The Effect of Hydrogen on Porosity Formation during Electron Beam Welding of Titanium Alloys Jianglin Huang*, Richard Turner, Jean-Christophe Gebelin, Nils Warnken, Martin Strangwood and Roger C. Reed PRISM2 Research Group, Department of Metallurgy & Materials, University of Birmingham, Birmingham, United Kingdom B15 2TT *Contact details: (Email):
[email protected], (Tel): (+44) 121 4143774.
Abstract Titanium and its alloys are prone to hydrogen-assisted porosity formation during welding, but this effect is not yet sufficiently understood. Research aimed at elucidating the behaviour of hydrogen during electron beam welding of Ti6Al-4V is presented. Characterisation is carried out using high resolution X-ray tomography, residual gas analysis and metallographic sectioning; this confirms that porosity formation is associated with hydrogen evolution. To quantify the dependence between porosity formation and hydrogen content in the base material, a hydrogen diffusion-controlled bubble growth model is used to simulate bubble growth in the melt, and thus to make predictions of the hydrogen concentration barrier needed for pore formation. The modeling results are supported up by experimentation on Ti-6Al-4V of different hydrogen levels, achieved by electrochemical charging. The results confirm that vigorous hydrogen degassing happens at high hydrogen levels. But porosity can be suppressed when welding is carried out with optimized welding parameters and perfect joint alignment; on the other hand, porosity is exacerbated when a small beam offset is employed. The influence of beam offset on porosity formation is discussed. It would appear that the nucleation rate in the liquid zone at the melting front determines the likelihood of porosity occurrence. Keywords: Welding Porosity, Keyhole, Electron Beam Welding, Titanium Alloys.
Introduction Titanium alloys are prone to porosity formation during fusion welding. Early work [1, 2] found that the occurrence of porosity in electron beam (EB) welded Ti-6Al-4V depended both on the intrinsic hydrogen content of the material being welded, and the quality of the edge preparation. Silvinskii [3] believed that titanium hydride was responsible for porosity formation; hydrogen was considered to be the root cause, but little was revealed about the critical hydrogen level needed to avoid porosity formation. Later, the appearance of pores during the fusion welding of a titanium alloy was attributed to the presence of gas forming substances (oil, grease, moisture) on the surfaces being welded and therefore inadequate cleanliness levels; presumably, decomposition of these might occur during processing [4]. Welding speed has also been found to correlate with porosity formation [5], which suggests that
porosity can be minimised if the dynamics of the bubble inside the weld pool can be understood and controlled. Unfortunately, although it is widely accepted that pore formation in titanium welds occurs due to hydrogen in the weld pool, the precise mechanism of formation is not yet well understood. With the progress made in the measurement/assessment of thermodynamic and kinetic data relating to hydrogen in titanium-based alloy systems [6-9], quantitative investigation of hydrogen behaviour during welding process and the mechanism of porosity formation is now possible, at least in principle. More recently, attempts have been made to understand the hydrogen migration and its effect on porosity formation during electron beam welding of titanium-based alloys [10, 11], which has shed light on the influence of hydrogen on porosity formation. This paper focuses on the influence of hydrogen on porosity formation during keyhole-mode electron beam (EB) welding of titanium and its alloys. Electron beam welds in commercially pure (CP) titanium are first studied, before proceeding to the welding of titanium alloys. Pores formed in electron beam welded titanium alloys are investigated; we believe this work to contain the first reported gas chemical analysis results of gas pores formed in electron beam welds of these materials. From the porosity characterisation results, the porosity formation mechanism is rationalised, with the role of hydrogen being shown to play a critical role. Due to the very complex phenomena occurring in the keyhole, a simplified hydrogen diffusion-controlled bubble growth model is proposed to quantify the dynamic growth of the bubbles in the weld pool, aiming to investigate the required hydrogen concentration needed to form a large bubble. To clarify the dependence between hydrogen content and porosity distribution, Ti-6Al-4V samples were electrochemically charged to achieve different hydrogen levels and then welded at different surface conditions and welding parameters. As will be seen, this has shed light on the role played by hydrogen in porosity formation.
Porosity Characterisation Commercially pure titanium (CP-Ti) EB welds for porosity characterisation were 6.5 mm-thick butt joints, which were fabricated using the electron beam welding facility at TWI, UK. The electron beam welding parameters are given in Table 1. Joint surfaces were prepared by wet machining. Before welding, the sample surfaces were washed using acetone wash
and then swabbed with 5% hydrofluoric acid, 35% nitric acid, and balance water, and finally washed in ethanol. The finished joint surfaces had a roughness of Ra 0.89 μm and Rz 4.46 μm. Table 1: EB parameters used in CP-Ti welding.
Accelerating Voltage (kV) 175
Beam current (mA) 80
Travel Speed (mm/s) 25
A. X-Ray Tomography. X-ray non-destructive testing (NDT) methods were used for porosity detection. For the rapid investigation of large pore formation, conventional X-ray film radiography was performed at Rolls-Royce plc, UK, Derby. The resolution of conventional X-ray radiography approaches approximately 2% of the thickness of the specimens. The major advantage of Xray radiography is the high efficiency for porosity detection in a large welded part; however, there also some disadvantages, e.g. the overlap of pores and surface defects, and relatively low spatial resolution. For more details of the pores, e.g. their three-dimensional morphology, precise location in threedimensions and high resolution imaging etc, it is necessary to resort to X-ray computed tomography (CT).
(RGA). A Hidden Analytical HAL 100-RC RGA (residual gas analysis) quadrupole gas analyser was employed with associated ultra high vacuum equipment, assembled by Leybold Ltd. Due to the small volume of the system and the power of the vacuum pumps, the system can be pumped down within a few hours and stabilised to a very high vacuum (10-8 mbar), before the analysis proceeds. The samples for RGA analysis should contain pores large enough which can be cracked easily inside the high vacuum chamber. The gas released from the pore is directed to the mass spectrometer, which is run in multi-ion detection (MID) mode until the detected background signal becomes stable. The released gases are detected by mass spectrometry, and the data collected and analysed using MASsoft (version 5) software provided by Hidden Analytical Ltd. In practice, samples for residual gas analysis need to be prepared with caution. To make sure the samples can be cracked to release the gas quickly, sample surfaces need to be ground down very close to the pores (within 100 μm), which requires the pore to be located precisely in 3-dimensions. For this purpose, high resolution (10 μm) X-ray tomography was carried out using the desktop SkyScan 1172 unit, which in practice was found to provide the necessary resolution.
C. Results of Pore Charactersation.
Based on conventional X-ray radiography, computed X-ray tomography uses projections obtained by conventional radiography many times in different orientations of the object, and then the X-ray attenuation is calculated at all points within a volume so that a 3D image is created. In this study, if X-ray tomography needed to be carried out, two different X-ray tomography facilities were used; a desktop micro-CT SkyScan 1172 unit (at the University of Birmingham) and an X-Tek Xray system XT H225L (at the µ-VIS {Multidisciplinary, Multiscale, Microtomographic Volume Imaging at Southampton} facility at the University of Southampton, UK). The SkyScan 1172 desktop micro-CT contains a maximum Xray source up to 100 kV, 100 mA, which enables 4mm-thick titanium sample analysis with good image quality. The maximum sample dimension is limited to 50 mm in the offset scanning mode employed. For larger samples which needed a larger working chamber, the X-Tek X-ray system was used. In both CT facilities, either a 1 mm Al filter or a 0.5 mm Al+ Cu filter was used. Depending on the thickness of sample, exposure time for each projection varied between 0.2~0.5 s. A suitable projection procedure was chosen according to the image acquired. Normally, a step angle of 0.45˚ was used, which led to a projection number of 800 for a 360˚ scanning. Increasing the projection number has a tendency to increase the image quality, but it increases dramatically the scanning and reconstructing time.
B. Residual Gas analysis. After locating the porosity in three dimensions with the X-ray tomography approaches, samples containing large pores (diameter>0.5 mm) were prepared for residual gas analysis
Figure 1: Reconstructed X-Ray Tomography showing three dimensional porosity distributions.
Figure 1 shows an example of a reconstructed X-ray tomography image. A large pore of diameter around 1.8 mm was identified for residual gas analysis, see Figure 2. The detected partial pressure confirms that most of the gas inside the pore is hydrogen. A small amount of CO2 was also detected. Beside H2 and CO2, gas with atomic mass unit (amu) of 28 was also found but with amount slightly less than for CO2. Since both N2 and CO have the same atomic mass unit of 28, it is not possible to distinguish these two gases from the mass spectrometry results. In principle, one would like to distinguish between these two gases, because they will be associated with completely different mechanisms of pore initiation. If N2 exists inside the pores, it indicates that air is involved with the porosity formation. Even though the EB welding process is performed at high vacuum condition, it is still possible that small air pockets can be entrapped between the two joint surfaces due to mechanical assembly before the welding chamber is pumped down. If the gas with amu of 28 is CO, it suggests that the porosity formation is related to the
decomposition of hydrocarbon. In both cases, it suggests that hydrogen favours porosity formation.
hydrogen diffusivity. In order to determine the dynamic bubble growth process, hydrogen diffusivity in liquid titanium is the hydrogen diffusivity value in liquid is assumed to be D = 2×10-7 m2/s, as used in [13]. Inside the bubble, the gas obeys ideal gas law:
PgVg ng RT
(2)
where Pg , Vg and ng are the total gas pressure, volume and amount respectively. The gases inside the bubble are of two types: (i) the insolvable gas with constant mass and (ii) the hydrogen from the melt. Due to ideal gas mixture law and pressure balance, one has:
Figure 2: Gas analysis result by using MID (Multiple Ion Detection) mode.
Hydrogen Diffusion-Controlled Bubble Growth The pore characterisation studies confirm that hydrogen plays a critical role in pore formation. In this section, a simple hydrogen diffusion-controlled bubble growth model has been developed to aid in the interpretation of the experimental results. It is assumed that a pre-existing bubble is surrounded by liquid metal, which contains a uniform hydrogen concentration C0 with temperature T and pressure PL. The surface tension between the liquid metal and gas bubble is σ. The bubble is stabilised by insoluble gas inside, at initial radius r0. Due to the hydrogen dissolved in the melt, diffusion towards the bubble occurs so that the pressure inside the bubble Pg (t) and gas amount ng (t) change. As long as the pressure inside the bubble is higher than the pressure induced by the melt and surface tension, bubbles will grow. To investigate the factors influencing bubble growth, the following major assumptions are made in the analysis: 1. The temperature is assumed to be isothermal. 2. Gas inside the bubble is treated as ideal. 3. The melt size is large compared to the bubble, so the bubble growth is unbounded and hydrogen content far from the bubble is constant. 4. Thermodynamic equilibrium is maintained at the gas/ liquid interface according to Henry’s law. 5. The liquid is incompressible.
PH P '
3nH RgT 4 R3
4 r03 PL 8 r02 2 PL 3 4 R R
(3)
where nH is the number of moles of hydrogen inside bubble, and PH and P’ are partial pressures of hydrogen and insoluble gas in the bubble respectively.
B. Boundary and Initial Conditions The initial and boundary conditions for Equation (1) are:
R(t ) r0 , CH (r , t ) C0 (t 0, r R(t )) CH (r , t ) CHL (t 0, r R(t )) CH (r , t ) C0 where
(4)
(t 0, r )
CHL is hydrogen concentration at the bubble/liquid
metal interface, which is assumed to obey Henry's law:
CHL K H ( PH )n
(5)
where PH is the hydrogen partial pressure inside the bubble. KH is Henry’s constant and n is set to 0.5 consistent with the reported hydrogen solubility in the titanium system [14]. At the gas-liquid interface, due to the mass conservation of hydrogen diffusion, we have:
dnH C 4 R 2 D H dt r
r R
(6)
Taking the equations (1) ~ (6) together, the equation set is solved using the finite difference method.
A. Governing Equation Under the above assumptions, hydrogen diffusion in the liquid metal obeys literature [12]. 2 2C|H 2 C|H CH R dR C|H D D t r dt r r 2 r r
(1)
where t is time, r is the spherical radial coordinate, R is the bubble radius, CH is the hydrogen concentration and D is the
C. Implementation of the Model The calculation procedure to solve above equation set is described as below: 1. Input temperature T, initial hydrogen concentration C0 and initial bubble radius r0 and surface tension σ. 2. Set up initial and boundary conditions. 3. Calculate insolvable gas amount n0 by using equation (2).
4. Perform hydrogen diffusion calculation within the melt by solving equation (1). 5. Calculate the amount of hydrogen diffusion through the gas-liquid interface via equation (6). 6. Track gas-liquid interface by using volume fraction method (VOF). 7. Check whether the hydrogen concentration at gas-liquid interface reaches the hydrogen concentration at the surrounding liquid. If not, go step 4 for the next iteration. The material properties and the physical constants were adopted consistent with values reported in the literature. The temperature for analysis is taken to be 2000K in this study, which is slightly above the liquidus of a typical titanium alloy. According to the reported temperature dependent surface tension of liquid Ti-6Al-4V [15], surface tension used here is 1.18 N.m-1, Henry’s constant has been estimated via the reported hydrogen solubility in pure liquid titanium [5], which has a value of 1.439 mol.l-1.atm-1.
the hydrogen concentration at the bubble/liquid metal interface, the hydrogen level at the interface evolves as shown in Figure 5. The peak hydrogen level at the interface as indicated in Figure 5 is around 260 ppm, which is lower than the initial hydrogen level in the liquid metal (300 ppm). For as long as the hydrogen level in the liquid metal is higher than that at interface, the hydrogen keeps diffusing from the liquid metal into the bubble, thus driving bubble growth. However, according to the evolution of the hydrogen level at the interface, there exists a hydrogen threshold in the liquid metal, which can limit bubble growth. The factors which affect this critical hydrogen level for bubble growth in the weld pool – pre-existing bubble size, surface tension and ambient pressure – are discussed in the next section.
D. Dynamic growth of bubble inside weld pool Figure 3 shows the predicted pore radius as function of time for growth of a bubble in unbounded liquid metal, with the hydrogen level of the melt set to 300 ppm. The pre-existing bubble size is assumed to be 20 μm, and ambient pressure to 10-4 mbar – typical of the vacuum conditions employed during the electron beam welding of titanium alloys. As the bubble grows, the associated pressure evolves inside the bubble as shown in Figure 4. The partial pressure of the pre-existing insoluble gas decreases, whilst the hydrogen partial pressure increases rapidly in the early stage of bubble growth due to the very small initial bubble volume. The rapidly increased hydrogen partial pressure increases the hydrogen level at the interface, which reduces the hydrogen concentration gradient in the liquid metal around the pores and thus slows down the bubble growth. After the hydrogen partial pressure reaches a peak value of around 6×10-4 Pa, it decreases thereafter with a further increase of the bubble radius.
Figure 4: Pressure evolution during bubble growth.
Figure 5: Hydrogen concentration evolution during the bubble growth.
E. Critical Hydrogen Level Analysis Figure 3: Calculated bubble evolution.
This is because an increasing bubble volume dominates the pressure change inside the bubble as it grows larger. A decreasing hydrogen level at the interface coupled with increasing area surface of the bubble favours the hydrogen diffusion into the bubble, which makes for faster bubble growth. Consistent with the local equilibrium assumption of
The above indicates that during bubble growth, a peak hydrogen level at the bubble/liquid metal can be identified. If the hydrogen level in the liquid metal is higher than this critical hydrogen level, the bubble keeps increasing in size until a balance between the hydrogen level at the interface and the residual hydrogen in the liquid metal is reached. It is of interest to determine the factors which influence this critical
hydrogen level. Figure 6 illustrates how the pre-existing bubble size, surface tension and ambient pressure affect the critical hydrogen concentration. When the pre-existing bubble radius decreases from 30 μm to 5 μm, the critical hydrogen level increases from 200 ppm to 450 ppm; hence the larger the pre-existing bubble, the lower the hydrogen level required in the liquid metal for pore formation. Our results indicate that surface tension will also play a crucial role in bubble formation; the critical hydrogen level drops with decreased surface tension. Since for titanium alloys the temperature coefficient of the surface tension has a gradient of -2.4×10-4 N/m-K [15], which means that high temperatures in the weld pool (high beam powers and/or low welding speed), will favour bubble formation. Finally, one can see that the ambient pressure used for welding will also influence bubble formation: a better vacuum decreases the critical hydrogen level for bubble formation.
Electrochemical Hydrogen Charge From the modelling results, one sees that the critical hydrogen level depends on several factors. Although it is difficult to predict the exact value of the critical hydrogen level by using the simplified model developed in this study, a series of targetted experiments were designed to investigate the dependence between the porosity formation and the critical hydrogen level. Thus, Ti-6Al-4V was electrochemically charged to achieve different hydrogen levels. The hydrogen charged samples were also prepared with different surface conditions, which may provide different sizes of nucleation site, which correlated to the pre-existing bubble size in the model. For example, one would expect more bubbles to exist when the sample has a rough surface and higher hydrogen content.
Figure 6: Effect of pre-existing bubble size, surface tension and ambient pressure on critical hydrogen level for bubble growth.
Materials used for hydrogen charge are Ti-6Al-4V plates, which were machined from a large forged compressor disc. Samples were rectilinear plates of length 300 mm, width around 30 mm and thickness of 4.0 mm. The joint surfaces of the as-received sample were very smooth, with Ra value around 0.5 µm. Hydrogen charging (cathodic) was performed
at room temperature, in a H3PO4: glycerine (1:2, v/v) electrolyte, with constant current density of 50 mA/cm2 applied by using the DC power supply. More details of the hydrogen charge employed can be found in [16]. Due to the limit of the maximum power supply and the size of electrolyte basin, samples for hydrogen charging were cut to have dimensions of 100 mm in length, 30 mm in width and 4 mm thickness. To increase the possibility of bubble nucleation at the melting front, the joint surfaces were ground off around 100 μm by using SiC paper (P60) to increase the surface roughness. A. Quantitative Hydrogen Measurement Quantitative hydrogen measurement is a challenge due to the significant hydrogen range (namely hundreds of ppm to thousand ppm in weight percentage) used in this study. For low hydrogen levels, the inert gas fusion method can be used to do the hydrogen measurement. Here samples are melted under an inert gas stream in a graphite crucible and the gas evolved measured by thermal conductivity or infra-red absorption. But due to the difficulty of calibration, the maximum hydrogen which can be measured in this way is usually limited to around 200 ppm; for higher hydrogen content measurement, it is necessary to resort to thermogravimetric analysis (TGA). In both of the above two methods, samples are heated to release the volatile components from the samples; in the former method, the release of volatile components is monitored and quantified by mass spectrometry, and in the latter one, the amount of the volatile component is quantified by measurement of the weight loss during the operation. To identify the gas in the TGA measurements, mass spectrometry (MS) can also be used in a so-called TGA-MS system. In this study, in order to determine the hydrogen content as a function of time, both the inert gas fusion method and the TGA-MS method have been used. The measurement strategy was as follows: first, hydrogen in the uncharged samples was measured using the inert gas fusion method, using a LECO RH404 analyser at IncoTest, UK. After determination of the initial hydrogen content in the base material, hydrogen in charged samples were measured in the same way. Unfortunately, with sufficient charging, the hydrogen contents in the charged samples were beyond the calibration of the LECO system (>200 ppm). For higher hydrogen content from the longest changing times, the hydrogen was initially measured using the TGA-MS system, at Hidden Isochema Ltd, UK, and then, the analysed samples were sent to IncoTest for residual hydrogen measurement by using the LECO RH404 system. Concerning the accuracy and reliability of the measurement results, the LECO RH404 can provide accuracy up to 2% of the reading. In the TGA-MS system, the resolution of weight measurement is 0.2 μg, and the maximum sample weight is limited to 4.5 g, so the accuracy depends upon the total weight of the sample tested. Samples for hydrogen analysis were cut using SiC cutting blades with water cooling. This might have a tendency to induce more hydrogen into the cut surfaces of the samples due
to high temperature at the front of the blade during the cutting process. To minimise any such effect, all cut surfaces of the samples were ground off by about 50 μm by using a MDPiano 220 grinding disc. After grinding, samples were washed using ultrasonic cleaning in ethanol and dried. The sample weight for the LECO RH404 system was around 0.1~0.2 g, leading the size of Ti-6Al-4V sample around 22~44 mm3. Considering a 4mm-thick plates, samples for hydrogen measurement were cut off parallel to the joint edges of the plates with 3 mm offset. So the samples for the hydrogen analysis in the LECO RH404 system had dimensions of 4 mm thickness, 3 mm in width, and a length of approximately 3 mm. The weight of samples for TGA-MS system was around 3.5 g and typical dimensions 10 mm× 2 mm× 4 mm. B. Characterisation of Hydrogen Charged Samples Hydrogen charged samples were characterised using standard metallographic methods. Figure 7 (a) shows the surface morphology of these samples prior to hydrogen charging. For short times (18 hours), cracks occurred at the sample edges, probably due to the formation of the titanium hydride phase.
Figure 8: Hydrogen measurement results of hydrogen charged sample.
C. EB Welding of the Hydrogen Charged Sample Figure 9 shows the electron beam current and detected pressure rise in the working chamber when electron beam welding a sample charged for 6 hours. When the beam reaches its maximum power, a very significant pressure rise is detected, which confirms that strong hydrogen degassing occurs. With an increase in charging time, the detected pressure level also increases dramatically, see Figure 10.
Figure 9: Monitored pressure rise in the working chamber.
Figure 7: SEM micrograph of uncharged (a) and 12-hours charged (b) sample surfaces.
The hydrogen measurement results obtained using the LECO RH404 system confirmed that the hydrogen uptake rate is approximately linear with time, at 50 ppm/hour as shown in Error! Reference source not found. For measurements of higher hydrogen content, the TGA-MS system was used. The hydrogen content in the charged samples was found to be not high enough to provide sufficient signal due to the sensitivity limit of the TGA-MS system.
Figure 10: Detected pressure rise during welding of sample charged with different time.
It can be concluded that the pressure rise in the working chamber is directly related to the hydrogen gas released from the sample during the welding process. Unfortunately it is not
totally clear whether hydrogen escaping from the weld pool induced bubbles and pores. If the bubbles were induced by hydrogen, one would expect that a few bubbles would be trapped during solidification or captured by the weld pool surface. None were identified. Figure 11 shows micrographs of the top surfaces of the weld beads after hydrogen charging to different times. As the hydrogen content increases, the quality of the weld joint was found to decrease. Charging for 12 hours caused extensive spattering. Nevertheless, no porosity was found either by conventional film X-ray radiography or high resolution X-ray tomography (5 µm). When looking at the cross-section of the 12-hours charged samples, significant undercut was observed, and very small pores (of diameter less than 1 µm) were located at the edge of the fusion zone, as shown in Figure 12. It seems that gas pores formed in the liquid metal when the hydrogen concentration was large, but due to the strong evaporation, hydrogen-rich liquid metal spattered away. This rationalises why there is no porosity found in samples with very high hydrogen content, but cannot explain why there is no porosity found in the samples charged for 3-hours and 6hours charged. The quality of the weld bead seemed normal, and no undercut was found.
A further point relates to the very significant pressure rise observed in the pressure chamber during welding of the hydrogen-charged samples; this confirms that hydrogen does indeed outgas during processing. In an optimised electron beam welding process, the keyhole is stabilised by an optimised energy input. The liquid metal film around the keyhole at the melting front is very thin, which enables the bubbles nucleated at the joint edges to break through the keyhole wall and thus to escape. To increase the possibility of bubble survival inside the weld pool, we postulate that a wider liquid region at the melting front is required. One possible way to achieve this condition is beam offset welding, as illustrated in Figure 13. To confirm whether the beam offset contributes to the porosity formation, beam offset welding was performed by preparing butt joint welds using material charged for 6 hours. In Figure 14(a), the beam offset value increased linearly from 0.3 mm to 1.0 mm within a welding length of 100 mm. High resolution X-ray tomography was used for porosity detection. Pores were indeed found in the beam offset welds. Figure 14(b) shows the cross section at a beam offset value of 0.65 mm, confirming lack of fusion and porosity formed at the joint edges. Figure 14(c) shows a gas pore from a beam offset value at 0.51 mm. To elucidate the influence of the beam offset effect on porosity formation more clearly, more work needs to be done. But from the results we have so far, it would appear the beam offset influences the probability of pore formation in a significant way.
Figure 13: Liquid zone size around joint edges at melting front changes due to beam offset (BOF).
Figure 11: Top surfaces of weld beads for samples charged after different time.
Figure 12: Undercut observed in 6-hours charged samples; note small pores at fusion edge.
Figure 14: Pores observed in beam offset welded samples.
Summary and Conclusions
[3].
The following specific conclusions have been drawn from this work:
[4].
1. Pores found in EB welds of titanium are round and have smooth inner surface, which suggests that porosity formation is associated with gas evolution during the welding process. Typical diameter around 100~300 μm; large pores (diameter > 0.5 mm) are rarely found.
[5].
2. Residual gas analysis has been performed on large pores found in EB welded CP-Ti. This has confirmed that hydrogen is the majority gas inside the pore; a small amount of CO2 also exists. 3. Electrochemical charging was used to achieve different hydrogen contents prior to welding; when optimised welding parameters were employed, it proved possible to suppress porosity formation by keyhole stabilisation. 4. Pores are found in hydrogen-charged samples welded with a small beam offset. It is postulated that this is because a greater beam offset increases the size of the liquid region around the joint edges at the melting front. A wider liquid region at the melt front favours bubble growth and their survival after bubbles nucleate from the joint edges. 5. A stabilised keyhole, narrow melting front and perfect beam alignment are suggested to be effective ways to minimise porosity formation during keyhole mode electron beam welding of titanium alloys, when the possible presence of hydrogen cannot be avoided.
[6]. [7].
[8].
[9].
[10].
[11].
[12].
Acknowledgments The authors are grateful to the Engineering and Physical Sciences Research Council (EPSRC) of the United Kingdom and to Rolls-Royce plc for sponsorship of this work, via a Dorothy Hodgkin Postgraduate Award. Support from Dr. Steve Beech and Alistair Smith at Rolls-Royce plc is appreciated. The authors also thank Dr. Mark Mavrogordato of the University of Southampton for his help and advice concerning X-ray tomography.
References [1].
[2].
Nikiforov, G.D. and Redchits V.V., "The Mechanism of Pore Formation When Thick Titanium Alloys are Fusion Welded". Avt. Svarka, Vol. 10 (1984). pp. 4245. Redchits, V.V., "Scientific fundamentals and measures used to prevent the formation of pores in fusion welded titanium and its alloys." Welding International, Vol. 11, No. 9 (1997), pp. 722-728.
[13].
[14].
[15].
[16].
Silvinskii, V.A., et al., "Role of hydrides in pore formation in welding titanium". Paton Welding Journal, Vol. 6 (1994), pp. 265-269. Gouret, N., et al., "Assessment of the Origin of Porosity in Electron-Beam-Welded TA6V Plates." Metallurgical and Materials Transactions A, Vol. 35A (2004), pp. 879-889. Mohandas, T., Banerjee D., and Rao V.V., "Fusion Zone Microstructure and Porosity in Electron Bean Welds of an α + β Titanium Alloy", Metallurgical and Materials Transactions A, Vol. 30A (1999), pp. 789. Okamoto, H., "H-Ti (Hydrogen-Titanium)". Journal of Phase Equilibria, Vol. 13 (1992), pp. 443. Wang, W.E., "Thermodynamic evaluation of the titanium-hydrogen system." Journal of Alloy and Compounds, vol. 238, (1996), pp. 6-12. Nishikiori, T., Nohira T., and Ito Y., "Kinetics Investigation of the Ti-H System by the Molten Salt Electrochemical Technique." Journal of the Electrochemical Society, Vol. 148 (2001), pp. E127E132. Kasajima, T., et al., "Thermodynamic Evaluation of Ti-H System at Medium-Range Temperature by Molten Salt Electrochemical Technique." Journal of The Electrochemical Society, Vol. 150 (2003), pp. E355-E359. Huang, J., et al., "Hydrogen Transport and Rationalisation of Porosity Formation during Welding of Titanium Alloys". Metallurgical and Materials Transaction A, Vol. 42, No. 2, (2012) pp. 582-591. Huang, J., et al., "On The Mechanism of Porosity Formation During Welding of Titanium Alloys." Acta Materialia, Vol. 60, (2012), pp. 3215-3225. Pai, V. and Favelukis M., "Dynamics of Spherical Bubble Growth". Journal of Cellular Plastics, Vol. 38 (2002), pp. 403-419. Fishgot, A.F., et al., "Diffusion of hydrogen during and after electron beam welding titanium alloys." Welding International, Vol. 7, (1993), pp. 230-233. Chen, Y., et al., "The behavior of hydrogen in high temperature titanium alloy Ti-60". Materials Science and Engineering A, Vol. 466, (2007), pp. 156-159. Westerberg, K.W., et al., "Analysis of the E-Beam Evaporation of Titanium and Ti-6Al-4V" Electron Beam Melting and Refining State of the Art Conference , Reno, NV, Oct. 1997. Eliezer, D., et al., "Hydrogen absorption and desorption in a duplex-annealed Ti-6Al-4V alloy during exposure to different hydrogen-containing environments", Materials Science and Engineering A. Vol. 433 (2006), pp. 298-304.
