Content uploaded by Kevin Distelhurst
Author content
All content in this area was uploaded by Kevin Distelhurst on Mar 15, 2025
Content may be subject to copyright.
Isolating Electrically Open Laminate Defects Using Space Domain Reflectometry
(SDR) with a Superconducting Quantum Interference Device (SQUID)
Kevin A. Distelhurst
GLOBALFOUNDRIES, Essex Junction VT, USA
kevin.distelhurst@globalfoundries.com
Abstract
An electrically open defect on a laminate may not always be
found timely or successfully due to the lack of fault isolation
techniques for this type of defect. This is partly due to needing
high frequency techniques to isolate the location of the open.
Magnetic field imaging (MFI) using a Superconducting
Quantum Interference Device (SQUID) is a technique that
maps, in this case, an RF signal through a trace, up until the
open defect boundary. Several obstacles are introduced when
using an RF signal, one of which is the shielding of the signal
from the external world. Despite this obstacle, analysis of an
open in an arbitrary location along a laminate under a copper
plane is proven successful using this technique.
Introduction
Several examples of electrically open defects that can be
found within a laminate include a break in a signal trace, via
or the laminate-die interface (e.g. wirebond). There is no
direct way to view these defects without destructive analysis
due to the opaque material that the traces are typically
embedded within. XRAY computed tomography (XRT) is one
method of indirectly observing the defect by taking series of
2D XRAY images and reconstructing them into a 3D render.
XRAY 2D imaging may be sufficient in detecting catastrophic
opens and even some less obvious breaks in a signal trace or
via. However, without sufficient isolation, the XRAY imaging
technique may not provide timely results. Several iterations of
XRT may be needed throughout the laminate to analyze all
stretches of the electrically open trace. In some cases, the
defect may be too subtle to detect. Techniques have been
developed, such as Time Domain Reflectometry (TDR), to
help isolate the defect [1] but XRAY imaging may still be
needed to confirm or further isolate the defect before
destructive, more direct analysis is performed.
MFI has been successful in isolating shorts in both laminate
[2] and die [3]. Typically, an AC signal is applied to the
shorting defect and the resulting magnetic field generated
from the current path is observed using a SQUID or Giant
Magnetoresistance (GMR). The technique has been extended
to using an RF signal. The signal is applied to the defective
trace and the resulting standing wave magnetic field is mapped
up until the open boundary which reflects the incoming RF
signal. This technique is called space domain reflectometry
(SDR). The resulting map of the magnetic field is used to
generate a current density map (CDM). Further descriptions
of the setup are found in [1] & [4]. Destructive, direct analysis
of the defect can be performed after MFI provides the
approximate location of the fault, circumventing the need for
lengthy XRT and electrical isolation.
SDR has been successful [1], [4] in finding opens in the
laminate. This work extends the concept by including an open
along a signal trace within the laminate, buried under a single
copper plane. The buried defect is of interest because it is one
example of the obstacles encountered when dealing with RF
signals (e.g. shielding, coupling, ect.). Analysis of a buried
defect compared to an open without a copper power plane
above reveals the magnetic field strength at the SQUID sensor
is impacted by the planes.
Setup of Laminate
A non-defective laminate is used as the device under test. The
laminate consists of a ball grid array (BGA) followed by
several layers of copper traces and power planes up until the
laminate-die interface to the silicon flip chip. An arbitrary net
is chosen in two corners of the laminate opposite one another.
Both nets start at a ball near the edge of the laminate on the
bottom (opposite the flipped die) and eventually transverse
most of the distance near the top (closest to die) of the
laminate. The top most portion of the laminate above the top
level trace has a power plane covering the nets. See
for a representation of the nets.
Figure 1: Cross-section representation of the laminate used.
Note that the trace through the laminate is just below the
copper power plane.
On one side of the sample denoted side 1, the copper power
plane is removed using surgical blades. Optical inspection
revealed little damage to the insulation above the traces and no
disturbance to the underlying traces. The other side of the
sample denoted side 2 only has the copper power plane
exposed. This is done so design features in the copper plane
are visible for alignment of CAD images if necessary.
Mechanical abrasion is used to expose a portion of the copper
power plane above the area the net will be cut. Both nets were
electrically measured for further verification that no defects
were introduced.
Next, a cut is made in each net using a laser cutting system.
An arbitrary location was chosen halfway along the side 1 net.
Exposed vias that connected to the removed copper power
plane were used to navigate and ensure the cut was made in
the appropriate location. Likewise, clearance features in the
copper plane were used to navigate to a portion of the net and
ensure the cut was made in the appropriate location on side 2.
The resulting cuts for side 1 and side 2 appear in Figure 2(a)
and Figure 2(b), respectively.
Figure 2: Optical images of the cuts on side 1 and side 2. (a)
The cut made in side 1. The vias that connected to the removed
power plane are visible. The underlying net that was cut is
partially visible through the insulator. (b) The cut made in
side 2. Features in the copper power plane are visible. Black
wax was added to both cuts after they were confirmed
electrically open.
The formation of the holes were monitored optically with a
camera system accompanying the laser head. The underlying
traces appeared as the cuts were being made and allowed the
status of the cut to be monitored. An electrical open was
confirmed on both sides after the lines appeared cut. Black
wax was subsequently filled into the holes to insulate the
exposed traces from conductive debris.
The part containing both sides was placed into the MFI tool. A
single probe contacted the BGA connecting to the side 1 net.
Several SDR scans using the SQUID were performed. The
initial, low res scan (100µm) of the area around entire length
of the net observed the magnetic field up-until the cut after
which the signal began to drop off. The scan size was reduced
and the resolution set to 25µm. Figure 3(a) provides the CDM
of this scan and Figure 3(b) provides the infrared image (IR)
image of the same area. Note the beginning of the copper
power plane near the left edge of Figure 3(b). An initial glance
of the CDM may indicate that no open exists. However, the
RF signal formed by the standing wave along the
superimposed line in Figure 3(a) is plotted with respect to
distance. Likewise to [4], the plot reveals a reduction in RF
signal over distance until it reaches a minimum. The linear fit
of the decay of this line is then used to find at what distance
the current drop may indicate an open. A cross in Figure 3
indicates this location.
Figure 3: (a) CDM with the superimposed line indicating the
distance plotted with respect to RF current. The cross
indicates the location the current reaches the first minimum.
(b) The IR image of the scan area with the cross translated
from (a).
Comparison of the cross location with the IR image of the cut
reveals an offset. Further, higher resolution SDR scans of
about 5µm are performed in an effort to reduce the offset. The
same linear fit is performed on the CDM and the probable
defect location is marked on an IR image of the scan area. The
IR image with the cross is overlaid with a CAD image in
Figure 4. The location of the cut in the net of interest is
highlighted by an arrow. The overlay reveals that the higher
resolution scan provided a closer approximation of the open.
In many cases, this approximation and even a lower resolution
approximation is sufficient to begin destructive analysis or
closer indirect inspection with techniques such as XRT.
Figure 4: IR image of side 1 with a cross indicating calculated
location of the open defect. The image is overlaid with a CAD
image to visualize the accuracy and the location of the open
(a) (b)
(a)
(b)
along the net. The arrow indicates the location of the
manufactured open.
An SDR scan of side 2 is then performed. Likewise to side 1, a
lower resolution scan of the entire net is obtained followed by
higher resolution (2µm), small area scan to reduce the offset.
Even with the copper power plane above the net, a signal is
still observed. The signal strength is reduced but the data is
sufficient in providing the approximate location of the open.
An IR and CAD image overlay is provided in Figure 5.
Figure 5: IR image of side 2 with a cross indicating calculated
location of the open defect. The image is overlaid with a CAD
image to visualize the accuracy and the location of the open
along the net. The arrow indicates the location of the
manufactured open.
Conclusions
Low resolution SDR scans of an open manufactured along a
net using a laser cutting system provides a reasonable
approximation of the defect spatial location. Higher resolution
scans reduce the offset which could be useful in applications
were more accurate results are needed (e.g. finding subtle
defects). A net buried under a copper power plane which may
act as an RF shield is also analyzed. Although a reduction in
signal is observed, accurate results are still obtained and useful
for further analysis. Further experimentation on thicker
laminates with more copper planes and different materials is
still needed to understand the viability of this method in more
diverse applications. However, this technique is capable of
locating open defects even with certain obstacles.
References
[1] D. P. Vallet, D. A. Bader, V. V. Talanov, J. Guadestad, N. Gagliolo and A. Orozco, "Localization of Dead Open in a Solder
Bump by Space Domain Reflectomery," ISTFA 2012: Conference proceedings from the 38th International Symposium for
Testing and Failure Analysis, p. 4, 2012.
[2] J. Gaudestad, A. Orozco, M. Kimball, K. Gopinadhan and T. Venkatesan, "Short Localization in a Multi Chip BGA Package,"
IEEE 21st International Symposium on the Physical and Failure Analysis of Integrated Circuits, p. 74, 2014.
[3] J. Gaudestad, N. Gagliolo and V. V. Talanov, "High Resolution Magnetic Current Imaging for Die Level Short Localization,"
20th IEEE Internaional Symposium on the Physical and Failure Analysis of Intergrated Circuits (ITSFA), p. 355, 2013.
[4] J. Gaudestad, V. Talanov, N. Gagliolo and A. Orozco, "Space Domain Reflectometry for Open Failure Localization," 19th
International Symposium Physical & Failure Analysis of Integrated Circuits, p. 1, 2012.