- Email: [email protected]
Overcoming geometric limitations in metallic glasses through stretch blow molding
Applied Materials Today 19 (2020) 100567
Contents lists available at ScienceDirect
Applied Materials Today journal homepage: www.elsevier.com/locate/apmt
Overcoming geometric limitations in metallic glasses through stretch blow molding Rodrigo Miguel Ojeda Mota a , Naijia Liu a , Sebastian Alexander Kube a , John Chay b , Hayley D. McClintock a , Jan Schroers a,∗ a b
Department of Mechanical Engineering & Materials Science, Yale University, New Haven, CT 06511, USA Supercool Metals, New Haven, CT, USA
a r t i c l e
i n f o
Article history: Received 27 August 2019 Received in revised form 12 January 2020 Accepted 14 January 2020 Keywords: Bulk metallic glass Thermoplastic forming Stretch blow molding Superplastic forming
a b s t r a c t Bulk metallic glasses (BMGs) exhibit remarkable mechanical properties, such as high strength and elasticity, which is often paired with fracture toughness. Their supercooled liquid region gives rise to plastic-like processing and suggests parts and shapes that can otherwise not be obtained for crystalline metals. However, current processing techniques only allow for limited options in terms of geometry, thicknesses uniformity, and shape complexity. Here we introduce a new processing technique, “stretch blow molding,” to expand the range of possible parts and increase the available geometries that can be fabricated with BMGs. Additionally, a model is derived that allows for the quantification and prediction of stretch blow molding and provides insight into its potential use and limitations. We demonstrate that with stretch blow molding overall strains exceeding 2000% are achievable, compared to the previously reported ∼150% of blow molding. With the ability to stretch blow mold shapes that were previously unachievable with any other metal fabrication technique in a fast and economical manner, and the superb properties of BMGs, we look forward to a broad commercial adaptation of this technique. © 2020 Published by Elsevier Ltd.
1. Introduction Open shape metal articles are in high demand for use in a wide range of applications including containers, casings, housings, covers, tubing, hoses, and bellows. The defining characteristic of these shapes is that their wall thickness is significantly thinner than their other dimensions. In contrast, closed shape articles exhibit similar size in all dimensions. While a wide range of processes exists for the fabrication of closed shape articles, including forging, casting, and stamping, processes to fabricate open shape metal parts are limited to high symmetry shapes, which are generally difficult to realize in a metal. Additive manufacturing alleviates the challenge to some extent; however, closed shape articles are much easier to fabricate than open shape metal articles with a thin wall thickness [1–4]. With the rise of super plastic forming (SPF), some limitations in fabricating open shape metal articles have been overcome in relation to size, thickness, and usable metals [5–8]. Extensive research
∗ Corresponding author. E-mail address: [email protected] (J. Schroers). https://doi.org/10.1016/j.apmt.2020.100567 2352-9407/© 2020 Published by Elsevier Ltd.
has concluded, however, that SPF suffers from material thinning and is limited to relatively low achievable strain rates [9]. Open shape articles can be readily achieved in thermoplastics [10]. One example is extrusion blow molding, which can fabricate high-strain parts within seconds, as demonstrated by the canonical example of the soda bottle. The reason for such ease is that thermoplastics exist in a viscosity range on the order of 102 –105 Pa·s, where they can be readily deformed under gas pressure. Such a viscosity range is generally not accessible in metals, which was suggested as the reason for the asymmetry in the difficulty of fabricating metals vs thermoplastics [11]. A metallic materials class that provides access to such a viscosity range is BMGs [12–14]. Their so-called supercooled liquid region has been widely explored for novel processing methods [15–21]. One such method is blow molding [11,22,23], which enables the molding of thin wall open shape BMG articles at low pressures (∼1 atm) and over short times (∼10 s) [11]. Here, the limiting factor is geometrical thinning, which restricts this process to relatively low strains, <150% when using sheet-like feedstock [22]. One strategy to overcome geometry limitations due to geometrical thinning is the use of a pre-shaped parison [11]. Though effective, this method requires an additional processing step to manufacture the parison, which increases cost and complexity, while consuming the
2
R.M. Ojeda Mota, N. Liu, S.A. Kube et al. / Applied Materials Today 19 (2020) 100567
time available in the supercooled liquid region before the onset of crystallization (thermal budget) [24]. Here, we overcome the limitation of the achievable shapes due to geometrical thinning by first stretching a sheet-like feedstock into the mold cavity, which we subsequently blow mold. The stretching step creates a basis pre-shape of the final article. To increase the uniformity of the stretched pre-shape, we locally vary temperature and the geometry of the parts used for stretching. This allows us to fabricate complex shapes from a sheet feedstock with overall strains exceeding 2000%. As much higher forces can be readily applied during stretching (up to 100 MPa) than during blow molding (up to 1 MPa), stretch blow molding can be carried out much faster than blow molding alone, hence a shorter cycle time in a commercial process can be achieved and crystallization can be readily circumvented.
2. Method and materials Fabricating open shape BMG parts with TPF based processing is challenging when using techniques like compression molding. This is due to the fact that during TPF, the BMG is subject to perfect stick conditions (zero velocity at BMG/mold interface) [25]. The resulting creep-flow exhibits a scaling of forming pressure, p, with wall thickness, d, of p = vL/d2 (v: velocity, h: viscosity, L: filling depth). Such scaling suggests that compression molding becomes rapidly more difficult when forming articles with thin walls [12,26]. An approach to drastically reduce the required shear strain and shear stress rates for the same geometry is the use of processes where the BMG is, for the most part, not in contact with the mold during the shaping operation [22]. One of these processes is blow molding, where low pressures (∼1 atm) and short times (∼10 s) have been demonstrated to suffice for the fabrication of thin wall open shape articles [27]. Furthermore, material thinning, reflected in the strain rate exponent of unity (m = d/dε˙ = 1), is essentially absent during blow molding of BMGs [28–30]. However, geometrical thinning limits this process to relatively low strain when using sheet-like feedstock. This issue, present in essentially all blowmolding processes, stems from the boundary conditions of the BMG to the mold; once the BMG touches the mold, it no longer deforms and the remaining deformation must be accomplished only with the BMG that has not yet touched the mold [22]. Thinning is particularly problematic for parts with high aspect ratios (Fig. 1A) and has therefore limited blow molding to BMG parts with a small overall strain. One strategy to overcome geometry limitations due to geometrical thinning is the use of pre-shaped parisons (Fig. 1B). When using parisons, which resemble the final shape of the article, all parts of the mold are filled at a similar time after a similar amount of strain. Hence, thinning is eliminated and the majority of deformation should be carried out without any contact to the mold. Though effective in realizing complex shapes that require high strains, the required additional processing step to fabricate the parison enhances costs, complexity of the process, and consumes thermal budget. In this work, we overcome the limitation imposed by geometrical thinning and expand the range of achievable shapes that can be realized with BMGs by first stretching a sheet-like BMG feedstock into the mold cavity. Subsequently, but in the same heating cycle, the stretched feedstock is blow molded against the mold, assuming and replicating its shape (Fig. 1C). The stretching step creates a pre-shape which can be designed to resemble aspects of the final shape such as length, aspect ratio and/or geometrical features. This allows us to fabricate complex shapes from sheet feedstock with one heating cycle (Fig. 1C), which we term “stretch blow molding”. To stretch blow mold BMGs, we developed the following protocol; for BMG feedstock we use Zr44 Ti11 Cu9.8 Ni10.2 Be25 , as it has high
formability [21,31], is commercially available, and is widely studied. A disc of this BMG feedstock is then clamped on top of the mold cavity. Subsequently, the feedstock is heated above the glass transition temperature Tg , where it reaches the supercooled liquid region, at which the feedstock softens and becomes moldable. For the stretch blow molding process, we considered temperatures ranging from 365 ◦ C to 430 ◦ C. The viscosity of Zr44 Ti11 Cu9.8 Ni10.2 Be25 in this temperature region ranges from 109 Pa·s (365 ◦ C) to 106 Pa·s (430 ◦ C) [32,33]. A concentric and axisymmetric poker then pushes and stretches the feedstock into the mold with a varying force of approximately 1 N–1500 N and a strain rate of ∼0.60 s−1 . This stretching is carried out without any contact of the BMG to the mold walls and creates a pre-shape (Fig. 1c middle). It is this step that enables the fabrication of high-aspect ratio BMG articles. Once the BMG material has been stretched to its final extension inside the cavity, the gas pressure applied in the subsequent blow molding step separates the BMG from the poker and molds it against the mold cavity, replicating its shape. Finally, the sample is cooled down below Tg for extraction from the mold. The amorphous nature of the material was confirmed to be retained after all processing steps using differential scanning calorimetry (TA instruments Q200) and X-ray diffraction (Rigaku SmartLab X-ray Diffractometer). Bending tests were performed on cut strips from stretch blow molded samples, which confirmed that the BMG maintained its superb mechanical properties during stretch blow molding. 3. Results and discussion 3.1. Thickness distribution in pre-shape after stretching An essential aspect of the introduced stretch blow molding process is the control of the thickness distribution in the pre-shape after stretching (Fig. 2). We used feedstock discs (25 mm diameter, 1.25 mm thickness) of amorphous Zr44 Ti11 Cu9.8 Ni10.2 Be25 , which were stretched at a temperature of 415 ◦ C to various depths of 25 mm, 50 mm and 75 mm. A circular poker with diameter of 10 mm was used. Two distinctive regions are observed upon stretching (Fig. 2). One of these regions, which we refer to as the contact region, is in contact with the poker and exhibits a relatively uniform thickness distribution. The contact region is thickest at the location that first touches the poker and thinnest at the location that last touches the poker. The region that does not touch the poker has a different thickness distribution and is referred to as the non-contact region. The shape and thickness distribution of the non-contact region is defined by the ratio of the outer rim to the poker diameter, and by the original feedstock thickness and depth of the contact region. To quantitatively describe the thickness distribution in the stretched pre-shape, we developed a 3D model of the stretching process (Fig. 3) based on the following assumptions: The depth of the non-contact region remains constant (Z = PC ), the thickness in the contact region is defined by the changing thickness of t = tc , and no further deformation occurs for material at z > PC . Finally, we also assume that the material is incompressible. A complete derivation of the model can be found in the supplementary material. The model predicts that the thickness evolution in the contact region decays exponentially with the stretch depth, z: 5
t(z) = tt e− 4
D0 −z Pc
(1)
Here, the thickness t at a depth z is a function of a critical point Pc (depth where BMG and poker make contact), the total stretching depth Do and the thickness value tt at the tip (initial thickness at the first point of contact with the poker wall vertically) (Fig. 3). The model shows reasonable agreement with experimental data
R.M. Ojeda Mota, N. Liu, S.A. Kube et al. / Applied Materials Today 19 (2020) 100567
3
Fig. 1. A) Thickness profile after blow molding, which exhibits significant thinning that increases with increasing aspect ratio. B) Thickness profile after blow molding when a parison is used. An approximately uniform thickness distribution can be achieved. C) Thickness profile after blow molding when a sheet-like feedstock is stretched prior to blow molding. Such stretch blow molding results in an approximately uniform thickness distribution even for very high overall strains.
Fig. 2. Thickness profile of stretched BMG to different stretching depths. Two distinct regions are observed. One is the contact region (blue), which comprises a purely extensional flow prior to contact with the poker and no deformation once the BMG touches the poker. The non-contact region (green) is the reservoir of BMG that has not touched the poker yet and deforms in a more homogenous fashion. Dashed lines represent predicted thickness evolution according to Eq. (1) for 50 mm (red) and 75 mm (blue). Right side shows the BMG stretched over the poker and a cross sectional cut of the sample.
(see dashed lines in Fig. 2), but overestimates the thickness variations and underestimates the overall thickness. We argue that the model’s overrepresentation of the thickness is due to the fact that it does not consider the tip effect, the rapidly changing contact angle due to the typical hemispherical tip shape of the poker.
3.2. Controlling the thickness profile distribution The thickness distribution predicted by Eq. (1) and experimentally observed in Fig. 2 is non-uniform. In order to control the thickness distribution, we first evaluated using a thermal gradient within the disc feedstock to locally vary the deformation resistance.
4
R.M. Ojeda Mota, N. Liu, S.A. Kube et al. / Applied Materials Today 19 (2020) 100567
out the stretched pre-shape increases again, particularly in the clamped region (depth 0 mm in Fig. 5). Here, the flow resistance at T1 is overcompensated relative to T2 . By imposing a thermal gradient during the stretching step, however, one has to consider the reduced time to the onset of crystallization at the highest temperature used in order to avoid any unwanted modification in the mechanical characteristics of the part.
Fig. 3. Schematics of the model used for the derivation of Eq. (1). h0 is the original thickness of the feedstock, R(z) is the radius from the center of the poker to the BMG, tc is the thickness at the critical point where the BMG first touches the poker at a depth z = Pc , t(z) is the thickness value at a certain depth, R0 is the radius of the mold cavity, r0 is the radius of the poker. The z-axis origin is placed at the top of the mold where the BMG is originally placed. Z = D0 is the final stretching depth.
As a second option, we also evaluated the ability of the poker geometry to manipulate the location of Pc , as t is a strong function of z. 3.2.1. Thickness control through thermal gradient The strong temperature dependence of the BMGs’ viscosity [34], so called fragility, can be used as a tool to control the thickness distribution in the stretching step. For example, a change in the temperature in the supercooled liquid region by 20 ◦ C results in a change of viscosity by approximately one order of magnitude [33]. The same effect through feedstock geometry (Fig. 4A), corresponds to a thickness variation in the initial feedstock of also one order of magnitude, such thickness variation would be difficult to realize and likely result in an instability [11,22,23], preventing the usage of highly non-uniform sheet feedstock material. To realize a controlled radial temperature gradient in the disc-shaped feedstock, we control the temperature on the outside of the sample (T1 , clamp temperature) and at the center of the feedstock (T2 , poker temperature) (Fig. 4b). In order to reduce the thickness variation in the stretched pre-shape (Fig. 2), a lower temperature in the center of the feedstock material, T2 , should be used relative to the clamp temperature T1 . Therefore, the center region experiences higher resistance to deformation and thins to a lesser degree than it would when both temperatures are identical. Likewise, the outer region will strain more, hence compensating for the geometrical thinning. Temperature differences T1 -T2 ranging from 15 ◦ C to 65 ◦ C were realized and their resulting thickness distribution in the stretched (50 mm deep) pre-shape were determined (Fig. 5A). Upon stretching, the overall thickness in non-contact region (Fig. 2) decreases more quickly as compared to samples where a single uniform temperature was imposed. With increasing temperature gradient from T1 -T2 = 15 ◦ C to T1 -T2 = 55 ◦ C, the thickness becomes more uniform. When further increasing T1 -T2 , the variation in thickness through-
3.2.2. Modification of the point of contact by poker geometry An alternative approach to control the thickness distribution in the stretched pre-shape is based on the modification of the poker geometry (Fig. 4C). By varying the poker geometry, the contact region of the poker with the feedstock BMG can be adjusted. For a constant poker radius, at an arbitrary time 1 the poker subtracts a certain amount of BMG from the non-contact region. Since the volume of the material is finite at a subsequent step at time 2, the poker subtracts a smaller BMG volume. To compensate for this continuous change, we can increase the poker radius to reach for BMG material at a higher position by displacing the contact point Pc upwards (Fig. 4CII). A tapered poker can therefore result in a more even thickness distribution. A schematic illustration is shown in Fig. 4C, where the diagram compares the use of a constant radius poker to a poker with increasing radius. For a cone shaped poker with a round tip, we estimated that an angle of 88.75 degrees would give the most uniform thickness distribution, based on the thickness profile obtained experimentally. Samples of Zr44 Ti11 Cu9.8 Ni10.2 Be25 with an initial thickness of 1.25 mm were stretched to 80 mm in depth. Fig. 5B shows the comparison of the thickness profile obtained for a sample stretched with a uniform poker radius vs. the thickness profile obtained for a sample stretched with a poker with a taper angle with increasing radius. With the tapered poker, we obtain a uniform wall thickness across a larger region of the samples. 3.3. Thermal, structural, and mechanical characterization of stretched BMG An important aspect in BMG processing is how processing affect properties [12,35–37]. It has been shown that changes in fictive temperature [38–41] and partial crystallization [42] can have dramatic effects on the BMGs properties, particularly on their mechanical properties. In addition, some research has suggested that crystallization can be affected by a strain rate during exposure in the supercooled liquid state [43–46]. To characterize the effect of stretch blow molding on thermal, structural, and mechanical properties we carried out a range of characterization at various stages of the stretch blow molding. To quantify their thermal characteristics, we carried out differential thermal analysis (DSC). Specifically, we characterized BMG after it was stretch blow molded and used material from various locations (Fig. 6A) at different stretched depths (a = 15 mm, b = 30 mm and c = 60 mm, sample cap). Based on their glass transition temperature, Tg, their crystallization temperature, Tx, and their heat of crystallization, H, the thermograms from the samples at the various considered locations after stretch blow molding are indistinguishable from the as cast feedstock material. This suggests that stretch blow molding can be carried out without introducing crystallization or even consuming large fractions of the thermal budget. These findings are supported by XRD characterization which reveal an amorphous structure for the stretch blow molded samples, indistinguishable from the as cast feedstock (Fig. 6B). A powerful indicator to determine processing effects on BMGs is the bending strain to failure [47,48]. Whereas BMGs typically do not show any tensile ductility in uniaxial loading, they can exhibit significant ductility in bending, particularly when their thickness is
R.M. Ojeda Mota, N. Liu, S.A. Kube et al. / Applied Materials Today 19 (2020) 100567
5
Fig. 4. Strategies to modify the thickness profile obtained through stretching of BMG. A) Modification of the initial thickness/geometry of the feedstock material. B) Imposing a temperature gradient on the feedstock to locally vary the viscosity. C) Alteration of the point of contact between the poker and the feedstock; CI) shows a zoomed representation of a typical profile obtained using a poker with a constant diameter. Here, the point of contact Pc remains at the same height from time 1 to time 3), CII) shows that by using a poker with an increasing diameter (tapered), the amount of BMG in contact at a given time can be adjusted. The initial Pc at time 1 is continuously moving upwards at the following times 2 and 3. Thereby, one can use the taper angle of the poker to generate a desired thickness distribution.
Fig. 5. A) Thickness distribution profile for samples with a radial thermal gradient where the outer temperature T1 is always higher than the center temperature T2 . B) Thickness distribution profiles for samples stretched with a constant radius poker (red) and a poker with increasing poker radius with an angle of 88.75◦ (black).
6
R.M. Ojeda Mota, N. Liu, S.A. Kube et al. / Applied Materials Today 19 (2020) 100567
Fig. 6. A) DSC thermograms of a stretch blow molded sample taken at three different heights: a) 15 mm, b) 30 mm and c) 60 mm, and for comparison the thermogram of the feedstock. The thermograms for the samples from the various locations after stretch blow molding are undistinguishable within experimental error from the as cast feedstock sample judging by their Tg, Tx, and H. B) XRD spectra of feedstock and stretched blow molded sample both reveal amorphous structure.
below 1 mm [47]. To quantify their bending ductility, bending tests around mandrels with various radii were performed on cut strips from stretch blow molded samples (Fig. 7). Strain to failure is calculated by ε = t/2r, where t is the thickness of the sample and r is the bending radius. Stretch blow molded samples, which were taken from various locations, as well as samples from as cast feedstock, all with an average thickness of ∼.125 mm start to deform plastically at a bending radius of 2.5 mm with a corresponding strain of ∼2.5% and fracture at a bending radius 0.25 mm at a failure strain of 25%. Such large plasticity in all samples further suggests that stretch blow molding can result in BMG parts that maintain their high properties during processing.
with conventional metal processing methods such as spinning or deep drawing (Fig. 6). For example, the versatility of our process allows for the fabrication of seamless open shape parts, which are only possible for high symmetry designs with other metal processing methods (Fig. 8). All BMG parts replicated the mold surface and details. In fact, recent studies of the fidelity of thermoplastic forming processes suggest that much smaller features can be achieved as the surface appearance [49,50]. This suggest that surface finishes, for example a mirror finish, can be incorporated into the stretch blow molding processing step, as oppose to conventionally where it is subsequently added at high costs. 4. Conclusion
3.4. Parts manufactured through stretch blow molding Using above described strategies to control the thickness distribution during the stretching step, we were able to fabricate thin-walled metal parts that are open shaped (Fig. 6). Starting with a flat, sheet-like feedstock, strains of up to 2000% and parts with an aspect ratio of four were achieved. It should be mentioned that likely even larger strains and aspect ratios can be achieved; here we were limited by the equipment design. By comparison, blow molding using sheet like feedstock can only provide parts with an aspect ratio of <1.5 [22]. This technique also allows fabricating single piece geometries with undercuts and side features, which are impossible
We have developed stretch blow molding which allows for the fabrication of high aspect ratio open shape structures from flat, sheet-like bulk metallic glass feedstock. Including a stretching step in the blow molding process eliminates previous limitations such as small aspect ratios and non-uniform thicknesses during blow molding. Achievable overall strains with reasonable thickness variations were increased from ∼150% achieved during blow molding to ∼2000% when incorporating the stretching processing step. Mechanical, thermal, and structural analysis revealed that during stretch blow molding the amorphous structure is maintained. Hence when using appropriate feedstock BMG in sheet-like form,
R.M. Ojeda Mota, N. Liu, S.A. Kube et al. / Applied Materials Today 19 (2020) 100567
7
Fig. 7. A) mandrels used for ductility analysis of the stretch blow molded and feedstock Zr-Based BMG parts. B) Example of a manually bent BMG strip from the stretch blow molded sample.
Fig. 8. Parts obtained through the stretch blow molding process. By starting with a flat sheet-like feedstock with a few mm in thickness as the one shown at the center of the image parts with aspect ratios of 4 and strains over 2000% were obtained.
thin walled shapes, previously unachievable with any metal process, can be realized and these shapes exhibit superb mechanical properties. Open shape geometries are ideal application of BMGs, as they yield exceptional bending ductility as long as the wall thickness is below ∼1 mm [47]. Additionally, open shape structures utilize less material, which is an important consideration due to the relatively high material costs and criticality of BMGs when compared to steel [51]. Therefore, stretch blow molding offers a highly practical and effective method to net-shape complex open shape articles from BMGs, and we are looking forward to a broad commercial adaptation of this method.
Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. CRediT authorship contribution statement Rodrigo Miguel Ojeda Mota: Conceptualization, Methodology, Validation, Formal analysis, Project administration, Visualization, Writing - original draft, Investigation. Naijia Liu: Methodology, Validation, Formal analysis. Sebastian Alexander Kube: Visualization, Methodology, Software, Writing - review & editing. John Chay:
8
R.M. Ojeda Mota, N. Liu, S.A. Kube et al. / Applied Materials Today 19 (2020) 100567
Conceptualization, Methodology. Hayley D. McClintock: Investigation, Writing - review & editing. Jan Schroers: Conceptualization, Methodology, Funding acquisition, Project administration, Supervision, Resources, Writing - review & editing. Acknowledgements The author Ojeda Mota would like to express his gratitude to Consejo Nacional de Ciencia y Tecnología and Secretaría de Energía (CONACYT-SENER) for the financial support provided in the pursuit of his doctoral studies. This work was supported by NSF IIP under grant #1601867. Characterization of the material was done under U.S. Department of Energy through the Office of Science, Basic Energy Sciences, Materials Science and Engineering Division #DE SC0004889. Appendix A. Supplementary data Supplementary material related to this article can be found, in the online version, at https://doi.org/10.1016/j.apmt.2020.100567. References [1] D. Herzog, V. Seyda, E. Wycisk, C. Emmelmann, Acta Mater. 117 (2016) 371–392. [2] W.E. King, A.T. Anderson, R.M. Ferencz, N.E. Hodge, C. Kamath, S.A. Khairallah, A.M. Rubenchik, Appl. Phys. Rev. 2 (2015). [3] M.A. Gibson, N.M. Mykulowycz, J. Shim, R. Fontana, P. Schmitt, A. Roberts, J. Ketkaew, L. Shao, W. Chen, P. Bordeenithikasem, J.S. Myerberg, R. Fulop, M.D. Verminski, E.M. Sachs, Y.M. Chiang, C.A. Schuh, A.J. Hart, J. Schroers, Mater. Today 21 (2018) 697–702. [4] S. Pauly, L. Lober, R. Petters, M. Stoica, S. Scudino, U. Kuhn, J. Eckert, Mater. Today 16 (2013) 37–41. [5] W.A. Backofen, I.R. Turner, D.H. Avery, JOM-J. Met. 16 (1964), 763-&. [6] C.H. Hamilton, A.K. Ghosh, M.W. Mahoney, J. Met. 32 (1980) 43. [7] A.H. Chokshi, A.K. Mukherjee, T.G. Langdon, Mater. Sci. Eng. R-Rep. 10 (1993) 237–274. [8] P. Zwigl, D.C. Dunand, Metall. Mater. Trans 29 (1998) 2571–2582. [9] A.J. Barnes, J. Mater. Eng. Perform. 10 (2007) 11665. [10] Y. Brechet, M. Ashby, Adv. Eng. Mater. 4 (2002) 319. [11] J. Schroers, T.M. Hodges, G. Kumar, H. Raman, A.J. Barnes, P. Quoc, T.A. Waniuk, Mater. Today 14 (2011) 14–19. [12] J. Schroers, Adv. Mater. 22 (2010) 1566–1597. [13] Y. Kawamura, T. Nakamura, H. Kato, H. Mano, A. Inoue, Mater. Sci. Eng. 304 (2001) 674–678. [14] H. Kato, T. Wada, M. Hasegawa, J. Saida, A. Inoue, H.S. Chen, Scr. Mater. 54 (2006) 2023–2027. [15] B. Sarac, J. Ketkaew, D.O. Popnoe, J. Schroers, Adv. Funct. Mater. 22 (2012) 3161–3169. [16] W. Chen, Z. Liu, J. Schroers, Acta Mater. 62 (2014) 49–57.
[17] R. Martinez, G. Kumar, J. Schroers, Scr. Mater. 59 (2008) 187–190. [18] M. Hasan, G. Kumar, Scr. Mater. 123 (2016) 140–143. [19] L.M.A. Douglas, C. Hofmann, Joanna Kolodziejska, Scott N. Roberts, W.L.J. John-Paul Borgonia, Kenneth S. Vecchio, Andrew Kennett, Adv. Eng. Mater. (2016). [20] E.R. Homer, M.B. Harris, S.A. Zirbel, J.A. Kolodziejska, H. Kozachkov, B.P. Trease, J.P.C. Borgonia, G.S. Agnes, L.L. Howell, D.C. Hofmann, Adv. Eng. Mater. 16 (2014) 850–856. [21] E.B. Pitt, G. Kumar, J. Schroers, J. Appl. Phys. 110 (2011). [22] J. Schroers, Q. Pham, A. Peker, N. Paton, R.V. Curtis, Scr. Mater. 57 (2007) 341–344. [23] B. Sarac, G. Kumar, T. Hodges, S.Y. Ding, A. Desai, J. Schroers, J. Microelectromech. Syst. 20 (2011) 28–36. [24] P. Bordeenithikasem, S. Sohn, Z. Liu, J. Schroers, Scr. Mater. 104 (2015) 56–59. [25] H.M. Chiu, G. Kumar, J. Blawzdziewicz, J. Schroers, Scr. Mater. 61 (2009) 28–31. [26] G. Kumar, J. Blawzdziewicz, J. Schroers, Nanotechnology 24 (2013). [27] M. Kanik, P. Bordeenithikasem, D. Kim, N. Selden, A. Desai, R. M’Closkey, J. Schroers, J. Microelectromech. Syst. 24 (2015) 19–28. [28] J. Lu, G. Ravichandran, W.L. Johnson, Acta Mater. 51 (2003) 3429–3443. [29] F. Yang, Appl. Phys. Lett. 91 (2007). [30] J. Lu, G. Ravichandran, W.L. Johnson, Acta Mater. 51 (2003) 3429–3443. [31] J. Schroers, Acta Mater. 56 (2008) 471–478. [32] R. Busch, E. Bakke, W. Johnson, Acta Mater. 46 (1998) 4725–4732. [33] T. Waniuk, J. Schroers, W.L. Johnson, Phys. Rev. B 67 (2003) 184203. [34] R. Busch, J. Schroers, W.H. Wang, MRS Bull. 32 (2007) 620–623. [35] S.V. Ketov, Y.H. Sun, S. Nachum, Z. Lu, A. Checchi, A.R. Beraldin, H.Y. Bai, W.H. Wang, D.V. Louzguine-Luzgin, M.A. Carpenter, A.L. Greer, Nature 524 (2015), 200-+. [36] M. Fan, M.L. Wang, K. Zhang, Y.H. Liu, J. Schroers, M.D. Shattuck, C.S. O’Hern, Phys. Rev. E 95 (2017). [37] J. Ketkaew, R. Yamada, H.Q. Wang, D. Kuldinow, B.S. Schroers, W. Dmowski, T. Egami, J. Schroers, Acta Mater. 184 (2020). [38] G. Kumar, P. Neibecker, L. Yanhui, J. Schroers, Nat. Commun. 4 (2013) 1536. [39] D.J. Magagnosc, G. Kumar, J. Schroers, D.S. Gianola, Acta Mater. 74 (2014) 165. [40] J. Ketkaew, W. Chen, H. Wang, A. Datye, M. Fan, G. Pereira, U.D. Schwarz, Z. Liu, R. Yamada, W. Dmowski, M.D. Shattuck, C.S. O’Hern, T. Egami, E. Bouchbinder, J. Schroers, Nat. Commun. 9 (2018). [41] D.V. Louzguine-Luzgin, T. Saito, J. Saida, A. Inoue, J. Mater. Res. 23 (2008) 515–522. [42] J. Ketkaew, Z. Liu, W. Chen, J. Schroers, Phys. Rev. Lett. 115 (2015). [43] J.F. Sun, M. Yan, J. Shen, Trans. Nonferrous Met. Soc. China 15 (2005) 115–119. [44] W.H. Jiang, F.E. Pinkerton, M. Atzmon, J. Appl. Phys. 93 (2003) 9287–9290. [45] B. Lohwongwatana, J. Schroers, W.L. Johnson, Phys. Rev. Lett. 96 (2006). [46] Z. Shao, J.P. Singer, Y.H. Liu, Z. Liu, H.P. Li, M. Gopinadhan, C.S. O’Hern, J. Schroers, C.O. Osuji, Phys. Rev. E 91 (2015). [47] R.D. Conner, W.L. Johnson, N.E. Paton, W.D. Nix, J. Appl. Phys. 94 (2003) 904–911. [48] G. Kumar, D. Rector, R.D. Conner, J. Schroers, Acta Mater. 57 (2009) 3572–3583. [49] G. Kumar, P.A. Staffier, J. Blawzdziewicz, U.D. Schwarz, J. Schroers, Appl. Phys. Lett. 97 (2010). [50] R. Li, Z. Chen, A. Datye, G.H. Simon, J. Ketkaew, E. Kinser, Z. Liu, C. Zhou, O.E. Dagdeviren, S. Sohn, J.P. Singer, C.O. Osuji, J. Schroers, U.D. Schwarz, Commun. Phys. 1 (2018). [51] R.M.O. Mota, T.E. Graedel, E. Pekarskaya, J. Schroers, JOM 69 (2017) 2156–2163.
Contents lists available at ScienceDirect
Applied Materials Today journal homepage: www.elsevier.com/locate/apmt
Overcoming geometric limitations in metallic glasses through stretch blow molding Rodrigo Miguel Ojeda Mota a , Naijia Liu a , Sebastian Alexander Kube a , John Chay b , Hayley D. McClintock a , Jan Schroers a,∗ a b
Department of Mechanical Engineering & Materials Science, Yale University, New Haven, CT 06511, USA Supercool Metals, New Haven, CT, USA
a r t i c l e
i n f o
Article history: Received 27 August 2019 Received in revised form 12 January 2020 Accepted 14 January 2020 Keywords: Bulk metallic glass Thermoplastic forming Stretch blow molding Superplastic forming
a b s t r a c t Bulk metallic glasses (BMGs) exhibit remarkable mechanical properties, such as high strength and elasticity, which is often paired with fracture toughness. Their supercooled liquid region gives rise to plastic-like processing and suggests parts and shapes that can otherwise not be obtained for crystalline metals. However, current processing techniques only allow for limited options in terms of geometry, thicknesses uniformity, and shape complexity. Here we introduce a new processing technique, “stretch blow molding,” to expand the range of possible parts and increase the available geometries that can be fabricated with BMGs. Additionally, a model is derived that allows for the quantification and prediction of stretch blow molding and provides insight into its potential use and limitations. We demonstrate that with stretch blow molding overall strains exceeding 2000% are achievable, compared to the previously reported ∼150% of blow molding. With the ability to stretch blow mold shapes that were previously unachievable with any other metal fabrication technique in a fast and economical manner, and the superb properties of BMGs, we look forward to a broad commercial adaptation of this technique. © 2020 Published by Elsevier Ltd.
1. Introduction Open shape metal articles are in high demand for use in a wide range of applications including containers, casings, housings, covers, tubing, hoses, and bellows. The defining characteristic of these shapes is that their wall thickness is significantly thinner than their other dimensions. In contrast, closed shape articles exhibit similar size in all dimensions. While a wide range of processes exists for the fabrication of closed shape articles, including forging, casting, and stamping, processes to fabricate open shape metal parts are limited to high symmetry shapes, which are generally difficult to realize in a metal. Additive manufacturing alleviates the challenge to some extent; however, closed shape articles are much easier to fabricate than open shape metal articles with a thin wall thickness [1–4]. With the rise of super plastic forming (SPF), some limitations in fabricating open shape metal articles have been overcome in relation to size, thickness, and usable metals [5–8]. Extensive research
∗ Corresponding author. E-mail address: [email protected] (J. Schroers). https://doi.org/10.1016/j.apmt.2020.100567 2352-9407/© 2020 Published by Elsevier Ltd.
has concluded, however, that SPF suffers from material thinning and is limited to relatively low achievable strain rates [9]. Open shape articles can be readily achieved in thermoplastics [10]. One example is extrusion blow molding, which can fabricate high-strain parts within seconds, as demonstrated by the canonical example of the soda bottle. The reason for such ease is that thermoplastics exist in a viscosity range on the order of 102 –105 Pa·s, where they can be readily deformed under gas pressure. Such a viscosity range is generally not accessible in metals, which was suggested as the reason for the asymmetry in the difficulty of fabricating metals vs thermoplastics [11]. A metallic materials class that provides access to such a viscosity range is BMGs [12–14]. Their so-called supercooled liquid region has been widely explored for novel processing methods [15–21]. One such method is blow molding [11,22,23], which enables the molding of thin wall open shape BMG articles at low pressures (∼1 atm) and over short times (∼10 s) [11]. Here, the limiting factor is geometrical thinning, which restricts this process to relatively low strains, <150% when using sheet-like feedstock [22]. One strategy to overcome geometry limitations due to geometrical thinning is the use of a pre-shaped parison [11]. Though effective, this method requires an additional processing step to manufacture the parison, which increases cost and complexity, while consuming the
2
R.M. Ojeda Mota, N. Liu, S.A. Kube et al. / Applied Materials Today 19 (2020) 100567
time available in the supercooled liquid region before the onset of crystallization (thermal budget) [24]. Here, we overcome the limitation of the achievable shapes due to geometrical thinning by first stretching a sheet-like feedstock into the mold cavity, which we subsequently blow mold. The stretching step creates a basis pre-shape of the final article. To increase the uniformity of the stretched pre-shape, we locally vary temperature and the geometry of the parts used for stretching. This allows us to fabricate complex shapes from a sheet feedstock with overall strains exceeding 2000%. As much higher forces can be readily applied during stretching (up to 100 MPa) than during blow molding (up to 1 MPa), stretch blow molding can be carried out much faster than blow molding alone, hence a shorter cycle time in a commercial process can be achieved and crystallization can be readily circumvented.
2. Method and materials Fabricating open shape BMG parts with TPF based processing is challenging when using techniques like compression molding. This is due to the fact that during TPF, the BMG is subject to perfect stick conditions (zero velocity at BMG/mold interface) [25]. The resulting creep-flow exhibits a scaling of forming pressure, p, with wall thickness, d, of p = vL/d2 (v: velocity, h: viscosity, L: filling depth). Such scaling suggests that compression molding becomes rapidly more difficult when forming articles with thin walls [12,26]. An approach to drastically reduce the required shear strain and shear stress rates for the same geometry is the use of processes where the BMG is, for the most part, not in contact with the mold during the shaping operation [22]. One of these processes is blow molding, where low pressures (∼1 atm) and short times (∼10 s) have been demonstrated to suffice for the fabrication of thin wall open shape articles [27]. Furthermore, material thinning, reflected in the strain rate exponent of unity (m = d/dε˙ = 1), is essentially absent during blow molding of BMGs [28–30]. However, geometrical thinning limits this process to relatively low strain when using sheet-like feedstock. This issue, present in essentially all blowmolding processes, stems from the boundary conditions of the BMG to the mold; once the BMG touches the mold, it no longer deforms and the remaining deformation must be accomplished only with the BMG that has not yet touched the mold [22]. Thinning is particularly problematic for parts with high aspect ratios (Fig. 1A) and has therefore limited blow molding to BMG parts with a small overall strain. One strategy to overcome geometry limitations due to geometrical thinning is the use of pre-shaped parisons (Fig. 1B). When using parisons, which resemble the final shape of the article, all parts of the mold are filled at a similar time after a similar amount of strain. Hence, thinning is eliminated and the majority of deformation should be carried out without any contact to the mold. Though effective in realizing complex shapes that require high strains, the required additional processing step to fabricate the parison enhances costs, complexity of the process, and consumes thermal budget. In this work, we overcome the limitation imposed by geometrical thinning and expand the range of achievable shapes that can be realized with BMGs by first stretching a sheet-like BMG feedstock into the mold cavity. Subsequently, but in the same heating cycle, the stretched feedstock is blow molded against the mold, assuming and replicating its shape (Fig. 1C). The stretching step creates a pre-shape which can be designed to resemble aspects of the final shape such as length, aspect ratio and/or geometrical features. This allows us to fabricate complex shapes from sheet feedstock with one heating cycle (Fig. 1C), which we term “stretch blow molding”. To stretch blow mold BMGs, we developed the following protocol; for BMG feedstock we use Zr44 Ti11 Cu9.8 Ni10.2 Be25 , as it has high
formability [21,31], is commercially available, and is widely studied. A disc of this BMG feedstock is then clamped on top of the mold cavity. Subsequently, the feedstock is heated above the glass transition temperature Tg , where it reaches the supercooled liquid region, at which the feedstock softens and becomes moldable. For the stretch blow molding process, we considered temperatures ranging from 365 ◦ C to 430 ◦ C. The viscosity of Zr44 Ti11 Cu9.8 Ni10.2 Be25 in this temperature region ranges from 109 Pa·s (365 ◦ C) to 106 Pa·s (430 ◦ C) [32,33]. A concentric and axisymmetric poker then pushes and stretches the feedstock into the mold with a varying force of approximately 1 N–1500 N and a strain rate of ∼0.60 s−1 . This stretching is carried out without any contact of the BMG to the mold walls and creates a pre-shape (Fig. 1c middle). It is this step that enables the fabrication of high-aspect ratio BMG articles. Once the BMG material has been stretched to its final extension inside the cavity, the gas pressure applied in the subsequent blow molding step separates the BMG from the poker and molds it against the mold cavity, replicating its shape. Finally, the sample is cooled down below Tg for extraction from the mold. The amorphous nature of the material was confirmed to be retained after all processing steps using differential scanning calorimetry (TA instruments Q200) and X-ray diffraction (Rigaku SmartLab X-ray Diffractometer). Bending tests were performed on cut strips from stretch blow molded samples, which confirmed that the BMG maintained its superb mechanical properties during stretch blow molding. 3. Results and discussion 3.1. Thickness distribution in pre-shape after stretching An essential aspect of the introduced stretch blow molding process is the control of the thickness distribution in the pre-shape after stretching (Fig. 2). We used feedstock discs (25 mm diameter, 1.25 mm thickness) of amorphous Zr44 Ti11 Cu9.8 Ni10.2 Be25 , which were stretched at a temperature of 415 ◦ C to various depths of 25 mm, 50 mm and 75 mm. A circular poker with diameter of 10 mm was used. Two distinctive regions are observed upon stretching (Fig. 2). One of these regions, which we refer to as the contact region, is in contact with the poker and exhibits a relatively uniform thickness distribution. The contact region is thickest at the location that first touches the poker and thinnest at the location that last touches the poker. The region that does not touch the poker has a different thickness distribution and is referred to as the non-contact region. The shape and thickness distribution of the non-contact region is defined by the ratio of the outer rim to the poker diameter, and by the original feedstock thickness and depth of the contact region. To quantitatively describe the thickness distribution in the stretched pre-shape, we developed a 3D model of the stretching process (Fig. 3) based on the following assumptions: The depth of the non-contact region remains constant (Z = PC ), the thickness in the contact region is defined by the changing thickness of t = tc , and no further deformation occurs for material at z > PC . Finally, we also assume that the material is incompressible. A complete derivation of the model can be found in the supplementary material. The model predicts that the thickness evolution in the contact region decays exponentially with the stretch depth, z: 5
t(z) = tt e− 4
D0 −z Pc
(1)
Here, the thickness t at a depth z is a function of a critical point Pc (depth where BMG and poker make contact), the total stretching depth Do and the thickness value tt at the tip (initial thickness at the first point of contact with the poker wall vertically) (Fig. 3). The model shows reasonable agreement with experimental data
R.M. Ojeda Mota, N. Liu, S.A. Kube et al. / Applied Materials Today 19 (2020) 100567
3
Fig. 1. A) Thickness profile after blow molding, which exhibits significant thinning that increases with increasing aspect ratio. B) Thickness profile after blow molding when a parison is used. An approximately uniform thickness distribution can be achieved. C) Thickness profile after blow molding when a sheet-like feedstock is stretched prior to blow molding. Such stretch blow molding results in an approximately uniform thickness distribution even for very high overall strains.
Fig. 2. Thickness profile of stretched BMG to different stretching depths. Two distinct regions are observed. One is the contact region (blue), which comprises a purely extensional flow prior to contact with the poker and no deformation once the BMG touches the poker. The non-contact region (green) is the reservoir of BMG that has not touched the poker yet and deforms in a more homogenous fashion. Dashed lines represent predicted thickness evolution according to Eq. (1) for 50 mm (red) and 75 mm (blue). Right side shows the BMG stretched over the poker and a cross sectional cut of the sample.
(see dashed lines in Fig. 2), but overestimates the thickness variations and underestimates the overall thickness. We argue that the model’s overrepresentation of the thickness is due to the fact that it does not consider the tip effect, the rapidly changing contact angle due to the typical hemispherical tip shape of the poker.
3.2. Controlling the thickness profile distribution The thickness distribution predicted by Eq. (1) and experimentally observed in Fig. 2 is non-uniform. In order to control the thickness distribution, we first evaluated using a thermal gradient within the disc feedstock to locally vary the deformation resistance.
4
R.M. Ojeda Mota, N. Liu, S.A. Kube et al. / Applied Materials Today 19 (2020) 100567
out the stretched pre-shape increases again, particularly in the clamped region (depth 0 mm in Fig. 5). Here, the flow resistance at T1 is overcompensated relative to T2 . By imposing a thermal gradient during the stretching step, however, one has to consider the reduced time to the onset of crystallization at the highest temperature used in order to avoid any unwanted modification in the mechanical characteristics of the part.
Fig. 3. Schematics of the model used for the derivation of Eq. (1). h0 is the original thickness of the feedstock, R(z) is the radius from the center of the poker to the BMG, tc is the thickness at the critical point where the BMG first touches the poker at a depth z = Pc , t(z) is the thickness value at a certain depth, R0 is the radius of the mold cavity, r0 is the radius of the poker. The z-axis origin is placed at the top of the mold where the BMG is originally placed. Z = D0 is the final stretching depth.
As a second option, we also evaluated the ability of the poker geometry to manipulate the location of Pc , as t is a strong function of z. 3.2.1. Thickness control through thermal gradient The strong temperature dependence of the BMGs’ viscosity [34], so called fragility, can be used as a tool to control the thickness distribution in the stretching step. For example, a change in the temperature in the supercooled liquid region by 20 ◦ C results in a change of viscosity by approximately one order of magnitude [33]. The same effect through feedstock geometry (Fig. 4A), corresponds to a thickness variation in the initial feedstock of also one order of magnitude, such thickness variation would be difficult to realize and likely result in an instability [11,22,23], preventing the usage of highly non-uniform sheet feedstock material. To realize a controlled radial temperature gradient in the disc-shaped feedstock, we control the temperature on the outside of the sample (T1 , clamp temperature) and at the center of the feedstock (T2 , poker temperature) (Fig. 4b). In order to reduce the thickness variation in the stretched pre-shape (Fig. 2), a lower temperature in the center of the feedstock material, T2 , should be used relative to the clamp temperature T1 . Therefore, the center region experiences higher resistance to deformation and thins to a lesser degree than it would when both temperatures are identical. Likewise, the outer region will strain more, hence compensating for the geometrical thinning. Temperature differences T1 -T2 ranging from 15 ◦ C to 65 ◦ C were realized and their resulting thickness distribution in the stretched (50 mm deep) pre-shape were determined (Fig. 5A). Upon stretching, the overall thickness in non-contact region (Fig. 2) decreases more quickly as compared to samples where a single uniform temperature was imposed. With increasing temperature gradient from T1 -T2 = 15 ◦ C to T1 -T2 = 55 ◦ C, the thickness becomes more uniform. When further increasing T1 -T2 , the variation in thickness through-
3.2.2. Modification of the point of contact by poker geometry An alternative approach to control the thickness distribution in the stretched pre-shape is based on the modification of the poker geometry (Fig. 4C). By varying the poker geometry, the contact region of the poker with the feedstock BMG can be adjusted. For a constant poker radius, at an arbitrary time 1 the poker subtracts a certain amount of BMG from the non-contact region. Since the volume of the material is finite at a subsequent step at time 2, the poker subtracts a smaller BMG volume. To compensate for this continuous change, we can increase the poker radius to reach for BMG material at a higher position by displacing the contact point Pc upwards (Fig. 4CII). A tapered poker can therefore result in a more even thickness distribution. A schematic illustration is shown in Fig. 4C, where the diagram compares the use of a constant radius poker to a poker with increasing radius. For a cone shaped poker with a round tip, we estimated that an angle of 88.75 degrees would give the most uniform thickness distribution, based on the thickness profile obtained experimentally. Samples of Zr44 Ti11 Cu9.8 Ni10.2 Be25 with an initial thickness of 1.25 mm were stretched to 80 mm in depth. Fig. 5B shows the comparison of the thickness profile obtained for a sample stretched with a uniform poker radius vs. the thickness profile obtained for a sample stretched with a poker with a taper angle with increasing radius. With the tapered poker, we obtain a uniform wall thickness across a larger region of the samples. 3.3. Thermal, structural, and mechanical characterization of stretched BMG An important aspect in BMG processing is how processing affect properties [12,35–37]. It has been shown that changes in fictive temperature [38–41] and partial crystallization [42] can have dramatic effects on the BMGs properties, particularly on their mechanical properties. In addition, some research has suggested that crystallization can be affected by a strain rate during exposure in the supercooled liquid state [43–46]. To characterize the effect of stretch blow molding on thermal, structural, and mechanical properties we carried out a range of characterization at various stages of the stretch blow molding. To quantify their thermal characteristics, we carried out differential thermal analysis (DSC). Specifically, we characterized BMG after it was stretch blow molded and used material from various locations (Fig. 6A) at different stretched depths (a = 15 mm, b = 30 mm and c = 60 mm, sample cap). Based on their glass transition temperature, Tg, their crystallization temperature, Tx, and their heat of crystallization, H, the thermograms from the samples at the various considered locations after stretch blow molding are indistinguishable from the as cast feedstock material. This suggests that stretch blow molding can be carried out without introducing crystallization or even consuming large fractions of the thermal budget. These findings are supported by XRD characterization which reveal an amorphous structure for the stretch blow molded samples, indistinguishable from the as cast feedstock (Fig. 6B). A powerful indicator to determine processing effects on BMGs is the bending strain to failure [47,48]. Whereas BMGs typically do not show any tensile ductility in uniaxial loading, they can exhibit significant ductility in bending, particularly when their thickness is
R.M. Ojeda Mota, N. Liu, S.A. Kube et al. / Applied Materials Today 19 (2020) 100567
5
Fig. 4. Strategies to modify the thickness profile obtained through stretching of BMG. A) Modification of the initial thickness/geometry of the feedstock material. B) Imposing a temperature gradient on the feedstock to locally vary the viscosity. C) Alteration of the point of contact between the poker and the feedstock; CI) shows a zoomed representation of a typical profile obtained using a poker with a constant diameter. Here, the point of contact Pc remains at the same height from time 1 to time 3), CII) shows that by using a poker with an increasing diameter (tapered), the amount of BMG in contact at a given time can be adjusted. The initial Pc at time 1 is continuously moving upwards at the following times 2 and 3. Thereby, one can use the taper angle of the poker to generate a desired thickness distribution.
Fig. 5. A) Thickness distribution profile for samples with a radial thermal gradient where the outer temperature T1 is always higher than the center temperature T2 . B) Thickness distribution profiles for samples stretched with a constant radius poker (red) and a poker with increasing poker radius with an angle of 88.75◦ (black).
6
R.M. Ojeda Mota, N. Liu, S.A. Kube et al. / Applied Materials Today 19 (2020) 100567
Fig. 6. A) DSC thermograms of a stretch blow molded sample taken at three different heights: a) 15 mm, b) 30 mm and c) 60 mm, and for comparison the thermogram of the feedstock. The thermograms for the samples from the various locations after stretch blow molding are undistinguishable within experimental error from the as cast feedstock sample judging by their Tg, Tx, and H. B) XRD spectra of feedstock and stretched blow molded sample both reveal amorphous structure.
below 1 mm [47]. To quantify their bending ductility, bending tests around mandrels with various radii were performed on cut strips from stretch blow molded samples (Fig. 7). Strain to failure is calculated by ε = t/2r, where t is the thickness of the sample and r is the bending radius. Stretch blow molded samples, which were taken from various locations, as well as samples from as cast feedstock, all with an average thickness of ∼.125 mm start to deform plastically at a bending radius of 2.5 mm with a corresponding strain of ∼2.5% and fracture at a bending radius 0.25 mm at a failure strain of 25%. Such large plasticity in all samples further suggests that stretch blow molding can result in BMG parts that maintain their high properties during processing.
with conventional metal processing methods such as spinning or deep drawing (Fig. 6). For example, the versatility of our process allows for the fabrication of seamless open shape parts, which are only possible for high symmetry designs with other metal processing methods (Fig. 8). All BMG parts replicated the mold surface and details. In fact, recent studies of the fidelity of thermoplastic forming processes suggest that much smaller features can be achieved as the surface appearance [49,50]. This suggest that surface finishes, for example a mirror finish, can be incorporated into the stretch blow molding processing step, as oppose to conventionally where it is subsequently added at high costs. 4. Conclusion
3.4. Parts manufactured through stretch blow molding Using above described strategies to control the thickness distribution during the stretching step, we were able to fabricate thin-walled metal parts that are open shaped (Fig. 6). Starting with a flat, sheet-like feedstock, strains of up to 2000% and parts with an aspect ratio of four were achieved. It should be mentioned that likely even larger strains and aspect ratios can be achieved; here we were limited by the equipment design. By comparison, blow molding using sheet like feedstock can only provide parts with an aspect ratio of <1.5 [22]. This technique also allows fabricating single piece geometries with undercuts and side features, which are impossible
We have developed stretch blow molding which allows for the fabrication of high aspect ratio open shape structures from flat, sheet-like bulk metallic glass feedstock. Including a stretching step in the blow molding process eliminates previous limitations such as small aspect ratios and non-uniform thicknesses during blow molding. Achievable overall strains with reasonable thickness variations were increased from ∼150% achieved during blow molding to ∼2000% when incorporating the stretching processing step. Mechanical, thermal, and structural analysis revealed that during stretch blow molding the amorphous structure is maintained. Hence when using appropriate feedstock BMG in sheet-like form,
R.M. Ojeda Mota, N. Liu, S.A. Kube et al. / Applied Materials Today 19 (2020) 100567
7
Fig. 7. A) mandrels used for ductility analysis of the stretch blow molded and feedstock Zr-Based BMG parts. B) Example of a manually bent BMG strip from the stretch blow molded sample.
Fig. 8. Parts obtained through the stretch blow molding process. By starting with a flat sheet-like feedstock with a few mm in thickness as the one shown at the center of the image parts with aspect ratios of 4 and strains over 2000% were obtained.
thin walled shapes, previously unachievable with any metal process, can be realized and these shapes exhibit superb mechanical properties. Open shape geometries are ideal application of BMGs, as they yield exceptional bending ductility as long as the wall thickness is below ∼1 mm [47]. Additionally, open shape structures utilize less material, which is an important consideration due to the relatively high material costs and criticality of BMGs when compared to steel [51]. Therefore, stretch blow molding offers a highly practical and effective method to net-shape complex open shape articles from BMGs, and we are looking forward to a broad commercial adaptation of this method.
Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. CRediT authorship contribution statement Rodrigo Miguel Ojeda Mota: Conceptualization, Methodology, Validation, Formal analysis, Project administration, Visualization, Writing - original draft, Investigation. Naijia Liu: Methodology, Validation, Formal analysis. Sebastian Alexander Kube: Visualization, Methodology, Software, Writing - review & editing. John Chay:
8
R.M. Ojeda Mota, N. Liu, S.A. Kube et al. / Applied Materials Today 19 (2020) 100567
Conceptualization, Methodology. Hayley D. McClintock: Investigation, Writing - review & editing. Jan Schroers: Conceptualization, Methodology, Funding acquisition, Project administration, Supervision, Resources, Writing - review & editing. Acknowledgements The author Ojeda Mota would like to express his gratitude to Consejo Nacional de Ciencia y Tecnología and Secretaría de Energía (CONACYT-SENER) for the financial support provided in the pursuit of his doctoral studies. This work was supported by NSF IIP under grant #1601867. Characterization of the material was done under U.S. Department of Energy through the Office of Science, Basic Energy Sciences, Materials Science and Engineering Division #DE SC0004889. Appendix A. Supplementary data Supplementary material related to this article can be found, in the online version, at https://doi.org/10.1016/j.apmt.2020.100567. References [1] D. Herzog, V. Seyda, E. Wycisk, C. Emmelmann, Acta Mater. 117 (2016) 371–392. [2] W.E. King, A.T. Anderson, R.M. Ferencz, N.E. Hodge, C. Kamath, S.A. Khairallah, A.M. Rubenchik, Appl. Phys. Rev. 2 (2015). [3] M.A. Gibson, N.M. Mykulowycz, J. Shim, R. Fontana, P. Schmitt, A. Roberts, J. Ketkaew, L. Shao, W. Chen, P. Bordeenithikasem, J.S. Myerberg, R. Fulop, M.D. Verminski, E.M. Sachs, Y.M. Chiang, C.A. Schuh, A.J. Hart, J. Schroers, Mater. Today 21 (2018) 697–702. [4] S. Pauly, L. Lober, R. Petters, M. Stoica, S. Scudino, U. Kuhn, J. Eckert, Mater. Today 16 (2013) 37–41. [5] W.A. Backofen, I.R. Turner, D.H. Avery, JOM-J. Met. 16 (1964), 763-&. [6] C.H. Hamilton, A.K. Ghosh, M.W. Mahoney, J. Met. 32 (1980) 43. [7] A.H. Chokshi, A.K. Mukherjee, T.G. Langdon, Mater. Sci. Eng. R-Rep. 10 (1993) 237–274. [8] P. Zwigl, D.C. Dunand, Metall. Mater. Trans 29 (1998) 2571–2582. [9] A.J. Barnes, J. Mater. Eng. Perform. 10 (2007) 11665. [10] Y. Brechet, M. Ashby, Adv. Eng. Mater. 4 (2002) 319. [11] J. Schroers, T.M. Hodges, G. Kumar, H. Raman, A.J. Barnes, P. Quoc, T.A. Waniuk, Mater. Today 14 (2011) 14–19. [12] J. Schroers, Adv. Mater. 22 (2010) 1566–1597. [13] Y. Kawamura, T. Nakamura, H. Kato, H. Mano, A. Inoue, Mater. Sci. Eng. 304 (2001) 674–678. [14] H. Kato, T. Wada, M. Hasegawa, J. Saida, A. Inoue, H.S. Chen, Scr. Mater. 54 (2006) 2023–2027. [15] B. Sarac, J. Ketkaew, D.O. Popnoe, J. Schroers, Adv. Funct. Mater. 22 (2012) 3161–3169. [16] W. Chen, Z. Liu, J. Schroers, Acta Mater. 62 (2014) 49–57.
[17] R. Martinez, G. Kumar, J. Schroers, Scr. Mater. 59 (2008) 187–190. [18] M. Hasan, G. Kumar, Scr. Mater. 123 (2016) 140–143. [19] L.M.A. Douglas, C. Hofmann, Joanna Kolodziejska, Scott N. Roberts, W.L.J. John-Paul Borgonia, Kenneth S. Vecchio, Andrew Kennett, Adv. Eng. Mater. (2016). [20] E.R. Homer, M.B. Harris, S.A. Zirbel, J.A. Kolodziejska, H. Kozachkov, B.P. Trease, J.P.C. Borgonia, G.S. Agnes, L.L. Howell, D.C. Hofmann, Adv. Eng. Mater. 16 (2014) 850–856. [21] E.B. Pitt, G. Kumar, J. Schroers, J. Appl. Phys. 110 (2011). [22] J. Schroers, Q. Pham, A. Peker, N. Paton, R.V. Curtis, Scr. Mater. 57 (2007) 341–344. [23] B. Sarac, G. Kumar, T. Hodges, S.Y. Ding, A. Desai, J. Schroers, J. Microelectromech. Syst. 20 (2011) 28–36. [24] P. Bordeenithikasem, S. Sohn, Z. Liu, J. Schroers, Scr. Mater. 104 (2015) 56–59. [25] H.M. Chiu, G. Kumar, J. Blawzdziewicz, J. Schroers, Scr. Mater. 61 (2009) 28–31. [26] G. Kumar, J. Blawzdziewicz, J. Schroers, Nanotechnology 24 (2013). [27] M. Kanik, P. Bordeenithikasem, D. Kim, N. Selden, A. Desai, R. M’Closkey, J. Schroers, J. Microelectromech. Syst. 24 (2015) 19–28. [28] J. Lu, G. Ravichandran, W.L. Johnson, Acta Mater. 51 (2003) 3429–3443. [29] F. Yang, Appl. Phys. Lett. 91 (2007). [30] J. Lu, G. Ravichandran, W.L. Johnson, Acta Mater. 51 (2003) 3429–3443. [31] J. Schroers, Acta Mater. 56 (2008) 471–478. [32] R. Busch, E. Bakke, W. Johnson, Acta Mater. 46 (1998) 4725–4732. [33] T. Waniuk, J. Schroers, W.L. Johnson, Phys. Rev. B 67 (2003) 184203. [34] R. Busch, J. Schroers, W.H. Wang, MRS Bull. 32 (2007) 620–623. [35] S.V. Ketov, Y.H. Sun, S. Nachum, Z. Lu, A. Checchi, A.R. Beraldin, H.Y. Bai, W.H. Wang, D.V. Louzguine-Luzgin, M.A. Carpenter, A.L. Greer, Nature 524 (2015), 200-+. [36] M. Fan, M.L. Wang, K. Zhang, Y.H. Liu, J. Schroers, M.D. Shattuck, C.S. O’Hern, Phys. Rev. E 95 (2017). [37] J. Ketkaew, R. Yamada, H.Q. Wang, D. Kuldinow, B.S. Schroers, W. Dmowski, T. Egami, J. Schroers, Acta Mater. 184 (2020). [38] G. Kumar, P. Neibecker, L. Yanhui, J. Schroers, Nat. Commun. 4 (2013) 1536. [39] D.J. Magagnosc, G. Kumar, J. Schroers, D.S. Gianola, Acta Mater. 74 (2014) 165. [40] J. Ketkaew, W. Chen, H. Wang, A. Datye, M. Fan, G. Pereira, U.D. Schwarz, Z. Liu, R. Yamada, W. Dmowski, M.D. Shattuck, C.S. O’Hern, T. Egami, E. Bouchbinder, J. Schroers, Nat. Commun. 9 (2018). [41] D.V. Louzguine-Luzgin, T. Saito, J. Saida, A. Inoue, J. Mater. Res. 23 (2008) 515–522. [42] J. Ketkaew, Z. Liu, W. Chen, J. Schroers, Phys. Rev. Lett. 115 (2015). [43] J.F. Sun, M. Yan, J. Shen, Trans. Nonferrous Met. Soc. China 15 (2005) 115–119. [44] W.H. Jiang, F.E. Pinkerton, M. Atzmon, J. Appl. Phys. 93 (2003) 9287–9290. [45] B. Lohwongwatana, J. Schroers, W.L. Johnson, Phys. Rev. Lett. 96 (2006). [46] Z. Shao, J.P. Singer, Y.H. Liu, Z. Liu, H.P. Li, M. Gopinadhan, C.S. O’Hern, J. Schroers, C.O. Osuji, Phys. Rev. E 91 (2015). [47] R.D. Conner, W.L. Johnson, N.E. Paton, W.D. Nix, J. Appl. Phys. 94 (2003) 904–911. [48] G. Kumar, D. Rector, R.D. Conner, J. Schroers, Acta Mater. 57 (2009) 3572–3583. [49] G. Kumar, P.A. Staffier, J. Blawzdziewicz, U.D. Schwarz, J. Schroers, Appl. Phys. Lett. 97 (2010). [50] R. Li, Z. Chen, A. Datye, G.H. Simon, J. Ketkaew, E. Kinser, Z. Liu, C. Zhou, O.E. Dagdeviren, S. Sohn, J.P. Singer, C.O. Osuji, J. Schroers, U.D. Schwarz, Commun. Phys. 1 (2018). [51] R.M.O. Mota, T.E. Graedel, E. Pekarskaya, J. Schroers, JOM 69 (2017) 2156–2163.