Int J Adv Manuf Technol (2017) 90:3055–3066

DOI 10.1007/s00170-016-9650-9

ORIGINAL ARTICLE

A multi-scale study of Inconel 625 powders HIP process

and construction of HIP maps
Min Wang 1 & Yajun Yin 1 & Jianxin Zhou 1 & Xiaoyuan Ji 1 &
Hai Nan 2 & Pengju Xue 1 & Yusheng Shi 1

Received: 3 July 2016 / Accepted: 17 October 2016 / Published online: 27 October 2016
# Springer-Verlag London 2016

Abstract A multi-scale analysis of Inconel 625 powders hot time. Effects of particle size on the densification have also
isostatic pressing (HIP) has been carried out numerically by been considered in the HIP maps.
finite element method base on the modified Shima model.
Numerical results have indicated the capsule deformation Key words Multi-scale . Hot isostatic pressing . Numerical
and powders movement at different HIP parameters. simulation . HIP maps
Furthermore, five interrupted experiments have given a vivid
description about the powder densification process in the mi-
cro level. Results have shown that at the initial stage of HIP, 1 Introduction
densification process is caused by the movement, arrange-
ment, and aggregation of powders. When the coordinate num- Nickel-based alloy has now been widely used in aerospace
ber reaches to the maximum, plastic deformation plays the due to its good mechanical properties, especially in the pro-
leading role in densification. When the applied pressure is duction of compressor disks and gas turbine blades. Current
bigger than the yield stress of powders, parts of the particles manufacturing technology of these complex structural com-
will be squeezed into the voids and all the powders particles ponents focused on the traditional processing method, like
are linked into a whole unit. At the last stage, plastic flows are casting. However, the castings generally present defects in
relatively small; diffusion and power law creep mechanisms the terms of coarse microstructure, composition segregation,
play the leading role in densification. Powder atoms are and porosity, which greatly deteriorate the mechanical prop-
squeezed into the residual porosity slowly with the help of erties. Hot isostatic pressing coupled with powder metallurgy
isostatic temperature and pressure. Based on the above micro technology has been put forward to improve the manufactur-
analysis, coupled with the original empirical densification ing efficiency and quality of the hipped components. As
model, the HIP maps of Inconel 625 powders have been con- shown in Fig. 1, the production process of powders HIP pro-
structed. And the HIP maps have depicted how the relative cess includes the following main operations: (a) design and
density changes with HIP temperature, pressure, and duration fabrication of capsule; (b) filling of the powders into capsule,
evacuation and sealing of the capsule; (c) consolidation of the
powder in the capsule by the combined action of high temper-
* Jianxin Zhou ature and pressure; and (d) acquisition of the component part
zhoujianxin@hust.edu.cn after removal of the capsule.
* Xiaoyuan Ji Comparing with the traditional manufacturing technology,
linnjxy@gmail.com component produced by hot isostatic pressing shows excellent
mechanical performance [1–3]. More specifically, powders
1
State Key Laboratory of Materials Processing and Die & Mould
HIP technology has the ability to form full dense and complex
Technology, Huazhong University of Science and Technology, shape parts with homogeneous microstructure, mechanical
Wuhan 430074, China properties of the fabricated component equivalent to forgings.
2
Beijing Institute of Aeronautical Materials, Aviation Industry of The ABB company from the USA has successfully produced
China, Beijing 100095, China chromium alloy turbine disk and blade based on the HIP
3056 Int J Adv Manuf Technol (2017) 90:3055–3066

Fig. 1 Schematic diagram of HIP

process

technology and reduced about 25–40 % of the manufacturing et al. [17] have numerically studied the powders hot isostatic
costs [4]; moreover, Russia has produced more than 25,000 pressing based on the above models. Recently, Kim et al. [18]
superalloy turbine disks [5]. As is known to all, powders den- have modified Abouaf’s creep densification model by consid-
sification degree in HIP process will directly determine the ering the stacking fault energy as a material parameter, and a
quality of the forming parts. Therefore, it is particularly im- good agreement with experimental results has been found.
portant to analyze the aggregations process. Powders densifi- Advantages of the macro finite element model to simulate
cation process is the growth of necks where the particles are in the powder hot isostatic pressing process are obvious.
contact and increase in the number of such contact points. However, the FEM method cannot reflect the aggregation
When the contact powders number reaches to the maximum and deformation process of individual powder; effects of ag-
number, the remaining porosity is reduced by the deformation gregation and deformation on the densification process cannot
of the dense powders. When the largest amount of plastic be reflected either. From the mesoscopic level, powders den-
deformation of powders has been reached, the diffusion mech- sification process includes powders cluster, rearrangement,
anism will promote the annihilation of the gap between parti- plastic deformation, creep, and diffusion mechanism [19].
cles. Although the densification process is the result of indi- For a better understanding of the densification process during
vidual powder cluster and deformation, the most popular nu- HIP process, researchers, like Molerus [20], Wilkinson et al.
merical method to study powders densification process during [21], Arzt E [11], Nair et al. [22], Helle et al. [12] and Gasik
HIP is based on a hypothesis that the powder body is a con- et al. [23], have divided the powder densification process into
tinuous medium. And traditionally, most of the researchers try two different stages according to the relative density. More
to study the deformation condition and the change of the rel- specifically, in the first stage, the relative density is smaller
ative density during HIP process with the help of a modified than 0.9 and deformation of powders particles are small; in
Von Mises yield criterion and elastic-plastic model in the mac- this stage, growth of the contact neck between powders as
roscopic level. These representative scholars include Kuhn well as the increase of contact number contribute to the den-
et al. [6], Green [7], Shima et al. [8], and Abouaf et al. [9]. sification. In the second densification stage, relative density is
For example, models proposed by Kuhn and Downey have bigger than 0.9. In this stage, the remaining porosity existed
considered the uniaxial yield stress which can be obtained by between powder particles is modeled as an array of spherical
the compression experiments. Abouaf has proposed an equiv- holes, and the following densification process can be modeled
alent stress equation based on the yield theory of porous by calculating the shrinkage of holes. Traditionally, the RDP
metals for FEA. Although the proposed continuum model model of spherical particles has been successfully adapted
cannot directly handle interactions between individual locally [24]. The contact particles number and their average contact
active mechanisms, possibly including granular behavior [5, area in the RDP model are functions of the instantaneous
10, 11], they can give a macroscopic analysis about the den- relative density. As we all know, powders aggregate continu-
sification process of powders during HIP. Moreover, Arzt [11] ally during the HIP densification with the help of applied
and Helle et al. [12] have predicted the overall macroscopic pressure at high temperature. An important function of RDP
deformation of powders by scaling the individual local mech- model is to calculate the average effective pressure. Based on
anisms approach. the RDP model, researchers have modeled the densification
Obviously, these works put their emphasis on the densifi- mechanism to analyze the densification process in the
cation mechanism of powders as well as the container defor- mesoscopic level.
mation. However, the mechanical behaviors, like stress and In this paper, a multi-scale research about the hot isostatic
strain, during HIP process were rarely involved. In order to press of Inconel 625 powders have been carried out. More
make a comprehensive analysis of powders densification specifically, the macroscopical deformation of container during
mechanisms during HIP process, the Arrhenius model and HIP process has been simulated by FEM method. And relative
Perzyna’s model have been proposed [13, 14]. Both of the density, temperature, stress, and strain distribution have been
models are used to describe the viscoplasticity deformation analyzed in the macroscopical level. Also, five sets of experi-
behavior under a relatively high temperature treatment. ments have been designed to study the powders densification
Researchers like Essa et al. [15], Wei et al. [16], and Zhou process; these experimental results can not only validate the
Int J Adv Manuf Technol (2017) 90:3055–3066 3057

accuracy of the FEM simulation method, but also give a vivid temperature and pressure drop to room temperature and pres-
depiction of the powders’ aggregation and deformation process sure directly; in other words, there is no maximum tempera-
during HIP. Moreover, based on the mesoscopic researches of ture and pressure keeping time. Maximum temperature and
HIP process, the HIP maps which can be used making the HIP pressure of the four experiments are 730/80, 850/95, 980/
technological plan have been constructed. 105, and 1100 °C/120 MPa, respectively. Theoretically, in
the fifth HIP experiment, the maximum temperature should
be bigger than 1100 °C and the maximum pressure should be
2 Simulation and experimental methods greater than 120 MPa; moreover. in the fifth HIP experiment,
the maximum temperature and pressure should be maintained
A 3D nonlinear finite element model based on Perzyna’s for 180 min. Unfortunately, it is very dangerous for the HIP
elastic-viscoplasticity equation was applied in this study by furnace to work in such condition. To be sure, purpose of the
using MSC.Marc program to predict the powders’ densifica- fifth experiment is to study influence of the maximum tem-
tion and flowing behavior. A thermo-mechanical coupling so- perature and pressure maintaining time on the densification
lution was used to simulate the thermal deformation during process. So for the safety of researchers and furnace, we have
HIP process. The influence of powder–container friction was reduced the maximum temperature and pressure in the fifth
solved by the use of stick-slip boundary model. To analyze the HIP experiment slightly but maximum temperature and pres-
numerical model, the updated Lagrangian incremental solu- sure maintaining time remains unchanged. So, during the fifth
tion was performed in MSC.Marc. Temperature-dependent HIP process, the temperature and pressures increased to
mechanical and thermal parameters were obtained from ex- 1050 °C and 110 MPa, respectively, within 180 min, and were
periments. After simulation, HIP experiments were performed then maintained at the same value for 180 min. Finally, the
and the relative density of the compact in different regions was temperature and pressure dropped to room temperature and
measured to validate the simulation results. A prep Inconel pressure, respectively, in 3 h. This is why the heating rate
625 powder was used as experiment material. The powder and pressurization rate of the HIP5 are different from other
composition is listed in Table 1. And the powders SEM pic- HIPs.
ture in the loose state can be found in Fig. 2. After HIP, containers have been removed, and specimens
The particle size ranges from 30 to 80 μm, and the melting (5 × 5 × 5 mm) located at the center of the HIP component
point of the powder is about 1300 °C. A designed 304 stain- were obtained by electrical discharge machining. After grind-
less steel cylinder capsule (Φ40 × 90 mm) with 5 mm wall was ing, the specimens were electro-polished at room temperature.
used as the HIP can; a 3D geometric model of the cylinder and The relative density of the specimens were measured based on
2D FEM mesh of the cylinder used in this study can be found Archimedes’ principle, and specimens of all the five experi-
in Fig. 3. Three small circles, colored in red, green, and blue, ments have been detected by the SEM technology to study the
shown in Fig. 3, are the three analysis points for later quanti- powders aggregation and deformation.
tative analysis.
After filling powders, the capsule was degassed at 600 °C
to remove the contaminants in the capsule and make sure that
3 FEM thermodynamic model
the vacuum degree is less than 10−3 Pa. Then, the process of
crimping and welding was carried out. Because the actual HIP
Metal powders show compressibility during the hot isostatic
process is conducted in the high temperature high pressure
pressing process; plastic deformation as well as the volume
furnace, it is difficult to trace the continuous powder aggrega-
contraction can be found between powders. So, both the hy-
tion and deformation process during HIP timely. Five HIP
drostatic pressure and shear stress will affect the yield behav-
experiments have been designed to study the HIP process.
ior of powder, and the equivalent stress for powder material is
The applied maximum temperature, pressure, and their dura-
expressed by a double parameter equation as follows [6]:
tion time of all the five HIP experiments can be found in
Fig. 4. The fifth experiment is a full densification experiment, σ2eq ¼ A J 2 þ BI 21 ð1Þ
and the first four experiments are meant to be the interruption
of the fifth one. HIP temperature and pressure in the first four where σeq is the equivalent stress. J2 and I1 are the second and
experiments increase proportionally, and when the tempera- first invariants of the stress tensor, respectively. Variables A
ture and pressure increase to the maximum value, then the and B are the functions of relative density. In this paper, the

Table 1 Composition of Inconel

625 powder (%) Element C Mo Cr Fe Si Nb Al Ti N O Mn

Mass ration 0.01 8.8 21.5 0.96 0.41 3.71 0.02 0.03 0.12 0.08 0.47
3058 Int J Adv Manuf Technol (2017) 90:3055–3066

Fig. 2 a SEM morphology of

PREP Inconel 625 powders and b
size distribution

modified Shima model [25] which is an intrinsic module in the pressing process is a coupled thermodynamic progress.
applied commercialized MSC.Marc FEM software has been Powders aggregate and deform under high temperature and
adopted. The model is regarded as unified visco plastic, and pressure during the HIP process. The process involves the
the yield function of the material is given by: nonlinearity of materials because of the changing temperature

0:5 and relative density. During the heating and cooling process of
1 3 d d σ2m HIP, the heat conduction could result in the thermal strain.
F¼ σ σ þ 2 −σy ð2Þ
γ 2 β Compared with the thermal deformation and viscoplasticity,
the elastic deformation is not significant and can be ignored.
where σy is the uniaxial yield stress, σd is the deviatoric stress MSC.Marc, a commercial finite element program was used to
tensor, σm is the hydrostatic stress, γ , β materials parameters solve the models described above. In the process of finite
which are the functions of relative packing density ρ [7]: element calculation, the effect of temperature on the material
 b4 properties and thermal strains was considered and material
γ ¼ b 1 þ b2 ρ b 3 ð3Þ parameters against temperature of hot isostatic pressing
β ¼ ð q1 þ q 2 ρ q 3 Þ q 4
Inconel 625 powder can be found in Table 2 [26].
In expression (3), b and q can be determined by fitting the During the HIP process, due to the aggregation of powders,
data from uniaxial compaction experiment. A special situation relative density becomes much higher; parameters of powders
is that when relative density of the porous material approaches especially the Young’s modulus and Poisson’s ratio change.
1.0, the classical Von Mises model can be expressed by the Equations for calculating Poisson’s ratio and Young’s modu-
above equation [16]. As we all know, powders hot isostatic lus at different relative densities are expressed as [23, 27]:
rffiffiffiffiffiffiffiffiffiffiffi
ρ
ν ¼ νm
(a) (b)  3−2ρ  ð4Þ
E ¼ E m 0:15 þ 0:85ρ12

Where νm and Em are Poisson’s ratio and Young’s modulus

for the dense material. A modified Coulomb friction model
can was adopted to handle the friction between metal powders and
inner surfaces of container. An updated Lagrangian incremen-
powder tal solution was adopted to solve the FEM thermodynamic
model.

4 Results and discussion

4.1 Numerical analysis of macroscopic deformation

Due to the high symmetry of the cylinder specimen, in order to

facilitate analysis, the 2D center cross section of the cylinder
Fig. 3 Schematic diagram of 3D (a) and 2D (b) FEM models used in the specimen has been chosen to conduct the numerical simula-
study tion. Figure 5 shows the eventually deformation condition and
Int J Adv Manuf Technol (2017) 90:3055–3066 3059

Fig. 4 Conditions for

temperature and pressure during
five sets of HIP experiments

relative density distribution under the five different HIP con- of the capsule wall is smaller. So, powders near the cylindrical
ditions. In these pictures, the pink solid lines represent the wall will be under bigger deformation driving force and pow-
original capsule shape and the black gray part represents the ders located at the corner will be under relative smaller defor-
deformed capsule. As can be seen, capsule in the first HIP mation pressure. This leads to the bigger powders densifica-
process shares the minimum deformation, while capsule in tion degree at the middle area and smaller relative density at
the fifth HIP process gets the maximum deformation. the corner. And the overall nonuniformity of relative density
Apparently, deformations of the five capsules increase along will decrease along with the increase of HIP temperature and
with the increase in the HIP temperature and pressure. On the pressure.
one hand, resistance of capsules to deformation is smaller In order to give a more intuitive depiction of relative den-
under greater temperature condition; on the other hand, the sity change in different positions, we have chosen three anal-
greater the pressure on the capsule, the greater the deformation ysis points (position of the three analysis point can be found in
driving force acting on the capsule. Aggregation degree of Fig. 3b) to conduct quantitative analysis of the relative density
powders will increase with the help of capsule deformation. changes. Figure 6 shows the relative density changes of the
So, the relative density of the five cylinder specimen increases three analysis points in all the five HIP process. Figure 6 con-
along with the increasing HIP temperature and pressure. As sists of three parts. Picture in the upper left area is the physical
can be seen in the pictures, relative density distribution is not part after the fifth HIP process, and lower right part is the
uniform in all the five HIP process, and powder relative den- relative density changes curves of the three analysis points
sity distribution appears obvious stratification phenomenon. in the third HIP process. The remaining curves are the relative
This is because powders densification process during hot iso- density changes curves of the three analysis points in all the
static pressing is defined by the deformation of capsule. And five HIP processes. Clearly, the final relative density of the
due to the structure characteristics of the capsule, deformation first HIP process is the smallest because of lower temperature
resistance at the intersecting areas between the capsule wall and pressure. As can be seen, in all the five HIP processes,
and capsule cover is bigger, while the deformation resistance relative density of the red analysis point is bigger than that of

Table 2 Material parameters

against temperature of hot Temperature (°C) Room temperature 200 400 600 800 1000
isostatic pressing Inconel 625
powder Poisson’s ratio 0.278
Young’s modulus (GPa) 207 197 185 172 157
Thermal expansion coefficient (10−6/°C) 12.3 13 13.9 14.5 14.7 16.2
Specific heat capacity (J/kg/°C) 410 455 506 559 608 659
Thermal conductivity (W/m/°C) 8 12.4 15.5 18.7 21.7 24.9
Yield stress (MPa) 653 496 442 441 401 102
3060 Int J Adv Manuf Technol (2017) 90:3055–3066

Fig. 5 Relative density distribution of five experiments

the green point, and the relative density of the blue analysis been picked out to give a detailed explanation. The most sig-
point is the smallest. This is because that distance between the nificant characteristic in stage I is that the relative density will
capsule and the red analysis point is smaller than that of the decrease in this stage and the duration time of this stage is
green one, and the distance between capsule and the blue point relatively small. In stage II, relative density runs in a very fast
is the largest. When capsule deforms, the capsule deformation increasing speed and increases almost linearly with the dura-
force will pass through the exterior powders to the inner pow- tion time; at the end of this stage, the densification degree of
ders, so the external powders densification process will pre- the powders are relatively high. In the stage III, the relative
cede the internal powder densification process. Meanwhile, density of powders keep increasing but the increasing speed is
during the deformation force transmission process, part of much smaller than the former stage. To be clear, the whole
the deformation force will be lose, and the effective deforma- HIP process is a continuum process; there are no clear bound-
tion force acting on the inner powders is smaller than that on aries between the above three stages in actual HIP process.
the exterior powders. This also leads to the smaller relative For much more detailed analysis about powder densifica-
density at the blue analysis point. tion process at the three different stages, the deformation ve-
Apparently, all the five sets of relative density changes locity vector of the capsule and powders have been traced,
curves are similar. It is easy to find that there are clearly three which will indicate the flowing law of the powders at three
stages at all the relative density changes curves. And the rel- different times, respectively. There are 200 FEM analysis files
ative density changes curves in the third HIP process have in the fifth HIP process; each file represents actual HIP time of

Fig. 6 Relative density change

with HIP time of three analysis HIP1 HIP2 HIP3 HIP4 HIP5
points in five experiments

HIP3
Int J Adv Manuf Technol (2017) 90:3055–3066 3061

165 s. Figure 7a represents the one moment in the first stage, 4.2 Results analysis of interrupt experiments
which has shown the velocity vector at HIP time of 27.5 min;
as can be seen, the powder and container did not move toward In order to describe the powder’s movement and deformation
the center of the components. On the contrary, the velocity characteristics during the HIP process from a mesoscopic lev-
vector pointed from the center to the wall. It can be explained el, microstructure morphology of the five sets of Inconel625
by the thermal expansion of both the powder and the contain- powders HIP-compacted components are analyzed; SEM im-
er. So, in stage I, relative density will decrease. Figure 7b ages of the five HIP experiments can be found in Fig. 8.
indicates the powder’s flowing characteristics at HIP time of The HIP parameters used in Fig. 8a are 80 MPa/730 °C; the
132 min, and the densification state depicted in Fig. 7b happed microstructure of the powders are similar with that at the ini-
in the second stage. As can be seen, with the increasing tem- tial condition. Only a small contact area between powder par-
perature and pressure, deformation degree of the container ticles can be found, and small pits produced by plastic defor-
increases. However, because of the high deformation resis- mation can be observed on the powder particles; powders
tance at the corner area of the cylinder, deformation at the connect each other through the mechanical engagement rather
corner is relatively small, which means that the effective driv- than diffusion bonding. When the HIP parameters are 95 MPa/
ing force at the center area is bigger than the effective driving 850 °C (shown in Fig. 8b), morphology of individual powder
force at the corner. So, powders flow from the center area to particle is similar with that in the first condition; however,
the corner area symmetrically. And this leads to the symmet- powder particles went through large movement and rearrange-
rical distribution of relative density. A relatively high gradient ment, and the coordination number has increased rapidly;
variation of flow velocity has been observed, because the tem- small particles are fully squeezed into large void space be-
perature and pressure loads exceeded the shielding limit of the tween lager particles; this leads to the increase in relative den-
container, and the deformation amount of the powder rapidly sity. Figure 8c shows the microstructure morphology when
increased. Figure 7c shows the velocity vector at the HIP time HIP parameters are 105 MPa/980 °C. At this situation,
of 484 min. Obviously, this densification state happened in the sintering neck has been formed between particles, and this
last stage. Because the internal powder densification degree is leads to mutual diffusion between particles to promote pow-
relatively high, free movement of powders will be limited and ders densification process. Distances between powder parti-
with the decrease in temperature and pressure at off-loaded cles are small and enclosure space appears between particles.
stage, densification process will no doubt slowdown in this When the parameters are 120 MPa/1100 °C (as depicted in
stage; however, the remaining temperature difference existing Fig. 8d), the enclosure space between particles can no longer
within the compacted powders led to the increased thermal be found; parts of the particles have been squeezed into the
strain. Thus, the deformation amount of the powder increased voids because of the plastic deformation. Although the
and the isostatic pressure would still impel the powder to flow sintering necks have been formed between particles, effects
toward the center. Thus, relative density increases at a rela- of the mutual diffusion are relatively small because of short
tively smaller speed at this stage. maintaining time. The white parts in the picture which

(a) 27.5min (b) 132min (c) 484min

Fig. 7 Powder velocity vector diagram at three different times
3062 Int J Adv Manuf Technol (2017) 90:3055–3066

Fig. 8 SEM image of five HIP

experiment results. a 80 MPa/ (a) (b)
730 °C. b 95 MPa/850 °C. c
105 MPa/980 °C. d 120 MPa/
1100 °C

(c) (d)

PPB

PPB

represent the previous particle boundary (PPB) are still iden- under the fifth HIP process. The sample was prepared by me-
tifiable. PPBs precipitation in HIPing processed superalloys is chanical polishing followed by chemical etching. A mixture of
attributed to the atomic segregation and oxygen-induced sur- 10 ml HNO3, 15 ml HCL, and 10 ml CH3COOH etchant was
face contamination of prealloyed powders. The elements such used. The porosity found in the picture is the result of particle
as Al and Ti react with the adsorbed oxygen and form their boundary corrosion. The punctuated white γ″ (Ni3Nb) phases
highly stable oxides (Al2O3 and TiO2) at the powder particle dispersedly distribute in the matrix. Black networks around the
surface. In this stage, voids between particles can no longer be particles which are the corrosion particle boundaries can be
found because of the full plastic deformation. Sample of the seen obviously. Compared with the fourth HIP experiments,
fifth HIP experiment has been prepared by electrochemical cor- PPBs in the fifth HIP experiment are much fewer because of
rosion. Figure 9 shows the EBSP microstructure morphology the long-time diffusion. However, the PPBs are believed to act
as the nuclei for the preferential precipitation of MC (Nb, Ti) C-
carbides along the particle boundaries during consolidation
[28]. As can be seen in Fig. 9, we can find recrystallization
phenomenon (marked in red block). This means that in this
stage all the powder particles are connected with each other
through the replies and recrystallization process; enclosure
space and powder boundary are no longer visible.
After HIP, specimens (5 × 5 × 5 mm) located at the center
of the HIP component were obtained by electrical discharge
machining. The relative density of the specimens were mea-
sured based on Archimedes’ principle, and the measured rel-
ative density has been compared with the FEM simulation
results; the comparison results can be found in Fig. 10. It
can be found that the measured density is higher than that of
the predicted one. This phenomenon maybe partially attribut-
ed to the fact that the measured density is an average value
while the predicted one is the average value of some analysis
Fig. 9 EBSP image of the fifth HIP experiment nodes. In addition, the FEM model has neglected the
Int J Adv Manuf Technol (2017) 90:3055–3066 3063

however, the densification effects of these mechanisms are

not the same at different stages of HIP. To facilitate the anal-
ysis of the mesoscopic densification process, researchers
[19–21, 24] have made the assumptions that diffusion and
creep mechanisms only come into play when the plastic de-
formation happened to the whole powder body. And the dif-
fusion and creep densification mechanisms have also been
divided into two stages. During the initial stage (D < 0.9),
the individual powder particles are still recognizable, and den-
sification is modeled by the growth of necks where the parti-
cles are in contact and by the increase in the number of such
contact points. During the final stage (0.9 < D < l), the remain-
ing porosity is modeled as an array of spherical holes which
Fig. 10 Comparisons of the relative densities of simulations and shrink in diameter as sintering proceeds [19]. The key points
experiments at five HIP conditions lay in the two stages are the change of coordination number
and effective pressure decided by the coordinate number.
rearrangement and movement of powder particles at the early
stage of HIP. Therefore, the simulation deviation can only be 4.4 Plastic yielding
decreased by taking the micro-effects into account. And devi-
ation of thermal physical parameters and numerical algorithm Plastic yielding is an instantaneous mechanism. Molerus
itself can also lead to these differences. and so on [20, 21] have studied the limit pressure which
can activate the plastic yielding mechanism and studied the
coordinate number changes regularity. And the relative
4.3 Mesoscopic modeling
density Dy after the plastic yielding is calculated by the
following expression:
Based on the above analysis, at the initial stage of HIP, powder
"
1 #
2 P =2
particles connect with each other through point contact; there
D0
are great amounts of spaces between particles, and the coor- Dy ¼ 1þ 1þ ð5Þ
2 3D0 σy
dinate number is relatively small at this stage. The applied
pressure will induce a progressive rearrangement of particles,
and small particles are fully squeezed into large void space
between lager particles; bridges and voids between big parti- 4.5 Power law creep mechanism
cles collapses. This leads to an increase in the coordinate
number. Moreover, the application of hydrostatical pressure When the applied pressure has squeezed the powders into an
at a high temperature induces a bringing closer of cores be- aggregation under the effect of a high temperature, creep
tween particles. This tightening leads to new contact forma- zones appear at the level of the contacts between particles.
tion. At this circumstance, contact area between particles as Then, a powder aggregate densification occurs by power law
well as the deformation resistance increase. In order to precede creep ruled by:
the densification process, we need to increase the HIP pressure :
or lower the critical shear stress of powder’s plastic flow by ε ¼ Aσn ð6Þ
increasing temperature. When the applied pressure is bigger Molerus has given the expression form of A in his re-
than the yield stress of powders, parts of the particles will be searches:
squeezed into the voids because of the plastic deformation.
: 0
Relative density increases rapidly in this stage and all the ε0 A bDv
A¼ n ¼ ð7Þ
powder particles are linked into a whole unit. The residual σ0 kTμn−1
porosities are isolated and dispersively distributed among the
:
powder body, and densification caused by plastic flow mech- where ε and σ respectively, the deformation rate and the ap-
anism can be neglected at this moment. However, diffusion plied stress; meanings of other symbols can be found in the
mechanisms and powder law creep mechanisms will play an appendix. Several studies have analyzed the densification ef-
important role in densification. Powder atoms are squeezed fects of power law creep in both the two stages. Ashby [12,
into the residual porosity slowly by the isostatic temperature 21] has made comprehensive studies about the power law
and pressure. In actual HIP process, all these mechanisms creep mechanisms and has given the expressions of the den-
contribute to the powders densification simultaneously; sification mechanisms. However, experimental results have
3064 Int J Adv Manuf Technol (2017) 90:3055–3066

shown that his calculated results are a little bigger. L. 4.8 Construction of HIP maps
Redouani and S. Boudrahem [19] have adopted the RDP mod-
el to modify the densification models by power law creep. Based on the above densification mechanisms, we have con-
And the contribution of the power law creep mechanism dur- structed the HIP maps at six different sets of HIP parameters.
ing the densification by HIP is expressed by: And the HIP maps can be found in Fig. 11. Figure 11a, b shows
the evolution of relative density changing with pressure for

1=3  

:І D0 DðD−D0 Þ 1−n P n different processing durations (0.25 h/0.5 h/1 h/2 h and 4 h)
Dcreep ¼ 2A
D D0
at 800 and 1000 °C, respectively. Figure 11c, d describes the
n 6 ð8Þ
:П 3 Dð1−DÞ 3 relative density change along with temperature for different
Dcreep ¼ A h in P
2 2n duration times at 80 and 100 MPa. Figure 11e, f shows the
1−ð1−DÞ1=n
effects of particle and grain size on the densification rate under
different temperature and pressure. In these maps, G represents
the grain size, R is the average powders diameter, and Rm
4.6 Diffusion mechanisms
represents the maximum powder diameter. Parameters used in
the HIP maps construction have been listed in the appendix.
At the last stage of HIP process, powders have been linked as an
As can be seen from the above HIP maps, along with the
aggregation unit, matter transport to the isolated voids by dif-
increase of duration time, temperature, and pressure, the relative
fusion mechanisms, and operate along the maintaining level for
density keeps increasing. Due to the differences in the densifi-
the duration of the application of temperature and pressure. Two
cation mechanism in the two stages, the calculated relative den-
ways, grain boundary diffusion and lattice diffusion, can con-
sity curves show two different periods. And from the slope of
tribute to the densification process by diffusion mechanisms:
the curves, we can find that the densification rate in the second
Densification rate by grain boundary diffusion mecha-
stage is smaller than the densification rate in the first stage
nisms:
obviously. The relative density curves changing rule is similar
5=3 with what we have found in the macro FEM simulation results.
: Ωδ D0 D1=3
DІbund ¼ 72Db P In other words, the densification model at the mesoscopic level
kTR3 ðD−D0 Þg ðDÞ can give a quantitative description on the densification process
: Ω D2 ð9Þ
DП
bund ¼ 27δDb 3
n oP during HIP. From the Fig. 11e, f, the comparison results have
kTR 1−5½ð1−DÞ=5D2=3 shown that at the same HIP temperature, pressure, and duration
time, the smaller the powder size, the greater relative density we
Densification rate by lattice diffusion mechanisms: can get. Moreover, differences in the initial stage are much
bigger than that in the second stage. This is because, in the first
: Ω D0 D1=3
5=3 stage, voids between the small particles are smaller than among
DІlattice ¼ 72Dv 2
P the bigger particles. Moreover, it is easy for the small particles
kTR ðD−D0 Þ1=2 gðDÞ
ð10Þ to be pushed into the void by the applied pressure. Although
: Ω D2 ½ð1−DÞ=5D1=3
DП
lattice ¼ 15Dv 2
n oP some assumptions have been made in these empirical models,
kTR 1−5½ð1−DÞ=5D2=3 HIP maps have been proved applicable in providing guidance
for real HIP process [12, 19].
   
1=3 1=3
where g ðDÞ ¼ D
D0 −1 2Z 0 þ C D
D0 −1 .
5 Conclusions

4.7 Nabarro–Herring creep mechanism In a multi-scale level, we have combined the numerical method
and physical experiment method to study the densification pro-
Experiments have shown that when the grain size of powder is cess during HIP of Inconel 625 powders. The densification
significantly smaller than the particle size, the Nabarro– mechanism has been analyzed in both macro and micro level;
Herring mechanism (or Coble creep) contributes to densifica- based on the mesoscopic empirical model, HIP maps have been
tion [12]. constructed. The main contents of this paper are as follows:

1 = " # A modified Shima model coupled with FEM technology
: 4:7Ω D0 3 πδDb has been adopted to simulate the Inconel 625 powders densi-
DІcoble ¼ Dv þ P
kTG2 D G" fication process under HIP based on the commercial software
# ð11Þ Marc. Based on the macro simulation results, powder move-
: 10:5Ω 1−D πδDb
DП
coble ¼ 2=3
Dv þ P ments caused by the capsule deformation have been analyzed.
kTG2 1−5½ð1−DÞ=5D G And the numerical and experimental comparison results have
Int J Adv Manuf Technol (2017) 90:3055–3066 3065

Fig. 11 HIP maps under different HIP conditions

proven the ability of the modified Shima model in predicting whole unit. At the last stage, plastic flows are relatively small-
the capsule deformation and powders densification. er; diffusion and powder law creep mechanisms play the lead-
Five physical experiments have given a vivid description ing role in densification. Powder atoms are squeezed into the
about the powder densification process in micro level. Results residual porosity slowly with the help of isostatic temperature
have shown that at the initial stage of HIP, powder particles and pressure.
connect with each other through point contact; densification Based on the micro analysis of the interrupted experiments,
process is caused by the movement, arrangement, and aggre- coupled with the original empirical densification models, we
gation of powders. When the coordinate number reaches to the have constructed the HIP maps of Inconel 625 powders. And
maximum state, plastic deformation plays the leading role in the HIP maps have depicted how the relative density changes
densification. When the applied pressure is bigger than the with HIP temperature, pressure, and duration time. Effects of
yield stress of powders, parts of the particles will be squeezed particle size on the densification have also been considered in
into the voids and all the powder particles are linked into a the HIP map.
3066 Int J Adv Manuf Technol (2017) 90:3055–3066

Acknowledgments This research was financially supported by the 4. Baccino R, Moret F, Pellerin F, Guichard D, Raisson G (2000) High
National Natural Science Foundation of China (No.51305149) and performance and high complexity net shape parts for gas turbines: the
National Science & Technology Key Projects of Numerical Control ISOPREC® powder metallurgy process. Mater Des 21:345–350
(2012ZX04012-011). And also thanks for the Investment Casting 5. Delo DP, Piehler HR (1999) Early stage consolidation mechanisms
Collaborating Laboratory on Advanced Lightweight Alloy Materials. during hot isostatic pressing of Ti–6Al–4V powder compacts. Acta
Mater 47:2841–2852
6. Kuhn HA, Downey CL (1971) Deformation characteristics and
plasticity theory of sintered powder materials. International
Appendix 1 Journal of Powder Metallurgy 7:15–25
7. Green RJ (1972) A plasticity theory for porous solids. Int J Mech
Sci 14:215–224
Table 3 Parameters and symbols used in this paper 8. Shima S, Oyane M (1976) Plasticity theory for porous metals. Int J
Mech Sci 18:285–291
Symbol Meaning Unit
9. Abouaf M, Chenot JL, Raisson G, Bauduin P (1988) Finite element
simulation of hot isostatic pressing of metal powders. Int J Numer
Tm Melting point K Methods Eng 25:191–212
μ0 Shear modulus at 300 K MPa 10. Dutton RE, Semiatin SL (1998) The effect of density anisotropy on
β Temperature dependence of shear modulus / the yielding and flow behavior of partially consolidated powder
compacts. Metall Mater Trans A 29:1471–1475
σy 0 Yield stress at 300 K MPa
11. Arzt E (1982) The influence of an increasing particle coordination on
σy Yield stress at a given temperature MPa the densification of spherical powders. Acta Metall 30:1883–1890
α Temperature dependence of yield stress / 12. Helle AS, Easterling KE, Ashby MF (1985) Hot-isostatic pressing
diagrams: new developments. Acta Metall 33:2163–2174
Ω Atomic volume m3
13. Sanchez L, Ouedraogo E, Dellis C, Federzoni L. (2013) Influence
D0v Lattice auto diffusion ratio m2/s of container on numerical simulation of hot isostatic pressing: final
Qv Activation energy of lattice diffusion kJ/mol shape profile comparison. Powder Metall
δD0b Grain boundary auto diffusion ration m3/s 14. Sanchez L, Ouedraogo E, Federzoni L, Stutz P (2002) New
viscoplastic model to simulate hot isostatic pressing. Powder
Qb Activation energy of grain boundary kJ/mol Metall 45:329–334
n Powder creep exponent / 15. Essa K, Khan R, Hassanin H, Attallah MM, Reed R (2016) An iter-
A′ Dorn constant / ative approach of hot isostatic pressing tooling design for net-shape
IN718 superalloy parts. Int J Adv Manuf Technol 83:1835–1845
b Burger’s vector m
16. Wei QS, Xue PJ, Liu GC, Lu H, Huang J, Shi YS (2014) Simulation
Z0 Original powder particles coordination number / and verification of near-net shaping a complex-shaped turbine disc by
C Radius radial distribution function in RDP / hot isostatic pressing process. Int J Adv Manuf Technol 74:1667–1677
The average grain size m 17. Zhou S, Song B, Xue P, Cai C, Liu J, Shi Y. (2016) Numerical
G simulation and experimental investigation on densification, shape
R Average radius of particles m deformation, and stress distribution of Ti6Al4V compacts during
D Relative density / hot isostatic pressing. Int J Adv Manuf Technol 1–13
18. ElRakayby H, Kim K (2016) Effect of stacking fault energy on
D0 Initial relative density /
densification behavior of metal powder during hot isostatic press-
:
D Densification rate s−1 ing. Mater Des 99:433–438
19. Redouani L, Boudrahem S (2012) Hot isostatic pressing process
Peff Effective pressure Pa simulation: application to metal powders. Can J Phys 90:573–583
A Steady-state power law creep constant / 20. Molerus O (1975) Theory of yield of cohesive powders. Powder
T Temperature during HIP K Technol 12:259–275
a Contact area between powders m2 21. Wilkinson DS, Ashby MF (1975) Pressure sintering by power law
creep. Acta Metall 23:1277–1285
:
ε0 Deformation rate at ambient temperature / 22. Nair SV, Tien JK (1987) Densification mechanism maps for hot
isostatic pressing (HIP) of unequal sized particles. Metallurgical
σ0 Yield strength at ambient temperature MPa
Transactions A 18:97–107
23. Scott GD (1960) Packing of spheres: packing of equal spheres.
Nature 188:908–909
24. Oyane M, Shima S, Kono Y (1973) Theory of plasticity for porous
References metals. Bulletin of JSME 16:1254–1262
25. An X, Zhang Y, Zhang Y, Yang S (2015) Finite element modeling
1. Cai C, Song B, Xue P, Wei Q, Yan C, Shi Y (2016) A novel near α-Ti on the compaction of copper powder under different conditions.
alloy prepared by hot isostatic pressing: microstructure evolution mech- Metall Mater Trans A 46:3744–3752
anism and high temperature tensile properties. Mater Des 106:371–379 26. Gasik M, Zhang B (2000) A constitutive model and FE simulation
2. Cai C, Song B, Xue P, Wei Q, Wu J, Li W, Shi Y. (2016) Effect of hot for the sintering process of powder compacts. Comput Mater Sci
isostatic pressing procedure on performance of Ti6Al4V: surface 18:93–101
qualities, microstructure and mechanical properties. J Alloys Compd 27. Aryanpour G. (2006) Constitutive modeling for hot isostatic press-
3. Cai C, Song B, Wei Q, Xue P, Wen S, Liu J, Shi Y (2015) In-situ ing of metal powders. J Porous Media 9
integrated fabrication of Ti-Ni coating during hot isostatic pressing 28. Rao GA, Srinivas M, Sarma DS (2006) Influence of modified pro-
of Ti6Al4V parts: microstructure and tribological behavior. Surf cessing on structure and properties of hot isostatically pressed su-
Coat Technol 280:194–200 peralloy Inconel 718. Mater Sci Eng A 418:282–291