¹ EDIP: Environmental Design of Industrial Products

² This corresponds to 0.6-0.7% of the world lead consumption and approx. 11% of the world tin production.

³ [6] has applied prints with adhesive and solder by using the same stencil. He scaled the prints and found a difference of about a factor two.

5.Recycling analysis

5.1 Material contents
5.2 Recycling process
5.2.1 Collection 
5.2.2 Dismantling facility 
5.2.3 Metal recovery - recycling
5.3 Differences in material value
5.4 Differences in environmental load

A recycling analysis has been carried out with the aim of identifying differences connected to recycling of electronics assembled with traditional solder and electrically conductive adhesives, respectively. The Danfoss EST (electronic thermostat) was used as a case. It was tried to answer the following questions:

	What are the contents of the two EST subjects?
	Do the recycling processes differ?
	What are the differences in material value?
	What are the differences in environmental load connected to the recycling?

5.1 Material contents

The complete composition of the two EST case subjects was inventoried by contacting the component suppliers. Data were obtained from most suppliers. Data on the remaining 4-5 components were estimated based on data for the other components. Material composition of the PCB’s were partly calculated and partly obtained from the suppliers.

Most of the supplier information has been forwarded to Danfoss A/S under confidentiality, why only a total can be presented. Differences are due to: The surface treatment of the PCB: An eutectic SnPb HAL (Hot Air Levelling) surface and a solder mask (resist) is used when soldering, whereas adhesive requires chemically applied Ni and Au PCB top layers. The joining material: Solder versus conductive adhesive.
Through-hole components: Two of the through hole components in the soldered version are substituted with SMT components in the adhesive version. NB! Care should be taken not to interpret data on these two components as a general picture of differences between functionally comparable through-hole and SMT components.

Component leading will ideally also differ. Traditional components contain an inner AgPd layer, a Ni-barrier and a SnPb layer on top. Optimum adhesive components only contain a AgPd layer [6]. It was, however, not possible to obtain components with this termination for the test case. Environmental consequences of using different component termination are addressed in the LCA case (Chapter 7).

5.2 Recycling process

For Danish conditions, end-of-life electronics are collected and forwarded to an electronics dismantling facility.

No differences are expected in relation to collection of electronics joined with solder and adhesive, respectively.

5.2.2 Dismantling facility

Danish dismantling facilities typically conduct a preliminary mechanical separation, where the mounted PCBs are separated from major metal and plastic parts (which are forwarded for recycling). The PCBs are then checked for known problematic components such as mercury batteries and electrolytic capacitors containing polychlorinated biphenyls. Problematic components are removed with a pair of tongs. Elektromiljø A/S carried out a test on the two subjects and did not find any differences in dismantling feasibility.

No differences are expected in relation to dismantling electronics joined with solder and adhesive, respectively.

The PCBs are sent for further resource recovery at abroad recycling facilities.

In general, extraction of metals from scrap is a very complex process, because the various types of raw materials used contain a combination of metals and impurities. The processes involve roasting/drying, smelting, converting and refining. After refining of one specific metal, the remaining impurities (sludge) containing several metals are treated in other processes for recovering.

PCB’s may be recovered by different recycling technologies, see e.g. [8]. Elektromiljø A/S has assessed that both subjects would ideally be recycled by the ‘copper process’. The copper process is applied at Boliden Rönnskär, Sweden and will briefly be described as typical for an up to date copper recovery plant.

The Boliden smelter, Rönnskär is one of the largest smelters for electronic scrap. However, the major input to the process is refined copper ore. Other raw materials, incl. electronic scrap consist about 25%.

5.2.3.1 Recovery at Boliden, Rönnskär

The copper process at Boliden is briefly described below. Focus is on metal recovery and especially those metals for which differences are seen between products interconnected with solder and adhesive, respectively.

Electronic scrap (circuit boards from computers, television sets etc.) is inspected to prevent the inclusion of dangerous materials such as mercury and radioactive isotopes. The collectors or dismantling facilities will normally remove electronic components containing dangerous materials before delivery.

The secondary raw materials incl. electronic scrap is flash smelted in a "Kaldo plant", primarily for recovering the copper. The flash smelting recovers energy from burning the organic materials in the scrap at around 1250 °C.

The melt is transferred to a copper plant for converting, casting and electro-refining. In a converter, sand is added as a slag-former at about 1200 ° C for removing iron and zinc. The removed material is further treated and purified in order to give the by-products ‘ironsand’ and ‘zink billets’.

The "white metal" (copper sulphide), formed in the converter, is oxidized (blowing) to 98% pure copper, followed by an injection of liquid ammonia for reducing the level of oxygen. The melt is casted into copper anodes containing 99% copper and 0,5% precious metals.

The copper anodes are electro refined in the copper refinery, using sulphuric acid and copper sulphate as the electrolyte. The sludge in the electrolyte tanks contains precious metal impurities and is forwarded for a separate precious metal plant, where gold, silver, platinum, palladium and selenium are recovered. Boliden has informed that close to 100% palladium and gold is recovered and between 90 and 99% silver. Another by-product from the refinery is a Nickel-fraction, which is treated in a nickel-sulphate plant. The nickel recovery percentage is about 80%.

Tin and lead mainly ends up as dust in gas streams from the Kaldo plant and from the converting. The gas streams are cleaned, capturing the major amount of the tin and lead. This fraction is sent for abroad metal recovery. At present, about 80% may be recovered from an economical point of view.

Metal amounts not recycled end up in waste products, as contaminants in the by-products or leave the plant as emissions.

Very little information is available on these waste streams. However, it is assumed that the far major amount ends up as solid waste products. At present waste materials are stored within the industrial area. A permit from the Swedish National Board of Franchise states that the company shall take measures to store and handle waste material in a manner that prevents nuisance from an environmental point of view via contamination of water, soil or air. The permanent disposal of these waste fractions is currently under investigation. Annually, a public environmental report with full description of the amounts and chemical composition of waste products is sent to the Swedish EPA.

5.3 Differences in material value

Demet Deutche Edelmetall Recycling AG & Co. (Alzenau, Germany) has estimated the following ‘recoverable material value’ for the two EST subjects:

	Soldered version: approx. 7 DM (German Mark)/kg (about 26,50 DKK/kg)
	Adhesive version: approx. 11 DM/ kg (about 41,50 DKK/kg)
	About 2 DM/kg should be subtracted for processing costs.

The results indicate that considerably more precious metal is used for the adhesive version. Interestingly, differences in component termination are not included in the test case. ‘Optimum’ component termination for the adhesive alternative (Ag/Pd) would increase the silver and palladium consumption further and result in an even larger difference.

The result should as such only be taken as indicative, as specific print layouts (PCB area, components applied, packing density etc.) may severely affect the metal content and thereby the value. However, adhesive prints must be assumed in general to contain a higher metal value.

Elektromiljø A/S has looked at the specific case subject and found that the PCB would not be separated from other thermostat parts (the chassis) due to labour cost at the dismantling facility. Therefore, the entire thermostat would be forwarded for recycling. The chassis parts do not contain precious metals and the value per kg thermostat is therefore much lower than the PCB value per kg. This influences the economic benefit as the recycling process costs to some extent are per kg electronic scrap delivered to the recycling facility.

Finally, it should be noticed that it would require a substantial substitution for adhesive technology before differences in material value becomes economically visible for the dismantlers and other interested parties. If only few ‘adhesive PCBs’ are present in the PCB scrap fraction, they would physically and economically ‘disappear’. Possible adhesive substitution potentials are discussed in [3].

5.4 Differences in environmental load

Different recycling facilities have been contacted in order to inventory differences in environmental loads (energy consumption, emissions and waste generation) from recycling solder and adhesive PCBs. However, except for metal recovery percentages (see previous paragraph ‘Metal recovery – recycling’), it has not been possible to obtain data on this issue.

The requested data are not available. It is very difficult to allocate the environmental burdens from the complex recycling processes between the inhomogeneous input materials varying dramatically in composition. Further, an allocation would probably require more detailed measurements than what is presently available.

It can be concluded that it would require a substantial developmental work to map and allocate environmental burdens in a way that would make it possible to differentiate burdens associated with solder and adhesive PCBs, respectively.

6. Occupational health aspects of adhesive and solder technology

6.1 Comparison
6.2  Recommendations
6.2.1 Principles for designing a production line with minimal impact on occupational health
6.3 Handling conductive adhesives safely

The objectives of the occupational health analysis have been to:

1. Compare occupational risk associated with the mounting processes applied when joining electronics with electrically conductive adhesives and tin/lead solders, respectively.
2. Provide recommendations on optimum design of production lines and manual handling when applying adhesives

6.1 Comparison

A detailed analysis for the Danfoss test case can be found in [4]. The conclusion states that it is difficult to judge between the alternatives as exposure data are lacking. However, if proper measures are taken, it is believed that the risk of handling the materials is low for both adhesive and soldering technology.

Another occupational health assessment was carried out at Mekoprint A/S. A comparison with soldering was not possible here as only adhesives are applied. Air measurements had been made here in order to detect possible inhalation exposure connected with adhesive application. Further, an evaluation on skin contact exposure was made. The conclusion – also here – was that proper handling of the adhesives results in a very low occupational risk.

Information on inherent toxic properties of a wide number of adhesives can be found in [5].

6.2 Recommendations

Based on experience from the two occupational health cases, recommendation can be given in relation to design of a production line and to manual handling of conductive adhesives.

6.2.1 Principles for designing a production line with minimal impact on occupational health.

When designing a new production line, a number of occupational health and safety issues have to be taken into consideration.

First of all, the choice of machinery and the design of the work process should support the establishing of a workplace that is feasible from an ergonomic and a psychological point of view. The machinery should be chosen to avoid noise- or vibration related problems.

These issues will not be discussed any further in this paper. The paper focuses on the possible impact on health and safety caused by the use of adhesives.

The conductive adhesives are dispersions of an appropriate conductor (e.g. silver) in a polymeric binder. The typical one-component binders, which are applied in the conductive adhesives, are epoxy, silicone, polyester and polyimide resins. Silicone and epoxy resins are used in two-component binders as well.

For some applications the use of adhesives will require cleaning of the printed wiring board (PWB) before the adhesive is applied. The recommended cleaning agent will most frequently be an organic solvent such as butanone or ethanol.

The units containing and applying the adhesives must be cleaned regularly using a proper solvent. This solvent will often be identical to the solvent used as a thinner for the adhesive.

For a general hazard and risk assessment of the chemicals, please refer to [5].

Recommendations:

	One-component adhesives should be used whenever possible from a technical point of view. This will render the mixing of two components superfluous and reduce the risk of exposure (skin contact and inhalation). In case a two-component adhesive is to be used, the manual mixture of the adhesives should be avoided by establishing a mechanical mixture unit in the assembly line.
	The adhesives should be bought from the supplier in containers ready for installation in the assembly line if convenient. This will eliminate the risk of exposure (skin contact and inhalation) while decanting the adhesive into other containers. Another possibility would be to decant the adhesive to containers for use in the assembly line in one step and the maybe freeze the containers until they should be used.
	Manual handling of the chemicals should be avoided to the greatest possible extent, e.g. when applying the adhesive, mixing or cleaning the accessories.

	Cleaning of the PWB should be carried out in a closed automatic cleaning machine. In different industries the problems related to cleaning technology have been solved in ways leaving only few health- and safety related problems. It is recommended to benefit from these experiences.
	Cleaning of the application part (e.g. a needle) should be done automatically. The cleaning process might be a periodically occurring and integrated part of the production flow.
	If cleaning is planned as a manual or semi-manual process, proper process ventilation will be required if any organic solvents are used. In addition measures have to be taken to avoid skin contact.

Most adhesives will cure only or most efficiently by heating. In case the curing process cannot be brought to an end on the automatically line, the PWB’s should be collected in a well ventilated area in order to ensure fresh air in the breathing zone of the people working there.

It is recommended to establish the possibility for ventilation of the assembly line already when designing since it is difficult to envisage which type of adhesive will be chosen from the technical point of view in due time. Even though a process has been automatized, the possibility for liberation of vapours to the working environment should be avoided. Whether the ventilation has to be process ventilation or a general ventilation of the room depends on the amount and the hazard of chemicals in use.

	The curing of the adhesive should take place in a ventilated area. Some adhesives will start curing as soon as applied requiring proper ventilation of the assembly line. Special notice has to be taken when silicone based adhesives are chosen, as these may liberate irritating vapours during curing.
	Any use of organic solvents requires ventilation as well.

The recommendations given adhere to the design- and planning process. Recommendations on how to handle the chemicals safely are given in the following paragraph.

6.3 Handling conductive adhesives safely

Recommendations on how to design a production line with minimal impact on the occupational health and safety was given above. By following these recommendations, the risk connected with working with conductive adhesives will be reduced.

However, a well considered working practice, the proper use of available technical preventive measures and personal protective equipment as well as good hygiene at work are also necessary in order to minimise the actual risk.

As a basic principle, the preventive measures should be prioritised as follows:

1. Designing the assembly line in a way that reduces risk as much as technical possible
2. Establishing technical measures (e.g. ventilation) where needed.
3. Prescribing the use of personal protection equipment.

Personal instruction at the workplace about the actual health hazards arising from the adhesives and other chemicals is a must in order to enable the employees to handle the hazards and to make the right use of the preventive measures offered. Knowledge about the chemicals and their inherent hazards will provide the necessary background for sound handling of the chemicals.

The instruction must include the following:

	Knowledge about the inherent hazards of the chemicals, e.g. whether the chemicals contain sensitising substances, must be available in order to decide how to work with the chemical and to motivate the employees for using the proper preventive measures.
	Skin contact with chemicals must be avoided. It is of utmost importance that gloves are used properly and that the gloves are resistant to the chemicals. Routines on how often to change the gloves and how to handle the gloves must be implemented.
	Inhalation of vapours and dust must be avoided. It is important for the employees to know how to control the efficiency of the ventilation.
	The hygiene at the workplace must be proper in order to avoid the risk for oral uptake of chemicals (e.g. the metals) due to contaminated cloths and contaminated workplaces.

Whenever required by legislation, the employees must receive the mandatory training as well as a general training is recommended.

7. Life cycle assessment of electrically conductive adhesive vs. traditional tin/lead solder

7.1 Goal and scope definition
7.1.1 Products/Alternatives
7.1.2 Functional unit
7.1.3 Process tree, data quality and limitation of the life cycles
7.1.4 Assessment method
7.1.5 Irregularities and accidents
7.1.6 Capital goods
7.1.7 Allocation
7.1.8 Critical review
7.2 Results – Resources
7.2.1 SMT scenario
7.2.2 Danfoss scenario
7.3 Results - External environment
7.3.1 SMT-scenario (disposal scenario: HW)
7.3.2 Danfoss scenario (HW)
7.3.3 SMT scenario (TH)
7.3.4 Danfoss scenario (TH)
7.4 Conclusion and Recommendations

Adhesives have during the latest decades become a potential substitution for lead containing solder in electronics joining. Lead is suspicious from a environmental point of view due to the human toxicological potential of the metal. Adhesives on the other hand contain silver, which is suspicious due to its high ecotoxic potential.

In the first phase of the Kamille project (Kamille I), a qualitative life cycle screening was carried out. The results indicated adhesive to be the better alternative. However, a quantitative assessment was assumed preferable [25]. Other works have been carried out in order to compare adhesive and solder technology. However, [22] did not include the differences in printed circuit board and component terminations, which are necessary when substituting to adhesives and [15] only focused on energy consumption as a parameter.

The aim of the present analysis has been to compare adhesive with solder joining technology in a life cycle perspective by inclusion of all possible differences and by conducting an impact assessment according to the EDIP-methodology [24]. This report only presents some of the LCA. Please consult the working report [2], which contains a careful discussion about limitations as well as the inventory and an energy analysis for full details.

7.1 Goal and scope definition

The overall aim of the LCA is to make an environmental comparison of the electrically conductive adhesive versus the traditional soldering interconnection technologies in a life cycle perspective.

The present analysis is made as environmental documentation in the Kamille project. The intended audience are environmental/LCA specialist; e.g. among authorities, larger companies and consultancies.

It has been aimed at following the main principles in the ISO 14040 standards.

7.1.1 Products/Alternatives

A Danfoss electronic thermostat EST (077F0301) was chosen as a case product. The EST is to be used in a refrigerator.

In the following, ‘EST-sold’ refers to the soldered product, whereas ‘EST-adh’ refers to the product interconnected by electrically conductive adhesive.

In the first run it was aimed at conducting the Danfoss case as representative for state-of-the-art soldering and adhesive technology

However, as pointed out elsewhere in the Kamille project [7], it has not been possible. It has e.g. not been possible to obtain components without PbSn-termination, where AgPd-terminations would have been preferable [6]. Further, the adhesive print and the SMT-stencil in the Danfoss case were not designed to optimise adhesive consumption. During the work it further showed up that the metal amounts estimated for Printed Circuit Board (PCB) surface finish were not representative for the general state-of-the-art.

Another ‘speciality’ of the Danfoss case product is related to the three largest components. In the EST-sold version, these components are mounted by through-hole technology, which include a wave-soldering step. All other components are mounted by SMT (Surface Mount Technology). In the EST-adh alternative, all components (incl. the three largest) are mounted by SMT. Substitution of wave-soldering is an example of an advantage with electrically conductive adhesive technology; that the curing temperature in the SMT/reflow-oven is much lower (and therefore does not harm large/sensitive components) than what is necessary when soldering. However, in most situations, it is assumed that adhesives will be used to simply substitute ‘pure’ SMT soldering.

Altogether, the Danfoss test case showed up not to be representative for the general comparison between technologies

It was therefore decided to model and assess two scenarios - EST-SMT and EST-Danfoss – in which solder and adhesive joining technologies will be compared

EST-SMT scenario

The purpose of this scenario is to model and assess state-of-the-art technology for the most obvious substitution application (SMT). The following assumptions characterises this scenario

	only SMT is considered
	it is assumed that adhesive components are terminated with AgPd

Important features of this scenario are that:

	literature data are used to estimate metal amounts used for components termination and PCB surfaces
	it is assumed that the adhesive amount by weight contributes 1/5 of the solder paste amount

The purpose of this scenario is to model and assess the specific Danfoss case. The following assumptions characterises this scenario:

	Substitution of both SMT and wave soldering with SMT adhesive technology
	Component termination is assumed not to differ

Because of these assumptions, care should be taken in interpreting the results for general purposes

In this scenario the measured adhesive amounts are used.

Disposal will be modelled as recycling, which is required by law in Denmark. It is also believed soon to be demanded in the European Union. Environmental impacts connected to other parts of the life cycle are assumed to be representative for regions with medium to good standards for production and emission handling. The results are therefore altogether assumed to be representative for regions with electronics recycling.

7.1.2 Functional unit

The function to be delivered by the alternatives is:

Possibility for regulation of the temperature in the entire life time of a refrigerator

The two alternatives are assumed not to differ in terms of performance, incl. energy consumption in the use phase and they are both expected to last the entire refrigerator life; i.e. the alternatives can be compared one to one.

7.1.3 Process tree, data quality and limitation of the life cycles

A rough presentation of the life cycles for the two alternatives is shown in Figure 7.1 and 7.2.

It is important to notice that the LCA aims at comparing the two alternatives. Therefore, identical life cycle phases for the two alternatives will not be considered. For instance, only the thermostat print is considered. The remaining parts are excluded. All inventory data are therefore for one print. Further, as described under ‘functional unit’, it is expected that the performance of the two alternatives in the use phase do not differ. The use phase is therefore excluded.

It was aimed at getting up-to-date on environmental input/outputs (resource consumption, emissions and waste generation) for all included processes. However, it showed up that most data are either non-existing or not available. Quite a few assumptions and estimations have therefore been made during the inventory analysis. Data limitations and general principles used for data collection/estimation in each life cycle phase are discussed below.

Figure 7.1: Life cycle for EST-sold

Figure 7.2: Life cycle for EST-adh

7.1.3.1 Printed circuit boards (PCBs)

The same basic substrate, FR-4 (epoxy/fibre glass), has been used for the two products. However, the PCBs differ in terms of surface treatment, where PCB for EST-sold is surface finished with tin/lead Hot Air Levelling (HAL), whereas the EST-adh PCB is finished with a nickel/gold surface applied chemically. EST-sold is further applied with an acrylate solder mask. The surface areas of the PCBs are identical.

7.1.3.2 Components

EST-SMT

It is assumed that traditional components with a top layer PbSn termination are used for EST-sold, whereas it is assumed that adhesive components are terminated with AgPd. Literature data have been used to estimate the amounts used for termination.

EST-Danfoss

Components in the test cases are identical except two – the electrolytic capacitor and the potentiometer. The two components used for the EST-adh are mounted by SMT, whereas the equivalent EST-sold components are "through-hole" mounted requiring an extra wave soldering process step^4.

Data on material content for the electrolytic capacitor has been supplied to Danfoss, but the data is confidential for others and can therefore not be included in this study, whereas the material composition of the potentiometers are available.

Energy consumption and other environmental input/outputs connected to the actual processing of the components have not been obtained from the suppliers.

As it has only been possible to obtain data for difference in composition for one of the components and because this is not considered representative for substitution of through-hole components for soldering with functionally equivalent SMT components for adhesive application, these differences will not be included.

Two components differ between EST-adh and EST-sold in the Danfoss scenario. The differences may be of importance but are NOT included due to lack of data.

7.1.3.3 Interconnection

Adhesive

The adhesive applied - based on epoxy and silver - is representative for state-of-the-art adhesives.

Solder

Solder paste has been used for SMT mounting and solder bars were used for the wave soldering process. The products applied are considered representative for state-of-the-art.

Metal mining

Extraction of metals is a very complex issue. Usually several metals are mined in the same mine with one metal being the reason why the mine is running. The same metal may in some mines be the major metal and in other mines a by-product. It is therefore not straightforward to allocate environmental burdens from mining to the different metals.

7.1.3.4 Production and use

Data from the Danfoss case is used to emulate the production. The two alternatives are assumed to deliver the same function, incl. having similar energy consumptions and life span (the entire refrigerator life). Environmental input/outputs in the use phase are therefore considered not to differ and have therefore not been included.

7.1.3.5 Disposal

Recycling of electronics is a demand in Denmark (regulated by law) as it is in a number of other European countries. It is expected to be a demand in the entire European Union in the near future.

The disposal phase is therefore modelled as a recycling scenario. A description of the processes can be found in other parts of the Kamille project [9].

7.1.3.6 Transport

Transport is considered not to vary significantly between the two alternatives as their weight and volume are approx. equal.

7.1.4 Assessment method

The EDIP-method [24] is used. EDIP applies the ‘environmental theme’ approach, where inputs/outputs are assessed in relation to their impact on:

	Resource availability
	Global warming
	Acidification
	Nutrient enrichment
	Photochemical ozone formation (photosmog)
	Bulk waste
	Hazardous waste
	Slag and ash
	Human toxicity via air
	Eco-toxicity, acute
	Persistent toxicity (covering human toxicity via water and soil, eco-toxicity via soil and chronic eco-toxicity via water)

Especially for toxicity it should be mentioned that quite a few inventory data are missing. It is assessed that estimation of these would be very uncertain as is the toxicity evaluation method by itself. Toxicity will therefore be discussed qualitatively on top of the quantified values.

Further, some toxicity assessment factors are lacking, which may especially affect the result for the metals. Toxicity factors for lead, nickel and silver (only human toxicity for the latter) can be found in [24]. Ecotoxicity factors for silver have mistakenly not been included, but have been obtained from the authors. Toxocity factors for tin, gold and palladium have not been included as they are assessed considerably less toxic than lead for human toxicity and silver for eco-toxicity.

It has not been possible to obtain specific information on occupational health for processes outside Danfoss. Therefore occupational health has only been included detailed for the production at Danfoss, for which an occupational health assessment has been conducted [4]. Occupational health for other parts of the life cycle will be discussed qualitatively. Only chemical exposures will be considered.

When interpreting the results, it should be remembered that the LCA only includes processes with possible differences between the alternatives. The assessment is therefore conducted for the difference between adhesive and solder technologies.

7.1.5 Irregularities and accidents

Irregularities and accidents are not included, as these are not assumed to influence the results significantly. This assumption requires that the production volumes are not too small. In that situation, start-up and close-down routines may substantially influence the environmental input/outputs per functional unit.

7.1.6 Capital goods

SMT-scenario

The same equipment can be used for SMT solder and adhesive joining [7]. Capital goods are therefore excluded in this scenario.

Danfoss scenario

The solder alternative requires wave soldering equipment, which the adhesive alternative does not. However, due to the high through-put, environmental impacts from capital good manufacturing is considered negligible and is therefore not included.

7.1.7 Allocation

Raw material phase

Allocation is often implicit in data obtained from raw material suppliers. Except for mining, the allocation principle used is believed to be co-product allocation by mass, where e.g. the APME⁵-data on epoxy production [14] is a good illustration.

As already described, allocation is very difficult when considering mining, as several metals are usually mined at the same location. The data applied are estimated based on literature studies assuming ‘pure’ mining of the considered metal. Palladium, however, is assumed mined in connection with nickel, platinum and rhodium and an allocation based on market prices has been applied.

Recycling

Recovered metals are subtracted from the inventory by system expansion and reflects thereby an ‘avoided production’ allocation.

7.1.8 Critical review

A critical review was carried out by Institute for Product Development, Lyngby, Denmark. The first part being a discussion of the goal and scope and data sources and the second a two step review of the final report.

7.2 Results – Resources

The weighting step for resources in the EDIP methodology assesses resources by comparing consumption with known global reserves per world citizen [24]. The unit for this weighting is (milli) Person Reserves (mPR).

7.2.1 SMT scenario

Figure 7.3 shows the resource weighting for the SMT scenario. The figure present differences between the alternatives (EST-adh minus EST-sold). E.g. it can be seen that more silver is used in the adhesive alternative. Thus, it does not mean that no silver is used in the solder alternative.

Further, metal columns represent differences in net consumption, i.e. overall metal consumption minus metals recovered during recycling.

Figure 7.3: Resource weighting (EST-adh minus EST-sold) – SMT scenario

Metal consumption dominates by far as compared to energy resources– by three or more orders of magnitude. The assumption that the energy requirements for metal mining/refining are delivered by oil combustion does therefore not affect the resource evaluation significantly.

The dominating among the metals are silver, tin and palladium. Gold and lead are less important and nickel is negligible.

For silver, about 1/3 of the net consumption is due to the silver loss during production (6-7% loss was estimated for start-up and lost by equipment cleaning), whereas the remaining 2/3 represents loss during recycling. For tin and lead, about 15% of the column heights are the result of solder paste loss (about 5% loss was estimated). Losses during production are considered worst case and would be reduced in case of larger production series.

The figure indicates, the adhesive alternative to be worse (sum of column heights for the adhesive alternative is largest). However, there are several uncertainties connected to the assessment:

	Metal losses during joining material, component and PCB manufacturing have not been included. These losses – if not recycled/recovered – may seriously affect the result. A 10% palladium or gold loss may increase net consumption (and thereby column heights in Figure 7.3) by a factor five. Silver loss during SMT production is included (see above), but a 10% loss during adhesive and component manufacturing would increase the overall net consumption by about 2/3. For tin and lead, a 10% loss would approx. increase the net consumption by 50%. It must be assumed that the 10% are worst case estimates, especially for the costly precious metals.
	EDIP weighing factor for silver. The weighting factors are based on an assumed amount of known global resources. As can be seen in other parts of the Kamille project [3], the estimate on global silver reserves used in EDIP (770.000 tonnnes) differs from other estimates obtained from Boliden (250.000 tonnes) and the Silver Institute (280.000 tonnes). If the alternative values are used, the silver column in Figure 7.3 would be approx. 3 times higher. Tin and lead reserves were also considered but for these metals, the references did not differ significantly from the EDIP value.
	The ‘net consumption’ is heavily depending on the recycling percentages. For silver, the recycling facility has informed that ‘more than 90% are recovered’. A default value of 95% has been used in the calculations. If the right percentage is 99%, the silver column in Figure 7.3 would have been about 1/3 in height, but if the right percentage is 90%, it would be about 60-70% higher. 80% recovery has been assumed for tin and lead - according to the recycling facility. The height of these columns will not change dramatically if this recovery percentage varies within 5%, but a considerably higher recycling percentage would decrease the Sn-column significantly. For palladium and gold, it was stated that close to 100% is recovered. 98% has been used as default. The Pd and gold columns are therefore much likely not underestimated. Possible higher recycling percentages would reduce the column heights. Altogether, especially the precious metals are sensitive towards fluctuations in recovery percentages. However, the major uncertainty is connected to metal recovery efficiency if other recycling facilities were considered, especially if one of the metals is not recovered at all. Altogether, the overall result may be affected in both directions by variations in recycling percentages.
	Joining material consumption. Data on solder paste are measured with high accuracy and therefore assumed accurate within a few percentages, whereas the adhesive consumption is estimated. The estimation is based on the assumption that by weight about 1/5 of the solder paste amount is needed. This is done according to own estimations and information from several other sources. The adhesive consumption is therefore considered pretty reliable (for optimised adhesive technology).
	Metal consumption for component termination. It has been very difficult to get reliable data on metal consumption for component termination. Further, termination amounts differ a lot between component types. The silver consumption for component termination is thus far more uncertain as compared to the silver consumption for adhesive. Further, silver for component termination accounts for about 69% of the silver difference between the alternatives. For palladium, which estimate is also very uncertain, it account for the entire difference. Altogether, it is assumed that the palladium figures (and thereby directly column height) may vary by a factor five. Silver figures may vary by up to a factor three and affect the silver column height by about a factor two. Tin and lead figures for component termination have been estimated based on an assumed distribution between solder material and component termination found in several references. These consumptions are assessed far more reliable than the precious metal consumptions.
	Metal consumption for PCB’s. As can be seen from Figure 7.3 the assessment is far more sensitive to the tin consumption in the soldered alternative than to the gold and nickel consumption in the adhesive alternative. The tin (and lead) consumption has been estimated based on an assumed distribution between solder material and PCB Hot Air Surface Levelling found in literature references. On average it is therefore considered pretty reliable. Uncertainties in PCB consumption figures are therefore assessed not to affect the overall result considerably. However, for special PCB designs (with a low packaging densities), the tin consumption may be underestimated.

Summary – SMT-scenario - Resources

Consumption of metal resources by far dominate over consumption of energy resources. Especially silver, tin and to some extent palladium are of importance.

Altogether, the results are too imprecise to show a significant difference between the alternatives, though the results indicate a more severe consumption of resources in the adhesive alternative.

The major uncertainties are:

	EDIP weighing factor for silver (may increase weighting factor by about a factor three)
	loss of precious metals in early parts of the life cycle (component, adhesive and PCB manufacturing) may increase net palladium consumption by up to a factor five, silver consumption by about 2/3 and tin consumption by about 50%
	palladium and silver content in component terminations (for Pd the uncertainty and thereby net consumption is assessed to be a factor 5, whereas for silver the uncertainty is assessed to be up to a factor three affected the consumption by about a factor two)
	recovery percentages, especially for silver (may affect silver consumption by about 2/3)
	other recycling technologies than the assumed may dramatically affect the result, especially if one of the metals is not recycled at all

7.2.2 Danfoss scenario

This scenario differs from the SMT scenario with respect to the following parameters: inclusion of the wave soldering step for EST-sold and the equivalent EST-adh processes measured/scaled adhesive consumptions is used component termination is not assumed to differ (SnPb-termination was also used for EST-adh)

The result of the resource assessment can be seen in Figure 7.4.

Figure 7.4: Resource weighting (EST-adh minus EST-sold) – Danfoss scenario

As for the SMT scenario, the metals and in particular tin and silver are dominating. For silver, about 45% of the column height is due to losses during manufacturing (the remaining losses during recycling), whereas only about 3% of tin column represent losses during the production phase. The losses may be decreased in case of larger production series.

Based on the figure, the solder alternative seems to be worse.

However, as for the SMT scenario, there are some major uncertainties. Especially the EDIP weighting factor for silver and metal recovery efficiencies (e.g. other recycling facilities) may affect the assessment.

Losses of silver during adhesive manufacturing (assuming 10% loss) would increase the overall silver consumption by about 50% (doubling the part of the silver column which is not represented by loss during production). Losses of tin during PCB and solder manufacturing (assuming 10%) would increase the net consumption by about 50%.

Another uncertainty of importance for this scenario is the metal consumption for Hot Air Levelling PCB surface. The amount, which contributes by 44% of the tin difference, is estimated based on scaling the weight and assumption of layer thickness. The uncertainty is assumed to be in the range of a factor two.

Summary – Danfoss-scenario - Resources

Consumption of metal resources by far dominate over consumption of energy resources. Tin and silver are dominating in between the metals.

Altogether, the results are too imprecise to show a significant difference between the alternatives, though the results indicate a more severe consumption of resources in the solder alternative.

The major uncertainties are:

	EDIP weighing factor for silver (may increase weighting by a factor three)
	Recovery percentages, especially if other recycling technologies (not recycling one or more of the metals) were applied
	Loss of silver and tin in the raw material phase (may worst case increase the net consumption by 50%)

Care should be taken not to generalize the results of the Danfoss scenario, because:

	The adhesive consumption for fastening the relay was very high. That type of component would most obviously not be fastened by adhesive in a ‘real production case’. Therefore the adhesive consumption for a substitution of wave soldering with an adhesive process would most obviously result in considerably less adhesive consumption – reducing the silver column in Figure 7.4.
	It is highly questionable whether full scale adhesive production would be carried out with SnPb-terminated components. Application of AgPd-terminated components would increase the consumption of precious metals for the adhesive alternative.
	Differences between through-hole components and equivalent SMT components have not been investigated. This may affect the results considerably.
	The applied PCB may not be representative as the packaging density may vary considerably.

In other parts of the project, it has been assessed whether a severe increase in adhesive application will considerably affect the global silver consumption and thereby potentially silver prices and the scarcity of silver [3].

7.3 Results - External environment

According to EDIP the following environmental parameters have been included:

	Global warming
	Acidification
	Nutrient enrichment
	Photochemical ozone formation (photosmog)
	Bulk waste
	Hazardous waste
	Slag and ash
	Human toxicity via air
	Eco-toxicity, acute
	Persistent toxicity6

The following figures will show the weighted results. The weighting principle in EDIP is based on a comparison with politically set target for pollution prevention (see [24] for further explanation). The unit for this weighting is (milli) Person Equivalent Target (mPET).

The weighting method is rather subjective and although, the included environmental impact effects have the same unit, they should not directly be added between effect categories, but can be used as an indication of ‘what is big and what is small’.

As pointed out earlier, the assessment methodology for toxicity is rather uncertain and results should therefore be taken with caution.

Disposal of the hazardous waste is modelled by two scenarios. The first scenario assumes the metal waste to be an amount of hazardous waste (named HW = Hazardous waste) and the second assumes that the entire metal content – with time – will leak to the environment surrounding the waste disposal site (named TH = Time Horizon). The "true" situation can be assumed to be somewhere in between these two extremes.

7.3.1 SMT-scenario (disposal scenario: HW)

Figure 7.5 shows the weighted result comparing the two alternatives (EST-adh minus EST-sold).

Figure 7.5: Weighted result (EST-adh minus EST-sold) – SMT scenario (HW)

The figure indicates the adhesive alternative to be better for all impact categories. However, the results are uncertain.

Energy related impact categories

Energy consumption is the only (or major contributor) to the impact categories global warming, acidification, nutrient enrichment, bulk waste as well as slag and ash. The same interpretation as in the energy analysis (see [2]) can therefore be made for these impact categories, i.e. differences between the alternatives are not significant.

Photosmog

Contributions to the effect category ‘Photosmog’ comes from Volatile Organic Compounds (VOC’s) emitted by combustion of fossil fuels or from manufacturing, handling or applying solvents. About 80% of the photosmog column comes from the difference in solvent emission in the production phase. Specific data on solvent emissions in the raw material phase have not been obtained. However, from the APME epoxy data, it can be seen that 5,8 g of hydrocarbons are emitted per kg epoxy. If these are assumed to contribute by an average photosmog potential (POCP=0,3) for solvent, it will only affect the result by 8%. Further, more solvent - and thereby much likely more emissions of solvents in the raw material phase – are used for the solder alternative. It is therefore assessed that the solder alternative is worse in relation to photosmog formation.

Hazardous waste

Most of the column representing hazardous waste in the figure results from the recycling step. It is larger for the soldering alternative due to the higher metal consumption in the solder alternative and due to the lower recycling percentages for tin and lead. Data on hazardous waste formation from several of the raw material processes have not been obtained. However, it is assumed to be higher for the solder alternative as larger amounts of materials have been used. It is therefore assessed that the solder alternative is worse regarding hazardous waste.

Toxicity

The toxicity parameters are difficult to interpret because there are uncertainties connected to the method. Among others, LCA toxicity potentials does not reflect ‘real’ but rather ‘potential’ impact. Whether an estimated impact will really cause harm, very much rely the local environment (background concentration, total emitted amount and sensitivity of the receiving environmental media) to which an emission takes place.

In this LCA case, it has further been very difficult to obtain emission data connected with mining/refining, raw material manufacturing and recycling.

However, if Figure 7.5 is taken as a starting point, it can be shown that emission of the solder paste solvent (for emission purposes modelled as diethylene glycol mono-n-butyl ether) contributes by about 87% and lead emission from mining by about 7% of the difference (i.e. Figure 7.5 column heights) between the alternatives for the impact category ‘human toxicity’. The remaining difference in toxicity comes from energy production.

Due to the relatively low total amounts of solder paste solvent emitted from an SMT-line, it is questionable whether the emission of solder paste solvent will cause an impact. The same consideration can be made for the adhesive solvent. The adhesive solvent (which is confidential) has not been assigned a toxicity factor. Its main toxicity is irritation by inhalation, but with the emission amounts possible it will definitely not be able to cause any harm to humans in the environment surrounding an SMT plant.

Several toxic emissions are not included in the inventory data. Toxic emissiones may take place during manufacturing of epoxy, solvents, PCBs, components and joining materials, and during manufacturing of input chemicals for these processes. Toxicity impacts from these phases may severely affect the overall evaluation. However, as most solvents and most reactive polymers (acrylate for the solder PCB) is used in the solder alternative, emission from these productions will most likely increase the human toxic load for the solder alternative more than for the adhesive alternative. Further, because lead is by far the most human toxic of the metals included, air emissions of metals (in the raw material phase and during recycling) will disfavour the solder alternative. Altogether, the solder alternative must be assumed to be the worse with respect to human toxicity.

Persistent toxicity is difficult to interpret as it covers several toxic impact categories (see introduction to this section). In Figure 7.5 about 75% of the column height results from Pb air emission during mining. It is not possible to judge whether emissions not inventoried will favour the one or the other alternative with respect to persistent toxicity.

Ecotoxicity, can hardly be seen on the figure. However, possible silver emission to water (silver is by far the more ecotoxic metal) will severely affect the ecotoxic load for the adhesive alternative.

A sensitivity analysis was therefore carried out assuming that 1 g of metal is emitted to water for every 1 kg of virgin metal production (assumed for all metals). In that case, the assessment will turn out as depicted in Figure 7.6.

As can be seen, the silver ecotoxicity (acute water) disfavours the adhesive alternative. Further, it can be seen that the difference in persistent toxicity has now decreased slightly, indicating that chronic toxiocity of silver via water affects this parameter.

Silver emissions may also take place during component and adhesive manufacturing as well as in connection with recycling. That would further disfavour the adhesive alternative.

A sensitivity analysis was also carried out assuming 0,65 g metal air emission per 1 kg metal production (this figures was already included in the lead data). This assumption does not affect the results significantly.

Summary – SMT (HW)-scenario – External environment

The immediate result indicate, the adhesive alternative to be better for the impact categories: global warming, acidification, nutrient enrichment, bulk waste as well as slag and ash. However, as these impact categories heavily or entirely rely on the energy consumption, the result is not significant.

Photosmog formation (due to more emissions of solvents), hazardous waste formation (more recycle waste and more assumed waste in the raw material phase) and probably human toxicity (mainly due to lead emission and to some degree emission of solder paste solvent) are assumed worse for the solder alternative.

Ecotoxicity is assessed worse for the adhesive alternative (due to possible silver emissions to the water environment).

It cannot be judged for which alternative ‘Persistent toxicity’ is better or worse.

The toxicity aspect will be discussed further in the TH-scenario.

7.3.2 Danfoss scenario (HW)

Figure 7.7 shows the weighted result.

Figure 7.7: Weighted result (EST-adh minus EST-sold) – Danfoss scenario (HW)

The figure indicates, the adhesive alternative to be better for all impact categories.

Energy related impact categories

In this scenario, the impact categories associated with energy consumption; i.e. global warming, acidification, nutrient enrichment, bulk waste as well as ash and slag, are judged to be significantly better for the adhesive alternative (see energy analysis in [2]. For metal mining, where the energy system supplying the energy was not known, it has been assumed that the energy is delivered by oil combustion. It could be argued that different energy sources in between the metals would give incompatible emissions (energy supply systems differ to some extent in emissions), but by comparing with the energy analysis, the result must still be assumed to be significant.

Photosmog

For photosmog, the major contributor to the difference is the flux solvent of 2-propanol. The solder alternative is thus the worse for photosmog formation.

Hazardous waste

The hazardous waste fraction is even more significant in this scenario as compared to SMT (HW); also here because of the larger consumption and lower recovery percentages for tin and lead. The solder alternative is thus assessed worse with respect to hazardous waste.

Toxicity

According to Figure 7.7, human toxicity is worse for the solder alternative. About 30% of the difference is the result of emission of solder paste solvent, about 40% is lead emission form mining, whereas the remaining part results from energy consumption. Similarly to the SMT (HW) scenario, but even stronger here, human toxicity is assessed to be worse for the solder alternative.

About 75% of the difference in persistent toxicity results from emission of the flux solvent 2-propanol. Whether this emission will really cause any significant harm in the environment surrounding a production site is highly questionable. Altogether, as for the SMT (HW) scenario, it is difficult to point out the worse alternative according to persistent toxicity.

Figure 7.8 represent the situation where water emission of 1 g/kg virgin metal has been assumed.

Figure 7.8: Weighted result (EST-adh minus EST-sold) – Danfoss (HW) - assuming silver and lead emission to water during virgin metal manufacturing

As for the SMT (HW) scenario, possible silver emissions to water will turn the ecotoxicity parameter to disfavour the adhesive alternative.

Summary – Danfoss (HW)-scenario – External environment

The Danfoss (HW) scenario shows that the adhesive alternative is the better concerning the energy related impact categories global warming, acidification, nutrient enrichment, bulk waste, as well as slag and ash, and it is also assessed better concerning photosmog (flux solvent emission), hazardous waste (big loss of tin and lead during recycling) and toxicity (mainly due to lead emission, but also due to energy consumption and solder paste solvent emission). It is not clear whether the persistent toxicity parameter is significant, whereas the adhesive alternative becomes worse regarding ecotoxicity, when silver emissions to water may appear during the life cycle.

Care should be taken not to generalize the results of the Danfoss scenario, because:

	It is highly questionable whether full scale adhesive production would be carried out with SnPb-terminated components [7]. Application of AgPd-terminated components would increase the consumption of precious metals for the adhesive alternative.
	Differences between through-hole components and equivalent SMT components have not been investigated. This may affect the results considerably.
	The applied PCB may not be representative as the packaging density may vary considerably.

7.3.3 SMT scenario (TH)

The TH (Time Horizon) disposal scenario models the situation, where it is assumed that the not recovered metals with time will leak to the surrounding environment.

Figure 7.9 shows the weighted result assuming that the metals end up in the soil environment.

Figure 7.9: Weighted result (EST-adh minus EST-sold) – SMT scenario (TH-soil)Although care should be taken when comparing in between impact categories, there is no doubt that the impact category ‘persistent toxicity’ has become dominant.

This is because of the assumed silver eco-toxicity in soil.

Figure 7.10 shows the result where the metals are assumed to end up in the water environment.

Figure 7.10: Weighted result (EST-adh minus EST-sold) – SMT scenario (TH-water)

Now the ecotoxicity parameter has become the dominant and again it is the silver toxicity that triggers the result.

Persistent toxicity is still high due to the expected chronic toxicity of silver in water. However, when going in more details with the data behind the figure, it can be seen that lead is pulling in the other direction (by about 1/3 of the silver value!) due to the chronic eco-toxicity potential of lead in water.

7.3.4 Danfoss scenario (TH)

Figure 7.11 shows the weighted result when assuming the metals to end in the soil compartment.

Figure 7.11: Weighted result (EST-adh minus EST-sold) – Danfoss scenario (TH-soil)

As for the SMT scenario, it can be seen that the expected silver eco-toxicity in soil, heavily influence the assessment.

Figure 7.12 shows the weighting by assuming metal leaking to water

Figure 7.12: Weighted result (EST-adh minus EST-sold) – Danfoss scenario (TH-water)

As compared to the SMT scenario, the relative difference between lead as compared to silver emission has increased (the wave soldering step requires substantial lead amounts and silver content in components is not assumed to differ). Therefore, the persistent toxicity of lead now dominates over silver. Further, when looking into the figures behind the ecotoxicity column, it can be seen that silver contributes by about 1,16 mPET, whereas lead with about 0,90 mPET, i.e. in reality no significant difference. It is however clear that the ecotoxicity parameter is important for both alternatives, if the waste can be assumed to leak to water.

Summary on the SMT and Danfoss scenarios (TH) – external environment

It is clear from the above discussion that the potential risk connected with waste containing silver and lead is substantial as compared to other life cycle impacts. However, actual risk will depend on the bio-availability; i.e. the dose/exposure that can be found in the environment. Among other things (under most conditions), both metals are assumed to be bound firmly in soil layers. However, the environmental fate of the metals is uncertain and no definite conclusion can be drawn here. These issues are discussed in more detail in other parts of the Kamille project [5].

Especially the behaviour of silver is interesting, as silver’s eco-toxicity may heavily influence the conclusions. However, with today’s knowledge, it is difficult to state, whether silver leaking from (hazardous) waste sites may cause substantial risks. In other parts of the life cycle (e.g. during component and adhesive manufacturing), potential silver emissions (especially to the fresh water environment) may severely affect the result of the LCA.

Occupational health

Only chemical impacts will be considered as other occupational impacts like e.g. noise and physical impacts are estimated not to differ significantly between the alternatives.

No information was obtained about chemical exposures in the life cycle phases outside Danfoss. Considerations about impacts in these phases are therefore qualitative and only an indication of potential impacts can be outlined (see [2]). An occupational health assessment for production of the two alternatives has been carried out at Danfoss. The assessment can be found in chapter 6.

7.4  Conclusion and Recommendations

Based on the present LCA, it cannot be judged whether adhesive technology is better or worse than solder technology. The available data on material consumption and emissions during the life cycle are too uncertain. Further, in relation to toxicity, which is a crucial impact category in this assessment, too little is known about the actual environmental fate of metals leaking to the environment from waste steams.

More specific conclusions may be drawn in relation to some of the impact categories. These are described below along with description of the major uncertainties.

Resources

The assessment showed that the weighted metal values and especially silver, tin and palladium consumption dominates consumption of energy resources by several orders of magnitude.

However, it has not been possible to judge between the alternatives, because of the following major uncertainties:

	evaluation of the scarcity of silver (different references vary by a factor three)
	consumption of silver and palladium in component termination for adhesives is very uncertain
	uncertainty about recovery percentages during recycling – especially application of different recycling technologies than the one considered, may heavily affect the result
	metal losses during manufacturing processes have not been included, but may to some extent affect the evaluation

A very interesting finding was that the component termination may account for a larger difference in silver consumption than the adhesive itself.

External environment

Toxicity is assessed to be the potentially major impact category concerning external environment.

It has major influence on the result how big metal emissions to the environment are and how the environmental fate of metals in hazardous wastes is assessed. Especially, the inherent ecotoxicity of silver is crucial. However, the knowledge of silvers behaviour in the environment is at the moment too weak to really quantify the risks connected to (hazardous) waste disposal of silver containing waste. Also the fate of lead affects the assessment.

For both the SMT and Danfoss scenarios, it is assessed that the solder alternative potentially is the worse in relation to human toxicity and the adhesive alternative potentially in relation to exotoxicity. Human toxicity is mainly affected by lead emissions and possibly to some extent by solder paste solvent emission during production and ecotoxicity mainly by silver emissions.

Both for the SMT and Danfoss scenarios, photosmog and hazardous waste generation are assessed worse for the solder alternative. For the Danfoss alternative especially because of substitution of the flux solvent emission during wave soldering.

For the energy related impact categories (global warming, nutrient enrichment, acidification, bulk waste as well as ash and slag generation), it is not possible to judge between the alternatives with the available data for the SMT scenario. For the Danfoss scenario, the solder alternative is assessed significantly worse due to the energy requirement for the wave soldering process.

Finally, it should be stressed that care should be taken not to overinterpret results from the Danfoss scenario, mainly because it has a pretty special print layout and differences in components are not considered

Occupational health

An occupational health assessment at Danfoss showed that both alternatives can be produced with potential low risk, if proper care is taken. For other parts of the life cycle, it has not been possible to judge which alternative is preferable based on the available information.

Recommendations

The study has lead to the following recommendations in relation to manufacturing and emission handling:

	Use components with as little silver and palladium termination as possible
	Optimise adhesive consumption (for SMT about 1/5 as compared to solder paste amount)
	Assure high recovery percentages for the metals during recycling and reduce losses during manufacturing operations
	Assure proper treatment of hazardous wastes
	During production: avoid skin contact with solder paste, flux and adhesives and avoid inhalation of solvent vapours and lead fumes (see also more specific guideline in other parts of the project [4])

4 It might be confusing that three through-hole components are mentioned in the goal description but only two here. The third component is the relay, which is the same for the two alternatives. The relay is originally a through-hole component, but the component leads were bended manually to become a ‘SMT component’ for the adhesive alternative.

⁵APME: Association of Plastics Manufacturers in Europe.

6covering human toxicity via water and soil and eco-toxicity via soil and chronic eco-toxicity via water

8.Test of electrical conductive adhesives

8.1 Survey of electrical conductive adhesives tested
8.2 Survey of test substrates
8.3 Survey of the mounted components
8.3.1 Passive components on each substrate variant
8.3.2 Active components on each substrate variants
8.4 Survey of the samples (adhesive/substrate)
8.5 Survey of the mounting process
8.5.1 Stencil printing of adhesives/solder paste
8.5.2 Dispensing of adhesives
8.6 Component mounting
8.7 Curing/soldering
8.8  Cleaning
8.9 Test plan
8.10 Description of the tests
8.10.1 Adhesion test
8.11 Current - temperature
8.12  Discussion
8.13  Comments to each electrical conductive adhesive variant

The present test is a quality assessment of 13 electrical conductive adhesives from ten different manufacturers. The variants include adhesives based on epoxy and silicone with silver or carbon particles. Further adhesives with silver on aluminium particles and silver on copper particles were included. The adhesives have been tested on four different substrates; ceramic and polyester substrates with silver conductors and FR-4 substrates with gold over nickel and tin lead conductors. The components attached to the substrates had silver palladium and tin-lead finish on the terminals.

The test programme included environmental testing of the component to substrate connection. A lead-free solder paste was tested as a reference.

The interconnection resistance was tested before and after all environmental exposures, and the shear strength was tested initially and after the final exposure.

8.1 Survey of electrical conductive adhesives tested

The tested electrical conductive adhesives are shown in table 8.1 (The information is from the manufacturer's datasheet).

	6 electrical conductive adhesive variants, nos.1, 2, 3, 4, 5 and 6, are epoxy with silver particles, all 1-component except for variant no. 6 which is a 2-component adhesive.
	1 variant, no. 7, is 'thermoplastic' 1-component electrical conductive adhesive with silver.
	1 variant, no. 8, is 'thermoset' 1-component electrical conductive adhesive with silver.
	3 electrical conductive adhesive variants, nos. 9, 10, and 11, are silicone, variant no. 9 is a 2-component adhesive with silver, variant no.10 is a 2-component adhesive with silver-plated aluminium particles and variant no. 11 is a 1-component adhesive with silver-plated copper particles.
	Two 1-component variants have carbon particles, one with epoxy, variant no. 12 and one with 'polymer', variant no. 13.
	No further specification was found for the thermoset, the thermoplastic, and the polymer variants, variant nos. 7, 8, and 13 respectively.
	Variant no. 14 is a lead-free solder paste (Sn / 3.8 Ag / 0.7 Cu), and is tested as a reference.

8.2 Survey of test substrates

Four different substrate types were included in the investigation:

1. Alumina (thickfilm) substrate with thickfilm silver conductors (Manufactured by Grundfos A/S)
2. Polyester substrate with polymer silver (Ag) conductors (Manufactured by Mekoprint A/S)
3. FR-4 printed circuit board with gold over nickel (Au/Ni) finish, on copper conductors (Manufactured by Bent Hede Elektronik A/S)
4. FR-4 printed circuit board with tin lead (Sn/Pb) finish, on copper conductors (Manufactured by Bent Hede Elektronik A/S)

The same layout was used for all four substrate types.

Description of the different test pattern is given in section 8.10.

Se her!

8.3 Survey of the mounted components

Three different passive components and two different active components were mounted on the test substrates.

8.3.1 Passive components on each substrate variant

0603 0-W chip resistors for contact resistance measurements:

	23 resistors with Sn/Pb terminal finish (21 for the ceramic substrate)
	23 resistors with Ag/Pd terminal finish (21 for the ceramic substrate)

Distance between terminals: Sn/Pb 0.95 mm and Ag/Pd 1.10 mm.

1206 0-W resistors for the shear test:

	16 resistors with Sn/Pb terminal finish, shear test was performed on 7 resistors initially and 7 resistors after the final environmental exposure.
	16 resistors with Ag/Pd terminal finish, shear test was performed on 7 resistors initially and 7 resistors after the final environmental exposure.

Distance between terminals: Sn/Pb 2.45 mm and Ag/Pd 2.50 mm.

3 surface mount electrolyte capacitors with tin lead finish, for shear test.

8.3.2 Active components on each substrate variants

	samples LQFP, 44 leads Plastic low profile quad flat package (Pitch: 0.8 mm, with smallest insulation distance of 0.45 mm between terminals. Terminal plating: Sn/Pb).
	4 samples SO 28 Plastic small outline with 2 times 14 leads (Pitch: 1.27 mm, with smallest insulation distance of 0.80 mm between terminals. Terminal plating: Sn/Pb).

Both types for visual inspection of the interconnections between the leads and the substrate finish, flow-out or reduced insulation distance, short circuit and colour change.

8.4 Survey of the samples (adhesive/substrate)

Table 8.2 shows on what substrate, the electrical conductive adhesive variants have been tested.

	3 variants and the solder reference were tested on ceramic substrate.
	5 variants were tested on polyester substrate.
	8 variants and the solder reference were tested on FR-4 with gold over nickel finish.
	8 variants and the solder reference were tested on FR-4 with tin lead finish.

These gives 24 electrical conductive adhesive/substrate samples and 3 solder reference samples.
Passive and active components were mounted on each of the 24 + 3 test samples.
The solder reference variant no. 14 was not tested on the polyester substrate, as the polyester could not withstand the solder temperature. Therefore, there is no reference for the polyester substrate.
Variant no. 2, Ablebond, and variant no. 3, Namics, were tested on ceramic and both FR-4 substrates, and variant no. 4, Loctite (3880), was tested on polyester and both FR-4 substrates. The other variants were either tested on the ceramic, the polyester or on the 2 FR-4 substrates.

8.5 Survey of the mounting process

The mounting process was done in the same way for all variants although variants, which were not suitable for stencil printing due to low viscosity, were dispensed.

8.5.1 Stencil printing of adhesives/solder paste

A manual stencil printer with stainless steel squeegees (100 mm stencil) was used for contact printing (i.e. the stencil is placed on the substrate during the process). The electrical conductive adhesives/solder paste was applied on the stencil and with the squeegee. The electrical conductive adhesive/solder paste was screened by hand through the opens in the stencil. The opens in the stencil in percentage of the footprint were 100%.

8.5.2 Dispensing of adhesives

A KEPRO 210B dispenser with 0.4 mm needle (1.7 bar pressure) was used. The electrical conductive adhesives were applied onto the test boards with a handheld dispenser activated by a footswitch.

8.6 Component mounting

The components were mounted by means of a KEPRO 210B vacuum tweezer. All components were mounted in the electrical conductive adhesive/solder paste by the handheld vacuum tweezer. Due to the hand-mounting, the mounting force could not be identical from component to component, which may give greater variations in the contact resistance measurements and the shear test for each variant as if the components were machine mounted. The printing, dispensing, and the mounting procedure were performed by the same person for all the variants.

8.7 Curing / soldering

A box oven with air circulation was used for the curing. All variants were cured just after the mounting process, according to the manufacturer's cure times (compare with table 8.1).

8.8 Cleaning

Cleaning was made by means of 2-propanol Acetone and compressed air. The screen and squeegee were wiped off with 2-propanol followed by Acetone applied on absorption paper. Finally, the opens in the stencil were blown clean with compressed air. The cleanliness of the stencil was inspected in microscope after each cleaning process to assure that no forcing materials are present before applying a new variant on the stencil.

8.9 Test plan

The test plan is shown on Figure 8.1.

Figure 8.1 KAMILLE test plan.

8.10 Description of the tests

8.10.1 Adhesion test

The intention was to make the adhesion test with a mandrel; this test however requires a hole beneath the component in order to push the mandrel through. This is unlikely to perform on production boards (substrates). Therefore the values from a mandrel test will be stand-alone values. To be able to compare the adhesion test performed in this project to what is possible on production boards, it was decided to perform the adhesion test as a shear strength test.

Adhesion test was performed as a shear strength test, like MIL-STD 883C method 2019.2 of 1206 resistors with both Ag/Pd and Sn/Pb component finish.

Seven resistors of each terminal finish were shear tested initially and after the H₂S exposure.

The test substrate layout is shown on Figure 8.2 with letters related to each test pattern described below.

Figure 8.2 Test patterns on test substrate.

Test pattern A: Left side of the substrate

Mounting of 23 0-W 0603 resistors with tin lead on nickel barrier finish for 4terminal contact resistance measurement.

Measurement through 1) Substrate conductor – 2) Adhesive – 3) Component terminal – 4) Component – 5) Component terminal – 6) Adhesive – 7) Substrate conductor, where the resistance of the substrate conductor, component terminal, component, and interface between component terminal and component are assumed to be constant throughout the test programme leaving the adhesive and its interface to the substrate and to the component terminal to vary.

Test pattern B: Right side of the substrate

Mounting of 23 0-W 0603 resistors with silver palladium finish for 4-terminal contact resistance measurement.
Measurement through 1) Substrate conductor – 2) Adhesive – 3) Component terminal – 4) Component – 5) Component terminal – 6) Adhesive – 7) Substrate conductor, where the resistance of the substrate conductor, component terminal, component and interface between component terminal and component are assumed to be constant throughout the test programme leaving the adhesive and its interface to the substrate and to the component terminal to vary.

Test pattern C: Middle/top of the substrate

Mounting of 16 0-W 1206 resistors with silver palladium finish and 16 0-W 1206 resistors with tin/lead on nickel barrier finish for shear test initially and after final the environmental exposure.
The footprint could not be used for shear test of soldered components, as there was no space for a solder meniscus on the footprint.

Test pattern D: Footprint for surface mounting of 3 electrolyte capacitors (from the Danfoss A/S case)

The electrical conductive adhesive/solder joint was visually inspected initially and after test exposure for degradation, colour change, cracks, etc.

Test pattern E: Footprint for 4 SOT319-1, Plastic quad flat package (QFP44)

Pitch: 1.0 mm.
The component was not available from the vendor. The printing on the footprint was visually inspected for printing quality, short circuit, degradation, colour change, cracks, etc.

Test pattern F: Footprint for 4 SOT389-1, low profile quad flat package (LQFP44)

Pitch: 0.8 mm.
The electrical conductive adhesive/solder joint was visually inspected initially and after test exposure for short circuit, degradation, colour change, cracks, etc.

Test pattern G: Footprint for 4 SOT136-1, plastic small outline package (SO28)

Pitch: 1.27 mm.
The electrical conductive adhesive/solder joint was visually inspected initially and after test exposure, for short circuit, degradation, colour change, cracks, etc.

Test pattern H: Middle/bottom of the substrate

Electrical conductive adhesive/solder paste was printed on and between the footprint making a short circuit with the adhesive/solder paste.
5 of the short circuit pattern was 4-terminal contact resistance measured before and after each environmental exposures.
After the final contact resistance measurements one of the short circuit pattern was current/temperature tested.

Test pattern for insulation resistance/silver migration

Insulation resistance / silver migration of electrical conductive adhesives has been investigated in a previous report [27]. The result from this report stated that: No migration or indication of migration could be identified using microscope up to 100x enlargement.

This was in general the result for all adhesive variants. The no-migration result was confirmed by the insulation resistance measurements.

The no-migration results are, however, valid as long as the adhesive media encapsulates the silver particles.

Migration of the silver conductors on both the ceramics and the polyester substrate is beyond this project and has therefore not been investigated.

An investigation of silver migration of min. 10 m m electroplated silver on copper conductors is found in [28].

8.11 Current - temperature

The results after the current/temperature test are shown in table 8.2.

To have adhesive joint heated up to 105°C in equipment will not be suitable for most applications, and will reduce the reliability of the whole equipment if not cooled. Therefore, the electrical conductive adhesives must be used with low currents and cannot be recommended for power use without cooling.

In table 8.2 it is seen that highest current values are found for the ceramic substrates, this was expected due to the temperature coefficient of ceramic substrates.

Se her!

8.12 Discussion

A very important parameter seems to be the component mounting pressure. A too high mounting pressure will tend the adhesive to flow-out, which may result in short circuit after the curing. On the other hand a too low mounting pressure may cause very high contact resistance of the interconnection.

Therefore, process optimising with electrical conductive adhesive may be a hard task.

The stencil printing was performed by hand as contact print with a stainless steel stencil. This may not be the most optimum process for large-scale production.

The use of snap-off printing with a polymer stencil or screen may give better printing results (recommended by Loctite). This has not been tested in this investigation, but should be considered.

Some flow-outs of the adhesives have been observed. This flow-out can be avoided by process optimising, either by reducing the stencil thickness or the printing area on the footprint. Reducing the stencil thickness will, however, result in a faster worn down of the stencil. Reducing the opens in the stencil may be a problem for the fine pitch components due to the distribution of the conductive particles in the adhesive during the printing process. With a sufficiently stirred adhesive with sufficiently small conductive particles this should not be a problem.

Regarding solder paste, it is even more important that the viscosity of the adhesives is constant over time and from lot to lot, as the self-alignment and the contraction effect are not present for the electrical conductive adhesives like it is for solder paste. The environmental humidity, which is an important parameter for solder paste, may not be as critical for the electrical conductive adhesive. This has, however, not been investigated. For the adhesive it is important that the adhesive will not create skin, especially from the printing process to the mounting process.

The visual inspection after the H₂S exposure has shown a colour change for all electrical conductive adhesives with silver and for the silver palladium component terminals. The colour has become dark or black; this is what could be expected for silver in hydrogen sulphide environment. This colour change may be expected over time for silver, as hydrogene sulphide is in the atmosphere.

Contact resistance for a component to substrate interconnection may depend on the circuit. Normally low contact resistance is required, but for components with relatively high input impedance (LCD display) a high resistance may be accepted.

Comparing the contact resistance values from the short circuit pattern with the values from the component mounted pattern, there is, in general, seen a somewhat lower standard deviation for the short circuit pattern.

It was assumed that this was due to the handmounting process, however if hand-mounting has caused the relative high standard deviation, handmounting of components in electrical conductive adhesives may be a serious problem. As a result of this, prototypes or engineering models with electrical conductive adhesives may first be manufactured after optimising the mounting process.

Single measured contact resistance and samples with no failure throughout the test sequence were observed.

The high standard deviation can therefore also be assumed to be due to an inhomogeneous distribution of the conductive particles in the adhesive after the curing process or due to different oxidation degrees of the surfaces to be glued.

The mounting process is assumed as follows: When the component is pressed into the adhesive the conductive particles will be pressed down as well, and stay there due to the gravity. This may result in less conductive particles on the top of the joint where the component is attached which may result in higher contact resistance. A high conductive particle density will of course minimise this effect.

The 2 kg and above limit for the shear strength test is estimated as an acceptable shear strength value for a 1206 chip resistor. The shear strength test is, as the contact resistance possible dependent on the hand-mounting process, and it will also be a question of mounting force contra flow-out of the adhesive.

In order to avoid the effect of shear strength of the electrical conductive adhesive, the components can be glued to the substrate with a non-conductive adhesive with high shear strength force value. This may, however, add an extra process to the production flow, if a dispenser for 2 adhesives is not used simultaneously.

The initial minimum shear force value may be due to the hand-mounting process, but may indicate that process optimising is necessary for a variant with minimum shear strength value below 2 kg.

For both contact resistance and shear strength the thickness of the adhesive joint is important, and the process optimising shall be related to this. Normally, a thin joint is preferred.

The curing time is important as well, if the manufacturer gives minimum times for the curing, due to the total processing time, and insufficient curing can result, this may give at higher contact resistance and a lower shear strength values.

Initially, the contact resistance was very high for some of the variants, but after the temperature cycling the contact resistance had decreased to 'normal'.

Due to the possibility of silver migration, it may be necessary to conformal coat circuits connected with electrical conductive adhesives. The reaction between a electrical conductive adhesive and a conformal coating must be reliability tested before production.

For the current/temperature test it shall be noted that the highest current values were found on the ceramic substrates, which was expected due to the temperature coefficient of the ceramic substrate. Power circuits with components connected with electrical conductive adhesives on FR-4 are not recommended, without cooling.

8.13 Comments to each electrical conductive adhesive variant

Comments to each electrical conductive adhesive variant are given below.
The conclusion below does not compare the actual measured contact resistance values, but only the increase of contact resistance from the initial value. This may give some variants a bad assessment even if the contact resistance is lower as a variant which have shown good results.

Variant no. 1, Amino CE 3511, tested on ceramic substrate with silver conductors

The adhesive has shown good results after all environmental exposures as used as a short circuit and to connect 0603 resistors with silver palladium finish. To connect the 0603 resistor with tin lead finish the results were acceptable. However, it is seen from the calculated mean value that for the tin lead finish the resistance has been lower and the standard deviation higher compared to silver palladium finish.
Shear strength values were acceptable for both component finishes.

Variant no. 2, Ablebond 8175A, tested on ceramic substrate with silver conductors and FR-4 substrate with gold over nickel conductors and tin lead conductors

For the ceramic substrate the adhesive has shown good results after all environmental exposures as used as a short circuit and to connect 0603 resistors with silver palladium finish. To connect the 0603 resistor with tin lead finish the results were acceptable after temperature cycling but not after the humidity and H₂S exposure. A greater fall in contact resistance was measured after temperature cycling for both 0603 resistor types.
For the FR-4 substrate with gold over nickel plating the adhesive has shown good results after all environmental exposure as used as a short circuit and to connect 0603 resistors with silver palladium finish. To connect the 0603 resistor with tin lead finish the results were unacceptable.

For the FR-4 substrate with tin lead plating the adhesive has shown acceptable results after all environmental exposure as used as a short circuit. To connect 0603 resistors with silver palladium and with tin lead finish the results were not acceptable.

Shear strength values were acceptable for both component finishes on ceramic and FR-4 substrate with gold over nickel finish. Other values were unacceptable.

Variant no. 3, Namics XH9626, tested on ceramic substrate with silver conductors and FR-4 substrate with gold over nickel conductors and tin lead conductors

For the ceramic substrate the adhesive has shown good results after all environmental exposures as used as a short circuit and to connect 0603 resistors with silver palladium finish. To connect the 0603 resistor with tin lead finish the results were unacceptable.For the FR-4 substrate with gold over nickel plating the adhesive has shown good results after all environmental exposure as used as a short circuit and to connect 0603 resistors with silver palladium finish. To connect the 0603 resistor with tin lead finish the results were unacceptable.
For the FR-4 substrate with tin lead plating the adhesive has shown acceptable results after all environmental exposures as used as a short circuit. To connect 0603 resistors with silver palladium and with tin lead finish the results were unacceptable.
Shear strength values were acceptable for both component finishes on all substrate types.

Variant no. 4, Loctite 3880, tested on polyester substrate with silver conductors and FR-4 substrate with gold over nickel conductors and tin lead conductors.

For the polyester substrate the adhesive has shown good results after all environmental exposures as used as a short circuit and to connect 0603 resistors with silver palladium finish. To connect the 0603 resistor with tin lead finish the results were unacceptable. For the FR-4 substrate with gold over nickel plating the adhesive has shown good results after all environmental exposures as used as a short circuit and to connect 0603 resistors with silver palladium finish. To connect the 0603 resistor with tin lead finish the results were unacceptable.
For the FR-4 substrate with tin lead plating the adhesive has shown unacceptable results after all environmental exposure as used as a short circuit and to connect 0603 resistors with silver palladium and with tin lead finish.
Shear strength values were acceptable for both component finishes on FR-4 substrate with gold over nickel finish and with tin lead finish. Shear strength values on polyester substrate were unacceptable

Variant no. 5, Loctite 3882, tested on polyester substrate with silver conductors

The adhesive has shown good results after all environmental exposures to connect 0603 resistors with silver palladium finish. To connect the 0603 resistor with tin lead finish the results were unacceptable.
When used as a short circuit the results were only acceptable after temperature cycling.
Shear strength values were acceptable for Ag/Pd component finish on polyester substrate but for Sn/Pb components finish the values were unacceptable.

Variant no. 6, Epo-Tek H20F, tested on polyester substrate with silver conductors

The adhesive has shown good results after all environmental exposures as used as a short circuit and to connect 0603 resistors with silver palladium finish. To connect the 0603 resistor with tin lead finish the results were acceptable, H₂S may increase the contact resistance when components with tin lead finish are used.
Shear strength values were unacceptable for both component finishes on polyester substrate.

Variant no. 7, Multicore M-4030-SR, tested on FR-4 substrate with gold over nickel conductors and tin lead conductors

For the FR-4 substrate with gold over nickel plating the adhesive has shown good results after all environmental exposures as used as a short circuit and to connect 0603 resistors with silver palladium finish. To connect the 0603 resistor with tin lead finish the results were acceptable with the same resistance values as for the components with silver palladium finish.
For the FR-4 substrate with tin lead plating the adhesive has shown acceptable results after all environmental exposures as used as a short circuit tin lead it and to connect 0603 resistors with silver palladium. To connect components with tin lead finish the results were unacceptable compared with the initial value. However, the contact resistance values were a factor 3 higher than the solder reference initially and a factor 6 higher than the solder reference after the exposures, which may be accepted for some applications.
Shear strength values were acceptable for both component finishes on FR-4 substrate with gold over nickel finish and with tin lead finish.

Variant no. 8, Acheson Electrodag SMD10, tested on FR-4 substrate with gold over nickel conductors and tin lead conductors

For the FR-4 substrate with gold plating the adhesive has shown good results after all environmental exposures as used as a short circuit and acceptable results to connect 0603 resistors with silver palladium finish. To connect 0603 resistor with tin lead finish the results were unacceptable.
For the FR-4 substrate with tin lead plating the adhesive has shown acceptable results after all environmental exposures as used as a short circuit. Connecting 0603 resistors with silver palladium has not been tested due to an assembly failure. To connect components with tin lead finish the results were unacceptable.
Shear strength values were acceptable for both component finish on FR-4 substrate with gold over nickel finish and for Sn/Pb component finish on FR-4 substrate with tin lead finish. Other values were unacceptable.

Variant no. 9, Wacker Semicosil 970EC, tested on FR-4 substrate with gold over nickel conductors and tin lead conductors

For the FR-4 substrate with gold plating the adhesive has shown unacceptable results after all environmental exposures as used as a short circuit to connect 0603 resistors with silver palladium finish and to connect 0603 resistor with tin lead finish.
For the FR-4 substrate with tin lead plating the adhesive has shown unacceptable results after all environmental exposures as used as a short circuit, connecting 0603 resistors with silver palladium, and to connect components with tin lead finish.
Shear strength values were unacceptable for both component finishees on FR-4 substrate with gold over nickel finish and with tin lead finish.

Variant no. 10, Tecknit 72-00236, tested on FR-4 substrate with gold over nickel conductors and tin lead conductors

For the FR-4 substrate with gold over nickel plating the adhesive has shown unacceptable results after all environmental exposures as used as a short circuit, to connect 0603 resistors with silver palladium finish and to connect 0603 resistor with tin lead finish.
For the FR-4 substrate with tin lead plating the adhesive has shown unacceptable results after all environmental exposures as used as a short circuit, connecting 0603 resistors with silver palladium, and to connect components with tin lead finish.
Shear strength values were unacceptable for both component finishes on FR-4 substrate with gold over nickel finish and with tin lead finish.

Variant no.11, Tecknit 72-00151, tested on polyester substrate with silver conductor

In general, the adhesive has shown unacceptable results after all environmental exposures as used as a short circuit and to connect 0603 resistors with silver palladium and tin lead finish. Process optimisation may prove the results for connecting components with silver palladium finish, as good results were found for 8 out of 23 measurements.
Shear strength values were unacceptable for both component finishes on polyester substrate.

Variant no. 12, Creative 106-22, tested on FR-4 substrate with gold over nickel conductors and tin lead conductors

For the FR-4 substrate with gold plating the adhesive has shown good results after all environmental exposures as used as a short circuit and to connect 0603 resistors with silver palladium finish. The measured contact resistance values were however relatively high, possibly due to the fact that the conductive material is carbon. To connect 0603 resistor with tin lead finish the results were unacceptable.
For the FR-4 substrate with tin lead plating the adhesive has shown unacceptable results after all environmental exposures as used as a short circuit, connecting 0603 resistors with silver palladium finish and with tin lead finish.
Shear strength values were acceptable for both component finishes on ceramic and FR-4 substrate with gold over nickel finish and with tin lead finish.

Variant no. 13, Tecknit 107-25, tested on polyester substrate with silver conductors

The adhesive has shown acceptable results after all environmental exposure as used as a short circuit and to connect 0603 resistors with silver palladium finish. To connect the 0603 resistor with tin lead finish the results were unacceptable. H₂S may increase the contact resistance when components with tin lead finish are used. The resistance value for the adhesive used as a short circuit was high compared with the other variants used on polyester substrate.
Shear strength values were unacceptable for both component finishes on polyester substrate.
Variant no. 14, Multicore 96SCMX39AGS88.5 Sn/Ag/Cu lead-free solder, tested on ceramic substrate with silver conductors and FR-4 substrate with gold over nickel conductors and tin lead conductors
All results were good

9. Conclusion

It is estimated that electrically conductive adhesives will not be a drop-in replacement for solder, but be used in niche applications like low temperature substrates and possibly where heat transfer is high, i.e. for shielding applications, ceramic substrates, and possibly for low current applications. A further development of the electrically conductive adhesives and process technology may be necessary if electrically conductive adhesives should more widely replace solder.

The overall result of the evaluation is that the reliability of the adhesives is very dependent on the substrate and the component plating. The silicone variants showed unacceptable interconnection resistance and the variants with carbon showed high interconnection resistance. In general, adhesives applied on tin-lead substrates or to tin-lead component terminals showed increased interconnection resistance.

The shear strength showed acceptable values for most of the variants especially when mounted on the ceramic substrate and on the FR-4 substrate with gold over nickel plating. However, the silicone variants have shown very low shear strength values on all substrates.

The overall results from the contact resistance measurement showed that the silicone variants were unacceptable for either the short circuit or mounting of chip resistors. The variants with carbon particles have relatively high contact resistance, and will only be suitable for components with high input impedance. The silver-filled epoxy, thermoplastic and thermoset variants showed good results when used with silver palladium component finish mounted on gold and silver conductors. Tin lead-plating is not compatible with the adhesives, as high contact resistance values were measured. However, one variant, the thermoplastic, showed reasonable contact resistance values when tested with tin-lead platings.

The overall results from the shear strength test have shown that when the adhesive is bonded to tin lead surfaces the fraction will be between the adhesive and the tin lead. When the bonding material is gold, silver or silver palladium the separation will be in the adhesive joint. All variants tested on the ceramic substrate showed very high shear strength force value. The variants tested on the polyester substrate has low shear strength value, except for one epoxy variant connecting to silver-palladium finish.

The variants tested on the two different FR-4 substrates have for the two silicone adhesives shown unacceptable results. The shear strength value for all other variants was acceptable when tested on the substrate with gold finish. When tested on the substrate with tin lead finish one of these variants has unacceptable shear strength values, while the other variants (except one with the same values on both substrate finishes) have lower values on the tin lead substrate.

From an environmental point of view, it is a question of release of lead and tin, or silver to the environment, unless carbon is used as the conductive material in the adhesives. All the three metals impose environmental hazards, but the actual risks are not easily evaluated. Both for solder and adhesive, recirculation of electronic equipment will reduce environmental release.

Based on the Life Cycle Assessment, it cannot be judged whether adhesive technology is better or worse than solder technology. The available data on material consumption and emissions during the life cycle are too uncertain. Further, in relation to toxicity, which is a crucial impact category in this assessment, too little is known about the actual environmental fate of metals leaking to the environment from waste steams.

The study has lead to the following recommendations in relation to manufacturing and emission handling:

	Use components with as little silver and palladium termination as possible
	Optimise adhesive consumption (for SMT about 1/5 as compared to solder paste amount)
	Assure high recovery percentages for the metals during recycling and reduce losses during manufacturing operations
	Assure proper treatment of hazardous wastes

From a working environmental point of view, the choice of technology is equivocal. The two technologies impose different hazards (the potential hazard impact of solder is slightly higher), but the occupational risks are minimal when adequate preventive measures are implemented.

A well considered working practice, the proper use of available technical preventive measures and personal protective equipment as well as good hygiene at work all reduce the actual risk. As a basic principle, the preventive measures should be prioritised as follows:

	Designing the assembly line in a way that reduces risk as much as technical possible (avoid skin contact with solder paste, flux and adhesives and avoid inhalation of solvent vapours and lead fumes)
	Establishing technical measures (e.g. ventilation) where needed.
	Prescribing the use of personal protection equipment.

From a working environmental point of view, the choice of adhesive resin should be made according to the following preference:

	Polyester
	one-component silicone / two-component silicone
	acrylate / two-component epoxy / one-component epoxy

From an environmental point of view, the choise of adhesive conductor should be made according to the following preference:

1. Carbon

2a. Silver-plated aluminium

2b. Silver-plated copper

3. Silver

From a technical point of view the adhesive systems should be selected with due respect to the below criteria:

	electrical conductive adhesive shall be applied on noble metal surfaces, i.e. tin-lead as a substrate or component terminal finish is not recommended, due to oxidation.
	redesign of the footprints on the substrates may be necessary.
	reduction of the opens in the stencil is necessary to reduce the applied amount of electrical conductive adhesive compared to solder.
	the existing production equipment can be used i.e. the stencil printer or the dispenser, the component mounting automate and the furnace. A furnace for adhesive curing is recommended, especially if both adhesive and solder shall be used in the production.
	process optimisation with electrical conductive adhesive is more difficult than with solder, especially due to no self alignment and no contraction of the adhesive.
	curing shall be performed according to manufacturers specification, i.e. the temperature shall be controlled.
	interconnection resistance is higher with electrical conductive adhesive than with solder.
	the adhesion between component/adhesive/substrate seems to be lower for electrical conductive adhesive than for solder, this may not be correct for some SMD (ceramic chip resistors and capacitors) components mounted to thickfilm silver conductors on ceramic substrate. Depending on the electrical conductive adhesives adhesion strength and the component, components shall be mechanical fixed.
	repair may be a problem, as hand mounting in electrical conductive adhesive may give an increased interconnection resistance, and revealed silver flakes may tend to migrate.
	repair is estimated to be much more costly with electrical conductive adhesives than with solder.
	protection against uncured epoxy and epoxy vapour is required. Personal shall be trained to handle epoxy.
	for low temperature substrates there is only one alternative, which is low melting solder containing Indium.
	the price for electrical conductive adhesives is about 5 times higher than for solder.

A widespread substitution of solder with silver based adhesives would increase the silver demand by approximately 11%. However, with today’s technology the demand will increase with less than 1%. The substitution for adhesive will most obviously also increase the silver demands for component termination as Ag/Pd terminated components are highly recommended for adhesives.

A widespread - future - substitution would therefore most obviously increase silver prices - when also comparing to the fact that it is estimated that the silver supply horizon is only about 14 years! However, with the present technology level, the increase in silver demand due to adhesive application would most obviously be marginal as compared to other mechanisms affecting silver prices.

10. References

1.	Helle Westphal et al, Environmental Impacts of Adhesives and Solders, Danish Toxicology Centre, June 1998, ISBN 87-7398-123-0
2.	Christensen F M (2000), Life Cycle Assessment of electrically conductive adhesive versus traditional tin/lead solder. Kamille II. DTC, Denmark
3.	Christensen F M, Jensen A B (2000). Silver as a resource as compared to tin and lead. Kamille II. DTC, Denmark.
4.	Christensen F M, Jaroszewski M, Syska J, Cohr K H (2000), Occupational health assessment. Kamille II. DTC, Denmark.
5.	Cohr K H, Christensen F M (2000). Toxicological aspects of adhesives as compared to soldering systems. DTC, Denmark.
6.	Jørgensen T (2000). Test report. Kamille II. DELTA, Denmark.
7.	Jørgensen T (2000). Production analysis. Kamille II. DELTA, Denmark.
8.	Legarth J B (1996). Recycling of electronic scrap. Ph.D-thesis. AP96.08/IPT.96.08-A. Department of Manufacturing Engineering, Technical University of Denmark.
9.	Christensen F M (2000). Recycling analysis. DTC, Denmark.
10.	Harris P G, Whitmore M A (1993). Alternative Solders for Electronics Assemblies. Part 1: Materials Selection. Circuit World. Vol. 19(2), p. 25-27.
11.	Lee N-C (1999). Lead-Free soldereing – Where the world is going. Advancing Microelectronics. 26(5), p. 29-35.
12.	MCS (1996). Mineral Commodity Summaries, January 1996. World Mine Production, Reserves and Base (http://minerals.er.usgs... odity/silver/880396.txt)
13.	World Resources, 1992. CEIDOCT, Mineral resources. (http://tiger.eea.int/projects/envwin/busindus/ceidoct/42.htm)
14.	Boustead I (1997). Eco-profiles of the European plastics industry. Report 12: Liquid epoxy resins. Association of Plastics Manufactures in Europe (APME), Brussels.
15.	Deubzer O, Middendorf A (1998). Environmentally Friendly Materials and Processes for SMT-Reflow Soldering and Conductive Adhesive Joining in Comparison, Proc. Int. Conf. on Electronic Assembly: Materials and Process Challenges,Atlanta/Georgia, June 16 - 18, 1998.
16.	Hauschild M, Wenzel H (1997). Environmental assessment of products. Volume 2: Scientific background. First edition. Chapman & Hall, United Kingdom.
17.	ISO 14040 (1997). Environmental Management – life cycle assessment – Principles and framework.
18.	ISO 14041 (1999). Environmental Management – life cycle assessment – Goal and scope definition and inventory analysis.
19.	ISO/DIS 14042 (1998a). Environmental Management – Life cycle assessment – Life cycle impact assessment. Draft international Standard.
20.	ISO/DIS 14043 (1998b). Environmental Management – Life cycle assessment – Life cycle interpretation. Draft international Standard.
21.	Pedersen M A (1998). Brugermanual til UMIP PC-værktøj (betaversion). Miljø- og Energiministeriet, Miljøstyrelsen. (User manual in Danish)
22.	Segerberg T (1995). Life Cycle Assessment of Tin-lead Solder and Conductive Adhesive in Surface Mount Technology. IVF Report 95051. The Swedish Institute of Production Engineering Research (IVF).
23.	Snowdon K (1999). Lead Free – The Nortel Experience. Technical paper and presentation at IPCWorks ’99, Minneapolis, Minnesota, USA. Association connecting electronics industries (IPC).
24.	Wenzel H, Hauschild M, Alting L (1997). Environmental assessment of products. Volume 1: Methodology, tools and case studies in product development. First edition. Chapman & Hall, United Kingdom.
25.	Westphal H, Christensen F M, Nielsen I R, Hvims H (1995). "Kan miljøet blive renere med ledende lime i elektronik-produktionen - KAMILLE", Danish EPA, Dept. for Products and Cleaner Technology. Copenhagen 1995. (Toxicological and environmental comparison of tin/lead solder with electrically conductive adhesives - in Danish).
26.	Cohr et al (1999), Lead-free solder, Health and safety properties of alternative metals and a technical status report.
27.	Hvims, Henrik (1994), Adhesives as solder replacement for SMT, DELTA
28.	Petersen, Jens Rytter et al (1992), "Electromigration and voltage induced corrosion in electronics" SPM-109, DELTA.
29.	AMM-Online, (1999), Silver Profile (http://www.amm.com/ref/silver.HTM)
30.	Boliden, a (1998) Våra viktigaste metaller. Information brochure.
31.	TSI (1998). The Silver Institute. Supply and Demand of Silver in 1997 (http://www.silverinstitute.org/demand.htm.)
32.	Kleinmann, Georg (1998). Silver lining revisited. (http://www.commodity.com/reports/98_02_silver.htm)
33.	Hauschild M, Wenzel H (1997). Environmental assessment of products. Volume 2: Scientific background. First edition. Chapman & Hall, United Kingdom.
34.	MCS (1996). Mineral Commodity Summaries, January 1996. World Mine Production, Reserves and Base (http://minerals.er.usgs... odity/silver/880396.txt)
35.	Knudsen, M.B., 1998, Price of silver may double in about a year. (http://www.sunshinemining.com)
36.	World Resources, 1992. CEIDOCT, Mineral resources. (http://tiger.eea.int/projects/envwin/busindus/ceidoct/42.htm)
37.	Lesser, Mikael (1999). Boliden, Bergsöe. Personal communication
38.	ITRI Market Data 1995 and 1996. (http://www.itri.co.uk/market.htm)
39.	Harris P G, Whitmore M A (1993). Alternative Solders for Electronics Assemblies. Part 1: Materials Selection. Circuit World. Vol. 19(2), p. 25-27.
40	Lee N-C (1999). Lead-Free soldereing – Where the world is going. Advancing Microelectronics. 26(5), p. 29-35.

 

[Front page] [Top]