Active metal brazing of ceramics

Kane Ironside joined TWI in 1997 as a Project Leader within the Advanced Materials and Processes department. Prior to this he studied for a BEng and PhD degrees at the University of Surrey.

Brazing is one of the most widely used processes for joining ceramics. John Fernie and Kane Ironside review the process and describe some of the best practice, applications and recent developments of the process.

The properties of engineering ceramics make them extremely attractive to industry. These properties include low density, high hardness, high strength (in tension), low friction and good chemical resistance. However, they are relatively expensive and thus are most frequently used to enhance the performance of another material, most often a metal; a good example would be the use of ceramics as inserts in cutting tools. Here ceramics, such as silicon nitride and diamond, are used as the cutting edge, supported by a metal. Since ceramics are most frequently used in conjunction with other families of materials, joining becomes a critical issue.

Of the many joining processes available to join ceramics, probably the primary and most adaptable technique used is brazing. Brazing is a liquid phase process particularly well suited to preparing joints and seals. The brazing process can be readily adapted to mass production of components, such as those used in the electronics and automotive industries.

Wetting and metallisation

If a braze alloy is melted between two ceramics, a poor joint is the most likely result. This is because of poor wetting. Wetting is measured in terms of the contact angle between the braze and substrate after melting. For good wetting the contact angle ( Θ) is less than 90°; for poor wetting, the angle is greater than 90°. Examples of good and bad wetting are shown in Fig.1.

Fig.1 Two wetting trials showing examples of good and bad wetting

The reason that most molten metals do not readily wet ceramics stems from the very strong ionic/covalent bonding found in ceramics. It is energetically unfavourable for the ceramic bonds to break and interact with the molten metal; thus, physics dictates that the molten metal adopts the lowest surface energy to volume ratio ie it balls up.

To encourage wetting there are two basic philosophies - apply something to the surface of the ceramics so that the braze will wet, or, put something in the braze that will induce wetting. Surface treatments include metallisation, metal coating and metal hydride treatment, while braze modification has led to a process known as active metal brazing. In either case the actual brazing operation takes place either in a controlled atmosphere, such as nitrogen or argon, or a vacuum better than 10 ^-4 Nm ^-2 .

One of the most widely used methods of metallisation is the so-called 'moly-manganese' process, which is particularly well suited to the metallising and subsequent brazing of alumina. In essence a series of thin layers is deposited onto the ceramic surface to be joined, each contributing to the enhancement of wetting between substrate and braze. Typically, a powder mixture of glass, molybdenum and manganese is screen printed onto the ceramic. On firing this leaves a manganese rich zone at the ceramic surface (to which it wets and bonds) and an outer region of molybdenum. Next the parts are plated with nickel, and then finally brazed. Despite the numerous steps required, the moly-manganese process can be automated. It is used extensively in the electronics and electrical industries to produce ceramic to metal seals for ultra high vacuum equipment, such as isolators and power semi-conductor housings.

The moly-manganese process is dependent on the presence of an intergranular 'glassy' phase, which is found in several ceramics notably alumina and silicon nitride, where the glass acts as a binding agent. However, where such glass is not present, ie zirconia and silicon carbide, a different approach is often used.

Active metal brazing

For the 'active metal' process to work an 'active' element is required that will change the interfacial energy between the braze and ceramic to such a level that wetting of the ceramic takes place. This occurs via migration of the active metal to the surface of the ceramic during the brazing process. A series of chemical reactions takes place, resulting in a number of intermetallic compounds being formed at the interface. These reaction products remain as thin layers between the ceramic and the rest of the braze alloy. The braze then wets to the reaction products. It is important to recognise that wetting and reacting are different, although in active metal brazing they are related.

Active elements include Ti, Zr, Hf, V and Al; the best known of these active elements is Ti, which is used in many commercially available active metal braze alloys. Commercial active braze alloys are mainly based on the Ag, Cu or Ag-Cu eutectic systems with added titanium (typically 1-5%). These and other available families of braze alloys are shown as a function of melting point in Fig.2. Brazing can be undertaken at a range of temperatures in controlled atmosphere, depending on the alloy used. Active braze alloy compositions are constantly being improved to permit metals to be joined to ceramics (for ever more demanding requirements) without the need for metallising the ceramic surface.

Fig.2 Principal braze alloy families including active metal alloys and their melting range

Typical heat treatments schedules are described in Fig.3 and the Table. In vacuum, or controlled atmosphere, the samples are heated to approximately 30-50°C below the solidus temperature, then held to allow homogenisation. Following which, the temperature is then increased slowly to the brazing temperature and held for a period of time, typically 5-30 minutes during which bonding takes place. Excessive brazing temperatures and times can be detrimental as discussed later. Cooling from the brazing temperature is important and must be relatively slow to avoid the inducement of strains associated with co-efficient of thermal expansion (CTE) mismatch and thermal shock.

Fig.3 Representation of the brazing cycle

Table: Typical brazing schedules

It is important to establish and maintain a good vacuum or inert atmosphere during brazing, otherwise the active metal will react with any impurities and not with the ceramic. For example, Ti will preferentially react with any oxygen, before it will react with a ceramic.

Many studies have been conducted to understand the wetting characteristics, reaction kinetics and joint behaviour of ceramics joined using these active metal alloys. Figure 4 shows a scanning electron microscope (SEM) image of a Ag-Cu-Ti alloy brazed to a silicon nitride substrate. The reaction layer is clearly visible and although many reaction products are possible, thermodynamic analysis can determine which products are most favourable. For silicon nitride, TiN, Ti ₅ Si ₃ and TiSi are the main reaction products. Studies have shown that control of the thickness of the reaction layer is critical to the final strength of the bond. Too little and there is not enough strength induced, too much and failure occurs by tearing along the interface. The reaction layer thickness is a function of the initial Ti content of the braze and the brazing temperature and time.

Fig.4 Scanning electron micrograph of the reaction layer between an active metal braze and silicon nitride

The upper working temperature of a brazed joint is usually about two-thirds of the melting point of the braze. This can vary with joint design and operating conditions, but acts as a guideline. Protection of the joint from shear is regarded as very good practice.

Joint design and best practice

A further factor to consider is the brittle nature of ceramics and the avoidance of sharp edges/changes of section in the joint. These points can act as stress concentrators, potentially acting as initiation sites for the fracture and failure of the ceramic or joint. Best practice in both selection of braze alloy and brazing temperature will help reduce the residual stresses that develop in the joint.

Modelling can be used to predict failure of brazed joints and compare many different joint designs. For simple component shapes and joint design then analytical methods may be used. However, in the majority of cases the component or joint is too complex so numerical methods are required, such as finite element analysis (FEA).

For FEA a complex component is usually modelled by considering it as an assembly of small simple parts, called elements. The element assembly or mesh gives a geometric representation of the component. The use of the computer allows a 'virtual laboratory' to be set-up to test different joint configurations, braze alloys and brazing conditions prior to any actual brazing. This technique is very powerful and is being used more in industrial development work, and not only for brazing, but other joining methods.

Applications

A relatively simple example of brazing is found with the use of ceramic cutting tools. Figure 5 shows a diamond brazed to a steel support. Brazes are relatively strong compared to adhesives and can survive temperatures of up to 500°C in harsh environments. On the cutting tool shown the joint area is relatively small. Thus the difference in thermal expansion between the two materials does not have a detrimental effect. 

Fig.5 A diamond brazed to a steel substrate for application as a cutting tool

Recent developments

Another innovation has a ceramic reinforcement incorporated into the braze alloy, to both increase its strength and reduce its CTE. This can be achieved by mixing a braze (in powder form) with the reinforcement and applying the mixture as a paste; on cooling, this forms an in-situ metal matrix composite. It has been shown that the joint strength increases by approximately 50% (from ~350 to 540MPa respectively) with the introduction of as little as 5% by volume of reinforcement (silicon carbide particulates), when compared with the unreinforced braze alloy. Figure 7 shows the microstructure of such a braze alloy which has been used to join silicon carbide. This work is currently being undertaken as a collaboration between TWI and Cambridge University as part of a Postgraduate Training Partnership (PTP). Work on this concept will be extended to include higher temperature braze alloys.

Fig.7 A scanning electron micrograph of a 5vol% SiC p reinforced active braze alloy

High temperature active metal brazes

The majority of the commercial active metal brazes have been developed for moderate temperature use, up to ~450°C. However, ceramics are able to survive high temperatures, for example, alumina typically has an upper use temperature of 1700°C, therefore joints should accommodate this. To take advantage of the high temperature properties of ceramics, the braze alloys need to have a higher temperature capability than is currently available. One method is to coat the ceramic with either a reactive or refractory metal (W, Mo, Ta, Cr) then braze using high temperature braze alloys, such as, palladium and platinium-based systems. This method has been used successfully for joining many high temperature ceramics.

A number of brazing alloys have recently entered the market designed for specific applications such as high temperature ceramic (silicon carbide) heat exchangers. These are based on transition metal-Si eutectic alloys, producing joints capable of surviving temperatures in excess of 1200°C.

Conclusions

One of the most commonly used technique for the joining of ceramics is brazing. There are two main techniques to ensure the wetting of braze alloys on to ceramics: metallisation and active metal brazing. Traditionally the moly-manganese process has been widely used by industry; however, active metal brazing is becoming a serious competitor. New brazing alloys for use at both low and high temperature uses are being developed.

References