Brazing Fundamentals

CHAPTER 2

BRAZING does not involve any melting or of the liquid in equilibrium with its saturated plastic state of the base metal. Brazing com- vapor, γ at the interface between the solid and sl γ prises a group of joining processes in which the liquid, and sv at the interface of the solid in coalescence is produced by heating to suitable equilibrium with the saturated vapor of the liq- temperatures above 450 °C (840 °F) and by uid. Hence: using a ferrous and/or nonferrous filler metal γ = γ cos θ + γ (Eq 1) that must have a liquidus temperature above 450 sv lv sl °C and below the solidus temperature(s) of the It is important to keep in mind that phases are base metal(s). The filler metal is distributed supposed to be mutually in equilibrium. The between the closely fitted surfaces of the joint γ designation sv is a reminder that the solid sur- by capillary attraction. Brazing is distinguished face near the liquid should have an equilibrium from soldering in that soldering employs a filler film of vapor due to the film pressure. Young’s metal having a liquidus below 450 °C. equation has been used extensively in literature, Brazing has four distinct characteristics: which reflects its general acceptance. However, Eq 1 has never been verified exper- • The coalescence, joining, or uniting of an imentally. The problem is that surface tensions assembly of two or more parts into one struc- of solids are not easy to measure due to the ture is achieved by heating the assembly or inevitable presence of the interfacial tension the region of the parts to be joined to a tem- between a solid and its liquid. More impor- perature of 450 °C or above. tantly, there is the difficulty that any tensile • Assembled parts and filler metal are heated to stresses existing in the surface of the solid a temperature high enough to melt the filler would prevent the system from being in equilib- metal but not the parts. rium. The surface tension at the solid-vapor • The molten filler metal spreads into the joint interface (γ ) has a relationship with surface and must wet the base-metal surfaces. sv tension of a solid in vacuum (γ ) as follows: • The parts are cooled to freeze the filler metal, s which is held in the joint by capillary attrac- γ γ π sv = s – e (Eq 2) tion and anchors the part together. π where e refers to the spreading pressure. Con- sequently, Young’s equation may be rewritten Adhesion, Wetting, Spreading, as: γ γ θ γ π and Capillary Attraction s = lv cos + sl + e (Eq 3) Because most of the solids have a negligible Metals π θ e, particularly when the contact angle ( ) is More than 195 years ago, Thomas Young greater than 10°, Young’s equation becomes: (Ref 1) proposed treating the contact angle (θ) γ = γ cos θ + γ (Eq 4) of a liquid as the result of the mechanical equi- s lv sl librium of a drop resting on a plain, solid surface A decrease of the contact angle causes an under the action of three surface tensions (Fig. increase of the liquid drop surface area and thus γ 2.1). The surface tensions are lv at the interface increases the total liquid surface free energy. © 2003 ASM International. All Rights Reserved. www.asminternational.org Brazing (#06955G) 8 / Brazing, Second Edition

The total surface free energy of the solid between the phases to be bonded (as with con- decreases concurrently. A balance of these two taminated solid surfaces, such as absorbed car- forces results in a steady-state condition repre- bonaceous layers), γ is larger than γ and also γ sl sv sented by an acute contact angle. Mathemati- lv. Then, a large, obtuse contact angle forms cally, this balance is expressed as Young’s equa- and approaches 180° with decreasing attractive tion (Eq 1) acting at the periphery of the drop. forces, as indicated by the absence of reduction The driving force for wetting thus is (γ – of either the solid or liquid surface free energy γ sv sl). The balancing resisting force is represented and very weak adherence. An optimal 180° by the horizontal component of the surface ten- angle would indicate no attractive forces γ θ sion of the liquid ( lv cos ), as shown in Fig. between the two phases. 2.1. A balance of vertical forces also exists, but Technologically, a nonwetting liquid is vertical forces do not play a role in this problem. highly unfavorable with regard to the formation γ γ γ γ γ When sv < lv and sv < sl < lv (again, in the of an intimate interface, due to its lack of capa- absence of a reaction), a steady-state condition bility of penetration of surface and grain-bound- results with θ > 90°. With a decrease of the ary irregularities because of the lack of capillary obtuse contact angle, the liquid drop surface behavior. Also, the liquid does not distribute area (and thus the total liquid surface free itself uniformly. energy) also decreases. In either case of a starting acute or obtuse Young’s equation (Eq 1) represents a steady- contact angle, the characteristic of wetting can state condition for a solid-liquid interface in sta- be achieved and enhanced by a reaction at the ble or metastable thermodynamic equilibrium. interface at an elevated temperature. However, there is no definite indication of Young’s equation (Eq 1) can be modified to whether chemical or van der Waals bonding represent a spreading coefficient, S1 (Ref 2), or exists, other than that the contact angle is gener- a work of spreading, Ws, by taking the extreme ally smaller with chemical bonding versus van case of wetting when the contact angle γ θ der Waals bonding (because when sl is smaller, approaches 0 and cos approaches 1. The re- the driving force for wetting is greater). sisting force to the extension of the drop then is In either wetting (γ > γ ) or nonwetting (γ γ , as discussed earlier. Young’s equation can γ sv lv lv lv > sv), if an intimate interface does not form then be expressed as: γ γ γ Ws = sv – sl – lv (Eq 5)

In order to have spreading occur, Ws has to be positive. Under these conditions, the driving γ γ γ force for wetting ( sv – sl) is greater than lv. If the Ws is negative, then the driving force for wetting is smaller, and spreading does not occur, but an acute angle forms. If a reaction occurs in which the substrate is an active participant, then the free energy of the reaction, ∆G /dAdt, contributes to the driving force for R γ wetting, which practically always exceeds lv. Spreading thus occurs. Young’s equation (Eq 1) for a nonreacting steady-state sessile drop can be modiﬁed to include the contribution of the free energy of reaction:

γ + G γ sl R ≥γ sv –lv cos q (Eq 6) dAsdt

Sessile drop conﬁgurations: (top) wetting, and (bot- Fig. 2.1 γ γ tom) nonwetting. sv and lv, surface tensions and The free energy required for the increase of the surface free energies of the solid-vapor and liquid-vapor, respec- γ surface area of the drop as the perimeter ex- tively; sl, interfacial energy of the solid-liquid; –dGR/dA · dt, free energy of reaction pands provides the only resisting force to the © 2003 ASM International. All Rights Reserved. www.asminternational.org Brazing (#06955G) Chapter 2: Brazing Fundamentals / 9

γ γ expansion. It can be shown thermodynamically ior corresponds to sv > lv, because, in a given that in the absence of a reaction, the driving system, the surface free energy of a liquid is less γ force for wetting does not exceed lv, resulting than that of a solid, due to its lack of long-range in a steady-state contact angle (Ref 3). The driv- order. The liquid thus has the opportunity to ing force with the contribution of the free rearrange its surface structure to a lower free- energy of reaction in most cases exceeds the energy state. However, when liquid C is placed γ θ resisting force represented by lv, because is on solid A, spreading occurs, because substrate 0° during spreading. A condition of an expand- A (as an active participant in the reaction) ing drop during a reaction is deﬁned as spread- changes its surface composition toward B. The ing. It can be seen that the free energy of a reac- third equation in Fig. 2.1 applies in this case. tion in which the substrate is a passive Another example is that of liquid D on solid participant does not contribute to the driving B. Liquid D is not in equilibrium with B and dis- force for wetting; thus, spreading does not solves some of the substrate to change its com- occur. The contact angle, however, adjusts to position to C. Even though a reaction occurs, conform with the surface-energy changes of the there is no spreading, because B is a passive par- liquid caused by composition changes due to the ticipant with no change in composition, even reaction. though it is being dissolved. However, with liq- Example: Copper-Silver System. The uid D on solid A, spreading occurs, because equilibrium phase diagram for the copper-silver both are active participants as they change to binary system (Fig. 2.2) can be used to illustrate equilibrium compositions C and B, respec- examples of wetting and spreading (Ref 3). The tively. In both of the latter examples, liquid D is system has a eutectic at 780 °C (1430 °F), with an active participant, because it dissolves some 72 wt% Ag. At 900 °C (1650 °F), the solid-solu- of the substrate to reach equilibrium composition limit is 5 wt% Ag in copper and 8 wt% Cu tions. It does not, however, contribute to spread- in silver. ing, which is controlled by the active participa- Several compositions are identiﬁed in the tion of the substrate. phase diagram by the letters A to D. When a drop of liquid C is placed on solid B at 900 °C (1650 °F), wetting occurs, with a contact angle Ceramics of 11° and no chemical reaction, because the Joining dissimilar materials invariably results phases are in chemical equilibrium. This behav- in high interfacial energy; that is, the work of

adhesion is not sufficient to maintain the joint problems, because interfacial compound forma- integrity. Most structural ceramic-metal inter- tion could create fragile layers (Ref 9), further faces are no exceptions; the liquid metal does complicating the development of good joint not readily wet the ceramic surface. Therefore, efficiency. The weakest among the adhesive the first challenge in metal-to-ceramic joining is strength at the ceramic-compound boundary, to alter the interfacial thermodynamics to render the cohesive strength of the compound, or the the ceramic surface wettable. There are two adhesive strength between compound and metal approaches available today (Ref 4) to accom- would determine the final joint strength. plish this task: metallization of the ceramic sur- Table 2.1 (Ref 10) summarizes the various face and reaction wetting. A thin layer of metal reaction products that have been identified in lit- alloy is deposited on the ceramic by vapor dep- erature for the common metal-ceramic systems. osition or sputtering. Brazing is then carried out Many researchers have concentrated their study by appropriate filler material or by simply melt- on alumina surfaces, because it is one of the ing the deposited layer (Ref 5). This two-step very few ceramics for which essential thermo- approach is not as enthusiastically embraced by dynamic data are available. One of the first industry as is direct reaction brazing. commercial applications of metal-to-ceramic In reaction brazing, the filler metal is care- brazed components is the turbocharger rotor fully chosen so as to facilitate compound for- (Ref 8), where a silicon nitride turbine blade is mation at the interface. A small percentage of brazed to a stainless steel shaft. Other ceramics reactive metals, such as aluminum and titanium, of interest to brazers are silicon carbide and zir- are added to the otherwise inert base alloys (Ref conia. 6, 7). The compounds that form are commonly Ceramics exhibit very different thermal spinels for the oxide ceramics and complex expansion behavior compared to metals; hence, nitrides for the ceramic nitrides (Ref 8, 9). considerable residual stress can build up during It is important to realize that wetting in such cooling. This thermal expansion mismatch systems is time dependent. Successful bond for- more or less dictates the use of a ductile filler mation relies on rapid transport of the reactive material. Most commercial brazing systems are metal to the interface and a rapid rate of com- therefore silver and copper base. The soft inter- pound formation. Reaction wetting may not layer might not be sufficient to compensate for be the solution to all metal-ceramic joining large differences in thermal expansion coeffi-

Table 2.1 Ceramic-metal interface formation and reaction products

Bonding conditions

System Temperature, K Load, MPa (ksi) Time, h Atmosphere Reaction products

Al2O3(a)-Nb(a) 1925 20 (3) 0.1 Argon (Al)Nb (b)-Nb(c) –4 Pa (O)Nb Al2O3 1973 10 (1.5) 2 Vacuum 10 –3 Al2O3(d)-Nb(a) 1973 6.4 (1.0) 1 Vacuum 10 Pa NbOx, (Al,O)Nb –3 Al2O3(b)-Nb(a) 1973 6.4 (1.0) 1 Vacuum 10 Pa NbOx, (Al,O)Nb –4 Al2O3(e)-Ti(a) 1250 . . . 60 Vacuum 10 Pa Ti3Al, TiO, (AlO)Ti –3 SiC(b)-Ti(a) 1773 0.34 (0.05) 1 Vacuum 10 Pa Ti3SiC2, Ti5Si(C), TiSi2 SiC(b)-Zr(a) 1773 0.56 (0.08) 1 Vacuum 10–3 Pa ZrSi, ZrC + (Si)Zr SiC(a)-Al/Ti/Al(a) 1273 0.56 (0.08) 1 Air TiC, TiAl3Si –2 SiC(a)-Al(a) 1373 0 (0) 1 Vacuum 10 Pa Al4C, (Si)Al Si N (a)-Fe(a) 3 4 1683 3000 (435) 1 Argon Fe3Si, (Si)Fe (a) –4 SiO2-Al 875 10 (1.5) 10 75N2/25H2 Not observed 2) α-Al O , (Si)Al ...... 24 (H2/H2O = 10 2 3 Al O (b)-Cu(a) × 5 2 3 1270 20 (3) 0.25 H2/H2O = 2 10 None (b)-Pt(a) 5 Al2O3 1773 20 (3) 0.25 H2/H2O = 10 None. After 6 h, (Al)Pt; after 1000 h, Pt3Al (b)-Pt(a) × 5 Al2O3 1773 20 (3) 0.25 H2/H2O = 5 10 Pt3Al, (Al)Pt (a) × 3 SiO2-Pt 1473 1 (0.15) 0.25 H2/H2O = 2 10 Pt3Si (a)-Pt 1273 0 (0) 0.6 H (Zr)Pt ZrO2 2 ...... 8 H Pt Zr, (Zr)Pt 2 3 Plane of interaction: (a) Polycrystal. (b) (0001). (c) (110). (d) (1010). (e) (1100). Source: Ref 10 © 2003 ASM International. All Rights Reserved. www.asminternational.org Brazing (#06955G) Chapter 2: Brazing Fundamentals / 11

cients (e.g., Si3N4 as compared to stainless metal to be drawn into the area that covers the steel). In such situations, laminated interlayers parallel surfaces that are to be brazed. that provide a continuous gradient thermal Capillarity is a result of surface tension expansion coefficient are used (Ref 6). between base metals(s), filler metal, flux, or Thermodynamic phenomena that occur at the atmosphere and the contact angle between base interface can be studied in terms of the contact metal and filler metal. In actual practice, filler- angle, θ, and the work of adhesion, W. These metal flow characteristics are also influenced by terms can be related to various surface or inter- dynamic considerations involving viscosity, facial energies. The general case for a liquid vapor pressure, gravity, and metallurgical reac- metal in contact with a solid ceramic is shown in tions between filler metal and base metal. Eq 1, where a balance of surface tension forces As a matter of fact, present-day brazing prac- results in the familiar Young’s equation. tices have evolved as the result of an empirical The Dupree equation is easily derived from approach to the phenomena of wetting and Eq 1: spreading, which are of prime importance in the formation of brazed joints. Classical, physical, W = γ (1 + cos θ) (Eq 7) lv and chemical principles led to equations gov- Attempts at understanding the nature of the erning the shape of liquid surfaces and the rate force of adhesion across the interface have not of filling a capillary gap in systems that do not been very successful. In 1965, researchers (Ref interreact. However, the extension of theory to 11) rationalized, on the basis of the work of practical systems necessitates the consideration adhesion data for an alumina-metal interface, of a number of complicating factors, which that the observed work of adhesion was the sum often arise in everyday practice. A few of these of two independent contributions arising from factors include the condition of the solid surface the van der Waals forces and a primary chemi- as to the presence of oxide films and their cal bond. effects on wetting and spreading, surface rough- Predicting adhesion data in joining an alu- ness, alloying between the filler metal and base mina-metal interface is of great importance in metal and the extent to which this affects the many applications. The objective of proposed thermodynamic properties of the liquid and research to predict wettability and bond strength solid surfaces, and the condition and properties from measurable parameters and bridge the gap of the brazing atmosphere. between a theoretical understanding and tech- The factors that control the rate at which wet- nology of observed work was undertaken (Ref ting, spreading, and capillary flow occur are of 12). great practical, as well as theoretical, interest. Researchers (Ref 13, 14) attempted to explain Studies have indicated profound influences of the entire work of adhesion across the metal- various kinds of surface activation that cannot ceramic boundary in terms of physical forces be explained in terms of surface energies or using the dielectric principle. Such models are alterations in equilibrium contact angle (Ref 15, not of much use to the brazing industry, because 16). Some of the most spectacular of these most commercial metal-to-ceramic bonds are effects have been observed in systems in which based on chemical bond formations. a finite contact angle is thermodynamically unstable, because the solid-vapor surface Effects of Capillary Attraction and energy exceeds the sum of the liquid-solid surface energies—that is, a system in which ther- Wetting on Brazing modynamics would predict complete spreading. Capillary attraction makes leak-tight In actual fact, spreading may or may not occur joints a simple proposition for brazing. In a in this type of system, and the rate of spreading properly designed joint, the molten filler metal can be markedly dependent on surface chem- is normally drawn completely through the joint istry, although the fundamental mechanisms of area without any voids or gaps, and brazed this dependence are not all clear. joints remain liquid- and gas-tight under heavy Wetting is, perhaps, best understood by pressures, even when the joint is subjected to example. If a solid is immersed in a liquid bath shock or vibrational types of loading. and wetting occurs, a thin, continuous layer of Capillary action results in the phenomenon liquid adheres to the solid when it is removed where surface tension causes molten braze filler from the liquid. Technically speaking, in the © 2003 ASM International. All Rights Reserved. www.asminternational.org Brazing (#06955G) 12 / Brazing, Second Edition

wetting process, the force of adhesion between Good wetting and spreading of the liquid the solid and the liquid is greater than the cohe- filler metal on the base metal are necessary in sive force of the liquid. In practical terms, with brazing, because the mechanics of the process respect to brazing, wetting implies that the liq- demand that the filler metal be brought uid filler metal spreads on the solid base metal smoothly, rapidly, and continuously to the joint instead of balling up on its surface (Fig. 2.3). It opening. If the conditions within the capillary has been demonstrated that wetting actually space of the joint do not promote good wetting, depends on a slight surface alloying of the base the filler metal is not drawn into the space by metal with the filler metal. capillary attraction. A comprehensive theory of the wetting or It all boils down to the fact that, for success- spreading of liquids on solid surfaces is pre- ful joining of components by brazing, the filler sented in Ref 17 and 18. metal selected must have a melting point above It can be concluded that wetting is the ability 450 °C (840 °F) and must also wet the base of the molten filler metal to adhere to the surface metal without melting it. Then, the joint must be of a metal in the solid state and, when cooled designed so that the mating surfaces of the com- below its solidus temperature, to make a strong ponents are parallel and close enough together bond with that metal. to cause capillary attraction. Wetting is a function not only of the filler metal but also of the nature of the metal or metals to be joined. There is considerable evidence Practical Experience, Work-Related that in order to wet well, a molten metal must be Tips, and Problem Solving capable of dissolving, or alloying with, some of the metal on which it flows. In order to braze tungsten carbide (WC) gran- Wetting is only one important facet of the ules or diamonds to 1010 carbon steel wheels brazing process. A very important factor affect- with BNi-2 filler metal, avoid the problem of ing wetting is the cleanliness of the surface to be filler metal sagging. Because the wheels must wetted. Oxide layers inhibit wetting and spread- rest flat in the furnace, the diamonds or carbides ing, as do grease, dirt, and other contaminants are then on the vertical surface of the wheel that prevent good contact between the filler diameter. metal and the base metal. One of the functions Typical brazing takes place at 1040 °C (1900 of a flux is to remove the oxide layer on the joint °F) in a pure dry hydrogen atmosphere. There area and to expose clean base metal. are a large number of variables that must be taken into consideration. It is much easier and more practical to use temperature as the con- trolling variable. Key variables that affect braze quality include:

• The chemistry which is controlled by a spec- ification and as a result the melting and flow characteristics of the filler metal are controlled by the chemistry. • Partial pressure of nitrogen will affect the melting characteristics of the filler metal. • A variation in the partial pressure of oxygen will affect the melting and flow characteristics of the filler. • The length of time in the oxidation range of 540 to 925 °C (1000 to 1700 °F) can also alter the melting and flow characteristics of the filler metal. • The heating rate, particularly at the high temperature where diffusion takes place, can Fig. 2.3 Wetting and dewetting alter the melting and flow characteristics. © 2003 ASM International. All Rights Reserved. www.asminternational.org Brazing (#06955G) Chapter 2: Brazing Fundamentals / 13

• The maximum brazing temperature is the 8. R.E. Loehman, Interfacial Reactions in best variable to control, because any one of Ceramic-Metal Systems, Ceram. Bull., the previously mentioned variables can Vol 68 (No. 4), 1989, p 891 change, requiring a change in the brazing 9. M.G. Nicholas and R.J. Lee, Joining Dis- temperature. A better braze filler metal for similar Materials, Met. Mater., Vol 5 (No. this type of application would be a very wide- 6), 1989, p 348 melting-range material and a filler metal of 10. J.T. Klomp, in Ceramic Microstructures Cr-Ni-B-Si-Fe. This filler metal has a melting 86: Role of Interfaces, J.A. Pask and A.G. range of 970 to 1160 °C (1780 to 2120 °F). Evans, Ed., Plenum Press, 1988, p 307 • Therefore, the large number of variables 11. J.E. McDonald and J.G. Eberhart, Adhe- presents a problem, but considering and tak- sion in Aluminum Oxide-Metal Systems, ing into account the various variables, control Trans. AIME, Vol 233, 1965, p 512 of the flow of the filler metal is feasible. 12. G.R. Edwards and J.J. Moore, “Investiga- tion of Brazing Alloys for Ceramic Sub- REFERENCES strates,” Research Proposal CSM 3264, Colorado School of Mines, Feb 1990, p 1. T. Young, Philos. Trans. R. Soc. (Lon- 54–70 don) A, Vol 95, 1805, p 65 13. R.G. Barrera and C.B. Duke, Dielectric 2. A.W. Adamson, Physical Chemistry of Continuum Theory of the Electronic Surfaces, 4th ed., John Wiley & Sons, Structure of Interfaces, Phys. Rev. B., Vol 1982, p 339 13 (No. 10), 1976, p 4477 3. P.R. Sharps, A.P. Tomsia, and J.A. Pask, 14. A.M. Stoneham and P.W. Tasker, in Wetting and Spreading in the Cu-Ag Sys- Ceramic Microstructures 86, Vol 21, tem, Acta Metall., Vol 29 (No. 7), 1981, p Materials Science Research, J.A. Pask 855–865 and A.G. Evans, Ed., Plenum Press, 1988, 4. M. Erg and A.W. Hennicke, Ceramics in p 155 Advanced Energy Technologies, A. 15. C.M. Adams, Jr., “Dynamics of Wetting Krockel et al., Ed., Dreidel Publishing, in Brazing and Soldering,” Technical 1982, p 138 Report WAL TR 650/1, Army Materials 5. M.E. Twentyman and P. Hancock, in Sur- Research Agency, Watertown Arsenal, faces and Interfaces in Ceramic and Watertown, MA, July 1962 Ceramic-Metal Systems, Vol 14, Materi- 16. S. Weiss and C.M. Adams, Jr., The Pro- als Science Research, J.A. Pask and A.G. motion of Wetting, Weld. J., Vol 46 (No. Evans, Ed., Plenum Press, 1981, p 535 2), Feb 1967, p 49s–57s 6. H. Mizuhara, Vacuum Brazing Ceramics 17. W.D. Hawkins, Physical Chemistry of to Metals, Adv. Mater. Process., Vol 131 Surface Films, Reinhold, 1952, p 1–413 (No. 2), Feb 1987, p 53–55 18. M.M. Schwartz, Fundamentals of Braz- 7. A.J. Moorhead and A. Keating, Direct ing, Welding, Brazing, and Soldering, Brazing of Ceramic for Advanced Heavy- Vol 6, ASM Handbook, ASM Interna- Duty Diesels, Weld. J., Oct 1986, p 117 tional, 1993, p 114–125