Failure in Multilayer Ceramic Capacitors (MLCCs) for AVX Ltd
Submitting InstitutionQueen's University Belfast
Unit of AssessmentPhysics
Summary Impact TypeTechnological
Research Subject Area(s)
Physical Sciences: Other Physical Sciences
Chemical Sciences: Inorganic Chemistry
Engineering: Materials Engineering
Summary of the impact
The provision of professional analysis and advice has created an economic
impact of $0.6M with AVX Ltd, a leading supplier of electronic components
within the Kyocera Group. This information was pivotal to AVX Ltd
retaining a major contract, for multi-layer ceramic capacitor (MLCC's)
supply through to the automotive manufacturer Volkswagen Group. Our
intervention addressed a reliability issue in the MLCCs and allowed them
to improve processes and revise manufacture protocols. The impact drew on
previous collaborative research with AVX Ltd and innovative methodologies
for preparation of micro and nanoscale samples of materials in capacitors
in academic research.
In the mid-1990's, academic staff, Bowman and Gregg, initiated a research
activity to map and understand the properties of micro- and nanoscale
ferroelectrics. Ferroelectrics were receiving significant commercial
interest in numerous capacitor applications/markets. A key element in the
early research was the fabrication and functional characterisation of very
simple thin film planar capacitor structures based on titanates. These
were investigated in an attempt to elucidate fundamental reasons for the
dramatic collapse in permittivity, that reduces device capacitance, with
decreasing device thickness and which was not understood at the time [Reference
1]. This was of great importance as titanates are of principal use
for both commercial ceramic capacitors (circa 1012 components
manufactured p.a.) such as the X7R devices and for novel high-K
perovskite-based Dynamic Random Access Memory (DRAM) systems
conceptualised at the time.
While capacitor structures made in-house did allow useful insights [Reference
2], it soon became clear that ferroelectric (dielectric) properties
were so sensitive to defects, introduced during thin film growth, that a
radically new experimental approach was needed. Thus, from 2002, a wholly
novel programme in which thin film capacitors were made from small slices
of ferroelectric cut from high purity single crystals using Focused Ion
Beam (FIB) milling [Reference 3] was initiated. Functional
characterisation of these relatively unclamped defect-free systems
demonstrated that the permittivity did not necessarily collapse with
dielectric thickness decrease at all and that the effect could be
engineered out in real devices. Such insights were published in a number
of articles between 2004 and 2009 [Reference 4, 5].
At the same time as this FIB micro-machining work progressed that
included further insights into damage mitigation [Reference 6] we
commenced a Knowledge Transfer Partnership (KTP) with AVX Ltd. The
three-year project from 2005-2007 worth £123K was titled "Critical
interfaces in currently produced capacitors". The project engaged
M.M. Saad, primary author of Reference 3, as the Knowledge Transfer
Associate. The project combined precision micro-machining by FIB to
isolate known and potential defects and regions of MLCCs for detailed
chemical and structural analysis that would then be followed up via TEM
(Transmission Electron Microscopy) and STEM (Scanning Transmission
Electron Microscopy). In the programme the parts investigated covered X7R
specifications for in consumer electronics applications through to those
for high voltage automotive usage. This provided Bowman and Gregg with
significant bilateral experience in handling of parts and establishing
common terminological ground for the use of the research methodologies and
results to be applied to evaluating the defects that arise in such
We note that associated research in thin film capacitors remains an
active area of endeavour, most recently via Technology Strategy Board
(TSB) support of a project "Advanced capacitors for energy storage"
with Syfer Technology worth £164K in 2010-12 and has received significant
recent press attention http://goo.gl/p1kp12.
Such collaborations have tapped various aspects of the ferroelectrics
expertise developed in Queen's, but microscopy and FIB-processing
knowledge has been crucial in all collaborative research to date.
References to the research
For each reference we provide data on the journal ranking within ISI
classification area/s by ISI Journal Impact Factor (IF), the actual
Journal Impact Factor and the number of cites the article received on the
ISI Web of Knowledge to September 2013.
References 1-3 illustrate both quality of the underpinning research and
evidence to the impact, while references 4-6 provide the latter.
 "Investigation of dead layer thickness in SrRuO3/Ba0.5Sr0.5TiO3/Au
thin film capacitors", L. J. Sinnamon et al., Applied Physics
Letters, 78, 1724 (2001). http://dx.doi.org/10.1063/1.1356731
(17th of 125 in Physics — Applied, IF 3.8, 123 cites).
 "Exploring grain size as a cause for "dead-layer" effects in thin
film capacitors", L.J. Sinnamon, et al, Applied Physics Letters, 81,
703 (2002). http://dx.doi.org/10.1063/1.1494837
(17th of 125 in Physics — Applied, IF 3.8, 77 cites).
 "Intrinsic dielectric response of ferroelectric nanocapacitors",
M. M. Saad, et al, Journal of Physics: Condensed Matter 16, L451
(18th of 69 in Physics — Condensed Matter, IF 2.55, 96 cites).
 "Size Effects on Thin Film Ferroelectrics: Experiments on
Isolated Single Crystal Sheets", L. W. Chang, et al, Applied Physics
Letters, 93, 132904 (2008). http://dx.doi.org/10.1063/1.2990760
(17th of 125 in Physics — Applied, IF 3.8, 31 cites)
 "Settling the Dead Layer Debate in Nanoscale Capacitors", L.
W. Chang, et al, Advanced Materials 21, 4911 (2009). http://dx.doi.org/10.1002/adma.200901756
(4th of 125 in Physics — Applied, IF 13.9, 25 cites).
 "Strategies for gallium removal after focused ion beam patterning
of ferroelectric oxide nanostructures", A. Schilling, et al,
Nanotechnology 18, 035301 (2007). http://dx.doi.org/10.1088/0957-4484/18/3/035301
(16th of 125 in Physics — Applied, IF 4.0, 16 cites)
Details of the impact
Through the longstanding interaction with AVX Ltd the company was aware
of our unique academic research involving FIB micromachining allied to
microscopy of ferroelectrics to reveal intrinsic behaviour and identify
consequences of defects to capacitor behaviour.
In 2011, AVX Ltd encountered a reliability problem concerning specific
MLCC product ranges. Through their Tier 2 customers Lear Corporation and
Johnson Controls Inc. who on behalf of their Tier 1 customer Volkswagen
Group (for VW and Audi) reported that the Volkswagen Group had revised the
environmental specification for MLCC (case sizes 0603 and 0805) for
automotive applications [Source 1].
Under revised environmental stress conditions at the customer site and
subsequently replicated in the AVX Ltd. factory, of increased humidity and
temperature, the supplied parts where exhibiting unacceptable failure
rates. "The automotive industry is one of the most demanding markets
for MLCC with failure rates often pushed toward no more than 1:109
parts" [Source 1]. This arises because both the owner and
the public on the road perceives particular brand of vehicle to have
broken down and to not associate the breakdown with failure of a supply
AVX Ltd made available a selection of operational and failed parts for
our investigation along with a technical brief on their electrical testing
to date on the basis that they "... were aware of the research in
ceramic capacitors and in particular the innovative focused ion beam
methods applied to ferroelectrics and capacitors that had been developed
at Queen's" [Source 1].
With our previous research in more sensitive nanoscale capacitors allied
to our expertise acquired via the KTP project "handling `green' and
finished parts" we were familiar at handling their commercial
product and that was seen as `beneficial" [Source 1].
A MLCC is comprised of a stack of several hundred interleaved
ferroelectric ceramic and nickel electrode layers. Electric contact to the
device is then made via end terminations of two layers of nickel and tin.
The tin layer provides the region for solder contact to be made. In the
particular product that exhibited failure the electrode is connected to
the MLCC via a stress relieving silver-polymer composite layer.
We were able to adapt and formulate a modification to our usual FIB based
micromachining protocols used in the creation of nanoscale capacitors to
develop a workflow that would allow us to extract representative sample
pieces for evaluation and inspection. This adapted workflow meant that we
were able to perform successional slicing of the MLCC's in a series of
Electron microscopy imaging and associated chemical analysis was
performed on slices extracted and in regions from where they had been
removed. It became evident that the failure under humidity test was not
arising from damage to or degradation of the more sensitive ferroelectric
ceramic material, but instead it was originating at the interface of the
electrode layers and the ferroelectric ceramic.
A major challenge, over the last fifteen or so years, in all research and
engineering of ferroelectric capacitors has been the move away from noble
metal electrodes and contacts to base metals such as nickel. Nickel will
unfavourably oxidise at the interface with the oxide-based titanate
ferroelectric. In this case of these failed parts further detailed
examination on FIB machined sections led to an unexpected cause of the
actual device failure. The final overcoat of tin was introducing
additional stress due to a high deposition rate that was exacerbating the
delamination of the nickel from the ferroelectric. That delamination then
provided a path for the metal in the polymer composite to escape under
application of voltage and caused part failure.
A report [Source 2] of the investigation and outcomes was prepared
and a de-briefing meeting at AVX Ltd arranged and the findings presented.
In addition to the technical staff of AVX Ltd at Coleraine also present
were engineers from their subsidiary plant in the Czech Republic that
undertakes the electro-plating of the metal electrode layers.
In addressing the reach and significance of this impact we note that
while AVX Ltd is based in Coleraine, Northern Ireland (NI), they are part
of AVX Corporation who are in turn part of Kyocera Electronic Devices in
the Kyocera Group.
As of 2012 AVX Ltd. employed 317 people in NI and had a £108M turnover;
no profit figures are available [Source 3]. The factory operates a
subsidiary activity in the Czech Republic. Product development and
principal manufacture in undertaken in Coleraine and then the parts are
shipped to the Czech Republic for the plating and sample testing before
supply to customers.
On the basis of our technical report and de-briefing meeting, informed by
data acquired using our FIB micromachining protocols AVX Ltd were able to
undertake timely revisions to their manufacturing program. This eradicated
the part failures to an accepted tolerance level and thus allowed them to
deliver on the contract to their Tier 2 customers Lear Corporation and
Johnson Controls Inc. who then in turn supply to their Tier 1 customers VW
and Audi [Source 1].
AVX Ltd. have confirmed that the value of the international contract was
"valued at $0.6M". [Source 1].
Sources to corroborate the impact
- Letter from Business Development Manager, AVX Ltd. Coleraine
- Technical Report to AVX Ltd., 19 May 2011
- Northern Ireland's Top 100 Companies — The Belfast Telegraph, 30 April