Powder Technology 239 (2013) 389–402

Contents lists available at SciVerse ScienceDirect

Powder Technology

j ourna l homepage: www.e lsev ie r .com/ locate /powtec

Gas-assisted powder injection molding: A study about residual wall thickness

Donghan Kim a, Seokyoung Ahn b,⁎, Kye Hwan Lee c, Rajiv Nambiar c, Sang Won Chung d,Seong Jin Park e, Randall M. German f

a Mechanical Engineering, Ajou University, South Koreab Manufacturing Engineering, University of Texas-Pan American, TX, USAc Mechanical Engineering, Pusan National University, South Koread Automotive Engineering, Nambu University, South Koreae Mechanical Engineering, Pohang University of Science and Technology, South Koreaf College of Engineering, San Diego State University, CA, USA

⁎ Corresponding author.E-mail address: sahn@pusan.ac.kr (S. Ahn).

0032-5910/$ – see front matter © 2013 Elsevier B.V. Allhttp://dx.doi.org/10.1016/j.powtec.2013.02.032

a b s t r a c t

a r t i c l e i n f o

Article history:Received 8 October 2012Received in revised form 12 February 2013Accepted 14 February 2013Available online 26 February 2013

Keywords:Powder injection moldingGas-assisted injection moldingGas penetration depthResidual wall thicknessStereolithography cavity

The effects of processing variables on gas penetration depth and residual wall thickness (RWT) of gas-assistedinjection molding (GAIM) parts were investigated for polypropylene (PP) and 316L stainless steel powder feed-stock (SUS316). The processing variables were melt temperature, shot size, gas pressure, and gas delay time. Byusing a Taguchi L9 array, the results were compared with the previous work. Mold material was also investigatedby molding with both aluminum and Stereolithography (SLA) mold inserts at varying temperatures. The mostsignificant parameter affecting RWT was melt temperature for PP and gas delay time for SUS316. Additionallyfor SUS316, it was found that gas penetration depth and RWTwere decreasedwith increasing mold temperature.While both computer simulation and experimental validation were carried out, the results from simulationsfailed to consider the effect of thermal conductivity differences between SLA and Al cavities due to lack of acoupled analysis capability in its module.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

The injectionmolding process is one of themost popularmanufactur-ing methods due to the capability to produce a high volume of precisionparts using awide range of plasticmaterials. The cycle time is also signif-icantly less compared to the other manufacturing techniques. As withthe material flexibility of injection molding, powder injection molding(PIM) has presented an effective andmature technology able to produceparts of complex geometry through injecting metal or ceramic powderblends in place of plastic into the cavity [1–3]. The PIM allows the fastcycle time and low cost production of small-sized complex componentsconsistently with high dimensional accuracy.

Gas-assisted injection molding (GAIM) is a modification of the con-ventional injection molding and a method of pressurizing an injectionmolding part with gas in order to provide the necessary packing forceto produce a quality injection molded part [4]. GAIM produces partswith hollow internal sections and this becomes particularly usefulfor PIM [5–7] as lower material usage is achieved. Required time fordebinding processes can be reduced as well [8]. Despite the advantages,theGAIMprocesswith powder feedstock, i.e., gas assisted powder injec-tion molding (GAPIM) process is not intensively studied. Also the effectof processing variables over GAPIM process is not well understood yet.

rights reserved.

Epoxy cavities made by Stereolithography (SLA) were used in theprior experiments by Lee et al. [7] due to low cost and easy fabrication.They found out that durability of SLA cavities can be significantly in-creased by the application of GAPIM due to reduced injection pressurein the experiments [9]. However, use of metal cavities is necessary inorder to accomplish mass production and commercialization. Accord-ingly, this research will focus on the effects of the processing param-eters such as melt temperature, shot size, gas pressure, and gas delaytime on gas penetration depth using an aluminum (Al) cavity insert.Simulations and experiments were conducted to find the effects ofprocessing variables and the results were compared to the previousstudy using an SLA cavity. Furthermore, residual wall thickness (RWT),an important parameter for commercial application of GAPIM,was inves-tigated. The effect of mold temperature on gas penetration and RWTwasalso studied.

1.1. Background

Karatas et al. [10] recently studied the moldability of various feed-stocks used in PIM. It was mentioned that feedstocks used in PIM havehigh thermal conductivity which leads to fast solidification and as a re-sult, very high injection rates were required. The high injection ratespromoted the accumulation and separation of binder in sudden directionchanges in the cavity during injection molding. This separation causeddefects in the sample which become apparent during debinding and

Fig. 1. Schematic diagram of GAPIM process.

Fig. 2. Schematic diagram of GAIM.

390 D. Kim et al. / Powder Technology 239 (2013) 389–402

sintering. In the experiment, the moldability increased with increasinginjection pressure, temperature, and flow rate.

Urval et al. [11] investigated the influence of decreasing part thicknesswith accompanying increase in aspect ratios on the process parameters,including melt temperature, mold temperature, fill time and switch-over position. This was done using the Taguchi design method and asimulation tool (PIMSolver) [12]. It is reported that as less part thickness,higher melt temperature and mold temperature would be necessary to

Fig. 3. Advantages provided by application of GAIM to PIM.

obtain complete parts due to the solidification rate increase. The moldtemperature was considered to be the most critical parameter.

1.2. Gas-assisted injection molding process

GAIM technology has been increasingly adopted due to many ad-vantages over conventional injection molding, such as reduction ofinjection-packing pressure, cooling time, and material usage. Injectedparts have less sink mark, shrinkage, warpage, and residual stresses,resulting in better final production at lower costs [13–16]. GAIM is atechnology which injects gas to form hollow cores in the thicker sec-tions of the part [17]. Fig. 1 shows a schematic of GAPIM and Fig. 2shows the process of GAIM in four stages.

At the first stage, a fixed amount of plastic melt is injected into themold cavity, less than the full volume of the cavity, which is called as“short shot”. The injection pressure required is reduced due to shortshot in the cavity as compared to conventional injection molding.

In the second stage, the nitrogen gas is injected and the plasticmelt is displaced by applied gas pressure. It takes the path of least

Table 1Material properties.

Property PP SUS316

Density (g/cm3) 0.9 7.76Specific heat capacity (J/kg-K) 2740 685Thermal conductivity (W/m-K) 0.16 1.84Thermal diffusivity (m2/s) 0.65×10−4 3.46×10−4

Fig. 4. Dimension for (a) mold cavity inserted in injection mold, (b) mold cavity inserted in ejector mold, and (c) the part. All dimensions are in mm.

391D. Kim et al. / Powder Technology 239 (2013) 389–402

resistance ideally along the center section of thicker channels that areat a relatively high temperature. In the third stage, the gas pushes theplastic melt from the thick section of the part to the unfilled extremitiesof the vented cavities, thereby filling the part and leaving a hollow sec-tion in the channel. The gas continues to apply pressure as the plasticcools, solidifies, and packs more efficiently. The pressure that is appliedagainst the walls of the mold cavity is lower than the packing pressureused in conventional injection molding. Further, the gas is compressibleand so applies a uniform pressure on the inside surface throughout thepart resulting in better packing, thus minimizing sink marks and surface

Fig. 5. Aluminum cavities.

blemishes which lead to a more aesthetically pleasing part [4,18]. In thefourth stage, the part is completely cooled and the gas being vented be-fore the mold opens.

Poslinski et al. [19] investigated gas-assisted displacement ofviscoplastic liquids in tubes. They found that RWT could be determinedby a capillary number (Ca) which shows asymptotic behavior. It wasalso concluded that RWT was determined by the gas penetration rate.Chen et al. [20] studied the characteristics of gas penetration in a spiraltube and found that the RWT behind the gas front was uniformly distrib-uted in the primary gas penetration stage, two types of characteristics onRWT were observed near the gas front. Yang and Chou [21] studied theuniformity of RWTdistribution arounddimensional transition and curvedsections in circular tubes as RWTwas appearing non-uniformnear transi-tions. The uniformity of RWT in the transition was improved by the addi-tion of fillets. Recently there was an experiment to improve the RWTuniformity around curved sections in a tube by varying mold tempera-tures [22]. According to this experiment, the RWT uniformity could be

Table 2Fixed processing conditions.

Parameter Value

Injection pressure 8.48 MPaHold pressure 6.89 MPaBack pressure 0.52 MPaClamping force 11.0 MPaScrew speed 50.0 rpm

Table 3DOE 34 factor L9 orthogonal array.

Trial Melt temperature Shot size Gas pressure Gas delay

1 1 (low) 1 (low) 1 (low) 1 (low)2 1 (low) 2 (medium) 2 (medium) 2 (medium)3 1 (low) 3 (high) 3 (high) 3 (high)4 2 (medium) 1 (low) 2 (medium) 3 (high)5 2 (medium) 2 (medium) 3 (high) 1 (low)6 2 (medium) 3 (high) 1 (low) 2 (medium)7 3 (high) 1 (low) 3 (high) 2 (medium)8 3 (high) 2 (medium) 1 (low) 3 (high)9 3 (high) 3 (high) 2 (medium) 1 (low)

Table 4Molding window for PP and SUS316.

Process Level Melt temperature Shot size Gas pressure Gas delay

PP 1 193 °C 73% 5.65 MPa 0.5 s2 204 °C 76% 5.79 MPa 1.0 s3 216 °C 79% 5.93 MPa 1.5 s

SUS316 1 150 °C 69% 6.21 MPa 0.0 s2 155 °C 72% 6.55 MPa 0.2 s3 160 °C 75% 6.89 MPa 0.4 s

392 D. Kim et al. / Powder Technology 239 (2013) 389–402

improved by the differential mold temperature. Since a highermold tem-perature caused a lower viscosity of the polymer melt, gas penetratedmore at the outer side of the tube in the curved section.

0

10

20

30

40

50

60

70

Gas

Pen

etra

tion

Dep

th (

mm

)

PP

Experiment

1 2 3 4 5

Fig. 6. Gas penetration depth fr

Table 5Process parameter rank of significance on penetration depth and RWT for PP and SUS316.

Materials

Penetration depth Simulation PP Portion (rank)SUS316

Experiment PPSUS316

RWT Simulation PPSUS316

Experiment PPSUS316

1.3. Gas-assisted powder injection molding process

GAPIM is a combination of GAIM and PIM with the advantages iden-tified in Fig. 3. While the PIM process has a high economic efficiency inthe manufacturing of small components, the efficiency decreases as thecomponent size increases due to high powder cost and long debindingtimes. With the application of GAIM technology to the PIM process, ma-terial cost can be lowered and the time for debinding can be also reduced.Additionally, a high aspect ratio and superior part surface quality can beobtained. Qingfa reports that the lower material consumption and theability to apply the technology to large components with thick wall sec-tions are great advantages of GAPIM [5].

Michaeli and Hopmann [6] performed GAPIM experiments using acurved spiral mold cavity and alumina powder feedstock. The thermalconductivity of the material was about five times greater than unfilledpolypropylene, which implied the availability of shorter gas delay timesdue to the faster solidification of the feedstock during the mold filling.The work focused on the wall thickness and its distribution along themelt-flow path in order to investigate the influence of the processingconditions, including gas pressure, gas holding time, shot size, and gasdelay time. Among the parameters, the gas delay time was the most sig-nificant factor for RWT, and the RWT increased according to the increaseof gas delay time but in a limited range. Additionally, the melt tempera-ture and mold temperature were considered influencing the RWT alsobut in a lesser degree.When the difference between themelt temperatureandmold temperature was high, themelt front solidifies fast with a thick

SUS316

al Trial Number

6 7 8 9

Optimal

om the simulation results.

Melt temperature Shot size Gas pressure Gas delay

9% (3) 57% (1) 6% (4) 28% (2)2% (4) 87% (1) 3% (3) 8% (2)

14% (3) 45% (1) 11% (4) 30% (2)24% (2) 23% (3) 17% (4) 35% (1)17% (4) 20% (3) 32% (1) 31% (2)6% (4) 19% (3) 36% (2) 39% (1)

47% (1) 16% (3) 13% (4) 24% (2)22% (2) 6% (4) 8% (3) 64% (1)

33

33.5

34

34.5

35

35.5

low med. high

S/N

Rat

io (

dB)

Setting

Melt Temp. Shot Size Gas Pressure Gas Delay

Fig. 7. S/N ratio values shown in relation to process settings for simulation of GAIM with PP.

393D. Kim et al. / Powder Technology 239 (2013) 389–402

frozen layer. On the other hand, when the difference between the tem-peratures was low, the feedstock froze slowly and the gas bubble pushedmore melt into the cavity, resulting in decreased wall thickness. Therewere less significant effects of other parameters on the wall thickness.

Recently, Lee et al. [7] studied the effects of parameters such as melttemperature, shot size, gas pressure, and gas delay time on the gas pen-etration depth andRWT in the GAPIMexperiment inwhich an SLA cavitywas introduced. The Taguchi L9 array based on Design of Experiments(DOE) was used in simulation runs with the AMI software (AutodeskMoldflow Insight, USA), and experiments for GAIM with PP and forGAPIM with SUS316. Despite the simulation prediction that shot sizewas the only significant parameter on gas penetration depth, in GAIMand GAPIM, he found that the shot size and gas delay timewere equally

33

33.5

34

34.5

35

35.5

36

low m

S/N

Rat

io (

dB)

S

Melt Temp. Shot Size

Fig. 8. Main effect plots for simulation on g

significant parameters in the GAIM experiment and the shot size, gaspressure, melt temperature were equally significant parameters inGAPIM experiment. The effects of parameters on RWT were insignifi-cant in the GAPIM experiment using the SLA cavity, and it indicatedthat the RWT was not in control with the SLA cavity.

2. Experimental procedures

Polypropylene (PP) random copolymer, 13T10Acs279 from FlintHills Resources (Odessa, TX), was used in the GAIM experiment. ThePP is one of the most widely used crystalline polymers and tends tohave higher shrinkage rate than amorphous polymers [30]. The PP hasexcellent impact resistance, flexural modulus, and clarity which allows

ed. high

etting

Gas Pressre Gas Delay

as penetration of GAPIM with SUS316.

Fig. 9. Gas penetration forMoldflowsimulations of (a)GAIMand (b)GAPIMunder optimumconditions.

394 D. Kim et al. / Powder Technology 239 (2013) 389–402

the easy observation of gas penetration. It is commercially applied forthe wide ranges of products from lab ware and food packaging to auto-motive applications. The material selected for the GAPIM experimentwas stainless steel powder SUS316L (CetaTech, South Korea) combinedwith a wax-polypropylene based binder and a powder loading of59 vol.%. Various medical parts and cellular phone parts are producedwith the 316L stainless steel powder feedstock (SUS316). Table 1shows the properties of PP and SUS316 used in this research. The simu-lation runs were also conducted based on these input data.

The part geometry shown in Fig. 4c and the cavities were built byCNC tooling with Al, as shown in Fig. 5.The mold inserts were mountedinto a quick change Master Unit Die consisting of an 84/90 ALU 210mold frame. Injection molding was conducted through a 30 ton Boy30M injection-molding machine. The unit has a maximum stroke of95 mm with a maximum shot capacity of 37 g and a screw diameterof 28 mm. The fixed processing conditions for the experiment are listed

Gas

Pen

etra

tion

Dep

th (

mm

)

Experiment

1 2 3 4 5

0

10

20

30

40

50

60

70PP

Fig. 10. Gas penetration depth fro

in Table 2. Higher injection pressure was required in this experimentcompared to Lee's [7] due to the faster cooling on the surface of the Almold. The faster cooling caused solidification of layer to be built tooearly and blocked the melt flow into the cavity.

A nitrogen supply for GAIM was obtained through a nitrogen gener-ator fromGain Technologies (GT-N2GA). Membrane separation technol-ogy separates compressed air into streams of 99.5% nitrogen and mixedoxygenwith carbon dioxide traces. A gas control system fromHEA Inter-national was used.

A K-Type Omega TT-J-30-SLE thermocouple was installed 2 mmbelow the surface of the part cavity used to measure themold tempera-ture. Real time temperature of the mold was recorded using a NationalInstruments data acquisition board. The temperaturemonitoring systemwas calibrated with Omega HH21A temperature meter with 0.5 °C res-olution. Mold temperature was maintained at 30 °C during the DOE(later the mold temperatures were varied from 30 °C to 60 °C to studythe effect of mold temperature over RWT). The mold temperature wasmonitored by the installed thermocouple and the cooling time was setto maintain constant mold temperature from shot to shot.

2.1. Design of Experiments

In the injection molding process, there are many parameters thathave an effect on the properties of the injected parts. In the GAIM pro-cess, additional parameters associatedwith gas control such as gas pres-sure, gas delay time, and gas inject timehave to be considered since theyhave a significant effect on the properties of the injected parts. Accord-ingly, it is helpful to use DOE methods such as partial/full factorial de-sign, or Taguchi approach to reduce experimental cost and time for theproduction. The Taguchi method, which is a “robust design method”,has been widely used in industry and research [11].

The processing parameters under investigation were: melt temper-ature, shot size, gas pressure, and gas delay time, as these are shown tobe the most significant parameters for GAIM [13,23,24]. The processingwindows for Taguchi approach were determined after preliminarymolding experiments. Low, medium, and high values that were chosenwithin the processingwindows are shown in Table 3. All processing var-iables used were the same as previous experiments [7] except injection

al Trial Number

6 7 8 9

Optimal

SUS316

m the experimental results.

35.1

35.2

35.3

35.4

35.5

35.6

35.7

35.8

35.9

36

low med. high

S/N

Rat

io (

dB)

Setting

Melt Temp. Shot Size Gas Pressure Gas Delay

Fig. 11. Main effect plots for experiment on gas penetration of GAIM with PP.

395D. Kim et al. / Powder Technology 239 (2013) 389–402

pressure. Process parameter variations were done through a DOE ap-proach in order to reduce the number of experiments while maintainingreliability. In this study a 34 factor L9 orthogonal array was adopted asshown in Table 3. From the results of the molding window study thevalues of the three levels of the processing parameters of the L9 orthog-onal array for both GAIM with PP and GAPIM with SUS316 were chosenas shown in Table 4. This processing window has been used for the ninesimulation runs and experiments.

In this study, our target function for optimization is to obtain maxi-mum gas penetration depth and RWT. The statistic Signal to Noise ratio(S/N ratio) is the ratio of the power of the signal to the power of thenoise. The larger-the-better S/N ratio calculation shown below wasused to achieve maximum gas permeation and RWT:

S=N ¼ −10 log1n

∑ 1y2i

ð1Þ

30

31

32

33

34

35

36

low

S/N

Rat

io (

dB)

Se

Melt Temp. Shot Size

Fig. 12. Main effect plots for experiment on

where n is the number of data points and yi is themeasured data. The S/Nratio is used to identify the parameter values for the optimal result.

2.2. Simulation studies

The Autodesk® Simulation Moldflow software was employed tounderstand the effects of processing parameters on gas penetrationdepth and RWT through simulation runs for both PP and SUS316. Thegas penetration depth and RWT values were obtained through the mea-surement tool in the Moldflow software, and were compared to the ex-perimental values. The material properties of PP used in this experimentwere readily available in the Moldflow material database. The viscosityand pressure–volume–temperature (pvT) data for SUS316were importedfrom the database file provided by CetaTech.

The properties of Al used for the mold insert were given in theMoldflow material database. The process parameter settings used arelisted in Table 4. From the results of the gas penetration and RWT, S/N

med. high

tting

Gas Pressure Gas Delay

gas penetration of GAPIM with SUS316.

Fig. 13. PP sample fabricated under the optimum processing conditions showing the gaspenetration.

Fig. 15. RWT measurement cross sections from (1) to (3).

396 D. Kim et al. / Powder Technology 239 (2013) 389–402

ratio analysiswas conducted to find the effects of the processing param-eters on the gas penetration depth and RWT to obtain the optimumprocess values. Additional simulation runswere conducted to investigatethe effect of mold temperature on gas penetration depth and RWT. Themold temperature was varied from 30 °C to 60 °C in the recommendedrange of mold temperature for SUS316.

2.3. Experiment

In order tomaintain a constant temperature for the cavity surface andmold, 60 s of cooling time was used. The experimental trials were ran-domized to minimize the noise. After the process was stabilized, thefirst five samples were discarded, and then ten good samples were col-lected from each trial. Five samples were used to measure the gas pene-tration depth and the other five sampleswere used tomeasure the RWT.The fivemolded parts for gas penetration depthmeasurements were cutin halves along parting lines to make the gas penetration visible, and theother five parts for the RWT measurements were also cut in four dif-ferent locations. Measurements were taken with a Vernier caliper. Themeasured gas penetration depth and RWT were then used to conductthe analysis based on a DOE orthogonal array. The same methodologiesused in the simulation studies were used for further analysis.

Fig. 14. SUS316 sample fabricated under the optimumprocessing conditions showing thegas penetration.

3. Results and discussions

The Al cavity insert, was used for both simulation runs and experi-ments in comparison to the previous work [7]. Due to the thermal con-ductivity of the Al cavity insert, mold temperature control was possible.The simulation runs and experiments for GAIM and GAPIM were con-ducted based on the processing parameter settings from Table 4 inorder to find the effects of the parameters on gas penetration depthand, the optimum processing conditions for gas penetration. The effectsof the processing parameters on RWT were then considered. Addition-ally, since mold temperature was a significant parameter in previousstudies [25], it was also varied in simulation runs and molding experi-ments in this work.

3.1. Effects of processing parameters on gas penetration

3.1.1. Simulation resultsThe gas penetration depth for each simulation was measured using

the measurement tool in the software, see Fig. 6 for GAIM and GAPIMrespectively. Trials 1, 4 and 7, which had the smallest shot size, havethe greatest penetration depth, this indicates that shot size is an impor-tant parameter for determining gas penetration depth.

S/N ratios were calculated from the penetration depth values, Fig. 6,and were used to get the average S/N ratios over three levels pertainingto low, medium, and, high parameter settings. Delta, the differencebetween themaximumandminimumS/N value, portions and parameterrankingswere then calculated as shown in Table 5. The significance of theeffect on gas penetration depthwas identified by the ranked portions. Ac-cordingly, the shot size was the most significant parameter with 57% forGAIM and 87% in GAPIM, and is in agreement with previous studies[13,17,23]. Gas delay time was a significant parameter in GAIM at 28%,but showed much lower significance at 8% in GAPIM. This difference is

0

0.2

0.4

0.6

0.8

1

1.2

1 2 3 4 5 6 7 8 9

RW

T (

mm

)

Simulation Trial Numbers

GAIM GAPIM

Fig. 16. RWT measurement from the simulation results.

-1.6

-1.4

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0

low med. high

S/N

Rat

io (

dB)

Setting

Melt Temp. Shot Size Gas Pressure Gas Delay

Fig. 17. Main effect plots for simulation on RWT of GAIM with PP.

397D. Kim et al. / Powder Technology 239 (2013) 389–402

mainly due to higher thermal diffusivity of the SUS316 compared to PP,which may be a result of fast cooling of injected SUS316.

Figs. 7 and 8 show the main effect plots from Table 5. Since the tar-get function is tomaximize gas penetration depth, the highest S/N ratiosin the plots will offer the optimum process settings for each processingparameter. From Fig. 7 for GAIM the optimum penetration depth can beobtained when the melt temperature is low and shot size is low, whilechanges in gas pressure and gas delay time were not significant. ForGAPIM, Fig. 8, the optimum condition for penetration depth was lowshot size, the other parameters were insignificant. There was no differ-ence between the main effect plot using the SLA cavity [7] and the Alcavity in the simulation runs of GAIM and GAPIM. The gas penetrationdepth of the parts processed under the optimum simulation conditionsis included in Fig. 6, and the part pictures are shown in Fig. 9. The shadedregions inside the parts represent the volume of the penetrated gasbubble.

3.1.2. Experimental resultsThe gas penetration depths of the PP parts were measured by dying

the hollow internal core of the parts with blue ink. The gas penetration

-5.1

-4.9

-4.7

-4.5

-4.3

-4.1

-3.9

-3.7

-3.5

low m

S/N

Rat

io (

dB)

Se

Melt Temp. Shot Size

Fig. 18. Main effect plots for simulatio

depths of the SUS316 parts were measured after cutting along the part-ing lines. The average gas penetration depth of all GAPIM samples ispresented in Fig. 10. We can observe that gas penetration depth ismore sensitive in GAPIM than GAIM.

The processing parameters are ranked by significance in Table 5, cal-culated from the depth values of Fig. 10. It indicates that shot size is themost significant parameterwith a portion of 45% and the gas delay timeis also a significant parameter with a portion of 30% in GAIM. This resultagrees with the simulation results for PP and is also in agreement withParvez et al.'s work [13]. The result of previous experiments using anSLA cavity by Lee [7] showed gas delay time with a portion of 38% andshot size with a portion of 37%, both parameters have shown to besignificant using an Al mold insert [26,27]. Gas pressure is the least sig-nificant parameter in both simulation and the experiment for PP andagrees with previous experiments [13,23]. For the GAPIM experimentusing an Al cavity, the gas delay time is the most significant parameterwith a portion of 35% and gas pressure is the least significant parameterwith a portion of 17%. This result is different from the above simulationresults and previous experiments using an SLA cavity that the gas pres-sure was the most significant parameter with a portion of 33% and the

ed. high

tting

Gas Press. Gas Delay

n on RWT of GAPIM with SUS316.

00.20.40.60.8

11.21.41.61.8

2

1 2 3 4 5 6 7 8 9

RW

T (

mm

)

Experiment Trial Numbers

GAIM GAPIM

Fig. 19. RWT from the experimental results of GAIM with PP and GAPIM with SUS316.

398 D. Kim et al. / Powder Technology 239 (2013) 389–402

delay time was the least significant parameter with a portion of 13%.The difference implies that the cooling effect is more critical in theGAPIM experiment using an Al cavity than using an SLA cavity. Further-more, the simulation results showed that the shot size was the mostsignificant parameter and the simulation was not able to predict the ef-fects of the other processing parameters associated with gas control onthe gas penetration depth in the GAPIM using SUS316.

The main effect plots in Fig. 11 show that optimum gas penetrationdepth can be obtained when melt temperature is low, shot size is low,gas pressure is high and gas delay time is high. Fig. 12 shows that opti-mumgas penetration depth for GAPIM can be achievedwhenmelt tem-perature is low, shot size is low, gas pressure is low and gas delay time ismedium.

Figs. 13 and 14 show the PP and SUS316 parts produced under theoptimum conditions using an Al cavity. The hollow core of the PP sam-ple was dyed with ink to make it visible.

3.2. Effects of processing parameters on RWT

Fig. 15 shows three RWTmeasurement locations on the part and thecross section at each location. RWT at each location was obtained by

0.7

0.9

1.1

1.3

1.5

1.7

1.9

2.1

low

S/N

Rat

io (

dB)

S

Melt Temp. Shot Size

Fig. 20. Main effect plots for experim

averaging wall thickness at the four points on the cross section. The ef-fects of processing parameters on the RWT were analyzed by means ofthe main effect plots and portion ranks. The target function is to maxi-mize RWT, and the highest S/N ratio in the plot will offer the optimumprocessing setting levels for each of the four processing parameters. Inthe simulation and experiment, only Section 1 was considered in theanalysis because gas penetration depths were not long enough for theRWTmeasurement in some trials. The other sectionswill be consideredlater in this chapter.

3.2.1. Simulation resultThe RWT of Section 1 for each trial was obtained from simulation

runs. Fig. 16 shows the RWT for simulations of GAIM andGAPIM, indicat-ing that overall RWT is higher in GAIM than in GAPIM.

Table 5 shows the parameter rankings of RWT for each simulation. Itshows that gas pressure and gas delay time are the most significant withportions of 32% and 31% in GAIM, respectively. It also indicates that gasdelay time and gas pressure are the most significant parameters withportion of 39% and 36% in GAPIM, respectively. Since the values of RWTin simulations were very low, S/N ratios were also very low therefore,the effects of the parameters on RWT cannot be estimated reliably.

Figs. 17 and 18 show the main effect plots from Table 5, respective-ly. For GAIM, low melt temperature, medium shot size, medium gaspressure, and high gas delay time yield the highest RWT. For GAPIM,medium melt temperature, low shot size, low gas pressure, and medi-um gas delay time result in the highest RWT.

3.2.2. Experimental resultFig. 19 shows the RWT for the GAIM and GAPIM experiments. Over-

all RWT for GAPIM is higher than for GAIM. This is due to faster coolingof SUS316 than PP, resulting in a thicker frozen layer. Refer to Table 5,the melt temperature is the most significant parameter with a portionof 47%, the gas delay time is also significant with a portion of 24%. Com-pared to the results of gas penetration depth, the significance of melttemperature increased, while the significance of shot size decreased. Itindicates that themelt flowproperties and buildup of frozen layer affectRWTmore than the shot size. Table 5 shows that the gas delay time hasthe most significant effect on RWT with a portion of 64%, and the melttemperature has a portion of 22%. The gas delay time, which was the

med. high

etting

Gas Pressure Gas Delay

ent on RWT of GAIM with PP.

2.7

2.9

3.1

3.3

3.5

3.7

3.9

4.1

4.3

4.5

low med. high

S/N

Rat

io (

dB)

Setting

Melt Temp. Shot Size Gas Pressure Gas Delay

Fig. 21. Main effect plots for experiment on RWT of GAPIM with SUS316.

399D. Kim et al. / Powder Technology 239 (2013) 389–402

most significant parameter on gas penetration depth, still played animportant role in RWT. The greater significance of gas delay time forSUS316 over PP is due to the thicker frozen layer.

In GAIMwith PP, Fig. 20, lowmelt temperature and high delay timeyield the highest RWT. In GAPIM with SUS316, Fig. 21, high gas delaytime results in a significantly larger RWT. This is mainly due to thethickness of the frozen layer increase with the gas delay time increase.This result also agrees with previous work [23,26,28] where it wasfound that the high thermal diffusivity of the feedstock leads to fastersolidification in the cavity during the gas delay period, although therewere no significant effects of shot size and gas pressure on the RWT.

Fig. 22 shows the cross section of themolded part from the results ofthe Moldflow simulation. The temperature changes in the part, and thetemperature difference between PP and SUS316 can be observed. Thesetemperature profiles represent 8.01 s inGAIM and6.45 s inGAPIM afterinjection started. It seems that SUS316 solidifies much faster than PPwhich explains the delay time and its importance in GAPIM than inGAIM. The simulation results using Al cavity and SLA cavity (Lee et. al.

a) PP

Fig. 22. Temperature distribution of melt c

[7]) showed no differences, this indicates Moldflow gas-assisted mod-ule is not able to consider difference between SLA andAl cavities lackingin coupled thermal analysis capability.

3.3. Effect of mold temperature on gas penetration depth and RWT

Fig. 23 shows the RWT at each section for simulation runs of GAPIMat variousmold temperatures and indicates that RWT is not controllableby changing mold temperature in simulation runs.

Fig. 24 shows the effect of mold temperature on the gas penetrationdepth in the experiment and that the gas penetration depth decreasesas the mold temperature increases. This result agrees with Chau'sstudy with plastics [29] and Michaeli's study with powder feedstock[6]. According to the previous studies, the RWT decreases as the moldtemperature increases due to the decreased solid layer during the in-jection stage, caused by the lower viscosity and slower cooling of themelt. As the frozen layer decreases, more melt of material is available

b) SUS316

ores after completion of filling stage.

0

10

20

30

40

50

60

70

30 35 40 45 50 55 60

Gas

Per

mea

tion

(m

m)

Mold Temperature

Fig. 24. Gas penetration depth versus mold temperature for experiment of GAPIM with SUS316.

0.45

0.5

0.55

0.6

0.65

0.7

0.75

0.8

1 2 3

RW

T (

mm

)

Measurement Section

30 C 35 C 40 C 45 C 50 C 55 C 60 C

Fig. 23. WT at each section for simulation of GAPIM with SUS316.

0.9

1

1.1

1.2

1.3

1.4

1.5

1.6

1.7

1 2 3

RW

T (

mm

)

Measurement Section

30 35 40 45 50 55 60

Fig. 25. RWT at each section for experiment of GAPIM with SUS316.

400 D. Kim et al. / Powder Technology 239 (2013) 389–402

Fig. 26. Samples fabricated under different mold temperatures.

401D. Kim et al. / Powder Technology 239 (2013) 389–402

tofill out the remaining volume of the cavity, and this leads to the lowergas penetration depth [25,29].

Fig. 25 shows the effect of the mold temperature on RWT at eachsection and it also shows that the RWT decreases as the mold tempera-ture increases. When the mold temperature changed from 30 to 60 °C,the RWT changed from 1.61 to 1.53 mm in Section 1, and from 1.54to 1.43 mm in Section 2. Since the gas penetration depth affected theRWT measurement for Section 3, the effect of mold temperature onRWT was insignificant. Fig. 26 shows the samples fabricated under thevarious mold temperatures in the GAPIM experiment using SUS316.All parts show the required gas penetration depth with optimum sur-face finish. The sectioned samples show some porosity, possibly dueto gas penetration during the gas injection molding stage with pow-der–binder separation which requires further study in the future.

4. Conclusions

While the simulation from Moldflow effectively predicted the gaspenetration patterns in gas-assisted injection molding (GAIM) withpolypropylene (PP), it was not sensitive to the gas pressure and gasdelay time in gas-assisted powder injection molding (GAPIM) withstainless steel powder feedstock (SUS316). The Moldflow simulationalso failed to consider the effect of thermal conductivity differencesbetween Stereolithography (SLA) and Al cavities.

In the GAIMof PP, therewas no significant difference of gas penetra-tion depth between the SLA cavity and Al cavity. However, in theGAPIMexperiment with SUS316, the significance of gas delay time was higherin the Al cavity than in the SLA cavity. This is because the thermal con-ductivity of Al cavity (190 W/m-C) is higher than that of the SLA cavity(1.47 W/m-C), showing that cooling is faster in the Al cavity. Generally,the gas channels were developed better with the Al than the SLA cavityinserts. The effect of gas delay time differed between GAIM and GAPIM,there was a greater significance in the GAPIM than in GAIM. This differ-ence is mainly due to the higher thermal diffusivity of SUS316 as com-pared to PP. The optimum gas penetration depth was obtained in theprocessing conditions identified by high gas delay time, low melt tem-perature, and low shot size. Gas pressurewas an insignificant parameterin all simulations and experiments.

The effect of processing conditions on RWTwas also considered. Forthe GAPIM experiment, gas delay timewas themost significant param-eter followed by melt temperature. The significance of gas delay timeshows that cooling from the surface of the cavity is more sensitive toRWT. As the gas delay time increases more time for heat transfer is pro-vided for solidification of themelt on the cavity wall resulting in thickerRWT.While the RWTwas not well controlled in the GAPIM experiment,RWT was thicker and more uniform with an Al cavity than as reportedin the SLA cavity study [7].

An increase in mold temperature yields a decrease in gas penetra-tion depth and RWT. When the mold temperature is low, the viscosityof themeltwill increase. The higher viscosity causesmore accumulationof melt on the cavity wall and less melt flows from the melt layer todownstream of the gas front giving low resistance to gas penetration.The accumulation also leads to a thicker frozen layer in the cavity wallresulting in high RWT. This phenomenon is observed in the GAPIM ex-periment with SUS316 and is the same as that of plastics.

Acknowledgment

This work was supported by the second phase of the Brain Korea21 Program and Pusan National University Research Grant, 2011.

References

[1] R.M. German, A. Bose, Injection Molding of Metal and Ceramics, Metal PowderIndustries Federation, Princeton, NJ, 1997.

[2] M. Bloemacher, D. Weinand, Catamold™: a new direction for powder injectionmolding, Journal of Materials Processing Technology 63 (1–3) (1997) 918–922.

[3] P.J. Vervoort, R. Vetter, J. Duszczyk, Overview of powder injection molding,Advanced Performance Materials 3 (2) (1996) 121–151.

[4] R.A. Malloy, Plastic Part Design for Injection Molding: An Introduction, Hanser,Munich, 1994.

[5] L. Qingfa, Gas-assisted PIM, SIMTech Technical Report, 2000.[6] W. Michaeli, C. Hopmann, New perspectives for ceramic injection molding with

gas injection, Advanced Engineering Materials 2 (12) (2000) 827–832.[7] K. Lee, M. De Hoyos, S. Ahn, R. Nambiar, M.A. Gonzalez, S.J. Park, R.M. German,

Gas-assisted powder injectionmolding: a study on the effect of processing variableson gas penetration, Powder Technology 200 (3) (2010) 128–135.

[8] H.P. Heim, H. Potente, SpecializedMolding Techniques, William Andrew Publishing,Norwich, NY, 2001, pp. 73–78.

[9] R.V. Nambiar, K.H. Lee, D. Nagarajan, Stereolithography mold life extension usinggas-assisted injection, Rapid Prototyping Journal 13 (2) (2007) 92–98.

[10] C. Karatas, A. Sozen, E. Arcaklioglu, S. Erguney, Investigation of moldability forfeedstocks used powder injection molding, Materials and Design 29 (9) (2008)1713–1724.

[11] R. Urval, S. Lee, S.V. Atre, S.-J. Park, R.M. German, Optimisation of process conditions inpowder injectionmoldingofmicrosystemcomponents using a robust designmethod:part I. primary design parameters, Powder Metallurgy 51 (2) (2008) 133–142.

[12] R. Ruprecht, T. Gietzelt, K. Muller, V. Piotter, J. Haubelt, Injection molding ofmicrostructured components from plastics, metals and ceramics, MicrosystemTechnologies 8 (4–5) (2002) 351–358.

[13] M.A. Parvez, N.S. Ong, Y.C. Lam, S.B. Tor, Gas-assisted injection molding: the ef-fects of process variables and gas channel geometry, Journal of Materials Process-ing Technology 21 (1) (2002) 27–35.

[14] F. Ilinca, J.F. Hetu, Three-dimensional simulation of multi-material injectionmolding: application to gas-assisted and co-injection molding, Polymer Engineer-ing and Science 43 (7) (2003) 1415–1427.

[15] Q. Li, X.Wang, C. Shen, J. Jiang, S. Hou, Influence of processing parameters on shrink-age of LDPE parts in gas-assisted injection molding, ANTEC, 2007, pp. 625–629.

[16] A. Polynkin, J.F.T. Pittman, J. Sienz, Approximate prediction of gas core geometryin gas assisted injection molding using a short cut method, Polymer Engineeringand Science 47 (5) (2007) 713–720.

[17] Lee, K. H. A study of fingering phenomena in the gas-assisted injection moldingprocess: the effects of processing variables and the fluidity ratio. PhD. Disserta-tion, U. Mass., Lowell.

402 D. Kim et al. / Powder Technology 239 (2013) 389–402

[18] J. Avery, Gas-assist Injection Molding: Principles and Applications, Hanser, Munich,2001.

[19] A.J. Poslinski, P.R. Oehler, V.K. Stokes, Isothermal gas-assisted displacement ofviscoplastic liquids in tubes, Polymer Engineering and Science 35 (11) (1995)877–892.

[20] S.C. Chen, K.S. Hsu, J.S. Huang, Experimental study on gas penetration characteristicsin a spiral tube during gas-assisted injection molding, Industrial and EngineeringChemistry Research 34 (1) (1995) 416–420.

[21] S.Y. Yang, H.L. Chou, Study on the residual wall thickness at dimensional transitionsand curved sections in gas-assisted molded circular tubes, Polymer Engineering andScience 42 (1) (2002) 111–119.

[22] K.Y. Lin, F.A. Chang, S.J. Liu, Using differential mold temperatures to improve the re-sidual wall thickness uniformity around curved sections of fluid assisted injectionmolded tubes, International Communications in Heat and Mass Transfer 36 (5)(2009) 491–497.

[23] N.S. Ong, H.L. Lee, M.A. Parvez, Influence of processing conditions and part design onthe gas-assisted injection molding process, Advances in Polymer Technology 20 (4)(2001) 270–280.

[24] S.Y. Yang, C.T. Lin, J.H. Chang, Secondary gas penetrations in ribs during full-shotgas-assisted injection molding, Advances in Polymer Technology 22 (3) (2003)225–237.

[25] Gas-injection molding with DuPont engineering polymer. DuPont Technical ReportTRG 3060 http://www2.dupont.com/Plastics/en_US/assets/downloads/processing/GasInj_e.pdf.

[26] A.J. Carrillo, A.I. Isayev, Birefringence in gas-assisted tubular injection moldings:simulation and experiment, Polymer Engineering and Science 49 (12) (2009)2350–2373.

[27] W. Van Hoye, K. Callewaert, R. Hick, R. Vermeersch, E. Beeckman, C. Emmerechts, E.Casteleyn, Fabrication of Thin Cavities with Gas Assisted Injection Molding: Experi-ments and Simulations, 2005, pp. 1–8.

[28] A. Polynkin, J.F.T. Pittman, J. Sienz, Gas assisted injection molding of a handle:three-dimensional simulation and experimental verification, Polymer Engineeringand Science 45 (8) (2005) 1049–1058.

[29] S.W. Chau, Three-dimensional simulation of primary and secondary penetrationin a clip-shaped square tube during a gas-assisted injection molding process,Polymer Engineering and Science 48 (9) (2008) 1801–1814.