roughness resulted in higher moldability with the metallic mold, regardless of the mold temperature or the feedstock temperature. This counterintuitive result was probably related to a change in the solidification rate of the feedstock (e.g., potentially explained by air trapped in the grooves of the mold surface, producing an insulating layer at the mold/binder interface). • The adhesion of feedstock was directly influenced by the interfacial energy between the mold and the binder. A low interfacial energy indicates a high level of compatibility between the binder and the mold, which could be used to predict difficulties in removing the part from the mold cavity after its injection and solidification. In this respect, the high and low interfacial energies measured respectively with the steel and PTFE molds were correlated to the non-adherence or full adherence behavior of the feedstock. For the steel mold, no adherence of the feedstock with the mold was visible for all typical injection conditions because of the high interfacial energy resulting in a less stable bond. Since the adhesion affects the demoldability of the injected parts, the high moldability potential obtained with the PTFE mold was significantly counterbalanced by its high feedstock adhesion. • The adhesion phenomenon is not affected by the solidification rate, but rather, is only related to the surface properties of the mold. The surface modification of the PTFE mold using a gold-coating produced high moldability (i.e., moldability of a polymeric mold) and no adhesion of the feedstock with the mold (i.e., adhesion properties of a metallic mold).