Joining techniques in zero gravity

TWI Bulletin, February 1986

by Paul Marshall

Paul Marshall, BSc, is a Research Engineer in the Stainless Steels and Corrosion Section of the Materials Department.

With the advent of the Space Transport System, better known as the Space Shuttle, there has been a rapid acceleration in the number of materials science experiments conducted in space, aimed both at fundamental study and at developing materials technology. Part of this work has been directed towards electron beam welding and vacuum brazing, as described in this article.

The initial motivation for investigating joining, techniques by the United States National Aeronautics and Space Administration (NASA) came from studies performed in the early 1960s on methods of assembling structures in space. At that time construction, maintenance and repair of large orbiting space stations were envisaged, and the consensus of opinion was that, of the methods involving a molten metal phase, electron beam welding and vacuum brazing would be the best processes for performing the necessary operations. Both these techniques would make use of one of the inherent qualities of the space environment - a high vacuum. (Even in low earth orbit, altitude about 200km, pressure is reduced to 1o-a torr because of an almost total absence of atmosphere).

Other special aspects of this environment are the reduction in gravitational acceleration experienced and the high level of incident radiation. The former would be expected to influence the behaviour of liquid phases and consequently affect these joining processes. The level of gravitational acceleration experienced by Earth orbiting vehicles is 10 ^-3 - 10 ^-4g and is frequently referred to as microgravity (g is the normal gravitational acceleration of 9.81 m/sec ² ).

Although simple materials experiments and scientific demonstrations had been conducted during the flights of Apollo 14 in 1970 and Apollo 17 in 1971, investigation of practical aspects of electron beam welding and vacuum brazing by American scientists waited until June 1973 and the first Skylab mission. In the meantime, Soviet workers had acquired considerable data not only on electron beam welding and brazing techniques but also on plasma arc and consumable electrode methods of joining metals in space. Unfortunately few details of this work have been released but it is known that experiments were performed on stainless steels, aluminium alloys and titanium alloys.

By comparison, a great volume of information on early American experiments has been released. This, together with more recent work by European workers serves to illustrate some of the interesting phenomena observed, and the differences in the behaviour of materials in space from those in more normal conditions.

Electron beam welding

An electron beam metals melting experiment (M551) was performed in June 1973 in the M512 metals processing facility of Skylab I ^[1] To maximise the data from this experiment, the apparatus was designed so that a constant power electron beam (at 8omA beam current) impinged on a slowly rotating disc of material which had been machined to present several different thicknesses to the beam. In this way regions of over, full and partial penetration were produced on the same sample. In addition a diffuse beam was held in one position to create a large molten pool which was subsequently allowed to solidify. Experiments were performed on a stainless steel (AISI type 304), an aluminium alloy (2219-T87) and pure tantalum. The results of these experiments were analysed in conjunction with those from a parallel set of experiments performed on the ground.

Bead shape

Analysis of the materials showed that gravity effects on the weld bead shape appeared to be small. However some reduction in the frequency of surface ripple, beading and spatter under microgravity conditions was observed. ^[1] These parameters are of interest to fundamental study of solidification but are relatively unimportant to practical aspects of welding. A detailed analysis of the ripple phenomena is contained in a paper by Garland and McKeown. ^[2] This investigation concluded that, in the absence of any microstructurally related feature, the ripples were formed by vertical oscillations of the pool surface caused by vibrations from equipment. An explanation for the reduced frequency of ripples observed under microgravity conditions may be found by considering the behaviour of waves in a molten pool. Surface waves can be introduced into a liquid metal by vibration (as stated above), by fluctuations in a welding power source, and also through surface tension forces (surface capillary waves). These waves are normally affected by gravity and their frequency may be expressed by a simplified Navier-Stokes equation thus

where w is wave frequency, σ is surface tension, λ is wavelength and ρ is density of the liquid ( Fig.1). It is clear that as the value of the gravitational acceleration parameter decreases, so the wave frequency also decreases.

Solidification structure

Metallurgical examinations conducted by several institutions produced some conflicting results. While samples prepared by The Welding Institute showed very small effects of gravity, ^[2] significant differences between results of normal gravity and microgravity experiments were found by other investigators in stainless steel and tantalum samples. ^[2] The ground based solidified melt showed large columnar grain growth with heavy banding along the centreline of the weld bead, and a relatively narrow region of epitaxially formed crystals at the fusion boundary. In contrast, the specimen which had been produced under microgravity displayed a major reduction in the quantity of large columnar grains, an increase in thickness of the epitaxial grain region at the fusion boundary and almost no banding.

The observations can be explained when the mechanisms governing metal solidification are considered. Under normal terrestrial conditions the behaviour of metals before and during solidification is governed predominantly by gravity driven phenomena. In addition to the obvious separation caused by density differences between compositionally dissimilar phases, temperature gradients cause transient changes in the density of single phase materials leading to convectioe flow. The turbulence produced by such buoyancy driven convection currents has three effects of concern in the study of grain growth in liquid metals. First, mass transport causes mixing and reduces the extent to which chemical gradients can form. This has implications on the degree to which constitutional supercooling can be expected. (Constitutional supercooling occurs when solute enrichment of the liquid just ahead of the growing solidification front causes the solidification temperature of the liquid to be lowered.) Second, temperature homogenisation controls the thickness of the thermal boundary layer and therefore the thermal gradient of the solidification interface. Finally, turbulent flow during dendrite growth causes fracture of the dendrites and results in crystal multiplication in the bulk of the melt ( Fig.2).

Under microgravity conditions, buoyancy driven convection through density differences between phases or from temperature fluctuations is significantly reduced. Therefore, greater constitutional supercooling would be expected through lack of mass transport; the development of a thicker thermal boundary layer reduces the temperature gradient in the melt and further increases supercooling ( Fig.3), and crystal multiplication by dendrite fracture is negligible, ( Fig.4).

The extent of constitutional supercooling has a pronounced effect on the solidification mode and resulting solidification structure. Figure 5, adapted from a paper by Savage ^[4] summarises the effects of supercooling on solidification mode. For any solute concentration it can be seen that by increasing supercooling the solidification mode may be changed from columnar to equiaxed dendrite growth. This, therefore, can be used to explain the reduction in columnar grain growth in space. The reduction in banding is most probably a consequence of enhanced microsegregation which leaves few impurities ahead of the solidification front. As these do not remain in the last liquid to solidify, banding does not occur.

The improved compositional homogeneity associated with high supercooling has been demonstrated in a separate microgravity experiment. ^[5]

The more equiaxed appearance of grains in the microgravity samples may also be a consequence of the absence of any gravitationally induced favourable growth patterns of the dendrites. In normal gravity, convective dispersion of the latent heat of fusion of a growing dendrite favours growth of 'downward' orientated dendrites. Under microgravity this does not occur. [6] Where there is no convection, latent heat can only be lost by conduction, and this occurs equally in all directions resulting in equal growth in all directions.

Crystal multiplication through dendrite fracture causes heterogeneous nucleation of grains at an early stage in solidification, and these grains therefore have a greater opportunity to grow into large crystals. This grain multiplication does not occur under microgravity, so that only heterogeneous nucleation on existing grains or homogeneous nucleation in the melt occurs. Absence of grain multiplication has been confirmed by direct observation of the solidification of a model system (NH 4 Cl) in microgravity during the Space Processing Applications Rocket programme, [7] experiment V/74-37/2R.

Mixing under microgravity

The discussion so far has suggested that under microgravity, in the absence of gravity induced convection currents, mass and heat transfer in a liquid would be expected to be by diffusion alone. This is indeed the case for closed systems and has been demonstrated by several experiments where diffusion of a radioactively labelled species in a liquid has been studied, the results from microgravity experiments being compared with terrestrial equivalents and mathematical models based on Fick's diffusion laws.

Self-diffusion of 65 Zn was studied on Skylab in experiment M558 [8] . During the Apollo-Soyuz Test Project, experiment MA-041 examined the diffusion of 198 Au in Pb, [9] and more recently diffusion of 112 Sn and 124 Sn in Sn has been observed during the first Spacelab mission in experiment ES 335. [10]

The composition profiles obtained in experiments conducted under microgravity closely matched those predicted by diffusion transport whereas the terrestrial experiments showed that convective transport was dominant.

Vacuum brazing

Behaviour of liquids

When a free surface or a liquid/liquid interface is present, surface tension may play a considerable role in controlling the behaviour of fluids in microgravity. The Bond number, B o , is a measure of the relative importance of gravity and surface tension in influencing the behaviour of liquids. [11]

B o = ρgd 2 / σ

where ρ is the density of the liquid, g is gravitational acceleration, d is the diameter of a liquid sphere and σ is surface tension. As the gravitational acceleration parameter decreases the Bond number decreases, (assuming that the other parameters remain constant), and when B o falls below unity surface tension predominates and gravitational influence becomes secondary.

Under terrestrial conditions the Bond number only falls below unity when d is small, so that a significant force may be exerted in a liquid by surface tension between two closely spaced solid bodies. This allows spreading by capillary action and forms the basis of the theory for joint filling in brazing. The equations which govern capillarity must take into account the properties of the liquid/solid interface (this was not necessary when considering a single isolated phase in the Bond equation) and to this end the contact angle between the two phases is introduced.

The maximum theoretical height of capillary rise is given by:

and the velocity of capillary rise by:

where h is the maximum theoretical height of rise, r is capillary radius, µ is the coefficient of viscosity, H is the actual height of rise, and θ is the contact angle ( Fig.6).

Reduction in gravity increases the maximum capillary rise for any specified capillary gap, or the same rise may be produced in a wider capillary, and finally the velocity of capillary filling increases, all other parameters remaining unchanged.

It was the prospect of rapid brazing of large gapped joints which made this an attractive possibility for joining materials in space.

Experiments

Exothermic brazing experiments were conducted during the Skylab I mission. Tube materials of 304L stainless steel and commercially pure nickel were examined. The braze employed was a near eutectic silver-copper alloy (71.8%Ag, 28%Cu and o.2%Li). The tubes, with a gap cut in the centre to represent two tubes end to end, were assembled in a joining sleeve with braze material in an annular groove. The surrounding container was filled with a heating material which, when ignited, produced sufficient heat to melt the braze. The experiments were performed in a working chamber open to the vacuum of space. In this way a structural joint was made between the tube and the sleeve. The braze alloy flow and dispersion patterns found in these experiments were compared with those obtained in a parallel control group of experiments conducted under similar vacuum conditions on earth. A radioactive tracer allowed braze mixing and flow to be assessed autoradiographically, and specimens were also destructively examined.

These experiments performed well and yielded promising results. A greater dispersion of braze was found for both stainless steel and nickel samples prepared in space. This is not surprising for wide gaps, since these would not be expected to fill under terrestrial conditions. However, several anomalous effects were observed, including increased solubility of base metal in the filler and significant porosity in the solidified braze metal itself.

Pure Ni and Ag-Cu-Li

The relative success of experiments in vacuum brazing under microgravity has led to several additional, more sophisticated, experiments being performed, first during sounding rocket programmes ^[12] and more recently on the first Spacelab flight. ^[13] The experiments were performed using commercially pure nickel and a variety of silver-copper-lithium brazes. In each case the experimental package consisted of a number of concentric nickel cylinders forming a specimen with annular gaps of varying width. A supply of radioactively labelled braze was placed in each cartridge and the whole assembly evacuated to 10 ^-2 Pa before each space flight. During the flights, experiments were performed inside the isothermal heating furnace of the respective craft.

Experimental results showed that more active mixing occurred in the microgravity environment, apparently contradicting an earlier statement that, in the absence of gravity driven convection, fluid currents would be dramatically reduced. However, the dominant role of surface tension forces has already been referred to and considerable fluid currents have been demonstrated to arise from these forces alone. ^[8]

Temperature affects surface tension, and the development of surface tension gradients causes flow to occur from hot to cold regions ( Fig.7). Gradients in composition also cause differences in surface tension leading to flow in multicomponent systems.

Surface flow is opposed by viscous shear stresses within the bulk of the liquid and gives rise to the phenomenon known as Marangoni convection. ^[14]

This can induce fluid circulation in the bulk of the liquid causing mixing unaided by gravity driven forces.

It appears that in this case, gravity induced convection under terrestrial conditions is negligible and may possibly be opposed by mixing induced by Marangoni convection. In addition, the density difference of CuNi dendrites when compared with. molten braze is large (~0.95 g/cm ³ ) and causes segregation under terrestrial conditions, preventing a homogeneous composition from being achieved.

Longitudinal sections of the concentric cylinders revealed the extent to which capillary filling had occurred. The sections showed that only one of the several capillary gaps had been filled in the microgravity experiment whereas under normal gravity conditions all of the capillary gaps were filled to the degree dictated by the equilibrium hydrostatic and capillary pressures. This unexpected result was attributed to a difference in heat flow between the normal gravity and microgravity experiments. Normally hydrostatic pressure provides constant contact of the molten braze with the tube material and hence a means for heat transfer through conduction. In the absence of gravity, voids may form from outgassing or from surface instabilities and since buoyancy forces are negligible these voids remain stable, preventing heat transfer by means other than radiation. Therefore the temperatures of the braze in microgravity will not have been maintained for the duration of the experiment.

Although seemingly less successful, these more recent brazing experiments have illustrated the extreme care needed in the design of experiments to be conducted in microgravity environments.

* * *

Experiments so far have demonstrated that joining of metals in space by electron beam welding and by vacuum brazing is feasible. A second and equally important outcome of the work has been the gain of fundamental knowledge to aid understanding of the role which gravity plays in the behaviour of liquids on earth, and in particular the behaviour of liquid metals in brazed and welded joints.

References