ISIJ International
Online ISSN : 1347-5460
Print ISSN : 0915-1559
ISSN-L : 0915-1559
Regular Article
Digital Image Correlation Measurement of Localized Deformation in Carbon Steel in the Presence of Hydrogen
Shinnosuke AomatsuRyosuke Matsumoto
Author information

2014 Volume 54 Issue 10 Pages 2411-2415


Although there have been many reports related to localized plastic deformation in the presence of hydrogen, this process has not been clearly elucidated to date. In our previous report, we conducted tensile tests on hydrogen-charged carbon steel S25C. Using the digital image correlation method, we were able to measure the time evolution of the equivalent strain distribution and the equivalent strain rate distribution up to the crack initiation around the notch bottom. We then revealed that the hydrogen induced an increase in the equivalent strain rate during the deformation process and a decrease in the equivalent strain when the time the crack started to grow. In this study, we performed similar measurements on carbon steels S15C and S55C and evaluated how the deformation progressed until the crack started to grow depending on the carbon content. The results showed a decrease in the local equivalent strain when the time the crack started to grow in all of the investigated materials, but we could not identify any clear trend with respect to a carbon-content dependence. However, we observed that smaller carbon contents (i.e., higher ferrite volume fractions) resulted in greater significance of the localization of the deformation due to the hydrogen and in a greater rate of increase in the local strain rate. In addition, on the basis of the equivalent strain distribution, we revealed that the deformation tended to localize in the ferrite phase.

1. Introduction

Despite previous and ongoing research, hydrogen embrittlement remains a topic of debate. The literature contains numerous previous reports concerning the localization of deformations in the presence of hydrogen.1,2,3,4,5) In this respect, various research studies that use microscopic experiments have been conducted, and various mechanisms have been suggested to explain this localization.6,7,8,9) In particular, many studies have been conducted that relate the effect of hydrogen on the movement of dislocations.10,11,12,13,14) Because the localized deformation is the result of interaction between many lattice defects, measurements on a macroscopic scale are also important. In addition, the strong interaction of hydrogen with lattice defects other than dislocations15,16,17,18) must not be overlooked.

Note that quantitative measurements using macroscopic observations are not sufficient to reveal either the degree of localization or the process involved in the localization of the deformation. Elucidation of the detailed process involved in this localization is important to explain the hydrogen embrittlement phenomena and to evaluate the strength of structural member exposed to hydrogen. In addition, the process involved in this localization and the extent to which it occurs are considered to be intimately dependent on the material texture and its composition.

In our previous report,19) while conducting slow-strain-rate testing (SSRT) of carbon steel S25, we could visualize the deformation field until the crack started to grow using the digital image correlation method (DICM).20,21,22) In addition, by comparing the time evolution of the distributions of both the equivalent strain and the equivalent strain rate in the presence or absence of hydrogen, we determined how hydrogen actually affects the macroscopic deformation behavior. The two obvious main effects were an accelerated increase in the equivalent strain rate during the deformation process and a decrease in the equivalent strain at the crack initiation. In the present work, while conducting similar measurements on S15C and S55C carbon steels, we demonstrate how the two hydrogen effects that we previously reported change according to variations in the carbon content or, equivalently, the ferrite volume fraction.

2. Experimental

2.1. Specimen Preparation

Because the experimental procedure used to prepare the specimens and the evaluation techniques are the same as those previously reported,19) we provide only a brief description of these methods here. Rods of S15C, S25C, and S55C carbon steel were processed into thin specimens with a thickness of 0.5 mm, as shown in Fig. 1, using wire electrical discharge machining. To limit the origination point of deformation for microscopic observation, a 0.26(± 0.02)-mm-wide notch with a length of 2.5 mm was introduced at the center section. After this specimen was annealed, the surface was polished mechanically to a mirror finish. To prepare the hydrogen-charged material, hydrogen charging was subsequently conducted for 24 h using an immersion method in an FIP (Federation International de la Precontrainte) bath23) (20%mass-NH4SCN). In the case of the hydrogen-uncharged material, nital etching was performed to form the surface pattern necessary for the DICM. Here, the hydrogen-uncharged and hydrogen-charged S15C materials are denoted as No_charged_15 and Charged_15, respectively. Similarly, the corresponding S25C and S55C specimens are denoted as No_charged_25, Charged_25, No_charged_55, and Charged_55, respectively.

Fig. 1.

Geometory of the specimen.

2.2. SSRT and DICM

For the hydrogen-charged material, SSRT with a tensile rate of 0.03 mm/min was started within 5 min after the hydrogen charge. During SSRT, the vicinity of the notch bottom was photographed at 15-s intervals with a microscope (KEYENCE, Digital Microscope VHX-900) at a magnification of 300× (0.63 μm/pixel, image resolution 1600 × 1200 pixels). Subsequently, application of the DICM to the collected images yielded the distributions of the equivalent strain and the equivalent strain rate. Because a large deformation occurred in the notch bottom, the equivalent strain was evaluated by taking the sum of the equivalent strain increments, which considers movement of the material points themselves.22)

The amount of the strain measured depends on the spatial resolution in which the strain is defined. In general, the variation in the spatial distribution of the strain increases as the spatial resolution is increased, and the absolute value also increases. In this study, the dimensions of the subset used in the strain measurement by the DICM are 35 × 35 pixels (≈ 22 × 22 μm2), which is similar to the crystal grain size of S15C and S25C and to the crystal size in the ferrite phase of S55C. Therefore, the strain measured here reasonably represents the deformation on the scale of the crystal grains.

2.3. Analysis of the Hydrogen Content by Thermal Desorption Analysis (TDA)

The hydrogen content of a fragment that had been immersed along with the specimen in the FIP bath was measured using thermal desorption analysis (TDA) at a heating rate of 5°C/min. TDA was started within 5 min after charging was completed. In this study, the hydrogen released at ≤200°C is considered diffusible hydrogen, and the hydrogen content of the specimen is represented by this amount.24,25) The hydrogen desorption spectra for each individual specimen, as measured by TDA, are shown in Fig. 2. The diffusible hydrogen concentrations estimated using TDA are summarized in Table 1. The hydrogen concentrations in the hydrogen-charged materials are approximately the same. However, comparing the hydrogen desorption spectra shown in Fig. 2, it is confirmed that Charged_15 contains more hydrogen which is released around 75°C and 100°C, and less hydrogen which is released above 125°C than Charged_25 and Charged_55. Thus, it is considered that Charged_15 contains more hydrogen with high diffusibility than the others. The diffusible hydrogen concentration in the hydrogen-uncharged materials was approximately 0.01 mass ppm for all of the specimens.

Fig. 2.

Hydrogen desorption spectra: The released hydrogen content was analyzed by the thermal conductivity detector at 5-min intervals.

Table 1. Diffusible hydrogen concentration in the specimens.
Hydrogen concentration [mass ppm]0.660.660.68

3. Experimental Results and Discussion

3.1. Stress–displacement Relation

The nominal stress curves are shown in Fig. 3. The vertical axis represents the nominal stress at the notched section, and the horizontal axis is the displacement of the tensile-testing-machine jig (i.e., the crosshead displacement). Because of the play of jig and the deflection of the specimen, the data acquired immediately after the start of the tensile tests are not shown. The crack started to grow from the notch bottom at the points shown as triangles (Point A, A’, B, B’, C, and C’ for No_charged_15, Charged_15, No_charged_25, Charged_25, No_charged_55, and Charged_55, respectively) in the figure (these points are hereafter referred to as crack initiation points). At the crack initiation points for S25C (No_charged_25 and Charged_25) and S55C (No_charged_55 and Charged_55), a crack started to grow immediately after they formed. In the case of S15C (No_charged_15 and Charged_15), a tiny crack was generated around the notch bottom immediately after macroscopic yielding of the sample and started to grow at the crack initiation point.

Fig. 3.

Nominal stress - crosshead displacement relation.

In all of the materials, the nominal stress at the crack initiation point and the elongation after yielding (lyie) were both reduced by the effect of hydrogen. The reduction in lyie is particularly significant in the case of S15C, where the lyie of the hydrogen-charged material decreased 73% relative to that of the hydrogen-uncharged material. The rate of reduction was approximately 25% in the cases of S25C and S55C. In contrast, almost no visible difference was observed between the stress curves of the hydrogen-charged and hydrogen-uncharged materials to the crack initiation point. The only notable difference was that the work-hardening coefficient was slightly larger in the hydrogen-charged case.

3.2. Equivalent Strain Distribution at the Crack Initiation Points

Figures 4 and 5 show the equivalent strain distributions in the vicinity of the notch bottom for the specimens S15C (No_charged_15 and Charged_15) and S55C (No_charged_55 and Charged_55), respectively, immediately preceding the crack initiation point (Point A, A’, C, and C’ in Fig. 3). In these figures, the distribution maps obtained from the DICM were made semi-transparent and overlaid on the micrographs such that the relationship between the texture and the strain distribution could be observed. In the cases of the No_charged_15 and No_charged_55 specimens, maximum equivalent strains of approximately 1.4 and 0.8, respectively, accumulated at the notch bottom (we let the maximum values of these equivalent strains be ε eq,c max ). In contrast, in the cases of the Charged_15 and Charged_55 specimens, ε eq,c max of approximately 0.7 and 0.4, respectively, accumulated. The cracks started to grow from these maximum-strain regions. In either case, the proportional decrease in ε eq,c max in the hydrogen-charged material was approximately 50%. Notably, as has been reported previously19) for S25C under the same conditions, ε eq,c max of the hydrogen-uncharged and hydrogen-charged material were 1.6 and 1.2, respectively, which is a decrease of 25%. We observed that all of the investigated materials, when charged with hydrogen, exhibited decreased equivalent strains at the crack initiation point. Nevertheless, a clear relationship between the carbon content and ε eq,c max was not readily apparent. The results obtained are summarized in Table 2.

Fig. 4.

Equivalent strain distribution around the notch bottom just before crack initiation point in S15C; (a) No_charged_15 (Point A in Fig. 3), and (b) Charged_15 (Point A’ in Fig. 3).

Fig. 5.

Equivalent strain distribution around the notch bottom just before crack initiation point in S55C; (a) No_charged_55 (Point C in Fig. 3), and (b) Charged_55 (Point C’ in Fig. 3).

Table 2. Summary of observied hydrogen effects.
H charge in FIP bathNo24 hNo24 hNo24 h
Elongation from yielding to crack initiation point lyie [mm]0.450.120.450.350.200.15
Decrease rate73%22%25%
Nominal stress at crack initiation point [MPa]389313420405548543
Decrease rate20%4%0%
Equivalent strain at crack initiation point ε eq,c max
Decrease rate50%25%50%
Localized deformationQuite significantSignificantLocalize to ferrite phase

Despite the ε eq,c max for S55C greatly decreasing in much the same way as that for S15C by hydrogen, the plastic elongation to the crack initiation point lyie did not decrease as much as in the case of S15C. This result indicates that, when measured on the scale of crystal grains, deformation is more localized in S15C by hydrogen.

3.3. Relationship between the State of Deformation and the Deformation Rate at a Local Material Point

The equivalent strain and equivalent strain rate were measured using the DICM at the measurement points in all the images, which were collected in 15-s intervals. We selected all the points that had an equivalent strain within a constant range (εeq ± 0.05) from the data collected during the SSRT up to the crack initiation point and then evaluated the average of the equivalent strain rates at these points ( ε ˙ ¯ eq ). The relationship between equivalent strain and ε ˙ ¯ eq is shown in Fig. 6. Notably, an adequate statistical average cannot be obtained for 50 or fewer evaluation points; therefore, such cases are not plotted in the figure. In Fig. 6(a), a rapid increase is evident in the deformation rate of the Charged_15 specimen with increasing equivalent strain. For example, although the deformation rate relative to that of the No_charged_15 specimen for 0.1 < εeq < 0.2 was approximately 1.8 times greater, it increased to approximately 3.1 times greater for 0.5 < εeq < 0.6. This rate of increase is greater than that previously reported for S25C.19) However, Fig. 6(b) reveals that no increase in the deformation rate due to hydrogen is evident in the case of S55C; i.e., deformations in materials with lower carbon contents (higher ferrite volume fractions) are localized by the further increase in the deformation rate at the deformed area.

Fig. 6.

Relationship between local equivalent strain and averaged equivalent strain-rate.

3.4. The Correspondence between Equivalent Strain Distribution and the Microstructure

In these measurement results, displacement of the material points within the measured region was also obtained. Thus, the equivalent strain distribution can be mapped on the micrographs taken prior to deformation and in which the ferrite or pearlite phase is easily discerned. Figures 7 and 8 show the equivalent strain distributions for S25C and S55C, respectively, immediately before crack initiation point (Point B, B’, C, and C’ in Fig. 3), where the distributions are superimposed on the sample photographs taken before deformation. The dark region is the pearlite phase, and the light region is the ferrite phase. Because the aim here is to know the correspondence of the microstructure with the extent of deformation, the transparency of the equivalent strain distributions was increased compared to Figs. 4 and 5. Now, in the case of S15C, which contains a high ferrite volume fraction, no correspondence could be discerned between the phase and the extent of deformation (data not been shown). However, as shown in Fig. 8, although the deformation was concentrated in the ferrite phase in the case of the Charged_55 specimen, a large strain accumulated in the pearlite phase in the case of the No_charged_55 specimen. In addition, the strain in the charged material also tends to be biased to the ferrite phase with S25C, although not as clear as in the S55C case (Fig. 7). Overall, we reasonably concluded that the equivalent strain in charged material tends to concentrate in the ferrite phase. This point qualitatively agrees with the point that localized deformation occurs to a greater extent where the ferrite volume fraction is higher.

Fig. 7.

The relationship between microstructure and equivalent strain distribution in S25C: Equivalent strain distributions just before crack initiation was superimposed on the microscope image captured before SSRT. (a) No_charged_25, and (b) Charged_25.

Fig. 8.

The relationship between microstructure and equivalent strain distribution in S55C: Equivalent strain distributions just before crack initiation was superimposed on the microscope image captured before SSRT. (a) No_charged_55, and (b) Charged_55.

4. Conclusion

SSRT was performed for three different types of carbon steel: S15C, S25C, and S55C. By applying the DICM to observation images obtained in situ, we evaluated the effect of hydrogen on the plastic deformation process.

(1) For all of the investigated materials, the equivalent strain at the crack started to grow decreased because of the effect of hydrogen. The proportional decrease was 50% for S15C, 25% for S25C, and 50% for S55C, and no dependence on the carbon content (the ferrite volume fraction) was observed.

(2) When deformation progresses in localized regions, the deformation rate increases more remarkably in materials with a high ferrite volume fraction. In particular, practically no change was observed in the case of S55C.

(3) By mapping the equivalent strain distribution onto the image captured prior to SSRT, we studied the correspondence between the extent of deformation and the microstructure. Consequently, we observed that the deformation tends to localize in the ferrite phase in the presence of hydrogen.

In this paper, we have evaluated the strain distribution at a spatial resolution on the scale of the grain size. We believe that the results presented in this paper have captured the characteristics of deformation in carbon steel as a dual-phase material. However, on the basis of measurements performed at a higher spatial resolution, greater equivalent strains and strain rates are believed to be observed locally.

Based on the results presented in sections 3.3 and 3.4, hydrogen appears to play a role locally in impeding work hardening in the ferrite phase. In contrast, in the stress curves that represent the averaged response of the material, the work-hardening coefficient tends to increase because of the hydrogen (for example, Fig. 3, or the results for pure iron in Ref. 15)). We believe a more multipronged approach is ultimately required to completely understand hydrogen embrittlement.


This study was partially supported by Grant-in-Aid for Young Scientists (A), 23686022, by the Ministry of Education, Culture, Sports, Science and Technology, Japan.

© 2014 by The Iron and Steel Institute of Japan