Termination Delamination of Surface-Mount Chip Resistors

Jude M. Runge-Marchese, Taussig Associates, Inc.


From: Handbook of Case Histories in Failure Analysis, Vol 2, K.A. Esakul, Ed., ASM International, 1992

Abstract: Several surface-mount chip resistor assemblies failed during monthly thermal shock testing and in the field. The resistor exhibited a failure mode characterized by a rise in resistance out of tolerance for the system. Representative samples from each step in the manufacturing process were selected for analysis, along with additional samples representing the various resistor failures. Visual examination revealed two different types of termination failures: total delamination and partial delamination. Electron probe microanalysis confirmed that the fracture occurred at the end of the termination. Transverse sections from each of the groups were examined metallographically. Consistent interfacial separation was noted. Fourier transform infrared and EDS analyses were also performed. It was concluded that low wraparound termination strength of the resistors had caused unacceptable increases in the resistance values, resulting in circuit nonperformance at inappropriate times. The low termination strength was attributed to deficient chip design for the intended materials and manufacturing process and exacerbated by the presence of polymeric contamination at the termination interface.

Keywords: Delaminating; Electronic devices

Materials: Gold (Gold); Lead-tin (Lead-tin); Polymer (Polymer, general)

Failure type: (Other, miscellaneous, or unspecified) failure


Background

Several surface-mount chip resistor assemblies failed during monthly thermal shock testing and in the field.

Circumstances leading to failure

The resistor, the only electrically active component in the final assembly, exhibited a failure mode characterized by a rise in resistance out of tolerance for the system. The failures occurred during monthly testing after 100 h of thermal shock (from -65 to 125°C, or -85 to 255°F, 30 min at each extreme for 100 cycles) and “suddenly” in the field (e.g., overnight).
As part of a quality control test program, a die shear test was performed on the resistors after solder attachment to the carrier strips. The most recent attachment shear forces varied from 0.4 to 2 kg (0.8 to 4.5 lb.). The resistors with low force values reportedly exhibited entire termination failure. The attachment never broke within the solder joints. Rather, the termination consistently came apart from the substrate.

Pertinent specifications

A chip detail drawing (Fig. 1) indicated that the chip substrate was high-purity alumina. The resistive element was a sintered thick film that was coated with a protective glass film after laser trimming and finished with an epoxy coating. Continuity through the resistive element was established by solder attachments to the assembly lead frame through edge terminations.
Figure 1

Fig. 1  Chip detail schematic, indicating chip construction as having both a top and bottom termination. Actual construction and sales literature indicated only a top land termination. The edge termination has a wraparound configuration. Dimensions given in inches

The terminations were of a lamellar design. Initial resistive element/termination contact was at the top land termination. The wraparound portion of the terminations coated the top land termination and “wrapped around” the ends of the substrate. No material was specified for the termination layers; however, it was specified that the wrap-around or edge terminations were to be coated with a nickel barrier layer (electroplated) and finished with electroplated tin deposit for solderability (Fig. 1).
Termination/carrier strip attachment was reportedly performed in a belt furnace. The carrier strip was stamped from coin silver strip. Solder paste (Pb-8Sn-2Ag) and type RMA flux were used to make the attachment. The reflow profile was such that the temperature ramped up to a maximum of 330°C (625°F), which is 40°C (70°F) higher than the liquidus temperature of the solder. The dwell time above the liquidus temperature was 110s.
The carrier strip/clip assembly was overmolded with Valox resin at a temperature of 250°C (480°F); a recommended operating range of 240 to 260°C (465 to 500°F) was specified. Valox is a General Electric polymer blend of Lexan, a GE polycarbonate, and polybutylene terepthalate.
Changes in resistance were often measured after solder attachment and encapsulation. Changes were also determined after thermal shock and were measured on returns from the field.

Specimen selection

Representative samples from each step in the manufacturing process were selected for analysis. Additional samples were selected representing the various resistor failures. Samples submitted for analysis were divided into the following groups:
Group No.
Description
1
As-received resistor chips from storage
2
Chips exposed to the solder reflow temperature of 330°C (625°F) for a dwell time of 60 to 90 s
3
Chips soldered on a silver carrier strip and washed in methylene chloride after attachment in order to remove flux residue
4
Carrier strip/chip assemblies as encapsulated with Valox resin
5
Encapsulated resistor assembly that exhibited a failure rise in resistance after thermal shock. A void in the solder fillet was noted.
6
Assembly that reportedly exhibited a cracked solder joint
7
Assembly that exhibited an increase in resistance after thermal shock
8
Several resistor chips that exhibited low shear force values during die shear testing. Some exhibited termination separation from the resistor die.

Testing and Results

Surface examination

Visual. The chips were soldered with the resistive element toward the carrier plate. The resistor/carrier assemblies were encapsulated with Valox and cut from the carrier strip.
The resistor chips were rectangular with metallic end terminations. The resistive element was coated with a gold-colored polymeric material. The material appeared soft, but could not be indented with a fingernail. The gold coating apparently turned brown when exposed to the heat of soldering (Fig. 2).
Figure 2

Fig. 2  Group 1 (left) and Group 2 resistors. The group 2 resistor had been exposed to the heat of soldering for 90 s. Note the dark appearance of the top polymer coating. This is due to the pyrolysis, or breakdown, of the epoxy coating. 19.25×

Examination of various group 8 resistors, which failed by die shear, revealed two different types of termination failures. The first, and apparently most prevalent, type was that of total delamination (Fig. 3). The end terminations peeled from the substrate, exposing the ceramic ends and fracturing the polymer coating. It was noticed that the polymer coating on the failed resistors appeared darker in color than those in the as-received condition. The gold polymer coating also appeared to have melted in the area of the delamination.
Figure 3

Fig. 3  Delaminated group 8 resistor that exhibited a low pushoff strength. Note the puckered appearance of the gold polymer coating at the left termination. This was caused by the heat of soldering exceeding the Tg of the epoxy. 21.6×

The second type of failure noted was that of partial determination. The gold-colored polymer coating did not appear as dark on these failures. The failed termination appeared to fracture at the end and around the bottom of the resistor.
Scanning Electron Microscopy/Fractography. In order to determine at which interface the resistors were delaminated, several samples from group 8 were examined using a scanning electron microscope (SEM). Examination of the fractured termination confirmed visual findings that the fracture occurred at the end of the termination, exposing the ceramic substrate. The top land termination adjacent to the resistive element was apparently intact (Fig. 4). Examination of the delaminated termination revealed no evidence of fracture. The edge termination appeared to have peeled from the end of the resistor (Fig. 5 and 6).
Figure 4

Fig. 4  SEM micrograph of a resistor with a partially delaminated termination. The light-appeahng portion is the lead-tin alloy on the surface of the termination. The dark-appearing area beneath the lead-tin alloy is exposed alumina.

Figure 5

Fig. 5  SEM micrograph of the delaminated termination. Note the gray scale difference between the surface of the delaminated termination and that of the alumina substrate in Fig. 6. This is because delamination occurred at the cermet/thick-film polymer interface.

Figure 6

Fig. 6  Higher-magnification examination of the surface of the delaminated portion of the resistor termination. Note the absence of evidence of any type of fracture or tearing on the surface.

Energy-dispersive x-ray spectrographic (EDS) analysis was performed on the delaminated surface and determined the presence of zirconium, silver, silicon, lead, manganese, magnesium, and aluminum. The aluminum was most likely from the alumina resistor substrate. Silicon, zirconium, magnesium, and manganese oxide are various glass formers and metal oxides used in forming a ceramic-to-metal bond on chip components, most likely the top land termination. The silver component may have been the metallic portion of the metallization, which enabled subsequent electro deposition of the nickel barrier and solderable tin coatings.

Metallography

Microstructural Analysis. In order to determine the construction integrity and material failure mode of the resistor chips, transverse sections of samples from each of the groups were metallographically prepared. Examination was performed with a calibrated metallurgical microscope with magnification capabilities up to 2000×.
Examination of the group 1 resistors, those which reportedly were in the as-received condition, revealed the following termination construction. The top land termination contacted the resistive element and ended at the edge of the substrate. It exhibited a typical cermet (ceramic-metal) metallization that had been fired to fuse the termination to the alumina substrate. The gold-colored polymer that coated the resistive element often overlapped the cermet top land termination.
The top land termination was the only portion of the termination that exhibited actual ceramic-to-metal bonding. The wraparound portion of the termination was achieved with what appeared to be conductive (metal-loaded) polymer-base thick-film ink. Small, dark-appearing amorphous areas dotted the interface between the thick-film ink and the cermet metallization. The termination appeared to have been plated with two distinct metallic layers (Fig. 7).
Figure 7

Fig. 7  Chip cross section from group l, those samples taken in the as-received condition prior to processing. As polished. 87×

Examination of the metallization interface determined that each of the layers composing the termination was separate and distinct. There was no evidence of intermetallic formation between the cermet and thick-film ink. The presence of the dark-appearing amorphous phase at this interface was confirmed (Fig. 8). Samples from group 2, those that had been processed through the soldering profile, exhibited similar characteristics.
Figure 8

Fig. 8  An interesting detail on the group 1 resistors was the presence of an amorphous dark-appearing phase dotting the interface between the cermet metallization and the thick-film polymer conductive ink. Fourier transform infrared analysis determined that this dark-appearing phase was actually resin separation from the epoxy binder in the thick-film ink. The following features can be discerned: area 1 is the alumina substrate; area 2 is the cermet metallization; area 3 is the thick-film polymer ink; area 4 is the electroplated nickel barrier; and area 5 is the lead-tin surface alloy 31.0×

In order to understand the metallurgical ramifications of a reported failure, a transverse section through a sample from group 6 was metallographically prepared. It was reported that the sample exhibited a cracked solder fillet. Examination of the cross section revealed gross delamination of the terminations (Fig. 9). So severe was the delamination that the encapsulant had flowed into the interface between the cermet metallization and the polymer thick-film ink. This expanded the plating and thick-film ink layers from the substrate to the point of rupture at the top of the termination.
Figure 9

Fig. 9  Delaminated termination on a group 6 sample, showing Valox (the pellet encapsulant) intruding on the cermet/polymer thick-film metallization interface. Metallization forced the expansion of the termination, rupturing the nickel barrier and giving the solder joint a cracked appearance

Metallographic examination of the group 6 samples and samples from the remaining groups determined the presence of consistent interfacial separation between the cermet and polymer thick-film layers. The condition ranged in severity from its presence as an amorphous layer of contamination to that of total delamination. Gross delamination was consistently observed in the group 3 samples (Fig. 10), indicating that the solder attachment of the chip to the carrier strip was the point in the manufacturing process that produced the failure condition.
Figure 10

Fig. 10  Consistent gross delamination in the group 3 samples, those that had been soldered to the carrier plate. Measurable separation was noted at the metallization interface. 124×

Representative sections from groups 2 and 3 were analyzed using electron probe microanalysis (EMA) to determine the constituents of the resistor materials and possibly to identify the contaminating layer present at the metallization interface. EMA of the group 2 sections (see Fig. 11) determined that the resistor substrate was high-purity alumina. The metal portion of the cermet metallization was silver rich; the ceramic portion was zirconium rich. Significant amounts of silicon were also detected, most likely present as an oxide or glass former. Segregation due to diffusion and mass transport, typical for a ceramic-to-metal bond, was observed. The heaviest concentration of the oxides and glass formers was toward the substrate, whereas the silver remained at the surface for subsequent metallic joining operations.
Figure 11

Fig. 11  SEM micrograph of a group 2 resistor termination. Areas of interest are as follows: area 1 is the alumina substrata area 2 is the cermet thick-film metallization; area 3 is the polymer thick-film ink; area 4 is the nickel barrier layer; and area 5 is the lead-tin surface coa ting. Arrows indicate the critical interface. Note at this interface, even in the group 2 components, the appearance of a dark thin layer at the cermet/polymer thick-film interface

EDS analysis of the wraparound metallization determined that the metallic portion was silver. The dark phase was carbon rich, supporting the visual observation that the binder in the thick-film ink was polymer based. EDS analysis of the electroplated layers confirmed that the barrier layer was nickel and that the surface layer was tin-lead alloy. The gold-colored polymer coating over the resistive element was a metal-oxide-loaded polymer with a significant amount of chlorine, probably an epoxy.
EMA of the sections of the group 3 samples that exhibited delamination (see Fig. 12) confirmed the metallographic analysis results by determining the presence of an amorphous contaminant at the cermet/polymer thick-film interface. Higher-magnification examination of this layer revealed the presence of voids within the amorphous material. EDS analysis of the material determined that it was organic in nature, with a significant amount of chlorine, and probably was an epoxy.
Figure 12

Fig. 12  Interface on a group 3 resistor. Note the increased thickness at the critical interface. This was caused by the forces exerted by the expansion and contraction of the silver carrier plate during soldering. The polymer contamination (most likely resin separation from the epoxy thick-film ink) at the surface of the polymer metallization layer created a nonadherent film at the interface

Chemical analysis/identification

Fourier Transform Infrared Analysis. In order to discern the compositions of the polymer constituents of the resistors, Fourier transform infrared (FTIR) analysis was performed. Specifically, the gold coating, the polymer-base thick-film ink, and the interfacial contamination layer were compared to try to determine the nature of the amorphous layer consistently observed at the metallization interface.
FTIR analysis of the gold coating material determined that it was an anhydride-cure epoxy. The significant chlorine content determined by way of EDS suggested that the curing agent for the epoxy was chlorendic anhydride. This material, thermo-setting in nature, will not melt when exposed to elevated temperatures. Instead, the epoxy exhibits a glass transition temperature, or T g. This is the point at which the amorphous phase of the polymer changes from a hard brittle condition to a viscous rubbery condition. This could explain the ductile, puckered appearance of the gold coating on the delaminated resistors.
The T g for most anhydride epoxies is about 130°C (265°F); some formulations (specifically, di-anhydride cure epoxies) exhibit T gs as high as 280°C (535°F). It is never advisable to subject epoxies to temperatures in excess of the T g for an extended period of time.
The polymer binder from the thick-film ink on several delaminated samples was removed by squeezing the silver flakes of the metallization together and extruding the binder between them. FTIR determined that this material consisted of polyamide and an epoxy similar to the polymer epoxy gold coating. These results were consistent for the binder in the thick-film ink on resistors from groups 1 and 2. These consistencies indicated that the epoxy contained a polyamide curing agent. This material, if fully cross-linked, would not melt, but would also exhibit a T g.
The metallization interface was analyzed by preparing samples from group 3, the group that exhibited consistent gross delamination. Samples were prepared for transmission FTIR by replication in amyl acetate on potassium chloride blanks. During preparation, it was noted that a significant amount of a dark yellow, oily substance was present at the interface. FTIR analysis determined that it was a triglyceride. Triglycerides are often added to epoxy formulations as drying oils. They are also typically found in animal or vegetable fats and can be attributed to handling. However, glycidyl groups are the most common constituents of epoxy resin functions; exposure to acids from soldering flux or residual cleaning solutions can convert un-cross-linked resin to triglycerides by way of nucleophilic substitution. This suggested that the amorphous contamination observed at the metallization interface was the result of epoxy degradation by way of resin separation from the thick-film ink.
To determine how early in the manufacturing process resin separation was apparent in the chip resistors, samples from groups 1 and 2 were prepared for analysis by forcing delamination and replicating the surface as described above. Smaller amounts of an oily substance, not yellow in color, were extracted from the samples. FTIR analysis again determined the oily substance to be a triglyceride, suggesting that degradation was present in chips in the as-received condition.
Analysis of the delaminated surface on the group 8 samples revealed the presence of human skin cells. These skin cells could certainly contribute to delamination failures; however, they could also be the result of improper handling of the resistors after failure and prior to receiving them for analysis.

Discussion

Metallographic analysis determined that the failure mode exhibited by the resistors, that of an unacceptable increase in resistance during test and during service, was due to delamination of the wraparound or edge portion of the terminations. The focus of the failure analysis thus turned to determination of the nature and cause of the delamination.
The consistent presence of an amorphous, apparently polymeric layer at the metallization interface required FTIR analysis. By determining the nature of the layer via FTIR, the source for failure was indicated as both material and process related. In other words, the resistor termination design, because it incorporated polymeric materials and thus limited metallurgical bonding, was not appropriate for the recommended soldering and encapsulation profiles. This inadequacy was amplified by the presence of the layer of polymer-based contamination at the metallization interface.
In order to support this conclusion, it is necessary to understand the failures. Two types of failures were observed: total and partial delamination.
Partial delamination occurred at the metallization interface, confined to the portion of the wraparound termination away from the top land termination. This was primarily due to the absence of a ceramic-to-metal bond in this portion of the termination. Termination fracture occurred when die shear forces exceeded the mechanical bond force between the ink and the substrate during die shear testing. During manufacturing, fracture occurred when the resistor assembly was manufactured with the top land termination up and shear forces exerted by the thermal expansion and contraction of the silver carrier plate during soldering and encapsulation exceeded the strength of the mechanical bond between the termination and the resistor substrate. These failures were the easiest to cull out, because intermittent performance or abrupt failure was often revealed during thermal shock testing.
The more subtle and most prevalent delamination type of failure also occurred at the metallization interface. The primary source for total delamination was also the absence of a true ceramic-to-metal bond at this interface. Exacerbating this lack of bond was the presence of interfacial polymer-based contamination.
The consistency of a concentrated film or layer of amorphous polymer film at the metallization interface suggested that the epoxy binder in the thick-film ink was neither homogeneous nor stable. Resin separation from the epoxy binder created an oily, nonadherent film at this critical interface. This meant that not only was there no ceramic-to-metal bond fixing the termination, but there also was no epoxy adhesive bond. The amount of resin separation varied from chip to chip, with those with worst-case adhesion exhibiting the most resin separation and those off the shelf with apparently good termination strength displaying only a small amount. It was observed that this layer became more pronounced as the resistors were exposed to the heat of processing. The dark yellow color of the oily residue observed at the metallization interface of the group 3 samples was most likely due to thermal breakdown of the resin during attachment to the carrier strip.
Gold coat epoxy overlapped the top land termination of many of the samples. While this was probably a process control problem during component manufacture, the dimpled, reflowed appearance of the material on failed components indicated that the heat of processing the resistor assemblies exceeded the T g for the gold coat. Its softening would certainly undermine termination strength in the area, aiding thermal expansion forces in pulling the terminations from the resistor substrate.
In the resistor assembly manufacturing process, as the components are soldered to the silver carrier plate, they are soaked above 290°C (555°F), peaking at 330°C (625°F) for 110 s. Even assuming the highest possible T g for the epoxy systems in the chip, 280°C (535°F), visual evidence supports the conclusion that the material certainly became viscous during this operation. Any polyamide curing agent or uncured resin that had not reacted during the cure cycle for the polymer-base thick-film ink degraded. Thermal expansion forces exerted by the carrier strip on the metallization interface pulled the terminations apart. Because nothing other than the wraparound termination was making contact with the top land termination and the resistive element, an open condition or measurable increase in the resistance value for the assembly was created.

Conclusion and Recommendations

Most probable cause

Analysis of several groups of surface-mount chip resistors determined that sudden unacceptable increases in the chip resistance values both during testing and in the field were caused by inadequate wraparound termination strength. The source of the low termination strength was twofold. The first and most prevalent cause appeared to be inadequate component design for the intended manufacturing process. The second cause was polymeric contamination at the termination metallization interface.

Remedial action

In order to manufacture a resistor assembly with chips of this design, the attachment solder and reflow profile must remain below the T g for the epoxy systems manufactured into the component. A new encapsulant and encapsulation procedure are needed that will not exceed the critical T g.
Variations in strength due to the presence of contamination at the metallization interface or degradation of the polymer constituents should be determined by nondestructive die shear testing. Chips soldered to carriers should be subjected to a nominal force for a period of time, released, and retested for shifts in the resistance values. Failed chips would be those that exhibit a shift in resistance out of the accepted range.
Rather than alter the materials and add tests to the current process, it is also recommended that a resistor that exhibits a complete ceramic-to-metal wraparound termination be used. Manufactured properly, the resistor should be able to withstand the heat of soldering and encapsulation without delaminating.
It is imperative when considering a new resistor design that the processing recommendations by the chip manufacturer be addressed. It still may be necessary to alter the assembly process and materials to reflect the temperature requirements of the chip.