Minimizing radiation to components: filters during x-ray can reduce radiation without compromising image quality.

X-ray inspection has become ever more important for checking product quality as it allows investigation within optically hidden areas, such as the quality of post-solder reflow of area array devices like ball grid arrays (BGAs), chip-scale packages (CSPs) and flip chips. During inspection, the sample is bathed in the ionizing radiation of high-energy photons and receives a dose of radiation. Certain devices are susceptible to damage by the ionizing radiation. Therefore, the user must consider the radiation dose during x-ray inspection to ensure that the device's critical threshold is not exceeded.

The total radiation dose to a device is the sum of each dose rate experienced by the device, multiplied by the time at that dose rate. A large radiation dose deposited into the silicon die could result in physical damage to the device and, therefore, its failure.

Large radiation doses cause almost certain failure, but, during shop floor x-ray inspection, the device is not subjected to doses large enough to cause failure. Instead, the concern lies in the subtler failure mechanisms, such as bit flips, loss of programmed data and erase margin, and leakage. Here, the failure of the device has some statistical, or random, chance of occurring, and physical damage is not visible. (1)

The difficulty with these random failure mechanisms for commercial-off-the shelf (COTS) products is that their occurrence could be caused by radiation doses that are orders of magnitude less than the gross-failure variety and potentially within the doses given during x-ray inspection. Therefore, if a device on the board is at risk, the dose achieved during the inspection procedure should be determined and checked against relevant critical threshold values. The radiation dose given to semiconductor devices is cumulative, (2) and repeating an inspection routine will double the dose to the device. So, repeat inspection, typical after rework, may have a detrimental effect on susceptible devices.

Before concluding that all COTS devices are at threat from radiation doses during x-ray inspection, consider the following:

* Some devices ate more susceptible to radiation dose than others, but the vast majority will not be affected by x-ray inspection. (3)

* The actual threshold of radiation dose levels for random events, as in x-ray inspection, is very difficult to define.

* The figures quoted in published literature vary by orders of magnitude and may be far in excess of what can be practically given during even the most extensive x-ray inspection routine.

* Exceeding the dose threshold of these random events only increases the probability of failure(s) occurring and not the certainty of failure.

X-ray inspection systems are basically x-ray shadow micrographs (Figure 1). An x-ray light source such as a tube produces x-rays, which pass through the device. Different materials within the device absorb more, or less, of the x-ray radiation depending on their density and atomic number and cast a shadow of the material on the detector. The denser the material is, the darker the shadow. The closer the sample is moved to the x-ray tube, the larger the shadow becomes due to geometric magnification and the larger the dose.

[FIGURE 1 OMITTED]

The factors affecting the dose are:

* distance of the device from the source of the x-rays

* x-ray tube power

* presence of filters between the tube and the sample

* repeat inspections.

Dose Rate and Distance

Radiation dose rate varies with the inverse square of distance. If you double the distance from where the initial dose rate is measured, then the dose rate will have decreased by a factor of four. Triple the distance, and it will have decreased by a factor of nine, and so on. The closeness of the device to the local point (source) of the x-rays has a dramatic effect on the dose rates and the doses that the device experiences. In addition, the total dose a device gets also includes the dose received from adjacent radiation fields when other devices are inspected. This dose could be significant, even though the device may be relatively distant from the source.

A simple prescription to minimize dose to samples is to limit the proximity of the sample from the source. This method, though, is countered by the continued reduction in device size and the need to move the device closer to the x-ray source to increase magnification. Therefore, a compromise between magnification and dose may have to be made during the inspection of specific devices.

Higher resolution and magnification open, of demountable transmission, x-ray tubes are standard within semiconductor and printed circuit board (PCB) x-ray systems. (4) To achieve the high magnification required by these small devices, a sample is placed closer to the tube focal point, which means higher radiation doses are probable.

Radiation dose rate varies linearly with the cosine of the angle of incidence of the x-ray beam and the sample under inspection. Therefore, if the sample is tilted to produce an oblique view, the dose rate given the device will be reduced.

The pad/ball interface of array devices is obscured during examination if only imaged from directly above, as is the case with normal x-ray. For these devices, oblique x-ray views are very important, particularly for determining if an open has occurred after the reflow process. However, because of the reduction in device size, rate reduction gained in using oblique views is minimized by the need to provide these views at higher magnifications.

Today, instead of tilting the sample to provide the required angled view, x-ray equipment manufacturers keep the sample perpendicular to the focal point and move the detector (Figure 2). In this way, the oblique view does not compromise the magnification and potentially limit analytical detail. However, this technique means no dose rate reduction from having an angled x-ray beam.

[FIGURE 2 OMITTED]

Dose Rate and Tube Power

The dose rate is also affected by the power of the tube. The power is calculated from the product of the accelerating potential used to make the electron strike the target (kV) and the filament current that produces the electrons. This parameter is used to set the x-ray tube at the level appropriate to a well-contrasted image at the detector. To a reasonable approximation, the dose rate is linear with power. So, for example, if the power doubles, the dose rate doubles. The more power that is available, the brighter the source.

Technical limitations do exist on the maximum power that the tubes can achieve, (4) but the greater the power used during inspection, the less time before a critical threshold is reached. Whatever tube conditions are used, repeating the inspection routine will double the dose the device receives.

Dose Rate and Filtration

The x-ray beam is absorbed at various degrees by different density materials in the device/board. Apart from attenuating the strength of the x-ray beam passing through the material, the x-ray energy spectrum can also be modified. This modification can be used to reduce the dose to devices substantially. The deliberate use of additional filtration by placing a material immediate[y in front of the sample can improve detection sensitivity. (5) This is achieved by optimizing the contrast of certain materials in the sample, such as the copper tracks on the PCB and the solder used on the joints. By using appropriate filters, the dose can be dramatically reduced without compromising the image quality needed for analysis.

An optimal filter for x-ray inspection is a 300 to 400 [micro]m thick zinc foil. At this thickness, the foil provides fui1 protection for most susceptible devices. (1,5) In practical terms, however, the thickness will reduce the x-ray flux passing through the device, resulting in low contrast images and long image acquisition times. Instead, a zinc filter of about 100 to 150 [micro]m thickness is proposed. This thickness would reduce the dose to the susceptible silicon by a factor of about 100X, and the change in brightness and contrast of the x-ray image produced is more modest, can be handled by today's x-ray imaging systems and still give useful analytical images. (5)

Dose Minimization

The following steps may minimize the dose imparted during inspection:

* Deliberately increase the distance between the sample and the tube focal point. Figures 3-5 and Table 1 show how a small increase in distance does not compromise the analytical quality but does dramatically reduce the dose rate.

[FIGURES 3-5 OMITTED]

* Add additional filtration of about 100 to 150 [micro]m thickness of zinc foil (Figures 6-7).

[FIGURES 6-7 OMITTED]

* Minimize inspection time by automating the procedure so that only those areas that need examining are examined.

* Because the dose is cumulative, reduce re-inspection.

* Make corroborative dose measurements to validate proposed inspection procedures and confirm the doses that are likely to be given to the samples.

Conclusion

The radiation doses given to electronic devices during x-ray inspection may be higher than expected because of the need for inspection at higher magnification. The susceptible part of the device, usually the silicon die, is brought closer to the source and is exposed to greater dose rates.

Determine if the devices being inspected are susceptible to random radiation damage from the dose levels during the inspection process. If you find that this susceptibility is the case, then estimate and, if possible, corroborate with measurements the doses delivered during the proposed inspection routine and minimize the dose.

TABLE 1: Changing the PCB to focal point distance affects the dose
rate that the sample receives. This change of distance has an
influence on the magnification of the final x-ray image.

Distance of Sample from         Dose Rate              Image
Focal Point of X-ray (mm)       Gy/min (ii)

         2.0 (i)                 D (iii)                 B
         3.1                     0.416 D
         4.4                     0.207 D
         6.9                8.4 x [10.sup.-2] D          C
        11.1                3.3 x [10.sup.-2] D
        17.5                1.3 x [10.sup.-2] D
        31.4                4.06 x [10.sup.-3] D
       147                  1.85 x [10.sup.-4] D
       275                  5.3 x [10.sup.-5] D          A

Distance of Sample from     Time of reach
Focal Point of X-ray (mm)   Dose D (min)

         2.0 (i)                1.0
         3.1                    2.4
         4.4                    4.8
         6.9                   11.9
        11.1                   30.3
        17.5                   76.9
        31.4                  246
       147                   5400
       275                  18868

(i.) Position of maximum magnification for this sample

(ii.) 1 Gy/min = 100 Rads/min

(iii.) The value of D will be kV, tube power and x ray system
dependent. These values of D ignore any effects of beam
filtration from other parts of the board/device.

References

(1.) R. Blish, IEEE International Reliability Physics Symposium, Dallas, TX, April 2002.

(2.) R V. Dressendorfer, Basic Mechanisms for the New Millennium, IEEE Press, NSREC Short Course, 1998.

(3.) http://nppp.jpl.nasa.gov/docs/Radcrs_Final5.pdf, from which the final two pages provide total dose thresholds for commercial semiconductor devices and citations.

(4.) D. Bernard, The Proceedings of SMTA International Conference, September 2002.

(5.) R. Blish et al. IEEE Trans. Devices & Materials Reliability, Vol. 2, #4, pp 102-106 (2002).

Richard G Blish, II, is an AMD Fellow with AMD, Sunnyvale, CA; [e-mail]. David Bernard is the x-ray systems product manager with Dage Precision Industries, Aylesbury, Buckinghamshire, UK; [e-mail].