Fundamental mechanical and microstructural observations in metallic glass coating production

Size: px
Start display at page:

Download "Fundamental mechanical and microstructural observations in metallic glass coating production"

Transcription

1 Computer Methods and Experimental Measurements for Surface Effects and Contact Mechanics VII 161 Fundamental mechanical and microstructural observations in metallic glass coating production D. T. A. Matthews, V. Ocelík & J. Th. M. De Hosson Department of Applied Physics, Materials Science Centre and The Netherlands Institute of Metals Research, University of Groningen, The Netherlands Abstract The production of a wide range of metallic Glass Forming Alloys (GFA) has been investigated by several processing routes including simple arc-casting and melt-spinning to form Bulk Metallic Glasses (BMG). The concepts surrounding such alloys have been directed towards the production of thick (>300µm) amorphous surface layers by high power laser treatments. Microstructural observation techniques include secondary electron microscopy, transmission electron microscopy and X-ray diffraction, which reveal that fully amorphous metals are attainable with a range of dimensions by arc- and induction-furnace fabrication, whilst the laser processing routes have also been proven to be powerful enough techniques to facilitate the production of glassy metallic surface layers. Thermo-dependant properties have been explored by differential scanning calorimetry and in-situ heating with transmission electron microscopy. Hardness and nano-indentation profiles reveal hardnesses up to 850 Vickers over the full depth of the coating. Shear band formation in the laser treated layers has also been observed and reported. Keywords: bulk metallic glasses, laser surface treatment, DSC, XRD, TEM, hardness. 1 Introduction Crystalline materials are inherently vulnerable to failure by (for example) fracture or corrosion due to their in built defects grain boundaries. Grain

2 162 Computer Methods and Experimental Measurements for Surface Effects and Contact Mechanics VII boundaries represent weak areas of less than optimal packing, and contain numerous small defects. It is possible to create materials which do not contain these weak spots, and these are termed amorphous, which descends from Greek language and means without regular form [α = not, µορφη = form]. The atoms in an amorphous metal are therefore considered to be arranged with random order. This class of metallic alloys has been grouped into two general categories, as cited by Perepezko and Herbert [1]. In the first case; large, bulk volumes may be slowly cooled to the glassy state, which signifies a nucleation controlled synthesis. The other class is represented by metallic glasses that can be synthesised upon rapid solidification processes such as melt spinning. These glasses are often called marginal glass formers that are synthesised under growth controlled kinetic conditions found the first metallic glass, developed by Klement et al. [2], with the discovery of the Au-Si BMG. In the ensuing years, amorphous alloys have been developed progressively, with one of the driving forces behind the development of metallic glasses being to lower their critical cooling rate, and thus facilitate their fabrication (and size) [3]. Whilst this critical cooling rate is being reduced, significant quench rates are still required ( Ks -1 ). The dimensions of these rods are currently limited to around 5 mm thicknesses, though rods up to 14 mm have been reported [4]. Of course, it is rarely the case that an entire work-piece should need to be formed wholly from one material, since, other than for structural properties, a manufactured article is, more often that not, only functional at the surface. By harnessing the properties of selected BMGs in the surfaces of tribologically poor materials, (such as titanium and aluminium) these materials are can be exploited in many more diverse ranges of applications than they currently find. One powerful surface engineering tool is that of the high power laser, which has been proven to produce adherent, hard, wear- corrosion- fatigue- and fracture resistant coatings on a diverse range of materials [5,6,7]. Another major attribute of the high power laser is the high cooling rates attainable [8,9]. These are certainly in the bounds of the quench rates necessary for amorphousisation, and hence surface engineering by high power laser provides the chosen tool for this investigation into the fabrication of functionally graded amorphous surface layers. The associated adhesion properties of an FGM ensure the prospects are exciting [10]. Numerous BMG compositions have been published to date. The subject of our intrigue, as stated previously, has been not only the possibility of producing metallic glasses, but producing glassy metallic surface layers by high power laser. It will be noticed that our main impetus is behind Ti-rich or Ticontaining compositions in the hope that we may laser clad to improve the inherently poor tribological properties of titanium. Aluminium is also an interesting substrate material, as is iron. Many of the BMG compositions published contain Zr, which is often used in tandem with Be, as this element considerably improves the glass forming ability of Zr containing alloys [11,12] by strong bonding between Zr-Be atomic pairs which suppress the formation of competing crystalline phases during solidification. Beryllium, however, unfortunately forms harmful (cancerous) oxides and therefore is deemed too dangerous for our chosen processing route.

3 Computer Methods and Experimental Measurements for Surface Effects and Contact Mechanics VII Experimental procedure Alloys are prepared by weighing the component elements, such that an approximately 1 cm 3 button may be produced by arc melting. The materials are of at least 99.99% purity and in sheet, plate, pellet or powder form prior to fabrication. The melting process is conducted in a Ti-gettered, high purity argon atmosphere. To ensure chemical and microstrucural homogeneity, the buttons are rotated and remelted 3-5 times within the furnace. The resultant buttons are then weighed and then (given negligible weight loss) analysed by optical and secondary electron microscopy (SEM Philips XL30 FEG with EDS). Ribbons of 2-8 mm width, with thicknesses in the region µm, are produced from the pre-alloyed buttons. They are reheated above their melting point in an argon or helium atmosphere by induction heating and injected by an overpressure of 500 mbar onto a rotating (1800 rpm) copper wheel (Ø = 50 cm). The buttons may also be cut to appropriate shapes and sizes for arc casting into water cooled copper moulds, with internal dimensions of 1 mm or 2 mm diameter cylinders 25 mm in length, or 0.5 mm, 0.75 mm or 1 mm thick plates 5 mm wide and 35 mm in length. The buttons have also been prepared for laser remelting by cutting the buttons to 15 cm diameter hemispheres, followed by grinding and fine polishing to produce a flat surface. Since, during laser treatments, some of the applied energy may be reflected away from the target, the surface is fine sand blasted to reduce the reflectivity, ergo improving the efficiency of the laser processing. The laser remelting process was conducted over a range of processing parameters which will be specified as appropriate with a 2 kw Rofin-Sinar Nd-YAG laser, however laser power is always kept at 1750 W, and argon shielding of 10 l/min is always applied. All resultant fabrications are investigated by optical microscopy, secondary electron microscopy with energy dispersive spectroscopy (SEM with EDS), (high resolution) transmission electron microscopy ((HR)TEM) (FEG Jeol 2010) with in-situ heating and EELS capability, X-ray diffraction (XRD) (Phillips PW1710) and differential scanning calorimetry (DSC) (Perkin-Elmer DSC 7), where the test specimens are placed in sealed aluminium pans and heated at rates of 10, 20 or 40 o C/min between 25 and 600 o C. Hardness and nano-indentation examinations are conducted on CSM Revetester and MTS Nanoindenter XP with CSM/LFM control respectively. 3 Results and discussion 3.1 Melt-spun ribbons The resultant ribbons from the melt-spinning process can be examined by XRD. A typical scan is shown in figure 1. Cu 47 Ti 33 Zr 11 Ni 8 Si 1 ribbon is represented in figure 1a and is seen to be fully X-ray amorphous, with the amorphous halo seen around 2theta = 40 o. Figure 1b shows a Cu 47 Ti 34 Zr 11 Ni 8 ribbon, which shows a small diffraction peak [highlighted by the rectangular block] alongside the expected amorphous halo. Such features may signify a nanocrystalline phase, and provide interest for TEM analysis. This result also highlights the effect of

4 164 Computer Methods and Experimental Measurements for Surface Effects and Contact Mechanics VII the addition of silicon to this system in quantities < 2 at % on its glass forming ability. A XRD Cu 47 Ti 34 Zr 11 Ni 8 B 1600 XRD Cu 47 Ti 33 Zr 11 Ni 8 Si Intensity (a.u.) Intensity (a.u.) theta (deg) 2theta (deg) Figure 1: XRD scans for (a) Cu 47 Ti 33 Zr 11 Ni 8 Si 1 and (b) Cu 47 Ti 34 Zr 11 Ni 8 melt spun ribbons. In this article, the TEM results shown are concerned with Cu 47 Ti 34 Zr 11 Ni 8 BMG alloy, part of a series first developed by Johnson et al. [13]. Many other works have been published in connection with this system and its recrystallization; however slight differences in production route or the state of the raw material can lead to widely differing results and indeed differing degrees of amorphousisation. The glass transition temperature, T g (endothermic reaction) and crystallisation temperature T x (exothermic reaction) are not strictly intrinsic properties of a system, but instead are found to lie within a range. From a detailed reading of published data, for most BMG alloys this range has been found to be in the order of 50K. This is highlighted in the DSC trace shown in figure 2. The ribbon was heated to discover information regarding the material s thermo-dependant properties. Not only does this DSC traces for the ribbon yield information surrounding T g (found to be ~375 o C) and T x found to be ~450 o C), but also proves the material to be amorphous, or at least metastable in state. The area of the first crystallisation gives a H of J/g. Heat Flow [endo. up] (mw/mg) 0.3 T x T g H Temperature (Celcius) Figure 2: DSC trace for Cu 47 Ti 34 Zr 11 Ni 8. EDS analysis was conducted and yielded results such that the matrix was seen to be very close to that of nominal, with a matrix composition of (in atomic %) Cu 48 Ti 33 Zr 11 Ni 8, whilst crystals found in the matrix were seen to be titanium rich,

5 Computer Methods and Experimental Measurements for Surface Effects and Contact Mechanics VII 165 with a composition (in atomic %) of Cu 22 Ti 66 Zr 8 Ni 4. Diffraction patterns from the matrix showed it to be amorphous. This was confirmed by high resolution TEM (HRTEM) imaging, which shows there were indeed crystals, embedded in an amorphous matrix (figure 3). A B Figure 3: [A] TEM micrograph showing a Ti rich crystal in an amorphous matrix; the HRTEM image of which is shown in [B]. ZL 250x250nm Ti Map 250x250nm SERIES A SERIES B ZL 250x250nm Ti Map 250x250nm Figure 4: TEM micrographs of the two matrices found to develop in a Cu 47 Ti 34 Zr 11 Ni 8 melt spun ribbon. [SERIES A] shows a chemically homogeneous matrix; [SERIES B] shows a matrix exhibiting spinodal decomposition. The matrix has been found to take two forms. These are represented in figure 4. The first is a homogeneous matrix (series A, figure 4), the second shows some decomposition (series B, figure 4) (possibly spinodal, due to the vein-like appearance of the microstructure) or segregation. By virtue of EELS analysis in the TEM, it was possible to show the vein-like structure to be Ti rich (white) and most likely Cu regions. Both pictorial series comprise an overview, a zero-loss (ZL) image and a Ti map to highlight the points outlined here. Again, the presence of Ti rich crystals is clearly evident in both cases.

6 166 Computer Methods and Experimental Measurements for Surface Effects and Contact Mechanics VII From the DSC values attained, it is possible to attempt knowledge-driven in-situ heating with TEM (figure 5). Heating began just above the given first crystallisation temperature of 450 oc. Imperfect thermal contact between the specimen holder and specimen however, limited crystallisation until 490 oc. Beyond this temperature, a clear densification of crystals is seen in the sample, however it appears retarded at some distance from the sample edge, until heating above 535 oc is achieved, at which point explosive crystallisation occurs of, once again, Ti-rich composition, revealed by EELS analysis Figure 5: TEM observation of in-situ heating of Cu47Ti34Zr11Ni8 ribbon with values shown in oc. By the size of the crystals grown, which will not grow larger than the thickness of the ribbon, it is possible to say that the thickness dependent crystallisation of Cu47Ti34Zr11Ni8 is limited to a thickness of 100 nm and requires and activation temperature of 45 oc higher than the first crystallisation. This result is included as an example, but many other compositions have also been investigated, and again amorphous metallics appear easily attainable with such a production method; however when the casting dimensions are increased, the amorphousisation is not so easily achieved, as will be seen in section Arc-cast rods and plates The fabrication or amorphous rods has been widely publicised as an appropriate method for BMG production, however in our attempts to recreate published data,

7 Computer Methods and Experimental Measurements for Surface Effects and Contact Mechanics VII 167 a deal of problems have arisen that have previously had no explanation or commentary. Suction cast rods, prepared as outlined previously, have been investigated by XRD, DSC, and microscopically. An example of the critical nature of the casting thickness is represented in figure 6a. Here, the DSC result of a 1 mm and 2 mm Zr 50 Cu 30 Al 10 Ni 10 rod is seen, with their second traces also shown. When a specimen is heated and crystallises, this crystallisation may be either reversible (stable) or irreversible (metastable such as amorphous). Therefore, if a specimen is cooled down and again reheated, the crystallisation activity will reappear if the process is stable (2 mm diameter rod), but it will not appear if the transition is metastable (1 mm diameter rod). When viewed microscopically, the 1 mm rod was consistent of very fine dendritic growth. The 2 mm rods were similar; however some larger dendrites (figure 6b) were also seen with high Zr contents (Zr 90 Cu 3.3 Al 3.5 Ni 3.2 ). The hardness of such rods is seen to reside in the region of 850 Vickers, and rather surprisingly does not vary between the two dimensions. 3 Heat Flow [Endo Up] (mw/mg) mmDSC1 2mmDSC2 1mmDSC1 1mmDSC2-4 A Temperature ( o C) B Figure 6: DSC trace for 1mm and 2mm arc-cast rods with compositions of Zr 50 Cu 30 Al 10 Ni 10 (A); the difference possibly caused by the growth of Zr-based dendrites shown in the micrograph (B). 3.3 Laser treatments All materials have been fabricated in accordance with published data citing compositions which appear attractive to the laser processing techniques under investigation (namely laser remelting and laser cladding). Chronologically, direct laser cladding was originally attempted, but failures occurred on a variety of levels. These were mainly due to differing powder efficiencies and the volatile nature of Zr powders. It was instead proposed that simple remelting of prealloyed buttons would permit a better understanding of the possibilities in producing glassy metallic layers by high power laser. It is however worthwhile briefly noting some of the laser cladding results attained. Of particular interest was work based upon a model which can be found elsewhere [14]. The work highlights the complexities of laser cladding. Figure 7 shows the result of the model in [14], developed for the application of up to n constituents. The model was used for only two powders; pure Cu, with regular dimensions and a mean

8 168 Computer Methods and Experimental Measurements for Surface Effects and Contact Mechanics VII particle size of 50µm and also >97% purity Zr 70 Ni 30 powder with an extremely fine mean particle size of 4µm. This small size, coupled with the volatile nature of Zr itself, both in reaction with moisture and air lead to highly ineffective processing, with coarse grained (50 µm), high purity (>98% Zr) powder now utilised due to these experiments. Laser Power (W) Zr 70 Ni 30 Cu Substrate Scan Speed (ms -1 ) Figure 7: Graph revealing the power required to melt Zr 70 Ni 30 powder, Cu powder and a pure Ti substrate as function of scan speed. The remelting of pre-alloyed specimens has proven very successful, in terms of creating glassy metallic layers, and therefore improving the mechanical properties (such as hardness). Figure 8 shows an optical micrograph of a single laser remelted Zr 50 Cu 30 Al 10 Ni 10 track (process parameters, P = 1750 W, scan speed = 15m/min, defocus = +3 mm), which has also been subjected to microhardness measurements. Due to the rapid processing, mushy regions are seen to be retained in the layer. 300µm Figure 8: Optical micrograph showing a microhardness tested laser remelted Zr 50 Cu 30 Al 10 Ni 10 layer. Maintaining the energy density of the laser beam at a constant, with changing parameters has been investigated. The overall depth of the surface layers is not seen to change for constant energy densities; however, the proportion of the affected area that was found to be amorphous was greatly reduced for slower scan speeds in a variety of compositions. This is beneficial in terms of creating an adherent gradient from coating to substrate; however the effective amorphous area is detrimentally depleted. The track width decreases with increasing scan

9 Computer Methods and Experimental Measurements for Surface Effects and Contact Mechanics VII 169 speed. It has been witnessed that some of the bulk glass forming alloys under investigation bear their highest hardness values at some distance below the surface (figure 9a and b). This is expected to coincide with nanoscale changes due to an internal cooling gradient, with the softer region of the track in the first 50µm possibly being slower cooled due to its close proximity with the immediate atmosphere, in comparison with the rapid conductive cooling afforded by the bulk material below the laser track. The result shown is for the compositions Cu 47 Ti 33 Zr 11 Ni 6 Sn 2 Si 1, [A] and Zr 50 Cu 30 Al 10 Ni 10 (from fig. 8) [B] but similar results have been found for Cu 47 Ti 34 Zr 11 Ni 8 and Cu 47 Ti 33 Zr 11 Ni 8 Si 1 regardless of the processing parameters. The result for the hardness of Cu 47 Ti 33 Zr 11 Ni 6 Sn 2 Si 1 shows data for 2 tracks, both produced with laser parameters of Power, P =1750 W, scan speed = 9m/min, defocus +6 mm. The hardness of the materials ranges between 750 and 850 Vickers, which will provide vast improvements in the surface properties of materials such as aluminium (<100 Vickers) and titanium (~210 Vickers) when applied by laser cladding techniques A 850 B Hardness (HV2) Interfacial region (T1) (T2) Hardness (HV2) Mushy Zone Heat Affected Zone 640 "Substrate" x from surface (µm) Distance from surface (µm) Figure 9: Hardness profiles for Cu 47 Ti 33 Zr 11 Ni 6 Sn 2 Si 1 (A) and Zr 50 Cu 30 Al 10 Ni 10 (B) laser melted surface layers. One limitation of metallic glasses is their lack of plasticity [15]; often induced through thin, sheet-like volumes in which very large strains can be concentrated, leading to the formation of shear bands. These shear bands have been formed in a number of our amorphous layers when subjected to indentation experiments, such as the microhardness test, shown in figure 9 with the related shear band formation. Averaged results for these found the Cu 47 Ti 33 Zr 11 Ni 6 Sn 2 Si 1 alloy to exhibit GPa hardness and elastic modulus to be GPa, giving H/E values between and through the depth of a coating. 4 Conclusions The production of metallic glasses by melt-spinning has been proven to be a successful and valuable source of reference in the quest for metallic glass coatings. Fundamental, but important information can be gained from the ribbons in terms of thermo-dependant, microstructural and some mechanical characteristics. The possible self-healing ability of amorphous Cu 47 Ti 34 Zr 11 Ni 8 at

10 170 Computer Methods and Experimental Measurements for Surface Effects and Contact Mechanics VII elevated temperatures has been proposed, and the high hardness of laser remelted amorphous layers is advantageous in many industrial applications. Acknowledgements We acknowledge the contribution of Dr. P. Svec and Dr. D. Janickovic at the Institute of Physics, Bratislava for their assistance in preparing the melt-spun ribbons and their thoughtful discussions. The authors also acknowledge financial support from the NIMR and FOM-Utrecht. References [1] Perepezko J. and Herbert R., Amorphous aluminium alloys synthesis and stability, Journal of Materials, March 2002 pp , (2002). [2] Klement W., Wilens R.H. and Duwez P., Nature 187, p. 869, (1960). [3] Telford M., The Case for Bulk Metallic Glasses, Materials Today March 2004, pp (2004). [4] Xu D., Duan G., and Johnson W., Unusual glass-forming ability of bulk amorphous alloys based on ordinary copper metal, Physical Review Letters, 92, (2004). [5] Ocelík V., Matthews D., and De Hosson J. Th. M., Sliding wear resistance of metal matrix composite layers prepared by high power laser cladding and laser melt injection, Surface and Coatings Technology, Article in Press (2005). [6] Vreeling J., Ocelík V., and De Hosson J. Th. M., Ti-6Al-4V strengthened by laser melt injection of WC p particles, Acta Materialia 50 pp (2002). [7] Pei Y.T., Ocelík V. and De Hosson J. Th. M., SiC p /Ti6Al4V Functionally Graded Materials Produced by Laser Melt Injection, Acta Materialia, 50 pp (2002). [8] Aubert F., Colaco R., Vilar R., and Sirkin H., Production of glassy metallic layers by laser surface treatment, Scripta Materialia 48 pp (2003). [9] Akamatsu H. and Yatsuzuka M., Simulation of surface temperature of metals irradiated by intense pulsed electron, ion and laser beams, Surface and Coatings Technology, Proceedings of Frontiers of Surface Engineering, pp (2003). [10] Pei Y.T., Ocelík V., and De Hosson J. Th. M., Interfacial adhesion of laser clad functionally graded materials, Materials and Engineering A 342, pp , (2003). [11] Johnson W., Bulk glass forming alloys: Science and Technology, MRS Bulletin October 1999, pp 42-56, (1999). [12] Tanner L. and Ray R., Metallic glass formation and properties in Zr and Ti alloyed with Be-I the binary Zr-Be and Ti-Be systems, Acta Metallurgica 27 pp (1979).

11 Computer Methods and Experimental Measurements for Surface Effects and Contact Mechanics VII 171 [13] X. H. Lin and W. L. Johnson, Formation of Ti Zr Cu Ni bulk metallic glasses, Journal of Applied Physics, Volume 78, Issue 11, pp , (1995). [14] de Oliveira U., Ocelík V., De Hosson J. Th. M., Analysis of coaxial laser cladding processing conditions, Surface and Coatings Technology, Article in Press (2005). [15] Schoers J. and Johnson W., Ductile Bulk Metallic Glass, Physical Review Letters, , (2004).