Nanoindentation of silicon implanted with hydrogen: effect of implantation dose on silicon's mechanical properties and nanoindentation-induced phase transformation

In the past, the mechanical behavior of single-crystal silicon (c-Si) implanted with hydrogen was studied from two aspects. The first one is the so-called 'smart cut' process, which is essential for silicon-on-insulator (SOI) technology [1–4]. In this process, hydrogen ions are implanted into silicon (Si) through a thin oxide layer. With later annealing, a film of Si on oxide is exfoliated, with mechanical cleavage happening at the depth of the projected implantation range. In the 'smart cut' process, the implanted layer is mainly investigated from a structural viewpoint [1–4]. Machining of Si implanted with hydrogen is the second aspect [5–7]. Crystalline Si is important for optical devices, but it is brittle and difficult to cut [8–11]. With hydrogen implantation, some of the cutting parameters like the critical depth of the cut and the cutting force were mitigated [5–7].

Implantation of hydrogen damages Si–Si bonds, and creates vacancies and various Si–H complexes [1–4]. The elastic modulus (E) is a property of the material and is determined by the nature of the bonds. Generally, the hardness (H) of material is considered not to be a material property and is affected by defects in the crystalline material. Therefore, it is natural to expect that implanted hydrogen can have an impact on Si's hardness and modulus. H and E are commonly obtained from nanoindentation experiments [12].

To the best of our knowledge, there is only one publication that addresses the hardness and elastic modulus of Si implanted with hydrogen [13]. In it, it was mentioned that the hardness was not modified by hydrogen implantation. A careful examination of the hardness-displacement graph from this reference, obtained from hydrogen-implanted Si, shows that in the implanted region the hardness is higher than in the bulk of Si, which contradicts the statement mentioned above. This statement could depend on the hydrogen concentration, and although Guo et al [13] performed implantation with different doses, the group did not present any relation between the implantation dose and H and E. In the previous work by Jelenković et al [7], high doses of hydrogen were implanted in a depth of ~1–1.55 µm, which is however well away from the ductile depth of cutting (~100–200 nm) of lightly doped Si [9]. In that work [7], based on load-displacement curves, it was suggested that implanted Si could have different mechanical properties, but hardness and the elastic modulus were not investigated.

In addition to H and E evaluation, nanoindentation is suitable to investigate the influence of high pressure on c-Si, which is an important scientific issue. Applying nanoindentation on c-Si, it was observed that during the unloading period, when the high pressure is released, the formation of a new Si phase may happen [14–18]. This observation is significant for ductile cutting of Si. In such cutting, under the cutting tool, high pressure is developed as well [8–10]. Also, upon traversing the cut Si region, the pressure of the cutting tool is released. This can lead to an amorphous layer formation, similar to the unloading process of nanoindentation. In relation to the nanoindentation-induced phase transformation in c-Si, it was reported that high hydrogen concentration in Si promotes an amorphous Si phase formation [19]. But the role of the hydrogen concentration on this phenomenon was not properly addressed [19].

Overall, a brief review of the literature proves that there is insufficient information on: (i) the mechanical properties of hydrogen-implanted Si, and (ii) its response to high pressures. In this article, these matters were addressed through appropriate variation of hydrogen ion implantation doses and energies and nanoindentation parameters. In addition, different physical analyses were applied in order to explain the mechanical and material properties of the implanted layer before and after nanoindentation.

Si wafers with (1 0 0) orientation and a resistivity of 5–25 ohm cm were used for the experiment. The wafers were boron doped with one side polished. Si wafers were wet oxidized to obtain 300 nm thick oxides. H⁺ ions were implanted into the Si through the oxide. The implantation was carried out with the double ion energy of 75 and 40 keV, with the 75 keV implantation executed first. It should be noted that the oxidized, but not the implanted sample, labeled OS, was used as a reference sample. Details of the samples' labeling and preparation are given in table 1. The wet oxides were removed from all samples in a diluted hydrofluoric acid solution before nanoindentation.

Nanoindentation was performed with indenters with a Berkovich tip. In the dynamic mode of indentation, the hardness and modulus depth profile were obtained with the maximum load (P_max) of 50 mN. The depth was recorded by a three-plate capacitive displacement sensor which was attached to the shaft with the nanoindentation tip at its end, with an estimated accuracy of 0.15 nm. For Si phase transformation studies and load-displacement curves analysis, the maximum load varied from 10 to 50 mN, with a load/unload rate of 0.5 mN s⁻¹ and a hold time of 1 s. A fused quartz standard of known hardness and modulus was used as a reference to monitor the performance of the nanoindenter system. The indenter's control program can automatically calibrate the tip area function and the load frame stiffness. Such calibrations are carried out according to the procedures proposed by Oliver et al [20]. The material properties of Si in different stages of the experiment were monitored by Rutherford backscattering (RBS), high resolution x-ray diffraction (HRXRD), secondary mass ion spectrometry (SIMS) and µ-Raman spectroscopy (532 nm laser wavelength with a spot size of ~1 µm in diameter). Stopping and range of ions in matter (SRIM) simulations were used to predict hydrogen and implantation-induced defect distribution [21].

3.1. Hardness and modulus

The SRIM simulation of H⁺ ions implantation was performed through the 300 nm thick oxide. With the implantation scheme, as outlined in table 1, the maximum depth of the implanted hydrogen, from the surface of the Si, was obtained as ~500 nm. Calculating from the oxide surface, the ion ranges were about 460 and 710 nm for 40 and 75 keV implantation energies, respectively.

To obtain a depth profile of hardness and modulus of the whole implantation depth, nanoindentations were performed in the dynamic mode up to the maximum load of 50 mN. Hardness (H) and elastic modulus (E) depth profiles of Si, implanted to different doses, are shown in figures 1 and 2, respectively. As shown in these two figures, the whole implanted depth range, predicted by the SRIM simulation, is probed by the indentation load. Based on the calibration of the nanoindentation system with fused quartz, the measured H and E are not reliable to the depth of 20–30 nm. In figure 2, a dip in E is observed at about 130 nm. No such dip is recorded for the reference fused quartz, and at this stage it is not clear why it appears in the bare and implanted Si. Curves for the hardness of samples B and C have broad peaks in the depth range of 100–500 nm which suggests the role of implanted hydrogen. However, none of the curves peaks at about 160 and at 410 nm as one may expect after SRIM simulation. The near surface peaking of hardness for the OS sample is observed; it is speculated that the surface defects generated during high temperature oxidation causes this. The same explanation can be implemented for the sluggish growth of H for the A sample. However, more evidence is needed to substantiate this speculation.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The averaged values of H and E, obtained in the range from 75 nm to 470 nm and for 10 tests for each sample, are given in figures 3 and 4.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

3.2. Pop-out analysis

During the nanoindentation of Si, a phase change may happen upon unloading under certain conditions. The change in phase is indicated by the appearance of a pop-out (a sharp discontinuity), kink (gradual discontinuity) or elbow (hysteresis like) in the unloading curve. The relation of the hydrogen implantation level on the pop-out appearance was studied through load/unload rates of 0.5, 1.5, 3 and 6 mN s⁻¹ and with a maximum load of 50 mN. The pristine Si (sample OS) is characterized by the presence of well-defined pop-outs with very few kinks, as presented in figure 5.

Zoom In Zoom Out Reset image size

It has to be mentioned that all samples have a low dopant concentration (5.2  ×  10¹⁴–2.7  ×  10¹⁵ cm³). Such boron concentration was selected intentionally in order to avoid premature amorphous Si formation at lower load/unload rates. It is known that for highly boron-doped Si, nanoindentation-induced amorphization can happen at lower unloading rates [22]. With this dopant concentration choice, the impact of implanted hydrogen on indentation-induced phase transformation is not masked by the boron dopant and enables better evaluation of pop-out, kink and elbow events.

At lower implantation doses (samples A and B), smaller pop outs, kinks and elbows appear during the unloading process (figures 6 and 7, respectively). The pop-outs or kinks occur at the load rates of 0.5 and 1.5 mN s⁻¹, while elbows appear at the rates of 3 and 6 mN s⁻¹.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The implantation with a dose of 4  ×  10¹⁶ cm⁻³ (sample C) leads to predominant elbow events (figure 8). As reported by many groups, the pop-out events indicate the creation of predominant crystalline Si phases: Si-XII and Si-III [14, 15]. The elbow is the sign that the Si-II phase is converted into a-Si during the unloading path [14–18], while the kink has both: the crystalline and amorphous phase [14].

Zoom In Zoom Out Reset image size

In the inset of figure 1 the typical behavior of load-displacement curves, probed with P_max  =  10 mN and a load/unload rate of 0.5 mN s⁻¹, is given. Only a small part of the load-displacement curve, close to the maximum penetration depth, is shown, which points out that the implanted Si is harder to penetrate. It should be noted that the rightmost curve in the inset is related to the OS sample and the leftmost to the C sample. Otherwise, all the load-displacement curves display only elbows in the unloading curve.

The load-displacement curves can be fitted with a power law at the very beginning of the unloading part, where the unloading process is purely elastic. The illustration of the fitting is given in figure 5 for OS samples at the 6 mN s⁻¹ load/unload rate, but can be applied to all investigated load–unload curves, as described in different publications [20, 23].

It should noted that the curves in figures 5–8 are displaced horizontally from each other by about 200 nm in order to better observe pop-out, kink or elbow effects.

In c-Si, nanoindentation under nominally identical probing conditions produces a dispersion in load/unload curves in regard to the pop-out length, the pop-out position in the unloading curve and other features. As figure 9 shows, there is some scattering in the distribution of pop-outs, kinks and elbows events, with a tendency of the low load/unload rate and no hydrogen implantation favoring pop-out events, while the high load/unload rates and high hydrogen implantation dose promote a predominant elbow effect.

Zoom In Zoom Out Reset image size

The length of the pop-outs can be related to the amount of crystalline phase [14–17, 22, 23] and the pop-out lengths outlined in this article are analyzed only for the lowest load/unload rate of 0.5 mN s⁻¹, as given in figure 10. According to the figure, the crystalline level of the Si-XII and Si-III phases in the impression reduces with the increase of hydrogen concentration and is leveled to zero for the highest implantation dose.

Zoom In Zoom Out Reset image size

3.3. Raman spectroscopy

Raman spectroscopy is a very convenient tool to study indentation-induced phase transformations [14–18, 22] and is applied in this study as well. From the above-mentioned figures, the most significant difference in pop-outs for load/unload rate-hydrogen concentration dependence is seen at the 0.5 mN s⁻¹ load/unload rate and these nanoindentation impressions were analyzed by Raman spectroscopy. Typical Raman spectra from the nanoindentation impressions are given in figure 11. The appearance of the Raman peaks for non-implanted Si is in line with the published results [14–18, 22]. Besides the Si-I phase at around 520 cm⁻¹, the most prominent peaks associated with the Si-XII phase are located at 355, 377, 399, 440 and 492 cm⁻¹. The bands at 440 and 492 cm⁻¹ are related to the Si-III phase. The incorporation of hydrogen reduces the intensity of the peaks related to the Si-III and Si-XII phases and they disappear with the highest implantation dose (see figure 12). The absence of these peaks in sample C is associated with the elbow presence. In other words, the amorphous phase is formed and should be observed as broad peaks at about 160, 300 and 470 cm⁻¹. In a more sensitive measurement these broad peaks were indeed observed, but these are not presented in this article.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The observed averaged hardness for non-implanted Si is about 12.8 GPa which is close to the published results [12]. A steady increase in hardness with the implantation dose could be the consequence of a pile-up effect during the indentation tests. A convenient way to pinpoint the pile-up effect is to calculate the h_f/h_m ratio, where h_f and h_m are the residual and maximum indentation depths in the unloading curve respectively. This ratio depends on the material property and should not depend on the probing depth [20]. For OS, A, B and C samples this ratio was found to be in the range of 0.41–0.51 for the maximum probing loads of 10, 20, 30 and 50 mN and with the load/unload rate of 0.5 mN s⁻¹. In all the 10 tests for each load, the scattering of h_f/h_m was significant and no correlation between the implantation dose and the h_f/h_m could be deduced. The h_f/h_m was also calculated from the cumulative load/displacement curves of the dynamic measurement. The obtained data for the four samples were 0.58, 0.62, 0.62 and 0.62, respectively. For h_f/h_m  ⩽  0.7, the pile-up is considered small [20] which suggests that the result for hardness in figure 1 is correct and is not affected by the pile-up effect. One of the possible reasons for the greater hardness for the implanted Si could be a vertical strain caused by implantation, which will be discussed later with the XRD measurement.

The hardening of Si with hydrogen implantation is implied by the inset in figure 5. It demonstrates the lower penetration depth for the applied maximum load, in this case 10 mN for a higher implantation dose. The measurements with different maximum loads and load/unload rates lead to similar behavior. This result differs from the work of Guo et al [13] in which a single energy implantation was used, but with similar implantation doses. The shift of load-displacement curves in this work (figure 5, inset) to a shallower depth could be expected following a similar finding in an investigation with triple energy implantation of hydrogen and a projection depth in the range ~1–1.55 µm [7]. In that publication, hydrogen surface concentration is estimated to be about five times lower than the lowest surface concentration in this work for which D  =  0.5  ×  10¹⁶ cm⁻².

By implanting hydrogen in Si, numerous changes are introduced in the target material which may influence the mechanical and material properties of Si. The changes in the electrical properties are not the concern of this article and are not investigated, though they may be affected. The alterations include (i) the presence of hydrogen, (ii) implantation-induced defects, (iii) lattice constant modification and (iv) the chemical reaction of hydrogen with broken Si bonds. In this article, the first of the three were registered by appropriated measurements, while the fourth will be discussed based on various publications. The material characterization of the implanted Si is therefore important to define before proceeding with the further analysis of nanoindentation-related findings.

The illustrative example of hydrogen distribution, obtained by SIMS profiling, is given for sample C in figure 13. It should be noted that the vertical axis has arbitrary units. The positions of the peaks in figure 13 are close to those obtained by SRIM simulation, as described earlier. It predicts hydrogen concentration at the deeper peak of about 3.2  ×  10²¹ cm⁻³ for the implantation dose of 4  ×  10¹⁶ cm⁻³. Here, the assumption is made that during ion implantation there was no significant distribution of hydrogen. During the implantation, samples were not cooled and self heating was below 70 °C. In the earlier work of the authors [7], hydrogen distribution was found to be marginally altered after annealing at 400 °C, which supports the speculation about hydrogen concentration at the deeper peak (at about 400 nm from the surface). The shallower hydrogen peak at ~200 nm is slightly higher due to the nature of the double energy implantation process. Therefore, at the very surface of the Si, the hydrogen concentration is about 3  ×  10²⁰ cm⁻³, while at the valley between the two peaks it is 7  ×  10²⁰cm⁻³. By varying the implantation doses from 0.5  ×  10¹⁶ to 4  ×  10¹⁶ cm⁻³, the hydrogen profile should vary approximately in proportion to the implantation dose.

Zoom In Zoom Out Reset image size

Figure 13 confirms that the maximum selected load of 50 mN for dynamic indentation and pop-out analyzes was sufficient to probe the whole implanted region by nanoindentation. However, the indentation scheme was not sensitive enough to reproduce a peak-like profile either for hardness or for modulus as in figure 13 for hydrogen, even for sample C, as is evident from figures 1 and 2. The average hardness and modulus shown in figures 3 and 4 are obtained from the depth range ~ 75–470 nm, as indicated in figure 12 by a horizontal line.

At the stopping ranges, implanted hydrogen ions create Si vacancies and different Si–H complexes [2, 3]. The formation of Si–H complexes is due to the high reactivity of hydrogen, which reacts with Si atoms or dangling bonds during implantation. Created defects by implantation disturb the crystallinity of Si. The generated defects were characterized by channeling the RBS technique and the result is shown in figure 14. No obvious defects are observed for D  =  0.5  ×  10¹⁶ cm⁻² while the disorder of Si grows with higher implantation doses, as depicted by the peaks at channel numbers 490 and 520. These peaks are supposed to be located at the depth of the hydrogen peaks in figure 12 [2, 3].

Zoom In Zoom Out Reset image size

The consequence of the implanted hydrogen and generated defects is that the crystalline lattice of Si becomes swollen, straining the lattice. The straining of the lattice can be easily detected using HRXRD measurement. Typical HRXRD records for implanted samples are shown in figure 15. In this figure, HRXRD was performed in a θ  −  2θ configuration around the Bragg reflection (4 0 0). The vertically-induced strain is indicated by the presence of fringes on the left-hand side of the peak at 69.14°. The presence of the fringes is widely published for hydrogen-implanted semiconductors [4]. From this figure it is seen that the smaller the dose is, the fewer fringes appear. The measurement for the pristine Si does not show any fringes [6] and is not included in figure 15. The maximum strain of the defective layer can be extracted from the further-most fringes which are 0.002, 0.006 and 0.015 for samples A, B and C, respectively. These values are close to the values reported for single energy hydrogen ion implantation with similar implantation doses [4].

Zoom In Zoom Out Reset image size

A change in the Si phase can also happen when machining Si [9, 10]. In a plunge cut experiment [9] and by the diamond turning [10] of Si, a-Si phase formation was observed on the surface of the cut Si. As shown in this work, hydrogenation by implantation stimulates the formation of a-Si in nanoindentation tests. Facilitating this phase might be useful in solid-state epitaxial re-growth of the damaged layer caused by machining. It is possible, because the a-Si/c-Si interface of the machined Si is Si dioxide free, as observed by transmission electron microscopy [9, 10]. It is known that only a polycrystalline Si layer is grown upon the annealing of a-Si deposited on Si dioxide [24]. In a SiO₂-free a-Si/c-Si interface, the epitaxial growth could happen only if the nucleation of a-Si is limited to the a-Si/c-Si interface. If this speculative scenario can be realized, such epitaxial re-crystallization could be important in diamond turning of hydrogen-implanted Si wafers for electronic grade Si, a process advocated by Arif et al [11].

It is accepted that the Si-II phase is created during nanoindentation of Si during the loading step [14–17, 23]. Upon unloading, the unloading curve follows elastic relaxation to the bifurcation point (BP) where it stops following the power law. An example of a BP is given in figure 5 for the unloading rate of 6 mN s⁻¹. At the point of bifurcation, different Si phases start being created in addition to Si-I. In this work, based on the fitting of two–three curves for each kind of sample, at the BP the pressure was calculated to be in the range of 7–8 GPa. No clear relation could be established between the implantation level and the pressure at the BP. The calculated pressure values were obtained with the corrected indentation contact area after calibration with fuse silica. Otherwise, the phase transformation pressure values in this work are similar to those reported for low-doped Si [23].

Based on Raman measurements, the full width at half maximum (FWHM) of the Si peak at 520 cm⁻¹ is slightly larger for implanted Si, varying as 4.5, 4,5 4.8 and 5.4 cm⁻¹ for OS, A, B and C samples, respectively. It is a consequence of the damaged Si lattice after ion implantation. The position of the peak is the same (520 cm⁻¹) for all un-indented samples. This peak is shifted slightly to higher values for all indented samples, an indication of a compressive stress [18]. However, in all indented samples the FWHM is increased (9.6, 9.7, 10 and 13.1 cm⁻¹ for samples OS A, B and C, respectively). A larger FWHM suggests a polycrystalline [25] nature of the Si-I phase of the indentation impression. The greatest disorder is for the sample with D  =  4  ×  10¹⁶ cm⁻¹, which implies that hydrogen, during the unloading path, suppresses crystallization, which is analogous to the role of hydrogen in the thermal crystallization of a-Si, where hydrogen delays crystallization [26].

Overall, from the above analysis, load-displacement curves (figures 5–8), statistics of pop-outs, kink and elbow events (figure 9), pop-out length (figure 10), and Raman spectroscopy of the nanoindentation impressions (figure 11), there is a straightforward conclusion that hydrogen implantation favors a-Si formation during nanoindentation tests under the same indentation parameters. It is well known that the presence of hydrogen in a-Si can vanquish crystallization in nanoindentation tests [26]. However, it has to be noted that a-Si phase evolution is expected in non-doped Si under other nanoindentation conditions. Specifically, after nanoindentation of low-doped Si, the formation of a-Si is observed in two cases: (i) at low maximum indentation loads and (ii) at high indentation unload rates for high maximum loads [1–10]. In case (i), it is speculated that a small volume of Si is deformed during the indentation cycle, which does not allow the nucleation of the c-Si phase upon unloading [14–16, 23]. For (ii), at high unload rates, while a sufficient volume of the Si-II phase is generated, there is not enough time for nucleation to start, resulting in only a-Si phase generation once the unloading is finished [15, 16, 23]. Therefore, since the presence of H retards nucleation [26], it is normal to see that suppression of c-Si phase generation after the nanoindentation cycle is enhanced with a higher hydrogen content and that hydrogen plays this role even at smaller unload rates.

Figures 14 and 15 point out that the defects and strain in Si, which are caused by implantation, are also factors that have to be taken in consideration when discussing the suppression of c-Si phases. Ruffel et al studied nanoindentation of a-Si doped layer with hydrogen [26]. The a-Si layer was created by the Si implantation of c-Si with subsequent triple-energy hydrogen implantation and annealing in order to flatten the hydrogen depth profile and make the layer stress-free. The implanted layer, however, remained amorphous after annealing. The used doses for hydrogen implantation were similar to those applied in this article. In both cases, [26] and this article, it is assumed that the Si-II phase is created during the loading portion. The main difference is in the target material, amorphous in [26] and partially-damaged c-Si in this report. Despite this difference, qualitatively similar observations can be made: (i) the peaks of Raman shifts for the Si-XII phase and the Si-III phases get weaker with the implantation dose and (ii) the probability of pop-out appearances is reduced with H-implantation. Since the work of Ruffel et al [26] the indented layer was stress-free, while the stress is present in A, B and C samples, the common trend, ((i) and (ii)), for the two target materials suggests that disorder and stress may not be important. However, Ruffel et al also did a comparative study in which the indentation of c-Si and relaxed a-Si fabricated by Si implantation and annealing were performed [27]. That study revealed that indented a-Si produces Si-XII and Si-III phases easier than indented c-Si. This implies that less-ordered Si is prone to pressure-induced c-Si phase formation. Since the crystalline disorder (figure 14) and strain (figure 15) grow simultaneously with higher implantation doses, in the case of this study it is not possible to separate the individual influence of stress and defect density on phase transformation. Under such circumstances, no independent evaluation of the impact of defects, stress and hydrogen introduced by implantation on Si phase formation is possible.

Final remarks in the discussion section concern the value of the elastic modulus in figures 2 and 4 and the relation of H and E.

The elastic modulus was calculated using

$$\begin{eqnarray}&&1/{{E}_{\text{r}}}=\left(1-\nu _{\text{d}}^{2}\right)/{{E}_{\text{d}}}+\left(1-\nu _{\text{s}}^{2}\right)/{{E}_{\text{s}}},\end{eqnarray} \tag{ 1 }$$

where E_r is the reduced modulus, E_s and E_d are the Young modulus of Si and diamond, and ν_s and ν_d are the Poisson's ratios of Si and diamond. The reduced modulus is obtained from nanoindentation measurement and the elastic modulus for OS, A, B and C samples (figures 1–4) were calculated based on equation (1) for v_s  =  0.28 and other constants as defined in published literature [13]. The Poisson's ratio of Si may deviate in hydrogen-implanted Si. Namely, it was found that for (1 1 0) Si implanted with hydrogen at high doses, the Poisson's ratio of Si linearly increases with the implantation dose [28]. For the dose of 4  ×  10¹⁶ cm⁻², the Poisson's ratio was found to be about 0.38 [28]. Assuming that this value can be applied to the C sample in this work, the re-calculated averaged elastic modulus is reduced to 160.8 GPa from the initial averaged value of 176.2 GPa, which is taken from figure 4. The decrease of the elastic modulus with hydrogen implantation was also observed in plasma-enhanced chemical-deposited a-Si implanted with hydrogen [29]. The same report points to the decrease of hardness of this material, which is opposite to our findings. Nevertheless, the hydrogen role in Si material could depend on its history, as it was shown that in sputter-deposited a-Si the hardness grows with plasma hydrogenation [30].

There were attempts to correlate hardness with material properties, such as the bulk modulus and elastic modulus [31, 32]. Often materials with high bulk modulus have high hardness, but there are many exceptions [31]. Chung et al [32] reported that in transition metal diborides there was an almost linear dependence of hardness on the elastic modulus, but at the same time hardness did not grow linearly with the bulk modulus. This example illustrates the difficulties in correlating hardness with the elastic modulus. In our report, H increases with increasing E for the implanted samples (not shown). However, the OS sample has the highest value of modulus and the lowest hardness.

Another example of a complicated relationship between H and E (or a reduced elastic modulus) is that some hard films can be produced with the same hardness, but with a different elastic modulus [33]. Sometimes, the relation of these parameters is expressed with the H to E ratio (H/E) [32]. Often materials with a higher ratio show better resistance to wear [33]. Using the averaged values in figures 3 and 4, the H/E ratio in this investigation is calculated to be about 0. 072 for the OS sample. It steadily grows with the implantation dose, to about 0.087 for the C sample. A higher H/E ratio was also observed for Si implanted with low doses of argon ions [34].

In summary, Si was implanted with hydrogen to a depth of about 500 nm with different doses. The effect of hydrogen implantation on Si's mechanical and material properties was studied by nanoindentation tests.

• 1.  
  It was found that the hardness of Si increases with the increase of the implantation dose. The elastic modulus of hydrogen-implanted silicon is reduced. This result concerning the elastic modulus should be interpreted carefully because it is possible that the Poisson's ratio of hydrogen-implanted Si is higher than in pristine c-Si [28].
• 2.  
  From nanoidentation tests and Raman analysis, it was observed that the probability of suppression of Si-XII and Si-III nanoindentation-induced formation grows with the concentration of implanted hydrogen. The reason for this is ascribed to hydrogen which slows down the nucleation of the Si-XII and Si-III phases from the Si-II phase during unloading, similar to the effects of hydrogen on nanoindentation-induced phase transformation in a-Si [26].
• 3.  
  Due to the nature of the implantation process, the role of implantation-created strain and defects could not be separated from the hydrogen concentration effect on nanoindentation-induced Si phase transformation and the mechanical properties of implanted Si.

The elimination of strain by annealing may shed more light on the impact of hydrogen on Si's mechanical properties and is a matter for future investigation. In addition, the obtained knowledge of the hardness and modulus dependence on hydrogen implantation will be implemented in taper cutting of hydrogen-implanted Si for a better understanding of the role of implanted hydrogen on the cutting force and structural and topological properties of the cut Si.

The authors wish to thank Dr Huang Hu of the Department of Mechanical Engineering, Faculty of Science and Technology, Keio University, for providing some of the Raman and nanoindentation measurements, and also A & P Instruments Ltd. Hong Kong for the use of the iNano indenter. We would like to acknowledge the technical expertise of Mr Jack Hendriks at Western Tandetron Accelerator Facility. The work described in this paper was partially supported by the Research Committee of The Hong Kong Polytechnic University project (G-YBEX and BBX5).