DENSIFICATION OF SELECTIVE LASER SINTERED METAL PARTS BY HOT ISOSTATIC PRESSING Mukesh K. Agarwala, and David L. Bourell, Center for Materials Siene and Engineering, Joseph J. Beaman, Department ofmehanial Engineering, The University of Texas at Austin, Austin, TX 78712. ABSTRACT Metal matrix alloy omposite parts were made from powders by Seletive Laser Sintering (SLS). In this study, partially dense (60%-80%) metal parts made by SLS were densified to full density (>98%) by hot isostati pressing (HIPping) without any loss of shape. HIPping was done by vauum sealing SLS samples in glass apsules. HIPping parameters, suh as, temperature, pressure, and time, were studied with respet to density, linear shrinkage, and mirostrutures. Anisotropy in linear shrinkage was orrelated to the SLS proessing parameters. Densifiation resulting from HIPping was orrelated to mirostrutures and theoretial HIP densifiation maps. A detailed analysis of suh maps is presented. INTRODUCTION The ultimate aim of rapid manufaturing proesses is to fabriate three-dimensional fully funtional parts diretly from metals and erami materials, without the use of any intermediate binders or any other materials whih may require additional proessing steps before or after the rapid prototyping operation. Seletive Laser Sintering is one of the few rapid manufaturing proesses whih possesses the apability of produing suh struturally sound parts diretly from metals and eramis [1-4]. Feasibility of produing metal parts diretly by SLS has been demonstrated using various metal systems [5-12]. However, SLS parts of metals and eramis are not fully dense and hene require postproessing to obtain nearfull density. Typially, infiltration and onventional sintering, suh as solid state or liquid phase sintering have been employed so far as post-proessing steps for SLS parts [1-4]. These post-proessing tehniques produe near-full density parts, but still leave behind losed pores and result in anisotropi and unpreditable shrinkage during densifiation. In this study, SLS metal parts have been post proessed to full density by Hot Isostati Pressing (HIPping). Hot isostati pressing is a materials proessing tehnique in whih high isostati pressure is applied to a powder part or ompat preform at elevated temperatures to produe partile bonding [13-15]. This proess usually results in a fully dense body, although partially dense bodies an also be intentionally produed. In hot isostati pressing, the ontat between the partiles is inreased due to deformation by the external fores and the thermal mobility of the atoms. Therefore, with HIPping it is possible to produe a part with a density approahing the theoretial density and with properties of bulk: metal. Based on ertain onstitutive equations [16-18], the exat mehanism driving the densifiation proess an easily be determined during HIPping at a given set of proess parameters. Using these equations, densifiation maps are onstruted for various material systems. From suh maps, shown later for single phase pre-alloyed bronze powders, it an be easily seen that at low temperatures and short time, yielding is the dominant mehanism by whih densifiation takes plae. At high temperatures and long times, power-law reep is the dominant mehanism by whih densifiation takes plae. Although suh maps existfor various single phase metal and erami systems, densifiation maps for multi-phase systems have rarely been studied. 65
Also, suh maps do not take into aount the effet of phase transformations on the densifiation proess. As the pressing temperatures inreases, the amount of pressure neessary for ompation dereases. Typially, HIPping temperatures range from 450 C for aluminum alloys to 1700 C for tungsten. Pressures are usually applied using argon gas as the pressure transmitting medium and ~IPping pr~ssures r~nge from 20MPa to 300MPa: The duration of the pressing yle is also Important, SIne at high temperatures lengthy pressing leads to reep. The longer the pressing is done, the greater the density and the better the properties of the pressing at a given temperature and pressure. Lengthy pressing also brings about more omplete redution of oxides, softening and rerystallization, whih lead to an improvement in the plasti properties of the part. In addition to temperature, pressure, and time, partile harateristis suh as partile size also play an important role in densifiation during HIPping. In this study, HIPping was explored as a post proessing route for SLS metal parts. SLS bronze-nikel parts, previously used to study the strutural integrity and effet of post proessing by onventional liquid phase sintering [10-12], were used for HIPping. SLS parts typially have surfae onneted and interonneted pores. Therefore, suh parts require the surfae to be sealed or the part to be enapsulated suh that the pressure transmitting medium, argon gas, does not flow into the pores in the part. Presene of high pressure gas in the interonneted pores will prevent them from "losing" during the HIP proessing. To seal suh parts, the parts are enapsulated in an envelope suh that the "empty spae" in the pores is "squeezed" out during HIPping and the pores lose to form a fully dense part. The basi requirements for the enapsulation material are that it should be relatively strong, gas tight, inert and plasti under the applied temperature and pressure onditions, ompatible with the material to be densified so as to minimize diffusion reations and easily removable. Several materials and tehniques are used for enapsulation of powders and ompat preforms for HIP proessing [13-15]. In this study, the SLS bronze-nikel parts for HIPping were ontained in a vauum-sealed glass apsule. Glass enapsulation has so far been used only on experimental basis. Commerially, sheet metal ontainers or erami ontainers are ommonly used. Glass enapsulation was used instead of sheet metal ontainers or erami ontainers beause of the ease with whih glass an be shaped to take the form of SLS parts irrespetive of the omplex nature of the part. Unlike sheet metal and eramis, glass beomes vitreous at high temperatures and hene an flow easily and onform to the shape of the preform or in this study, the SLS part. EXPERIMENTAL PROCEDURE GLASS ENCAPSULATION OF SLS BRONZE-NICKEL PARTS SLS bronze-nikel samples of -70% density were ut from the shoulder setion of broken tension test oupons, previously tension tested [lo-12], into small setions of 1.5m x 1.5m x O.5m dimensions. The samples were degassed by annealing them at 450 C in for 2 hours. Commerial borosiliate glass tubes of O.D. 19 mm and LD. 15.5 mm were used as the ontainer material. The tube and the sample were thoroughly dried in an oven at IS0 C. One end of the glass tube was sealed by a onventional glass blowing method using an oxy-aetylene torh. The SLS sample was then plaed in the tube and a small onstrition was made in the tube, -2-4 em above the sample, by the usual glass blowing tehnique using a torh. Figures la and IB show shematially the stages involved in vauum sealed glass enapsulation of the SLS parts. The onstrition was as small as possible. 66
The glass tube with the sample and the onstrition was then evauated using a diffusion pump whih was baked up by a mehanial pump, Figure lb. A liquid nitrogen old-trap was used in the evauation to ensure effiient evauation and prevent any ontamination of the diffusion pump. The glass tube with the sample was evauated to a vauum of -10-5 torr. One the desired vauum level was reahed, the onstrited part of the tube was sealed ompletely, using the torh, and removed from the remaining part of the tube. HOT ISOSTATIC PRESSING OF SLS BRONZE-NICKEL PARTS HIPping of the vauum-sealed glass enapsulated SLS bronze-nikel samples was done at three different temperatures of 750 C, 825 C, and 900 C with a pressure of 124MPa (l8ksi) for 1, 2 or 3 hours. Similarly, HIPping of SLS bronze-nikel samples of starting density 70% was done with three different pressures of 69MPa (loksi), 124MPa (l8ksi), and 180MPa (26ksi) at 825 C for 1, 2 or 3 hours. Sample dimensions and weight were measured before and after HIPping to determine densifiation. RESULTS AND DISCUSSION SLS bronze-nikel samples exhibited densifiation at the various HIPping temperatures, pressures and times used in this study. The extent of densifiation observed varied with these variables. As shown in Figures 2, density inreased with inreasing temperatures, pressures and times and approahed theoretial density at high values of these variables. Densifiation of SLS bronze-nikel parts during HIPping was aompanied with phase transformation between bronze and nikel phases, Figures 3 and 4. Bronze, predominantly opper (90 Wt.%), and nikel homogenize at high temperatures by interdiffusion between bronze and nikel to form a homogeneous solid solution. The degree of homogenization depends on temperature and time of sintering or HIPping. As evidened by (311) peak broadening of opper and nikel, shown in Figures 3, at low temperatures and short times only partial homogenization ours and omplete homogenization ours at high temperatures and long times. As a result of hemial homogenization, the bronze phase disappears ompletely leaving behind pores in its plae and an expanded solid solution of bronze in nikel. This happens due to a faster diffusion rate of opper into nikel than that of nikel into opper, whih results in the Kirkendall effet. Therefore, the pores reated at this stage are referred to as Kirkendall porosity. Due to the generation of Kirkendall pores and expansion of nikel phase, the samples may undergo growth or swelling instead of shrinkage, as has been observed in SLS bronze-nikel parts during onventional sintering below 1000 C [10-12]. Although, as shown in Figures 3 and 4, hemial homogenozation between bronze and nikel was observed in the SLS bronze-nikel parts during HIPping also, no overall expansion or growth of the SLS parts was observed during HIPping. Parts proessed at lower temperatures and pressures, whih exhibit low density, show partial homogenization with Kirkendall pores present at the boundaries of homogenized and nonhomogenized phases, Figure 4b. However, parts proessed at high temperatures and pressures, whih exhibit nearly full density, also show partial homogenization but absene of any Kirkendall pores, Figure 4. As a result of hemial homogenization between bronze and nikel, and generation of Kirkendall pores, the densifiation of the bronze-nikel samples is onsiderably slower than predited by theoretial densifiation maps for single phase pre-alloyed bronze. Figures 5 show the densifiation of bronze-nikel samples observed in this study relative to theoretial densifiation urve for pure bronze of 50Jlm partile size. The theoretial HIP densifiation map for bronze was 67
Figure 1: [A] A shemati representation of the glass enapsulation used in this study (a) glass tube with both ends open, (b) glass tube with one end sealed and the sample in the tube, () glass tube with the sample in it and a onstrition, and (d) vauum sealed glass apsule with the sample in it, and [B] A shemati representation of evauation proedure for glass enapsulation of SLS part. t 9S i- ~ At 124MPa ~ 90 l / t /.;" 8S l / / '" g; r / t / Q 80~' / / / 75~ /' /' 1 Hour :1 Hour 3 Hour A 100 [ 95 90 7S 70 At 825"C if I 1 Hour 70 w...-'-'-'-~...~'-'--'-~"'-'--'-""-'--'-~...~... 200 300 400 500 600 700 800 900 1000 Temperature ( C) 65 L-~~~~U-..--,--"-~...,,",----..~-'-' 1 10 100 Pressure (MPa) Figure 2: Density of SLS bronze-nikel parts as a funtion of (A) HIPping temperature and time at a onstant pressure of 124MPa, (B) HIPping pressure and time at a onstant temperature of 825 C. 68
SLS Bronze-Nikel Ni a 750 C, 124MPa, 1 Hour Cu Ni b Cu 85 87 89 91 93 95 85 87 89 91 93 95 900 C, 124MPa, 3 Hour Solid Solution Figure 3: XRD patterns of (311) peaks of Bronze and Nikel showing peak broadening due to homogenization (a) SLS Bronze-Nikel part showing eu and Ni as separate phases, (b) SLS Bronze-Nikel sample HIPped at 750 C for 1hour at 124MPa, showing partial homogenization, and () SLS Bronze-Nikel sample HIPped at 900 C, for 1hour at 124MPa, showing 85 87 89 91 93 95 omplete homogenization.
Figure 4: Optial mirographs of (a) 70% dense SLS bronze-nikel sample, (b) SLS sample HIPped at 825 C, 69MPa and 1 hour, and () SLS sample HIPped at 825 C, 69MPa and 2 hour. [B=Bronze, N=Nikel, S=Solid Solution, P=SLS Pores, K=Kirkendall Pores]
validated experimentally by HIPping of a single phase pre-alloyed bronze sample at 750 C at 69MPa for 1 hour whih yielded nearly full density in aordane with the map. However, the experimental data for SLS bronze-nikel parts are well to the right of the theoretial urves indiating a muh slower densifiation in the bronze-nikel samples. Therefore, densifiation is slowed down during HIPping of mixed phases, suh as bronze-nikel, whih form a solid solution at HIPping temperatures resulting in Kirkendall pores whih oppose the on-going densifiation. Theoretial HIP maps do not aount for suh phase transformations and their effet on densifiation during HIPping. Low HIPping temperatures and pressures and short time of HIPping ause nearly omplete removal of pores present due to SLS but do not result in removal of Kirkendall pores reated due to homogenization during HIpping. Therefore, suh proessing onditions whih do not remove the Kirkendall pores ompletely result in less than full density. However, high temperatures and pressures result in removal of both the pores due to SLS as well as the Kirkendall pores reated due to homogenization during HIPping. Net volume hanges (shrinkage) of SLS parts during HIPping was in aordane with the observed density hanges. However, the linear dimensional hanges,.1.l/lo, were not the same in all three diretions, as observed earlier in onventional liquid phase sintering of suh parts. Figure 6A shows the shrinkage in parts HIPped at 825 C and 69MPa for varying times. The linear dimensional shrinkage in these parts was anisotropi with least shrinkage in the san diretion and maximum shrinkage in the thikness or build-up diretion. Similar shrinkages have been observed in SLS parts when subjeted to onventional liquid phase sintering [10-12]. In either ase, maximum shrinkage in thikness diretion is observed due to lowest density along that diretion due to poor sintering between layers during SLS and minimum shrinkage in the san diretion due higher degree of sintering during SLS along the san diretion. However, the degree of anisotropi shrinkage observed during HIpping is lower than that observed during liquid phase sintering. Comparision of Figures 6A and 6B shows that the differene in shrinkage between thikness and san diretion is lower in HIpped parts. HIpping is done at temperatures well below the liquid phase sintering temperatures, therefore there is little or no liquid phase formed during HIPping. Also, HIPping is done under isostati pressures, therefore there is minimal effet of gravity aiding the densifiation or flow of liquid in the diretion of gravity whih oinides with the thikness or build-up diretion during both the HIPping as well as liquid phase sintering proesses. Therefore, the anisotropy observed in HIPping of SLS parts represents a more preise piture of density gradient present in parts due to SLS. CONCLUSIONS A framework for enapsulation of SLS parts for HIPping has been established using glass ontainers. Glass enapsulation is an easy, flexible, and eonomial method for enapsulating omplex shapes. Densifiation studies of SLS bronze-nikel parts by HIpping resulted in higher density parts with inreasing temperatures, pressures and times. However, the densifiation of SLS bronze-nikel parts was slower than that predited by theoretial HIP densifiation maps due to generation of Kirkendall pores arising from the hemial homogenization between bronze and nikel phases during HIPping. The anisotropy in linear dimensional hanges during HIPping is not influened by liquid phase or effet of gravity, thus giving a preise piture of anisotropy present in parts due to SLS. ACKNOWLEDGMENTS This study was made possible by a researh grant from State oftexas Advaned Researh Projet, ATP-116: Seletive Laser Sintering: DiretMetal Fabriation 71
0.95 A 0.95 Power Law Creep (II) Meh. Driven B -...J N 0.9 0;:;.; 8 '" 0.85 ;>.~ ~ 0.8 Power Law Creep (I) Meh I ~ONZE 0.75 1 Yielding!Part.Dia. =50 flm iext.p =124 MPa i InLP=0 MPa 0.7 1-1_-..._...,..--_-+-_-+-_-+- _ Temperature, C 81-800 0.9 or;; 8 '" 0.85.::: '"~ 0.8 i:: "U Yielding 0.75 BRONZE IPart.Dia. =50 flm, Temp = 825 C I Int.P =OMPa 0.7.==::;;,J.._-=====::::::..."--_...l_~-=-=~ Pressure, MPa Figure 5: Comparision of densifiation urves of SLS bronze-nikel samples (symbols) with the theoretial HIP densifiation urves for single phase pre-alloyed bronze, as a funtion of (A) HIPping temperature and time, and (8) HIPping pressure and time.,... ~ '-'... Oil CIS.=.::... VJ... CIS... 0-2 -4.:. -6-8 -10 :::s -12 ~ Transverse --e. Longltudnal - e--. Thikness 3 (A) 4 0-2,-... ~ '-' -4... Oil CIS.:. -6 "i:. VJ -8 Ek...... C!S '-0... - ] - ~-10...... ;::s "- E> _14 '---'~...L-,--,-~-,--"~--,-~-,--'--.L-J,---,"--,--'---L-J o 2 Time (Hour) -12-14 o '- <> 1 2 ~ Transverse (B) --e - Longltudnal - e-- - Thikness -B----EJ --- <>- - - - - - - - - - -E> 3 4 Time (Hour) Figure 6: Anisotropi linear dimensional shrinkages observed in SLS bronze-nikel parts subjeted to (A) HIPping at 825 C and 69MPa, and (B) Conventional liquid phase sintering at I030 C. s 6 7
1. 2. 3. 4. 5. 6 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 1 1990 Solid Freeform Fabriation Symposium Pro., by J.J.Beaman, H.L.Marus, D.L.Bourell, and l.w.barlow, Aug. 6-8, 1990, University of at Austin, Austin, Texas. 1991 Solid Freeform Fabriation Symposium Pro., JI-J..U.",",,,",," J.W.Barlow, D.L.Bourell, and R.H.Crawford, Aug. at Austin, Austin, Texas. 1992 Solid Freeforn1 Fabriation Symposium Pro., Jl-JUA.""",,U J.W.Barlow, D.L.Bourell, and R.H.Crawford, Aug. Austin, Austin, Texas. 1993 Solid Freeform Fabriation Symposium Pro., by J.W.Barlow, D.L.Bourell, and R.H.Crawford, Aug. 9-11, 1993, at Austin, Austin, Texas. J.A.Manriquez-Frayre and D.L.Bourell, "Seletive Laser Sintering ofbinary Powder," ibid., Referene #1, pp. 99-106. J.A.Manriquez-Frayre and D.L.Bourell, "Seletive Sintering of Powders," ibid., Referene #2, pp. 236-244. W.Weiss and D.L.Bourell, "Seletive Laser Sintering to Produe ibid., Referene #2, pp. 251-258. G.Zong, Y.Wu, N.Tran. LLee, D.L.Bourell, J.J.Beaman, and Seletive Laser Sintering of High Temperature Materials," ibid., Referene # 3, G.Prabhu and D.L.Bourell, "Supersolidus Liquid Phase of Powder," ibid., Referene #4, pp.317-324. M.K.Agarwala, D.L.Bourell, B.Wu and J.J.Beaman, "An Evaluation Behavior of Bronze-Nikel Composites Produed by Seletive J.J.Beaman, University of Texas Solder lntermetalli...:,1"1,1-""'... Referene # 4, pp.193-202. M.K.Agarwala, D.L.Bourell, B.Wu and J.J.Beaman, "Struturally Sound Metal by Seletive Laser Sintering," The Congress 1994, of a symposium sponsored by the Extration and Proessing Division, At the 123rd TMS Annual Meeting, Mar. 1994, San Franiso, CA, pp. 833-851. M.K.Agarwala, D.L.Bourell, J.J.Beaman, B.Wu, J.W.Barlow, "Seletive Laser Sintering of a Bronze-Nikel Powder Mixture," IMS lnt. Conf. on Rapid Produt Development, Feb. 1994, Stuttgart, P.E.Rie and S.P.Kohler, "Hot Isostati Pressing of Metal Powders," Metals pp. 419-443, Vol.7, 9th Ed., Am. of Metals, Park, Ohio. H.V.Atkinson, B.A.Rikinson, "Hot Isostati Pressing," Adam A A"'...p.,"""... Manufaturing Proesses and Materials, Adam Publishing 1991. "Isostati Pressing: Tehnology and Appliations,"...".u...'"" Elsevier Applied Siene, London, 1991. M.F.Ashby, "A First Report on Sintering Diagrams," F.B.Swinkels and M.F.Ashby, "A Seond Report on...,... AIC""""...A.. ;;;" pp. 259-81, 1981. A.S.Helle, K.E.Easterling, and M.F.Ashby, "Hot-Isostati Pressing Developments," Ata Metall., 33, pp. 2163-74, 1985....,.,...,...,. New on New York, M.Nishihara, " 73