Sei sulla pagina 1di 9

Bridging Length Scales in the Analysis of

Transient Tests for Metallic Materials

Aditya Gokhale Krishnaswamy Hariharan∗


Applied Mechanics, Mechanical Engineering
Indian Institute of Technology Delhi Indian Institute of Technology Madras
New Delhi 110016, India Chennai 600036, India
Email: amz148021@am.iitd.ac.in Email: hariharan@iitm.ac.in

Jayant Jain Rajesh Prasad


Applied Mechanics, Applied Mechanics,
Indian Institute of Technology Delhi Indian Institute of Technology Delhi
New Delhi 110016, India New Delhi 110016, India
Email: jayantj@iitd.ac.in Email: rajesh@am.iitd.ac.in

Heung Nam Han


Seoul National University and Research Institute of Advanced Materials
Seoul 08826, Republic of Korea
Email: hnhan@snu.ac.kr

ABSTRACT
The transient data obtained during stress relaxation test of polycrystalline materials has broader implications.
The test is influenced by the material length scale. Efforts to mathematically bridge data at different length scales
is scarce. In the present work, it is attempted to modify a recently proposed stress relaxation model with additional
coefficients to accommodate the mechanical behaviour at different length scales. The macro scale stress relaxation
test was performed using a tensile testing machine whereas the micro and nano scale specimens were tested using
indentation technique. Assuming power law rate behaviour, a scaling relation is derived initially to correlate the
indentation pressure and flow stress.

Nomenclature
σ Flow stress
σ∗ Thermal component of flow stress
σi Internal stress
ε̇ Strain rate
ε̇0 , m material constants
F Reduced contact pressure
θ angle of conical indenter
L Load applied during indentation
h penetration depth during indentation
Lcorrected Corrected load of indentation for comparison
a indenter contact radius
Pnom Nominal indentation pressure
α, β Material constants in stress relaxation model
λ Material constant corresponding to evolution of mobile dislocation density

∗ Address all correspondence for other issues to this author.


V ∗ Apparent activation volume
ξ1 to 3 Coupling coefficients that link different length scales

1 Introduction
Stress relaxation is one of the commonly used transient tests to determine the parameters such as activation volume
and long range internal stress [1, 2] that characterize thermally activated plastic deformation. In a typical stress relaxation
test, the material is deformed monotonically to a predefined strain (in the uniform deformation range) and intermittently
stopped for a certain time interval, ensuring the constancy of total strain (εtotal ). The continuous decrement of flow stress
(σ) during relaxation is attributed to the rate effect of thermal stress component (σ∗ ). This phenomenon is found to be
one of the major contributing factors in improving the formability of metallic materials when deformed using servopress
[3,4]. Systematic investigations on stress relaxation tests indicate considerable improvement in ductility, especially at higher
strain rate for a range of materials [5–7]. Ductility improvement is attributed to the annihilation of mobile dislocations
and subsequent homogenization of long range internal stress during relaxation [5]. An alternate viewpoint [8] is that the
ductility improvement is only apparent and is due to the non-isothermal condition prevailing in standard tests. In general,
stress relaxation behaviour is characterized using macroscopic tensile tests. However, microstructural studies indicate that
the mechanisms related to relaxation are substructural [9, 10]. In a recent study, Varma et al. [11] used a novel indentation
technique to explain the sub-structural mechanisms that could contribute to ductility improvement post relaxation. Therefore,
it will be of interest to compare the σ vs. t at different length scales to evaluate any additional local effect due to reduced
scale.
While other nano-scale time dependent tests such as creep has gained much attention in the past [12–15], limited attempts
have been made to study nano-indentation based relaxation [16, 17]. In general [16, 17], the transient pressure (P vs.t)
obtained from the indentation relaxation tests has been directly used to estimate the deformation parameters such as strain
rate sensitivity and activation volume. However, these parameters differ distinctly from those obtained from macroscopic
tests [16]. To the best of author’s knowledge, till date there has been no attempt to link the parameters obtained at different
scales. In the present work, it is attempted to link the experimental data obtained from different length scales using a common
mathematical relation.

2 Material and methods


The objective of the present work is to perform stress relaxation at macro, micro and nano scale lengths and compare the
relaxation data obtained thereof. Commercially available SS 316 material is chosen for the present study. The macroscopic
test was performed following the procedure described in [5]. The micro and nano scale tests were performed by indentation
technique. The micro scale indentation was conducted using Zwick Roell ZHU0.2 with a Diamond Vickers indenter. The
nanoindentation test was conducted using ASMEC UNAT Nanoindenter with a Berkovich diamond indenter of tip radius
≈ 184 nm. The nano-indentation test was carried out in selected interior locations sufficiently far from the grain boundary.
The initial penetration depth prior to stress relaxation was 25 µm (at 0.01 mm/s) and 1.98 µm (at 25 nm/s) respectively for
micro and nano indentation. All the relaxation tests were continued for 100 s. Nanosurf Nanite B atomic force microscope
was employed for scanning the indent surface. Silicon tip of size 7 nm was used to scan the surface laterally, with a constant
force of 20 nN. The image analysis and processing was done using Image Metrology’s Scanning Probe Image Processing
software (Hrsholm, Denmark).

3 Results and Discussion


The deformation strain in an indentation test is estimated from its penetration depth (h). During indentation tests,
the experiment is intermittently stopped at a predefined penetration depth without unloading the indenter. The penetration
depth indicating total strain would be expected to remain constant during the time interval, analogous to the macroscopic
tests. However, experimental measurements during indentation (Fig.1) showed that the penetration depth varied during the
relaxation test 1 . Similar observations were made in both the micro and nano-indentation relaxation tests. This is attributed to
the time independent deformation [18] due to the inhomogenous strain state surrounding the indenter. The highly localized
plastic deformation zone below the indenter is surrounded by a relatively larger elastic zone [19]. As the load (L) drops,
the contact area changes leading to transient penetration depth during relaxation. Similar observations have been reported
elsewhere [20].
The non-constancy in the total indentation depth during relaxation prevents direct comparison with the macroscopic
data. To overcome this difficulty, the L vs.t data under constant penetration depth has been corrected. Assuming the depth
change is purely due to time-independent plastic deformation, an equivalent load proportional to the depth change h(t) is

1 This is not applicable to flat indenter where the contact radius remains invariant and a constant penetration depth with time can be observed [16].
Fig. 1. Change in penetration depth during relaxation shown along with ideal constant depth

Fig. 2. Variation of penetration depth as a function of hold time. Best fit curve of the form h(t) = h0 + λ1 exp(λ2 t). [h0 = 1.98 µm, λ1 =
0.01, λ2 = −0.11]

superposed with the experimentally measured load data. The equivalent load can be estimated from the depth change (∆h)
and stiffness (S), Eq.1.

∆L = ∆h.S (1)

where ∆L is the additional load to compensate the time independent deformation.


The depth change was initially fit using a suitable non-linear equation, ∆h = h(t) − h0 to eliminate the noise. An
exponential fit, h(t) = h0 + λ1 exp(λ2t), where h0 is the initial penetration
 depth and λ1,2 are fitting constants, was found to
be suitable (Fig.2). The fit data was then multiplied with the stiffness S = ∂L
∂h obtained from the loading portion of the
indentation test (Eq.1). ∆L thus calculated is added with the experimentally obtained load vs. time data (Eq.2) to obtain the
corrected load. The original and corrected L vs.t obtained during nano-indentation test is shown in Fig.3

Lcorrected = Lexp + ∆L (2)

Lcorrected was used for further analysis including the estimation of indentation pressure, P = L/πa2c , where ac is the actual
contact radius of the indenter. As mentioned, the parameters obtained directly from the transient P vs.t data [16, 17] differed
distinctly from that obtained from macroscopic tests. The data at reduced length scales were used to infer the underlying
mechanisms during indentation [16]. However, our present objective is to establish a scaling relation that can be used to
identify the same parameters irrespective of the length scale at which the tests were performed.
Fig. 3. Experimental and corrected load vs time during relaxation

3.1 Correlation between indentation pressure and macroscopic flow stress


It is necessary to correlate the indentation pressure (P) and macroscopic flow stress, σ to compare the relaxation be-
haviour at different length scales. Following Estrin [21], the kinetic equation2 of a rate dependent material is given by
Eq.3
 1/m
ε̇
σ = σi (3)
ε̇0

where σi = σ − σ∗ is the internal stress, ε̇0 and m are material constants. For a material whose uniaxial response follows
Eq.3, Bower et.al [18] has derived the scaling relation between indentation pressure and flow stress for different indenter
geometry. The scaling relation for a conical indenter, which closely approximates the self-similar indenter such as Berkovich
and Vickers used in the present work is given by
! 1/m
σi ḣ
P=F 1/m
(4)
ε̇0 ac

where P is the indentation pressure, ḣ is the loading rate during indentation and ac is the contact radius. F is a non-
dimensional parameter referred to as “reduced contact pressure”. Su et al. [22] expressed Eq.4 in terms of nominal parameters
(indicated by subscript ‘nom’) as
! 1/m
σi c2m−1 1/m
Pnom = F 1/m
ε̇e f f (5)
ε̇0 tanθ

where ε̇e f f = ḣ/h, c = ac /anom = hc /h is the pile-up or sink-in ratio and tanθ = anom /h = ac /hc is calculated from the
geometry of the conical indenter.
In their original analysis [18], it was suggested that F and c are independent of the indenter geometry. Their non-linear
finite element analysis based on flat punch ignored the finite geometry effect, which plays a role when using axisymmetric
indenters. In a more recent work, Su et al. [22] performed full field finite element analysis for conical indenter and observed
that F and c also depend on geometry (θ) of the indenter. Both the above cited works [18, 22] aimed at establishing the
relation between the nominal indentation pressure and indentation strain rate. However, the present work intends to compare
σ vs. t at different length scales, which can be obtained directly by substituting Eq.3 in Eq.5.
The scaling relation thus obtained between Pnom and σ can be expressed as,

F m c2m−1
 
m ḣ/h
Pnom = σm (6)
tanθ ε̇

2 This is mathematically similar to the widely used power law, σ = Cε̇m to characterize strain rate dependency in metals
Fig. 4. 2D view of indent impression

Fig. 5. Line profiles on AFM image shown in (a)

The equivalent angle of conical indenter corresponding to Berkovich indenter is 2θ ≈ 70.3o [23]. In the present work,
negligible sink-in was observed as confirmed from AFM topographic analysis (Figs.4 & 5) and hence c ≈ 1. The exponent
m was obtained from the uniaxial tensile tests performed at different strain rates. ε̇e f f = ḣ/h and ε̇ in Eq.6 were calculated
from their respective tests prior to the onset of stress relaxation. The representative plastic strain when using Berkovich
indenter is ≈ 8%. The macroscopic flow stress at ε = 8% and ε̇ = 1e−2 s−1 for SS 316 was σ = 584 MPa. Relating Pnom(t=t0 )
to σε=8% , Eq.6 yields Pnom = 3.45 σ for nanoindentation and Pnom = 2.59 σ for microindentation. Substituting the above
values, ‘Pnom vs.t’ obtained from indentation relaxation tests can be directly converted to ‘σ vs.t’. The σ vs.t thus obtained
can be fit using a suitable stress relaxation equation, provided ε̇e f f /ε̇ remains constant during the entire relaxation period.
However experimental data indicate that ε̇e f f and ε̇ evolve independently with time and their relation is non-linear. We
propose to modify an existing logarithmic stress relaxation equation with coupling coefficients to accommodate the transient
nature of ε̇e f f /ε̇ during relaxation.

3.2 Scaling relation for stress relaxation


Many materials follow a logarithmic σ vs.t relation during relaxation [24], although other forms have been reported [1].
The logarithmic model reported in [1] is recently modified [5] following the procedure discussed in [9] to include the
variation of mobile dislocation density and internal stress. The modified model (Eq.7) is found to fit the experimental data
very well [5, 25].

σ(t) = σ(0) − ᾱ ln(1 + β¯ (t)t) (7)


Fig. 6. σ vs.t obtained across different length scales. The data corresponding to nano and micro indentation are scaled using eq.6

where σ0 is the flow stress at the beginning of relaxation when t = 0, ᾱ and β¯ are material constants obtained by fitting
the experimental data. The constant ᾱ = kα accounts for the variation of internal stress with time. From the definition, the
apparent activation volume, (V ∗ ) at a given temperature (T ) is related to ᾱ as V ∗ = kT /ᾱ. β¯ (t) = β¯ 0 eλt accommodates the
evolution of mobile dislocation density during relaxation, where λ is a fitting constant.
In the above equation (Eq.7), the time dependent variation of ε̇e f f /ε̇ influences the shape of the relaxation curve and
therefore the fitting constants, ᾱ , β¯ and λ obtained from the indentation data will be different from that obtained at macroscale
uniaxial relaxation test. It is proposed to modify Eq.7 as follows

σ(t) = σ(0) − ξ1 ᾱ ln(1 + ξ2 β¯ 0 eξ3 λt t) (8)

where ξ1 to 3 are the constants to accommodate the transient ε̇e f f /ε̇. ξi (i = 1 to 3) reduces to unity (ξ1 = ξ2 = ξ3 = 1) in the
original model [1, 24] corresponding to macroscopic test.
The Pnom vs.t obtained in micro and nano indentation was scaled using Eq.6 and the σ vs.t obtained across different
length scales are compared in Fig.6. As expected, the three curves did not overlap and is attributed to the transient ε̇e f f /ε̇.
Dislocation annihilation and homogenization of internal stress occur in parallel during stress relaxation [11]. In the case
of SS 316, the strain induced martensite leads to strain ageing [5] and influences the kinetics of stress relaxation [7]. The
relaxation data was fit using Eq.8 and the constants obtained are tabulated in Table 1. It is observed from Fig.6 that the scaled
nanoindentation data corresponds closely with the macroscopic data. ξ1 for nano indentation is 3.15, which is similar to the
scaling factor obtained from Eq.6. The λ value also matches with macroscropic data, ξ3 ≈ 1, indicating similar reduction
in mobile dislocation density with time. However, the scaled data from micro indentation is different both from the nano
and macro indentation. This is counter intuitive as it is expected that the micro scale data would better represent the average
macroscopic behaviour than the nanoscale data. The total stress drop in micro indentation is much less in the order 40 MPa,
against the 60 MPa in macroscale and nanoscale relaxation. Also, λ ≈ 0 indicates no change in mobile dislocation density
during relaxation.
As discussed earlier, the mechanisms related to stress relaxation are annihilation of mobile dislocations and homoge-
nization of internal stress [11], both of which are substructural in nature depending upon the evolution of dislocation mean
free path and inhomogeneity of internal stress. Therefore the similarity of the mechanical behaviour depends more on the
nature of obstacles and their distribution rather than the length scale of the sample. Albeit the different deformation mech-
anisms [16], the behaviour of obstacles during stress relaxation at nano and macro scale appears to be similar, evident from
the value of ξ3 . The deviation in the mechanical behaviour of micro indentation is presumed to be originated from two broad
effects. Although the indenter geometry of micro and nano indents are self-similar, the effect of tip radius and their contact
area are not studied in detail. The applicability of empirical relations converting the indentation depth to average strain dur-
ing relaxation test has to be validated considering this tip radius effect. The other effect is the local stress state which could
have possibly altered the distribution and strength of obstacles in the test region. It has been reported that stronger obstacles
resists relaxation [26, 27] and reduces the stress drop, similar to that observed in micro indentation. The exact mechanism in
micro scale indentation leading to this deviation in the σ vs.t however will require additional work.
Table 1. Material parameters across different length scales following eq.8
Nano Micro Macro (8%)

σ0 (MPa) 583.53
ᾱ 7.56
β¯ 0 60.36
λ -0.011
ξ1 3.15 1.32 1
ξ2 0.005 0.009 1
ξ3 1 0.16 1

4 Conclusion
In summary, stress relaxation across different length scales is described using a single logarithmic law with additional
constants (ξi ). The major advantage of the proposed method compared to other previous work [16, 17] is that the constants
ᾱ, β¯ 0 and λ remain unaffected by the length scales. Determining the relaxation parameters directly from Pnom vs.t [16, 17]
yields different values of activation volume V ∗ at different length scales. By following the proposed method, V ∗ across
different length scales is coupled with the constant ξ1 without affecting the stress relaxation parameters. It is however to be
noted that the physical interpretation of the constants ξi and the parameters that influence them needs further detailed study
and will be our scope for future work. It was also observed that the stress relaxation curve in nanoindentation was sensitive
to the indent location. Further statistical studies are required to understand the distribution of ξi within the grain.

Acknowledgements
KH would like to acknowledge the support received from Department of Science & Technology (DST) India through the
SERB project no. YSS/2015/001342. KH would also like to thank Prof.George Pharr for the technical clarification through
email. HNH was supported by the Engineering Research Center (ERC) program of the National Research Foundation of
Korea (NRF) grant funded by the Ministry of Science, ICT & Future Planning (MSIP) (NO. NRF-2015R1A5A1037627).
References
[1] Dotsenko, V. I., 1979. “Stress Relaxation in Crystals”. Physica Status Solidi (B), 11, pp. 11–43.
[2] Guiu, F., 1969. “On the measurement of internal stress by stress relaxation”. Scripta Metallurgica, 3, pp. 753–755.
[3] Bong, H. J., Barlat, F., and Lee, M.-G., 2016. “Probing Formability Improvement of Ultra-thin Ferritic Stainless Steel
Bipolar Plate of PEMFC in Non-conventional Forming Process”. Metallurgical and Materials Transactions A, 47(8),
pp. 4160–4174.
[4] Yamashita, H., and Ueno, H., 2013. “Enhancing deep drawability through strain dispersion using stress relaxation”.
AIP Conference Proceedings, 1567(1).
[5] Hariharan, K., Dubey, P., and Jain, J., 2016. “Time dependent ductility improvement of stainless steel SS 316 using
stress relaxation”. Materials Science and Engineering: A, 673, sep, pp. 250–256.
[6] Eipert, I., Sivaswamy, G., Bhattacharya, R., Amir, M., and Blackwell, P., 2014. “Improvement in Ductility in Commer-
cially Pure Titanium Alloys by Stress Relaxation at Room Temperature”. Key Engineering Materials, 611-612(April),
pp. 92–98.
[7] Hariharan, K., Majidi, O., Kim, C., Lee, M., and Barlat, F., 2013. “Stress relaxation and its effect on tensile deformation
of steels”. Materials & Design, 52, pp. 284–288.
[8] Majidi, O., Barlat, F., Korkolis, Y. P., Fu, J., and Lee, M. G., 2016. “Thermal effects on the enhanced ductility in
non-monotonic uniaxial tension of DP780 steel sheet”. Metals and Materials International, 22(6), pp. 968–973.
[9] Mohebbi, M. S., Akbarzadeh, A., Yoon, Y.-O., and Kim, S.-K., 2015. “Stress relaxation and flow behavior of ultrafine
grained AA 1050”. Mechanics of Materials, 89, oct, pp. 23–34.
[10] Xiao, L., and Bai, J., 1998. “Stress relaxation properties and microscopic deformation structure of H68 and QSn6.50.1
copper alloys at 353 K”. Materials Science and Engineering: A, 244, pp. 250–256.
[11] Varma, A., Gokhale, A. R., Jain, J., Hariharan, K., Cizek, P., and Barnett, M., 2018. “Investigation of stress relaxation
mechanisms for ductility improvement in SS316L”. Philosophical Magazine, 98, pp. 165–181.
[12] Choi, I.-C., Yoo, B.-G., Kim, Y.-J., and Jang, J.-i., 2011. “Indentation creep revisited”. Journal of Materials Research,
27(01), pp. 1–9.
[13] Lee, J.-A., Seo, B. B., Choi, I.-C., Seok, M.-Y., Zhao, Y., Jahed, Z., Ramamurty, U., Tsui, T. Y., and Jang, J.-i., 2016.
“Time-dependent nanoscale plasticity in nanocrystalline nickel rods and tubes”. Scripta Materialia, 112, pp. 79–82.
[14] Nautiyal, P., Jain, J., and Agarwal, A., 2016. “Influence of loading path and precipitates on indentation creep behavior
of wrought Mg6wt% Al1wt% Zn magnesium alloy”. Materials Science and Engineering: A, 650, pp. 183–189.
[15] Nautiyal, P., Jain, J., and Agarwal, A., 2015. “A comparative study of indentation induced creep in pure magnesium
and AZ61 alloy”. Materials Science and Engineering: A, 630, pp. 131–138.
[16] Xu, B., Yue, Z., and Chen, X., 2010. “Characterization of strain rate sensitivity and activation volume using the
indentation relaxation test”. Journal of Physics D: Applied Physics, 43(24), p. 245401.
[17] Gu, C. D., Lian, J. S., Jiang, Q., and Zheng, W. T., 2007. “Experimental and modelling investigations on strain rate
sensitivity of an electrodeposited 20 nm grain sized Ni”. J. Phys. D: Appl. Phys. J. Phys. D: Appl. Phys, 40, pp. 7440–
7446.
[18] Bower, A. F., Fleck, N. A., Needleman, A., and Ogbonna, N., 1993. “Indentation of a power law creeping solid”.
Proceedings of the Royal Society of London Series A Mathematical and Physical Sciences, 441(1911), pp. 97–124.
[19] Oliver, W., and Pharr, G., 2004. “Measurement of hardness and elastic modulus by instrumented indentation: Advances
in understanding and refinements to methodology”. Journal of Materials Research, 19(01), pp. 3–20.
[20] Alkorta, J., Martı́nez-Esnaola, J. M., and Gil Sevillano, J., 2008. “Critical examination of strain-rate sensitivity mea-
surement by nanoindentation methods: Application to severely deformed niobium”. Acta Materialia, 56(4), pp. 884–
893.
[21] Estrin, Y., 1998. “Dislocation theory based constitutive modelling: foundations and applications”. Journal of Materials
Processing Technology, 80-81, pp. 33–39.
[22] Su, C., Herbert, E. G., Sohn, S., LaManna, J. A., Oliver, W. C., and Pharr, G. M., 2013. “Measurement of power-
law creep parameters by instrumented indentation methods”. Journal of the Mechanics and Physics of Solids, 61(2),
pp. 517–536.
[23] Shi, Z., Feng, X., Huang, Y., Xiao, J., and Hwang, K. C., 2010. “The equivalent axisymmetric model for Berkovich
indenters in power-law hardening materials”. International Journal of Plasticity, 26(1), pp. 141–148.
[24] Caillard, D., and Martin, J., 2003. Thermally acticated mechanisms in crystal plasticity, Vol. 8. Pergamon.
[25] Prasad, K., Krishnaswamy, H., and Jain, J., 2018. “Leveraging transient mechanical effects during stress relaxation for
ductility improvement in aluminium AA 8011 alloy”. Journal of Materials Processing Technology, 255, pp. 1–7.
[26] Solberg, J., and Thon, H., 1985. “Stress relaxation and creep of some aluminium alloys”. Materials Science and
Engineering, 75, pp. 105–116.
[27] Yong, L., and Jingchuan, Z., 2008. “Effects of triple heat treatment on stress relaxation resistance of BT20 alloy”.
Mechanics of Materials, 40, pp. 792–795.
List of Figures
1 Change in penetration depth during relaxation shown along with ideal constant depth . . . . . . . . . . . . 3
2 Variation of penetration depth as a function of hold time. Best fit curve of the form h(t) = h0 + λ1 exp(λ2t).
[h0 = 1.98 µm, λ1 = 0.01, λ2 = −0.11] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
3 Experimental and corrected load vs time during relaxation . . . . . . . . . . . . . . . . . . . . . . . . . . 4
4 2D view of indent impression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
5 Line profiles on AFM image shown in (a) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
6 σ vs.t obtained across different length scales. The data corresponding to nano and micro indentation are
scaled using eq.6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

List of Tables
1 Material parameters across different length scales following eq.8 . . . . . . . . . . . . . . . . . . . . . . . 7

Potrebbero piacerti anche