Jiali Zhang, and Stefan Zaefferer, Influence of sample preparation on nanoindentation results of twinning-induced plasticity steel, Int. J. Miner. Metall. Mater., 28(2021), No. 5, pp.877-887. https://dx.doi.org/10.1007/s12613-021-2260-z
Cite this article as:
Jiali Zhang, and Stefan Zaefferer, Influence of sample preparation on nanoindentation results of twinning-induced plasticity steel, Int. J. Miner. Metall. Mater., 28(2021), No. 5, pp.877-887. https://dx.doi.org/10.1007/s12613-021-2260-z
Jiali Zhang, and Stefan Zaefferer, Influence of sample preparation on nanoindentation results of twinning-induced plasticity steel, Int. J. Miner. Metall. Mater., 28(2021), No. 5, pp.877-887. https://dx.doi.org/10.1007/s12613-021-2260-z
Cite this article as:
Jiali Zhang, and Stefan Zaefferer, Influence of sample preparation on nanoindentation results of twinning-induced plasticity steel, Int. J. Miner. Metall. Mater., 28(2021), No. 5, pp.877-887. https://dx.doi.org/10.1007/s12613-021-2260-z
Institute for Materials Applications in Mechanical Engineering, RWTH Aachen University, Aachen 52062, Germany
2)
Max-Planck-Institut für Eisenforschung, Düsseldorf 40237, Germany
Funds: The authors would like to express their appreciation to Dr.-Ing. Nahid-Nora Elhami for active discussion of the results. Acknowledgment is also given to Ms. Heidi Bögershausen for her help in the nanoindentation experiments and to Ms. Monika Nellessen and Ms. Katja Angenendt for their help in the sample preparation and SEM operation. Financial support was provided by the Max-Planck-Institute for Iron Research
Nanoindentation is an attractive characterization technique, as it not only measures the local properties of a material but also facilitates understanding of deformation mechanisms at submicron scales. However, because of the complex stress–strain field and the small scale of the deformation under the nanoindenter, the results can be easily influenced by artifacts induced during sample preparation. In this work, a systematic study was conducted to better understand the influence of sample preparation methods on the nanoindentation results of ductile metals. All experiments were conducted on a steel (Fe–22Mn–0.65C, wt%) with twinning-induced plasticity (TWIP), which was selected for its large grain size and sensitivity to different surface preparation methods. By grouping the results obtained from each nanoindent, chemical polishing was found to be the best sample preparation method with respect to the resulting mechanical properties of the material. In contrast, the presence of a deformation layer left by mechanical polishing and surface damage induced by focused ion beam (FIB) scanning were confirmed by the dislocation-nucleation-induced pop-in events of nanoindentation.
Nano-mechanical experiments are popular for mechanically characterizing materials at the submicron scale. Among the existing nanotechniques, nanoindentation (load- and displacement-sensing indentation tests), from which the hardness, elastic modulus, fracture toughness, and surface force can be estimated, is one of the most powerful for characterizing localized mechanical properties of materials [1–2]. As conventional nanoindentation involves a statistical strategy for measuring the properties of materials, it is generally performed by mapping many indents over an area of interest. The small scale of investigation makes it possible to obtain the mechanical properties of different, potentially closely spaced phases [3–5] or of the microstructural hierarchy [6–7] within a material. In addition, the local properties measured by nanoindentation are often used to calibrate the parameters implemented in crystal plasticity simulations [8–10]. Furthermore, in-situ nanoindentation tests are attracting interest for their use in the on-site observation of the deformation behavior of material under the nanoindenter [11–12].
Although nanoindentation is a powerful technique in numerous aspects, its results depend on the surface quality of the tested sample. Sample preparation can determine to different extent the accuracy of the results of any kind of surface measurement methods. For nanoindentation, the influence of sample preparation is even more critical, since the experimental scale is small, and the essential calculation equation of the contact area is derived based on the assumption of a flat surface [1–2,13–14]. Therefore, the magnitude of the measured displacements and the tolerance of uncertainty in the contact area usually determine the allowable surface roughness in the sample preparation. The greatest problems are encountered when the characteristic wavelength of roughness is comparable to the contact diameter [15]. Typically, the last step in the sample preparation of metals is either mechanical polishing (MP) or chemical polishing (CP). On the one hand, MP generally leads to a surface roughness similar to the grit size of the particles contained in the suspension used for polishing and a surface modification because of strain hardening or cold working, especially in metals. This surface modification generally appears as an uneven distribution of defects (dislocations, stacking faults, nanotwins, etc.) in the layer immediately beneath the surface, which is the volume probed by nanoindentation.
On the other hand, CP can undoubtedly remove surface deformation, but it also induces surface roughness of larger wavelengths compared to that associated with MP. For a nanoindent with a contact diameter significantly different from the wavelength of the surface roughness, the pop-in events evident in its load–displacement (P–h) curve can be useful for characterizing the mechanical state of its surface. These pop-in events are useful, because the presence of defects may lead to various effects on nanoindentation results, such as missing pop-ins and higher hardness in areas with high defect densities and the presence of pop-ins due to delayed dislocation nucleation in areas of low defect densities [16–17]. Wang et al. [16] confirmed the sensitivity of nanoindentation pop-in events to sample preparation by examining the influences of sample preparation on the pop-in events of single-crystal Mo. Pop-in events were also used to facilitate the identification of surface defects induced by low dose irradiation [17], thus proving its usefulness as a tool for verifying surface conditions.
In this study, we systematically investigated the influence of sample preparation methods on the nanoindentation results of a coarse-grained twinning-induced plasticity (TWIP) steel (Fe–22Mn–0.65C, wt%). This TWIP steel grade was selected for its large grain size and sensitivity to different surface preparation methods. In addition, interest regarding the study of TWIP steels has grown, because these steels exhibit relatively high strength (up to 1.1 GPa) and plasticity (elongation up to 95%) [18–19] due to the formation of mechanical twins under internal or external stress. However, TWIP steels are reported to suffer strong hydrogen embrittlement and delayed fracture during the cup-drawing process [20–21], which greatly hinders their wider application. Therefore, this work is also expected to contribute to the investigation of the embrittlement and damage mechanisms of TWIP steels by local experimental methodologies like nanoindentation. Unlike most other investigations on the influence of surface quality on nanoindentation results [15–16,22–24], in this work, we used a spherical indenter. This has the advantage over the more commonly used Berkovich or Vickers indenter [25] that the results are independent of the rotation of the investigated crystal orientations about the indentation axis. Pathak et al. [14] are the only other researchers who have investigated the effect of spherical indenters. However, in their work, no details were provided regarding the relevant error measures, which is an important focus of the present work.
2.
Experiments
In this study, we used a TWIP steel composed of Fe–22Mn–0.65C (wt%). Solidification after casting was performed in an induction furnace under Ar. The ingot was hot rolled at 1100°C for a total thickness reduction of up to 50%. After hot rolling, the material was annealed for 6 h at 1150°C under Ar and subsequently water quenched to room temperature. Thereafter, the sample was hot rolled again at 1100°C to achieve a total thickness reduction of up to 55%, followed by water quenching after an annealing treatment at 1150°C for 24 h. This procedure resulted in a homogeneous single fcc-phase microstructure with an average grain size of 150 µm, which enabled the placement of many nanoindents inside one grain.
The bulk material was first cut into samples 1 mm × 6 mm × 7 mm in size, which were then metallographically prepared. As noted above, the aim of this study was to understand the influence of sample preparation on the indentation data; therefore, first, a careful investigation of the different types of sample preparation was conducted. Notably, in many studies reported in the literature, sample preparation by electrochemical polishing was found to yield samples with benchmark-grade surface quality [14,16,25]. However, our attempts to perform effective electrochemical polishing resulted in samples with relatively thick oxide layers. Because CP removes the defects close to the surface similar to electropolishing, it was used as the benchmark in this work. In general, there are two types of sample preparation methods: (a) MP with a final polishing step using a suspension of nanosized silica particles; (b) additional CP in a solution of 90vol% H2O2 and 10vol% HF as the last step, which we refer to as “MP + CP”. For both sample preparation types, all the samples were prepared by a sequence of grinding, diamond polishing, and final polishing with a silica-particle suspension. Grinding was performed using silicon carbide papers with grits ranging from 600 to 2500. To prevent the formation of twinning during grinding, very coarse grinding papers (180–400) were not used. After grinding, the samples were cleaned in an ultrasonic bath, washed with water and ethanol, and dried before polishing to prevent the contamination of the polishing plate by particles from the previous grinding steps. Diamond polishing was performed with a mixture of the suspension containing 3 µm diamond particles and a lubricant solution. The final MP was performed with a suspension containing 50 nm silica particles (Struers OP-S) for 15 min to remove the deformation layer induced in the previous polishing steps. In particular, OP-S polishing is also a form of CP in which the silica particles mainly act to remove the products associated with the chemical attack of the polishing solution. After final polishing, some samples were directly taken for observation by scanning electron microscopy (SEM). Other samples were further etched in the H2O2–HF solution for 10 s and then quickly cleaned in a H2O2 solution and dried in warm air to be ready for further investigation.
To confirm the influence of the preparation process on the sample surface quality, we used the electron channeling contrast imaging (ECCI) technique [26] in the SEM to directly observe the defects (in particular dislocations and stacking faults) close to the sample surface. As CP would degrade the quality of the ECC-images, some of the etched samples were again mechanically polished with a suspension of silica particles for 2 min to ensure high image qualities. Hereafter, we refer this preparation combination as “MP + CP + MP”.
All the SEM observations were performed in a Zeiss cross-beam instrument (XB 1540; Carl Zeiss SMT AG, Germany) with a Gemini-type field emission gun electron column and a focused ion beam (FIB) (Orsay Physics). To eliminate the influence of grain orientation on the nanoindentation data, electron backscatter diffraction (EBSD) was used to determine the orientation of the grains and enable the selection of {111}-, {101}-, {001}-oriented grains. The FIB technique was applied with an accelerating voltage of 30 kV and a current of ~500 pA on some samples to mill markers, which are necessary for finding the precise position of pre-selected grains under an optical microscope equipped on the nanoindentation machine. During this process, the areas of interest were quickly scanned by FIB (<5 s) with an inclination angle of 36° prior to milling the markers to investigate the influence of FIB scanning on the nanoindentation results. This scanning step is generally unavoidable in FIB operations, although the beam parameters and scanning duration can be controlled.
A Hysitron TriboScope 950 was used to perform the nanoindentation tests. A conical indenter with a spherical tip which has a tip angle of 90° and tip radius of 1 µm was used, which enables a smooth transition from elastic to elastic–plastic contact. All the nanoindentation experiments were carried out using a load-controlled mode with a maximum load of 1000 µN. For each indent, after the surface was detected by the indenter, the nanoindenter was pressed into the material for 10 s to exert maximum load and was then held at maximum load for another 10 s before being retracted. A grid pattern of 10 × 10 indents was made with 10 µm between neighboring indents to enable statistical analysis of each measurement. Each time the machine setting was changed, the device was calibrated on a quartz single crystal to obtain the corresponding area function. Origin software equipped with a Hysitron plug-in unit was used in all the hardness and elastic modulus calculations and P–h curve plotting.
3.
Results
After the indentation test, we obtained SEM images of the indent patterns to identify invalid indents so that they could be excluded from the statistical analysis. The identified invalid indents included those for which no indent had been made at the given coordinates, indents located on or near a grain boundary or a scratch, and double or triple indents.
Regarding the influence of the sample surface condition, three preparation methods were considered: MP, MP + CP in the H2O2–HF solution, and MP + CP + MP. As some of the samples were also scanned by electron beam and FIB to select specifically oriented grains for the nanoindentation tests, electron beam (E) and FIB (F) scanning were also considered due to the induced respective hydrocarbon contamination layer and surface structural damage [27]. Table 1 lists the information regarding the nanoindentation experiments, samples, and preparation methods.
Table
1.
Nanoindentation series and corresponding sample preparation methods used on individual sample.
The MP + CP + MP sample was subjected to an additional test, as it had been prepared using the particular preparation method. Although the nanoindentation test on the MP + CP + MP sample had the same experimental setup as the other samples, a different area function was obtained following a new calibration, as the indenter has been taken out, cleaned, and reinstalled. Therefore, we obtained two series of nanoindentation test results.
Fig. 1(a) shows a column plot of the average reduced elastic modulus and hardness values of all the samples. All the results of the samples in nanoindentation series-1 lie within the same range with only subtle differences between them, whereas both the reduced elastic modulus and hardness values of the MP + CP + MP sample are clearly larger than those of the other samples. This data shift is even more obvious in the scatter plot of the results of each indent (Fig. 1(b)), which might be due to the utilization of different area functions. This originates from the method used to calculate the nanoindentation data, the most widely used of which was established by Oliver and Pharr as follows [13].
Fig.
1.
(a) Column plot of the average reduced elastic modulus and hardness of all the studied samples; (b) scatter plot of the reduced elastic modulus vs. hardness of all the valid indents. The data obtained for the MP + CP + MP sample are probably outliers because of different calibration or contaminated indenter tip.
The hardness (H) is calculated by dividing the maximum load (Pm) by the projected area of the contact of the residual hardness imprint (Ac).
H=Pm/Ac
(1)
The modulus directly calculated from the P–h curve is the reduced elastic modulus (Er). The elastic modulus of the tested material, E, can be calculated from Er as follows:
1Er=1−v2E+1−v2iEi
(2)
where v is the Poisson’s ratio of the sample, and Ei and vi are the elastic modulus and Poisson’s ratio of the indenter, respectively. For diamond, Ei = 1141 GPa and vi = 0.07 are often used. Er is calculated using Eq. (3).
Er=√π2βS√Ac
(3)
where β is a geometric correction factor that depends on the geometry of the indenter. S is the contact stiffness, which is determined as the slope of the straight line fitted to the upper 25%–50% of the unloading data.
From the above given equations, it can be seen that the accuracy of the calculated values depends on the accuracy of the measured contact stiffness (S) and the projected contact area (Ac), which is a function of the contact depth (hc). The contact depth of each indent can be easily calculated using the contact stiffness and the maximum depth [13]. As the contact stiffness and the maximum depth are directly obtained from the nanoindentation tests and thus not influenced by the calibration, the accuracy of the nanoindentation tests generally depends on the precision of the calibrated area function Ac(hc). Therefore, we considered the difference in the calibrated area functions between the different series of nanoindentation tests for the comparison of different datasets, although this difference is generally small enough to be neglected. Possible underlying reasons for obtaining different area functions through calibration are discussed in section 4.1.
3.1
Influence of sample surface quality
3.1.1
Mechanical and chemical polishings
To illustrate the influence of the aforementioned calibration problem and the relationship between the raw and the calculated data, the reduced elastic modulus (Er) vs. hardness (H) values obtained from nanoindentation series-1 and the contact stiffness (S) vs. maximum depth (hmax) values of all samples are plotted in Fig. 2 as scatter points.
Fig.
2.
Scatter plots of (a) the calculated data of reduced elastic modulus vs. hardness from nanoindentation series-1 and (b) the raw data of contact stiffness vs. maximum depth from all samples.
Fig. 2(a) shows that the scatter points group in two fractions, triangles and circles, which belong to the MP surfaces and MP + CP surfaces, respectively. Besides the calculated data from the nanoindentation series-1, the raw data of all samples in Fig. 2(b) show the same classification with the additional data from the MP + CP + MP sample (green squares) in the same group of the MP surfaces. The data obtained from the MP (including MP + CP + MP) surfaces are distinguished by higher contact stiffness and smaller maximum depth compared to the MP + CP surfaces, which results in a higher reduced elastic modulus and higher hardness for the MP samples. To directly observe the surface quality of the MP and MP + CP samples and to obtain information such as the defect densities, these samples were observed using ECCI (Fig. 3).
The ECC-image of the MP surface (Fig. 3(a)) shows a good contrast and reveals fine details of the microstructure, i.e., individual defects, whereas the image quality of the CP surface (Fig. 3(b)) is strongly degraded by the contrast from some blurred white and black dots induced by the strong surface roughness and residual oxides on the surface. Dislocation density determinations from five ECC-images of each sample preparation condition reveal dislocation densities of 10.24 × 1012 m−2 and 3.36 × 1012 m−2 for the MP and MP + CP surfaces, respectively. Note that due to the low image quality of the MP + CP surfaces, its measured dislocation density can be to some extent lower than the real value. However, this error should not be large enough to reach the density measured on the MP surfaces, as reported in another work of the authors [28]. The MP+ CP + MP prepared sample surface, which aimed for smaller surface roughness than on the MP + CP surface as well as for a smaller dislocation density than the MP surface, was also imaged as shown in Fig. 3(c). The calculated dislocation density is 9.05 × 1012 m−2, which is not significantly different from that of the MP surface. However, the surface flatness is much better than that of the MP + CP surface.
Fig.
3.
Controlled electron channeling contrast images of (a) MP, (b) MP + CP, and (c) MP + CP + MP samples.
3.1.2
Electron beam scanning and focused ion beam scanning
Electron beam scanning can leave a hydrocarbon contamination layer on sample surfaces. The heavy Ga+ ions in FIB can also cause lattice damage and induce dislocations in the surface [27]. In this study, the influences of these two factors on the nanoindentation results were also considered.
As EBSD scans generally leave more severe contamination than normal secondary or backscatter electron imaging raster scans of the electron beam, so the hardness data of an EBSD scanned sample and a non-EBSD scanned sample from the MP and MP + CP samples were plotted in Fig. 4 for comparison. As expected, the reduced elastic modulus values are not affected by the electron beam. In contrast, the hardness values show some influences. While the values from the generally harder MP surface show only marginal changes, the values from the more sensitive CP surface show a clear increase of ~5% from 2.96 GPa (±0.17 GPa) without contamination to 3.09 GPa (±0.21 GPa) with contamination. The shifts in these data are rather small compared to the statistical scatter of each dataset, therefore only a marginal influence of the hydrocarbon layer on the nanoindentation results was observed.
Fig.
4.
Scatter plots of the calculated data: reduced elastic modulus vs. hardness of (a) MP + CP samples and (b) MP samples for the comparison between non-EBSD scanning and EBSD scanning.
It is important to emphasize that FIB milling is a standard sample preparation method for many micro-mechanical tests, therefore, the investigation of its effect on the indentation results is of high importance to the field. Fig. 5 shows the scatter plot of the nanoindentation results of non-FIB scanned and FIB scanned surfaces from both the MP and MP + CP samples. In Fig. 5(a), the hardness data of the FIB scanned surface (3.17 GPa ± 0.17 GPa) clearly shift by ~5% towards higher values compared to that of an only chemically polished sample (2.96 GPa ± 0.17 GPa), while the distribution of the data becomes significantly smaller. In contrast, the data obtained for the two MP samples in Fig. 5(b) show almost no change in their hardness values with or without FIB irradiation. Therefore, there appears to be no strong influence of FIB on the hardness data. However, to confirm this, the pop-in events were further analyzed in this perspective to obtain more detailed information about each nanoindent.
Fig.
5.
Scatter plots of the calculated data: reduced elastic modulus vs. hardness of (a) MP + CP samples and (b) MP samples for the comparison between non-FIB scanning and FIB scanning.
The pop-in phenomenon corresponds to a sudden indentation displacement of the indenter under a certain constant load, i.e., the pop-in load, in load-controlled nanoindentation tests. The study of the pop-in events is very important in analyzing nanoindentation data, as it provides information on the local conditions of the indented surface [16,29–31]. Normally there is not only one pop-in along a P–h curve, as shown by the black curve in Fig. 6. The mechanisms responsible for the appearance of each pop-in event are relatively difficult to define, however, the well agreed mechanism responsible for the first pop-in is the nucleation of dislocations [29,31–32]. Thus, here we consider only the first pop-in event. Two main variables, pop-in load and pop-in displacement (Fig. 6) were obtained by checking the displacement using a macro in an Excel document of each P–h curve. In the macro, a pop-in was defined as a point-to-point displacement change greater than 3 nm. In this case, the instant load and displacement change were exported as a pop-in data point.
Fig.
6.
Typical P–h curves with multiple pop-ins, no pop-in, and one pop-in obtained from the MP + CP sample.
In addition, to compare the results of different samples, we also used the pop-in ratio (Fig. 7), which is the ratio of the number of indents that show a pop-in to the total number of the valid indents (10 × 10 pattern) in one nanoindentation test, which is not calibration dependent.
Fig.
7.
Pop-in ratio and sample preparation method for each sample.
Fig. 7 shows that the MP + CP sample surfaces (MP + CP, MP + CP + F, MP + CP + E, and MP + CP + E + F) have relatively high pop-in ratios, whereas the ratios of the MP samples (MP, MP + E, MP + E + F, and MP + CP + MP) are relatively low. This agrees with the aforementioned mechanism of the first pop-in event, as the MP surfaces contain a high number of pre-existing dislocations (Fig. 3). On this basis, Fig. 8 shows plots of the average pop-in loads and their standard deviations for all the samples.
Fig.
8.
Average pop-in loads and their standard deviations for the samples.
In Fig. 8, the average pop-in loads of the MP + CP + F, MP + CP + E + F, and MP + E + F samples, which were briefly scanned by FIB before the nanoindentation tests, are lower than those of the other samples. This indicates a strong influence of FIB on the surface quality. The underlying mechanisms for this will be discussed in section 4.2.2.
4.
Discussion
4.1
Statistics and data shift for two nanoindentation series
Nanoindentation is a statistical method for determining the mechanical properties of materials, with each data point indicating only the very local properties of the material in the vicinity of each indent. Here the quality of the measured data was evaluated by calculating the median and the full width at half maximum (FWHM) of the reduced elastic modulus and hardness value distributions for each sample. For this, we performed a Gaussian fitting to the frequency plot of each dataset. The obtained median and FWHM-to-median values are plotted as columns in Fig. 9.
Fig.
9.
Calculated (a) median and (b) FWHM/median in % for the reduced elastic modulus and hardness data of each sample.
The calculated median values in Fig. 9(a) provide the same information as the average values shown in Fig. 1(a), which reveal slightly higher (5%–10%) elastic modulus and hardness values for the MP surfaces than those for the CP surfaces. However, the influence of the FIB scanning cannot be determined without analyzing the pop-in events. The FWHM/median values in Fig. 9(b), in contrast, show quite a large influence of surface quality: the hardness values of the MP surfaces have a smaller scatter than those of the CP surfaces. This could be because of the greater roughness of the CP surfaces in comparison with the MP surfaces. Meanwhile, it is also possible that pre-existing defects in the surface lead to more homogenous plastic deformation behavior and decrease the scatter of the hardness values. However, surprisingly, a larger scatter of the FWHM/median values for the reduced elastic modulus than for those of the hardness is observed, which indicates that the unloading behaviors (contact stiffness) are also influenced by the surface quality.
A phenomenon that can be observed in all data plots is the inconsistency of the results of the MP + CP + MP sample with those of the other samples. It has been indicated in section 3 that the reason lies in the calibration, which was newly performed prior to this particular measurement on a fused quartz sample. The inconsistency may also be because of an indenter contamination. The preservation of the geometry of the indenter is well recognized to be essential for nanoindentation testing, as it determines the obtained contact area function from the calibration. Any indenter contamination would lead to an incorrectly calculated contact area and thus to inaccurate results. Therefore, the indenter requires routine cleaning.
Fig. 10(a) shows the top view of the indenter tip prior to nanoindentation test series-1 (Table 1). As the upper tip itself was not contaminated, except for the presence of a small fiber, the area function obtained from the calibration is most likely correct, hence, the measured results should be accurate. However, the image of the indenter tip captured directly after the measurement of the second nanoindentation series-2 (Fig. 10(b)) shows that a large amount of dirt had been collected at the very tip, which may have already been present before the calibration. This dirt might have significantly changed the area function obtained from the calibration and further induced large errors in the data. As the data shift between the two nanoindentation series was only observed on the calculated values of the reduced elastic modulus and hardness, but not on the directly measured values of maximum depth and contact stiffness, an incorrect area function obtained from the calibration with a contaminated indenter tip is the most probable cause for the data shift. We recommend, for future work, to repeat already performed measurements after each calibration to ensure the comparability of the datasets.
Fig.
10.
Secondary electron images of the indenter tip from the top viewport: (a) prior to nanoindentation test series-1; (b) directly after nanoindentation test series-2 (on the MP + CP + MP sample).
The influences of CP and MP can be observed in the scatter plots of the hardness and reduced elastic modulus (Fig. 2). The electron beam scanning shows only a marginal influence on the obtained data (Fig. 4) and is not going to be discussed in more detail. The influence of FIB scanning is discussed based on the data obtained for the first pop-in events (Figs. 7 and 8).
4.2.1
Mechanical and chemical polishing
Although hardness and reduced elastic modulus are the two main output values from the nanoindentation test, the raw data directly shown in the P–h curves can generally provide more information regarding the deformation behavior of the materials under the indenter. Fig. 2(b) shows that the scatter data of the MP surfaces are grouping up at a smaller maximum depth and larger contact stiffness range than the data of the MP + CP surfaces. By calculating the average maximum depths of the MP surfaces (52.8 nm ± 3 nm) and the MP + CP surfaces (59.3 nm ± 3 nm), it can be seen that with the same maximum load and loading function the indenter could go ~6 nm deeper into the MP + CP surface than into the MP surface.
When the indenter is applied to the surface, the material generally undergoes an initial elastic deformation, then passes the transition point from elastic to plastic deformation and continues with plastic deformation. A small segment at the beginning of the loading part of the P–h curve that can be described by the Hertzian law reveals the elastic behavior of the material, regardless of whether a deformation layer exists or not. This is the small segment in which all three P–h curves overlap in Fig. 6. Theoretically, with respect to the elastic–plastic transition, if there are no dislocations present in the tested region, a high load is necessary to nucleate dislocation sources which can then emit dislocations easily. As a result, a pop-in appears. If dislocation sources are already present close to the surface (including ledges in the surface itself), then the dislocation nucleation induced pop-in will not appear or exhibit a much smaller displacement. This behavior is generally used to make a statement about the amount of the dislocation sources present in the material prior to nanoindentation [16].
In this study, we used the pop-in ratio (Fig. 7) as an indication of the presence of dislocation sources prior to indentation, whereby the higher is the pop-in ratio, the more indents exhibit pop-ins and the lower is the amount of dislocation sources in the affected volume. The differences in the pop-in ratios of the differently prepared samples in Fig. 7 are in accordance with this hypothesis. The MP + CP samples reveal relatively high pop-in ratios of ~80%, whereas the MP samples show lower pop-in ratios of ~50%. Therefore, we confirm that MP leaves dislocation sources in the indenter affected volume. These dislocations also ease the yielding of the material. When a certain load is applied to the material, the pre-existing dislocations directly start to glide, i.e., no source nucleation needs to occur; only the bow-out stress on the sources needs to be exceeded. Since several sources may act simultaneously, the systems more quickly interact and form Lomer-locks and further Frank–Read sources [33]. As a result, strain hardening quickly sets in and larger loads are reached for a given indentation depth (red curve in comparison with blue and black curves with displacements less than 15 nm in Fig. 6). In contrast, no deformation layer exists close to the surface in most of the areas of the MP + CP surface. Therefore, the material undergoes elastic deformation until sufficient stress is reached to form dislocation sources. This appears, most probably first at the surface of the material beneath the indent where the shear forces are highest. Once an effective source has been nucleated, it emits dislocations, as visible by the onset of a pop-in event. The pop-in event ends when the back stresses on this active slip system become higher than the stress being applied.
Besides the analysis of deformation mechanisms, pop-in events are also often used to identify the onset of nano- or micro-cracks in a material [1]. When the oxide layer on a CP surface is thick or hard enough to influence the nanoindentation results, a pop-in event induced by fracture of this surface layer can be observed at the very beginning of the loading curve of each indent. However, this was not observed, thus no clear influence of the oxide layer on the nanoindentation results can be identified. On the other hand, it should be noted that the image quality of the CP samples in the SEM is strongly degraded because of the presence of the oxide layer, as shown in Fig. 3(b). The reason for the discrepancy between the visible presence of an oxide film and the lack of an observed pop-in event is that the oxide/hydroxide film formed on the sample surface is not continuous and dense but soft and sponge-like.
In summary, MP leaves dislocation sources close to the sample surface, which leads to overestimated hardness and reduced elastic modulus of the material. When the mechanical properties close to those of a perfect crystal are aimed to be measured by nanoindentation without a high demand on imaging the indents (i.e., if the slight thin oxide film is not a concern), CP should be selected as the final sample preparation step. If good images of the surface are also required, then a very light MP step should be applied after CP to remove the oxide film. Note that these suggestions are valid only for the class of steels used in this study and should not be regarded as a general recipe for surface preparation.
4.2.2
Focused ion beam scanning
In section 3.2, it has been shown by the pop-in results that slight irradiation of the surface with FIB has a strong influence on the surface quality, i.e., the pop-in loads of all the FIB-irradiated samples are reduced by ~20% (Fig. 8), which indicates that a relatively low force is required on these samples to activate dislocation sources. In contrast, the pop-in ratio is not much affected by FIB scanning: the FIB-irradiated samples show both high and low pop-in ratios (Fig. 7). In fact, crystal structural damage by FIB is a well-known phenomenon. The highly energetic Ga+ ions do not only sputter materials, for which they are designed, but some of them also create significant atomic displacement and vacancies in the material. This is most evident in transmission electron microscope (TEM) samples prepared by FIB, wherein a rather undefined “cloudy” contrast overlays the usual TEM image. In the case of metals, many of these atomic displacements are healed. It may be assumed that the produced vacancies collapse into vacancy platelets and form prismatic dislocation loops that can act as dislocation sources and obstacles for the gliding dislocations.
As commonly agreed, the first pop-in is assigned to the nucleation of the dislocations, and the pop-in load depends on the dislocation nucleation energy [29,31–32]. However, in addition to the nucleation, this load is also combined with the dislocation propagation, as pop-in displacements as large as 30–40 nm can be obtained within one pop-in event. To further understand the influence of the FIB irradiation on the pop-in behavior of the material, we plotted the discrete pop-in data of all the samples from nanoindentation series-1separated for non-FIB and FIB scanned cases, as shown in Fig. 11.
Fig.
11.
Scatter plot of the pop-in data of (a) non-FIB scanned samples and (b) FIB scanned samples obtained in nanoindentation series-1.
Both scatter plots in Fig. 11 show clear load-displacement delimitations of the indents with pop-ins, which are marked by the black dotted lines. It should be point out that these black dotted lines are drawn only to visually separate the two regions, no physical meaning with respect to the nanoindentation measurements should be applied to them. The pop-in data in the vicinity of the dotted lines can be understood by the dislocation nucleation mechanism described in the previous section. When no dislocation source is present in the indentation-affected volume, the combination of high pop-in loads and large pop-in displacements is observed, i.e., data points lying in the right-upper corner in Fig. 11(a). For these points high forces are required to nucleate dislocations. Once nucleated, these dislocations move for long distances and in large amounts, thus creating a large pop-in displacement. For pop-in events with low pop-in loads and displacements, which is the most frequently observed case (data points lying in the left-lower corner in Fig. 11), pre-existing dislocation sources are easily activated under low pop-in loads. Meanwhile, there is not much free dislocation path length for the activated dislocations to move, which results in generally small pop-in displacements. In those cases where high pop-in loads are observed in combination with small displacements (data points lying in the left-upper corner in Fig. 11), we assume that twinning occurred even though twinning was not observed in the SEM images. It is known that twinning requires a higher activation energy than dislocation nucleation, and the total shear experienced by deformation twins is small.
Based on the understandings given above, it can be confirmed that the ion dose used for the FIB irradiation in this study was sufficiently large to influence the pop-in behavior of the material. However, it was not large enough to induce a sufficient number of dislocation sources to eliminate the pop-in events. Therefore, the pop-in ratios of the irradiated samples remain comparable to those of the non-irradiated samples.
5.
Conclusions
In this study, the influence of sample preparation on the nanoindentation results of a traditional TWIP steel was systematically studied through statistical analyses of the data and pop-in events. ECCI observations were conducted to confirm the surface quality of the samples. The conclusions are as follows.
(1) Mechanical polishing leaves a deformation layer on the sample surface, which shifts the measured hardness and the reduced elastic modulus to larger values. Chemical polishing can effectively remove this deformation layer and help to obtain the mechanical properties that are close to those of a perfect crystal. However, it also induces a thin oxide layer and a large surface roughness that strongly degrade the quality of ECC-images. A brief mechanical polishing of the surface after chemical polishing is a good compromise to balance the obtained nanoindentation experimental results against the desired SEM image quality when both are demanded.
(2) No strong influence on the results due to the hydrocarbon contamination induced by electron beam scanning was observed.
(3) FIB scanning, even at low doses, introduces dislocations into the surface, which increases the measured hardness and reduces the pop-in loads of the first pop-in events.
(4) Repeated calibration during the experiments or the contamination of the indenter could lead to errors in the calibrated area function and inaccuracy of the measured data. Therefore, it is important to check and clean the indenter regularly and perform all measurements either with the same calibration or repeat some of the measurements after recalibration.
Acknowledgements
The authors would like to express their appreciation to Dr.-Ing. Nahid-Nora Elhami for active discussion of the results. Acknowledgment is also given to Ms. Heidi Bögershausen for her help in the nanoindentation experiments and to Ms. Monika Nellessen and Ms. Katja Angenendt for their help in the sample preparation and SEM operation. Financial support was provided by the Max-Planck-Institute for Iron Research.
A.C. Fischer-Cripps, Nanoindentation, Springer, Lindfield, Australia, 2004.
[2]
C.A. Schuh, Nanoindentation studies of materials, Mater. Today, 9(2006), No. 5, p. 32. DOI: 10.1016/S1369-7021(06)71495-X
[3]
J.L. Zhang, C.C. Tasan, M.J. Lai, D. Yan, and D. Raabe, Partial recrystallization of gum metal to achieve enhanced strength and ductility, Acta Mater., 135(2017), p. 400. DOI: 10.1016/j.actamat.2017.06.051
[4]
Z.C. Wu, S. Sandlöbes, J. Rao, J.S.K.L. Gibson, B. Berkels, and S. Korte-Kerzel, Local mechanical properties and plasticity mechanisms in a Zn–Al eutectic alloy, Mater. Des., 157(2018), p. 337. DOI: 10.1016/j.matdes.2018.07.051
[5]
L.S. de Vasconcelos, R. Xu, J.L. Li, and K.J. Zhao, Grid indentation analysis of mechanical properties of composite electrodes in Li-ion batteries, Extreme Mech. Lett., 9(2016), Part3, p. 495. DOI: 10.1016/j.eml.2016.03.002
[6]
L. Morsdorf, C.C. Tasan, D. Ponge, and D. Raabe, 3D structural and atomic-scale analysis of lath martensite: Effect of the transformation sequence, Acta Mater., 95(2015), p. 366. DOI: 10.1016/j.actamat.2015.05.023
[7]
A. Khosravani, L. Morsdorf, C.C. Tasan, and S.R. Kalidindi, Multiresolution mechanical characterization of hierarchical materials: Spherical nanoindentation on martensitic Fe–Ni–C steels, Acta Mater., 153(2018), p. 257. DOI: 10.1016/j.actamat.2018.04.063
[8]
C. Zambaldi and D. Raabe, Plastic anisotropy of γ-TiAl revealed by axisymmetric indentation, Acta Mater., 58(2010), No. 9, p. 3516. DOI: 10.1016/j.actamat.2010.02.025
[9]
C.C. Tasan, M. Diehl, D. Yan, C. Zambaldi, P. Shanthraj, F. Roters, and D. Raabe, Integrated experimental-simulation analysis of stress and strain partitioning in multiphase alloys, Acta Mater., 81(2014), p. 386. DOI: 10.1016/j.actamat.2014.07.071
[10]
W.Q. Liu, J.H. Lian, N. Aravas, and S. Münstermann, A strategy for synthetic microstructure generation and crystal plasticity parameter calibration of fine-grain-structured dual-phase steel, Int. J. Plast., 126(2020), art. No. 102614. DOI: 10.1016/j.ijplas.2019.10.002
[11]
J.D. Nowak, K.A. Rzepiejewska-Malyska, R.C. Major, O.L. Warren, and J. Michler, In-situ nanoindentation in the SEM, Mater. Today, 12(2010), Suppl. 1, p. 44. DOI: 10.1016/S1369-7021(10)70144-9
[12]
H. Nili, K. Kalantar-zadeh, M. Bhaskaran, and S. Sriram, In situ nanoindentation: Probing nanoscale multifunctionality, Prog. Mater. Sci., 58(2013), No. 1, p. 1. DOI: 10.1016/j.pmatsci.2012.08.001
[13]
W.C. Oliver and G.M. Pharr, An improved technique for determining hardness and elastic modulus, J. Mater. Res., 7(1992), No. 6, p. 1564. DOI: 10.1557/JMR.1992.1564
[14]
S. Pathak, D. Stojakovic, R. Doherty, and S.R. Kalidindi, Importance of surface preparation on the nano-indentation stress-strain curves measured in metals, J. Mater. Res., 24(2009), No. 3, p. 1142. DOI: 10.1557/jmr.2009.0137
[15]
J.L. Hay and G.M. Pharr, Instrumented indentation testing, ASM Handbook, 8(2000), p. 232.
[16]
Z.G. Wang, H. Bei, E.P. George, and G.M. Pharr, Influences of surface preparation on nanoindentation pop-in in single-crystal Mo, Scripta Mater., 65(2011), No. 6, p. 469. DOI: 10.1016/j.scriptamat.2011.05.030
[17]
K. Jin, Y. Xia, M. Crespillo, H. Xue, Y. Zhang, Y.F. Gao, and H. Bei, Quantifying early stage irradiation damage from nanoindentation pop-in tests, Scripta Mater., 157(2018), p. 49. DOI: 10.1016/j.scriptamat.2018.07.035
[18]
O. Grässel, L. Krüger, G. Frommeyer, and L.W. Meyer, High strength Fe–Mn–(Al, Si) TRIP/TWIP steels development—properties—application, Int. J. Plast., 16(2000), No. 10-11, p. 1391. DOI: 10.1016/S0749-6419(00)00015-2
[19]
I. Gutierrez-Urrutia and D. Raabe, Dislocation and twin substructure evolution during strain hardening of an Fe–22wt.%Mn–0.6wt.%C TWIP steel observed by electron channeling contrast imaging, Acta Mater., 59(2011), No. 16, p. 6449. DOI: 10.1016/j.actamat.2011.07.009
[20]
K.G. Chin, C.Y. Kang, S.Y. Shin, S. Hong, S. Lee, H.S. Kim, K.H. Kim, and N.J. Kim, Effects of Al addition on deformation and fracture mechanisms in two high manganese TWIP steels, Mater. Sci. Eng. A, 528(2011), No. 6, p. 2922. DOI: 10.1016/j.msea.2010.12.085
[21]
M. Koyama, E. Akiyama, and K. Tsuzaki, Hydrogen embrittlement in a Fe–Mn–C ternary twinning-induced plasticity steel, Corros. Sci., 54(2012), p. 1. DOI: 10.1016/j.corsci.2011.09.022
[22]
X.F. Wang, X.P. Yang, Z.D. Guo, Y.C. Zhou, and H.W. Song, Nanoindentation characterization of mechanical properties of ferrite and austenite in duplex stainless steel, Adv. Mater. Res., 26-28(2007), p. 1165. DOI: 10.4028/www.scientific.net/AMR.26-28.1165
[23]
Y. Xia, M. Bigerelle, J. Marteau, P.E. Mazeran, S. Bouvier, and A. Iost, Effect of surface roughness in the determination of the mechanical properties of material using nanoindentation test, Scanning, 36(2014), No. 1, p. 134. DOI: 10.1002/sca.21111
[24]
M. Laurent-Brocq, E. Béjanin, and Y. Champion, Influence of roughness and tilt on nanoindentation measurements: A quantitative model, Scanning, 37(2015), No. 5, p. 350. DOI: 10.1002/sca.21220
[25]
P. Filippov and U. Koch, Nanoindentation of aluminum single crystals: Experimental study on influencing factors, Materials (Basel), 12(2019), No. 22, p. 3688. DOI: 10.3390/ma12223688
[26]
I. Gutierrez-Urrutia, S. Zaefferer, and D.Raabe, Electron channeling constrast imaging of twins and dislocations in twinning-induced plasticity steels under controlled diffraction conditions in a scanning electron microscope, Scripta Mater., 61(2009), No. 7, p. 737. DOI: 10.1016/j.scriptamat.2009.06.018
[27]
J.S. Yu, J.L. Liu, J.X. Zhang, and J.S. Wu, TEM investigation of FIB induced damages in preparation of metal material TEM specimens by FIB, Mater. Lett., 60(2006), No. 2, p. 206. DOI: 10.1016/j.matlet.2005.08.018
[28]
J.L. Zhang, S. Zaefferer, and D. Raabe, A study on the geometry of dislocation patterns in the surrounding of nanoindents in a TWIP steel using electron channeling contrast imaging and discrete dislocation dynamics simulations, Mater. Sci. Eng. A, 636(2015), p. 231. DOI: 10.1016/j.msea.2015.03.078
[29]
A. Barnoush, M.T. Welsch, and H. Vehoff, Correlation between dislocation density and pop-in phenomena in aluminum studied by nanoindentation and electron channeling contrast imaging, Scripta Mater., 63(2010), No. 5, p. 465. DOI: 10.1016/j.scriptamat.2010.04.048
[30]
A. Montagne, V. Audurier, and C. Tromas, Influence of pre-existing dislocations on the pop-in phenomenon during nanoindentation in MgO, Acta Mater., 61(2013), No. 13, p. 4778. DOI: 10.1016/j.actamat.2013.05.004
[31]
R.D.K. Misra, Z. Zhang, Z. Jia, P.K.C. Venkat Surya, M.C. Somani, and L. P. Karjalainen, Nanomechanical insights into the deformation behavior of austenitic alloys with different stacking fault energies and austenitic stability, Mater. Sci. Eng. A, 528(2011), No. 22-23, p. 6958. DOI: 10.1016/j.msea.2011.05.068
[32]
D. Lorenz, A. Zeckzer, U. Hilpert, P. Grau, H. Johansen, and H. Leipner, Pop-in Effect as Homogeneous Nucleation of Dislocations During Nanoindentation, Phys. Rev. B: Condens. Matter, 67(2003), No. 17, art. No. 172101. DOI: 10.1103/PhysRevB.67.172101
[33]
G. Gottstein, Physical Foundations of Materials Science, Springer, Berlin, 2004.
Xu Long, Yaxi Li, Ziyi Shen, et al. Review of uniqueness challenge in inverse analysis of nanoindentation. Journal of Manufacturing Processes, 2024, 131: 1897.
DOI:10.1016/j.jmapro.2024.10.005
2.
Federico Bruno, Georgios Konstantoupoulos, Edoardo Rossi, et al. Advanced microstructural characterization in high-strength steels via machine learning-enhanced high-speed nanoindentation and EBSD mapping. Materials Today Communications, 2024, 39: 109192.
DOI:10.1016/j.mtcomm.2024.109192
3.
Yan-ping Bao. Editorial for special issue on USTB-RWTH Aachen’s 40 years of collaboration. International Journal of Minerals, Metallurgy and Materials, 2021, 28(5): 739.
DOI:10.1007/s12613-021-2297-z
Search
DownLoad:
