IEEE TRANSACTIONS ON ADVANCED PACKAGING, VOL. 28, NO. 3, AUGUST 2005
503
Environmental Qualification Testing and Failure
Analysis of Embedded Resistors
Lawrence John Salzano, II, Chris Wilkinson, and Peter A. Sandborn, Senior Member, IEEE
Abstract—Embedding passive components (capacitors, resistors, and inductors) within printed wiring boards (PWBs) is one of
a series of technology advances enabling performance increases,
size and weight reductions, and potentially economic advantages in
electronic systems. This paper explores the reliability testing and
subsequent failure analysis for laser-trimmed Gould subtractive
nickel chromium and MacDermid additive nickel phosphorous
embedded resistor technologies within a PWB. Laser-trimmed
resistors that have been “reworked” using an inkjet printing
process to add material to their surface to reduce resistance
have also been considered. Environmental qualification testing
performed included: thermal characterization, stabilization bake,
temperature cycling, thermal shock and temperature/humidity
aging. In addition, a pre/post-lamination analysis was performed
to determine the effects of the board manufacturing process on
the embedded resistors. A failure analysis consisting of optical
inspection, scanning acoustic microscope (SAM) and environmental scanning electron microscope (ESEM) imaging, and PWB
cross-sectioning was employed to determine failure mechanisms.
All the embedded resistors were trimmed and the test samples
included resistors fabricated both parallel and perpendicular
to the weave of the board dielectric material. Material stability
assessment and a comparison with discrete resistor technologies
was performed.
Index Terms—Embedded passives, environmental testing, integral passives, laser trimming, reliability, resistors.
I. INTRODUCTION
D
ISCRETE passive components are continuing to increase
in use in electronic systems even though manufacturers
are increasing the degree of system integration. In 2001, passive devices accounted for 91% of components, 41% of board
area, and 92% of all solder joints in an electronic system, but
only 2.6% were integrated in some fashion [1]. Driven by performance, size, and economic concerns, embedded passives were
introduced to the market in the early 1980s. Embedded passives,
also known as integral passives, are passive components buried
in interconnecting substrate materials. The potential advantages
offered as a result of embedding passives include: increased circuit density, improved electrical properties, cost reduction, increased product quality, and improved reliability.
Embedded resistors are manufactured from both thin- and
thick-film technologies by depositing and pattering layers of
Manuscript received June 28, 2004; revised December 9, 2004. This work
was supported in by the Computer-Aided Life Cycle Engineering (CALCE)
Electronic Products and Systems Center and, specifically, the members of the
CALCE Consortium.
The authors are with the CALCE Electronic Products and Systems Center,
Department of Mechanical Engineering, University of Maryland, College Park,
MD 20742 USA.
Digital Object Identifier 10.1109/TADVP.2005.848387
Fig. 1.
TV-1R embedded resistor test vehicle [13].
resistive material in conjunction with interconnect lines within
a substrate. The primary driver for replacing surface-mount
resistors with embedded resistors is to enable faster bus speeds.
As bus speeds increase, the electrical termination requirements
increase, and thus the number of resistors required increases. In
order to reduce transmission times and improve performance,
embedded resistors can be buried in the substrate to minimize
the parasitics (inductance and capacitance) generated by interconnects. In addition, embedded resistors free valuable board
surface area and potentially improve reliability through the
elimination of solder joints and plated through-holes characteristic of surface mount and through-hole technologies.
One significant factor governing the applicability of embedded resistors is their tolerance level. The tolerance to which
a resistor can be fabricated determines the applications for
which it can be used. Tolerances of 10 or larger are readily
achievable with today’s embedded resistor technologies, however, achieving 1 is a challenge [2]. While surface-mount
resistors can be presorted by value, or even replaced during
assembly when their value is not within the required range,
embedded resistors provide no such opportunity and must
be within design tolerance value before the board fabrication
process is completed. One possible impediment to the widespread use of embedded resistors is the ability (and expense) of
tuning or trimming the resistors to the appropriate value range
(as defined by the design tolerances) prior to the lamination of
the layer pair containing them into the board [3].
Laser trimming of film resistors has been performed for many
years with application to resistors on silicon and trimming of
surface mount discrete resistors prior to packaging, e.g., [4].
1521-3323/$20.00 © 2005 IEEE
504
IEEE TRANSACTIONS ON ADVANCED PACKAGING, VOL. 28, NO. 3, AUGUST 2005
Fig. 2. Reliability test plan. The numbers located in the arrow paths indicate the quantity of each type of board subjected to the proceeding test.
However, only recently, highly automated laser trimming technologies have been developed and demonstrated for trimming
of embedded resistors during the board fabrication process [2].
Resistors are normally trimmed by micromachining a trough
in the resistive element. The length and path of the trough determine the resistance change obtained. As the laser cuts the
trough, the resistor value is measured and used as feedback to
control the trimming process. Several differently shaped paths
can be used depending on the specific trimming requirements.
High-precision laser trimming of buried resistors can tailor resistance values to within less than 1% tolerance of target values
[2].
Laser trimming is limited to only increasing the resistance
value of embedded resistors—the value of a resistor cannot be
decreased by removing material. If laser trimming is a manufacturing option, embedded resistors are usually designed so
that the distribution of resistance values resulting from their
manufacturing process is centered on a value that is lower
than that required by the application—trimming is then used
to increase the resistance to the desired value. Nonetheless,
some fraction of manufactured resistors will have higher than
required resistance values due to the distribution of the original manufacturing process, material voids encountered during
trimming, or trimming errors. Prior to completion of the board
fabrication process, it is also possible to perform a material-ad-
dition “rework” process on embedded resistors that have too
high a value. One method of reworking embedded resistors is
to print conductive ink on the surface of an embedded resistor,
thus, adding a parallel resistor that effectively “trims down”
the resistor value [5].
A. Embedded Resistor Reliability
Since embedded passives cannot be replaced after the board or
substrate is completed, long-term reliability and yield are major
concerns for manufacturers. One of the key reliability issues for
embedded resistors is how well the resistive materials are able
to adhere to the surface of the substrate onto which they are
deposited [6]. Furthermore, delaminations between the resistor
and copper leads are possible in addition to chemical interactions
between resistor and copper electrode materials [7]. These
failure mechanisms are a result of manufacturing processes
and application-specific environmental and mechanical loads
that subject the embedded resistors to temperature variations and
substrate deformations. Testing embedded passive components
for these particular failure mechanisms can be achieved through
temperature shock, temperature cycling and various bending
and torsion loading tests [6]. Zhou et al. [8] conducted thermal
cycling and electrostatic discharge (ESD) tests on laser-trimmed,
embedded ceramic paste resistors measuring 100, 1000, and
SALZANO et al.: ENVIRONMENTAL QUALIFICATION TESTING AND FAILURE ANALYSIS OF EMBEDDED RESISTORS
505
Fig. 3. Pre/post lamination analysis. These results are for all 40 fabricated boards.
10 000 and ceramic capacitors of large and small sizes buried
in FR4 boards. No induced failures were observed in any of
the tested materials. Gould Electronics, Inc. reviewed in-house
reliability assessments of two different types of subtractive
thin-film nickel–chromium alloy resistors subject to humidity,
thermal cycling, conventional reflow, and soldering heat tests
[9]. The effect of power dissipation on embedded thin-film
nickel–chromium resistors has been touched upon; however, no
environmental qualification was performed [10]. Felten et al.
[11] conducted thermal cycling and ESD testing on ceramic
embedded resistors and capacitors; however, their research
was limited in the same manner as Zhou et al. Fairchild
et al. [12] conducted reliability tests on flexible thin-film
embedded resistors and electrical characterization of thin-film
embedded capacitors and inductors; however, laser-trimmed
(and material-addition reworked) embedded resistors were not
included within the study.
Generally, all the previous reliability assessment work on
embedded resistors has only treated subtractive technologies
and not considered reworking. The previous studies do not
consider thermal characteristics, long-term stability, thermal
shocking, and humidity exposure, and since no failures were
observed, subsequent failure analysis was not performed. This
paper presents more extensive environmental qualification
testing results for both additive and subtractive embedded
resistor technologies than previously reported and includes
detailed failure analysis.
For embedded passives to insert themselves into the mainstream market, performance characteristics and material
stability must be comparable to, or out perform existing discrete passive technologies. Therefore, long-term operation
and storage reliability needs to be assessed for tight-tolerance
embedded resistors. Section II of this paper summarizes the
test vehicle and test plan. Section III summarizes the results
506
IEEE TRANSACTIONS ON ADVANCED PACKAGING, VOL. 28, NO. 3, AUGUST 2005
TABLE I
TCR COMPARISON BETWEEN GOULD AND MACDERMID EMBEDDED RESISTORS IN ppm= C ( 25 C TO 125 C RAMPED OVER A 4-h TIME PERIOD)
0
obtained from the testing, and Section IV provides the results
of the associated failure analysis.
II. ENVIRONMENTAL TESTING
A. Test Vehicle
The TV-1R test vehicle is a multilayer printed wiring
board (PWB), approximately 5-in square, manufactured from
etched FR-406 copper panels originally designed by Delphi
Automotive Systems for the NIST Advanced Embedded
Passives Technology Consortium [13]. The FR-406 material
is an epoxy–glass specifically selected for its characteristic
high glass transition temperature (T
170 C) enabling the
TV-1R board to withstand severe environmental regimens.
Layer 2 of the PWB contains an array of 20 cells containing
approximately 600 embedded resistors of the two distinct material/fabrication approaches described in the next paragraph.
The 20 cells are equally divided into ten horizontal and ten
vertical orientations. Fig. 1 shows the TV-1R test vehicle with
one of the resistor cells magnified. The TV-1R test vehicle is
constructed with only square aspect ratio resistors; however,
an assortment of different size resistor squares and I/O track
widths exist throughout the board. Resistor sizes range from
10 to 50 mils, and the I/O track widths vary from 5 to 20 mils.
Cell arrangements contain two distinctive internal patterns;
either a daisy-chained configuration consisting of four strings
of nine resistors (36 resistors total) or isolated resistors. A 100
contact edge connector is located at the top of the PWB through
which the resistance of daisy-chained and/or isolated resistor
configurations can be measured.
Fabrication of the inner layer pairs and final board lamination was completed by Merix Interconnect Solutions. TV-1R
test vehicles were fabricated using two types of embedded
resistor materials/approaches, provided by Gould Electronics
and MacDermid Inc. The MacDermid M-Pass nickel–phosphorous (NiP) material is plated directly onto the inner layer
pairs of the FR-406 PWB using an additive process, [14].
Meanwhile, Gould (TCR Thin Film Embedded Resistor Foil)
uses a dedicated layer pair in conjunction with a subtractive
technique starting with a copper foil coated with resistive
nickel–chromium (NiCr) material that requires a multistage
photoresist and etching process to fabricate the required pattern
[9].
Electro Scientific Industries (ESI) conducted the embedded
resistor trimming process for the TV-1R test vehicle. MicroFab
Technologies, Inc. performed the embedded resistor materialaddition rework for the TV-1R test vehicle by inkjet printing
a proprietary conductive polyimide-based ink onto the surface
of embedded resistors to lower individual resistor resistance
values.
B. Manufacturing Process and Test Plan
Forty printed wiring boards of each type of embedded resistor
technology/material (MacDermid and Gould) were fabricated
and 27 of each type (54 boards total) were subjected to
environmental testing. All of the reliability tests were conducted
according to the process flow in Fig. 2. Testing commenced
with initial metrology to determine baseline resistance values.
Then, two boards of each type were used to establish upper
and lower bound temperatures of both the TV-1R PWB test
samples and the environmental testing chambers. Then, subsets
of samples (five boards of each type, ten boards total) were
subjected to linearly increasing temperature, stabilization bake,
thermal shock, temperature cycling, and temperature/humidity
accelerated aging profiles. During predetermined, periodic time
intervals, the TV-1R boards were measured to extract resistance
values. Finally, after running each of the qualification tests,
data and failure analysis was performed on the test samples to
determine material stability, operating performance, reliability
attributes, and failure mechanisms characteristic to Gould’s
subtractive NiCr and MacDermid’s additive NiP laser trimmed,
and material-addition reworked and nonreworked embedded
SALZANO et al.: ENVIRONMENTAL QUALIFICATION TESTING AND FAILURE ANALYSIS OF EMBEDDED RESISTORS
507
Fig. 4. Gould embedded resistor stabilization bake analysis after 2100 h at 105 C.
resistors in daisy-chained or isolated configurations, oriented
horizontally or vertically to the board weave of various size
resistive elements and I/O tracks.
No preconditioning of the boards was performed prior to
testing in order to avoid compounding the results with additional unknown variables. It would, however, be beneficial
to conduct a series of experiments with preconditioned test
vehicles subjected to a reflow profile to simulate a full electronic system manufacturing process before environmental
qualification commenced. It should also be noted that the
qualification testing reported in this paper (as with the previous
work referenced in Section I) is for unbiased resistors, i.e.,
all testing was performed with no current passing through the
resistors—biased qualification testing could yield differing
results.
III. TEST RESULTS
Each test was conducted as described by the reliability test
plan presented in Section II. Before environmental testing
commenced, every TV-1R PWB was tested for initial baseline
resistance values. Then, each board was subjected to a particular
environmental qualification test and periodically retested as
prescribed in the appropriate test procedure. Finally, the acquired
data was grouped and sorted according to individual embedded
resistor characteristics (technology, material-addition reworked,
connection, and orientation) in order to generate time-dependent
and statistical representations presenting aggregate resistance
fluctuation within the sample population.
In order to present the acquired embedded resistor results in
a compact fashion, histograms were generated for each of the
508
Fig. 5.
IEEE TRANSACTIONS ON ADVANCED PACKAGING, VOL. 28, NO. 3, AUGUST 2005
MacDermid embedded resistor stabilization bake analysis after 2100 h at 105 C.
qualification tests. For all of the tests except the thermal characterization regimen, a percent change in resistance for each
test location was calculated. A percentage increase indicates
an increase in resistance while a decrease reflects a decrease
in resistance. Vertical lines are provided on each histogram
representing zero change in resistance. The total number of
data points used to create a histogram appears in the corner of
each distribution and represents the total quantity of measured
embedded resistors, which is a combination of isolated and
daisy-chained measurements. Justification for combining isolated and daisy-chained measurements into single data counts
was the result of analyzing each environmental qualification
test’s data sorted into isolated and daisy-chained configurations.
No quantifiable shift occurred in any portion of the distributions as a result of segregating the data; therefore, combined
isolated and daisy-chained resistor results are represented in
the histograms.
Since trend data demonstrates dynamic behavior over time,
time-dependent plots showing dynamic resistance change are
included for each of the qualification tests in addition to the
histograms presenting static final resistance change.
A. Pre/Post Lamination Analysis
In order to determine the effect on the embedded resistors due
to the PWB lamination process, each individual resistor on the
resistive layer was measured before the layer pairs were laminated together. The resistors were then measured again after
lamination.
Fig. 3 presents the aggregated resistance percent change distribution for both Gould and MacDermid resistor technologies.
The results are ranged over 50 for Gould and 10 for
MacDermid in order to show the histograms clearly. 195 values
or 1.23% (Gould) and 325 values or 1.27% (MacDermid) of percent change values are outside these ranges.
SALZANO et al.: ENVIRONMENTAL QUALIFICATION TESTING AND FAILURE ANALYSIS OF EMBEDDED RESISTORS
Fig. 6.
Gould embedded resistor temperature cycling analysis after 500 cycles between
Gould’s subtractive embedded resistors exhibited a positive
shift in mean resistance (3.85%), while MacDermid’s additive
resistors showed a negative mean shift ( 2.68 ). Gould’s man7.46 ,
ufacturing process has a larger standard deviation,
compared to MacDermid’s,
1.94 .
B. Thermal Characteristics
To determine the effective temperature coefficient of resistance (TCR) for both embedded resistor technologies, five
samples of each type were subjected to a linearly increasing
temperature profile ranging from 25 C to 125 C ramped
over a 4-h period. The value of the TCR (ppm C) was deter-
065
509
C and 125 C.
mined using the method in [15] using a reference temperature
of 25 C.
Table I provides a comparison between the mean, standard deviation, and median values for both manufacturers’
embedded resistor technologies for the two orientations and
reworked/nonreworked. Gould embedded resistors exhibited
an aggregate mean TCR of 190.81 ppm C, while MacDermid embedded resistors revealed a more even distribution
between negative and positive TCR with an aggregate mean
of 6.28 ppm C. The Gould TCR is more than an order of
magnitude greater than MacDermid’s. The standard deviation
of TCR for MacDermid is greater than that of Gould (26.23
and 7.76 ppm C, respectively).
510
Fig. 7.
IEEE TRANSACTIONS ON ADVANCED PACKAGING, VOL. 28, NO. 3, AUGUST 2005
MacDermid embedded resistor temperature cycling analysis after 500 cycles between
We also noted that the TCR profile of the Gould resistors
was essentially positive linear, while the overall trend line of
the MacDermid characteristic was marginally negative.
C. Stabilization Bake Analysis
To determine the sensitivity of the TV-1R test vehicle to
storage at elevated temperatures without electrical stress applied, a stabilization bake at 105 C for 2100 h was performed.
Table II provides a comparison between the most frequently
occurring peak interval of the distribution and standard deviation for both manufacturers’ materials/approaches. Gould
065
C and 125 C.
embedded resistors exhibited a symmetric distribution of resistance change, while MacDermid resistors are weighted toward
increases in resistance (positively). Figs. 4 and 5 provide percent change versus time for a median case resistor as well as
aggregate percent change after 2100 h for all embedded resistor
measurements.1 Gould’s resistors stabilized more quickly than
MacDermid’s at 750 h to MacDermid’s 1500 h. MacDermid
embedded resistors have one less annealing procedure during
the manufacturing process than Gould embedded resistors, this
may be the reason why they take longer to stabilize.
1The bar labels on the histograms in Figs. 4–11 represent the value of the end
of the interval associated with the bar.
SALZANO et al.: ENVIRONMENTAL QUALIFICATION TESTING AND FAILURE ANALYSIS OF EMBEDDED RESISTORS
Fig. 8. Gould embedded resistor thermal shock analysis after 500 cycles between
D. Temperature Cycling Analysis
To determine the sensitivity of the TV-1R test vehicle to
extremes of high and low temperatures and the effect of
alternate exposures to those extremes, temperature cycling
between 65 C and 125 C for 500 cycles was performed.
Table III provides a comparison between the mean, median, and
standard deviation of the distribution for both manufacturers’
embedded resistor materials/approaches. Figs. 6 and 7 provide
percent change versus cycles for a median case resistor as well
as aggregate percent change after 500 cycles for all embedded
resistor measurements. Gould embedded resistors exhibited a
symmetric distribution between negative and positive response
040
511
C and 120 C.
to the temperature cycling profile while MacDermid resistors
were weighted toward increases in resistance (positively). Both
manufacturers’ technologies responded with an approximate,
most frequently occurring 1% change after 500 cycles. In
addition, Gould’s embedded resistors stabilized immediately
after 20 cycles, meanwhile MacDermid’s took almost 450
cycles.
E. Thermal Shock Analysis
To determine the sensitivity of the TV-1R test vehicle
to sudden exposure to extreme changes in temperature and
512
Fig. 9.
IEEE TRANSACTIONS ON ADVANCED PACKAGING, VOL. 28, NO. 3, AUGUST 2005
MacDermid embedded resistor thermal shock analysis after 500 cycles between
the effect of alternate exposure to those extremes, thermal
shocking between 40 C and 120 C for 500 cycles was
performed. Table IV provides a comparison between the mean,
median, and and standard deviation of the distribution for both
manufacturers’ embedded resistor materials/approaches. Gould
embedded resistors exhibited a negatively weighted distribution
with a positive peak interval, while MacDermid resistors were
weighted positively except for a few measurements falling into
the negative region. Neither technology stabilized after 500
cycles. Figs. 8 and 9 provide percent change versus cycles for a
median case resistor as well as aggregate percent change after
500 cycles for all embedded resistor measurements.
040
C and 120 C.
F. Temperature/Humidity Analysis
To determine the sensitivity of the TV-1R test vehicle to
humid environments, temperature/humidity accelerated aging
was performed at 130 C with 85% relative humidity under
2.325 atm. Table V provides a comparison between the mean,
median, and standard deviation of the distribution for both manufacturers’ embedded resistor materials/approaches. Figs. 10
and 11 provide percent change versus time for a median case
resistor as well as aggregate percent change after 1244 h for all
embedded resistor measurements. Both Gould and MacDermid
embedded resistors exhibited positively weighted distributions
with a few measurements falling into the negative region.
SALZANO et al.: ENVIRONMENTAL QUALIFICATION TESTING AND FAILURE ANALYSIS OF EMBEDDED RESISTORS
Fig. 10.
513
Gould embedded resistor temperature/humidity analysis after 1244 h at 130 C with 85% relative humidity under 2.325 atm.
Gould’s resistor technology stabilized after 800 h, meanwhile
MacDermid’s resistors continued to increase beyond the 1244-h
test time.
G. Material Stability Assessment
The stability of a resistor refers to how its resistance changes
with time under stressed conditions, e.g., temperature. These
changes may be due to recrystallization, hydration, oxidation,
and/or other chemical alterations of the resistor material as well
as effects at the conductor–resistor interface [16]. Traditional
resistor technologies are classified according to several parameters depending on the specific needs of electronic system manufacturers including: resistance value, tolerance, stability during
storage or operation, and dissipation [17]. Table VI provides an
overview of several resistor technologies as a function of the required precision.
After performing an extensive environmental qualification
regimen on the TV-1R test boards, Gould and MacDermid
embedded resistor technologies performance characteristics
appear to fit within the semi-precision classification for traditional resistor technologies. Yet, for high-speed electronic
systems, resistor tolerance is not as important as the elimination
of the inductive reactance of surface-mount chip components,
vias, and traces. Actual tolerance is a combination of the initial
mismatch of the device value and line impedance, the device
tolerance, and the series inductance and inductive reactance
514
Fig. 11.
IEEE TRANSACTIONS ON ADVANCED PACKAGING, VOL. 28, NO. 3, AUGUST 2005
MacDermid embedded resistor temperature/humidity analysis after 1244 h at 130 C with 85% relative humidity under 2.325 atm.
of the device. Embedded resistors have been found to eliminate almost all of the inductance normally associated with
discrete resistors and their vias, therefore enabling a 10%–15%
embedded resistor to exhibit significant signal integrity improvements over chip and discrete resistors of a 1%–2%
tolerance [18]. Table VII provides a breakdown of both Gould
and MacDermid embedded resistor technologies and compares them against Ohmega Technologies’ well-established
Ohmega-Ply 50 square planar resistor technology [19].
IV. FAILURE ANALYSIS
In this paper, we defined a failure as a shift in embedded
resistance value greater than 50 of the initial unstressed
resistor condition, or the development of an open or short
circuit as a consequence of a particular qualification test. After
conducting data analysis, failed embedded resistor cell locations
were identified within the TV-1R test vehicle’s 20-cell matrix
(see Fig. 1). Table VIII provides the quantity and configuration
of failed resistor measurements for a given test. Two numbers
appear below each embedded resistor configuration in Table
VIII. The number to the left of a backslash denotes the quantity of
failed embedded resistors, and the number to the right signifies
the quantity of resistors adhering to the specified tolerance
limit (nonfailed resistors).
Visual, scanning acoustic microscope (SAM) imaging, and
cross-sectioning of representative failed resistors was performed. Two characteristically different failure characteristics
SALZANO et al.: ENVIRONMENTAL QUALIFICATION TESTING AND FAILURE ANALYSIS OF EMBEDDED RESISTORS
515
TABLE II
STABILIZATION BAKE COMPARISON AFTER 2100 h AT 105 C (PERCENT CHANGE)
TABLE III
TEMPERATURE CYCLING COMPARISON AFTER 500 CYCLES BETWEEN
were apparent upon visually inspection of Gould and MacDermid boards. Failed MacDermid embedded resistors contained light discolorations within the resistor material as
illustrated in Fig. 12(a). Meanwhile, failed Gould embedded
resistors contained dark discolorations in thin and thick jagged
lines as illustrated in Fig. 14(a). SAM revealed possible delamination between the embedded resistors and the FR-406 material
within MacDermid test vehicles as shown in Fig. 12(b)–(d). In
order to identify the failure mechanism, the failed embedded resistor was cross sectioned and observed with an environmental
scanning electron microscope (ESEM). The ESEM pictures
indicated that delamination occurred between the embedded
resistor and the FR-406 inner layer pair. Fig. 13 shows images
of both “good” and “bad” embedded resistors making contact
065
C AND 125 C (PERCENT CHANGE)
with either a left or right trace. As a result of the delamination,
the value of the resistor cross sectioned in Fig. 13 increased
from 52 to 705 .
The SAM also revealed material inconsistencies within the
Gould test vehicles; however, they have a different nature than
those found within the MacDermid boards. Fig. 14(b)–(d) illustrate a failure instance within the Gould sample population. The
boxed region in the optical image provided by Fig. 14(a) suggests the occurrence of an anomaly; however, upon examining
the through transmission C-scan of the site, no additional
information could be determined about the failure mechanism.
Meanwhile, conducting a pulse echo, peak amplitude C-scan
at the resistor level disclosed the possibility that the darkening
region within the resistor was a metallic growth between copper
516
IEEE TRANSACTIONS ON ADVANCED PACKAGING, VOL. 28, NO. 3, AUGUST 2005
TABLE IV
THERMAL SHOCK COMPARISON AFTER 500 CYCLES BETWEEN
040
C AND 120 C (PERCENT CHANGE)
TABLE V
TEMPERATURE/HUMIDITY COMPARISON AFTER 1244 h AT 130 C WITH 85% RELATIVE HUMIDITY UNDER 2.325 ATM (PERCENT CHANGE)
TABLE VI
BREAKDOWN OF ESTABLISHED RESISTOR TYPES
FUNCTION OF PRECISION [17]
AS A
traces either above or below the embedded resistor. When a
pulse echo, phase inversion C-scan was performed and the
TABLE VII
BREAKDOWN OF EMBEDDED RESISTOR TECHNOLOGIES
corresponding darkened region in the optical image was overlaid with the various colorations generated, the result suggests
material inconsistency at the resistor level. Moreover, the same
SALZANO et al.: ENVIRONMENTAL QUALIFICATION TESTING AND FAILURE ANALYSIS OF EMBEDDED RESISTORS
517
TABLE VIII
FAILURE SUMMARY
coloration patterning was present throughout the copper traces
on the rest of the TV-1R board suggesting the metallic growth
was possibly copper.
In order to confirm the metallic growth hypothesis, the
embedded resistor was cross sectioned and observed under
an optical microscope. Metallic growth appeared to occur
between traces on the upper surface of the embedded resistor.
Fig. 15(a) provides an optical image of a “good” resistor while
images b, c, and d illustrate the “bad” resistor investigated
with the optical microscope. Likewise, Fig. 16 provides further
investigation with ESEM imaging. Looking more closely at
the left and right traces, it is difficult to conclude that the
metallic growth started from both traces equally since more of
the copper builds at the right trace and thins as it approaches
the left. In addition, since the cross-sectioned resistor was the
product of the temperature/humidity qualification test and no
bias was used during testing, it cannot be concluded that a
current flow starting from the right trace and ending at the
left caused the thinning effect on the growth. One possibility
is that during the measurement phase of the test plan, the
copper traces were still malleable and the resulting current flow
produced by the data acquisition equipment caused the transfer
of copper as the resistance dropped and the current increased.
In the cross-sectioned case, the value of the embedded resistor
decreased from 52 to 0.01
representing a short circuit.
The example resistors shown in Figs. 12–16 were from temperature/humidity testing; however, the remainder of the Gould
and MacDermid sample populations (from the other environmental tests) exhibited the same types of failure mechanisms.
While MacDermid embedded resistors exhibited delamination
above the devices between the nickel phosphorous resistor
material and FR-406 inner layer that increases resistive value,
Fig. 12. Optical and SAM imaging of a failed MacDermid resistor. The failed
resistor is located in the box (good resistors appear to the left and right of the
box).
failed Gould embedded resistors have decreasing resistive
value as the result of metallic growth on the upper surface
of the nickel–chromium resistor material between copper
traces. Accordingly, (see Section III), the majority of the
Gould histograms represent negatively weighted distributions
signifying decreases in resistance. Conversely, the majority of
MacDermid histograms represent positively weighted distributions signifying increases in resistance. A correlation between
the distribution of resistance values illustrated within the data
analysis histograms and the observed failure mechanisms on
the embedded resistors can be concluded.
518
Fig. 13.
IEEE TRANSACTIONS ON ADVANCED PACKAGING, VOL. 28, NO. 3, AUGUST 2005
ESEM images of a “good” and “bad” MacDermid embedded resistor delaminating from an FR-406 inner layer.
Fig. 14. Optical and SAM imaging of a failed Gould resistor. The failed
resistor is located in the box (good resistors appear to the left of the box).
V. DISCUSSION AND CONCLUSION
The experimental results obtained through selected environmental qualification experiments were used to determine the
operating characteristics and failure mechanisms associated
with Gould’s subtractive nickel–chromium and MacDermid’s
additive nickel–phosphorous embedded resistor technologies
used in printed wiring boards. Gould and MacDermid embedded
resistor technologies performance characteristics were found
to be comparable to the semi-precision classification for traditional resistor technologies. In addition, the embedded resistors
were inspected for and found to be statistically insensitive to
configuration in horizontal/vertical orientations, material-added
reworked/nonreworked conditions, and isolated/daisy-chained
connections. While the majority of Gould’s resistors responded
to environmental stressing with decreasing resistive values, the
majority of MacDermid’s resistors responded with increasing
resistive values. Gould failures appeared to be the result of
metallic growth on the upper surface of the embedded resistor,
meanwhile MacDermid failures were the result of delamination
between the upper surface of the embedded resistor and the
FR-406 inner layer. Both failure mechanisms could possibly be
diminished using tighter tolerance manufacturing constraints
to eliminate unnecessary defects that historically lead to delamination and metallic growth. Cleaner, smoother substrate
surfaces, more adhesive bonding methods, additional annealing
processes, more pure, untarnished materials, etc., could also
help to alleviate these unwanted defects. Even though Gould
and MacDermid’s technologies have less tolerance adherence
than high-precision surface-mount discrete resistor technologies (approximately two orders of magnitude less), embedded
resistors will undoubtedly establish themselves in the high
speed electronic systems market since they have proven to
eliminate almost all of the parasitic inductance, normally
associated with discrete resistors and their vias.
SALZANO et al.: ENVIRONMENTAL QUALIFICATION TESTING AND FAILURE ANALYSIS OF EMBEDDED RESISTORS
Fig. 15.
Optical images of a “good” and “bad” Gould embedded resistor.
Fig. 16.
ESEM images of a “good” and “bad” Gould embedded resistor illustrating metallic growth.
519
ACKNOWLEDGMENT
The authors would like to thank the NIST Advanced Embedded Passives Technology (AEPT) Consortium – NCMS,
ITRI, 3M, Compaq Computer, Delphi Delco Electronics,
DuPont Photopolymer and Electronic Materials, DuPont High
Performance Films, ESI, Foresight Systems, MacDermid,
Merix Corporation, MicroFab, Nortel Networks, ORMET
520
IEEE TRANSACTIONS ON ADVANCED PACKAGING, VOL. 28, NO. 3, AUGUST 2005
Corporation, and Sanmina. They would also like to thank R.
Hilburn at Gould Electronics and B. Greenlee at Merix.
REFERENCES
[1] T. Finnerty, “2001 Company Presentation,” Prismark Partners, LLC,
Cold Spring Harbor, NY.
[2] K. Fjeldsted and S. L. Chase, “Embedded passives: Laser trimmed resistors,” CircuiTree, pp. 70–76, Mar. 2002.
[3] “Passive Components Technology Roadmap,” National Electronics
Manufacturing Technology Roadmaps, NEMI, Inc., Herndon, NY,
2000.
[4] R. W. Berry, P. M. Hall, and T. H. Murray, Thin Film Technology. Princeton, NJ: Van Nostrand, 1968.
[5] V. G. Shah and D. J. Hayes, “Trimming and printing of embedded resistors using demand-mode ink-jet technology and conductor polymer,”
in Proc. Technical Conf. IPC Printed Circuits Expo., Mar. 2002, pp.
S14-4-1–S14-4-5.
[6] R. Ulrich, W. Brown, S. Ang, F. Barlow, A. Elshabini, T. Lenihan, H.
Naseem, D. Nelms, J. Parkerson, L. Schaper, and G. Morcan, “Getting
aggressive with passive devices,” IEEE Circuits Devices Mag., vol. 16,
no. 5, pp. 16–25, Sep. 2000.
[7] “Final Report, National Center for Manufacturing Sciences Advanced
Embedded Passives Technology Consortium,” Ormet Corporation,
Wheeling, WV, 2001.
[8] J. Zhou, J. D. Myers, and G. R. Dickinson, “Thermal cycling and ESD
evaluation of embedded resistors and capacitors in PWB,” in Proc. IPC
Annu.Conf., Aug. 2001, pp. 502–505.
[9] J. Wang and S. Clouser, “Thin film embedded resistors,” in Proc. IPC
Expo., Apr. 2001, p. S08-1.
[10] J. Wang, M. Davis, R. Hilburn, and S. Clouser, “Power dissipation of
embedded resistors,” in IPC Printed Circuits Expo., Mar. 2003, pp.
515–516.
[11] J. Felten, R. Snogren, and J. Zhou, “Embedded ceramic resistors and
capacitors in PWB: Process and performance,” in Proc. IPC Annual
Meeting, Oct. 2001, pp. 502–506.
[12] K. Fairchild, G. Morcan, T. Lenihan, W. Brown, L. Schaper, S. Ang,
W. Sommers, J. Parkenson, and M. Glover, “Reliability of flexible
thin-film embedded resistors and electrical characterization of thin-film
embedded capacitors and inductors,” in Proc. 47th Electronics Components and Technology Conf. (ECTC), 1997, pp. 730–738.
[13] Test Vehicle TV-1R, Design Files, The National Center for Manufacturing Sciences, Ann Arbor, MI.
[14] J. D’Ambrisi, D. Fritz, and D. Sawoska, “Plated embedded resistors for
high speed circuit application,” in Proc. IPC Annu. Meeting, Oct. 2001,
pp. 502–510.
[15] Resistance-Temperature Characteristic, Method 304, MIL-STD-202G,
Oct. 24, 1956.
[16] R. Ulrich and L. Schaper, Integrated Passive Component Technology. Piscataway, NJ: IEEE Press, 2003.
[17] F. Zandman, P. Simon, and J. Szwarc, Resistor Theory and Technology. Park Ridge, NJ: SciTech, 2001.
[18] Ohmega-Ply Resistor Tolerance (2004, Jun. 28). [Online]. Available:
http://www.ohmega.com/newtosite2.html
[19] Specifications
and
Properties,
Ohmega-Ply
Resistor-Conductor
Material
(2004,
Jun.
12).
[Online].
Available:
http://www.ohmega.com/OhmPly.html
Lawrence John Salzano, II received the B.S. and
M.S. degrees in mechanical engineering from the
University of Maryland, College Park, in 2002 and
2003, respectively. He is currently pursuing the
law degree at The George Washington University
Law School, Washington, DC, where his interests
concentrate on litigation in intellectual property and
technology law.
At the University of Maryland, he worked on embedded resistors reliability analysis and co-developed
a GPS-based intelligent tracking system harnessing
both location and data management technologies for use in the criminal justice
system as a part of TRX Systems.
Chris Wilkinson received the M.I.E.E. (electronics
engineering, B.Sc.) from the University of London,
London, U.K., in 1971.
He is a Member of the Research Faculty of mechanical engineering at the University of Maryland,
College Park. His research experience and interests
are in avionics design, maintenance-free operating
period, electronics prognostics, parts obsolescence
management, and uprating. He has contributed
papers in the areas of reliability assessment, parts
obsolescence and management, and uprating. He
was previously with Smiths Aerospace, Cheltenham, U.K., most recently as a
Principal Research Engineer in the Corporate Research Department.
Peter A. Sandborn (M’87–SM’01) received the B.S.
degree in engineering physics from the University of
Colorado, Boulder, in 1982, and the M.S. degree in
electrical science and Ph.D. degree in electrical engineering, both from the University of Michigan, Ann
Arbor, in 1983 and 1987, respectively.
He is an Associate Professor in the CALCE
Electronic Products and Systems Center (EPSC),
University of Maryland, College Park, where his
interests include technology tradeoff analysis for
electronic packaging, embedded passive component
analysis, system life cycle economics, and virtual qualification of electronic
components and systems. Prior to joining the University of Maryland, he was
a Founder and Chief Technical Officer of Savantage, Austin, TX, and a Senior
Member of Technical Staff at the Microelectronics and Computer Technology
Corporation, Austin. He is the author of over 100 technical publications and
books on multichip module design and part obsolescence forecasting.
Dr. Sandborn is an Associate Editor of the IEEE TRANSACTIONS ON
ELECTRONICS PACKAGING MANUFACTURING and a Member of the Editorial
Board for the International Journal of Performability Engineering.