Developing Extreme Hardness (> 15 GPa) in Iron Based Nanocomposites D.J. Branagan * Idaho National Engineering and Environmental Laboratory, Idaho Falls, ID, 83415 and Yali Tang Argonne National Laboratory, Argonne, IL, 60439 ABSTRACT Through a unique methodology, novel nanocomposite microstructures were created in a bulk iron based alloy by first processing into a glass condition followed by devitrifying the glass through heat treating above the crystallization temperature. The ascrystallized microstructure was made up of three nanoscale phases; α-fe, Fe 23 C 6, and Fe 3 B phases. Vickers hardness testing revealed a maximum hardness of 16.2 GPa which is significantly harder than existing commercial steel alloys and hardmetals. Detailed structural studies uncovered two important factors which contribute to the development of this extreme hardness; reductions in microstructure scale to the nanometer regime and supersaturation of transition metal alloying elements significantly above their equilibrium solubility limits. Keywords: A) Glasses, Nanostructures, B) Hardness, E) Rapid Solidification * Corresponding Author 1
Introduction Recently, a new type of steel called Devitrified Nanocomposite Steel was reported which exhibits extreme hardness (> 15 GPa). 1 This class of steel was developed by first overquenching to form a metallic glass and then heat treating the glass precursor above its crystallization temperature to devitrify it into a multiphase crystalline microstructure. Since the solid state transformation occurs at low temperatures, diffusion is limited and since the glass is thermodynamically metastable, the driving force is extremely high. Thus, a very high nucleation frequency results with limited time for growth before impingement between neighboring grains resulting in the formation of nanoscale phases and novel nanocomposite microstructures. In this paper, a study was launched to expand on earlier work to explain the key factors in developing extreme hardness in iron based metal matrix nanocomposites. To facilitate this a simplified iron based composition was developed which exhibits sufficiently reduced critical cooling rates for metallic glass formation to allow the formation of metallic glasses during melt-spinning. The design of the alloy was partially modeled on the class of materials called hardmetals (i.e. cemented carbides) which consist of refractory carbides (mainly WC) which are held together (i.e. cemented) by a tough binder metal. The vast majority of hardmetals contain from 75 to 97 wt% WC bonded in a cobalt matrix. 2 Cobalt is usually used as the binder due to the high solubility of WC in cobalt giving it a good ability to wet and join high fractions of WC. Iron is not commonly used to bind the carbides due to their much lower solubility in iron resulting in an ineffective ability of iron to wet and hold the large fractions of carbides found in commercial hardmetals. In this study, iron is used as the base element since similar (or higher) levels of hardness may be achieved with vastly reduced amounts of carbide through this unique nanotechnology approach due to simultaneous reductions in microstructural scale and optimization of the second phase distribution. EXPERIMENTAL APPROACH The alloy composition designed for the studies described in this paper had the following atomic stoichiometry; (Fe 0.8 Cr 0.2 ) 79 B17W 2 C 2. Its glass forming ability was based on previous studies which indicated that its critical cooling rate was sufficiently low to 1,3 enable glass formation by melt-spinning under the processing conditions used. The amount of W and C additions in the alloy was determined by the maximum reported 4 solubility (7.95 wt%) of WC in iron. The targeted alloy was processed from high purity constituents (>99.9%) into two sample forms, ingots and ribbons. By arc-melting in ultra high purity (UHP) argon on a water cooled copper hearth, ingots were formed. Small 5 gram ingots were used for ingot studies and larger 15 gram ingots were produced for melt-spinning. Melt-spin ribbons were formed by first induction melting, followed by ejecting the liquid melt onto a rapidly moving copper chill wheel. The melt-spinning parameters used were chamber atmosphere of 1/3 atm helium, 15 m/s tangential wheel velocity, melt-ejection temperature of 1400 C, and an ejection pressure of 150 torr. Heat-treating was done on the ribbon and ingot samples in a radiating vacuum furnace -6 at 10 torr. The samples were heated from room temperature at a 120 C/min heating rate to the heat treatment temperature, held for a constant 1 hour annealing time, and 2
then furnace cooled. Differential thermal analysis (DTA) and Differential Thermal Calorimetry (DSC) of the ribbon samples was performed in a Perkin Elmer DTA-7 from 30 C to 1375 C at a heating rate of 10 C/min in a 50 ml/s flowrate of UHP Ar. X-ray diffraction was carried out on powdered samples after incorporation of a silicon standard using a Bruker X-ray Diffractometer with filtered Cu-K α radiation and a LiF monochromometer to reduce fluorescence. Analysis of the experimental X-ray patterns was done by Rietveld analysis using SIROQUANT V 2.0 software and the calculated patterns were refined until the total χ 2 was less than 2.5 over the entire two-theta range (30-85 ). Vickers microhardness measurements were done with a 100 g load on metallographally mounted cross sections of the ingots or ribbons using a Leco M-400- H1 System. For each sample, 10 hardness indentations were made and then the average of the measurements was reported. The average standard deviation for the hardness measured on the ingots was 0.75 and for the measurements of the ribbons the standard deviation was 0.35. Transmission Electron microscopy (TEM) was performed on a Philips CM30T analytical electron microscope attached with an EDAX energy dispersive spectrometer (EDS) using an ultra thin window detector. TEM samples were first dimpled from the free side of the ribbons to reduce the thickness to 20 µm and then ion milled by a Gatan PIPS low angle ion miller from two directions. For composition analysis using EDS, a separate spectrum was collected for the beam directed through the hole of each thin specimen and then this hole count was subtracted from the unknown and standard spectra to minimize effects of spurious X-ray generation within the AEM. Peak intensities were converted to compositions using the Cliff Lorimer approach, I A /I B =k AB *C A /C B where B is the base element (iron in this case) for which the k-factors are determined. 5 Convergent beam electron diffraction (CBED) was employed to analyze the structure and identify the nano-sized phases. To determine the average phase size, 50 measurements of each phase was done on printed photographs by taking the average of the perpendicular chord lengths. RESULTS Hardness measurements were taken on the cross sections of the as-solidified and heat treated ingot and ribbon samples (Table 1). An example hardness indentation on the cross section of a melt-spun ribbon (700 C sample) is shown in Figure 1. Note that in spite of the high hardness, cracking was not observed in any of the measurements (110 total) taken. The as-cast ingot exhibited a relatively high hardness of 10.3 GPa which became softer after heat treatment. The as-spun ribbons exhibited a hardness of 11.0 GPa which arises from the unique short range order as a result of their as-solidified amorphous structure (see section below). After devitrification, the hardness of the ribbons could be increased considerably and reached a maximum of 16.2 GPa after the 700 C heat treatment. Interestingly, the hardness level eclipses any reported hardness from existing commercial steels or hardmetals. A significant drop in hardness was observed in increasing the heat treatment temperature from 700 C to 750 C with a smaller drop in hardness observed at higher temperatures up to 850 C. Based on the hardness results, the microstructure of selected samples was studied in detail using electron microscopy and X-ray diffraction. The microstructure of the as- 3
solidified ingot consists of three phases with an average phase size 4 µm (Figure 2). Note that a gradient in phase size was found with the finest grains near the chill surface and the largest grains near the free surface of the ingot. The three phases were identified using X-ray diffraction and were determined to be: cubic α-fe with a Im3m space group, cubic Fe 23 C 6 with a Fm3m space group, and tetragonal Fe 3 B with a I-4 space group. A summary of the lattice parameters is given in Table 2. The as-spun ribbons were found to be nominally 100% amorphous as shown by X-ray diffraction and DTA/DSC thermal analysis. The DTA scan showing the glass to crystalline and melting peaks in the as-spun alloy is shown in Figure 3. At the onset temperature of 536 C, the glass was found to exhibit a high enthalpy of crystallization ( 119 J/g) and crystallized at a very rapid rate (transformation rate 0.019/s). Based on the hardness results, detailed TEM and X-ray studies were performed on three key heat treatments (700 C, 750 C, and 850 C). Due to the extremely high transformation rate (0.019/s) found in this alloy, it was expected that the as-crystallized microstructure would be nanoscale and because of the starting composition it should be multiphase. This was verified in the heat-treated microstructures of the selected ribbon samples as shown in the TEM micrographs of Figure 4. As the heat treatment temperature was increased, the microstructure was found to coarsen via normal grain growth with no abnormal growth observed. From X-ray diffraction studies on the heat treated samples, the same three phases were found in the crystallized ribbons as was found in the as-cast ingot (Table 2). A typical Rietveld refined scan with the Bragg diffraction peaks labeled is given in Figure 5 for the 700 C heat treated sample. Note that all of the phases had significantly different lattice parameters compared to their respective unalloyed binary phases indicating significant amounts of dissolved solute atoms. In the TEM, microdiffraction in conjunction with EDS was used to determine the identity of the individual phases and the same three phases were found in agreement with the X-ray results. Once identified, the individual phases were easily distinguishable due to their distinct appearance/morphology (Figure 6). The α-fe phase forms a distinctly mottled structure, the Fe 23 C 6 type phase forms a featureless smooth structure, and the Fe 3 B phase forms a heavily multi-twinned structure. Note that the Fe 23 C 6 and Fe 3 B phases are not known to form in their respective binary systems but can form as stable phases due to the presence of impurities or solute atoms. 6,7 Additionally, the EDS scans taken on the individual phases revealed that each phase also contained dissolved Cr, W, B, and C atoms. As the heat treatment temperature was increased, several important changes were found in the microstructure. There was only a small increase in phase size when comparing the 700 C and 750 C heat treatments but a significant increase after the 850 C heat treatment (Table 3). Additionally, consistent with the changes in lattice parameters, there were significant changes in solute content in each phase as the heat treatment temperature was changed (Table 4). Some of the biggest changes occurred in the narrow temperature interval between the 700 C and 750 C heat treatments where a redistribution of tungsten atoms into the Fe 23 C 6 phase and chromium atoms into the 4
Fe 3 B phase was found. Additionally, observations of the microstructure indicated that only slight changes in the volume fraction of each phase were occurring with heat treatment with a small decrease in the total amount of Fe 3 B and a slight increase in the amount of Fe 23 C 6 observed with increasing annealing temperature. DISCUSSION The microstructural studies clearly show that all of the phases which form are iron based phases. This result was in contrast to the original intent on the alloy design since no individual tungsten carbide (or tungsten boride) phases form. Thus the hardness mechanism in these nanocomposite steel alloys is clearly different than that found in cemented carbides which is based primarily on the hardness of the tungsten carbide. 8 An analysis of the data reveals two key factors resulting in the development of the extremely high hardness observed in this nanocomposite steel alloy when heat treated above its crystallization temperature and below 750 C. One important factor found in obtaining extreme hardness was reductions in microstructural scale to the nanometer level (< 150 nm). The X-ray studies revealed that the same three phases formed in the as-cast ingot as the crystallized melt-spun ribbons. However, the ribbons heat treated 700 C or below, developed much higher hardness (16.2 GPa) than found in the as-cast ingot (10.3 GPa), which could not be improved after heat treatment. Since the phase equilbria was the same, the difference in hardness can be related to the differences in phase size observed in the ingots. The hardest ribbon sample (700 C heat treatment) had phase sizes from 100 to 130 nm while the hardest ingot (as-cast) had an average phase size of 4 µm (4000 nm). Additionally, the effect on phase size can be observed in the ribbon samples which were heat treated at 750 C and higher. Above 750 C, there is a continuous decrease in hardness with increases in heat treatment temperature. While the changes in phase content were minimal, the TEM results (Table 3) clearly show that there is significant coarsening due to Ostwald ripening on increasing the heat treatment temperature from 750 C to 850 C. A second important factor is related to solid solution strengthening through nonequilibrium solid solubility. In the melt-spun ribbons, there is a large decrease in hardness observed as the heat treatment temperature is increased from 700 C to 750 C. From the measurements on particle size, it is found that there is insignificant particle coarsening ( 25nm) occurring at this temperature interval. However, the EDS results on the individual phases clearly show that there is significant redistribution of the transition metal elements occurring at 750 C (Table 4). While this temperature represents only 65% of the melting temperature, the diffusivity of the solute atoms is probably assisted greatly by the presence of the extremely high volume fraction of high diffusivity phase boundary paths. These observed changes in solute content indicate that during the extremely rapid crystallization, the three phase microstructure is extremely supersaturated in solute elements, since there is limited time for the atoms to diffuse and redistribute among the newly formed phases. Upon subsequent heat treatment, when the transition metal atoms acquire enough thermal activation energy, significant amounts of diffusion occur returning the dissolved elements to near 5
equilibrium solubility levels. Note that the small interstitial carbon and boron atoms should be able to diffuse at temperatures much lower than 750 C but at these low temperatures little change in hardness is found. This indicates that supersaturation of the transition metal elements and not the interstitial elements is the key factor in attaining the extreme hardness levels. CONCLUSIONS A multicomponent iron based alloy with a sufficiently low critical cooling rate for metallic glass formation was developed and processed into a metallic glass by melt-spinning. The amorphous precursor was then transformed via a solid / solid state glass devitrification transformation into a three phase nanoscale composite microstructure. The hardness levels obtained by this nanotechnology approach (16.2 GPa) was much harder than possible by conventional ingot casting (10.3 GPa). Analysis of the results revealed two important factors which contributed to the development of this extreme hardness: reductions in microstructure scale to the nanometer regime and supersaturation of transition metal alloying elements significantly above their equilibrium solubility limits. Due to the extreme hardness, this new class of nanocomposite steels may be useful for a wide variety of applications involving two and three body abrasive wear and may represent low cost alternatives for cobalt based hardmetals. REFERENCES [1] Branagan DJ, Devitrified Nanocomposite Steel Powder, Powder Metallurgy Alloys and Particulate Materials for Industrial Application, St. Louis, MO, 2000, ed. By David E. Alman and Joseph W. Newkirk, TMS, 111-122. [2] Santhanam AT, Tierney P, and Hunt JL, Cemented Carbides, Metals Handbook, 10 ed. Vol (2), Properties and Selection: Nonferrous Alloys and Special-Purpose Materials, ASM International, 1990. [3] Branagan DJ, Hyde TA, Sellers CH, and McCallum RW, Developing Rare Earth Permanent Magnet Alloys For Gas Atomization, J. Phys. D: Appl. Phys., 29(1996), 2376. [4] Exner HE, Physical and Chemical Nature of Cemented Carbides, International Metals Reviews, 4(1979), 162. [5] G. Cliff and G.W. Lorimer, J. Microsc. 103, 203 (1975). [6] Coehoorn R, De Mooij DB, and De Waard C, Melt-Spun Permanent Magnet Materials Containing Fe 3 B as the Main Phase, J. Magn. Magn. Mater., 80(1989), 101-104. [7] Honeycombe RWK and Bhadeshia HKDH, STEELS Microstructure and Properties, Halsted Press, New York, 2 nd ed., 1995, p192. [8] Boyer Howard E and Gall Timothy L, Metals Handbook Desk Edition, American Society for Metals, Metals Park, Ohio, 1985. ACKNOWLEDGMENTS Research was supported through the Defense Advanced Research Projects Agency (DARPA) under the DOE Idaho Operations Office Contract No. DE-AC07-99ID13727. TEM studies were done at the Electron Microscopy Center at Argonne National Laboratory and the authors thank Dean Miller, director of the TEM center for his support 6
in the use of these facilities. Argonne is operated by the University of Chicago under Contract No. W-31-109-ENG-38. This document has been approved for Public Release, Distribution Unlimited. 7
Table 1 Hardness (GPa) of the As-Solidified and Heat Treated Samples Sample As-solidified 600 C 650 C 700 C 750 C 800 C 850 C Ingot 10.3 8.0 --- 6.6 --- 6.5 --- Ribbon 11.0 15.6 15.6 16.2 12.2 12.0 10.5 Table 2 Lattice Parameters from X-Ray Diffraction Sample α-fe Fe 23 C 6 Fe 3 B Ingot as-cast a = 2.870 A = 10.641 a = 8.584 c = 4.342 Ribbon 700 C a = 2.870 A = 10.627 a = 8.638 c = 4.283 Ribbon 750 C a = 2.869 A = 10.625 a = 8.619 c = 4.312 Ribbon 850 C a = 2.871 A = 10.634 a = 8.725 c = 4.221 Table 3 Average Phase Sizes (nm) in the Heat Treated Melt-Spun Ribbons Heat Treatment α- Fe Fe 23 C 6 Fe 3 B 700 C 130 120 100 750 C 150 140 150 850 C 240 240 260 Table 4 EDS Results on the Phases found in the Melt-Spun Ribbons Heat Treatment Phase At% Fe At% Cr At% W α-fe 700 C 94.6 4.3 1.1 750 C 95.8 3.3 1.0 850C 95.9 3.4 0.8 Fe 3 B 700 C 76.3 19.2 4.5 750 C 65.7 30.7 3.6 850C 65.2 32.8 2.0 Fe 23 C 6 700 C 72.0 22.1 5.8 750 C 65.5 25.3 9.3 850C 74.7 16.2 9.2 8
Figure 1 Example hardness indentation on the cross section of a melt-spun ribbon which has been heat treated at 700 C. Notice the very uniform indentation and absence of cracking at the cube corners. Figure 2 A backscattered SEM micrograph of the as-cast ingot. Three distinct phases are found with an average microstructure scale of 4 µm. 9
Figure 3 DTA scan of the as-spun melt-spun ribbon. The extremely sharp glass to crystalline peak and melting peaks are identified on the diagram. Figure 4 TEM micrographs and the corresponding selected area diffraction (SAD) patterns for the ribbons heat treated at 700 C (left), 750 C (middle), and 850 C (right). Note the increase in phase size observed in the microstructure and the SAD patterns. 10
Figure 5 Experimental and Rietveld calculated patterns for the ribbon sample heat treated at 700 C. Note that silicon was added to the powdered sample as a standard to adjust for instrumental two-theta errors. Figure 6 TEM micrograph of the DAR18 sample heat treated at 850 C for 1 hour. The three phases in the microstructure are labeled with their corresponding CBED patterns. 11