During tempering of martensite a complex carbide precipitation sequence appeared in the steel particularly containing microalloyed elements such as V, Nb and Ti. The alloy carbide, which usually precipitates following cementite precipitation in certain temperature range, has been designed to maximize the number density and to retard the coarsening for increasing soften resistance. During the nucleation stage of the alloy carbide, the dislocations and interfaces of distinct phases are the actively precipitated position. However, because of extremely small sizes, their characterization is restricted by the analytic resolution of conventional methods. The 3D atom probe (3DAP) is a particularly helpful instrument with atomic spatial resolution and high componential sensitivity in the characterization of the early stages of precipitation reactions. In this paper, the 3DAP companied with TEM and micro-hardness test was applied to characterize the early nucleation stage of the alloy carbides precipitated during tempering of Nb-V microalloyed steel after quenched from solution treatment at 1200 ℃ for 0.5 h. With the tempering time prolonged from 0.5 to 100 h at 450 ℃, the micro-hardness of the experimental steel changes with the microstructure recovery and carbide evolution (from cementite to alloy carbide). The two peak hardness values appeared at 4 and 100 h tempering are related to precipitate cementite and alloy carbide, respectively. The nucleation of the alloy carbides happens during 30 h tempering at 450 ℃. The alloyed elements dynamically redistributed in the existed remnant austenite, that is, non-carbide-forming elements such as Si and Al diffuse to matrix from the cementite, whereas the carbide-forming elements such as Mo, Nb and V enriched in the cementite, resulting in in situ transformation of alloy carbides. The intragranular defects such as high density dislocation in martensite also act as nucleation sites of alloy carbide, at which V and Nb directly combine with C and lead to the formation of G.P. zone before formation of alloy carbides. Besides, the interfaces of the remnant austenite/matrix and the undissolved AlN/matrix are also energetically favorable nucleation sites, resulting in heterogeneous nucleation of alloy carbides. With the decrease of dislocation density and the dissolutions of cementite and remnant austenite, the consumption of the potential nucleation sites ends the nucleation stage of alloy carbide when tempering for 100 h.

The mechanical properties of quench-tempered high-strength low-alloy steels are commonly optimized by fine and dispersively distributed alloy carbides. The role of the alloying elements in determining the alloy carbide precipitation sequence is of great significance. The co-addition of carbide-forming elements such as Mo, V and Nb complicates the precipitation behavior. The mutual inter--solutions and growth rates of various MC- and/or M₂C-type carbides are qualitatively affected by the intrinsic solubility and diffusion at certain tempering condition. However, comprehensive study of the precipitation sequence must be followed with atomic scale resolution techniques. The 3D atom probe (3DAP) is a unique tool capable of obtaining chemical information at the atomic level, offering a powerful method to investigate microstructural and compositional changes occurring at nano-scale. And the sizes, morphology and composition of individual alloy carbide may be visualized and quantified by 3DAP. In this paper, a quenched Nb-V microalloyed steel was chosen to investigate the precipitation behavior of alloy carbide after tempering at 450-650 ℃ for different times. 3DAP, micro-hardness test and TEM were applied to characterize the phenomena of hardening and softening during tempering, and the composition evolution and growth behaviors of the alloyed carbides were also studied. The results indicated the second hardening of the 500-600 ℃ tempering martensite is mainly resulting from precipitation strengthening of alloy carbides. The alloy carbides composition dynamically changed with elevated temperature or prolonged time, that is, the stronger carbide-forming elements replaced or partly replaced the weaker ones. At first V and Nb replaced Mo, and then Nb partly replaced V, and at last the carbides with certain composition were formed. Tempering time has relatively less effect on the carbides composition compared with temperature. When the tempering temperature elevated or tempering time prolonged, the alloying elements can obtain adequate diffusion energy and/or time, the plate-like carbides preferentially grow along the radial direction, and then grow along the thickness direction and start to coarsen.

It has been confirmed that the fcc MC-type carbides such as VC and NbC have plate-like morphology and are mutually soluble when precipitated in tempering steel with martensite microstructure. Over-tempering makes the plate--like carbide change to spherical shape because of Ostwald coarsening. As coarsening is strongly linked to the diffusion rate of the carbide-forming elements, it is easy to understand that inhomogeneous structure may be formed when more kinds of elements were added, such as V, Nb and Mo. Besides, non-carbide-forming elements such as Si and Al tend to diffuse toward matrix. The morphology and lattice structure of carbide change simultaneously companying with the compositional redistribution of alloy elements. The detail compositional and nano-structural informations of the carbide can be obtained by 3DAP and HRTEM. In this paper, 3DAP and HRTEM were applied to characterize the composition, morphology and nanostructure of the carbide precipitated during 650 ℃ tempering of as-quenched Nb-V microalloyed steel. The results indicated that the martensite lath morphology was replaced by defect--free polygonal ferrite due to the recovery and recrystallization of the as-quenched microstructure. Simultaneously, the carbide-forming elements Mo and V diffused from smaller carbides to larger ones, resulting in the co-existence of carbides with different sizes and compositions. With the diffusion and redistribution of alloy elements, the prior-formed plate-like carbides grew along the radial direction, and a kind of transition carbide (P_outer) which is semi-coherent with ferrite matrix (M_bcc) was consequently formed. 3DAP constructional atom map demonstrated that Si and Al are rejected from the alloy carbide, whereas Mn and V were inhomogeneously distributed. That is, the coarsening carbide has a core-shell complex nanostructure, the core contains V, Mn, Mo and Nb, and the external shell contains Mo and Nb.

Though electron backscattering diffraction (EBSD) is widely employed for the orientation analysis of deformed microstructures in many metallic materials, its applications to Mg and its alloys are not widespread because of the difficulties involved in sample preparation. In this work, uniaxial compression tests were performed on samples cut along the extrusion direction from AZ31 Mg alloy bars, and then the twins and their intersections were analyzed by SEM/EBSD microscopy. Deformation and deformation mechanisms operating in Mg alloy depend sensitively on temperatures. Flow curve at 523 K shows a sharper and larger stress peak following slow strain hardening and rapid strain hardening. At slow hardening stage (0.02≦ε<0.06) a few of {1012} twinning was commonly observed, but at rapid hardening stage (0.06≦ε<0.22$) more twins and twin intersections appeared. There are five possible types of twin intersections, which are strongly depended on the stress direction. The two types of (1012)-(0112) and (1012)-(0112) twin intersections were observed at stress axis near to <1120> and <1010> in the samples compressed to a true strain of 0.06.

The computation method of the interaction force between adjacent dislocation segments and the discretization method of the dislocation line in the 3D model of discrete dislocation dynamics were improved in order to simulate the evolution of Frank-Read (FR) dislocation source on the meso-scale. The simulation results show that the improved model can more accurately simulate the critical shear stresses and dislocation configurations of the FR dislocation sources with different initial lengths. Moreover, the smaller dislocation loop is formed by FR dislocation source when larger stress loaded, and approaches the dislocation loop formed without considering the interactions among the dislocation segments. Under the same loaded stress, the larger loop is formed with increasing the Peierls stress. However, the drag coefficient only affects the formation time of the loop, but can not affect the loop size. This kind of nonlinear evolution of dislocation is caused by the interactions among the dislocation segments.

Previous researches indicated that the mechanical property of dual phase steel is not only depended on the volume fractions and grain sizes of ferrite and martensite but also the morphology and distribution of martensite island. Therefore, it is desired to obtain dispersive distribution of fine martensite islands in the matrix of fine grained ferrite. Generally, there are two methods to refine ferrite grain. First, γ/α dynamic transformation is promoted by increasing austenite free energy through heavy deformation at low temperature region. Second, fine ferrite grain is achieved by refining the initial austenite grain which can be obtained by microalloying, recrystallizing and cyclic heat treatment. In this paper, a low carbon Nb-microalloyed steel was cyclic-heat-treated to obtain 4.2 μm sized initial austenite grain and then cooled to different temperatures (810-720 ℃) to compressively deform. The effects of deformation temperature on flow stress curve, and the morphologies and distributions of ferrite and martensite island, two constituted phases in the steel, were investigated. The flow stress curves possess peak stress which increases first and then decreases with decreasing of deformation temperature. And the volume fraction of ferrite also decreases first and then increases with decreasing of deformation temperature, but the change is slight. At the lowest deformation temperature of 720 ℃, the size of ferrite grain was decreased to 2.8 μm and the volume fraction of fine martensite island which is dispersively distributed around the boundaries of ferrite was increased up to 22.7%. The inhomogeneity of the hardness of ferrite grains lowers with increasing of deformation temperature, and the hardness approaches a small stable value at last. The EBSD orientation maps show that the fraction of low angle grain boundary increases with decreasing of deformation temperature.

PVD or CVD Me-Si-N nanocomposite films synthesized by doping Si element in metallic nitride matrix have exhibited good oxidation resistance and wear resistance. As melting the alloy target containing Si is not easy, it is difficulty to dope much more Si in the films by PVD techniques. In addition, the Me-Si-N films do not have enough lubrication. In this paper, Cr-Si-C-N films were prepared by cathode arc ion deposition technique, in which tetramethylsilane (TMS) was used as Si and C sources, and their concentrations in the Cr-Si-C-N films can be controlled by TMS flow. The state of chemical bonding, microstructure and microhardness were investigated by XPS, XRD, HRTEM and microindentation hardness tester. Results show that the Si and C contents increase monotonicly with the increase of TMS flow. When the TMS flow is lower than\linebreak 90 mL/min, the Cr-Si-C-N film has a composite structure of Cr(C, N) nanocrystals dispersing in the amorphous Si₃N₄ (nc-Cr(C, N)/a-Si₃N₄), and the microhardness increases to 4500 HK with increasing TMS flow. Such high hardness originates from the solid solution hardening of the doping fewer element and the Veprek nanocomposite structure hardening mechanism. With the further increase of TMS flow, the hardness decreases because of the appearance of free C.

Steel 1Cr13Mn13 was treated by electropulsing (EP) with high density after annealing+furnace cooling (AF) or solution+oil quenching+tempering (SOT). Tensile experiments show that the mechanical properties of the sample treated by SOT+EP process with a maximum current intensity about 8.46 kA/mm² were greatly improved, compared with the sample only treated by SOT process, its tensile strength and elongation increased from (1250±10) MPa and (20±1)\% to 1400 MPa and 53%, respectively. Metallographs show that the equiaxed grains formed under high temperature solid solution treatment were remained in both the SOT and SOT+EP samples. But the coarse lamellar martensite in the SOT sample was greatly refined after EP treatment. As the samples were treated by EP treatment, they successively underwent a rapidly “heating-cooling” process with a heating time less than 0.001 s and a cooling time less than 0.6 s, resulting in a successive phase transformation of α´→γ→α´. The lamellar α´ martensites formed during rapidly cooling process have more refined size than that of the primary martensite laths, which further enhanced the tensile ductility and tensile strength of steel 1Cr13Mn13. Moreover, the maximum value of current intensity for each EP treatment can also influence the mechanical properties of the samples by controlling the amount of the α´ phase in the transformation. A higher current intensity usually produces much more amount of α´ phase with smaller lamellar size and the mechanical properties of the samples are more improved. But if the maximum current intensity is much higher than an optimal range about 8.30-8.46 kA/mm², the sample is overheated and the mechanical properties will be lowered.

An induced coil surrounding a segmented mold used in soft--contact electromagnetic casting
(soft-contact EMC) was used to produce a high frequency magnetic field for reducing ferrostatic pressure
between the mold and melt. The distribution of magnetic field in the mold was examined using a magnetic
probe of the induction coil type. Then mathematical model was developed to study the distributions of
magnetic field, electromagnetic force and flowing velocity of molten steel in the mold. Finally, continuous casting
experiments were conducted with alloy constructional steel 15CrMo in the laboratory caster. The surface
morphologies and macrostructure were examined and analyzed. Based on the comprehension of the
distributions of magnetic field, electromagnetic force and flowing velocity of molten steel in the mold through
measurements and numerical simulation, the effects  of electromagnetic field were systematically investigated.
The results indicate that when the electromagnetic field was applied in the initially solidified area,
the mold flux consumption was increased dramatically. As a result, the surface quality of continuously
cast billets is greatly improved, for example, oscillation marks disappeared due to the decrease of flux pressure. Moreover, the
growth of columnar grains is suppressed for two main reasons. The first one is that the mold near meniscus is
heated by Joule heat generated by the high frequency electromagnetic field. The other one is that the thermal
resistance between mold and the solidified shell is increased as the increase of mold flux thickness.
Inhomogeneous distributions of magnetic field in the mold along the casting direction were confirmed both by
measurement and numerical simulation. And the Lorentz force on the molten steel along the casting direction is
uneven likewise. Under the drive of Lorentz force, two counter-rotational vortices are formed below the
meniscus. Moreover, the temperature gradient in front of the solid/liquid interface is decreased as a result of the
circulation of liquid steel. Therefore, composition supercooling is easily obtained in the liquid core, which is
helpful to the growth of equiaxed crystals.

Ti-Al alloys as the high temperature structural material with the most prospective development are
widely used in aerospace. Further study should been conducted on their formation of fully lamellar structure in
directional solidification and lamellar orientation control for a good balance of mechanical properties. Directional solidification
experiments were conducted for Ti-50Al (atomic fraction, %) alloy in a relatively wide range of growth rates. The
effects of growth rate on interfacial morphology, microstructure evolution and formation of lamellar structure were
investigated. A single-phase growth of cellular α was observed in a growth rate range of 1-5 μm/s, and finally a fully
lamellar structure was formed. When the growth rate reached 10 μm/s, a single-phase growth of cellular α was
also observed during a relatively long distance after initial solidification, but as solidification proceeded, intercellular
solute enrichment became so severe that γ phase precipitated from liquid appeared between α cells, and finally a full
lamella can not be formed. When the growth rate was higher than 15 μm/s, a dendritic growth of α phase and γ
phase between $\alpha$ dendrites were observed.  The analysis on the final lamellar orientations at different growth rates
showed that the lamellar orientation is history-dependant on the orientation of as-cast grain at the started interface of direcitonal
solidification. Based on the above rules, Ti-50Al alloy, also as seed, was solidified under controlling the
lamellar orientation, and a relatively low growth rate of 8 $\mu$m/s was chosen to ensure a single-phase growth of $\alpha$.
Finally, a fully lamellar structure with an orientation parallel to the growth direction was obtained.

The effect of a static magnetic field on the solidification of monotectic alloy has attracted great attentions. But up to date little is known about the details of the magnetic field influences on the microstructure development during the solidification of a monotectic alloy. In this paper, rapid directional solidification experiments are carried out with Al-Pb alloys under a static magnetic field. The dependence of the solidification microstructure on the intensity of the magnetic field is investigated. The mechanism through which the magnetic field affects the microstructure formation is analyzed. It is indicated that the static magnetic field causes only a small decrease in the Marangoni migration velocity and the Stokes settlement velocity of the minority phase droplets during the liquid-liquid decomposition. Such a small change in the moving velocity of the droplets has only a negligible effect on the microstructure formation. The static magnetic field can suppress the convection efficiently. It enhances the spatial homogeneity of the liquid-liquid phase transformation along the radial direction of the sample, reduces the collision and coagulation frequency between the minority phase droplets and, therefore, causes a decrease in the largest and average sizes of the Pb-rich particles in the solidified sample as well as in the width of the particles size distribution. A static magnetic field is favorable for obtaining a monotectic alloy with well dispersed microstructure.

The creep behavior of a Ni base single crystal superalloy with [011] orientation under three conditions of temperature and stress level has been investigated. Creep deformation of the tested alloy occurs largely through dislocation activity in the γ matrix channel. Shearing of the γ´ precipitates is observed, while at higher temperature, this does not occur until late in life by means of the passage of superpartial dislocation.  At lower temperature (750 ℃) and high stress level, shearing of the  γ´ precipitates is observed in the relatively early creep through the passage of 1/3<112> dislocation, which leaves superlattice stacking faults (SSFs) in the precipitates. The creep behavior is closely related to microstructure evolution, the creep curve at 750 ℃ exhibits higher primary and steady creep rates, and thereby the creep life is shorter. Under the condition of 870 ℃ and 500 MPa, the steady-stage creep does not appear, it is suggested that the creep life is greatly influenced by the inhomogeneous slip band. At higher temperature and lower stress, such as 980 ℃ and 200 MPa, the alloy has longer creep life and lower steady creep rate. Observation of the dislocation configuration shows that the hexagonal dislocation network appears on the  γ/ γ´ interface at the early creep stage, the regular and denser dislocation networks can inhibit dislocation cutting into  γ´ phase and enhance the resistance of dislocation movement. In the later stage, $\gamma'$ precipitates are sheared by dislocation, which leads to an acceleration of creep rate.

Very-high-cycle fatigue of metallic materials is commonly regarded as fatigue failure occurs at stress levels below conventional fatigue limit and the relevant fatigue lives are above 10⁷ cyc. Rotary bending fatigue tests for a structural steel 40Cr were performed in laboratory air, fresh water and 3.5\%NaCl aqueous solution, respectively, to investigate the influence of environmental media on fatigue behaviors of the steel in high cycle and very-igh-cycle fatigue regimes. The results show that the fatigue strength of the steel in water is remarkably degraded compared with that in air, and the fatigue strength in 3.5%NaCl solution is even lower than that in water. The fracture surface observations show that for the specimens tested in water and 3.5%NaCl solution, multiple crack originations exist and cracks propagate along grain boundary with widespread secondary cracks in their steady propagation period.

Constant load creep tests were performed on the [001] and [011] oriented Ni-Co-Cr-Mo-W-Al-Ti-Ta single crystal superalloys. The [001] oriented alloy has much longer creep life than that of [011] oriented alloy, but the elongation of [011] oriented alloy is slightly higher under the condition of 750 ℃/750 MPa. The average creep life and elongation of [001] oriented alloy are both higher than that of [011] oriented alloy at 982 ℃/248 MPa, and the anisotropy occurs mainly during the accelerating creep stage, but anisotropic degree decreases obviously. The SEM analysis revealsγ ´phases are rafted in the two directions, which blocks the glide/climb of dislocations and causes creep hardening, the rafting of γ´ phase is the dominant reason to decrease of creep anisotropy at higher temperature. The TEM observation indicates deformation twins formed in [011] oriented alloy, which lowers the plasticity of the sample and induces the sample to fracture rapidly.

The yield characteristic and the plastic flow direction of a polycrystal copper are investigated, in which the anisotropy and random orientation of each grain in the polycrystal are taken into account, while the microstructure evolvement and the slip deformation mechanism are also analyzed. Applying the crystal plasticity theory associated with representative volume element (RVE) of a polycrystal aggregate, which consists of 200 polyhedral grains with irregular shape and orientation, the plastic deformation of polycrystalline copper is calculated through applying biaxial load along different paths to the RVE aggregate, stage by stage to simulate the material's biaxial stress state and the sub-stage load path. Then the yield surface and the subsequent yield surface for the RVE under preloading are obtained by the simulation through FEM calculation with the user crystalline material subroutine. The calculation results of the subsequent yield surface shape and the plastic flow direction are resolved and are discussed further. According to the results of yield surface and plastic flow direction of the polycrystal RVE, it can be concluded that the corner may appear on the subsequent yield surface at the preload point and the corner's appearance is dependent on the yield definition and the preload direction on the $\pi$ plane; the classical normality description for plastic flow is proved to be reasonable for the polycrystal aggregate but there is a difference between the flow direction and the surface normal vector, which is analyzed by statistical calculation, and the statistical difference between the plastic flow direction and the normal vector of subsequent yield surface is related with both the yield definition and direction of preloading.

In order to analyze the fatigue property and residual stress relaxation process of microshot peened medium carbon steel, the 10⁹ cyc fatigue tests of the specimens unpeened and shot peened by steel balls and ceramics balls with 100 $\mu$m in diameter were carried out by rotary bending fatigue machine in air at room temperature. The result shows that the fatigue limits of specimens peened by steel balls and ceramics balls are improved by 35% and 23% respectively, compared with that of unpeened specimen. Based on the test result of the residual stress in the fatigue process, the process and mechanism of residual stress relaxation are analyzed in detail. The cyclic yield strength of the material is the main factor controled the improvement level of the fatigue limit.

From the application point of view, corrosion resistance of materials in corrosive environments, especially in Cl^- containing medium, has great significance when used as mechanical components serving in marine and other aggressive environments. The corrosion behavior of a material is largely controlled by the presence or absence of protective surface film, which may act as a protective barrier against corrosion attacks. Therefore, the corrosion resistance of an alloy is closely related to the particular composition of the passive film and the synergistic interaction between the cations of alloy components in the passive film. In the present study, a γ-toughened Cr₁₃Ni₅Si₂ metal silicide alloy, consisting of Cr₁₃Ni₅Si₂, Ni base solid solution $\gamma$ and Cr₃Ni₅Si₂ was designed and fabricated by the introduction melting and die-casting prosess. Corrosion behaviors of the alloy in a series of Na₂SO₄+NaCl solutions were investigated by anodic polarization, Tafel plot and electrochemical impedance spectroscopy (EIS) experiments. Chemical composition of the passive film and the surface of polarized samples were examined by X-ray photoelectron spectroscopy (XPS) and scanning electron microscopy (SEM), respectively. A commercial solution treated single phase austenitic stainless steel 1Cr18Ni9Ti was selected as the reference material for all the electrochemical tests. Results showed that the metal silicide alloy exhibited high corrosion resistance in all testing solutions due to the formation of a compact and protective passive film composed mainly of Cr₂O₃, as well as the high chemical stability of Cr₁₃Ni₅Si₂ and Cr₃Ni₅Si₂ phases. No evidence of localized corrosion occurred even after anodic polarization in 0.1 mol/L Na₂SO₄+1 mol/L NaCl solution. Moreover, the corrosion potential, breakdown potential and anodic current density are almost invariant with increasing Cl^- concentration, which means the alloy has excellent corrosion resistance in neutral Cl^- containing solutions.

Minor Fe and Mn additions are necessary to enhance the corrosion resistance of commercial Cu-Ni alloys. The present paper aims at optimizing the addition amounts of Fe and Mn in Cu₇₀Ni₃₀ (atomic fraction, %) alloy using a cluster-based solid solution model. In this model it assumed that one Fe(Mn) atom and twelve Ni atoms formed a cluster consisted of Fe(Mn)-centered and Ni-surrounded cube-octahedron and the limit solid solution would be composed of isolated Fe(Mn)Ni₁₂ clusters embedded in the Cu matrix. The ratio of the Fe(Mn) atoms and its surrounding Ni atoms is 1∶12, and the limit solid solution composition of Fe(Mn)-modified Cu₇₀Ni₃₀ alloy is [M_1/13Ni_12/13]₃₀Cu₇₀=[(Fe_1-xMn_x)Ni₁₂]Cu_30.3, M=(Fe_1-xMn_x). The OM, XRD and electrochemical corrosion measurements were used to characterize the microstructure and corrosion resistance performance of [(Fe_1-xMn_x)Ni₁₂]Cu_30.3. The results indicated that the solid solubility limitative alloys [(Fe_0.75Mn_0.25)Ni₁₂]Cu_30.3 has the best corrosion resistance in 3.5%NaCl aqueous solution.

Casting Mg-Al-Zn alloys are promising for the automotive components loaded under both high strength and high ductility, as well as under high temperatures. The relationship of compositions, phase constituents and solidification paths of casting Mg-Al-Zn alloys were investigated by SEM/EDS, XRD and thermodynamic calculations. It is shown that the phase constituents of Mg-Al-Zn alloys are related to Zn/Al ratio; with the Zn/Al ratio increasing, the secondary phase γ-Mg₁₇Al₁₂ is gradually replaced by Φ-Mg₂₁(Zn, Al)₁₇ and eventually disappears completely. The phase consituents and solidification paths of the alloys under various conditions, including equilibrium, Scheil and permanent mould casting, were examined by thermodynamic calculation software Pandat with the availability of thermodynamic description of Mg-Al-Zn ternary system. The practical casting process deviates from the equilibrium; however, the practical phase constituents of the experimental alloys except ZA65 can be predicted by the Scheil model. Because of no considering the peritectic reaction in which solid participates as reactant, the Scheil model can not correctly predict the phase constituents of ZA65 alloy under permanent mould casting condition.

Chemical vapor infiltrated carbon felt with the density of 1.0 g/cm³ can be further dendified by resin impregnation-carbonization more times. Mercury proximity was used to detect changes of total porosity and pore size during the process. The fractal method was used to characterize the evolution model of pores. As a proof, the changes of morphology and size of pores with impregnation-carbonization times were observed by SEM. Results indicated that the total porosity decreased with the increase of resin impregnation-carbonization times. On the contrary, the pores with size ranging from 0.04 to 6 μm increased during repeated treatment. The pore evolution law can be characterized by the fractal, and the pore fractal dimension is constantly decreased with the increase of impregnation-carbonization times. Moreover, plenty of crack-like pores formed at the first resin impregnation-carbonization were gradually filled during further impregnation-carbonization processes, which was conformed with evolution model of pores proposed by fractal method.