Tetsuya Matsunaga

Abstract The near-α Ti–6Al–4Zr–4Nb (wt pct) alloy is a recently developed alloy with potential for aerospace applications. This study evaluates the microstructure and mechanical properties of Ti–6Al–4Zr–4Nb produced by spark plasma sintering (SPS). The SPS samples sintered in the α + β regions exhibited equiaxed α with a neighboring thin β phase. Above the β-transus temperature, the α/β lamellar structure formed, allowing control of the grain size (100 ~ 200 μm). The compressive strength and the creep property of the SPS samples were compared with the LBPDed and the forged samples. The compressive strength of the SPS sample was lower than that of the LPBFed sample but similar to the forged sample. The SPS samples exhibited longer creep rupture life (2220 hours) than LPBFed samples (1730 hours), but shorter than the forged sample (4109 hours). The creep deformation mechanism of the lamellar structure in the SPS sample was dislocation creep.

Powder bed fusion using a laser beam (PBF-LB) was performed for Ti6Al4Nb4Zr (mass%) developed by our group to improve the oxidation resistance at temperatures greater than 600°C by adding Nb and Zr to near-α alloys. Microstructure evolution of the PBF-LB samples by heat treatment was investigated, especially for heat treatment duration in the α + β phase, cooling rate, and heat treatment in the β phase. The equiaxed α phase formed during heat treatment along the melting-pool boundaries. The high volume fraction of the α phase and high Nb contents in the β phase was obtained by slow cooling (furnace cooling) compared with fast cooling (air cooling). The α/β lamellar structure formed in the melting pool boundaries with 100μm in size and no equiaxed α phase formed along the boundaries by heat treatment in the β phase regime. Creep life at 600°C and 137 MPa was similar for the air-cooled and furnace-cooled samples, but the slightly slower deformation was obtained in the furnace-cooled sample. Creep life of the sample in the β phase region drastically increased due to the absence of the equiaxed α phase. Dominant deformation mechanism of creep was grain boundary sliding. The small equiaxed α phase accelerated grain boundary sliding.

Heat-resistant Ti-Al-Nb-Zr alloys, which don’t contain Sn, have been designed to obtain good oxidation resistance above 600 °C. In addition, to design Ti alloys with best balance of creep and fatigue properties, prior β grain size which affects fatigue properties and lamellar microstructure which affects creep properties were controlled by heat treatment. In the present study, the effect of microstructure on creep properties of one of the alloys, i.e., Ti-7.5Al-4Nb-4Zr alloy, with the bimodal (B), the lamellar structures in small prior β grains (L_S), and the lamellar in large prior β grains (L_L) were investigated at 600 °C. The creep deformation mechanism for each microstructure was a power-law creep. However, the creep life varied depending on the microstructures. The longest creep life was obtained in L_S with prior β grain size of 90 μm and interlamellar spacing of approximately 10 μm, while the shortest creep life was obtained in L_L with prior β grain size of 550 μm and fine interlamellar spacing of less than 2~3 μm. This suggests that creep life is more affected by interlamellar spacing than by prior β grain size.

A microstructure evolution based on the processing and heat-treatment conditions was investigated for Ti-13Al-2Nb-2Zr (at%) alloy, which has a promising oxidation resistance. Three processing temperatures, 900 degrees C and 1000 degrees C in the alpha+beta phase field, and 1080 degrees C in the beta phase field, and two rolling reduction ratios, 93% and 67%, were selected as the processing conditions. In the samples processed and heat-treated in the alpha+beta phase field, an almost fully equiaxed structure, i.e., the equiaxed or ellipsoid alpha phase surrounded by the beta phase, was formed through furnace cooling, and a bi-modal structure was formed using air cooling. The morphology of the alpha phase in the near fully equiaxed and lamellar structure depends on the rolling reduction ratio; in other words, the equiaxed and ellipsoid alpha phases are formed at rolling reduction ratios of 93% and 67%, respectively. The volume fraction of the equiaxed alpha phase in the bi-modal structure is processed at 900 degrees C, which is higher than that of the bi-modal structure processed at 1000 degrees C despite the same heat-treatment temperature applied. This is because the induced strain when processed at 1000 degrees C is smaller than that when processed at 900 degrees C. By contrast, in the samples processed in the beta phase field and heat-treated in either the alpha+beta or beta phase field, a lamellar structure is formed. The creep behavior of the bi-modal structure obtained upon processing at 900 degrees C and 1000 degrees C for up to a 93% rolling reduction ratio was investigated. The creep life of the sample processed at 1000 degrees C was two-times longer than the sample processed at 900 degrees C. This is because a smaller volume fraction of the equiaxed alpha phase in the sample processed at 1000 degrees C than that of the sample processed at 900 degrees C.

The effect of mixed partial occupation of metal sites on the phase stability of the γ-Cr23-x Fe x C6 (x = 0-3) carbides is explored as function of composition and temperature. Ab initio calculations combined with statistical thermodynamics approaches reveal that the site occupation of the carbides may be incorrectly predicted when only the commonly used approach of full sublattice occupation is considered. We found that the γ-M23C6 structure can be understood as a familiar sodium chloride structure with positively charged rhombic dodecahedron (M(4a) M12 (48h)) and negatively charged cubo-octahedron (M8 (32f) C6 (24e)) super-ion clusters, together with interstitial metal atoms at the 8c sites. The stability of the partially occupied phase can be easily rationalized on the basis of a super-ion analysis of the carbide phase. This new understanding of γ-M23C6 carbides may facilitate further development of high-chromium heat-resistant steels.

The advanced ultra-supercritical (A-USC) power generation system is expected to become the next-generation base-load power station in Japan. Dissimilar weld joints between high-Cr heat-resistant steels and nickel-based alloys with a nickel-based filler metal (Alloy 82) will need to be adopted for this purpose. However, interfacial failure between the steels and weld metal has been observed under high-temperature creep conditions. Fractography and microstructure observations showed the failure initiated in a brittle manner by an oxide notch at the bottom of the U-groove. The fracture then proceeded along the bond line in a ductile manner with shallow dimples, where micro-Vickers hardness tests showed remarkable softening in the steel next to the bond line. In addition, the steel showed a much larger total elongation and reduction of area than the weld metal at low stresses under long-term creep conditions, leading to mismatch deformation at the interface. According to the results, it can be concluded that the interfacial failure between the 9Cr steels and Alloy 82 weld metal is initiated by an oxide notch and promoted by softening and the difference in the plasticity of the steels and weld metal.

In advanced ultra-super critical (A-USC) power generation system, the steam condition becomes severe for conventional structural materials, i.e., high chromium heat resistant steels, because the steam temperature and pressure are 973 K and 35 MPa, respectively. It means that dissimilar welding between nickel base alloys and the heat resistant steels is inevitable. However, because the dissimilar welding engenders interface fracture at high-temperature creep conditions, understanding a relation between creep behavior and ductility of the weld metal as well as the base metals is important to ascertain safety of the plant. Alloy 82 weld metal decreased ductility with increasing creep lifetime because of Ni3Nb in the matrix and two kinds of GB precipitates: M23C6, with Cr, Ni, and Fe MX with Ni, Nb, Cr and Mn. On the other hand, high B-9Cr steel showed high ductility even after long term creep test. Therefore, the difference of mechanical property leads to incompatibility of strain at the interface between steel and weld metal, resulting in interfacial crack in dissimilar welding condition.

Mod. 9Cr-1Mo steel (ASME Gr.91 steel) is widely used as a main structural material for boiler components in ultra-supercritical (USC) thermal power plants at about 600 degrees C. Decrease of the creep strength of welded joints due to Type-IV failure is a critical issue, however creep rupture data longer than 10,000 h are scarcely obtained. The present paper conducted long-term creep tests of welded joints from 30,000 to about 70,000 h at 600 and 650 degrees C. For the specimens fractured after about 35,000 h at 600 and 650 degrees C, creep damage was mainly observed in the fine-grained heat-affected zone (HAZ) (Type-IV), however damage was also observed in the weld metal. The welded joints crept at 600 degrees C and 50 MPa for about 50,000 and 70,000 h fractured in the weld metal. It was found that the hardness of 9Cr weld metal decreased after about 30,000 h creep at 600 degrees C due to the recovery of microstructures. Therefore it is considered to be necessary to monitor creep damage not only in the HAZ but also in the weld metal.

To identify the creep behavior of Zircaloy-4 during operating time of nuclear power plants, creep tests were performed at 673 K. The alloy showed an abnormal creep behavior, i.e., two-stage steady state creep, at low stresses of less than 66 MPa. The first steady state was observed at strain of 0.05, whereas the second one appeared at that of 0.15 at 55 MPa. Moreover, the first steady state creep rate was faster than the second one. It resulted in the changing of the stress exponent from 4.9 to 14. Microscopy after the creep tests revealed that dislocation structures at each stage were totally different. In the first stage, EBSD results showed that strain distributed uniformly in grain interiors, where dislocations moved individually. In the second stage, the interaction among dislocations became stronger with increasing strain because cell structure was generated. Therefore, the creep was suppressed and the creep rate decreased in the second stage. It concluded that the changing of dislocation structure during creep led to two-stage steady state creep behavior in Zircaloy-4.

Small amounts of boron improve the mechanical properties in high-chromium ferritic heat-resistant steels. In this work, the stabilizing mechanism by boron in body-centered cubic iron (bcc Fe) through (Fe,Cr)(23)(C,B)(6) precipitates was investigated by first-principles calculations. Formation energy analysis of (Fe,Cr)(23)(C,B)(6) reveals that the compounds become more stable to elemental solids as the boron concentration increases. Furthermore, the interface energy of bcc Fe(110) parallel to Fe-23(C,B)(6)(111) also decreases with boron concentration in the compounds. The decreased interface energy caused by boron addition is explained by the balance between the change in the phase stability of the precipitates and the change in the misfit parameter for the bcc Fe matrix and the precipitates. These results show that boron stabilizes the microstructure of heat-resistant steels, which is important for understanding the origins of the creep strength in ferritic steels. (C) The Minerals, Metals & Materials Society and ASM International 2016

Prompt phase transformations make grains in the heat-affected zone (HAZ) smaller during welding of 9% chromium (9Cr) heat-resistant steels leading to premature failure under creep conditions, which is well known as a type IV fracture. Because the type IV fracture shortens the creep lifetime of the steels, suppressing the fracture is an urgent task in the energy industry. The present study shows that boron addition and nitrogen reduction inhibit grain refinement after welding because of a change in the morphology of the precipitate at prior austenite grain boundaries. In conventional 9Cr steel (ASME Gr. 92 steel), a high amount of MX was unable to pin interface migration of the phase transformation and generated fine grains in the HAZ. In the new B-added steels, B-stabilized M23C6 became the dominant precipitate and showed a larger pinning effect of the phase transformation than MX, which resulted in coarse grains in the HAZ. This suggests that designing stabilized M23C6 forms a superior welded microstructure and results in a longer creep lifetime of 9Cr steels. (C) 2015 Elsevier B.V. All rights reserved.

Mo-modified Zr-1.2Nb alloys were prepared and the effects of Mo addition on precipitation in Zr-Nb alloys were elucidated. It was found that an increase of Mo content decreased alpha ->(alpha+beta) phase transformation temperature in Zr-Nb alloys. A new kind of precipitate, beta-(Nb, Mo, Zr), was identified in the Mo-containing alloys. These particles are formed because Mo atoms replace Nb atoms in the beta-(Nb, Zr) phase. Further, it was found that Mo retarded the growth of the precipitates but increased their number density. (C) 2015 Elsevier B.V. All rights reserved.

At room temperature, the strain-rate sensitivity (SRS) was almost negligible at high strain rates in as-extruded Mg-3Al-1Zn alloy because of the exponent (m) of 0.008. However, SRS became remarkable with decreasing strain rate down to 10(-8) s(-1), where the m value increases up to 0.06. (C) 2015 Elsevier B.V. All rights reserved.

For self-interstitial atom (SIA) clusters in various concentrated alloys, one-dimensional (1D) migration is induced by electron irradiation around 300K. But at elevated temperatures, the 1D migration frequency decreases to less than one-tenth of that around 300K in iron-based bcc alloys. In this study, we examined mechanisms of 1D migration at elevated temperatures using in situ observation of SUS316L and its model alloys with high-voltage electron microscopy. First, for elevated temperatures, we examined the effects of annealing and short-term electron irradiation of SIA clusters on their subsequent 1D migration. In annealed SUS316L, 1D migration was suppressed and then recovered by prolonged irradiation at 300K. In high-purity model alloy Fe-18Cr-13Ni, annealing or irradiation had no effect. Addition of carbon or oxygen to the model alloy suppressed 1D migration after annealing. Manganese and silicon did not suppress 1D migration after annealing but after short-term electron irradiation. The suppression was attributable to the pinning of SIA clusters by segregated solute elements, and the recovery was to the dissolution of the segregation by interatomic mixing under electron irradiation. Next, we examined 1D migration of SIA clusters in SUS316L under continuous electron irradiation at elevated temperatures. The 1D migration frequency at 673K was proportional to the irradiation intensity. It was as high as half of that at 300K. We proposed that 1D migration is controlled by the competition of two effects: induction of 1D migration by interatomic mixing and suppression by solute segregation.

In this work, thermal aging experiments on 9Cr-ODS steel were carried out at 973 K from 100 h up to 10,000 h, and grain structural evolution was investigated mainly focusing on analyses by electron backscatter diffraction (EBSD). The EBSD maps revealed no significant morphological changes in the matrix grain structure under the long-term aging. Grain size increased slightly with increasing periods of aging time from 100 h to 10,000 h. Distribution of misorientation angles was not affected by the thermal aging. The fraction of low misorientation angles comprised of subgrains was about 40% before aging, and decreased slightly by aging time up to 10,000 h. Total grain boundary length and density per unit area did not change significantly by aging up to 5000 h. They decreased slightly after aging for 10,000 h, which may be caused by slight recovery of microstructure. However, the effects of slight changes in the grain structure on mechanical properties are expected to be negligible. (C) 2014 Elsevier B.V. All rights reserved.

Zr-Si alloys were designed to contain eutectic phase surrounding the parent phase to suppress creep behavior of claddings. Creep tests conducted at 294-573 K showed that creep behavior was inhibited and that the creep failure time of new Zr alloy became longer than that of a conventional alloy: Zircaloy-4. Results show that the eutectic phase can suppress creep at operating temperatures prevailing in current nuclear power plants. (C) 2013 Elsevier B.V. All rights reserved.

Creep tests were performed at less than 0.4 Tm (Tm is the melting temperature) for 99.999%, 99.57%, and 99.52% aluminum with several grain sizes in the range of 50-330 μm. These Al materials show remarkable creep behavior with an apparent activation energy (Q) of 30 kJ/mol, a stress exponent of 4, and a grain-size exponent of zero, and with a larger creep rate with increasing purity. These parameters resemble those of conventional dislocation creep, which is rate-controlled by the usual diffusion processes, except for the extra-low Q value. This means that a non-diffusional process affects the steady state deformation in this temperature region. Transmission electron microscopy revealed the development of a cell structure in the steady state and dislocations without any tangles in the cell interiors. Therefore, because the rate-controlling process could not occur inside of the cells, dislocation annihilation occurred through cross slip around the cell walls. According to these creep parameters and microstructural observations, the observed creep region is suggested to be a new creep region occurring through a non-diffusional process within the existing deformation mechanism map of Al at less than 0.4 Tm. © 2014 The Japan Institute of Light Metals.

We propose a mechanism for glide motion, i.e. one-dimensional (1D) migration, of interstitial clusters in concentrated alloys driven by high-energy particle irradiation. Interstitial clusters are fundamentally mobile on their respective 1D migration tracks, but in concentrated random alloys they are stationary at the position where the fluctuating formation energy achieves a local minimum. Irradiation changes the microscopic distribution of solute atoms through atomic displacement and recovery of the produced Frenkel pairs, which causes cluster 1D migration into a new stable position. In molecular dynamics simulations of interstitial clusters up to 217i in Fe-Cu alloys, stepwise 1D migration was observed under interatomic mixing or shrinkage of the cluster: a single 1D migration was induced by two exchanges per atom or cluster radius change by two interatomic distances. The 1D migration distance ranged up to several nanometers. We compared the frequency and distance of 1D migration with those for in situ observation using high-voltage electron microscopy, allowing for the extremely large rate of interatomic mixing and cluster shrinkage in the present simulation.

To examine effects of the grain boundary (GB) and dislocation on the deformation mechanism for ultrafine-grained (UFG) and coarse-grained (CG) interstitial-free (IF) steels at room temperature, tensile tests and several types of microscopy were conducted for each steel. Atomic force microscopy revealed that the contribution of grain-boundary sliding (GBS) on deformation increased more prominently in UFG region than in CG region. Moreover, transmission electron microscopy revealed that dislocation motion was dominant in CG steel, where cell structure was formed with increasing strain. On the other hand, although dislocations moved in UFG steel, they did not tangle and piled up at GB, where interaction between GB and dislocation occurred markedly, causing significant GBS. Therefore, the dominant deformation mechanism changed from dislocation motion to GBS by decreasing grain size in IF steel.

Fcc pure metals were irradiated with spallation neutrons (energies up to 500 MeV) at room temperature to a neutron fluence of 1 × 1018 n m-2 at KENS, High Energy Accelerator Research Organization (KEK). Defect clusters induced by large collision cascades were examined using transmission electron microscopy (TEM). In Au, large groups of defects included more than 10 clusters, and the damage zone extended over 50 nm, which was larger than that induced by fusion neutron irradiation (&lt 20 nm). Although small stacking fault tetrahedra (SFT) are formed in subcascades by fission and fusion neutron irradiation, dislocation loops were also observed in the present experiments. Large dislocation loops (&gt 10 nm) were identified as vacancy type by the conventional inside-outside contrast method. Because of the low neutron fluence, spatial overlapping of collision cascades was ignored. Large vacancy loops are formed through cooperative reactions among subcascades in a single collision cascade with large recoil energy. © 2013 Elsevier B.V. All rights reserved.

Creep tests were performed at less than 0.4 Tm (Tm is the melting temperature) for 99.999 99.57 and 99.52% aluminum with several grain sizes in the range of 50-330 μm. These Al materials show remarkable creep behavior with an apparent activation energy (Q) of 30kJ/mol a stress exponent of 4 and a grain-size exponent of zero and with a larger creep rate with increasing purity. These parameters resemble those of conventional dislocation creep which is rate-controlled by the usual diffusion processes except for the extra-low Q value. This means that a non-diffusional process affects the steady state deformation in this temperature region. Transmission electron microscopy revealed the development of a cell structure in the steady state and dislocations without any tangles in the cell interiors. Therefore because the ratecontrolling process could not occur inside of the cells dislocation annihilation occurred through cross slip around the cell walls. According to these creep parameters and microstructural observations the observed creep region is suggested to be a new creep region occurring through a non-diffusional process within the existing deformation mechanism map of Al at less than 0.4 Tm. © 2013 The Japan Society for Technology of Plasticity.

To reveal grain-size dependency and mechanism of creep of 1050 aluminum at low temperatures, creep tests were performed for the samples with grain sizes (d) of 1.0&ndash;47 &micro;m at 233&ndash;473K. Dislocation creep rate-controlled by non-diffusion process was observed at low temperatures, i.e., T<400K and 280K for coarse (CG) and ultrafine grained (UFG) specimens, respectively. The former showed creep at more than 0.2% proof stress, whereas the latter did it at less than that stress. Creep behavior of UFG aluminum was similar to ambient-temperature creep of hexagonal close-packed metals because apparent activation energy was about 30 kJ/mol. Although grain-size exponent was small, i.e. p=0&ndash;0.3, in CG and UFG regions, transient region was observed at d=1.7&ndash;10 &micro;m and creep rate decreased of about one order. At high temperatures, CG and UFG aluminum showed conventional dislocation creep rate-controlled by dislocation-core diffusion.

Creep behavior was investigated carefully for typical face-centered cubic (fee) metals, i e, high purity Al, Cu, and Pb, below 0 3T(m), where T(m) is the melting temperature. All samples showed marked creep behavior at such low temperatures, with an apparent activation energy of 15-30 kJ mol(-1), a stress exponent of 2-5 and a grain size exponent of 0. These results show a new creep region that has not appeared in deformation mechanism maps of pure fcc metals (C) 2010 Acta Materialia Inc Published by Elsevier Ltd All rights reserved.

The constitutive equation in ambient temperature creep region of α-Ti was investigated by performing creep tests on solute-strengthened and/or cold-rolled titanium. Increasing solute content and/or thickness reduction decreases the steady state creep rate. The stress exponent increases with solute content, but it is independent of thickness reduction in low stress. Then, the microyielding stress, σmy, is introduced to express the stress at which dislocations start moving. The stress exponent in ε̇ s - (σ - σmy) graph becomes almost constant with the value of three even in the solute-strengthened and/or cold-rolled titanium. The constitutive equation for ambient temperature creep in α-titanium is proposed as ε̇s= A(S) × (b/d) P × {(σ-σmy)/E}n ×exp(-Q/RT), where n≃3, P≃1 and (Q≃20kJ/mol. The deformation mechanism map of titanium with ambient temperature creep region including microyielding stress is proposed.

Even at ambient temperature or less, below their 0.2% proof stresses all hexagonal close-packed metals and alloys show creep behaviour because they have dislocation arrays lying on a single slip system with no tangled dislocation inside each grain. In this case, lattice dislocations move without obstacles and pile-up in front of a grain boundary. Then these dislocations must be accommodated at the grain boundary to continue creep deformation. Atomic force microscopy revealed the occurrence of grain boundary sliding (GBS) in the ambient-temperature creep region. Lattice rotation of 5 degrees was observed near grain boundaries by electron backscatter diffraction pattern analyses. Because of an extra low apparent activation energy of 20 kJ/mol, conventional diffusion processes are not activated. To accommodate these piled-up dislocations without diffusion processes, lattice dislocations must be absorbed by grain boundaries through a slip-induced GBS mechanism.

This paper reports creep tests on three kinds of polycrystalline hexagonal close-packed metals, i.e. commercially pure titanium, pure magnesium, and pure zinc, in the vicinity of ambient temperature even below their 0.2% proof stresses. These materials showed significant steady state creep rates around 10-9 s(-1) and had stress exponents of about 3.0. Arrhenius plots in the vicinity of ambient temperature indicate extremely low apparent activation energies. Q, of about 20 kJ/mol, which is at least one-fourth of the Q of dislocation-core diffusion. Ambient-temperature creep also has a grain-size effect with an exponent of 1.0. These parameters indicate that ambient-temperature creep is a new creep deformation mechanism in h.c.p. materials. [doi:10.2320/matertrans.M2009223]

Intragranular deformation mechanisms were investigated for ambient-temperature creep of pure hexagonal close-packed (h.c.p.) metals. i.e. commercially pure titanium, pure magnesium and pure zinc, by transmission electron microscopy and electron back-scatter diffract ion pattern mapping analysis. First, straightly aligned dislocation arrays were observed in all of the specimens. Second, although the Burgers vectors of (a) and several slip systems were observed, only one slip system was activated inside of each grain. Third, the deformation twins that form during creep hinder creep strain. Therefore, the dominant intragranular deformation mechanism of ambient-temperature, creep is a planner slip of dislocations inside of a grain. [doi: 10.2320/matertrans.M2009224]

The suppressing effect of ambient-temperature creep of CP-Ti by cold-rolling was reported. Annealed plates of CP-Ti grade 2 were cold-rolled with thickness reductions, and then creep tests under the applied stresses of 0.6-0-9 sigma(0.2) were performed at ambient temperature. With increasing the thickness reduction, the twin, dislocation density and sigma(0.2) were found to increase. At the same time, the steady-state creep rates under the applied stress for constant sigma/sigma(0.2) were decreased. The cold-rolled sample with 20% thickness reduction was then annealed at 813 K for 2400 s to decrease only the dislocation density. After the annealing, the steady-state creep rate remained constant, suggesting that the reduction of the steady-state creep is associated with the increasing twin density. (C) 2008 Elsevier B.V. All rights reserved.

The role of grain boundaries (GBs) for ambient-temperature creep of h.c.p. metals was investigated using pure Zn with several grain sizes. To reveal the relaxation mechanism of ambient-temperature creep, scanning electron microscopy and atomic force microscopy were performed to evaluate the amount of grain boundary sliding. Grain orientation variations were then measured using electron backscatter diffraction to investigate a lattice rotation after ambient-temperature creep. The results obtained by these experiments are as follows: (I) Strong grain size dependency, i.e. larger grain size showed lower total true strain. This is different from high temperature dislocation creep. (2) Grain boundary steps of a few tenths of a micrometer gave evidence of grain boundary sliding during ambient-temperature creep. (3) Lattice rotation of a few degrees was observed near GBs, which indicates that dislocations piled up at GBs. (4) Grain boundary sliding is considered as accommodation process of piled-up dislocations with an apparent activation energy of 18 kJ/mol. (C) 2008 Elsevier B.V. All rights reserved.