Investigation of plasma immersion ion implanted surfacesby instrumented indentation R. Hutchings, K. T. Short, J. Tendys Australian Nuclear Science and Technology Organisation, Lucas Heights Reseal& Laboratories, Private Mail Bag 1, Menai, NSW2234, Australia
Abstract Plasma immersion ion implantation (P13) involves the energetic implantation of nitrogen at energies up to 50 keV combined with the thermochemical absorption of nitrogen from a low pressure r.f. plasma at temperatures up to 550 “C. At low temperatures the treated surfaces are similar to those obtained by conventional ion beam implantation, but at temperatures where the mobility of nitrogen is significant, diffusion becomes important. In this case, thicker treated layers are obtained which give greater improvements in mechanical properties to depths of several micrometres. The development of instruments which measure continuously the force and displacement as an indentation is made in a material has led to great advances in probing the mechanical properties of materials on the submicrometre scale. We have used such an instrumented indenter at loads up to 500 g, in order to characterize the surface layers produced by PI3 in a range of alloys. Results are given for standard test blocks, demonstrating the comparability of hardness values derived from force vs. displacement data with those measured by the conventional Vickers technique. It is also shown that the instrumented indenter can be used to investigate the variability of the hardness profile across an engineered surface. Specific examples are given for an austenitic stainless steel treated at elevated temperatures. Keywords: Plasma immersion; Hardness; Stainless steel
1. Intrcduction The process of ion implantation appeared on the metallurgical scene in the 1970s as a result of the pioneering work at Harwell in the UK [l] and the Naval Research Laboratory in the USA . This work was driven by the desire for improved wear resistance in industrial tools and the initial focus was on the implantation of the nitrogen into steelsusing accelerating voltages of around 100 kV and at temperatures of 100 “C or lower, resulting in penetration of the ions to depths of the order of 0.1 pm. Metallurgical implantations were performed in the early days by using accelerators, isotope separators or semiconductor ion implanters. However, for nitrogen implantation the emphasis switched to the development of simpler units based on gaseousion sources providing an unanalysed beam of ions . More recently plasmabased processeshave been developed [4,5] as an alternative to ion beam implantation, initially with the view of overcoming problems associated with the line-of-sight nature of conventional implanters . In these newer techniques the components to be treated are immersed 0257-8972/96/$15.00 0 1996ElsevierScienceS.A.AU rights reserved
in a nitrogen plasma and biased to a high negative voltage. Ions from the plasma are accelerated towards the target and implanted approximately uniformly over all the exposed surfaces. A number of methods have been used to characterize the thin layers produced by implantation or comparable coating techniques . The need to investigate the mechanical behaviour of the very thin layers produced by ambient temperature implantation resulted in the development of ultra-sensitive microhardness testers capable of measuring hardness at very shallow penetration depths [ 81. These instruments work on the principle of measuring load vs. displacement as a diamond indenter is driven into the surface of a specimen, and have been used to explore hardness and wear resistance correlations in nitrogen implanted alloys . More recently such instrumented indentation devices have been used to investigate not only hardness but also elastic behaviour [lo] and fracture properties [ 111. However, the major emphasis has continued to remain on the use of very light applied loads in the ultramicrohardness, or nanohardness, range. Although plasma-based implantation techniques were
et al./Surfnce and Coatings Techology
initially developed as a direct alternative to conventional ion implantation, unique features have emerged. For example, in plasma immersion ion implantation (or P13) an r.f. generated plasma is used with accelerating voltages of up to 50 kV and, under normal conditions of operation, the applied bias produces a stable sheath with a width of a few centimetres . However, the high voltage is pulsed because of the large current densities drawn from the plasma, enabling the treatment temperature to be controlled in the range 150-600 “C by regulating the pulse frequency, or duty cycle. Elevated temperature implantation brings a significant advantage for a number of materials. Early trials with both mild steel and austenitic stainless steel [ 131 showed diffusion of nitrogen well beyond the implanted range at temperatures above 250 “C. Improvements in surface hardness achieved at these elevated temperatures could be easily shown by conventional microhardness testers. Such “macroscopic” hardening effectsare produced by the formation of a nitrogenstrengthened diffusion zone underlying the implanted layer. It is now apparent that the PI3 process offers the possibility of producing surface structures typical of ion implantation, plasma nitriding or a combination of both, depending on the treatment temperature. The mechanical properties of the specific structures produced clearly require further investigation in order to optimize the application of the PI3 process. This paper presents the initial results obtained using an instrumented indentation unit capable of applying loads of up to a mass of 500 g to study the response of an austenitic stainless steel to plasma immersion ion implantation at elevated temperatures.
83 (1996) 243-249
the desired temperature. A standard temperature controller was used to regulate the pulse frequency to maintain the samples at the selected treatment temperature. The surface temperature of the sample was monitored continuously by an IR pyrometer viewed through a quartz window. The treatment time at temperature was varied between 30 min and 5 h. It should be emphasized that the implantation dose over a specific time (i.e. the integrated high voltage nitrogen ion current delivered to the samples) depends on the treatment temperature. This means that samples treated for 3 h at 520 “C received 2.3 times the high voltage implantation dose that the samples treated at 350 “C! received.
3. Instrumented indentation testing at loads up to 5 N The experiments presented in this study were performed using a Nano IndenterTb’ 11s (registered trademark of Nano Instruments Inc., Knoxville, TN) equipped with a high load head and a Berkovich diamond indenter. The instrument is depicted schematically in Fig. 1. Indentation experiments were performed by first positioning the region of interest on a sample under the indenter tip using a motorized X-Y stage. The sample was then raised by a motorized 2 drive until contact
High LoadHead MagnetandCoil
High LoadCoil ConstantCurrentSource
II M (1
Springs CapacitiveDisplacement SensingSystem
2. Elevated temperature treatment by plasma immersion ion implantation As mentioned above, PI3 treatment enables a combination of ion implantation and thermochemical processesto be applied simultaneously. This duplex process has been found to be particularly effective for the hardening of austenitic stainless steels, without compromising the important corrosion resistance of these alloys [ 141. Samples were taken from a range of specimens which had been prepared for detailed microstructural studies which have been previously reported elsewhere [ 151. PI3 treatment was carried out in an industrial prototype facility which represents a development of the system described previously [ 161, but with an improved pumping system and a complete process control system [ 171. Samples were treated at a filling pressure of 0.13 Pa in a plasma of density 5 x 1Ol5 ions mW3. High voltage pulses of -45 kV amplitude and 100 ps duration were applied to the samples, initially at a frequency of 200 Hz for 15-25 min in order to heat the specimensto
I” I SSIlPlC
X Y 2 MotorisedTable VibrationIsolationBase
1piTzq 5 l/2 Digit DVM
Signal Multiplexer _
Fig. 1. Schematic representation of the instcumcnted indentation equipment.
R. Hutchings et al./Swface
and Coatings Technology 83 (1996) 243-249
was detected between the indenter tip and the sample surface. The indenter was then driven into the sample by applying a controlled force. This loading force is generated by changing, in a precise and controlled manner, the current passing through a coil located in the bore of a permanent magnet. The coil is connected to the indenter tip via a shaft supported by leaf springs which restrict lateral movement but allow vertical displacement with a low, reproducible stiffness. Tip displacement is monitored by a parallel-plate capacitive sensor, one plate of which is attached to the indenter shaft. The total vertical displacement of the tip is limited to approximately 200 urn, with a central linear region of plus or minus 20 urn within which measurements are made. The instrument was capable of applying controllable loads of less than 1 mN to in excess of 5 N using a programmable linear loading rate, while monitoring simultaneously the indenter tip displacement with a resolution of better than 0.5 nm. All operation of the instrument was controlled by a data acquisition and control computer, which could perform a series of preprogrammed indentation experiments without operator intervention. To achieve the high level of resolution, the indenter stage was mounted on a vibration isolation table, and located in a stable temperature enclosure. The resultant indentations were viewed using a microscope mounted above the indenter stage. For each indentation experiment, the tabulated tip displacement as a function of applied load can be used to determine various mechanical properties of the sample. The raw data provided by the respective sensors was analysed for each indentation as follows. (a) The voltages were first converted to physical units using the calibration constants for the detectors. The measurements of the load were then corrected for the stiffness of the leaf springs which support the indenter assembly, while the measured depth was corrected for the compliance of the load frame. (b) The projected area of the diamond in contact with the material was calculated from the depth of the indentation below the surface, using the shape function for the indenter. For this calculation to be performed accurately, it was found to be essential to know accurately the point at which the indenter contacts the surface. Advantage was taken of the fact that the stiffness increases rapidly after surface contact and, for steel, it was possible to determine the position of the surface to within several nanometres. Once the point of contact was located, the load at each depth below the surface was found, thereby enabling the curves of loading and unloading to be calculated as a function of depth for each indentation. Fig. 2 shows typical loading and unloading curves for maximum loads of 0.25 and 1.0 N. (c) The hardness was then calculated from the relation-
Depth( m 1 Fig. 2. Typical loading and unloading curves for maximum loads of 250 and 1000 mN.
ship that hardness is equal to load divided by projected contact area, where the load is in newtons and the contact area in square metres. As mentioned in (b) above, the area is derived from the penetration depth and a knowledge of the tip shape function. Two common models have been used for this calculation [lo]: a “plastic model” in which it is assumed that all the deformation is plastic and no recovery occurs as the indenter is withdrawn; an “elastic correction model” in which it is assumedthat some elastic recovery occurs and a net depth due to plastic deformation alone is calculated. Following the method of Oliver and Pharr, the correction is derived from the initial slope of the unloading curve and is strictly applicable only at the maximum load. Often it is reasonable to assume that the elastic modulus is constant with depth and under such conditions a correction can be derived for all depths. The accuracy of this assumption can be validated by unloading at several smaller loads. A comparison of these two basic methods is made below for standard test blocks. Five groups of ten indents were made on each of two standard microhardness test blocks. Hardness values were then calculated using both of the models identified above. It should be stressedthat these values are expressed in gigapascals and are based on the applied force divided by the projected contact area. In contrast, the Vickers hardness is defined in kilograms per millimetre squared and is based on the full contact area. Thus the values determined from the instrumented indenter require conversion to effective Vickers hardnesses for comparison with data measured by conventional means.
4. Results Table 1 summarizes the results obtained for two standard, steel, microhardness test blocks. Indent groups A
R Hutchings et al./Suflace and Coatings Technology 83 (1996) 243-249
Table 1 Comparison of the hardnessesmeasured for two standard test blocks, as determined using instrumented indentation and a conventional Vickers microhardness tester Indent group
Mean hardness, plastic model VW
Conversion to Vickers (Hv100)
8.43 8.34 8.12 7.89
776 768 141 125 752 321 310
D E F
G H I J
8.17 3.49 3.31
5 6 6
11 8 6 7 5 5 6
to E represent the results from a hardened steel with a hardness of around 800 Vickers. Column 2 lists the mean hardnessvalues determined using the plastic model described above, and column 3 gives the results of conversion to an effective Vickers value. The final column gives the data provided on the test block certification. It is clear that the plastic model underestimates the hardness for this test block, by a little under 10% when compared with the certified value. In contrast, use of the elastic model, as summarized by the results in columns 5 and 6, gives an overestimate of slightly above 10%. In the case of the softer test block, indent groups F to J show that instrumented indentation gives a similar overestimate of hardness irrespective of whether the plastic or elastic model is used. In both casesthe error is around 15%. Initial analysesclearly demonstrated the need to locate the surface accurately. Small errors in detecting the point of contact seriously affect the accuracy and reproducibility of the hardnesses calculated for shallow indents. A number of algorithms were used to define the point of contact from the change in slope of the load vs. displacement curve. However, it was found that the only way to avoid errors with absolute certainty was to view the curves and selectthe point of surface contact visually, as shown in Fig. 3. Having established the overall reliability of the technique, a number of PI3 treated sampleswere investigated. Fig. 4 shows a seriesof three indents in type 316 stainless steel, each indent at increasingly higher loads. It can be seen that at the lighter loads the indents are regularly shaped with the slightly concave edges typically seen where there is a significant degree of elastic recovery. However, for the 1 N indent the process has resulted in fracture of the outer layer. Re-examination of the loading and unloading curves in Fig. 2 clearly shows the point of fracture, as evidenced by the discontinuity in the loading curve at a load of around 800 mN. In order to examine possible grain to grain variation
Mean hardness, elastic model VW
Conversion to Vickers (HvlOO)
Certified hardness (HvlOO)
s13 & 53
10.24 9.71 9.59 9.85 3.54 3.45 3.44 3.60 3.56
942 899 582 907 326 317 316 332 327
513 so3 752 803 272& 11 274 276 276 274
Surface Contact \
-200 -300 i
-1.5 -1.0 -a,5 Displacement ( pm )
Fig. 3. A load-displacement curve covering the period of approach to, and contact with, the surface.
in the PI3 treated type 316 stainless steel, a large array of 50 mN indents were made in a number of samples. Fig. 5 shows such an array in a sample treated at 350 “C for 3 h. Except in those caseswhere the indenter struck an obvious flaw, or contaminant particle on the surface, the hardness profiles were found to be very consistent for all grains. This is clearly demonstrated by the hardnessvs. depth traces for this sample as shown in Fig. 6(b). The degree of hardening produced by this particular PI3 treatment is shown by comparison with the results for an untreated sample, Fig. 6(a). However, treatment at 520 “C produced a wide range of surface hardnesses, Fig, 6(c), even after filtering out those where the indentations were clearly defective.
5. Discussion Experience gained in the operation of the instrumented indentation unit has revealed a number of points that require care. Among the calibration constants for the
R Hutchings et al./Swface
and Coatings Technology 83 (1996) 243-249
(b) Fig. 5. (a) Array of 50 mN load indentations in type 316 stainless steel after PI3 treatment for 3 h at 350 “C, and (b) detail of one small region of the array.
Fig. 4. High load indentations in type 316 stainless steel after PI3 treatment for 3 h at 350 “C at loads of (a) 250 mN, (b) 500 mN, (c) 1000 mN.
various machine properties, the most difficult to obtain with great accuracy is the function relating the projected area A of the indenter to the depth d. The Berkovich indenter has an ideal shape function given by A = 24.5d2 and, for depths greater than about 1.5 pm, this gives sufficient accuracy. However, as the depth decreasesthe error increases, resulting in an apparently steeply rising hardness at small depths. Interactive procedures with which it is possible to determine an empirical shape function for a given diamond tip using a suitable material of known elastic properties have been proposed [lo], but their reliability is not yet well proven. One practical advantage in measuring the hardness by the instrumented indentation technique is that the procedure can be automated, and the statistical significance of the results improved by making a large number of indentations. We have found that two problems
R. Hutchings et al.JSurfaceand Coathgs Technology83 (1996) 243-249
-‘4g 12 ;;I0
g a5 . 26 42-
01 I ’ ” 0
300 400 Depth(nm)
- 1500 c, > z
-14 g 12 ;;;I0
- 1000 : 4
g 8 * I"6 4
i - 500
' ' ' 4 ' * ' ' ' ' ' ' 100 200 300 400 500 600 70: Depth(nm)
18 16 -
""'s 300 400 Depth(nm)
Fig. 6. Hardness as a function of depth for a number of indentations in type 316 stainless steel after PI3 treatment: (a) untreated, (b) 350 “C, (c) 520 “C. The “elastic correction model” was used, the elastic modulus was assumed to be constant with depth and the maximum load in all caseswas 50 mN.
conspire to make such blind testing an unsafe procedure-the indenter can strike imperfections in the material which are unrepresentativeof the surfaceas a whole, and errors may occur in detecting the point of contact with the surface accurately. For both these
reasons the analysis requires operator intervention for reliable results, although with suitable software tools and imaging hardware this need not be time consuming or a particular strain on the eyes. Each indentation should be inspectedto ensure that it satisfiesthe conditions for a satisfactory hardness measurement.Besides imperfectionsor grain boundariesfor small indentations, heavyloads may causevarious degreesof surfacecracking, especiallyin materials where a hard, brittle layer overlies a softer substrate. This can interfere with the interpretation of hardnessfrom the loading curves. At the larger loads, comparisons can be made with more conventional measurements of hardness, e.g. Vickers, with some assumptionsabout the comparison scale.However, it must be borne in mind that the tests are inherently different and such comparisonsshould be treated with caution, especiallywhen dealing with materials which show a high degree of elastic recovery. It was found that, for the type of steel which we normally treat, the elastic correction procedure gives measurements that are approximately 15% greater than the conventional Vickers hardnessvalue.It should be noted that at least one system is available which uses the instrumentedindentation approach to measurewhat has been termed a “Vickers under load” hardness. Brief mention of the elevatedtemperature PI3 treatment of austenitic stainlesssteelswas made previously in Sections 1 and 2. Early results on AISI type 304 stainless steel indicated the formation of a highly expanded austenite lattice with a large supersaturation of nitrogen present . More recently, a thorough study of the behaviour of PI3 treated type 316 stainless has beenreported [ 191,In addition, Samandiet rrl. [ 141 have compared the results of PI3 treatment with those obtained by plasma nitriding for the samematerial. For both types of treatment a form of nitrogen expanded austenite is produced. In the case of PI3 treatment at around 350 “C, the presenceof an amorphous layer has beenrevealedby transmissionelectronmicroscopy . The current study shows that these conditions produce a very consistent level of hardening. In contrast, at 520 “C chromium nitride begins to precipitate with the matrix transforming to what has been describedas a’martensite[ 141.This coincideswith the great variations in surfacehardnessas shown in Fig. 6(c).
These initial investigations with a high load instrumented indenter have shown the following. l High load instrumented indentation provides an efficient method for characterizing the duplex surfaces produced by plasma immersion ion implantation at elevatedtemperatures.
R Hzrtchingset al.jSusface and Coatings Technology83 (1996) 243-249
Visual inspection of the indentations and the data produced is essential for reliable data. Direct comparison with conventional Vickers hardnessmeasurementsis useful, but only at a semiquantitative level. PI3 treatment of austenitic stainless steel at 350 “C produces consistent hardening across the treated surface.
Acknowledgments The authors gratefully acknowledge the contributions of their colleagues. In particular, we appreciate the practical assistance of George Collins in producing this paper.
References [l] N.E.W. Hartley, Wkaj; 34 (1975) 427.  J.K. Hirvonen, J. Vat. Sci. Technol., 15 (1975) 1662.
131 G. Dearnaley, in CM. Preece and J.K. Hirvonen (eds.), ion Implantation bletallzrrgy, TMS-AIME, Warrendale, PA, 1980,p.1. r.41 J.R. Conrad, J.L. Radke, R.A. Dodd, F.J. Worzala and NC. Tran, J. .4ppl. Phys., 62 (1987) 4591. c51 J. Tendys, LJ. Donnelly, M.J. Kenny and J.T.A. Pollock, Appl. Phys. Lett., 53 (1988) 2143. 161J. Conrad, R.A. Dodd, F.J. Worzala and X. Qiu, Surf. Coat. Technol., 36 (1988) 927. 171 S.R.J. Saunders, Su$ Eng., 9 (1993) 293. 181J.B. Pethica, WC. Oliver and R. Hutchings, Philos. Mug. A, 48 (1983) 593. II91 R. Hutchings, WC. Oliver and J.B. Pethica, in R. Kossowsky and SC. Singhal (eds.), Surface Engineering, Martinus Nijhoff, Dordrecht, 1954,p. 170. WC. Oliver and G.M. Pharr, J. Mater. Res., 7 (1992) 1564. R.D. Dukino and M.V. Swain, J. Am. Ceram.Sot., 75 (1992) 3299. G.A. Collins and J. Tendys, J. Vat. Sci. Technol. B, 12 (1994) 875. G.A. Collins, R. Hutchings and J. Tendys, Muter. Sci. Eng., Al39 (1991) 171. [I41 M. Samandi, B.A. Shedden, T. Bell, G.A. Collins, R. Hutchings and J. Tendys, J. Vae Sci. Technol. B, 12 (1994) 935. [I51 G.A. Collins, R. Hutchings, K.T. Short, J. Tendys, X. Li and M.Samandi, Surf Coat. Technol., in press. Cl61 G.A. Collins, R. Hutchings and J. Tendys, Surf. Coat. Technol., 59 (1993) 267. Cl71 G.A. Collins, R. Hutchings, K.T. Short, J. Tendys and C.H. Van Der Valk, Surfi Coat. Technol., in press. Cl81 W. Weiler, Br. J. Non-Destr. Test., 31 (1989) 253. Cl91 M. Samandi, B.A. Shedden, D.I. Smith, G.A. Collins, R. Hutchings and 5. Tendys, Stlrf. Coat. Technol., 59 (1993) 261.