You are on page 1of 7

Effect of large-strain deformation on surface remodeling characteristics in

porous materials

P. Ganguly
Schlumberger-Doll Research, Cambridge

F.-J. Ulm
Dept. of Civil and Environmental Eng., Massachusetts Institute of Technology, Cambridge

ABSTRACT: The paper develops constitutive relationships for deformation coupled surface growth/dissolution
(remodeling) in porous solids. Such phenomena have been observed in a wide range of natural and technological
processes, including corrosion in metallic materials and healing of bone and skin. In all these situations, the
surface remodeling takes place in a biochemical environment, and the combined mechanical and biochemical
effects need to be modeled for determining the remodeling characteristics. The present paper develops a consis-
tent model for exploring large-strain deformation coupled surface remodeling, and examines different conditions
that may hinder or encourage surface growth/dissolution in porous materials. It is observed that both tensile and
compressive deformation aid dissolution (or mass removal) at the material surface. Thus, concurrent deformation
can enhance the rate of corrosion in materials or cement leaching. On the other hand, suitable chemical envi-
ronments can induce surface growth. The model helps develop an understanding of the effects of these different
factors, and can aid the design of suitable environments (through alloying, changing solvent acidity/alkalinity,
surface treatments) for pre-determined surface characteristics in a variety of technological applications.

1 INTRODUCTION tissues. The load bearing capabilities of such materials


are profoundly influenced by the geometrical charac-
Surface modification (remodeling) of materials teristics of the solid struts (Gibson & Ashby 1997).
through mass addition/removal is technologically sig- Deformation-remodeling induced local thinning of the
nificant in a number of material applications. Gen- struts can cause catastrophic failure of porous materi-
erally, such remodeling occurs due to the action of als, with minimal changes in the macroscopic material
a chemical or biological agent, and is driven by densities.
the chemical potential. Examples include corrosion/ Modeling of deformation-coupled surface remodel-
oxidation of metals (Fontana 1986), cement leaching ing has been earlier undertaken by Ulm and co-workers
(Adenot & Buil 1992), or physiological processes like (Silva & Ulm 2002, Ulm et al. 2003) in the small-
wound healing in skin and bone (Blair 1998). In most strain regime. On the other hand, elastic large-strain
of these processes, experimental evidence suggests deformations may govern the remodeling process in
phenomenological connections between imposed soft materials, e.g. in healing of the skin wounds or
deformation (stress/strain) and the remodeling char- hypertrophy of the heart (i.e. dilation of overloaded
acteristics. Imposed stresses have been observed to heart muscles, Lund & Tomanek 1978). Hoger and
enhance corrosion rates in metals (Liu et al. 2004) and co-workers (Chen & Hoger 2000, Lubarda & Hoger
the influence of contractile forces generated during 2002) have modeled deformation-coupled growth in
skin wound healing on scar tissue formation has been biomaterials through a multiplicative split of the total
recorded (Yannas et al. 1989). Analysis of the coupling deformation gradient into the growth and deforma-
between deformation and the remodeling rates is thus tion components. This approach is similar to the
significant. multiplicative split in elasto-plastic materials (Lee
The deformation-remodeling coupling assumes 1969) and assumes a volumetric (or global-scale)
even greater significance for porous materials, i.e. growth/removal, rather than a surface (or local-scale)
materials with a continuous solid framework, com- remodeling.
posed of struts, enclosing a large volume fraction of The present paper develops a consistent frame-
pores. Many natural and technological materials are work for constitutive relationships defining large-
porous, including cement, wood and the bone and skin strain deformation coupled surface growth in porous

121
Copyright 2005 Taylor & Francis Group plc, London, UK
At any given time t, the mass of the solid is defined by
ms (t) and is given by
solid
strut

where s (t) and vs (t) are the density and the volume of
the solid strut, respectively. The rate of mass change
pore
can be determined by differentiating Equation 1 w.r.t. t,
space
(a) and applying the generalized divergence theorem (Ulm
and Coussy 2003). The rate of mass change would
include two rates, (a) due to deformation, and (b) due
to growth. The rate of mass change due to deformation
is given by
solvent

solid strut
(b) where ud is the deformation-induced velocity of the
solid particles. Assuming that the deformation is mass-
Figure 1. The deformation-remodeling phenomena in
porous materials at (a) macroscale, and (b) microscale. At
conserving, the rate of mass change given by Equation
the macroscale, the porous structure is composed of solid 2 is zero. Thus mass change, determined by the
struts and the solvent filled pore space. The porous represen- growth/dissolution, is defined by the growth-induced
tative volume element is subjected to traction, as shown by the volume change and is given by
block arrows in (a). The macroscopic traction is transferred
to the microscopic element (shown by the dashed rectangle
in (a)) through force balance and strain compatibility in the
structure. The microscopic element (shown in (b)) is com-
posed of a single solid strut exposed to chemical action at the
solid-solvent interface. where ug is the growth velocity of the solid-solvent
front (positive for growth and negative for dissolution),
n is the outward normal to the front and (t) is the
area of the solid-solvent surface that is subjected to
materials. The material is assumed to be elastic, i.e. remodeling (i.e. ug nd(t) is the infinitesimal change
only geometric non-linearity in included in the model. in volume due to remodeling).
The porous material is treated as multiscale, and In the above equations, all the variables are
the deformation/remodeling phenomena is modeled expressed in the current reference frame (x, t). In the
at two relevant length scales (the macroscopic and undeformed reference frame (X, t0 ), the rate of mass
the microscopic, see Figure 1). Physically, the porous change can be defined as
microstructure includes individual solid struts, while
the macrostructure encompasses the representative
porous material. The mass loss/gain due to surface
remodeling is assumed to occur at the microscale, and
the effect at the macroscale is determined through con- g
where u0 and N are the growth velocity and the outward
sistent upscaling. Such a model description defines normal, described in the undeformed reference frame.
most surface remodeling phenomena in elastic porous The volume conversion between the undeformed and
materials, and may have wide applications in structural the deformed configurations is given by
and bioengineering designs.

where j is the determinant of the deformation gradient


2 MICROSCALE MODEL tensor (f ). The corresponding surface area conversion
is defined by
The representative volume element at the microscale
is shown in Figure 1b, and includes a deforming solid
strut and the solvent. Part of the solid boundary is
also exposed to chemical action through exposure to where (f 1 )t is the inverse-transpose of the micro-
a solvent that deposits/ removes mass from the solid. scopic deformation gradient tensor (f ). Since the rate

122
Copyright 2005 Taylor & Francis Group plc, London, UK
of mass change is invariant with change in reference where d is the Eulerian strain rate tensor, and p is the
configuration, Equations 34 can be equated to yield solvent pressure at the solid-solvent interface. The first
(using Eqs. 56) integral in Equation 11 is deformation work, while the
second part evaluates the work done as the solid front
expands (or contracts) against (or aided by) the fluid
pressure. Thus, the first term is mechanical in origin,
Equation 7 simply states that the solid-solvent front while the second term defines the mechano-chemical
velocity in the current configuration would need to be coupling. Note that the partial pressure of the solute
scaled by the deformation gradient (which relates the in solution has been neglected in this formulation.
unit vectors, x and X, in the current and undeformed The change in free energy is due to free energy
configurations) to yield the velocity measurement in change in the bulk as well as due to the chemi-
the undeformed configuration. cal change (which results in mass addition/removal)
Volume change in the solid can be due to defor- occurring at the solid-solvent surface.
mation as well as mass change, and the rate of
change in volume (in the current configuration) can be
expressed as
where  is the difference in the chemical poten-
tial of the solid atoms (i s , where i and s
are the potentials of the atoms in solution and in
the solid, respectively), and m s is the rate of solid
where the first integral represents the volume change mass change (per unit interface area, m s = s (t)ug n).
due to deformation and the second integral that due to Thus, the first term in Equation 12 defines the free
growth/dissolution. The same volume change can be energy change in the bulk (independent of the remod-
expressed in terms of quantities measured in the unde- eling process), while the second term denotes the free
formed configuration, through suitable transformation energy change due to surface growth/dissolution (i.e.
of Equation 8 using Equations 57. the change in potential of ion/atoms that move between
solid and solvent).
The rate of dissipation can be determined from
Equations 1012, and would include the volume and
the surface effects. The volume phenomenon is given
Note that while the deformation and growth/disso- by Dv
lution contributions can be clearly demarcated in the
current configuration, the formulation is more mixed
in the undeformed configuration as the volume change
due to growth has to account for the deformation
induced density change. while the surface phenomenon can be defined by Ds

2.1 Thermodynamic considerations


The constitutive relationships for deformation coupled The dissipation rates in the undeformed reference
growth/dissolution in the solid strut can be derived by frame can be determined from Equations 57, 13.
application of the Claussius-Duhem inequality which
states that for any process, the rate of change of
(Helmholtz) free energy (
) is equal to or less than
the external work (Pext ) provided to the system, thus
where f is the deformation gradient rate and is
yielding a non-negative rate of dissipation D.
Boussinesq tensor (which is the energy compliment to
the deformation gradient rate). The similarity between
the second terms in Equations 13, 15 is evident, and is
a result of mass conservation in the bulk, as the defor-
The external work includes work done in deforming mation coupled surface remodeling progresses. The
the solid strut, as well as the work done (or released) surface dissipation rate in the undeformed reference
due to the movement of the solid-solvent interface. frame can be evaluated from Equation 14.

123
Copyright 2005 Taylor & Francis Group plc, London, UK
g
where m s0 = s (t0 )u0 N is the rate of mass gain/loss where Jm is the molar flux
per unit interface area in the undeformed frame.
The total dissipation rate, D, is a sum of the volume
and surface dissipations given in Equations 1316.

The effects of the process variables on the surface


remodeling characteristics can be estimated from
The volume integrals given in Equations 13, 15 are Equations 1921. Equation 20 indicates that the signs
standard integrals seen in elasticity theory, and indi- of the driving force (A) and the flux (Jm ) must be
cate that the energy supplied to the solid bulk is stored similar: growth (positive Jm ) can take place when the
as stressstrain energy or dissipated as heat. The vol- driving force is positive, while dissolution (negative
ume dissipations are non-negative and independent of Jm ) will result from a negative driving force. Thus,
the surface phenomena. Given that the overall dissi- from Equation 19, deformation and fluid pressure
pation is non-negative, the surface integrals given in would encourage dissolution at the solid surface, while
Equations 14, 16 must be non-negative. growth would be encouraged if the ions are at a lower
chemical potential at the solid surface than in the fluid.
The effect of change in driving force on the mass
flux need to be determined from a kinetic equation,
thus requiring simplifying assumptions on the rate
limiting step. Assuming that the remodeling kinetics
Eq. (18) indicates the relationship between mass addi- is governed by the diffusion of the chemical species
tion and the mechano-chemical conditions of the between the solid and the solution, a linear equation
surroundings. Following standard thermodynamic relating the driving force and the flux (in the form of
arguments, (i s p/s (t)) can be identified as Ficks Law) may be appropriate. In terms of measur-
the thermodynamic force causing the mass change able quantities, such an equation would link the driving
(dms /dt). The free energy in the solid (s ) is a sum force and the solid-liquid front speed (ug , ug = ug n,
of the elastic free energy (el ) and the chemical free given that ug is linked to Jm by Eqs. 3, 21), i.e.
energy (gs , defined per unit mass). The free energy of
the ion (i ) is primarily composed of the chemical free
energy (gi ) of the ions in solution (Atkins 1994). The
thermodynamic driving force (in terms of per mole) The constant k would be determined by the mechanism
can be represented by the chemical affinity (Coussy of diffusion and would be independent of the physi-
1995) of the deformation coupled remodeling process. cal quantities defining the driving force, i.e. pressure,
deformation and chemical potentials.
Equations 1922 indicate the connection between
the remodeling kinetics and the driving force. Defor-
mation resulting in elastic free energy would encour-
age dissolution at the surface. According to Equations
where A is the chemical affinity and Mm is the molec- 1922 higher deformation strains would increase the
ular weight of the species involved in mass addition/ rate of dissolution. Thus, in deformation coupled sur-
dissolution at surface. The driving force (A) for face remodeling, surface dissolution would be higher
remodeling can be decomposed into chemical and in regions with maximum deformation, which may be
mechanical effects. The first term in Equation 19 is regions of concentration of stress/strain and damage.
purely chemical, while the second term is primarily of This synergistic linkage between deformation and
mechanical origin. The chemical contribution to the mass-loss (i.e. the weaker regions are made progres-
driving force (gi gs ) arises from the difference in the sively weaker due to deformation-dissolution cou-
Gibbs free energy between an atom (or ion) in solution, pling) may cause localized failures, and is the cause for
and that same atom (or ion) in the solid structure. The concern in stress-corrosion cracking, cement leaching
mechanical driving force arises from the deformation and osteoporosis. In such situations, as seen in Equa-
of the ions (or atoms) at the surface and the resultant tions 1922, the dissolution rate of the solid can be
change in their potentials, and the work done due to lowered by changing the chemical potential of the ions
the expanding (or contracting) solid against the fluid in solution (e.g. by changing pH) or by lowering the
pressure. fluid pressure.
Equation 18 can be rewritten in the standard ther- The field and flux quantities in Equations 1822
modynamic format have been defined in the current configuration. The
corresponding quantities in the undeformed configu-
ration can be evaluated using the appropriate volume

124
Copyright 2005 Taylor & Francis Group plc, London, UK
and surface area conversion relationships (Eqs. 56). primarily signifies the contribution of the chemical
The driving force (in the undeformed configuration) processes (mass addition/removal) to the porosity
would be given by change.
The work done on the porous solid would include
the work done by the externally imposed tractions, as
well as the work done by the fluid pressure (on the
change in porosity). In terms of quantities measured
in the undeformed reference frame, the total external
as the chemical and elastic free energies (measured work would be
in terms of per unit mass) are invariant with change
in configuration, and s (t0 ) = js (t). The mass flux Jm
would be defined by Equation 21, although the appro-
priate formulation for rate of mass change (Eq. 4
instead of Eq. 3) needs to be considered. Finally,
the kinetic equation relating the driving force and where F and " are the macroscopic deformation gra-
the solid-liquid front motion can be rewritten in the dient rate and Boussinesq tensor, respectively. The
undeformed configuration, free energy can be evaluated by upscaling the micro-
scopic free energy change (Eq. 12 expressed in the
undeformed configuration).

g g
where u0 = u0 N, and the constants of proportionali-
ties (k and k0 ) can be related using Equations 57.

where sv is the solid free energy per unit volume


(i.e. sv = s (t0 )s ). As in Section 2.1, the dissipation
can be divided into volume and surface components.
3 MACROSCALE MODEL In this case (i.e. macroscopic analysis), the volume
phenomena include physical processes that occur in
The formulations derived above consider the defor- a smeared or distributed manner, and can be deter-
mation-remodeling occurring at a single strut in a mined from macroscopic or volume-averaged mea-
porous solid. In order to estimate the effect of this surements using standard poromechanics theory. The
phenomenon on the material characteristics at the surface phenomena involve the growth/dissolution
macroscale (porous material scale), the equations need processes that need to be modeled in a local man-
to be consistently upscaled. The representative vol- ner. The volume dissipation Dv will be given by (from
ume element at the macroscale (total volume V) is Eqs. 2729)
composed of a solid of volume Vs (t), and pore-space
volume of Vf (t). The rate of change of Lagrangian
porosity ( = Vf (t)/V(t0 )) can be defined as

where is the porosity change due to mechanical


causes only, and is given by the first two terms in Equa-
Using Equations 59, the rate of change of porosity tion 27. The volume dissipation is due to mechanical
can be derived from the microscopic measurements. forces, and indicates that a part of the external work
done on the porous material is stored as elastic energy
of the solid structure, while another part is expended
in porosity change (against or for the fluid pressure),
assuming that the system is under equilibrium.
The dissipation at the surface involves the driving
where J is the determinant of the macroscopic defor- force for the surface remodeling and the mass-flux for
mation gradient, cs0 indicate the original solid volume the remodeling process. From Equations 2730, the
fraction, and <> indicate averaging over the appro- surface dissipation can be written as
priate volume. The first two terms in Equation 27 are
routine in poromechanics, and define the change in
porosity due to differences in macroscopic and micro-
scopic deformation (Coussy 1995). The third term

125
Copyright 2005 Taylor & Francis Group plc, London, UK
1000 for bone remodeling can be estimated to be approx.
500 from Silva & Ulm (2002). Figure 2 (evalu-
Relative dissolution rate (u0g/u0g0)

ated from Eq. 33) shows that imposed deformation


= -1000
(tensile/compressive) accelerates the dissolution
100 kinetics of the strut, and the effect of deformation
increases with increasing value. However, the rel-
= -100 ative impact of different deformation modes on the
dissolution kinetics (e.g. change per unit strain in ten-
10 sion vs. compression) depend on the chosen reference
= -10 frame, as is evident from replacing the Lagrangian
strain measure (stretch, 1 ) in Equation 33 with the
= -1 Eulerian measure, true strain ( = ln (1 )).
1
0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25
Stretch in loading direction
5 CONCLUSION
Figure 2. Effect of deformation (tension/compression) on
the dissolution kinetics for different ( = EMm /s (t0 )A0 ).
A consistent model for analysis of deformation cou-
The stretch measure is computed as the ratio of the final
length and the initial length in the loading direction. The pled surface remodeling in porous materials has been
range of stretch values in this figure includes uniaxial defor- developed in the present work. The phenomenon is ini-
mations with final lengths ranging from half to twice the tially analyzed at the microscopic level i.e. the effect of
initial lengths. deformation on surface growth/dissolution in a single
strut. It is evident that deformation (both compressive
and tensile) would encourage dissolution (and mass
4 APPLICATION removal) at the solid surface. This can accelerate thin-
ning of struts in regions of maximum deformation
The application of the constitutive equations devel- or damage and can cause premature, localized fail-
oped above to the deformation/dissolution of a cylin- ure of the porous material. The analysis was extended
drical strut is explored in this section. Neglecting the to the macroscopic scale through consistent upscaling
effect of deformation on the chemical free energies and of the microscale formulation, and the factors defining
the fluid pressure, the deformation induced change in the increase in porosity of the material was analyzed.
the driving force for dissolution can be determined
from Equation 23.
REFERENCES

Adenot, F & Buil, M. 1992. Modelling of the corrosion of


0
the cement paste by deionized water. Cement Concr Res.
where A and A are the driving forces for the unde- 22: 489496.
formed and the deforming strut. Assuming that the Atkins, P.W. 1994. Physical chemistry. New York: W.H.
elastic energy of the strut is due to uniaxial deforma- Freeman and Company.
tion of an isotropic, Hookean material (Attard 2003) Attard, M.M. 2003. Finite strain beam theory. Int. J. Solids
and using Equation 24 to relate the driving force and and Struct. 40: 45634584.
the dissolution velocities, the velocity change due to Blair, H.C. 1998. How the osteoclast degrades bone. Bio
Essays. 20: 837846.
dissolution can be defined from Equation 32.
Chen, Y.-C. & Hoger, A. 2000. Constitutive functions
of elastic materials in finite growth and deformation.
J. Elasticity. 59: 175193.
Coussy, O. 1995. Mechanics of porous continua. Chichester
(UK): John Wiley & Sons.
Fontana, M.G. (3rd ed.) 1986. Corrosion Engineering. New
g g0
where u0 and u0 are the dissolution velocities for the York: McGraw Hill Pub.
undeformed and the deformed material (measured in Gibson, L.J. & Ashby, M. F. (2nd ed.) 1997. Cellular Solids.
the undeformed reference frame), and 1 is the stretch Cambridge (UK): Cambridge Univ. Press.
(=final length/initial length) in the loading direction. Lee, E.H. 1969. Elasto-plastic deformation at finite strains.
J. Appl. Mech. 36: 16.
The ratio of the elastic modulus and the chemical driv- Liu, X, Frankel, G.S., Zoofan, B. & Rokhlin, S.I. 2004.
ing force ( = EMm /s (t0 )A0 ) yields the significance Effect of applied tensile stress on intergranular corrosion
of the different effects (mechanical vs. chemical) on of AA2024-T3. Corrosion Science. 46: 405425.
the dissolution kinetics. The value of is unavailable Lubarda, V.A. & Hoger, A. 2002. On the mechanics of solids
for remodeling in soft materials. However, the value with a growing mass. Int. J.Solids Struct. 39: 46274664.

126
Copyright 2005 Taylor & Francis Group plc, London, UK
Lund, D. & Tomanek, R.J. 1978. Myocardial morphology Ulm, F.J., Lemarchand, E. & Heukamp, F.H. 2003. Ele-
in spontaneously hypertensive and aortic-constricted rats. ments of chemomechanics of calcium leaching of cement-
Amer. J. Anatomy. 152: 141151. based materials at different scales. Eng. Fract. Mech. 70:
Silva, E.C. & Ulm, F.-J. 2002. A bio-chemo-mechanics 871889.
approach to bone remodeling and fracture. In: Karihaloo, Yannas, I.V., Lee, E., Orgill, D.P., Skrabut, E.M. &
B. L. (ed.), Proc. IUTAM Symp. on Analytical and Com- Murphy, G.F. 1989. Synthesis and characterization of a
putational Fracture Mechanics of Non-Homogeneous model extracellular matrix that induces partial regenera-
Materials. London: Kluwer Acad. Pub. 355366. tion of adult mammalian skin. Proc. Natl. Acad. Sci. USA.
Ulm, F.J. & Coussy, O. 2003. Mechanics and Durability of 86: 933937.
Solids (vol. 1). Upper Saddle River (NJ): Prentice Hall.

127
Copyright 2005 Taylor & Francis Group plc, London, UK

You might also like