«Please cite this article as: Mendes, R., Ribeiro, J.B., Loureiro, A., Effect of explosive characteristics on the explosive welding of stainless steel ...»
The area of the melted regions varies with the type of sensitizer and also with the collision point velocity. For the welds using HGMB, the area of the melted regions increases as Vc increases (see Figs 9a to 9c). For weld 3, melted zones of successive waves are even connected, as illustrated in Fig. 9c. Welds performed with explosive containing EPS show large melted areas at the top and base of the waves (see Figs 9d and 9e). In addition, those melted area show voids that occur as the result of shrinkage during solidification, increasing their size as the volume of melted metal increases. Weld 6, performed with ANFO, and presenting the highest impact velocity (see table 3) shows waves identical to the weld carried out with the explosive sensitized with the highest amount of HGMB, but has very small melted areas, and even an absence of fusion, as illustrated in Fig. 9f. These results suggest that the morphology of the weld interfaces is affected not only by impact velocity, but also by the nature of the sensitizer added to the explosive. The characterization of wave geometry was done in a simplified manner using the amplitude A and wavelength λ, as illustrated in Fig. 9a.
The effect of impact velocity on the amplitude and length of the waves at the weld interfaces is illustrated in Fig. 10.
Fig. 10 – Effect of impact velocity on the amplitude of waves at the weld interfaces.
Fig. 10 shows that welds done using explosive sensitized with HGMB result in increasing amplitude of the waves with increasing impact velocity, though wavelength displayed only a modest increase. The welds performed with explosive sensitized with EPS showed the same trend.
However, these increasing rates are much higher for the welds performed with explosive sensitized with EPS than for the explosive sensitized with HGMB. In addition, the values of the amplitude and wavelength are themselves greater for the welds performed with the explosive sensitized with EPS than for the welds performed with explosive sensitized with HGMS.
The increase in wavelength and amplitude with the explosive load was also observed by Kahraman et al.  in welds between aluminum and titanium plates. According to Manikandan et al. , it is reasonable to link the loss of kinetic energy with the wavelength and wave amplitude, which is an essential requirement for inducing plastic flow for welding. Manikandan et al.  showed a consistent increase of the wavelength and amplitude of the interfacial waves with the loss of kinetic energy in titanium to stainless steel welds. As stated by Hokamoto et al. , once the loss of kinetic energy at the welding interface is a function of the impact velocity and of the mass of the colliding plates per unit of area it can be said that, for each kind of sensitizing agent, the results obtained are in accordance with the results of Manikandan et al. . Zamani and Liaghat  also reported an increase in the length and amplitude of the waves at the interface with the increase of explosive loads for the welding of stainless steel to carbon steel coaxial pipes.
However, none of the authors mention the effect of the nature of the explosive.
The effect of the collision angle on the morphology of the waves at the weld interfaces is illustrated in Fig. 11. The wave amplitude decreases as the collision angle increases in welds performed with explosive sensitized with HGMB, as opposed to welds done with explosive sensitized with EPS. However, weld 6, the one with the highest collision angle, displayed a low wave amplitude, similar to that of weld 3. The wavelength is not significantly affected by the collision angle for welds performed with explosive sensitized with HGMB: in fact, although it was performed with the highest collision angle, weld 6 presents a wavelength similar to welds 1 to 3.
Welds 4 and 5 (explosive sensitized with EPS) displayed the longest wavelength, regardless of the collision angle.
These results show that the variation of wave amplitude and length is much more a function of the explosive type than of the collision angle or impact velocity. Most of the previous attempts to link the welding parameters with the wave amplitude and length did not consider the influence of the explosive type. The most current parameters correlating with wave amplitude and length were the collision angle and the flyer and base plate thicknesses; amplitude and wave length increase with both parameters [30; 37].
However, Plaksin et al.  stated that the Kelvin-Helmoltz instabilities, which are nowadays accepted as one of the most likely mechanisms of the wave formation, are triggered by the flyer plate instabilities which, in turn, are a consequence of the oscillatory nature of the propagation of the detonation wave.
Fig. 11 – Effect of collision angle β on the amplitude (a) and wavelength (b) at the weld interface.
Those oscillations, related with the periodic variation of values of the local detonation velocity [24;
59], are much greater for the emulsion explosives sensitized with EPS than for those sensitized with HGMB. The much greater roughness of the external surface of the flyer tube for the welds performed with the explosives sensitized with EPS, when compared with the welds performed with explosive sensitized with HGMB, is an evidence of that difference. The erosion figures on the surface of metallic materials in contact with detonating explosive compositions are directly related with the spatial dimension of the detonation oscillations . The deeper and further apart the surface dents are, the greater the oscillations in the detonation process. In accordance with what was observed, greater oscillations during the propagation of the detonation wave are believed to induce greater oscillations in the flight velocity of the flyer plate and in turn, larger interfacial waves.
Although the results do not show a clear variation of the wave size with the collision angle for each kind of explosive, they suggest that the size of the interfacial waves increases with the impact velocity. Higher impact velocities are expected to result in conditions at the collision interface closer to the hydrodynamic state, and thus more suitable for the “actuation” of the KelvinHelmoltz instabilities, and the formation of bigger waves.
3.5. Hardness at the interface
A substantial increase in hardness was observed at the interface of the welds, both on the sides of the stainless steel and the low alloy steel, as illustrated in Fig. 12. On the low alloy steel side, the hardness increases from 220 to 300 HV0.05 at the interface, and extends over a width of a few mm. On the stainless steel side, the hardness rises from 165 HV0.05 (the hardness of the base material shown by a dotted line in the Fig. 12) to values between 300 and 450 HV0.05, with the highest near the interface and decreasing with distance from it, as illustrated in Fig. 12. This increase in hardness is consistent with the large plastic deformation experienced by the flyer during the impact. In Fig. 9, it is noticeable that deformation lines parallel to the weld interface are present in the stainless steel, caused by strong plastic deformation of the flyer during impact against the base rod. The deformation is so marked that micro-cracks occur in the flyer. Such marked plastic deformation should alter the mechanical properties of the flyer, particularly its hardness. Kahraman et al.  observed similar plastic deformation in oblique explosion welding between titanium and stainless steel, with the grains close to the interface, elongated and oriented parallel to the detonation direction. In order to verify that hypothesis, hardness measurements were carried out on the most plastically deformed zone of the AISI 304 stainless steel tensile specimens tested up to rupture. The values measured reached 320 HV0.5, which are close to the minimum values measured on the stainless steel flyer tube. However, it was not possible to establish any correlation between the impact velocity and the hardness achieved in stainless steel, because of the gradient of hardness observed. A similar explanation was given by Kaçar et al. [12; 62] and Durgutlu et al.  for the hardness increase in stainless steel flyers.
Fig. 12 – Microhardness profiles in the cross section of the welds. CS – carbon steel; SS – stainless steel. Base materials hardness is indicated by slash-dot lines.
In the regions close to the weld interface where vortices occurred, hardness values measured were significantly above those mentioned in Fig. 12. Fig. 13 illustrates an example of a line of hardness measurements on weld 5, which shows that, in the melted and solidified zone at the base of a wave, the hardness indentations are much smaller than in carbon steel or stainless steel.
Hardness values of approximately 700 HV0.05 were measured in all welds in which there was significant formation of melted and solidified areas. According to the equations of Yurioka et al.
, these values of hardness correspond to a martensitic structure containing 0.49% C, which is the carbon content of the carbon steel analyzed. However, the hardness within the melted regions is not uniform, as illustrated by the variation in the size indentation. This suggests variations in chemical composition inside melted region.
Fig. 13 – Hardness variation in a melted and solidified region of weld 4.
3.6. Analysis of melted zones As shown, for all the welds, except ANFO, the crest and the base of the interfacial waves show the formation of melted zones with a high degree of hardness. The formation of these local melted areas results from the dissipation of kinetic energy of the impacting materials in the form of heat (Crossland  and Hokomato et al. ), and takes place according to the mechanism described by Bahrani et al. . The formation of these zones is expected to be created with material from both the flyer and the base plates: stainless and carbon steels, respectively, although those contributions may not be uniform, as suggested by the variation in hardness measurements mentioned above. Fig. 14 shows the qualitative elemental distribution in the crest of a wave in weld 5. The area studied was 50 μm by 50 μm in order to include part of a melted region. The elements analyzed were those present in varying quantities in the carbon and stainless steels used in the study, such as iron, silicon, nickel and chromium.
The crest of the wave is located at the top right of the images and the molten zone at the bottom right, while the stainless steel is on the left side. The image of the elemental distribution of iron shows that the content of this element in the molten zone is higher than that of stainless steel, but lower than that of carbon steel according to the colour scale shown at the bottom of the Fig. 14.
Since silicon, nickel and chromium appear in greater quantities in the melted and solidified region than in the carbon steel, but in lower values than in stainless steel, it can be stated that both materials have contributed to the formation of the melted and solidified area.
Fig. 14 – Elemental mapping in the crest and in the molten zone of a wave of the weld 5.
Fig. 15 – SEM image of the melted zone in the crest of a wave of a weld 5.
The image of the melted zone, which contains shrinkage holes of a wave of the weld 5, is illustrated in Fig. 15. The analysis of chemical composition at different points (1 and 2) within the melted area, using energy dispersive X-ray spectroscopy, showed small differences in chemical composition, with the chromium varying between 9.74% and 7.19% wt and nickel between 4.49% and 3.84% wt. These results are consistent with the observations of Kaçar and Acarer  in welds of duplex stainless steel to carbon steel, where the molten zones resulted from the melting and mixing of both the flyer and the base material. These compositional differences are understandable, assuming that the heating and cooling rates experienced in the area are very high, and do not allow for complete mixing of the molten metal. In fact, according to Crossland  cooling rates about 105-107 K/s can be reached during solidification. Song et al.  also observed strong gradients in the chemical composition inside the molten regions in titanium to steel cladding.
The explosive welding of stainless steel AISI 304 to low alloy 51CrV4 steel in a cylindrical configuration was studied. A weldability window was determined for the experimental conditions
used. The following conclusions can be drawn:
- The final appearance and roughness of the external surface of the flyer is greatly affected by the amount and nature of the sensitizing agent of the emulsion explosive.
- All the welds exhibited wavy interface morphology. Pockets of melted and solidified material could be found for the welds performed in conditions outside the weldability window, while for the welds within the window the melted regions are absent or insignificant.
- Interfacial wave amplitude and length were shown to be, for each kind of explosive/sensitizing agent, a function of the impact velocity.
- No relation was established between wave length and amplitude and collision angle.
- The wavelength and amplitude were much more affected by the type and size of explosive/sensitizing agent than by any other process parameter.
- The emulsion explosive sensitized with EPS exhibited the highest values for wave length and amplitude at the weld interface.
- Significant hardening was observed at the interface of both materials, due to severe plastic deformation suffered during the wave formation.
- High hardness values, typical of high carbon/high alloy martensite microstructure, were found at the melted and solidified zones of the interface.
- Those melted and solidified zones displayed shrinkage holes, as well as chemical compositions which are a result of the contributions of both tube and rod compositions.
Acknowledgements This research is sponsored by FEDER through the program COMPETE – Programa Operacional Factores de Competitividade – and nationally through FCT – Fundação para a Ciência e a Tecnologia – under the project PEst-C/EME/UI0285/2011 The authors thank the MSc students in Mechanical Engineering Hugo Santos and João Pedro Carvalho for having carried out part of the experimental work presented in the article.