Surfaceand Coatings Technology83 (1996) 250-256
High current density, low energy, ion implantation of AN-M2 for tribological applications
P.J. Wilbur a,* , J.A. Davis a, R. Wei b, J.J. Vajo b, D.L. Williamson ’ a Department
of Mechanical Engineering, Colorado State Univemity, Fort Collins, CO S0523, USA b Hughes Research Laboratories, Malibu, CA, USA ’ Colorado School of Mines, Golderl, CO, USA
Abstract The effectsof the temperaturesof fully hardened and mill-annealed AISI Iv12 steels during N implantation at low energy and high current density on their hardnesses,sliding wear behaviors, microstructures and N concentration profiles \sere investigated. It is shownthat N implantation at 1 keV and 2 mA cm-’ can increase the hardness of the steel in either condition, but does not improve the wear resistanceof the fully hardenedmaterial. Implantation of the mill-annealedmaterial, however, yields a surface that is aswear resistantasthe fully hardenedsurface.Strengtheningand hardeningare inducedthrough enrichmentof the matrix that bonds hard (Fe,Cr),(W,Mo),C and V4C3particles with an iron nitride phasethat appearsto be predominantly +Fe,+,N. The implantation temperaturethat yields the greatestsurfacehardnessincreasein either state is near 450“C. The exponential prefactor and activation energy associatedwith N diffusion into heat-treatedM2 steelare approximately 1.4x 10s5cm2s-l and approximately 0.70eV respectively.In order for ions to be implanted,it is pointed out that their energymust be great enoughto
carry them through any barriers that may exist on a surface. It is shown that atomic B from an ion source can condense on a surfaceand produce a surfacebarrier that inhibits subsequentB+ implantation at the nominal N implantation energyof 1 keV. Keywords: Low energyimplantation; M2 tool steel
1. Introduction Nitrogen ions implanted under elevated temperature, high current density, low energy conditions yield treated layers on 304 stainless steel that are both thicker and -potentially less expensive than those produced using traditional N-implantation parameters [ 11, These improvements, which are critical to widespread acceptance of implantation technology in tribological applications, are realized because the elevated temperature enables thermal diffusion, and the high current density, low energy operation enables rapid processing with less expensive equipment. In order to increase the acceptance of this technology, however, its viability with other substrate materials and implanted ions needs to be demonstrated. This paper addresses this need for the case of a widely used tool steel (AISI M2) implanted with N. Implantation with B is also considered briefly. Conventional N implantation (i.e. at high energy, low current density and near-ambient temperatures) has been reported to improve the sliding wear performance of * Corresponding author. 0257-8972/96/$15.00 0 1996ElsevierScience S.A.All rightsreserved 00~T-~r~-~clinr>n*nIn I
tool steels in some cases and not in others [2-51. Hence, there is no clear extrapolation suggesting that hardness and wear improvements can be expected in this study. A single study conducted under elevated temperature, low energy conditions did, however, demonstrate substantial improvements in both the wear and friction characteristics of a tool steel similar to that used here . Boron is considered worthy of study because (1) it is less studied than N and (2) conventional B implantation to sufficiently high doses into ferrous materials yields iron borides  that are known to be wear resistant in the form produced by the boron&g process [S].
apparatus and procedures
Blocks that were to be ion implanted, wear tested and subjected to surface analysis were fabricated from AISI M2 steel (4.4 Cr, 3.0 MO, 2.3 V, 1.9 W, 4.1 C inat.%) in the as-received (mill-annealed) condition. After the blocks had been machined to specification (10.16 mm x 6.35 mm x 15.75 mm)  some were kept
P.J. Wilbur et al./Swface
and Coatings Technology 83 (1996) 250-256
in their as-received (AR) metallurgical state and some were heat treated (HT). The sequenceof heat treatment conditions was as follows: 1.25h preheat in Hz at 840 “C, 1.25 h austenization in H2 at 1180 “C, oil quench, 4 h temper at 530 “C in H,, air cool, 3.5 h temper at 530°C in HZ, air cool. Both the AR and HT blocks were wear tested in a block-on-ring tribometer  against 35 mm diameter, M2 steel rings that had been fully heat treated using the same procedure as used for the HT blocks. The heat treatment yielded a hardness of R, 62.5 on both rings and blocks. Final finishes on the contacting surfaces of the blocks and rings were 0.05 and 0.3 pm respectively. Implantation was accomplished on a 6.35 mm x 15.75mm block surface using either a broad-beam gaseous ion implanter [lo] for N or a metal ion implanter [ 111 for B. Rings were not implanted. Ions from the gaseous implanter are molecular (approximately 70%) and atomic (N’ approximately 30%). Ionic B is assumed to be monatomic. For all of the data presented herein, the ion energy, current density and implantation time were held constant at 1 keV, 2 mA cm-‘, and 40 min respectively. In order to operate the metal ion implanter at these conditions, the ion extraction grids were modified by reducing the grid separation (approximately 1 mm) and using an array of holes (2 mm diameter) that were much smaller than that used in the original design. Further, the surface of a block was about 5 cm downstream of the ion extraction grids rather than about 50 cm as it was in the original design. During implantation, a block was held at an implantation temperature, that was constant to within approximately 5 “C, of 400 “C under the prevailing beam heating condition by either heating or cooling the structure on which the block was mounted. The implantation temperature was sensed by a thermocouple attached to the block within approximately 2 mm of the surface being implanted. Heat-up to the implantation temperature was accomplished by simultaneously sputter cleaning the surface in a 3 mAcm-‘, 1 keV argon ion beam and using a mounting-plate heater. This typically required about 15 min. The initial cool-down rate after implantation, which was typically around 100 “C min-r, was achieved by water cooling the mounting plate or, in the case of blocks implanted with nitrogen at the higher temperatures: by immediate venting of the vacuum using liquid N, directed onto them. Sliding wear tests were conducted using a block-onring tribometer operating with the ring rotating at 300 rev min- 1 (0.55 m s-l relative speed) and a normal load of 222 N (271 MPa Hertzian stress) between the block and ring. All tests were conducted using a water-soluble boundary lubricant used for machining (long-life 20/20T”, diluted 2O:l with water). Under these conditions, friction coefficients were all generally near 0.12. The test duration was 1 h and the volume of material
removed from each block was determined using the standard technique  based on three measurements of the width of the wear scar. Independent mass-lossmeasurements yielded volumes that agreed within about 10%. Tests were repeated up to four times at some temperatures to demonstrate reproducibility and uncertainty in the results. Periodic removal and wear-scar measurement on several blocks treated in various ways suggested the wear rates are relatively uniform throughout the tests. The microhardness of a block was established using a minimum of ten measurementswith a Vickers indenter at a load of 50 g and a dwell time of 10 s. At this load, the indentation depth for a typical implanted surface (approximately 1 pm) was in most casesmuch less than the thickness of the treated layer, so the hardness could be considered representative of the layer itself. Block microstructures were investigated using a conventional X-ray diffractometer operating with Cu Ku (8 keV) X-rays in the Bragg-Brentano configuration. Under these conditions, the resulting spectra should be representative of l-3 pm thick surface layers. Nitrogen concentrations were measured as a function of depth using Auger electron microscopy (AES) coupled with sputter erosion from the surface (to approximately 3 pm) or block cross-sectioning coupled with AES analysis across the exposed edge of the implanted layer (to approximately 25 urn). Layer thicknesses were determined both from the AES data and visually in a microscope after nital (1.5% HNO,) etching blocks that had been crosssectioned and polished.
3. Results 3.1. Hardness The effectsof implantation temperature on the microhardness of as-received and heat-treated surfaces after implantation with nitrogen ions at a current density of 2 mA cmB2 and an energy of 1 keV for 40 min are shown in Fig. 1. The data show that N implantation induces four- to five-fold and approximately 60% increases in hardness of the as-received (AR) and heat-treated (HT) materials respectively, at the 400 to 450 “C condition. Above 450 “C the hardness of both materials drops off and although the AR steel appears to be slightly harder for higher temperature implantation, differencesin hardness between the two are within the experimental error. Hardnesses measured in the bulk material (i.e. at a surface other than the implanted surface) of the same HT blocks that yielded the data of Fig. 1 are shown in Fig. 2. These data indicate that significant annealing (transformation of tempered martensite) begins to occur at about 500 “C and induces a substantial loss of hardness in the 600 to 650 “C range. This transformation could explain the loss in hardness above around 500 “C
P.J. Wilbur et al,/Surfaee and Coatings Technolog) 83 (1996) 250-256
INDENTER LOAD - 0.5 N SOI-ID SYMBOLS - UNIMPLANTED opt :N SYMBOLS - N IMPLANTED 1 keV 2 mA/cm2 40 min
Fig. 1. Effect of implantation temperature on surface layer hardness,comparison of heat-treated and as-received M2 steels.
TIME AT TEMPERATURE
550 PROCESSING TEMPERATURE (C)
- 40 min
Fig, 2. Effect of processing temperature on bulk hardness for heat-treated M2 steel.
shown in Fig. 1 for the HT blocks. Since the AR (millannealed) blocks also lost hardness in this temperature range but had not been austenized, quenched and tempered, this explanation does not address their behavior.
3.2. Sliding wem The effectsof implantation temperature on the sliding wear rates of the same blocks that yielded the data of
P. J. Wilbur et al./Surface and Coatings Technology 83 (1996) 250-256 NORMAL LOAD - 220 N SLIDING VELOCITY - 55 cm/s BOUNDARY LUBRICATED SOLID SYMBOLS - UNIMPLANTED OPEN SYMBOLS - N IMPLANTED 1 keV 2 mA/cm* 40 min
450 400 500 IMPLANTATION TEMPERATURE
Fig. 3. Effect of implantation temperature on block wear rate, comparison of heat-treated and as-received M2 steels.
Figs. 1 and 2 are compared for the HT and AR materials in Fig. 3. Wear rates are computed as the volume of metal removed from the block (mm”) per unit normal load (FN in N) per unit sliding distance (m). The figure shows that wear rates that are very different for the unimplanted materials (solid symbols) become essentially the same above a temperature of about 400 “C!. In fact, the HT material exhibits the same wear rate within experimental error over the full temperature range, in spite of the fact that the hardness varies by 60%. Corresponding counterface (ring) wear rates were all near 2 x lo-’ mm3 N-l m-l, regardless of the block treatment. Regarding the data of Figs. 1 and 3, it is important to note that a relatively brief (40 min) ion implantation process yields a surface on the AR steel with wear and hardness characteristics that appear to be as good as those achieved using conventional heat treatment. By comparison, however, heat treatment requires several processing steps and an order-ofmagnitude greater time than implantation. 3.3. Near-surface analysis Insight into the hardness and wear behavior shown in Figs. 1 and 3 can be obtained by considering X-ray diffraction (XRD) and AES data from the blocks that produced this behavior. Fig. 4 compares XRD spectra for the HT blocks that were implanted at the various temperatures with the spectrum for the unimplanted material. The unimplanted material shows diffraction peaks representing b.c.c.-ferrite/b.c.t.-martensite (a), f.c.c.-
Fig. 4. X-ray diffraction spectra for heat-treated M2 steel. Diffraction peaks of the following phases are indicated: v: b.c.c.-ferrite/b.c.t.martensite, 4 f.c.c.-(Fe,Cr),(W,Mo),C, V Ec.c.-V,C,, E hex-Fe,N (dashed lines).
P. J. Wilbur et al./Swface
and Coatings Technology 83 (199G) 250-256
metal carbides (Fe,Cr),(W,Mo),C (q), and f.c.c.-V,C, (V). These carbide phasesappear to remain unchanged by implantation at any of the temperaturesinvestigated. The spectraof the implanted blocks clearly show through intensity changes,however, that the cc-phasenearly disappearsand a hexagonal E-Fe2+.Nphase (E)with x M 1 appears as the implantation temperature increasesinto the 400 to 450 “C range. This is the range where Fig. 1 showed the HT material was hardest. As the implantation temperatureis increasedto 500°C and above,the data indicate more a-phase remains and less of the c-phaseis formed. Fig. 5 shows XRD spectra for blocks made from AR material.The unimplanted spectrumis similar to that for the HT material except for the fact that the a-peak is sharper. Broadening of this peak in the HT material is probably a consequenceof the b.c.t.temperedmartensite which would not be present in the matrix of the AR, CIferrite material. The effectof increasingthe implantation temperatureof the AR material shown in Fig. 5 is again to reduce the cl-ferrite content and to produce a peak near the &-Fe2+JN peak observedin Fig. 4. It is suggested that this peak is due to E-Fe2+,Nbut is shifted because substantialalloying of the c-phasewith other constituents
of the steelis occurring. The developmentof this shifted s-peak is seen clearly in the spectra measuredfor the 400 “C case and for the 450 ‘C for 13 min case.It is noteworthy that increasesin implantation temperature above 450 “C appear to causelesstransformation of the c-phase to the a-phase for AR material than for HT material (compare the 500 “C spectra in Figs. 4 and 5). This is consistentwith increasedthermal stability of the E-phasedue to alloying of a-iron that has been demonstrated recently [ 121. Finally, it is noted that the Fez03 peak on the 500 “C spectrum of Fig. 5 is probably a consequenceof surface oxidation that occurred when liquid nitrogen containing some oxygen was directed onto the block to cool it after implantation wascomplete. Nitrogen concentration profiles measuredin the HT material implanted at the various temperaturesare plotted in Fig. 6. They indicate the concentrationsare in the 10 to 22 at.% range within a few micrometers of the surface.At temperaturesof 450 “C and above, the concentrations are all near 10 at,% and they extend deeper beneath the surface as the temperature is increased, reaching over 20 llrn for the layer implanted at 600 “C. Becausethe N concentrations in Fig. 6 are generally below 20 at.%, one expects f.c.c.y’-Fe4N to form. No evidence of this phase is observed in the XRD data, however, even for the higher temperature caseswhere concentrationswithin the 1 \tm X-ray diffraction probing thickness are lower. During a typical 1 h wear test the rotating ring surfacewore approximately 10 pm into the block surface,so the layer thicknessinformation of Fig. 6 suggestsa significantfraction of test durations on blocks implanted at lower temperaturesmay have involved the wearing of some steel that contained negligible N. 3.4. Boyon implantation
Fig. 5. X-ray diffraction spectra for as-received M2 steel N implanted at 1 keV, 2 mA crnm2for 40 min, unless noted otherwise. Diffraction peaks of the following phases are indicated: x b.c.c.-ferrite/b.c.t.martensite, 11f.c.c.-(Fe,Cr),(W,Mo)& V Ec.c.-V,C,, o Fe,OB, E hexFe,N (dashed lines).
Blocks wereexposedto a 1 keV, 2 mA cmB2boron ion beamfor timesranging to 40 min. SinceB and N exhibit similar diffusion behavior in x-ferrite [ 131, an implantation temperature of 450 “C was used. The results of subsequenttesting,which showedthe wear and hardness properties of treated surfaces were unaffected by the treatment, were initially surprising considering the fact that boronizing is known to be an effective process. Surfaceanalysis indicated that very little B penetrated beneath the implanted surface and this supported the wear and hardnessresults.Subsequentreflection has led to the hypothesisthat the surfacewas so closeto the ion sourceduring processingthat a visibleB layer condensed on it and preventedpenetration of the low-energyB ions. This was not a problem during low-energy N implantation becauseN is gaseousat prevailing surfaceconditions. Future B processingwill involve placementof the surfacebeing treated further from the ion extraction grids and increasing the ion energy to facilitate increased surfacesputtering and deeperion penetration. f
P.J. Wilbur et al./Surface and Coatings Technology 83 (1996) 250-256 0.25
IMPLANTATION CONDITIONS - 1 keV - 2 mA/cm’ - 40 min
10 DEPTH (pm)
Fig. 6. Nitrogen concentration profiles determined using Auger electron spectrography.
4. Discussion The concentration profiles of Fig. 6 do not appear to be consistent with Fick’s Law of diffusion, presumably because the diffusion parameters are concentration dependent. In spite of this, approximate diffusion parameters describing N migration in M2 steel can be estimated from the experimental data. Fig. 7, a semi-log plot of diffusion coefficient D vs. reciprocal temperature,
was obtained using AES data (solid symbols corresponding to 5% N-concentration points of Fig. 6) and microscopic observations (open symbols) of layer thickness (x). Diffusion coefficients were determined using the expression D = (~)~/(2t). The least square line fit in the figure suggests an exponential prefactor D, of 1.4 x low5 cm2 s-i and an activation energy E of (8116)(1.38 x 10-23)=1.1 x 10-l’ J=O.70 eV. Both of these values are below published values N IMPLANTATION TIME - 2400 set OPEN SYMBOLS - MICROSCOPIC OBSERVATION SOLID SYMBOLS - AES DATA
I 1.4 l/-T- (K-‘1
Fig. 7. Diffusion coefficient vs. reciprocal temperature plot for heat-treated M2 steel.
J x10-3 1.7
P.J. Wilbur et al./Surjace and Coatings Techtlology83 (1996) 250-256
(DO=3 x 10m3cm2 s-1,E=0.79eV)forNdiffusion into b.c.c. a-ferrite, which may be representative of the matrix of M2 steel. Observed increases in hardness of HT M2 steel with implantation temperature to about 450 “C (Fig. 1) are consistent with corresponding increases in &-Fe2+.,N layer thickness (Fig. 6). Decreasesin hardness for both HT and AR materials beyond about 500 “C (Fig. 1) are consistent with both decreasesin N concentration near the surface (Fig. 6) and general tempering (Fig. 2) that accompany such increases in temperature. It is noteworthy that Byeli et nl. have performed low-energy N implantation experiments on a similar HT tool steel and observed about a 60% increase in hardness and l/3 of the wear rate of the unimplanted material [ 61. Although these results seem inconsistent with the results of Figs. 1 and 3 at first glance, they are not, because they were performed on a steel that had been heat treated to a lesser initial hardness (8 GPa). The fact that the wear rate of the HT material does not change with the composition and thickness of the treated layer is worthy of comment. It is believed that this occurs because wear of this material is determined primarily by the rate at which hard carbide particles are dislodged rather than by the more gradual wearing of either these particles or the matrix material. It is argued, therefore, that the wear rate is unaffected because implantation of the HT material does not alter the bond strength between the matrix and the particles. Different wear test parameters (lubrication, load, relative sliding speed and counterface material) could change the wear mechanism and yield different behavior. Becausethe AR and HT blocks were both worn against the same HT rings, the results suggest that implantation of the AR material produces a matrix that bonds the carbide particles as well as the tempered martensite matrix of the HT material does. Microscopic examination of wear scars also suggesteda wear process that involves carbide detachment and plowing through the matrix materials.
5. Conclusions Low energy, high current density N implantation can
be used to make the surface of a widely used, millannealed tool steel as hard and resistant to sliding wear
as a much more complex and time-consuming heat treatment process.The strengthening and hardening are induced through enrichment of the a-ferrite matrix containing hard carbide particles with an iron nitride phase that appears to be predominantly r-Fez+,N. The implantation temperature that yields the greatest surface hardness is near 450 “C. Implantation of the steel in the fully hardened state also yields this nitride and although it increases the hardness of the material, it does not affect its wear resistance significantly. The exponential prefactor and activation energy associated with N diffusion into heat-treated M2 steel are 1.4 x 10B5cm2 s-l and 0.70 eV respectively. In order for ions to be implanted their energy must be great enough to carry them through any barriers that may exist on a surface. Atomic B from an ion source can condense on a surface and produce a surface barrier that inhibits B’ implantation at low energy.
Acknowledgement The financial support of the National Science Foundation under Grant CMS-9414459 for this work is gratefully acknowledged.
References  R. Wei, J.J. Vajo, J.N. Matossian, P.J. Wilbur, J.A. Davis, D.L. Williamson and G.A. Collins, SIU$ Coat. Teclrnol., 53 (1996) 235.  I.L. Singer, Mater. Res. Sot. Synp. Proc,, 27 (19S4) 5S5. [ 31 M. Iwaki, Mater. Sci. Eng., 69 (1985) 211.  F. Alonso, J.L. Viviente, J.I. OAate, B. Torp and B,R, Nielsen, Nd, lt?strlr/fl.Methods B, SO-81 ( 1993) 254.  S. Ohtani, Y. Mizutani and T. Takagi, Nucl. Instr’m Ali,thods B, 80-81 (1993) 336.  A.V. Byeli, 0-V. Lobodaeva, S.K. Shyky and V.A. Kukareko, !Year, 181-183 (1995) 632.  B. Rauschenbach, N~rcl.Imtrrw. Mefhorls B, SO-S1(1993) 303. [S] K. Budinski, Wear, 162-164 (1993) 757,  ASTM, Anrwal Book o~“disT,WSmrla&, Designation G77-83, v. 03.02, ASTM, Philadelphia, PA, 1983,p. 453. [lo] P.J. Wilbur and B.W. Buchholtz, Surface engineering using ion thruster technology, Paper 94-3235, 1994 (American Institute of Aeronautics and Astronautics), [ 111 P.J. Wilbur and R. Wei, Rec. Sci. Insttwn., 63 (1992) 2491.  M. Kopcewicz, J. Jagielski, G. Gawlik and A. Grabias, J. Appl. Phys., 78 (1995) 1312. [ 131 J. KuEera and K. StrBnskjr, Mater. Sci. Eng., 52 (1952) 1.