Journal of Materials Processing Technology 143–144 (2003) 651–655

Laser brazing of aluminium

T. Markovitsa,∗, J. Takácsa, A. Lovasa, J. Beltba Budapest University of Technology and Economics, Muegyetem rkp. 3-9, Budapest H-1111, Hungary

b SOLVAY Flour und Derivate GmbH, Germany

Abstract

Aluminium brazing is a method used for joining Al-parts by applying liquid alloys which have lower melting points than aluminium. Themain task during Al brazing is the successful elimination of the oxide layer which forms on the free Al surfaces even at low oxygen partialpressures. This layer is continuous resulting in a poor wetting ability of the surfaces to be joined. Hence, the elimination of this oxide layer isinevitable for successful brazing. This can be realised by the application of suitable flux material dissolving the oxide layer, simultaneouslyensuring temporal protection against the reconstruction of oxide layer during the heating period. In this paper, the CO2 laser beam is appliedas an energy source which yields rapid local surface melting. However, the absorption ability of Al is very low around the characteristicwavelength of this energy beam. In order to improve the absorption ability, the composition of NOCOLOK® flux was properly modified bySi powder addition. The absorption ability of the flux-covered surfaces was determined indirectly by measuring the emission coefficient atthe wavelength of CO2 laser (10.6�m). On the basis of surface emission measurements, the flux-composition can be successfully tailoredaccording to the technical requirement dictated by the joining of tubes or sheets with different diameter and wall-thickness.© 2003 Elsevier Science B.V. All rights reserved.

Keywords: Joining; Brazing of aluminium; Absorption

1. Introduction

The great variety of applied laser brazing process is con-nected with the different absorption ability of the irradiatedmetals and alloys around the characteristic wavelength ofapplied laser. For example, CO2 laser cannot be used forbrazing of Al and its alloys. While the absorption ability forsteels is appreciable at 10.6�m (∼10%), this value is verylow for Al (lower than 2%)[1].

The absorption ability can be raised by applying appro-priate coatings (flux). The flux is a two-component eutecticsystem (KF+ AlF3) having lower melting point than thatfor the eutectic Al–Si filler alloy. Consequently, the flux ismelted at first and dissolves the aluminium oxide beforethe melting of the filler material has started. In this way,the eutectic joint (formed from the melted filler material) isprotected against oxidation during the whole process. Theproperties of NOCOLOK® Flux produced by SOLVAY(composition of KAlF4 + K2AlF5 · H2O salts at the ratioof 13:1) fit to the outlined requirements. In the case ofNOCOLOK® Sil Flux, Si powder is mixed into the conven-tional flux (the maximal diameter of the powder particlesis 1�m) [2]. The mechanism of Al2O3 dissolution remains

∗ Corresponding author. Present address: Budapest University of Tech-nology and Economics 1111, Budapest, Børtalan 1, u.2, Z.ep.608,Hungary.

the same. In this case, the formation of eutectic joint isthe consequence of interaction between Si particles and theoxide-free Al surface, as illustrated inFig. 1.

The brazing process can be described as follows. Theflux is melted at 577◦C as the sample is heated, dissolv-ing the Al2O3 surface layer. A melting point depressionthrough Si-dissolution is thermodynamically dictated ac-cording toFig. 2, resulting in a thin eutectic melt betweenthe surfaces to be joined. This melt represents the fillerlayer.

The eutectic Al–Si melt can also be produced by the di-rect addition of filler alloy (Al/Si, 12 wt.%). After the com-pletion of the brazing process, the surface is covered by theremaining part of the flux (no need to remove).

2. Experiments and results

2.1. Absorption ability and compositional changesat the surfaces

Direct determination of absorption ability is not possible.Therefore, the relative emission of the aluminium surfacewas measured using a thermovision system. As transmissionis absent, the absorption and the emission is equal (aλ,T =ελ,T ) according to the Kirchoff’s law[3]. In Fig. 3, therelative emission coefficients as a function of temperature are

0924-0136/$ – see front matter © 2003 Elsevier Science B.V. All rights reserved.doi:10.1016/S0924-0136(03)00310-8

652 T. Markovits et al. / Journal of Materials Processing Technology 143–144 (2003) 651–655

Fig. 1. The NOCOLOK® aluminium brazing process using Sil Flux[2].

Fig. 2. Al–Si phase diagram[2].

Fig. 3. Relative emission coefficient as a function of temperature (coating conditions: 15% flux, dried in air).

plotted for samples covered by suspension (flux concentra-tion 15%). The curves were obtained by continuously heat-ing the sample from room temperature to 700◦C, and thencooling it below 100◦C.

A gradual increase of emission was found in every case,especially during the heating run. In principle, the drasticchange in the relative emission coefficient [ε(T)] is not ex-pected if phase transformation or compositional change isabsent. Obviously, this is not the case in our experiments.The increase in emission is connected with compositionalchange. The slope ofε(T) curves is especially high duringthe first period of heating, due to the dehydration in the fluxlayer. On the other hand, the slope of the heating curve de-creases as the concentration of flux suspension in the layeris increased. This tendency is verified on several samples,changing systematically the flux content between 5 and 20%.

Contrary to the tendency observed during the heating runs,the character ofε(T) curves is very similar in the coolingperiod. (The value ofε at room temperature is nearly equalirrespective of the initial flux content, which hints at thecompositional similarity of flux layers after drying and de-hydration at high temperature.) The absorption coefficientcan be further improved by raising the Si-content of flux.

Besides dehydration, additional transformations andchemical reactions take place on the sample surface. Inthe case of flux coated sample, the Al2O3 surface layer isdissolved by the molten flux. In the case of silicon powderaddition, an eutectic Al–Si surface layer is formed whenthe Al2O3 is already removed by the flux. Then, Si particlesdissolve into the Al surface, causing compositional change.As the composition is modified near the surface, it mayalso contribute to the increase in surface emissivity at theinvestigated wavelength.

T. Markovits et al. / Journal of Materials Processing Technology 143–144 (2003) 651–655 653

Fig. 4. The sample arrangements during the experiments: (a) sheets, (b) T-joints and (c) tube.

2.2. Laser brazing experiment

The experiments have been performed with OPL 1800type of CO2 laser equipment. The input laser power was upto 1.5 kW. Sample arrangements for the brazing of plates andtubes are illustrated inFig. 4. After mechanical flattening,burring-off, and degreasing in acetone, the samples werecladded, using NOCOLOK® Flux or Sil Flux. The claddingwas carried out by placing the flux-dispersion on the surfaceusing a dropper. Filler wire was also added in several cases.Addition of brazing paste is necessary for the successfulbrazing of tubes.

2.3. Testing of joints

The quality of joint is acceptable if the gap betweenthe surfaces is completely filled up by the liquid al-loy (Fig. 5a). Comparing the power densities requiredfor the brazing of 1–3 mm thick sheets, one can es-

Fig. 5. Micrographs of laser brazed sheet samples: (a) butt joint and (b) T-joint.

tablish that the necessary power density is lower forthe 3 mm thick sheets than expected. This is a conse-quence of the abundant quantity of filler material.Fig. 5bshows the micrograph of the T-joint brazed by the laserbeam.

2.3.1. Microhardness resultFig. 6 shows a scan of microhardness testing across the

brazed seam formed between two 2 mm thick aluminiumsheets.

The y-axis is placed in the middle of the joint. It can beseen that microhardness around the joint (within the heat-affected zone) is lower as a consequence of the local soft-ening. Though this softening phenomenon is harmful, theapplication of an additional (post) heat treatment can beconsidered as a compensation of this softening effect. Onthe other hand, significant hardening can be detected in thebrazing zone, which is the consequence of harder eutecticstructure.

654 T. Markovits et al. / Journal of Materials Processing Technology 143–144 (2003) 651–655

Fig. 6. The change of microhardness across the brazing seam (2 mm thick sheet).

2.3.2. Tensile testTensile tests do also supply important information on the

softening phenomenon arising from the recrystallisation tak-ing place in the heat-affected zone. After the brazing of1 mm thick sheets, tearing occurs in the basic material in thevicinity of the joint, so the strength of joint is satisfactory. InFig. 7, the tensile strength is plotted for the brazed and forthe “as-prepared” Al sheets 1–3 mm thick. The 1 mm thicksheet is not softened due to the laser heat input. On the con-trary, the strength of 2 and 3 mm thick sheets is lowered andthe probes were torn in the heat-affected zone.

Fig. 7. Tensile strength of pure aluminium sheets before and after thebrazing.

Fig. 8. Photo of the torn 1 mm thick aluminium sheet.

Fig. 9. Tensile diagram of the 1 mm thick aluminium sheet.

The tensile strength decrease is the result of the localsoftening. The use of high power density is inevitable for thesuccessful brazing of 2–3 mm thick sheets. Unfortunately,the applied high energy density causes an intensive heatshock near the joint.Fig. 8shows the picture of the torn 1 mmthick aluminium sheet. The appropriate tensile diagram isplotted inFig. 9 (the tensile stress is about 112 MPa).

3. Conclusions

• Using NOCOLOK® Flux or Sil Flux materials the laserenergy can penetrate into the aluminium. In this way, braz-ing of aluminium can be achieved.

• The emissivity (so the absorption ability) is improved byincreasing the flux concentration of the layer.

• As a consequence of chemical reactions and phase trans-formations, the run ofε(T) curves is different in the heat-ing and cooling periods.

T. Markovits et al. / Journal of Materials Processing Technology 143–144 (2003) 651–655 655

• The heat induced softening is negligible in the case of1 mm thick pure aluminium sheets. On the contrary, signif-icant softening (10–20%) can be detected after the brazingof already 2 and 3 mm thick pure Al, or Al(Mg) sheets.

• The frequently used brazing positions (butt- andT-joining) can be easily realised by using this laser pro-cedure.

Acknowledgements

This work has been supported by SOLVAY Fluor undDerivate GmbH, Germany. The authors are grateful for the

financial support of the OTKA (Hungarian Scientific Re-search Fund), project no. T035041.

References

[1] Gy. Kiss, A. Sklánitz, A. Szilágyi, J. Takács, Surface treatment andmodification by laser beam, Bewertung, Behandlung und anwendungvon Konstruktionswerkstoffen in Fahrzeugbauteilen VII, Kolloquium,Dresden, 1990, pp. 1–15.

[2] The NOCOLOK® Flux brazing process, Promotion Paper, SOLVAYFluor und Derivate GmbH.

[3] G. Rudowski, Az infratelevızió és alkalmazásai, Muszaki Könyvkiadó,Budapest, 1982, p. 226.