Design and characterization of novel iron‐based amorphous brazing foils based on thermodynamic predictions

Amorphous brazing foils facilitate the brazing process due to their high flexibility. Nickel‐based amorphous brazing foils are already in industrial use. However, as the raw material price for nickel is continuously increasing, more cost‐efficient iron‐based amorphous brazing foils are currently the focus of research. In this study, three alloys in the quinary system iron‐nickel‐chrom‐silicon‐boron will be investigated, which, according to previous predictions of newly developed thermodynamic databases, should exhibit an increased glass forming ability. The production and the characterization of the new iron‐based foils will in turn serve to validate the newly developed thermodynamic databases. With one of the alloys, a new iron‐based partly amorphous brazing foil can successfully be produced, but the other two alloys are too brittle for the brazing process. The results of this study not only show that the newly developed thermodynamic database of the quinary iron‐nickel‐chrome‐silicon‐boron alloy system can provide suitable prediction on the microstructure and melting behavior of the corresponding alloys, but also that it is possible to produce a quaternary iron‐based partly amorphous foil.


| INTRODUCTION
In order for materials to meet the ever-increasing requirements in terms of efficiency, environmental compatibility and material costs, alloying concepts must be continuously adapted and improved.This also applies in particular to brazing alloys, because brazed joints must also be able to withstand the same mechanical, thermal and corrosive stresses which the base material is subjected to.For high temperature brazing, nickel-based brazing alloys are commonly used because of their favorable properties for brazing.Nickel-based brazing alloys are very suitable for brazing joints that need to meet the high requirements in terms of corrosion resistance and high-temperature strength.They are therefore frequently used in steam, gas or aircraft turbine applications, but also in nuclear, space and reactor technology [1].However, since the raw material price of nickel is increasing rapidly, research into the less widely used iron-based brazing alloys is of great interest [2].
It has already been proven, that iron-based brazing alloys can offer advantages.Further studies have shown, for example, that iron-based ledeburite brazing alloys are particularly suitable for joining high-speed steels and, due to their similar chemical composition, ensure that the interface between the brazing gap and the base material is hardly recognizable [3].In order to replace nickelbased brazing alloys in exhaust gas coolers, the two ironchrome-based brazing alloys Fe-20Cr-43Ni-10P and Fe-20Cr-20Ni-8P-5Si-2Mo have been developed [4].However, these are more susceptible to corrosion than the Ni-29Cr-6P-4Si, which was used here as a comparative brazing alloy for joining X5CrNi18-10.
Both iron-based and nickel-based brazing alloys have included melting point depressants such as boron, phosphorous or silicon to lower the melting point so that the substrate is thermally influenced as little as possible.These elements support the formation of hard brittle phases in the form of borides, silicides or phosphides, which do not allow the production of brazing foils by cold-or hot rolling [5,6].For this reason, brazing alloys with melting point lowering elements were for a long time only available as powders.The disadvantages of powders are the use of organic, environmentally harmful binders and the high material consumption.Both can be significantly reduced by using brazing foils.Brazing material savings of up to 50 %-66 % are possible compared to pastes as a powder binder mixture [6,7].Amorphous brazing foils, which are flexible due to their microstructure can be easily adapted to a wide range of component geometries and have proved particularly successful, especially in aerospace engineering [7].Various nickel-based amorphous brazing foils out of the system nickel-chrome-silicon-boron can already be found in many application areas such as aerospace [8].
In the 1960s, an Au 75 Si 25 metal with a non-crystalline structure, which was produced by rapid solidification, was discovered [9].After that, many metallic glasses have been discovered among others also iron-based metallic glasses [10].The amorphous microstructure differs from the crystalline one.There is no three-dimensional periodically arranged lattice, like in a crystalline structure.The atoms are randomly distributed without long-range order, comparable to the structure of glasses, in the frozen state [11].The lack of typical crystalline defects like grain boundaries or dislocations ensures outstanding mechanical and corrosive properties [12,13].Since an amorphous microstructure can technically not be produced with all combinations of alloying elements despite a rapid cooling rate of _ T = 10 5 K s À 1 , a variety of parameters and formulas have been established to estimate the tendency of an alloy to form a glassy structure [11].All in all, the following empirical rules have emerged [14]: Another important factor is the eutectic principle.The probability of an alloy to be amorphous increases if it has a eutectic or near-eutectic composition with a low melting point, which impairs diffusion and thus nucleus bonding [15].
The development of iron-based amorphous brazing foils should combine cost-efficiency and flexibility to overcome problems of expensive nickel-based brazing foils and oxidation of iron-based powder.In regard to nickel-based amorphous brazing foils, chrome and nickel are chosen as additional elements to enhance corrosion resistance and silicon and boron should perform as melting point depressants [8].Since amorphous alloys consist of multicomponent systems, it is hardly possible to find compositions with a high glass forming ability by trial and error.Therefore, simulation techniques are used.Lately, novel cost-efficient iron-based amorphous ribbons which can be used as thermal barrier coatings in thermal spraying have been developed, using simulation techniques.Pseudo-ternary diagrams for the quinary system iron-niobium-silicon-carbon-boron were plotted, in which eutectic compositions can be located by interpolating isotherms with the Thin Plate Spline method.Thermal data of ten quinary alloying compositions were used for the interpolation in the pseudo-ternary system [16].These studies show that using simulation techniques can help to identify possible locations of high glass forming ability in multicomponent iron-based systems.
Another method to avoid a large number of trials, is the CALPHAD (CALculation of PHAse Diagram) -method.
The CALPHAD method combines experimental and theoretical knowledge of thermodynamic and physical chemistry to model the constitution of complex chemical systems as a function of temperature, composition and pressure.The backbone are Gibbs free energy models to be formulated for all phases.The free model parameters are then carefully fitted to reproduce critically assessed experimental data at best.Eventually, the CALPHAD thermodynamic databases represent a self-consistent set of model equations for the Gibbs free energies which can be used along with computational software tools for alloy design, thermodynamic property prediction and more [17,18].The CAL-PHAD method was successfully used in further studies to adjust brazing process parameters in the brazing of nickel-based brazing alloys in such a way that the formation of brittle phases is minimized [19].The challenge of using these databases is that there are no databases specifically designed for brazing.Therefore, for nickelbased brazing alloys, the database of nickel-based superalloys, which contains phase diagrams of similar elements, is most commonly used.the crucial problem here is the boron content in brazing alloys, which is not considered in this database and thus causes deviations between simulation and experimental results [19].In a further study, the CALPHAD-method was successfully used to predict the development of phases during the brazing process of nickel-based superalloy Inconel-718 with the BNi-2 brazing alloy [20].Both studies show that simulation techniques are also relevant for the brazing process, especially for predicting phase development during and after heating processes.
Most amorphous brazing alloys have eutectic or neareutectic compositions [14].The probability of identifying eutectic regions in a multicomponent system increases with the number of binary and ternary systems that are fully thermodynamically described [21].The quinary database used in this study is unpublished and will not be published entirely.It was used however for computing several experimental alloy compositions with the software Thermo-Calc, among them the alloys presented in this work [22].As a result, each constituent binary and ternary database were revised, re-modelled and incorporated into a high order quinary iron-nickel-chromium-silicon-boron database, Table 1.In particular, available thermodynamic descriptions as well as published data on phase equilibria in ternary iron-nickel-silicon system were scarce, calling for additional experimental study.The experimental work dedicated to complement thermodynamic description iron-nickel-silicon system by means of key experiments and CALPHAD method is presented in details in our earlier publication [32].
The motivation of this study is to characterize thermal and microstructural properties of three computed alloys in the quinary system iron-nickel-chrome-siliconboron, that have been calculated using the new thermodynamic database that is unpublished.The database predicts the investigated alloys to be eutectic, so the glass forming ability should be high in regard to Turnbull's criterion and the alloys can produce an amorphous or partly amorphous structure by using the melt-spin-process [15].The data obtained will be used to validate and adapt the new developed thermodynamic database.

| MATERIALS AND METHODS
Based on the results of the completed thermodynamic modelling for the ternary system iron-nickel-silicon the authors have further elaborated a proprietary thermodynamic database for the quinary system ironnickel-chrome-silicon-boron [32].This database was used for computing several experimental alloy compositions, among them the three alloys presented in this work.The major objective was to keep the nickel-content of the alloys as low as possible and essentially below ω = 35 wt.%, i. e. χ � 30 at.%, while searching for low liquidus temperatures along eutectic reactions, as follows: Alloy 1 and Alloy 3 were selected along the eutectic reaction "Liquid $ FeSi + (Fe, Ni) 2 Si" with the two solid phases being silicides.Alloy 2 was selected along the eutectic reaction "Liquid $ FCC-Fe ss + M 2 B", the two solid phases being an iron-, nickel-rich solid solution with face-centered-cubic crystal structure and the (Fe, Ni) 2 B boride (Pearson symbol tI12, I4/mcm), respectively.The computed phase equilibria for all investigated alloys has been plotted as a function of temperature, Figure 1.For multicomponent systems this representation of phase equilibria is the method of choice, while graphical phase diagrams along predefined section planes are more instructive for ternary alloy systems.For iron-nickel-silicon such phase diagrams are given in further studies [32].
Noteworthy, these reactions are not invariant, but in contrary, well extended in composition space.Alloy selection was therefore targeting compositions with low nickel and low liquidus temperature.The alloy compositions are listed along with the computed liquidus temperature, Table 2.In order to verify this and validate the thermodynamic database, these are to be produced and characterized.For the production, granulate (1 mm < d < 5 mm) with a purity of over 99.98 %, with the exception of boron (99.40 %) is used.
The raw input materials are melted in a ceramic boron-nitride-crucible under high-purity argon atmosphere T A B L E 1 List of binary and ternary systems that could be derived out of the quinary system iron-nickel-chrome-siliconboron.

Binary system
using R50-250-13 tube furnace from Nabertherm GmbH.The materials are heated up to T = 1,250 °C and homogenized at this temperature for t = 20 min.The ironbased brazing foils are then produced with the Melt-Spinner SC from Edmund Bühler GmbH.The argon atmosphere is set to a pressure of p = 0.6 bar.For all investigations, the same set of process parameters is applied, Table 3.
Differential scanning calorimetry (DSC) is performed with DSC/TG system SETARAM SETSYS Evolution from Setaram Instruments to identify phase transformation temperatures, especially a transition temperature between amorphous and crystalline states.DSC experimental procedure includes two heating and cooling cycles with heating and cooling rates of _ T = 0.10 K/s À 1 and _ T = 0.15 K/s À 1 respectively.Due to the high cooling rate in the melt-spinning process, the microstructure of the foil inherits strong micro-segregation profile, while the phase transformation path deviates from equilibrium.As a result, the solidus and liquidus temperatures determined during the first heating cycle in differential scanning calorimetry analysis can be affected by chemical inhomogeneity and the presence of metastable phases.Therefore, the heating curve from the second heating cycle to experimentally determine the solidus and liquidus temperatures are used.The iron-based brazing foils with mass m~0.005 g are placed in a ceramic aluminiumoxide-crucible and heated to a maximum temperature of T = 1,250 °C during a heating stage.Microstructure analysis of the brazing foils is performed using scanning electron microscope (SEM) Ultra 55 from Carl Zeiss AG operating in back-scatter-electrone mode with accelerating voltage of U = 15 kV.In addition, the x-ray diffractometer C3000 from GE Energy Germany GmbH operating with a voltage of U = 40 kV, current of I = 40 mA and radiation source of cobalt-Kα is used for microstructure analysis.In order to compare the actual composition with the nominal composition and to determine possible deviations, a wet chemical analysis is carried out by Elektrowerk Weisweiler GmbH.

| RESULTS AND DISCUSSION
The following chapter presents the results of melt-spinning-process and characterizing of the three different alloys.The characterization includes melt-spinning-process, thermal analysis, microstructure analysis and comparison between experimental and calculated values.

| Melt-spinning-process
The prepared iron-based foils have a thickness of d F � 50 μm, Figure 2. Optically Alloy 1 appears to develop a crystalline structure under the applied melt-spinningprocess parameters leading to brittle behavior.The foil from Alloy 1 cracks directly by bending, Figure 2a.Based on its elasticity and lack of evidence of transversal cracking during deformation by bending, Alloy 2 is most likely to have an amorphous microstructure, Figure 2b.However, the foil exhibits post-manufacturing longitudinal cracks and its width does not correspond to the width of the slit die of approx.d s = 10 mm.Nevertheless, it has a foil-like appearance, while Alloy 3, has a granulate-like appearance and is brittle, Figure 2b, c.Here it can already be assumed, without further investigations, that no amorphous structure could form under the given process parameter using the melt-spinning process.

| Thermal analysis
Based on its brittle behavior Alloy 1 solidified likely with a crystalline structure under applied melt-spinning process conditions.The result of differential scanning calorimetry shows no evidence of low temperature exothermic reaction, typical for transition from amorphous to crystalline state, Figure 3.A foil-like appearance of Alloy 2, its elastic behavior and absence of transversal cracking during deformation by bending, suggests a partly amorphous structure.Indeed, differential scanning calorimetry curve after the first heating cycle shows a characteristic exothermal reaction at ca. T exo = 493 °C produced by a bulk solid-state transformation into a crystalline microstructure, Figure 3 [34].Such an exothermic peak is not present in differential scanning calorimetry curve after the second heating cycle.Melt-spinning of Alloy 3 produced granulate-like pieces clearly indicating brittle, nonamorphous microstructure being unsuitable for brazing application.
The CALPHAD-computed solidus temperature corresponds well to differential scanning calorimetry measurement of Alloy 2, while the experimental liquidus temperature is lower than the computed one, Table 4. Nevertheless, a eutectic composition was predicted and Alloy 2 seems to have a partly amorphous microstructure which validates the criterion that eutectic and near-eutectic alloys have a high glass forming ability.Meanwhile solidus temperature of Alloy 1 shows stronger discrepancy calling for further work towards optimization of the CALPHAD database.Note that for Alloy 1 and Alloy 2 the temperatures were computed using experimentally determined composition.

| X-ray diffraction analysis
The x-ray diffraction spectrum of all alloys is plotted, Figure 4. Alloy 1 shows a variety of peaks which partially overlap, Figure 4a.This makes it difficult to detect the exact phases existing in Alloy 1.While the spectrum hardly matches diiron-silicide and dinickel-silicide, it fits very well with iron-silicide, so this is an existing phase in this alloy.This phase was also predicted by the thermodynamic database for the eutectic reaction.Due to the low chromium content, chromium should be present are also possible phases that can occur in Alloy 1 in a very small amount, as has already been shown by experimental studies of the ternary system iron-nickel-silicon [32].Alloy 2, which could have a partly amorphous or nanocrystalline structure, shows significantly less peaks than the spectrum of Alloy 1, Figure 4b.The identified peaks fit quite well with the face-centered-cubic nickel, which indicates that Alloy 2 is partly crystalline.Nevertheless, the absence of sharp reflections and the broad peaks are a further hint of the existence of a partly amorphous microstructure within Alloy 2 [35].Since the peak is not sharp due to the partial amorphous structure, several peaks may overlap, so that the further phases, that have been predicted face-centered-cubic iron and diiron-boride as well as dinickel-boride, cannot be identified exactly.

| Secondary electron microscopy analysis
The microstructure of the sample from Alloy 1 consists of two distinct morphologies which solidified with dendritic and eutectic morphologies, Figure 5.Although the microstructure of Alloy 1 did not solidify with amorphous structure, it suggests that the new CALPHAD-database is capable to predict low temperature eutectic alloy compositions.Further screening of process parameters for melt-spinning-process is needed aiming towards amorphization of Alloy 1.
Under the same process condition for melt-spinningprocess, a foil with composition of Alloy 2 achieved a nearly full amorphization with only single micro-scale crystalline area detected near the edge of the foil, Figure 6a, b.Formation of such crystalline area is possibly related to fluctuations of the cooling rate during the melt-spinning-process caused by a contact between the alloy and copper wheel.
Nominal and measured compositions for Alloy 1 and Alloy 2 are compared, Table 5.As mentioned previously, the element of boron limits application of energy-   dispersive-spectroscopy technique and as a result the composition of Alloy 2 was determined using the wetchemical-analysis method.Because Alloy 3 has been the least successful, secondary-electron-microscopy/energydispersive-spectroscopy has not been performed.

| CONCLUSION
A novel iron-based partly amorphous brazing foil, Alloy 2, was calculated and successfully produced using new thermodynamic databases.The successful fabrication of the foil and the predicted melting temperatures, that are fitting to experimental data, validate the accuracy of the thermodynamic databases, which predicted eutectic compositions for the investigated alloys.The partly amorphous structure of Alloy 2 was evidenced by different measurement methods: * Thermal analysis shows exothermic peaks at temperatures of T RX1 = 401 °C and T RX2 = 493 °C, suggesting a solid phase transition from amorphous to crystalline.* Computed data for Alloy 2 fits well with regard to the solidus temperature, while the experimental liquidus temperature is lower than the computed one.The data will be used to validate and correct the prototype database.* X-ray-diffraction-measurements show crystalline structure but the reflections are broad with low intensities, which is also common for amorphous structures.Due to the broadness of the peak, phase identification is difficult.Probably face-centered cubic nickel/iron and dinickel-boride as well as diiron-boride are existing phases.* Secondary-electron-microscopy images show that a crystalline structure with grains is present only at the edges of the foil.The inner areas of the foil are homogeneous and do not show typical crystalline structural features.
Under the given process parameters, it has not been possible to develop an amorphous iron-based brazing foil with Alloy 1 and 3, probably due to the high stability of silicide phases.

| OUTLOOK
Alloy 2, which is the most successful should be produced by melt-spinning with different process parameters to define an individual process window for this composition.The main aim is to vary the superheat, i. e. the difference between the homogenization temperature in the crucible and the melting temperature of the alloy.The higher the superheat, the higher the cooling rate and the probability of forming an amorphous structure.Other process parameters that can be varied are the speed of the copper wheel v and the distance between the copper wheel and the crucible d 1 .Thus, an individual process window for the production of an iron-based amorphous brazing foil can be defined for Alloy 2. With the new iron-based amorphous brazing foil, brazing tests can then be carried out, followed by an evaluation of the brazing joint in the form of wetting tests, investigation of the mechanical properties and microstructure analysis using secondary-electron-microscopy/energy-dispersivespectroscopy in order to generate a correlation between microstructure formation and mechanical properties.Based on the results, Alloy 1 and 3 will not be investigated further.Both have a very high silicon content and, according to thermodynamic predictions and x-raydiffraction measurements, form stoichiometric silicides, which counteract the formation of the amorphous structure.With the help of newly developed thermodynamic databases in the quinary system iron-nickel-chrome-silicon-boron, further alloy compositions with high glass forming ability will be calculated and produced using the melt-spinning process.This will validate and improve the thermodynamic databases and novel quaternary and quinary iron-based amorphous foils can be developed and subsequently characterized.Potential alloys considered in the future should have a significantly reduced silicon content compared to Alloy 1 and Alloy 3 and thus be oriented towards Alloy 2 in terms of chemical composition.The aim is to increase the chrome content in order to be able to join corrosion-resistant steels containing chrome.

T A B L E 2 Fe
Nominal compositions of experimental alloys computed by the proprietary database for the quinary system iron-nickelchrome-silicon-boron.The computed values of the liquidus temperature are listed in °C.

F I G U R E 2
Brazing foils from iron-nickel-chrome-silicon-boron-alloy-system produced by the melt-spinning process a) Alloy 1; b) Alloy 2; c) Alloy 3. F I G U R E 3 Differential-scanning-calorimetry-measurement of Alloy 1 and 2 for two heating cycles at heating rate _ T = 10 K/s À 1 .T A B L E 4 CALPHAD-computed and differential scanning analyzed solidus and liquidus temperatures for Alloys 1-2.The temperatures were calculated using experimentally measured composition for Alloy 1-2. of the solid solution phases.The second phase is probably a mixture of diiron-and dinickel-silicide.Nevertheless, different nickel-silicides like Ni 31 Si 12

F
I G U R E 4 X-ray diffraction-spectrum of the investigated iron-based brazing foils.

F I G U R E 5
Microstructure of the brazing foil produced from Alloy 1 -a).The microstructure consists of two distinct crystalline phases with high and low contrast, while two solidification morphologies i. e. dendritic and eutectic, are well distinguished.Magnified secondary electron microscopy image of the microstructure is shown in b).

F I G U R E 6
Microstructure of the brazing foil produced from Alloy 2 -a).The microstructure is nearly fully amorphous with only microsize crystalline area found near the edge of the foil shown in b).T A B L E 5 Nominal and measured composition in samples produced from Alloy 1 and Alloy 2. Note, the composition in Alloy 2 was measured by wet-chemical-analysis method due to limitations of energy-dispersive-spectroscopy technique to accurately measure light elements i. e. boron.