2022 Volume 63 Issue 10 Pages 1477-1483
Flux-free brazing of aluminum alloys was carried out under high purity nitrogen gas with ultra-low oxygen partial pressure prepared by zirconia oxygen pump. The flux-free brazability of the aluminum alloys was improved by adding a small amount of magnesium into the sample, which was more effective in the core alloy than in the brazing filler alloys. Higher heating rate of the sample increased the fillet length. When the oxygen partial pressure of atmospheric gas was much reduced to the order of 10−25 Pa using a zirconia oxygen pump, a long fillet, comparable to that formed by flux brazing, was obtained. The reason for the improved brazability was discussed from the viewpoint of the fine segmentation of the oxide caused by the combined effects of reduction of the oxide, thermal expansion at melting the filler alloy, and gas phase formation reaction from the molten filler alloys.
This Paper was Originally Published in Japanese in J. Japan Inst. Met. Mater. 85 (2021) 352–358. Figure 6 is modified, and eq. (9) is added.
Fig. 6 Schematic of the fragmentation progress of oxide film on the brazing filler alloy during flux-free brazing of aluminum alloy under ultra-low oxygen partial pressure.
Most modern automotive heat exchangers are manufactured by brazing aluminum alloys in order to improve its thermal conductivity and reduce its weight.1,2) Since oxidation of the base materials to be joined and brazing filler aluminum alloys drastically deteriorates the brazability, using of fluoride-based flux is indispensable to maintain a clean surface, free from any oxide during the brazing process.3–6) Aluminum alloys are easily oxidized even at room temperature. However, a post-braze washing of the residual flux is very inefficient in terms of manufacturing time and cost.
Vacuum brazing has been practically applied for fluxless or flux-free brazing of some aluminum alloys.7–11) In this technique, the atmospheric pressure is reduced to about 10−3 Pa during the brazing of aluminum alloys to inhibit the oxidation of base materials and molten filler aluminum alloys as much as possible. Furthermore, magnesium is added to the brazing filler alloys and/or the base materials as an oxygen and moisture getter. However, the chamber is significantly contaminated by magnesium oxide deposits formed from vaporized magnesium during the brazing,9–11) and thus the majority of maintenance time spent on a brazing furnace is devoted to cleaning it. Furthermore, zinc is often used to improve the corrosion resistance of aluminum alloys, however, similar to magnesium, it also vaporizes when vacuum brazing is performed. Therefore, a new flux-free brazing of aluminum alloys must be developed as an alternative to vacuum brazing.
It is reported that the flux-free brazability of the aluminum alloys is improved by adding bismuth and magnesium to the Al–Si brazing filler alloys even under ordinary atmospheric pressure of inert gas.8–25) One possible effect of bismuth addition is to lower the liquidus temperature of the brazing filler alloys, which may contribute to enhance the fluidity of molten filler alloy if the brazing temperature is the same. The vaporized bismuth may act as an oxygen getter pump, as with the case of magnesium addition. Furthermore, it is expected that the addition of bismuth improves the wetting and spreading of the molten brazing filler alloy on the base material through a decrease in the surface tension. Our group has confirmed that a small amount of bismuth addition indeed lowers the surface tension of molten Al–Si brazing filler alloy by oscillating droplet method using the electromagnetic levitation technique.26)
The addition of magnesium is believed to induce the destruction of an oxide layer formed on the molten brazing filler alloys through the following reduction reactions of Al2O3(s),
| \begin{equation} \text{3Mg(s)} + \text{4Al$_{2}$O$_{3}$(s)} \rightleftarrows \text{3MgAl$_{2}$O$_{4}$(s)} + \text{2Al(s)} \end{equation} | (1) |
| \begin{equation} \Delta G_{(1)}^{\circ} = - 194420 - 21.45T\ [\text{J/mol}]\ (300\unicode{x2013}900\,\text{K})^{27)} \end{equation} | (2) |
| \begin{equation} \text{3Mg(s)} + \text{Al$_{2}$O$_{3}$(s)} \rightleftarrows \text{3MgO(s)} + \text{2Al(s)} \end{equation} | (3) |
| \begin{equation} \Delta G_{(2)}^{\circ} = - 127248 + 9.277T\ [\text{J/mol}]\ (300\unicode{x2013}900\,\text{K})^{27)} \end{equation} | (4) |
Since zirconia type oxygen pump enables us to dramatically decrease $P_{\text{O}_{2}}$ down to 10−25 Pa at about 873 K near the typical aluminum brazing temperature even under ordinary atmospheric pressure of 1.013 × 105 Pa,31) it could be used to inhibit the oxidization of the molten filler aluminum alloys during flux-free brazing. In this apparatus, a voltage put across a dividing wall of a solid stabilized zirconia electrolyte makes oxygen ions migrate through the wall from the cathode to the anode according to the Nernst equation.33)
In the present study, flux-free brazing of aluminum alloys was carried out under ultra-low $P_{\text{O}_{2}}$ nitrogen gas at ordinary atmospheric pressure prepared using the zirconia type oxygen pump. One purpose of this investigation was to confirm the effectiveness of ultra-low $P_{\text{O}_{2}}$ conditions using the oxygen pump on the improvement of flux-free brazability of aluminum alloys. Furthermore, the influence of heating time on the flux-free brazability of aluminum alloys was also investigated, since high temperature aluminum alloys usually oxidize with time. Based on these results, we discussed the reason why magnesium addition in the sample and the decrease in $P_{\text{O}_{2}}$ of atmospheric gas improve the flux-free brazability of aluminum alloys, in light of the inhibited oxidization of the sample and destruction of an oxide layer.
Figure 1 shows a schematic of the specimen for clearance filling test used in this study to investigate flux-free brazing of aluminum alloys, which is according to LWS T 8801:1991 by Japan Light Metal Welding Association.32) The core and brazing filler alloys with their basic compositions of Al–1.2%Mn and Al–10%Si–0.02%Bi by mass concentration were cast, at which 0.6% magnesium was added to both samples. A 0.4 mm thick brazing sheet was made by cladding the filler alloy on the core alloy, where the cladding ratio of the filler alloy was precisely controlled to be 10%.
Schematic of clearance filling test specimen for flux-free brazing of aluminum alloys.
After the brazing sheet and a 1 mm thick vertical plate of Al–1.2%Mn alloy were chemically cleaned using nitric hydrofluoric acid, the vertical plate was fixed on the brazing sheet together with a 1.6 mm diameter cylinder of type 304 stainless steel using a thin wire of type 304 stainless steel, to make a clearance between the brazing sheet and the vertical plate.
Figure 2 shows a schematic of experimental set-up used for flux-free brazing of aluminum alloys in this study. After the assembled specimen was placed in a quartz chamber, the $P_{\text{O}_{2}}$ of atmosphere was decreased to 10−25 Pa by circulating a high purity nitrogen gas through the zirconia type oxygen pump (ULOCE-500, Canon Machinery Inc.) operated at 873 K. The $P_{\text{O}_{2}}$ of the gas was confirmed using zirconia type oxygen sensors installed at the inlet and outlet of the chamber. The sensors were calibrated using the gas phase equilibrium of H2–CO2 and in-situ observation of oxidation/reduction reactions of several metals.31) After the $P_{\text{O}_{2}}$ of the circulating gas had stabilized, a preheated electric furnace was moved over the assembled sample to be heated under a 2 L/min gas flow. The specimen was heated and maintained at 873 K for 9 minutes to carry out the flux-free brazing after heating from 623 K to 850 K for 7 to 22 minutes. For comparison, the specimen was also heated under a flow of high purity commercial nitrogen gas at $P_{\text{O}_{2}}$ of 10−2 Pa.
Schematic of experimental set-up.
After the specimen was rapidly cooled to room temperature in the chamber under well controlled $P_{\text{O}_{2}}$ atmosphere by removing the furnace, the fillet length was examined. Furthermore, the surface of the specimen after the flux-free brazing was examined using a field emission scanning electron microscope (FE-SEM) and transmission electron microscope (TEM) equipped with an energy dispersive X-ray spectrometer (EDX).
Figure 3 shows the fillet length of the specimens maintained at 873 K for 9 min under a high purity nitrogen gas flow with $P_{\text{O}_{2}} \approx 10^{ - 2}$ Pa, after heating from 623 K to 850 K for 7 min. When a brazing sheet consisting of Al–10%Si filler alloy and Al–1.2%Mn core alloy is used for flux brazing under nitrogen gas at $P_{\text{O}_{2}} \approx 10^{ - 2}$ Pa, a 38 mm long fillet is formed (Fig. 3(a)). However, no fillet is formed for the brazing sheet in flux-free brazing (Fig. 3(b)).
Fillet length of the specimens maintained at 873 K for 9 min after heating from 623 K to 850 K for 7 min.
When using the brazing sheet with 0.02% bismuth and 0.6% magnesium added in the filler alloys, a 20 mm long fillet is formed even for flux-free brazing (Fig. 3(c)). This confirms that the addition of these elements is effective to improve flux-free brazability of aluminum alloys as reported.
Furthermore, the effect of magnesium addition is enhanced in the core alloy, as can be seen from the increase in fillet length to 32 mm (Fig. 3(d)). This result agrees with the report by Doko et al.24)
3.2 Influences of heating time and $P_{\text{O}_{2}}$The flux-free brazability was suggested to comparatively improve by adding bismuth in the brazing filler alloy and magnesium in the core alloy as shown in Fig. 3(d). However, further improvement of the flux-free brazability is required to make it an alternative to flux brazing. The influence of the heating time and $P_{\text{O}_{2}}$ on the fillet length was investigated in flux-free brazing of the specimen as in Fig. 3(d). The heating time and $P_{\text{O}_{2}}$ would affect flux-free brazability by oxidizing the specimen. The results are shown in Fig. 4 together with some typical SEM images for the surface of the brazing sheet after the experiment. In the flux-free brazing under flowing nitrogen gas at $P_{\text{O}_{2}} \approx 10^{ - 2}$ Pa (○), the fillet length increases from 22 mm to 32 mm when the heating time of the specimen from 623 K to 850 K, before melting the brazing filler alloy, is decreased from 21 to 7 minutes.
Influences of heating time and $P_{\text{O}_{2}}$ on the fillet length after flux-free brazing of the specimen using the brazing sheet consisting of Al–10%Si–0.02%Bi filler alloy and Al–1.2%Mn–0.6%Mg core alloy. Also shown are the representative SEM images for the surface of the brazing sheet after brazing.
Although the surface of the brazing sheet is entirely covered with an oxide layer when the heating time is 21 minutes (Fig. 4(d)), it becomes finely fragmented when the heating time is decreased to 7 minutes (Fig. 4(c)).
When the $P_{\text{O}_{2}}$ of the flowing nitrogen gas is much decreased to 10−25 Pa through the zirconia type oxygen pump operated at 873 K (■), flux-free brazability of the specimen is greatly improved. Of note was the specimen with the heating time of 7 minutes shows a long, 39 mm, fillet comparable to that for flux brazing. Furthermore, a clean brazing filler alloy surface is observed together with the fine fragmented oxide, suggesting the obviously inhibited oxidation of the brazing sheet.
3.3 TEM–EDX analysis after flux-free brazingAs mentioned above, flux-free brazing of aluminum alloy was significantly improved by simultaneously shortening the heating time and decreasing the $P_{\text{O}_{2}}$, used in conjunction with a brazing sheet consisting of Al–10%Si–0.02%Bi brazing filler alloy and Al–1.2%Mn–0.6%Mg core alloy. The oxide fragmentation was observed at the surface of the brazing sheet after the flux-free brazing. The surface of the brazing sheet after the experiment was analyzed using TEM–EDX to examine the fragmentation of the oxide layer in more detail. Figure 5 shows the representative TEM–EDX images of the fragmented oxide observed at the surface of the brazing sheet after the experiment, corresponding to Fig. 4(a). Aluminum and magnesium are simultaneously detected at the region where fragmented oxide is observed. Selected area electron diffraction analysis revealed that the fragmented oxide is magnesium aluminate spinel (MgAl2O4(s)). This suggests that magnesium added to the core alloy diffuses toward to the surface of the brazing sheet when the specimen is heated, followed by the formation of MgAl2O4(s) through a chemical reaction with Al2O3(s) formed at the surface of the brazing filler alloy. This result is consistent with the fact that the standard Gibbs energy for formation of MgAl2O4(s) at the brazing temperature of 873 K shows negative values in reaction (1) between Al2O3(s) and Mg(s) and in reaction (5) between Al2O3(s) and MgO(s) obtained from the combination of reactions (1) and (3).
| \begin{equation} \text{Al$_{2}$O$_{3}$(s)} + \text{MgO(s)} \rightleftarrows \text{MgAl$_{2}$O$_{4}$(s)} \end{equation} | (5) |
| \begin{equation} \Delta G_{(5)}^{\circ} = - 22392 - 10.240T\ [\text{J/mol}]\ (300\unicode{x2013}900\,\text{K})^{27)} \end{equation} | (6) |
TEM–EDX images of the fragmented oxide observed at the surface of the brazing sheet after the experiment, corresponding to Fig. 4(a). Selected electron diffraction image for the oxide is also shown.
The flux-free brazability of aluminum alloys is reported to be improved by adding some magnesium to the specimen due to reduction of Al2O3(s), formed at the surface of molten brazing filler alloy, through reactions (1) and (3) for forming MgAl2O4(s) and MgO(s). However, only 0.6% magnesium added in the brazing sheet used in this study should be able to reduce only a very limited amount of Al2O3(s). Although the $P_{\text{O}_{2}}$ of atmospheric nitrogen gas was greatly reduced to ∼10−25 Pa using a zirconia type oxygen pump, it is still much higher than the equilibrium oxygen partial pressure of liquid aluminum. In this case, it is quite possible that the molten brazing filler alloy reoxidizes even once the added magnesium reduces the Al2O3(s) formed at the surface, followed by a deterioration of flux-free brazability. Nevertheless, excellent flux-free brazability of the aluminum alloy, comparable to that of flux brazing, was obtained in this study when using a brazing sheet consisting of Al–10%Si–0.02%Bi brazing filler alloy and Al–1.2%Mn–0.6%Mg core alloy at $P_{\text{O}_{2}}$ of 10−25 Pa. In this section, we will discuss a possible mechanism for flux-free brazing of aluminum alloy with a small amount of added magnesium, in light of the fact that oxide layer was finely fragmented after the experiment.
4.1 Effects of shrinkage of the oxide layerAccording to Pilling and Bedworth,34) the oxide layer shows porosity and/or cracking due to its volume shrinking when the ratio of the molecular volume of oxide to the atomic volume of the metal from which the oxide is formed is less than unity. Using the idea of Pilling and Bedworth, the ratio of total molecular volumes of the products to that of reactants was evaluated for the chemical reaction that added magnesium to the brazing sheet reduces 1 mol of Al2O3(s) to form MgAl2O4(s), corresponding to reaction (1), using the following eq. (7),
| \begin{equation} R_{(1)} = \frac{3/4V_{\text{MgAl${_{2}}$O${_{4}}$}} + 1/2V_{\text{Al}}}{3/4V_{\text{Mg}} + V_{\text{Al${_{2}}$O${_{3}}$}}} \end{equation} | (7) |
| \begin{equation} R_{(2)} = \frac{3V_{\text{MgO}} + 2V_{\text{Al}}}{3V_{\text{Mg}} + V_{\text{Al${_{2}}$O${_{3}}$}}} \end{equation} | (8) |
Schematic of the fragmentation progress of oxide film on the brazing filler alloy during flux-free brazing of aluminum alloy under ultra-low oxygen partial pressure.
Since the added magnesium is very small amount of 0.6% and volume shrinkage is only 4%, it may be difficult to induce a porous and/or crack in a thick oxide layer formed at comparatively high $P_{\text{O}_{2}}$ and a prolonged heating. Even in this case, at least a shrinkage stress would be occurred near the region where the MgAl2O4(s) forms. As a result, 6% volumetric expansion of the brazing filler alloy at its melting stage37,38) can promote the cracking and fragmentation of oxide layer together with the synergistic effect of the shrinkage stress (Fig. 6(c)).
In addition, Al(s) formed together with MgAl2O4(s) is usually susceptible to erosion during eutectic melting of Al–Si brazing filler alloy, which would also contribute to the destruction of oxide layer. Once the molten brazing filler alloy flows out to the surface of oxide layer due to wetting, natural convection, and the Marangoni convection, the destruction of the oxide layer would be further accelerated.
As mentioned above, adding magnesium to the brazing sheet contributes to induce not only the reduction of Al2O3(s) formed at the surface of the brazing sheet but also shrinkage stress in the oxide layer, which result in the formation of cracks and consequent fragmentation of the oxide layer.
4.2 Destruction and inhibited reoxidation of oxide layer by gas formation reaction at gas–liquid interfaceAlthough the formation of cracks and consequent fragmentation of the oxide layer, induced by the shrinkage stress at the reduction of Al2O3(s) using the added magnesium and thermal expansion of the brazing filler alloy at the melting stage, would expose the molten brazing filler alloys at the surface, it should reoxidize even under ultra-low $P_{\text{O}_{2}}$ nitrogen gas at 10−25 Pa prepared by a zirconia oxygen pump because it is still much high than the equilibrium oxygen partial pressure of oxidation reaction of liquid aluminum. In this regard, we considered that gas formation reaction from the melt could contribute to inhibit reoxidation. When an oxide layer completely covers the surface of the molten brazing filler alloy, gas phase formation reaction from the melt is kinetically inhibited since the formed gas phases cannot escape to the atmosphere. Whereas once a gas–liquid interface appears due to cracking and/or fragmentation of the oxide layer the gas phase formation reactions from the melt becomes possible (Fig. 6(d)). At this stage, the added magnesium, showing a high vapor pressure, in the brazing sheet evaporates followed by formation of MgO(s) with oxygen gas contained as impurities in the atmosphere. Since this oxygen getter effect through evaporation of magnesium can lower the oxygen partial pressure in the vicinity of the melt surface, $P_{\text{O}_{2}}^{\text{sur}}$, it would be favorable to inhibit reoxidization of molten brazing filler alloy. Conversely, exposing a gas–liquid interface is indispensable to expect the oxygen getter effect of magnesium vapor during the flux-free brazing of aluminum alloy, which was predicted in many conventional studies of improving the flux-free brazability of aluminum alloy. Evaporation of aluminum and silicon from the molten brazing filler alloy would also show a similar oxygen getter effect.44,45)
Research group of Arato and Ricci46–48) theoretically proposed that $P_{\text{O}_{2}}^{\text{sur}}$ becomes dramatically lower than $P_{\text{O}_{2}}$ of atmospheric gas in metal systems that form volatile metal oxide like aluminum, tin, and silicon. In their study they considered the mass transport mechanisms between liquid metals and the surrounding atmosphere involving the combined effects of vaporization phenomena, and chemical reactions forming metal oxide gases. Their theoretical model suggested that liquid aluminum, the main component of the brazing filler alloy used in this study, does not oxidize to Al2O3(s) due to the decreased $P_{\text{O}_{2}}^{\text{sur}}$ through a formation of Al2O(g), unless the $P_{\text{O}_{2}}$ of atmospheric gas is increased to around 10−4 Pa, which is much higher than the thermodynamically calculated equilibrium oxygen partial pressure of 10−51 Pa for formation of Al2O3(s). Similarly, the $P_{\text{O}_{2}}^{\text{sur}}$ for liquid silicon, which is another main component of brazing filler alloy, was also calculated to dramatically decrease due to the formation of SiO(g) and SiO2(g), the validity of which was experimentally confirmed.49)
Moreover, it was reported that the formation of Al2O(g) by the following reduction reaction of Al2O3(s) can also proceed in an open system.
| \begin{equation} \text{4Al(l)} + \text{Al$_{2}$O$_{3}$(s)} \rightleftarrows \text{3Al$_{2}$O(g)} \end{equation} | (9) |
Our group experimentally confirmed that the oxide layer of aluminum and Al–Si alloys, formed to cover the entire surface before melting, was reduced even under relatively high $P_{\text{O}_{2}}$ of 10−2 Pa once small cracks appeared in the oxide layer after melting the sample, by in-situ observation of the electromagnetically levitated sample.26) Since the observation was carried out at around 1000 K, the reduction of the oxide layer was thought to be because of the formation of gas phases described above, rather than thermal decomposition.
In summary, the added magnesium in the brazing sheet is supposed to form some MgAl2O4(s) at the surface of the brazing filler alloy through reduction of Al2O3(s), which induces a shrinkage stress that allows the oxide layer to crack and fragment. Volumetric expansion of the brazing filler alloy at the melting stage promotes the cracking and subsequent fragmentation of the oxide layer. Gas phase formation reactions becomes possible after the appearance of a gas–liquid interface, which induces a decrease of oxygen partial pressure in the vicinity of the melt surface together with reduction of Al2O3(s). As a result, further fragmentation of the oxide layer becomes possible and inhibition of reoxidation of exposed melt followed by the improved flux-free brazability of aluminum alloys.
Flux-free brazing of aluminum alloy was carried out using a clearance filling test to investigate the effect of a decrease in the $P_{\text{O}_{2}}$ of atmospheric gas, shorter heating time, and adding magnesium to the specimen. As a result, a small amount of magnesium added to the brazing sheet made flux-free brazing of aluminum possible. The magnesium addition was more effective in the core alloy than in the brazing filler alloy.
The decrease in the $P_{\text{O}_{2}}$ of atmospheric gas using a zirconia type oxygen pump and shortened heating time before melting the brazing sheet improved the flux-free brazability of aluminum alloys. Of note was the excellent flux-free brazability, comparable to that using flux, was obtained under ultra-low $P_{\text{O}_{2}}$ of 10−25 Pa with a heating time of 7 min.
A fragmented oxide layer was observed at the surface of the brazing filler alloy after flux-free brazing, in which the formation of MgAl2O4(s) was detected from the reduction of Al2O3(s) by adding magnesium in the brazing sheet.
From these results, the flux-free brazing was made possible in this study due to a combination of the following four effects;
This work was partially supported by JSPS KAKENHI under Grant No. 15H04136 and P20H02453.
