Making the right choice

Buying equipment with price tags of thousands or millions of euros requires careful consideration. System purchasers will be confronted by an array of marketing brochures, sales reps, spec sheets and offers to test equipment.

Numbers can be comforting, but there is a danger. Although different manufacturers use many of the same words, they often dont mean exactly the same thing. This is particularly the case where there are no agreed industry standards. Manufacturers may then resort to standards designed for another industry or even invent their own. Further, the specs are usually produced under the best possible conditions, and these often differ significantly from those under which the equipment is used.

X-ray inspection is important in electronics since it can see inside defective devices without the laborious reopening or dismantling of components required by optical techniques. Reopening can remove mode of failure evidence. The uses of highresolution X-ray inspection include checking of wire bonds, wafer bumps, wire sweep, chip-scale packaging (CSP), ball grid arrays (BGAs), multi-chip modules and micro leadframes (MLFs). Among the newer applications are searching for voids in solder balls and, with the push for lead-free processing, whiskering and ‘champagne voiding in tin-based solders.

What are the specifications of choice for the x-ray inspection industry in terms of systems aimed at the electronics market? While physical specifications such as the units dimensions, ergonomics and weight are often key considerations for cramped IC package and assembly facilities and for shipping to the site of use, we shall not concern ourselves with these. Rather we shall look at the factors that define X-ray inspection for use in electronics manufacturing. Here the key components are the magnification, X-ray tube and detection systems, each with their own characteristics. Although magnification is one of the main aims of the process, a systems effectiveness in this regard depends on the source and detection subsystems. This article will concentrate on the influence of the sources characteristics on "geometric magnification" (see box).

Tubes are specified in terms of accelerating voltage range, target power and type, focal spot size, beam cone angle, filament lifetime and even cooling system. Since X-rays are also a health hazard, safety compliance is an important consideration. Characteristics for the vacuum system used by the tube may also be given.

While most of these measures are uncontroversial, except perhaps in terms of the cut off points for quantities that grade off like beam cone angle, there is one specification that is used to characterise the whole image chain – feature recognition (Feinfocus, phoenix|x-ray Systems, XTek), minimum feature recognition (Dage), detail recognition (Viscom) or detail detectability (phoenix), terms apparently synonymous – that needs more careful consideration. Connected in most cases with these terms is the measurement of the focal spot size in the X-ray tube (see Open X-ray tubes).

Marketing term
Stefan Becker, sales and marketing director at phoenix|x-ray, believes that feature recognition is more a marketing term than a standard, since there is no clear definition. For an accurate comparison between systems, one should use focal spot size, he says. A minimum feature recognition value can then be estimated at roughly half the focal spot size, "but may be even smaller for certain objects", according to a glossary on the companys web site.

However, even the focal spot size has no global definition. To determine its focal spot sizes, phoenix uses a test sample and methodology from the Japan Inspection Instruments Manufacturers Association (JIMA RT RC-01). The sample consists of an easy to handle 2x2cm frame containing a periodic structure of bars separated by 1 micron. The sample image is recorded on high definition film for contrast resolution. The developed film is then examined with a densitometer. The transmission function is then used to determine the focal spot size from a linear interpolation between the 86% and 14% relative intensity points.

Phoenix claims the registered trademark "nanofocus" for its X-ray tubes aimed at the electronics inspection and other high definition markets, although other companies also use the term to describe some of their tubes. Further, a German business magazine ("Focus") and a manufacturer of digital projectors (InFocus) have separately lodged opposition to the trademark with the European Union Office for Harmonisation in the Internal Market (OHIM).

Phoenix tubes give focal spot sizes under the JIMA measurement down to 0.8 microns, although for real-time imaging a 1.2 micron focal spot is better. Therefore, the "nano" label is a far stretch from the US National Nanotechnology Initiatives "research and technology development at the atomic, molecular or macromolecular levels, in the length scale of approximately 1-100nm range" (http://www.nano.gov/html/facts/whatIsNano.html). Europes NANOCMOS project is working "from the 45nm node down to the limits".

Among the applications for nanofocus x-ray inspection, Becker highlights the search for voids in solder joints that have sizes at the 1-2 micron level. This requires nanofocus technology unlike the more traditional X-ray inspection of 25 micron diameter bond wires that can be adequately imaged with microfocus tubes. The company quotes minimum feature recognition values down to 0.2 microns.

Feinfocus, recently taken over by Swiss holding company Comet, also uses the relation that minimum feature recognition is about half the focal spot size. Jon Dupree, Feinfocus business manager for Comet North America, reports that the company uses a special mask with 100nm resolution to determine the focal spot size. The film image is examined by eye. Feinfocus claims a spot-size of less than 1 micron, and minimum feature recognitions down to 0.3 microns are quoted on its tubes.

Dupree also emphasises the need to consider the entire image chain since efforts to reduce spot sizes are approaching the limit for tungsten targets. The problem is that the heat transfer from the electron beam increases as the spot size shrinks. Unfortunately, tungsten is the most common target because it has a high melting point so the ability to find alternative materials that could allow even smaller spots is strictly limited. Heating therefore limits the number of electrons hitting the target, reducing the x-ray dose.

Another factor is the thickness of the target – a thin target is used for the smallest spots. However, a thin target produces less X-rays, making detection and image formation slower, and needs more frequent replacement through damage by the electron beam.

Apart from giving clearer images, a bit like highdefinition TV, Dupree believes that high-resolution systems will enable the imaging of subtle changes in packaged ICs such as whisker and dendritic growths in lead-free solders. However, one has to balance between the ease of defect identification with the slow speed of image capture at the highest resolutions.

Viscoms "detail recognition" is comparable to "feature resolution", confirms Thomas Gruhl, manager for X-ray system research and development. While "detail recognition" is not as precise a term as "resolution", it is what the customer needs. Further, the human eye has the ability to "easily" perceive a 10% difference in light intensity, while Rayleighs criterion for resolution results from a 19% intensity change in spectroscopic (slit aperture) and 26% in image forming systems (circular aperture). Hence, one can expect detail recognition to be better than theoretical resolutions. The resolution is around half the focal spot size, according to Gruhl.

The company uses two standards for measuring focal spot size, the microfocus section of European EN 12543-5:1999 standard and the JIMA technique used by phoenix. The European standard involves imaging a 1mm wire and looking for the 90% contrast points on a CMOS panel digital detector or an amorphous silicon device. Gruhl reports that the two standards give comparable results, although the JIMA technique yields the parameter faster. The company has tubes ranging from its standard "less than 5 microns" down to a "so-called" nanofocus device with a spot size less than 1 micron.

Viscom seeks to provide highly stable x-ray emission tubes, because of the companys big interest in automatic inspection systems. Such equipment needs a high level of repeatability between runs and between different systems.

Always take a test drive
While minimum feature recognition is not an exact term, Dr David Bernard, Dages X-ray system product manager, believes that it can be a useful measure. "We choose to use the term ‘feature recognition on our literature, as we believe it provides a meaningful indication to the market of the capabilities of our range of x-ray inspection systems. It means that we are able to image samples of appropriate dimensions (traceable to standards), as a way of proving what our systems can offer."

This company gives a minimum feature recognition value of 600nm on its latest systems. Dage entered the X-ray market six years ago, buying a medical X-ray tube producer. It also produces tubes for other equipment manufacturers and for its own systems aimed at the electronics market. However, the company has more than 35 years experience in bond testing. Bernard reports that on the basis of this, Dage has found that practical, application-relevant tests of product quality are much better than more esoteric "defined" procedures. He sees X-ray spot size as one such measure that is often meaningless to the customer and cannot often be reproduced with ease by the user, if at all, to prove the claims made.

Bernard thought it unnecessary to discuss the companys methodology to measure its X-ray spot size, saying that it is "company confidential". He adds: "We do make spot size measurements as part of our quality control. However, we do not provide this information externally. This is because there are no agreed standards for spot size measurements and therefore comparing one companys spot size measurement to another is fraught with the potential for inconsistency and confusion for the customer."

A very low kV and a narrow range for determining the spots edge (eg 30%-70% rather than 20%-80%) are among the definitional tricks that an unscrupulous manufacturer could use to squeeze spot size without improving in any way the technology. Such low kVs cant be used for inspecting real samples – there is insufficient penetrating power for the X-rays to pass through the sample and register on the detector.

The overall imaging chain is equally as important as the tube resolution (feature recognition), according to Bernard. He advises customers to "test drive" any system with their own samples to see if the equipment meets their particular needs.

Looking for a new standard
UK firm X-Tek, celebrating its 20th anniversary next year, has developed a test piece that consists of a 1 micron-thick gold pattern on a silicon nitride substrate. The company believes that since most open tube systems employ roughly the same source to detector distances and the same types of imaging system, the relationship between focal spot and feature recognition should apply to all such systems equally. Thus the focal spot size can be estimated from the feature recognition in the same way that feature recognition can be estimated from the focal spot size. This leaves the measurement of the focal spot or feature recognition using a shared standard as the key to providing a meaningful comparison between systems.

Dr Geraint Dermody, senior development manager at X-Tek, reports: "Having consulted with electron optics designers and other system manufacturers, X-Tek decided to standardise on a simple standard which can be used for both optical and X-ray inspection resolution tests. The test piece consists of a series of tapering lines in a starburst pattern formed from 1 micron-thick gold on a silicon substrate. The test will qualify sources with focal spot sizes from 5 down to 0.5 microns.

"As manufacturers we try to specify our products accurately and in the absence of an industry standard for microfocus X-ray source focal spot measurement, we had extrapolated an old British Standard using a tungsten ball bearing imaged onto high definition film designed for conventional sources (the 20%-80% points with linear interpolation yielding the spot-size). This gave a very pessimistic figure for the focal spot compared with other tests. Users were being confused when comparing a 2 micron system with a nanofocus system and getting the same image quality.

"The beauty of our new test piece is that it can give a simple, quick estimate of the systems resolution and contrast sensitivity, visually for users and mathematically for manufacturers. We were doing ourselves an injustice using the old standard and by adopting the new starburst pattern we have reclassified our old 2 micron tubes as 1 micron in line with other suppliers. We believe it is the only useable test piece that is easily available to all."

X-Tek has commissioned the manufacture of a large number of the standard test pieces to be available to all users and other x-ray system manufacturers.

Customer choice
All the above arguments have merit, but the customer must be aware that these viewpoints are made from the position of a vendors strengths and market focus. As x-ray technology develops into new applications, there will be a need to reassess, evolve and create meaningful measures of system ability and performance for electronics customers.

Broken 25 micron gold bond wire inspected with a nanofocus tube on an phoenix system. The gap is only a few microns wide. Such defects are much harder if not impossible to detect with standard microfocus equipment, according to the company’s sales and marketing director Stefan Becker.

X-Tek has produced a test piece to measure the resolution of x-ray inspection systems as a whole. Shown (top to bottom) are images from an electron microscope control and from x-ray inspection systems using a “2 micron” transmission target and a “5-micron” reflection target. A rule of thumb for determining resolution is that if 25% of the length of the wedges in the circular pattern is resolved, the spot size is 4 microns, if 50%, the spot is 3 microns, and if 75%, 2 microns. If the wedges are completely visible, the system resolution is equal to or better than 1 micron.