Electrostatic bonding for electronic applications

TWI Bulletin, April 1985

Gillian White, Andrew Newsam and Gill Taylor

Gillian White, BEng, is a research engineer in the Microjoining Section of the Sheet and Precision Processes Department. Andrew Newsam was a technician, and Gill Taylor was section leader (research), in the Microjoining Section, both having now left The Welding Institute.

The results of a feasibility study to determine whether electrostatic bonding can be used to join materials of interest to the electronics industry are described, together with the background to the process.

Electrostatic bonding (ESB) has been in use since 1969, primarily for the joining of metals, or semiconductors, to glasses. Process applications include solar cell and pressure transducer fabrication. Commercial bonding equipment is available, but equipment designed and developed at The Welding Institute was used in a recent feasibility study to determine how applicable ESB is for joining materials of interest to the electronics industry. The results of this research are described below.

One of the first references ^[1] relating to the process was a US patent, cited in 1969 by the Mallory Co Inc, for the joining of glasses to semiconductors or metals. Claims regarding the process included the joining of material combinations such as Be or Al sheet to borosilicate glass.

ESB is similar to diffusion bonding in that temperature, time and pressure are essential to the bonding process. However, ESB incorporates a DC voltage applied across the components during bonding. The metal, or semiconductor, is held at a positive potential with respect to the glass. The applied DC voltage can vary from a few hundred to several thousand volts with bonding times varying from one minute to several hours. Current is also an important parameter, since the product of current and time must be sufficient to enable bonding to occur. Current densities of the order of µA/mm ² are normally sufficient, but will depend on the materials which are to be bonded.

Mechanism of joint formation

The essential requirements for joint formation by electrostatic bonding are that components must be flat and highly polished and heated to a temperature at which ionic conduction, in the form of migration of ions, is possible. At these temperatures, most glasses behave as electrolytes which possess mobile Na+ ions. The glass is then slightly electrically conductive and capable of passing the low currents needed for bonding. The forces used are lower than in diffusion bonding, since it is electrostatic forces which lead to the intimate contact between components.

During bonding, the metal, or semiconductor, must be positive with respect to the glass. Reversing the polarity does not degrade the materials, it merely prevents bonding. It is possible to envisage a sandwich joint configuration, (metal-glass-metal) which can also be produced by this bonding technique.

The exact mechanism of ESB joint formation is still unclear but there are two main theories. These are described in detail in ref. ^[2] , but may be summarised as follows. The first ^[3,4] proposes that electrostatic forces, generated by ionic migration, bring the components into intimate contact. The application of a DC voltage causes positive ions within the glass to move towards the cathode. This leads to a deficiency of positive ions at the anode and a space charge effect is created. This effect gradually accumulates and a large proportion of the applied DC voltage is generated across this thin polarised region near the anode. Distribution of the voltage then occurs across the polarised region and the gap between the glass and metal.

According to the second theory ^[5] the high spots at the interface are electrochemically etched such that the gap between the glass and metal is closed. An oxide layer forms on the metals as the components are heated in air, which dissolves as the glass makes contact with the oxidised metal surface. When the glass meets bare metal, electrochemical oxidation begins. A bond is created by the polarised DC field which introduces additional metal ions into the glass producing an anodic bond.

Applications

Current applications of ESB include solar cell encapsulation, pressure transducer fabrication and opto-electronic device manufacture. Use of the process is still, therefore, rather limited and specialised. However, for those applications for which ESB is successfully used, it is unrivalled. A brief description of these applications, together with the reasons for ESB being chosen as the most suitable joining process are given below.

Solar cells

Solar cells require a glass cover for protection from the various harsh environments in which they must operate. Adhesives are commonly used but problems both technical and economic are encountered. High stress levels in the cover material and reduction in cell performance are technical disadvantages. Reduced cell output and low cover glass application rate are the economic ones.

Together, these have led to the adoption of ESB as the most suitable joining process. Bonds have been achieved ^[6] between 7070 borosilicate glass and either Si or Si coated with an anti-reflective layer, used to improve cell efficiency.

Figure 1 shows a solar cell contact which can be electrostatically bonded. The cover glass is deformed around a silver mesh creating a permanent, optically coupled, hermetic bond along the cell edges. This design ^[7,8] followed the original concept of grooved solar cell covers which accommodated the cell metallisation.

Alternative designs have been suggested for cells used in rugged conditions. ^[9] Once again, ESB has been used to provide hermetic sealing of the cell unit. Bonded cells which were subjected to subsequent rigorous environmental testing survived the tests without failure.

Capacitive pressure transducers

Figure 2 shows the configuration of a typical capacitive pressure transducer. A Si die is bonded to an opposing glass plate, the two materials having a fairly good match of thermal coefficient of expansion. The essential requirement of the joining process used for these devices is to provide a hermetic seal.

The thermal coefficients of the two materials to be bonded should match as closely as possible, thereby reducing the stresses induced during bonding. ESB of Si and Pyrex can meet these criteria [10] , and be used as a large scale production process.

Opto-electronic devices

Light-activated switches and photodiodes are examples of opto-electronic devices. Where the opto-electronic device is housed in a FeNiCo alloy can, the window (in the top of the can) is currently glued, or crimped, into position. A hermetic seal, between the window and the can, is required which, by the present joining techniques, is difficult to achieve. Furthermore, it is not easy to avoid contamination of the opto-electronic device during bonding. ESB has been shown [9] to be a means of successfully placing a 747 borosilicate glass window into the FeNiCo alloy can, both hermetically and without contaminating or damaging the device.

Research at the welding institute

A UK MoD sponsored project was placed at The Welding Institute in 1982/83 [11] to assess the feasibility of using ESB to join combinations of materials commonly used within the electronics industry. A range of materials was investigated at The Welding Institute under the above project and it is the results of these studies which will be discussed.

Equipment

Figure 3 shows the ESB equipment designed and built at The Welding Institute. The equipment incorporates cartridge heated Nimonic platens which can be heated to 600°C and a pneumatic pressurising system enables the use of loads of up to 7kN during bonding. Voltages up to a maximum of 3kV can be applied across the components during the bonding sequence, with a maximum current of 30mA. Timing of the bonding cycles is achieved by means of electronic timers and is, therefore, accurate and repeatable.

Components of up to 75mm diameter can be bonded on The Welding Institute's ESB equipment but, for production work, commercially available equipment exists [12] which can be used to bond as many as sixteen 75mm diameter Si wafers simultaneously.

Bonding trials and results

Electrostatic bonding of the material combinations listed in the Table, was investigated. For most combinations, the specimens were heated prior to bonding. This allowed the components to stabilise at the required bonding temperature before initiation of the bonding cycle. Following bonding the components were visually inspected and mechanically tested. The results showed that ESB was most successful for joining borosilicate glass to FeNiCo alloy and Pyrex.

For the former, bonding temperatures between 450 and 550°C gave reasonably strong joints. Polished FeNiCo alloy was used, and it was evident that the better the polished finish of the FeNiCo alloy, the stronger the resulting bond. This is illustrated in Fig. 4. Different resistivity Si specimens, ranging between 0.06-0.12 Ωcm and 10-20 Ωcm, were used in bonding trials with Pyrex and no differences in joint quality were obvious from the results of mechanical testing. 300°C was an acceptable bonding temperature.

For 747 borosilicate glass/6µm polished FeNiCo alloy joints, the maximum strength achieved were ~2.5 N/mm 2 whilst maximum strengths of ~ 0.15 N/mm 2 were achieved from Pyrex/Si joints.

Bonding to alumina

Alumina, Al 2 0 3 , is commonly used in the electronics industry as a substrate material, usually with metallised tracks. It. is also used in the fabrication of packages for hybrid circuits where it is often necessary for such packages to have ceramic or FeNiCo alloy lids and to be completely hermetic.

This is an essential consideration especially when packaging very sensitive circuits e.g. for aircraft systems.

Successful electrostatic bonding of ß-Al 2 0 3 to metal, in the form of foils, has been reported. [13] In these trials, ESB of 96% pure Al 2 0 3 to various metal foils or FeNiCo alloy was attempted as shown in the Table. 96% pure Al 2 0 3 did not bond successfully to either Cu or Al foil, since presumably its structure did not allow sufficient ionic conduction to occur. Three attempts were made to bond Al 2 0 3 , in different forms, to FeNiCo alloy but no bonds were achieved. It is likely that the lack of ionic conduction of the Al 2 0 3 , together with its insulating properties, prevented bonding.

A thin, 5µm, layer of Pyrex was therefore sputtered on the surface of the Al 2 0 3 and a metal foil inserted between the insulating Al 2 0 3 and the platens to allow electrical contact, such that a potential could be established across the glass-metal interface. The bonding conditions which had given bonds between Pyrex, in the form 0f discs, and FeNiCo alloy, were repeated in these trials. The current measured across the components was negligible, with the maximum value being 0.25mA.

It is likely that the molecular structure of the sputtered Pyrex is sufficiently different to that of Pyrex discs to cause these lower current values and hence, none of the attempted bonds were successful.

The diffusion bonding of Al thick film on Al 2 0 3 has previously been investigated at The Welding Institute [14] to determine whether the development of an all-Al electronic system, free from solder or brazed joints, would be feasible. However, the results were disappointing enough to consider ESB as a joining process for these materials.

Unfortunately, in common with the attempts of diffusion bonding, the ESB results were unsuccessful.

Material combinations investigated under electrostatic bonding feasibility study. [11]

Summary

From the results obtained, it seems that electrostatic bonding is most applicable to joining Pyrex or 747 borosilicate glass to Si or FeNiCo alloy. These material combinations relate directly to capacitive transducer or solar cell fabrication, which are already in production.

Attempts to bond Al ₂ 0 ₃ to metal foils (Cu or Al) and FeNiCo alloy sheet were unsuccessful. This indicates that more work would be necessary to find an Al ₂ 0 ₃ composition or an interlayer material (glass) which are more suited to the ESB process i.e. with better ionic conduction properties.

Member companies interested in this process are invited to contact Gillian White or Norman Stockham in the Microjoining Section at Abington.

Acknowledgements

This work was carried out with the support of the UK Ministry of Defence, Procurement Executive, Directorate of Components Valves and Devices.

References