Porosity in Spray-Formed Copper-Alloy Billets Hilmar R. Müller, Klaus Ohla, Robert Zauter, Michael Ebner Wieland-Werke AG Graf-Arco-Strasse 36, D-89079 Ulm e-mail:
[email protected]
Abstract In spray-formed pre-forms porosity is unavoidable. The atomising gas is entrapped during the process. Because usually nitrogen is used the voids are filled with this gas. Unfortunately nitrogen is not soluble in copper. To get dense material after further forming processes like extrusion, forging or drawing the porosity has to be limited to a certain level. There are means to influence the porosity in the spray-formed billet. Up to a certain extent the porosity can be influenced by process parameters like metal mass flow and/or G/M (Gas/Metal flow rate ratio). Reactive elements in the alloy (e.g. Ti) have an additional effect. A theory of porosity formation is proved by analytical means and calculation. For quality testing usually a disc is cut from each end of the billet. These discs are used as samples for chemical analysis and porosity measurement. The results should be representative of the whole billet (typically 2000 mm long, dia. between 160 and 500 mm). But in fact there are differences from bottom to top which cannot be detected. Ultrasonic (US) testing is a powerful tool to overcome this disadvantage. DC-cast material cannot be checked by USprobes in the as-cast condition due to the coarse structure. The US-signal is deflected by each grain boundary. Therefore porosity and other defects are hidden behind this signal level. The spray-formed structure is so fine, that porosity and defects like cracks are detectable.
Introduction Quality issues require to limit the as-sprayed porosity for subsequent processing of sprayformed copper-alloy billets to a certain level. This porosity can only be minimised when the effecting parameters are identified. These are described previously by various authors. Pores can be formed in the droplets by gas entrapment during droplet formation, dissolution of gas from the molten metal, solidification shrinkage or collision between larger liquid and smaller solidified droplets [Akh90].
During deposition the fraction of solid fs is most important. At high fraction of solid fs the pores arise from poor spreading of droplets and insufficient liquid feeding [Akh90]. Interstices between adjacent droplets arriving at the surface leave irregular cavities [Wat90]. This can be amplified by a high heat extracting substrate. The layers near the substrate are more porous than the bulk material [Wat90], [War96], [Mah90]. At low fraction of solid fs the interstices between solid particles can be fed by liquid metal and the porosity decreases. Apparently there are some contrary results. For example Akhlaghi [Akh90], Doherty [Doh97] and Warner [War96] describe, that a low liquid fraction fl results in high percentages of pre-solidified droplets and porosity. On the contrary, Chu [Chu97] reports that bulk porosity increases as fraction of liquid fl in the spray cone increases when impacting on mushy surface. It is explained by the state of the mushy layer on the top of the deposit. At high fraction of liquid fl the mushy layer is thick and hot. Therefore the viscosity is low. The mushy layer is continuously disturbed by the high velocity gas jet and atomised droplets. When the droplets hit the mushy surface, they entrain the surrounding gas [Rei98]. So it is expected, that fl has an upper and a lower limit for preventing either gas porosity or cold porosity. Another kind of porosity is caused by the so-called cauliflower-effect. It occurs during spray deposition in the centre of the billet. There are many solidified droplets collected [Jor98]. The minimising effect on porosity by reactive elements is reported by Watson [Wats90], [Wat90] and Cookey [Coo90]. The latter recommends strong nitride formers like silicon and chromium. Watson named as preferred elements aluminium, silicon, titanium, chromium and zirconium. The forming of nitrides changes the surface tension of the droplets. This influences the behaviour of the droplets during the impact of the mushy layer and therefore the entrainment of gas [Rei98]. All the reported types of porosity can be classified in two classes: cold and hot porosity (Fig.1).
Porosity
Cold Porosity § § § §
At low fraction liquid fl Near substrate At high G/M-ratio Interstices between adjacent solidified droplets are not fed by liquid metal
Hot Porosity § § § §
Gas entrainment by splashing droplets Injection of gas into the deposit by droplets Dissolution of gas during solidification Coalescence of fine bubbles to bigger ones
Figure 1: Classification of porosity in spray-formed deposits
Porosity in Copper-Billets Definition and Measurement of Porosity Porosity is defined by the following equation: Pt = (1 −
ρ ) *100 ρ0
(1 )
Pt : Porosity in %
ρ 0 : Mass density of material without pores. Not the theoretical but the measured density of hot and cold worked material is used. ρ : Mass density as measured in the deposit The density is measured with a buoyancy weighing-machine according to DIN EN 6018. The specimens are 10 x 10 x 10 mm³ cubes [Mat01]. These results are compared with pictures of the microstructure at 50 x magnification. Fig. 2 - 4 give an impression of the measured data for 3 different alloy types:
ρ = 6.44 Pt = 10.52%
ρ = 6.91 Pt = 4.0%
ρ = 7.11 Pt = 1.23%
ρ = 7.13 Pt = 1.03%
Figure 2: Microstructure and porosity of CuAl13Fe4.5CoMn, ρ 0 = 7.20 g/cm³
ρ = 8.37 Pt = 2.21%
ρ = 8.74 Pt = 2.13%
ρ = 8.83 Pt = 1.07%
ρ = 8.85 Pt = 0.89%
Figure 3: Microstructure and porosity of CuSn14.5, ρ 0 = 8.93 g/cm³
ρ = 8.04 Pt = 2.5%
ρ = 8.14 Pt = 1.34%
ρ = 8.20 Pt = 0.61%
ρ = 8.21 Pt = 0.54%
Figure 4: Microstructure and porosity of CuMn20Ni20, ρ 0 = 8.25 g/cm³ Process Parameters effecting porosity The most important parameter is the Gas-to-Metal flowrate G/M as reported elsewhere [Mue00]. It is not only important for porosity but for segregation too. In production plants the minimum G/M ratio is not limited by an increasing porosity caused by the lower viscosity of the mushy layer and therefore an increased gas entrainment (see Introduction, [Chu97] ). The
limiting factor is damage of the billet by centrifugal forces. Big pieces of partly solidified material are expelled. But now we will focus on the effect of reactive elements. Reactive elements which have an effect on porosity We think that the entrainment of gas by the droplets hitting the liquid surface of the deposit is the main reason for hot porosity. This effect is boosted by cavities in the solidified droplets. Partly solidified droplets collide with rigid particles. Some of them are embedded, other break out and leave b a deepenings. In Fig. 5 at position a, a droplet is embedded and completely welded. At position b, a particle broke out and left a crater. It cannot be recognized if this happened during flight or after preparation of the c sample. At position c, the particle which formed the crater lost contact before the collision with particle a. Crater b is deformed by this collision. Such craters increase the gas entrainment, when the droplets dive into the liquid layer on the deposit. Figure 5: Overspray particle with embedded particle and two craters The elements zirconium and titanium were tested in preliminary trials. Their effect on porosity is evaluated as equivalent. So for the following tests only titanium was added during normal production to numerous billets. From each end one slice was cut and at three radial positions (edge, median, centre) the density was measured. This was done for three different alloy groups represented by CuAl13Fe4.5CoMn, CuSn13.5 and CuMn20Ni20. All results are collected in data files, the average and standard deviation is computed and plotted in diagrams (Fig. 6 –8).
8,00
Density in g/cm³
7,50
7,00
6,50
6,00 without Pb and Ti Average: 6,90
5,50
with Pb Average:
with Pb and Ti Average: 7,10
7,00
Standard deviation: 0,067
Standard deviation: 0,374
Standard deviation: 0,069
Figure 6: Density of CuAl13Fe4.5MnCo with and without Ti
1866.4r
1866.2z
1864.2r
1865.2m
1863.4z
1861.4r
Run-Nr.
1862.4m
1861.2z
1859.2r
1860.2m
1858.4z
1856.4r
1857.4m
1856.2z
1854.2r
1855.2m
1853.4z
1683.4r
1852.4m
1683.2z
1681.2r
1682.2m
1680.4z
1678.4r
1679.4m
1678.2z
1607.2r
1608.2m
5,00
10,00
Density in g/cm³
9,50
9,00
8,50
8,00
with Ti
without Ti
Average: Standard deviation:
7,50
8,85 0,16
Average: Standard deviation:
8,69 0,08
11 55 .2 11 z 56 .2 11 z 57 .2 11 z 58 .2 11 z 59 .2 11 z 60 .2 11 z 61 .2 11 z 62 .2 11 z 63 .2 11 z 64 .2 11 z 65 .2 11 z 66 .2 11 z 67 .2 11 z 68 .2 11 z 69 .2 11 z 70 .2 11 z 71 .2 11 z 72 .2 11 z 73 .2 12 z 38 . 12 2z 38 .4 m 12 39 .2 12 r 40 . 12 2z 40 .4 m
7,00
Run-Nr.
Figure 7: Density of CuSn13.5Pb with and without Ti
9,0 0 8,8 0 8,6 0
Density g/cm³
8,4 0 8,2 0 8,0 0 7,8 0 7,6 0 without T i
7,4 0 with T i
8,05 0,19
A verage:
7,2 0
Standard deviation:
with T i
8,15
Average:
0,15
Standard deviation:
Figure 8: Density of CuMn20Ni20 with and without Ti
1783.4r
1788.4m
1781.4r
1779.4r
1774.4r
R un-N r.
1776.4m
1770.4z
1772.4m
1766.4z
1768.2r
1764.4m
1761.4r
1759.4r
1757.4r
1755.4m
1753.4r
1652.2r
1650.2r
1648.2m
1646.2m
890.FR
1137.2z
889.KZ
7,0 0
The addition of 0.25 % Pb to CuAl13Fe4.5CoMn increases the density significantly. It is not a reactive element, but it increases the fluidity of the mushy zone by the low melting point. So the interstices between solidified particles are filled. The higher density of lead in comparison with copper is negligible because of the little content. A further increase of density is observed after adding 0.15 % titanium into the melt. The alloy CuSn13,5Pb contains about 0.5 % lead. The addition of 0.15 % titanium in the melting furnace reduces porosity as described for CuAl13Fe4.5CoMn (Fig. 7). A similar result is shown by Fig. 8 for the alloy CuMn20Ni20. In the Introduction it is reported of the effect of reactive elements on porosity by forming nitrides. In the following this theory is proved by measurement and calculation. Analysis of Nitrides The method described by Hedges [Hed03] was adapted to CuSn. Samples for phase extraction were accurately machined cylinders 30 mm x 6 mm in diameter. Each cylinder was cleaned using 400 µm SiC paper before extraction. Secondary phases were extracted electrolytically at a current density of 3 mA/mm2 in a solution of 10 % HCl in methanol. The sample, which acted as the anode, was suspended in the acid solution such that a known surface area was submerged below a reference mark, and was concentrically located in a shaped stainless steel sheet that was used as the cathode (Fig. 9). The CuSn matrix was gradually dissolved and the secondary phases settled to the bottom of the beaker. The electrolyte was then filtered, as shown in Fig. 10, and the secondary phase particles were collected on a glass microfibre paper using a filter connected to a vacuum pump. +
-
Power pack Extracted phases
Stainless steel cathod Reference mark
Filter support
Electrolyte with extracted phases Microfibre filter Clip
Extraction sample Glass beaker
To vacuum pump
Electrolyte Extracted phases
Figure 9: Schematic diagram of the experimental set up for the electrolytic extraction of secondary phases
Electrolyte
Figure 10: Schematic diagram of the experimental set up for the filtration of extracted secondary phases from the electrolyte
To give crystallographic information, an analysis was performed on the extracted phases with the X-ray diffractometer STADI P (STOE &CIE GmbH, Darmstadt). The X-ray diffraction diagram (Fig. 11) shows the result. It is a striking evidence for TiN in the extract. This is confirmed by X-ray diffraction on the undissolved spray-formed sample with the ARL diffractometer X`TRA [Mul02].
Figure 11: X-ray diffraction and theoretical diffraction diagram for TiN (Osbornite)
What is the effect of titanium on porosity? The following assumptions are made for a calculation of the entrained nitrogen and reaction with titanium: § The entrained gas bubbles adapt their temperature to the surrounding metal immediately. § Solidification time of the mushy layer allows the reaction of titanium and nitrogen. Watson [Wats90] reports of 5 to 22 seconds for a strip and Doherty [Doh97] 10 to 200 seconds for a billet. Compared with this the reaction time during flight is only a split-second (droplet speed 50 - 100 m/s, flight distance 600 mm). § During solidification the bubble pressure is constant. Because of the relative thin liquid layer the metallostatic pressure is neglected. The initial pressure in the bubble is equal to the spray chamber pressure. § The bubble volume changes isobar until solidus temperature is reached. When they are enclosed in regions of higher solid fraction fs, they are fed with liquid metal through channels with low melting phases (Fig. 12). With further cooling, the bubble (pore) volume changes only by shrinkage of the surrounding metal. § The pores are regularly (like a ball) or irregularly shaped. § The gas in the bubble behaves ideally. The overall volume VP of the pores in a deposit of the mass mD can be calculated from the measured density, using the definition of porosity in equation (1).
VD = m D / ρ
(2)
VP = Pt *VD
(3)
Bubble, initial size at Tliqu
Sucked in liquid metal
Bubble at Tsol
Bubble, in mushy zone Solidified crystals
Figure 12: Scheme of the mushy layer on top of the deposit with embedded gas bubbles VD : volume of the deposit m D : mass of the deposit VP : volume of all pores in the deposit Therefore the density of the gas is kept constant during cooling to solidus temperature, the density at environmental temperature can be calculated by considering the shrinkage of the surrounding metal during cooling from solidus to environmental temperature:
ρ E = ρ sol (1 + 3α (Tsol − TE ))
(4)
ρ E : density of gas at environmental temperature ρ sol : density at solidus temperature Tsol : absolute solidus temperature α : linear thermal expansion coefficient The cooling from solidus to environmental temperature decreases the pressure in the bubbles. This effect is counteracted by the shrinkage of the surrounding metal. According to the law of ideal gas the pressure is: PE = Psol
ρ E TE ρ sol Tsol
(5)
PE : pressure in the pores at environmental temperature TE : absolute environmental temperature
The number of moles N2 in the deposit is now defined: n=
PEVP RTE
(6)
n : number of moles N2 R : molar gas constant The mass and concentration of N2 in the deposit can now be calculated using the molar mass of N: m N 2 = 2nM N c N2 =
(7)
mN2
(8)
mD
m N 2 : mass of nitrogen in the deposit
M N : molar mass of nitrogen c N2
: concentration of nitrogen in the deposit
The concentration was measured by hot extraction with the LECO ON-Analysator TC 600 [Ebn02]. In Table 1, these results are compared with the calculated concentration c N 2 . Table 1: Comparison of calculated with measured nitrogen concentration in the deposit Alloy Run ρ0 ρ min ρ max Pt min Pt max c N 2 ,min c N 2 ,max c N 2 ,m CuSn13.5Pb CuSn13.5
g/cm³ g/cm³ g/cm³ 1528 8.95 8.60 8.71 1736 8.93 8.65 8.81
% 2.68 1.34
% % 3.91 0.00010 3.14 0.00005
% % 0.00015 0.00015 0.00012 0.00020
Because the samples for density measurement and nitrogen analysis are from the same billet but not identical, the nitrogen concentration is calculated with the minimum and maximum density measured in the billet. The measured nitrogen concentration c N 2 ,m is nearly as high as the calculated maximum c N 2 ,max . This confirms the theory that in copper-base alloys without reactive elements the nitrogen is entrapped in the pores. The reaction of titanium with nitrogen can be proved by a similar calculation. Using Equation (3) the pore volume is calculated putting in the measured density with and without titanium. The volume difference is the amount of nitrogen, which reacted with titanium to TiN. ∆VP = V P − VP ,Ti
(9)
mTi = 2nM Ti
(10)
cTi =
mTi mD
∆VP : pore volume difference between deposits with and without titanium
(11)
VP : pore volume without titanium V P ,Ti : pore volume with titanium mTi
: mass
of titanium in the deposit
cTi
: concentration
of titanium, which is necessary for reducing the porosity from Pt to PtTi
The mass and concentration of TiN are: mTiN = 2n( M Ti + M N ) cTiN =
mTiN mD
(12)
(13)
mTiN : mass of TiN in the deposit cTiN : concentration of TiN in the deposit Table 2 shows an example calculation for CuMn20Ni20. The grey coloured fields are input data. The measured densities come from Fig. 8. The difference in density with and without Ti seems to be very small but in terms of porosity there is a factor of 3. The minimum concentration of Ti in the melt for this reduction of porosity is very low. The presence of TiN is proved (Fig. 11). The calculated quantity of formed TiN is so little that the reaction is a probable cause of reducing porosity. We are going on with quantitative analysis of the TiN concentration in CuMn20Ni20Ti to prove the assumptions of calculation. Detection of porosity For quality assurance, a slice from top and bottom is cut from each spray-formed billet and samples for density measurement are punched out at three different positions. This gives only a rough idea of the real porosity in the billet. More information is provided by a nondestructive method like ultrasonic testing (US). Due to the fine microstructure it is possible to test the as-sprayed material by US. First tests are encouraging. Fig. 13 shows an longitudinal section of a spray-formed billet and the US-test result. The test was performed before the longitudinal cut. On the right there is a very strong porosity, on the left the billet seems to be sound. But the US-diagram shows a number of porous layers.
Figure 13: Longitudinal Section of partly porous deposit dia. 100 mm (above) and US-test result (below)
Table 2: Example calculation for CuMn20Ni20 with and without Ti
ρ0 ρ ρ Ti mD ρN ϑN ϑ sol α ϑE Pliqu
8.20 g/cm³
Density of porefree deposit
8.05 g/cm³ 8.15 g/cm³
Measured mean value of deposit without Ti Measured mean value of deposit with Ti
1500 kg 1.17 kg/m³
Mass of deposit
15 °C 1005 °C 1.86E-05 1/K 25 °C 0.1 MPa
Density of gas at ϑ N , 1 bar Standard temperature Solidus temperature Thermal linear expansion coefficient Environmental temperature Pressure at entrainment of bubble (chamber pressure)
R M Ti
8.31 J/mol K Molar gas constant 47.90 g/mol Molar mass of Ti
MN Pt PtTi
14.01 g/mol
ρ sol ρE VD V D ,Ti
0.000264 g/cm³
Density of gas in pore at solidus temperature
0.000278 g/cm³
Density of gas in pore at environmental temperature
186,335.40 cm³ 184,049.08 cm³
VP V P ,Ti
3,409 cm³ 1,122 cm³
Volume of the pores in the deposit without Ti Volume of the pores in the deposit with Ti
∆V
nTi
2,286 cm³ 0.0246 MPa 0.023 mol 0.045 mol
Volume of N2 reacted with Ti Pressure in pore at environmental temperature Number of moles N2 reacted with Ti Number of moles Ti reacted with N
mTi
2.17 g
PE n
cTi
1.83 % 0.61 %
0.00014 %
Molar mass of N Porosity without Ti Porosity with Ti
Volume of deposit without Ti Volume of the deposit with Ti
Mass of Ti reacted with N2 Concentration of Ti needed for reaction
nN
0.068 mol
Total number of moles N in pores of the deposit
mN2
0.948 g
Total mass N2 in pores of the deposit
c N2
mTiN cTiN
0.000063 % 2.810 g 0.00019 %
Total concentration N2 in the deposit Mass of TiN in the deposit Concentration of TiN in the deposit
Conclusion The porosity in spray-formed copper-base alloys is classified in two classes: cold and hot porosity. It is influenced by process parameters and by addition of reactive elements, e.g.Ti. The most important process parameter is the Gas-/Metal flow rate ratio G/M. The effect of reactive elements can superpose this parameter. Titanium (Ti) and zirconium (Zr) have equivalent effect on reduction of porosity. A large number of CuSn-, CuAlFe- and CuMnNi-alloy billets were spray-formed with and without Ti. The density was measured by buoyancy balance. The data from these billets were used for calculating the mean porosity. The comparison of calculated with analysed nitrogen and titanium-nitride concentration in the billet gives a clear indication of the mechanism of hot porosity formation. It is believed that the gas (nitrogen) is entrained by the droplets hitting the mushy layer on top of the substrate. This nitrogen reacts with titanium in the melt. Most of this happens between liquidus and solidus temperature. Reaction below solidus temperature does not decrease porosity but can decrease the pressure in the pores. The calculations and measurements prove the theory of nitrogen filled pores and demonstrate how addition of Ti helps to reduce porosity. First non-destructive tests with an ultrasonic probe are encouraging. Porous regions are clearly detectable. This opens the window for 100% testing of spray-formed copper-base alloy billets.
References [Akh90]
[Chu97] [Coo90] [Doh97]
[Ebn02] [Hed03] [Jor98]
F. Akhlaghi, J. Beech, H. Jones: Influence of Operating Parameters on Characteristics of Aluminium Powders and Spray-Cast Deposits, Proceedings of the First International Conference on Spray Forming, 17th – 19th September 1990, Swansea, U.K. M. G. Chu: Microstructure of Aluminium Alloy Deposit Produced by Spray Forming Using a Linear Nozzle, Kolloquiumsband Sprühkompaktieren, Universität Bremen, Band 2 (1997), 115 - 130 R.H. Cookey, J.V. Wood: Production and Development of Copper-Base Alloys by the Osprey Process, Proceedings of the First International Conference on Spray Forming, 17th – 19th September 1990, Swansea, U.K. R. Doherty, S. Annavarapu, C. Cai, L. K. Kohler: Modelling Based Studies for Control and Microstructure Development in Spray Forming, Kolloquiumsband Sprühkompaktieren, Universität Bremen, Band 2 (1997), 45 - 78 M. Ebner: Nachweis von TiN-Phasen im sprühkompaktierten Material BC1, interner Laborbericht Nr. 1859 der Wieland-Werke AG, Ulm, 22.8.2002 M. Hedges: Spray Forming and Microstructure of Nickel Superalloys IN718 and RS5, D. Phil thesis, Oxford University, Dept. of Materials, 2003 N. Jordan, H. Harig: Sprühkompaktierte Kupferbasis-Werkstoffe – Stand der Forschungs- und Entwicklungsarbeiten, Kolloquiumsband Sprühkompaktieren, Universität Bremen, Band 3 (1998), 31 - 52
[Mah90]
[Mat01] [Mue00]
[Mul02] [Rei98] [War96]
[Wat90] [Wats90]
P.P. Maher, P.S. Grant, B. Cantor, L. Katgerman: Manufacture of SprayFormed Al Based Alloys and Composites, Proceedings of the First International Conference on Spray Forming, 17th – 19th September 1990, Swansea, U.K. E. Mathei-Schulz, A. Schulz, P. Mayer: Gefügeauswertung an sprühkompaktierten Werkstoffen mit bildanalytischen Methoden, Kolloquiumsband Sprühkompaktieren, Universität Bremen, Band 5 (2001), 179 - 191 H.R. Müller, S. Hansmann, K. Ohla: Influence of Process Parameters on Segregation and Porosity in Spray-Formed Cu-Sn-Billets, Spray Deposition and Melt Atomization Conference, 26. – 28. June 2000, Bremen, Germany, 205 - 218 J. Muller: XRD measurements on the ARL X’TRA diffractometer, Customer Result Report 0602, ARL Applied Research Laboratories S.A. En Vallaire Ouest C, CH-1024 Ecublens, Switzerland, 20.06.2002 M. Rein: Spray Deposition: The Importance of Droplet impact Phenomena, Kolloquiumsband Sprühkompaktieren, Universität Bremen, Band 3 (1998), 115 - 128 L. Warner, C. Cai, S. Annavarapu, R. Doherty: Modelling Microstructural Development in Spray Forming: Experimental Verification, Proceedings of the Third International Conference on Spray Forming, 1996, Cardiff, U.K. G. Watson: Thermal and Microstructural Characterization of Spray Cast Copper Alloy Strip, Proceedings of the First International Conference on Spray Forming, 17th – 19th September 1990, Swansea, U.K. W.G. Watson, S. Ashok, H.P. Cheskis: Method to Reduce Porosity in a Spray Cast Deposit, U.S. Patent Number 4 961 457, 9th Oct. 1990
