For over 40 years, the electronics industry has been seeking ways to determine an answer to the thorny old question: how clean is clean? Meanwhile, others were plotting ways that they could monitor the quality of their production
process with specific regard to the presence of ionics (salts) that conspire to cause circuit failure.
They figured that using a blend of alcohol with de-ionised water would be an ideal medium, measuring its conductivity before and after. Why alcohol? Because conventional rosin based fluxes used at that time were soluble in alcohol. Why water? Because salt dissolves in water.
So the process amounted to mixing 75% propan-2-ol (IPA) with 25% de-ionised water and using a mixed resin filter (a mixture of cation, anion and chelate) to ‘strip out’ ionics as they passed through the medium; then cleaning the test tank and its contents to a determined conductivity level, expressed as micro-Siemens (µS); then putting the object to be tested into the tank and measuring the change in conductivity.
The result could then be extrapolated as an equivalence of NaCl (plain salt) to obtain a record of the amount of ionics the process in question would leave on, or put on, the circuit assembly. The names employed for this test are ROSE (Resistivity of Solvent Extract) or SEC (Solvent Extract Conductivity).
Those seeking better process control had found an ideal tool: a measurement of ionics that might be present on a selected sample. Then, during the working day, changes in the level detected would be a really good indicator that the process was in or out of control.
Ionics take several forms and from many and various process steps, some are more soluble than others. If they are present on a circuit and are exposed to moisture in the presence of electricity, then an electrolyte is formed, electrochemical reactions occur, and that results in dendrites – inter-metallics between the cathode and anode that provide a path of lower resistance leading to short circuits and/or circuit failure.
In the 1970s the US Department of Defense (DoD) considered that this might be a useful test to control cleanliness in production and established a ‘pass/fail’ threshold at a level of 3,1 µg/cm2 of NaCl equivalence with ‘dynamic’ testing and 1,56 μg/cm² of NaCl for ‘static’ testing.
This was not such a good decision for two reasons: firstly because, by logical extension, it indicated that one could safely leave up to that amount of measurable ionics (salt) on assembly surfaces; and secondly there is no difference between dynamic and static, both test methods now require 1,56 μg/cm² NaCl equivalence and the dictum remains “it’s okay to leave that amount of salt on my assembly”.
As many have found to their considerable cost, it is not.
This is important because exposing any circuit assembly to a mixture of alcohol and water for 15 minutes or more significantly increases the risk of other ionics leaching out of the laminate and onto the surface (Swedish Institute for Production Engineering Research – IVF report 1990).
Hot or not?
It has been suggested that the test solution used may be heated to 400°C or more, but this is not acceptable since, apart from the fundamental changes in conductivity and its effect upon test accuracy, it significantly increases the risk of sub-surface ionic leeching as well as posing an explosion risk. Note that the test solution flashpoint is 190°C at ambient temperature.
Static vs dynamic: what’s the difference?
IPC-TM-650 2.3.25 and 18.104.22.168 permit the use of either static or dynamic test methods. Whilst both methods should yield the same result, good test methods should have only one variable – that being the item under test.
The dynamic test method, by its very nature, has an additional variable in that it behaves more like a cleaner with the test solution passing the conductivity probe, the filter and back into the tank. In this way the test solution is being continuously cleaned during the test.
The static test, by contrast, re-circulates the test solution via the conductivity sensor but bypasses the ionic exchange filter, thereby removing this important variable.
Is saturation a problem?
Some talk about a ‘saturation’ problem with static testing. Given that the test system may commonly be calibrated up to 30 μg/cm² of NaCl equivalence, and that the pass/fail level is 1,56 μg/cm² of NaCl equivalence, this means that if the process was producing boards that had a contamination that was going to send the solution into saturation, it would be well above 30 μg/cm².
One should be more concerned about stopping the production process than worrying about the solution saturation.
Test tank size
When selecting a test system, it is important to use the smallest possible tank size for the circuit under test. As outlined in IPC-TM-650 22.214.171.124:6.10 there is some concern regarding ROSE tester cell size. Testing a 2 x 2 cm board in a 20 000 mL cell causes such a severe dilution as to cause the signal to be lost in the noise.
A recommended cell size is 5000 mL or less. Smaller cell volumes will allow for a more measurable result. If a smaller cell or running with a smaller test volume are not an option, then the number of bare boards can be increased, all extracted separately, and the extract solutions all tested at once.
In a further attempt to destroy any arguments regarding that most heinous subject – the ‘Equivalency Factor’ – I argue that there is no such thing! If the test system is working correctly, during calibration a quantified amount of NaCl solution is put into the test tank and the system should be capable of recording the amount precisely. If it doesn’t, then there is something fundamentally wrong with the machine.
From IPC-TR-583 ‘An In-Depth Look At Ionic Cleanliness Testing’:
Once the test method and the pass/fail criteria were established, equipment manufacturers began designing and building systems to do this type of testing. Due to efficiency, or perhaps the slightly different measuring process, it as noted that the new equipment would typically give higher results than that of the beaker/funnel technique.
In 1978, a second study was performed at NAWC to establish ‘equivalency factors’ for some of the new equipment which would be incorporated into various military standards, such as MIL-P-28809 and WS-6536.
The theory behind the ‘equivalency factors’ was that the same PWA that measured 10,06 μg/in² using the beaker/funnel test would have measured 14,00 μg/in² in a static system and 20.00 μg/in² in a dynamic system under the conditions of study.
As the years progressed, more advances were made to the equipment such as the incorporation of solvent heaters, microprocessors and sprays. As the efficiency of the systems increased, it became increasingly apparent that the equivalency factors established for the 1978 equipment did not apply to current equipment. In addition, equipment introduced to the market after the study are not mentioned, even in the revised standards, and erroneously not considered as accepted test equipment by potential users.
Good cleanliness test systems need to be accurate, reliable, repeatable, simple to use and easy to maintain. They also need to reduce test time to a minimum, take account of temperature, circuit volume and atmospheric absorption of iogenic gasses, and avoid polarisation effects between electrodes.
When making a purchasing decision on a system it is important to keep in mind whether it uses ‘curve fitting’ algorithms that reduce test time, uses a pure gold sensor to improve accuracy and reduce maintenance , and measures at accuracies of better than 0,005 μS.
One should ignore any suggestions of using heated test solution, or that there is a difference between dynamic and static testing, or suggestions of saturation effects (this is irrelevant as the level of contamination to achieve that condition would be so great).
But is this the only way to answer the original question, ‘how clean is clean?’ Well, no, it isn’t. It is surely the quickest and simplest check that a production process is under control but it is important to recognise that these ‘cleanliness testers’ are in fact ionic contamination testers. They do not inform about the presence of non-ionic contaminants, of which there are many.
These include an enormous variety of surfactant additives in various process chemistries, including solder resist, solder flux, wire and paste, adhesives and cleaning chemistries. They are mostly used as wetting, levelling or de-wetting agents that can contribute to failure by adverse electrochemical reactions, especially if the manufacturing process does not include cleaning.
So what are the alternatives?
It is not the intention of the author to now examine in detail each of the alternative test methods, rather it is to simply highlight these alternatives to better appreciate why ROSE testing is such a preferred technique.
Ion chromatography (IC) is based on the use of specialised column packing for separation of ions that is able to separate, identify and quantify ions in a sample matrix, allowing the separation of ions and polar molecules based on their charge.
Advantages of IC include the following:
* Highly accurate, it can identify exactly what kind of contaminants are on the boards, and help to trace back the root cause of the problem through each production process.
* Can be used as a quality control tool, eg. for goods inwards inspection on a sampling basis.
* Excellent species differentiation.
* Can be employed to test for localised contamination.
But the process has limitations:
* Requires highly skilled operators.
* Expensive to run.
* Time to run test exceeds 15 minutes.
* Will identify exactly what is on the surface under test, but not whether the end product will be reliable.
Fourier transform infrared spectroscopy
Fourier transform infrared spectroscopy (FTIR) is a measurement technique whereby spectra are collected based on measurements of the temporal coherence of a radiative source, using time-domain measurements of the electromagnetic radiation or other type of radiation.
It involves collecting infrared spectra, but instead of recording the amount of energy absorbed when the frequency of the infrared light is varied, the IR light is guided through an interferometer. After passing through the sample, the measured signal is the interferogram.
* Highly accurate, it can identify exactly what polymers may be on the boards.
* Requires scientifically trained operators.
* Expensive to run.
* Generally run off-site at independent laboratories.
* Like IC, it will identify exactly what it has found (polymer contaminant) but it will not determine whether the end product will be reliable with its presence.
Surface insulation resistance
The principle of Surface Insulation Resistance (SIR) involves an inter-digitated test pattern to which is applied an electrical bias. The degradation or changes to surface insulation resistance are then measured.
* Determines the effects of both ionic and non-ionic contamination.
* Demonstrates the electrochemical compatibility between all process materials.
* Can be used to monitor material trends.
* A quantitative, not qualitative, test method.
* Works in conjunction with ROSE / SEC.
* Predicts whether an end product will be electrochemically reliable.
* Requires skilled operators.
* Requires dedicated equipment.
* Is carried out on dedicated test coupons that are representative of the end product.
* Takes a long time – not less than 72 hours.
* Will determine if the end product will be reliable, but not what’s causing a failure; for that IC or FTIR are required.
In summary and by this author’s recommendation:
1. Decide on the preferred process material mix and run SIR qualification tests; use the IPC B52 test coupon/vehicle that includes ROSE and IC snap-off coupons.
2. Analyse any failures using IC or FTIR.
3. Use ROSE / SEC tests to maintain process control.
4. Use SIR to monitor material quality by trend analysis.