A study of 15 systems finds most – but not all – offer a viable solution.

The requirement to comply with Europe’s RoHS regulations has driven adoption of a range of new materials in electronics components. A company failing to comply with RoHS can be fined. Hence, to ensure only RoHS-compliant materials are used, the industry has turned to energy-dispersive x-ray fluorescence (XRF) for incoming goods inspection. However, the technical capabilities of the related instruments are not well understood by the electronics manufacturing community.

A jointly funded industry/DTI collaborative project, led by the National Physical Laboratory, has been undertaken to determine the suitability of these techniques for determining the presence and levels of any restricted substances in typical electronics components. The project focused on an inter-comparison of different XRF equipment and test sites in a matrix experiment.

XRF is used throughout a range of industries for fast, nondestructive materials elemental analysis. Samples are bombarded with high-energy x-rays, and some of the x-rays are absorbed by the atoms of the sample. If the captured x-ray is of sufficient energy, an electron will be ejected from an inner shell of the atom, creating vacancies. To stabilize the atom, electrons from the outer shells fall to the inner shells, giving off an x-ray whose energy is the difference between the two binding energies of the inner and outer electron shells. As each element has a unique combination of electron shell energy levels, the spectrum of emitted x-rays is characteristic of the elements contained in the sample. The peak intensities of the emitted x-rays provide information about the concentration of the elements present.

The incident x-rays are usually provided by one of two alternative sources: an x-ray tube or a radioactive isotope. The x-ray tube is inert until activated by the operator, while the isotope source needs shielding to prevent operator exposure. Examples of both types were included in this study.

The emitted x-rays were analyzed using three types of detectors:

SiLi detectors. These have the best resolution, but require liquid nitrogen cooling to maintain stability.

Si-PIN photodiodes (formed from p-type/intrinsic/n-type semiconductor). These can be cooled using Peltier devices and are less expensive, but lack the resolution of SiLi detectors. These detectors form the majority of the systems evaluated here.

Proportional counters. These use photo-ionization of gases within the counters to detect x-rays. They are the least expensive of the options explored in this study and are widely used in PCB fabrication for plating thickness measurements.

Initially, a range of samples known to contain RoHS-banned substances was assembled by NPL. For each sample type, one sister specimen (i.e., from the same batch) was chemically analyzed to determine its composition and the levels of any banned substances. A master sample from each batch was then given to each partner for “blind” evaluation using their own XRF system. At each test site, therefore, the same sample was tested for each sample type. When all the XRF trials had been completed, the actual samples used were chemically analyzed using mass spectroscopy and SEM electron-dispersive x-ray analysis to confirm their elemental compositions.

A range of nearly 40 samples was used in the evaluation process, of which 22 contained RoHS-prohibited substances (Figure 1). The latter included Pb/Br/Cd/Hg in plastics (plasticizers/pigments) and lead in solders (bulk, joints and coatings). Fifteen RoHS-compliant samples were included to help evaluate the performance of the XRF systems.


Each sample was presented to the XRF systems three times, with the sample being removed from the equipment between tests. The XRF system parameters such as incident x-ray energy, analysis area (spot size) and emitted x-ray acquisition times, were set by the end-users/suppliers consistent with their own best practices. In total, 15 systems (11 different instruments from six manufacturers) were tested: 11 benchtop instruments (seven Si-PIN, one SiLi and three proportional counters) and four less expensive, portable machines (all Si-PIN). All employed x-ray tubes as the source for the incident x-rays, except one system, which used a Co57 source.

RoHS Screening of Plastic Components

In choosing plastic samples for these tests, a range of typical electronics components was evaluated before the final selection was made. During this procedure, it was noted that noncompliant components did not typically contain lead, mercury, bromine or chromium at levels around the RoHS limit of 1000 ppm. Rather, while typical values for noncompliant components were usually above 0.25% (2500 ppm), those for compliant components were typically less than 0.05% (500 ppm).

The Si-PIN or SiLi detector-based systems were capable of identifying the noncompliant components containing lead, mercury or cadmium. Eight typical RoHS-noncompliant plastics from electronic components were examined, and all 12 systems correctly indicated the samples were noncompliant. Three typical components containing bromine or chromium were also all correctly identified as containing these elements, hence requiring alternative tests for speciation of the hexavalent chromium and bromine compounds (PBB/PBDE/other).

Three RoHS-compliant components were tested, and all 12 PIN/SiLi systems indicated compliance for lead and mercury. Results regarding false negatives (i.e., incorrect indications of noncompliance) were also encouraging (Table 1).


In the case of cadmium, both benchtop and portable XRF systems were able to distinguish noncompliant systems for levels above 1000 ppm Cd. All the false indications of noncompliance occurred when the level was <260 ppm Cd.

The three proportional counter-based systems did not perform well for testing plastics. Although not specifically designed for testing for RoHS compliance in plastics, all the systems incorrectly indicated the presence of RoHS-banned elements. Moreover, quantitative results were unavailable or inaccurate. It is recommended proportional counter systems not be used for RoHS compliance measurements.

How the sample is presented to the test equipment is important. When a cable strain relief was tested while still attached to its accompanying cable (Figure 2), the lead level recorded was reduced by the effect of the material within the strain relief itself. In this instance, the lower results were noncompliant, but should this example have had lead content closer to the RoHS limit, this modification could have led the sample to be incorrectly marked as compliant. Where recorded contaminant levels are close to the RoHS limit, it is recommended the plastic be tested in isolation.


RoHS Screening of Bulk Solder Alloys

For larger samples, large analysis spot sizes can be an advantage. Sampling over a larger area can reduce any effects of segregation within the sample and give a more meaningful analysis. For the lead-contaminated, bulk alloy samples tested, the Si-PIN/SiLi-based XRF systems were able to detect lead levels to around 500 ppm. Some systems even achieved good repeatability at 50 ppm lead. All the systems indicated noncompliance in all instances of lead present in tin at levels of 2000 ppm. At 1000 ppm lead, 11 of 12 systems indicated noncompliance for lead, or within 10% of RoHS limit. Thus, all Si-PIN or SiLi detector-based systems proved suitable for screening bulk solder samples for RoHS compliance for lead, detecting lead at or above 2000 ppm. For levels between 500 and 2000 ppm, additional techniques are recommended if accurate elemental analysis is required.

Incorrect indications of noncompliance were again recorded for cadmium in SnPb. Ten Si-PIN/SiLi systems were used to analyze cadmium levels, and of these, 50% gave false indications of noncompliance. All the proportional counter-based systems were able to detect lead in bulk solder samples at 0.2% or above. But, only one system was able to detect lead to 0.1%, and none was able to detect less than 0.1% Pb. Some proportional counter-based systems were not able to detect less than 0.2% lead in bulk alloy samples; therefore, care should be taken when using them for RoHS compliance screening.

RoHS Screening of Solder Joints

For the much smaller SM solder joints, the measurement window size (or spot size) is important. For accurate determination of lead content, samples need to fill the measurement window. Solder joints by their nature tend to be thin, and the signal recorded may contain contributions from other materials beneath the solder joint.

Screening SOIC joints with large spot sizes. With Si-PIN/SiLi detector-based systems, the spot size for portable systems tended to be greater than for the benchtop machines (at least 3 mm diameter). With the toe of the SOIC solder joints tested approximately 0.65 x 1 mm, such joints filled only around 8% of the measurement window of a typical portable system. A further complication is that the joint areas examined were not of constant thickness due to the shape of the fillets. Although one of the portable systems did indicate noncompliance for lead for a sample SOIC joint containing ~3% lead, the recorded values were still very low at <0.2% lead. Similarly, for a SOIC joint containing ~11% lead, three portable systems indicated noncompliance, but gave recorded lead levels at least 10 times lower than the actual value. Thus, for identifying samples containing high lead levels (40% or more), the majority of the handheld systems are adequate, but the systems tested could not be relied on to identify noncompliant joints where lead levels were below 3%. Such systems may be suitable for analyzing joints if sufficient joints were removed from the assembly and collected together to fill the measurement window.

Screening SOIC joints with smaller spot sizes. The benchtop Si-PIN/SiLi detector-based systems, with their generally smaller spot sizes (0.1 to 2 mm diameter), were able to identify noncompliant joints, with all eight systems tested giving correct identifications (Figure 3).


Screening resistor joints. Spot size also played a part in the analysis of R1206 solder joints. The latter are 3 x 2 mm, filling around 85% of the typical measurement window of a portable system. Although no lead (<0.1%) was present in one sample, six systems recorded lead at levels of 0.1 to 0.8% in that sample (i.e., RoHS-noncompliant). The majority of these systems had larger spot sizes, and consequently, the measurement window extended beyond the joint area and included part of the component and PCB base (Figure 4). The lead signal therefore included some contribution from the resistor, which was known to contain lead in the passivation layer of the resistive element. Thus, incorrect RoHS noncompliances were generated.


Screening BGA joints. Several assembled BGAs were tested to determine if the XRF systems could detect and measure the lead content of the assembled BGA balls, when measured through the top of a BGA. This approach might permit nondestructive testing of completed assemblies, as may be required on imports into the EU. Only one system, using a Co57 x-ray source, recorded any lead signal sufficient to indicate noncompliance. However, the measured values from this system were significantly below the actual sample values (1.98% for 40% lead sample, and 0.32% for a 10% lead sample). Hence, if lead contamination levels were low, none of the systems could be relied on to determine RoHS compliance.

Incorrect identifications of cadmium. Again, the detection of cadmium during joint testing was difficult. Six systems gave 19 (of 121 measurements) false indications of RoHS noncompliance. All the recorded values were less than 0.08%.

Solder joints and proportional counter-based systems. Performance of proportional counter-based systems was variable. Only two systems provided a full set of measurements, neither of which proved adequate for identifying RoHS-noncompliant joints.

RoHS Screening of Other Components

Screening components/solder pastes in packaging. At incoming inspection, there are distinct advantages in being able to test components/materials without removing them from secondary packaging (tapes, reels, sticks, etc.). This may avoid deterioration after opening, as in the case of solder pastes, or prevent waste, as in the case of components in reels. Handling damage in removing components from sticks can also be avoided. When testing resistors on the reel, five of the 11 XRF systems evaluated (including one proportional counter-based system) did not record a sufficiently high level of lead to identify the RoHS-noncompliant components. Of the remaining systems, four recorded significantly lower lead levels present than were actually the case. One proportional counter-based system recorded significantly higher lead (34%) and the remaining system (proportional counter-based) did not quantify the amount of lead present. When known RoHS-compliant chip resistors (Pb-free terminations but with lead in the passivation of the resistive element) were tested on the reel, four of nine systems incorrectly indicated the resistors were not RoHS-compliant (up to 2.4% lead). Thus when RoHS compliance testing of chip resistors, removal of components from the reel is recommended.

SOIC components in sticks or tubes were also studied. Four of the 10 quantitative systems could not identify noncompliant components. All four were portable, with larger spot sizes. As discussed with the SOIC solder joints, these systems’ measurement windows were relatively large compared to the component termination size, and thus, surrounding materials attenuated the lead signal. The other systems, with generally smaller measurement windows, did not record accurate values for the lead in the terminations with values between 1.5 and 21% for a termination with 36% lead. For compliant components in tubes, four of the 10 systems evaluated indicated lead was present above 0.1%, or recorded lead peaks in the relevant spectra, suggesting an incorrect RoHS noncompliance. Thus for RoHS compliance testing, removing the components from the stick or tube would be required. For sensible analysis in systems with larger measurement windows, several component terminations may be required to fill the window.

When a sample of Sn62 (36% lead) solder paste in an unopened pot was tested, all 10 systems evaluated indicated the paste was not RoHS-compliant or contained lead peaks in the relevant spectra. But the measured values were, with one exception, well below the true value (in two cases as low as 1%). For a paste contaminated with lead at a lower level (~1%) in an unopened pot, four of the 10 systems did not correctly identify the paste as noncompliant. When an unopened Pb-free (0.040% lead) solder paste was examined, three systems falsely indicated the paste was noncompliant. In view of the degree of uncertainty with these solder paste measurements, it is recommended a sample sufficient to fill the measurement window of the test instrument be removed from the pot for testing.

Screening resistors. Even when removed from their reels, the analysis of some chip resistors still presented problems. Some components with pure tin terminations were found to contain lead, the signal clearly emanating from the passivation of the resistive element. When the component was tested with the resistive element toward the detector, all 12 systems under evaluation falsely indicated the component was RoHS-noncompliant, or recorded lead peaks in the spectra. The lead values varied between 0.14 and 30%. The higher levels of lead were generally associated with those systems that used a larger measurement area. Two systems also falsely indicated the terminations contained noncompliant levels of cadmium. Generally, it is recommended that when testing resistor component terminations for RoHS compliance, the resistor should be examined from the reverse side, with the resistive element facing away from the instrument detector.

Care must also be exercised when testing components with multiple layers. A spacer tested during the project was RoHS-compliant because its surface comprised Sn-plating on brass containing lead at <4% (brass has an exemption in the legislation). However, when evaluated, 12 of 14 systems indicated the component was noncompliant or recorded lead peaks in the spectra. This sample illustrates operators need a sound understanding of the materials involved in the samples to ensure sensible analyses. Challenging samples may require destructive testing by removing the coating and then testing separately.

Finally, a Cr-passivated, Zn-coated screw was tested (Figure 5). Chemical tests proved inconclusive for CrVI, but chromium was clearly present on the surface. The quantitative test instruments recorded levels of chromium varying from 0.77 to 1.8%. Clearly these XRF systems can detect the presence of chromium, but because of the lack of speciation, are unable to confirm RoHS compliance. The three proportional counter systems failed to detect any chromium.


Conclusions

XRF systems offer a viable method of screening for RoHS compliance. Compared to chemical analysis, these systems offer lower unit cost, lower running costs and faster results. Smaller sample sizes are also possible. Use of these systems, however, requires a semi-skilled operator with a sound understanding of the technique/equipment capabilities and the likely composition of materials involved in component assembly.

Nondestructive testing is the norm, but some samples may need separation into constituent parts, or removal from placement packaging, for meaningful testing.

Acknowledgments
The NPL would like to thank the following project partners, without whose help the project would not have been possible: Alcatel Alenia Space Italia; EADS Astrium; Fischer Instrumentation (GB) Ltd.; MBDA (UK) Ltd.; Oxford Instruments Analytical; Research in Motion; RMD Instruments; Roentgenanalytik; Rolls Royce Marine; RS Components Ltd.; Thermo Fisher Scientific Niton Analyzers; Tin Technology.

The work was carried out as part of a project in the Materials Processing Metrology Programme of the UK Department for Innovation, Universities and Skills.

Martin Wickham is a consultant and Dr. Christopher Hunt is group leader, Electronics Interconnection Group, at the National Physical Laboratory (npl.co.uk); martin.wickham@ntlworld.com.

• Prev
• Next