N R Stockham and C J Dawes
Paper presented at ISHM International Microelectronics Symposium, Philadelphia, PA, 31 Oct.-2 Nov.1983 and published in International Journal for Hybrid Microelectronics, vol.6, no.1. Oct.1983. pp.509-519
An investigation into resistance seam sealing and laser welding of Fe-Ni-Co alloy packages, plated with Au or electroless or electrolytic Ni is described. The resistance seam sealing evaluation trials have been conducted on an opposed electrode machine using both a.c. and d.c. power supplies. These results have shown that the braze formed by resistance heating is complex in nature, with a structure and mechanical strength that is influenced significantly by lid thickness. Stepped lids were found to be tolerant to quite marked changes in machine parameters. Flat lids (0.38mm) however, could not be joined successfully with the standard d.c. capacitor discharge power source, but the use of an a.c. power supply, allowing longer pulse times and a slower rate of energy input, enabled strong sound joints to be made.
Detailed laser welding trials have been conducted using a fast axial flow CO 2 laser. The results show that a high level of package leak tightness, satisfying BS 9450, was achieved in the unplated and Au-plated packages when welded at ~3m/min and that the temperatures measured in the package circuit attachment area do not exceed 110°C. Solidification cracking was, however, obtained with the Ni-plated packages. The results obtained are compared with those achieved by pulsed CO 2 laser welding which indicates that cracking can be eliminated when welding electrolytic Ni-plated packages and much reduced when welding electroless Ni-plated packages.
The rapid advances in hybrid circuit technology have created a demand by the defence, aerospace, and telecommunications industries for large hermetic metal circuit packages (> 150mm around the perimeter). Resistance seam sealing and laser welding are two important techniques for sealing these encapsulations. Other techniques available include electron beam welding and soldering. Resistance seam sealing is the more conventional process, and has been used in the hybrid industry for several years. Laser welding is being viewed as a possible future method of encapsulating components.
Resistance seam sealing comprises a miniature conventional seam welding machine which is capable of welding, brazing, and reflow soldering. Currently two techniques are commercially available: opposed electrode mode and series electrode mode. This paper briefly describes the operating principles of resistance seam sealing and relates the results of a three-year project (1979-1982) to examine the reliability of this process when sealing (on an opposed electrode machine) the conventional stepped lid and the cheaper more rigid flat lid solid sidewall packages.
Laser welding of large metal packages is an attractive alternative to resistance welding because reliable versatile laser welding equipments are available allowing very rapid welding speeds with limited heat input to the package. Furthermore, the technique is potentially very clean, giving little weld spatter, and like resistance seam sealing can operate in inert and dry atmospheres.
Other advantages of laser welding are that it is a non-contact process and therefore does not have electrode wear/reliability problems, it has the ability to seal joint configurations which are not possible by resistance welding, Fig.1, absence of X-rays, and the low electric and magnetic fields at the weld. Because of these potential advantages work was started at The Welding Institute in 1977 to assess the industrial interest in the CO 2 laser sealing of large packages and to establish the reliability and advantage of this technique by optimising conditions for laser welding of unplated and plated metal packages. The work reported here was conducted using CO 2 lasers since such equipments were already available at the Institute. However, the results indicate the viability of the more recently developed solid-state Nd-YAG lasers for this application.
Fig.1. Lid-to-base joint configurations for large packages and suitable welding techniques. Hatched areas indicate that controlled atmosphere can be sealed in package during welding/brazing
Resistance seam sealing of metal packages
Principles of operation
There are two types of seam sealing machine suitable for the hermetic sealing of electronic packages commercially available. These machines operate in what is classified as either a series or opposed wheel mode. In the series wheel mode, Fig.2a, the package and lid to be sealed are passed under a pair of small tapered copper electrode wheels. The transformer (power supply) produces a series of energy pulses (1kHz a.c.) which are conducted from one electrode across the package lid to the other electrode. The generation of heat at the wheel/lid interface is conducted to the lid/package interface to form either a weld or braze type seal, the nickel or gold plating acting as the brazing medium.
Fig.2. Arrangement of electrode wheel and current connections for the two available seam sealing equipments: a) series electrode wheel mode b) opposed electrode wheel mode
An alternative to series wheel mode is the opposed electrode mode, Fig.2b. Here again the package to be sealed is moved under a pair of larger electrode wheels. The power supply produces a series of welding pulses (normally d.c. capacitor discharge) which pass from the electrode wheel across the lid/package sidewall interface and return via the work table, generating heat at the electrode/lid interface and the lid/package sidewall interface. It is a common practice to have a power supply connected to each wheel. As can be seen from Fig.3a, a line of braze is formed between the package sidewall and the lid. The continuous seal which extends the full length of the package wall is formed by a series of overlapping spots (energy pulses) as the package is moved under the electrode wheels. At the end of each seam the package rotates through 90° and the other two sides are sealed. A plan view, Fig.3b, of the edge of the lid on a typical package shows the appearance of these overlapping spots. The weld/braze pitch, or the distance between consecutive spots, can be varied to control the amount of overlap.
Fig.3. Au-plated stepped lid package: 3a) microsection showing melted plating forming braze at lid/sidewall interface and outer fillet,
3b) plan view showing series of overlapping spots along edge of package lid to form continuous seal
The objective of this three-year programme was to evaluate the resistance seam sealing technique and to establish the effect of the major parameters on joint integrity. The trials were carried out on an opposed wheel seam sealer fitted with two d.c. capacitor discharge power supplies, Fig.4. To establish whether joint integrity can be improved (on flat lid packages) by the use of longer pulse times, an a.c. power supply (1½kVA transformer connected to a ½-11 cycle welding timer incorporating a phase shift control) was connected to one of the electrode wheels. Instrumentation was attached to the machine to allow calibration and monitoring of current, voltage, pulse length, traverse speed, and electrode force during seam sealing.
Fig.4. Package profile following head on seam sealing machine
The trials were conducted on solid sidewall Fe-Ni-Co alloy packages, Fig.5. These packages are single-piece construction with pin connections sealed through the base via glass seals. The width of the sidewall is lmm. The work was carried out on electroless Ni or Au-plated 20 (30 x 30mm) or 30 (30 x 45mm) pin packages with matching etched (periphery etched down from 0.38 to 0.13mm thick) or flat (0.38mm thick) lids. These packages were wipe degreased in acetone, assembled in the machine holder, and clamped into position for seam sealing. The machine stops were adjusted so that, when the electrode head was down in its welding position, the electrode wheels rose and fell 0.3mm at the stop and start of each weld run. This procedure was found to give sufficient and consistent overlap of the corner seams without arcing between the electrodes and the package.
Seal quality was assessed in five ways:
- Gross and fine leak testing (bubble testing and He leak testing to BS 9450, 5 x 10 7 atmosphere cm 3/sec)
- Hydraulic pressure testing
- Pull testing
- Thermal shock and temperature cycling to MIL-STD883B (-55° to +125°C, 50cycles)
- Metallurgical assessment.
Hydraulic pressure testing consisted of drilling and tapping a hole near the centre of the base and screwing the package to a hydraulic line, immersing it in water, and slowly pressurising with soluble oil until a leak is indicated by a jet of oil emerging from the device into the water. This test provides a simple means of testing the whole seam-sealed joint (sides and corners).
Fig.5. Solid sidewall packages used during resistance seam sealing trials: 5a) 30-pin DIL 5b) 20-pin dummy package
5c) package lid and sidewall dimension
Mechanical pull tests were also performed on 16mm wide specimens cut (and polished to reduce edge effects) from the centre of sealed packages. These specimens were mounted in a jig which restricted the bending of the lid (which will vary for each lid thickness) and a uniform shear-peel test of the joint carried out in a Hounsfield tensometer.
Stepped lid sealing trials
To determine the effect of the sealing parameters on joint integrity when using the d.c. power supply, a 'typical' brazing condition was selected from the machine manufacturer and current users' recommendations. Maximum and minimum values were selected each side of this condition, Table 1. These parameters were varied individually to establish tolerance to change of each parameter. A series of specimens was then brazed with a minimum of two samples joined at each condition.
Table 1 Range of machine parameters used for seam sealing solid sidewall packages with stepped lids when using a d.c. power supply
|Machine variables||Typical||Approximate range investigated, %|
Pulse width, msec
Step length, mm
Pulse frequency, pulse/min
Electrode force, N
Four series of trials were carried out on stepped lids using this procedure: two involved Au-plated lids and two were electroless Ni-plated. Nearly 50% of the first series of twenty-three Au-plated packages failed the leak test because of holes fused through the lid on the inside edge of the seam seal. This overheating fault was attributed to new electrode wheels and bearings which would tend to run hot (and fuse the Fe-Ni-Co lids) as a result of the higher contact resistance until they have 'bedded in'.
This series was repeated with bedded-in electrodes and all the packages passed the helium leak test. The two series of electroless Ni-plated packages (total of forty-six) sealed using the above procedure resulted in four He leak test failures. There was no obvious pattern to the package failures and, as it was not possible to locate these leaks, the type of failure mode could not be established.
The hydraulic pressure test results for the above (using 'bedded in' electrodes) Au and electroless Ni-plated stepped lid packages revealed high strength joints (average 41 bar) with nearly all failures occurring through the lid material away from the brazed joint, Fig.6.
Fig.6. Typical Au-plated stepped lid package showing failure through Ni-Fe-Co alloy lid during pressure testing
To assess the reliability of these joints a limited consistency trial (thirty-one packages) was carried out on electroless Ni-plated stepped lid packages sealed at 40Wsec, 14msec, 125 pulses/mm, 89N. All but one of the packages satisfied the leak requirements of BS 9450, the one failure stemming from a fine leak at a glass-to-metal seal.
Limited trials on Au and electroless Ni-plated stepped lid packages using an a.c. power supply (Au 2100A, 2 cycles, Ni 2000A, 2 cycles) showed the joints to be consistent and of similar strength to the equivalent d.c. capacitor discharge sealed packages.
Flat lid trials
d.c. power supply
Trials on Au and electroless Ni-plated flat lids (0.38mm thick) covered a range of machine parameters (using 2°, 7° and 15° electrode wheels), including weld energies from 40 to 160Wsec, with the aim of establishing a suitable sealing condition. However, a sealing schedule was not found which gave reliable, strong, leak-tight joints using the d.c. capacitor discharge power supply. The electroless Ni-plated lids were mechanically relatively weak (0.7-3.4 bar) and brittle compared with the stepped lids (average 41 bar). The Au-plated packages were stronger than the electroless Ni (approximately 10 bar) but with poor hermeticity. To aid the understanding of the brazing process when used on 0.13mm stepped and 0.38mm flat lids a batch of 0.25mm electroless Ni-plated flat lid packages was sealed. The pressure test results were between those for the 0.38mm flat lids (5.5-28 bar, average 13 bar). A higher proportion of joints also passed the He leak test when 0.25mm flat lids were used than when 0.38mm flat lids were employed.
Metallurgical examination of the joint interfaces from these packages (the lid having been peeled from the sidewall) showed that the structure of the brazed joint, which from sections seemed relatively simple, is complex. The brazed joints contain a range of structures which vary markedly along the length of the seam; Fig.7 shows the joint faces from a 0.25mm thick electroless Ni-plated package. At the inner edge of the joint there had apparently been a pressure weld between unmelted plating. Next to this is an area where some partial melting of plating has perhaps occurred with the formation of multiple small pores. Local analysis showed high Ni and P with only a little Fe, confirming no marked alloying.
Fig.7. Joint face from electroless Ni-plated 0.25mm thick flat lid package showing complex braze structure
Nearer the centre of the joint the pore size was much larger: within the pores the plating had melted and local analysis showed a higher Fe content than in the material between the pores, which had at most only partially melted and was frequently cracked. At the centre of the joint there was more evidence of melting in the big pores, as shown by the dendritic structure, but there were still large areas of cracked, partially fused plating, and such areas could be seen even at the outer edge of the joint.
These features were common to flat and stepped lids. Figure 8 shows the sidewall of a stepped lid package near the outer edge of the joint: pores and partially melted plating are visible right out to the edge, with shear failure through the fillet. In the centre a whiter area can be seen in which intergranular failure and some microvoid coalescence are visible. Local analysis showed that the predominant element was now Fe. In these areas melting had clearly been more extensive and failure was partly in the Fe-Ni-Co itself. When comparing the three lid thicknesses in electroless Ni-plated Fe-Ni-Co materials it becomes apparent that the structure and mechanical strength of seam sealed joints is influenced significantly by lid thickness. The Ni-P alloy formed by electroless plating is brittle, and to obtain joints which are not weak, conditions must be such as to give the type of failure shown in Fig.8. When using the d.c. capacitor discharge power supply this has been consistently possible only on stepped lids. It appears that, with thicker lids, the heat generated at the joint interface is insufficient to allow the formation of a good braze without a significant amount of additional heat being conducted down from the electrode/lid interface to the joint (as in series wheel mode resistance heating). As this lid thickness is increased the effect of this secondary heating source is reduced. If the input power is increased to compensate for this loss, the Fe-Ni-Co alloy base material begins to fuse at the electrode/lid interface.
Fig.8. Joint interface from electroless Ni-plated stepped lid package (40Wsec) showing structure of 'strong' area of joint interface, x150
a.c. power supply
Trials were conducted on both Au and electroless Ni-plated flat lids (0.38mm) to establish whether joint integrity could be improved by the use of the longer pulse times (> 14msec) and slower rise times possible with an a.c. power supply. These trials covered a range of machine parameters including current levels of 800 to 2700A (2 cycles). A sealing condition was selected from these trials which gave leak-tight joints (to BS 9450) with maximum strength without raising the temperature to a level which might cause component degradation or metallurgical damage to the package. These trials showed that hermetic seals can be achieved in both Au and electroless Ni-plated packages over a relatively wide range of machine parameters, Table 2, when using the a.c. power supply. The mechanical strength of the electroless Ni-plated lids, however, was similar to that obtained with the d.c. power supply (a.c.: generally 100-600N, d.c.: generally 60-900N). These results suggest that the increased current and time employed on the a.c. trials was sufficient to maintain the plating molten for long enough to form a leak-tight joint (unlike the d.c. results) but not to reduce the P content (which embrittles the plating) by diffusion and/or alloying with the Fe-Ni-Co alloy parent material. The joint strength could be increased either by appreciably raising the energy to the package by reducing the step length to 0.1 mm (pull strength > 4500N) or by doubling the pulse duration from 2 to 4 cycles (pull strength 1500N). This, however, resulted in excessive heating and severe metallurgical damage.
Table 2 Range of tolerance to changes in machine parameters for seam sealing solid sidewall packages with flat lids (0.38mm) based on hermeticity, using an a.c. power supply
|Machine variables||Electroless Ni-plated packages||Au-plated packages|
|Optimised schedule||Approximate range of tolerance, %||Optimised schedule||Approximate range of tolerance, %|
|Weld time, cycles
|Step length, mm
|Pulse frequency, pulse/min
|Electrode force, N
|*Most extreme value investigated, no leak
Considerably higher strength joints, Fig.9, were obtained when sealing Au-plated 0.38mm flat lids (generally 160-3500N). The Au-plated packages were more tolerant to higher heat inputs (metallurgical damage) but required a higher initial current to form a hermetic seal. Consistency trials were carried out on thirty Au and thirty electroless Ni-plated flat lid (0.38mm) packages sealed using the optimum condition shown in Table 2. All passed the He leak test. Four packages (two Au, two Ni) were then thermal shock tested and four packages (two Au, two Ni) were temperature cycled (MIL-STD883B, 50 cycles, -55° to +125°C) without loss of hermeticity or strength.
Fig.9. Au-plated flat lid (0.38mm) packages sealed using a.c. power supply at optimum condition (2100A, 2~): 9a) microsection showing braze on outer 50% of sidewall
9b) package pressure tested to 80 bar without failure
The temperatures measured at five positions in stepped and flat lid packages ( Fig.10a shows the positions) when sealing using the d.c. capacitor discharge (stepped lid) and a.c. power supplied (flat lid) are shown in Fig.10b. When sealing stepped lid packages at the typical brazing energy level of 40Wsec the maximum base temperature is approximately 70°C (side/base), but if components inside the package touch the sidewall they may reach 115°C. To obtain hermetic joints when sealing flat lid packages using the a.c. power supply higher energies and consequently higher package temperatures were required, Fig.10b. At the optimised condition of 2100A, 2 cycles, the maximum base temperature is 165°C (side/base) and sidewall (lmm from joint) 240°C. The plateau on the a.c. power supply curve corresponds to a change in transformer tap and a reduction in percentage weld heat to achieve 2300A p-p.
Fig.10. Temperature measured during seam sealing at various energy setting: 10a) solid sidewall stepped lid package sealed using d.c. capacitor power supply (14msec, 0.2m/min)
10b) solid sidewall flat lid package sealed using a.c. power supply (2~, 0.2m/min)
The trials on stepped lid packages with bedded-in electrodes suggest that they are insensitive to machine setting variations within rather broad limits when sealing with both d.c. capacitor discharge and a.c. power supplies. Examination of the structures present in joints made at various sealing conditions has shown that the formation of a strong brazed joint requires sufficient heat at the interface to melt approximately 50% of the sidewall width. Melting only the outer edge is inadequate, although a fillet at the outer edge may be sufficient to form a leak-tight package. In electroless Ni-plating, the heat must also be sufficient to maintain the plating molten for long enough for the P content (which embrittles the plating) to be reduced by diffusion and/or alloying with the Fe-Ni-Co alloy parent material. Since the heat generated at the electrode/lid interface is an important part of the total needed, the short pulses of the d.c. power supply are not as suitable as the longer pulses of the a.c. system which allow diffusion of the heat from the electrode/lid interface to the lid/sidewall interface and also cause less melting and cracking at the former interface (which should also reduce electrode wear).
The results of the trials carried out on 0.38mm thick flat lids have confirmed that, with the standard d.c. capacitor discharge power source, it is virtually impossible to make strong reliable leak-tight joints, irrespective of whether the package is Ni- or Au-plated, because of its inability to generate sufficient heat for long enough at the lid/sidewall interface. With an a.c. power supply, reasonably tolerant and consistent leak-tight joints can be produced in 0.38mm flat lids without excessive temperature rise within the package (maximum base temperature 165°C, air temperature 75°C). Strong joints (up to 100 bar) could be produced with Au-plating but with electroless Ni it was not found possible to reduce the P content sufficiently to form a very strong (7-28 bar) hermetic seal without causing significant damage (surface melting, cracking) to the package. In this respect alternative platings such as electrolytic Ni or electroless Ni-B may improve joint strength.
- Sealing stepped lids using a d.c. capacitor discharge power supply at the typical level of 40Wsec resulted in package base temperatures of between 50° and 70°C. Sealing flat lids (0.38mm) at a higher current of 2100Ausing an a.c. power supply resulted in package base temperatures of between 65° and 165°C.
- New electrodes should be 'run in' on several unwanted packages before sealing the required components.
- Electroless Ni- and Au-plated stepped lid packages were found to be tolerant to reasonably large variations in machine parameters ( Table 1) when sealed using a d.c. capacitor discharge power supply. Similar strength and hermeticity were obtained when using an a.c. power supply.
- The mechanism by which the brazed joint is formed during the resistance seam sealing process is complex. Its structure and mechanical strength are influenced significantly by lid thickness.
- With the standard d.c. capacitor discharge power supply it proved impossible to make strong, reliable, leak-tight joints with 0.38mm electroless Ni- and Au-plated flat lids.
- Au and electroless Ni-plated flat lids (0.38mm) when sealed using an a.c. power supply were found to be consistent and tolerant to reasonably large variations in machine parameters ( Table 2), satisfying the leak requirements of BS 9450. The Au-plating produced strong joints (43-100 bar) but even using the longer pulse times of the a.c. power supply the electroless Ni-plating resulted in relatively weakjoints (7-28 bar), although the strength is probably adequate for most applications.
- Au and electroless Ni-plated stepped and flat lid packages sealed at optimised conditions using the a.c. power supply passed thermal shock and temperature cycling tests (MIL-STD883B, -55° to +125°C, 50 test cycles).
Continuous laser sealing of large hybrid packages
To establish the potential of continuous wave (CW) laser welding trials were conducted on a 2kW, CW, fast axial flow laser, Fig.11a, which was designed and developed at the Institute in 1971-72. The main features of this laser are its high resonator efficiency (approximately 25%) and the model purity of its output beam. The latter feature permits the beam to be focused to a highly concentrated spot with sufficient power density to allow high speed seam welding with very low heat input to the surrounding material. A 75mm focal length KCL lens was used to focus the beam. The focused spot diameter produced by this lens is approximately 0.3mm. The point of focus was set at the package lid surface.
The alignment of the package under the laser beam is of paramount importance during high speed automatic welding. Good alignment depends upon the dimensions of the package, jigs, and work movement systems. The packages were located in recessed jigs for these trials with the lids held firmly in place by a spring-loaded finger clamp. The clamp pivoted horizontally allowing it to be slid off the package lid, by contact with the welding gun, during the latter stages of the welding cycle.
An inert gas shield was applied over the weld area to prevent oxidation of the weld metal and enhance the laser beam transfer efficiency, Fig.11b. The coaxial beam gas shield and the package location jig gas manifold were fed at a rate of 25litre/min. The package location jig was mounted on a double axis worktable, controlled by a programmable microcomputer which moved the package under the laser beam at constant velocity with a positional accuracy of ±0.lmm and work movement speeds of 0-12m/min.
Fig.11 a). The Welding Institute 2kW laser
Fig.11 b) schematic principle of gas shield arrangement
Welding trials were conducted on two sizes of unplated Fe-Ni-Co alloy packages. One was approximately 50 x 19 x 6mm deep (corner radii approximately 2mm). The wall thickness was approximately 1mm. The second package was approximately 31 x 29 x 4mm deep (corner radii approximately 3.5mm) with a wall thickness of lmm. Packages of this size were also used with electrolytic Au, electrolytic Ni, and electroless Ni platings approximately 5µm thick. The matching flat lids were 0.4mm thick Fe-Ni-Co alloy (unplated and plated).
Weld quality was assessed by helium leak testing to BS 9450, and by visual examination, and metallurgical assessment.
The welding trials were conducted with four different package surface finishes:
- Plain Fe-Ni-Co
- Electroless Ni-plated Fe-Ni-Co
- Electrolytic Ni-plated Fe-Ni-Co
- Electrolytic Au-plated Fe-Ni-Co
To allow analysis of the welding trials it was necessary to define an acceptable weld size. Welds were judged acceptable if the weld width (at the joint surface) was between 25 and 75% of the sidewall thickness. This means that the weld width was approximately equal to the lid thickness (0.4mm, Fig.12a). Weld widths of less than 0.25mm were considered to be potentially poor seals, Fig.12b, and those greater than 0.75mm unacceptable because of the risk of metal vapour entering the package interior, Fig.12c.
Fig.12. Microsections of laser welded Fe-Ni-Co packages for profile assessment: a) acceptable weld, 1.3kW, 50mm/sec
b) insufficient penetration, 1.0kW, 60mm/sec
c) excessive penetration, 1.5kW, 40mm/sec
At established CW laser welding conditions ten unplated 50 x 19mm flat lid packages were welded at each of the following conditions: 0.8kW at 1.8m/min, 1kW at 3m/min, 1.2kW at 4.5m/min, and 1.6kW at 6m/min. All these welds were of an acceptable size with similar profiles within each group of ten welds.
Unplated packages with flat lids
The above procedure was repeated using 29 x 30mm flat lid unplated packages with very similar results. The weld bead width on the corners, however, was more uniform, possibly because of the larger corner radii on the second batch of packages. A typical welded unplated flat lid Fe-Ni-Co alloy package is shown in Fig.13
. In general there was very little difference in weld profiles made at the first three conditions given in the previous paragraph. However, the weld widths at 1.6kW at 6m/min all approached the maximum acceptable limit (75% wall thickness).
Fig.13. Typical laser welded unplated Fe-Ni-Co package
To fill packages with dry N 2 a few tests were made replacing the normally used He shielding gas with N 2. However, this resulted in heavy spatter and, consequently, porosity of the weld bead. To overcome this problem He was fed coaxially with the laser beam onto the weld surface, and O 2-free N 2 was flooded around the periphery of the package via the manifold of the package location jig (see Fig.11b). Welds made on unplated Fe-Ni-Co alloy packages (29 x 30mm) at the above optimised condition (1kW at 3m/min) were found to have smooth crack-free weld profiles identical to those achieved with the all He atmosphere. Fifty unplated packages (29 x 30mm) were welded at 1kW, 3m/min. then leak tested to BS 9450 (5 x 10 -7 atmospheres cm 3/sec); forty-nine packages passed the test. Examination of one which failed showed a large pore in the weld bead. The pore was of a nature which indicated the presence of organic contamination at the joint interface. This stresses the importance of cleanliness prior to welding.
Plated packages with flat lids
When welding Au-plated packages it was found necessary to reduce the energy from 1kW, 3m/min, to 0.8kW, 3m/min, to reduce the weld width from >75% of the wall thickness to approximately 60%. It should however be noted that, at this condition, the Au-plating formed a braze on the inside of the weld to form a joint across the whole sidewall thickness. The weld bead showed no sign of cracks, Fig.14
. Some pitting is evident but this is not considered to be detrimental.
Fig.14. CW weld bead profile on Au-plated Kovar package; condition 0.8kW, 3m/min
A joint width of approximately 35 % of the wall thickness was achieved when welding electrolytic Ni-plated packages at 1kW, 3m/min. The weld bead appeared to be smooth but detailed examination revealed subsolidification cracks running transverse to the direction of welding, Fig.15. Those examined did not appear to have penetrated far below the surface and terminated well before the joint interfaces. These cracks were not thought to be detrimental.
Fig.15. CW weld bead profile on electrolytic Ni plated Kovar packages; condition 1kW, 3m/min, x124
When welding electroless Ni-plating it was again found necessary to reduce the energy from 1kW, 3m/min, to 0.8kW, 3m/min, to reduce weld widths to just over 50% (from >75%). At this condition the joint is extended almost across the entire interface by an additional Ni-P braze. Examination of weld profiles showed extensive solidification cracking transverse to the direction of welding, Fig.16. These cracks extended through the full thickness of the weld and propagated into the HAZ beneath the weld. The difference in mechanism and the increase in cracking severity when compared with electrolytic Fe-Ni-Co can be attributed to the presence of P in the electroless plating.
Fig.16. CW weld bead profile on electroless Ni-plated package showing solidification cracks
The temperatures measured at four positions in an unplated package when welded at 1kW, 3m/min, are given in Fig.17. The highest temperature recorded (163°C) was at the base of the package corner. The air temperature within the package did not rise above 45°C. The package temperature decayed to below 50°C in less than 35sec. These temperatures are similar to those recorded when resistance seam sealing flat lid packages using an a.c. power supply and should not inflict thermal damage on any enclosed circuitry.
Fig.17. Temperatures measured at four positions in unplated package when welding a 1kW, 3m/min
Pulsed laser welding
Limited trials were conducted using a 575W pulsed CO 2 laser (minimum spot diameter 0.12mm, maximum pulse length 100msec, minimum pulse rate 1kHz) in an attempt to reduce cracking in Ni-plated packages by using discrete overlapping pulses or spots to form a seam weld. A range of welding conditions was employed including pulses of lmsec applied at frequencies of 150 and 200Hz over a power range of 120-250W and at welding speeds of 0.5 and lm/min. These trials indicated that welds could be made which are of acceptable size and that cracking could be eliminated when using lmsec pulses at 150 and 200Hz at 0.5 and lm/min, 150-200W, on electrolytic Ni-plated packages (see for example Fig.18). Some fine cracking was encountered at the start/end weld at some conditions studied and irregular weld penetration observed in welds made at the lower powers employed (120W).
Fig.18. Pulse welded electrolytic Ni-plated package; condition 150W, 0.5m/min, lmsec pulse at 150Hz, x37
The results of pulsed laser welding electroless Ni-plated packages showed a much reduced incidence of weld cracking from that obtained when CW laser welding. Figure 19 shows a pulsed weld made at 0.5m/min, 150Hz, lmsec pulses, 150W, with no apparent cracking. The limited cracking that was encountered was very fine and probably insignificant. These results indicate that the pulsed laser welding is more likely to produce hermetic sealing of electroless Ni-plated Fe-Ni-Co alloy packages than in CW laser welding, although further optimisation of conditions is clearly needed before this can be firmly established.
Fig.19. General appearance of pulsed weld on electroless Ni-plated Kovar package; condition 150W, 0.5m/min, lmsec pulse at 150Hz, x37
- Continuous wave CO 2 laser welding conditions have been established which give welds with a uniform bead of greater than 35% width of package wall thickness in unplated, electrolytic, and electroless Ni-and electrolytic, Au-plated,Fe-Ni-Co alloy flat lid (0.4mm) packages (29 x 30mm). These conditions comprise a speed of 3m/min, a power of 0.8-1kW, and He shielding.
- Metallurgical examination of packages welded at the above conditions showed acceptable structures in the unplated and electrolytic Au-plated packages. However, welds in electrolytic Ni-plated packages showed transverse subsolidification cracks which would probably not be detrimental, but welds in electroless Ni-plated packages showed extensive transverse solidification cracking which would probably not give leak-tight welds.
- Pulsed CO 2 laser welding can eliminate weld cracking in electrolytic Ni-plated packages and give much reduced incidence in electroless Ni-plated packages from that obtained with CW laser welding. The pulse conditions comprised 0.5-1.0m/min, lmsec pulsed at 150-200Hz and 150-200W.
The results of these programmes demonstrate the ability of the two sealing processes. Resistance seam sealing was found to be capable of producing strong hermetic joints which were tolerant to large variations in machine parameters when operated at optimised conditions (sealing schedule/lid thickness). It does however have the limitation of electrode wear which can introduce reliability problems if good housekeeping is not maintained (cleaning/running in electrodes). Laser welding being a non-contact process avoids this problem. It is also capable of sealing thicker lids (>0.4mm) and joint configurations not possible by resistance seam sealing which may be an increasing advantage as packages increase in size (e.g. SAW packages). Weighed against these advantages is the potential corrosion problem where the plating has been removed from the weld area (unless a non-corrosive lid material is used, e.g. stainless steel or Ni). Also as a result of the small focused spot diameter of the laser the jigging and work handling systems have to be more accurate than with resistance seam sealing and consequently tend to be more expensive.
This work was carried out with the support of the Procurement Executive, UK Ministry of Defence, Directorate of Components, Valves and Devices.