Skip to main content

Environmentally Sustainable Manufacturing of Ultra-Functional Miniaturised Consumer Electronics Using Hyper-Fine Solder Powders

Final Report Summary - HYPERSOL (Environmentally Sustainable Manufacturing of Ultra-Functional Miniaturised Consumer Electronics Using Hyper-Fine Solder Powders)

Executive Summary:
Consumer electronics as exemplified by the cell phone have continuously decreased in size whilst offering more and more functions. The future cell phones will offer phone/video/TV/medical diagnostics/computer power using roll-up flexible screens. Sustainability is of paramount importance as electronics manufacture consumes enormous resources. However, currently advancement is being limited by the lack of hyper fine solder powders to enable ultra-high interconnect density on Printed Circuit Boards (PCB) which is needed for further miniaturisation and greater functionality. This limitation offers the SME participants an excellent opportunity to innovate and achieve business growth via offering enabling technology for:
➢ Reduction in electronics size, weight and therefore the amount of raw materials used.
➢ Finer interconnects, higher interconnect density, freeing-up of real estate on the printed circuit board and therefore enabling increase in component density.
In order to make accessible this technology research has to be undertaken in the next fields:
i) Metal atomisation: This project aims to achieve a commercially viable process and prototype for manufacture of solder sphere in sizes Type 8 and 9.
ii) Anti-oxidation protective coatings: HYPESOL aims to progress beyond the state-of-the-art by coating solder powder spheres with thin protective films based on unique non-acidic substances such as organic ligands.
iii) Sizing of Type 8 and 9 solder powders: State-of-the-art will be advanced by inventing methods of sizing based on wet sieving of fluidised bed sieving.
iv) Low rosin solder pastes: HYPERSOL solder pastes will contain <1.5% rosins, and therefore greatly reduce harmful rosin vapour in factories.

Project Context and Objectives:
HYPERSOL aims to deliver a prototype production line for solder powders of type 7 & 8. The line must simultaneously atomise the very fine powder (2-11, 2-8 micron range), separate the particles outside the desired size range to permit conformity with permitted tolerances for over- and under-size % permitted, and coat the product with a protective film of an organic substance that will inhibit oxidation of the surface of the powder.
The system is planned to be capable of processing about 50kg/hour of molten solder. It should be capable of continuous operation for several hours, so a 200kg capacity melter has been selected. Similar designs are on hand at ASL for fully continuous melt delivery at up to 200kg/hour using two melters.

In the first phase of the project the partners consisting of 4 SMEs, 1 industrial end user and 4 RTDs focused on the generation of specification and requirements needed for the progress of the project. Thus, specifications on engineering drawing of the HYPERSOL production prototype, the classes of coating and coating sub-assembly for the protection of the solder spheres, of the sieving sub-assembly, the real time process control system and of the bespoke stencil as well as of PCB reliability testing methods were set.
An atomizing sub assembly production line has been developed in order to produce the fine solder particles. The production is based on gas atomization while the system has also an optional centrifugal atomization. The atomization production line, simultaneously atomizes the very fine powder (2-11, 2-8 micron range), separate the particles outside the desired size range to permit conformity with permitted tolerances for over- and under-size %, and coat the product with a protective film of an organic substance that will inhibit oxidation of the surface of the powder. The system is capable of processing about 50kg/hour of molten solder. Moreover, the system is capable of continuous operation for several hours, so a 200kg capacity melter has been selected.
Concerning the coating sub assembly a vaporizer was designed and manufactured in order to be integrated with the atomizing system. The vaporizer can be used with both liquid and solid agents and is capable to provide the necessary concentration of the organic agent in the nitrogen stream (according to the specifications set) and thus the solder spheres that will be produced in the atomizer will be coated. Some final tests should be made to ensure the smooth operation of the coating sub-assembly.

Moreover, systematic research was undertaken in order to define the most suitable organic agent to be used as protective, anti-oxidized coating in the fine solder spheres. After assiduous literature research in both scientific papers and relevant patents, experimental work took place in order to test the anti-oxidation ability of various protective coatings, targeting in the minimization of the oxidation of the surface of solder spheres.

Project Results:
Oxide and protective surface layers


We plan to make low-oxygen powder with size range 2-11 or 2-8microns (Type 7, 8). ASL has data on making ultrasonically and centrifugally atomised powder that may be of some help, so the implications of this information are discussed here.

Oxygen in Ultrasonic and centrifugal atomisation

ASL has found that, no matter how low the oxygen level is kept in the atomising atmosphere during ultrasonic atomisation, the measured oxygen content of the resulting powder had never fallen below about 60ppm (as atomised, size range 20-100microns, D50 ~50microns, Sauter mean about 40microns). In ultrasonic atomisation, where the spray moves at only ~1m/s, it has been found that atmosphere oxygen levels much above 100ppm lead to distorted particle shapes. If the oxygen level is allowed (due to gettering) to fall into the 1-10ppm range (at which level, the slow circulation of gas in the vessel means that near-zero levels may occur locally), the resulting powder sticks together very strongly. Thus we find it necessary to inject air to control the oxygen level in the nitrogen atmosphere in the range 20-50ppm.

In centrifugal atomisation we have found that the optimum level of oxygen is rather higher, about 100-300ppm. This may reflect the far higher velocity of the droplets (ca 100m/s) which must lead to much more rapid freezing.

The amount of air/oxygen needed to keep the atmosphere in the ultrasonic atomiser at safe levels has been measured and found to equate to a rate of injection of oxygen of 6ppm related to the melt flow rate. The fact that this is ~1/10th of the measured oxygen levels in the powder tells us that there are two critical levels of oxide on the surface;
1 The lowest level to prevent sintering/cold welding of the super-clean surfaces
2 The lowest level stable in air.

Number crunching

The lowest possible level of oxygen in the powder is desirable, and we plan to attempt to reduce/control this by some chemical coating. We have two options for selecting our reagents:-
1 Attempt to run super-clean and bond to metal atoms on the surface
2 Run the 6ppm, non-stick level of oxygen, and bond to this oxide film

If we took the second option, then we would expect oxygen levels in the order of 60ppm at 4microns (as the surface area is ten times that at 40microns).

If we look at the surface area of a 40micron sphere and calculate the oxide thickness with 6ppm oxygen content we find:-
Surface area = π d2; Volume = π d3/6; weight = π d3/6 x 7400
Surface area/kg = (π d2)/(π d3/6 x 7400) = 6/7400d m2/kg Eq 1
For d = 40microns or 4 x 10-5m, Surface area/kg = 7 x 105/(7400 x 4) m2/kg = 23.6 m2/kg

From the above equations, we see that, if we reduce the Sauter mean diameter by a factor of 10 to 4 microns, the surface area/kg will rise tenfold to 236 m2/kg.
Now if we take 6ppm of oxygen and spread it over the 23.6m2 we find that we have a weight per square m of 6 x 10-6 kg/23.6 = 0.254 mg/m2. Eq 2
The atomic weight of oxygen is 16. report the empirical covalent radius as 73pm and the Van der Waals radius as 152pm. So if we had a layer of oxygen one atom thick, then it contains one atom in each area of 0.15 x 0.15 (nm)2 (ignoring questions of packing pattern). So one square metre of oxygen monolayer contains 1/(0.15 x 10-9)2 atoms or 44 x 1018 atoms of oxygen. Avogadro’s number is 6.0 x 1023 atoms/gm mole. Thus we can say that the weight of oxygen in one square metre of a monomolecular layer is given (in grams) by:-
44 x 1018 x 16/6.0 x 1023 gm = 117 x 10-5 gm = 1.17 mg/m2 Eq 3

Now we calculated above that an addition of 0.254mg/m2 is sufficient to prevent sintering, so we must conclude from the estimate of 1.17mg/m2 for a monolayer that there cannot possible be a full oxide layer on the surface, as this would require about 1.17/0.254 = 4.6 times more oxygen. Put another way, the oxygen can only cover about 20% of the surface area.

Now the above estimates are necessarily somewhat approximate, but it seems clear that the minimum level of oxygen to prevent sticking is well below monolayer level, so it is hardly surprising that such a powder is highly reactive in air and, by the time it has been exposed to air to get into an oxygen analyser, proves to have an oxygen content some 10 times higher, as this indicates a coverage of approximately 1-2 atoms thick.


This raises the intriguing question; “Why does a ~20% coverage of the surface prevent welding/sintering?” Perhaps we must just accept that it does, and remember that the clean tin surface will have adsorbed gas on it as well, presumably nitrogen atoms.

Now the point of these calculations is to understand what is happening in the atomisation process and the implications for reagent selection. From the above it now appears that we cannot select a reagent to bond to an oxide film, as one cannot exist if we are to hit our ultra-low target for oxygen content (not yet agreed, but 100ppm might be acceptable?)

Implications for reagent use

Using the same methodology, we can estimate the amount of reagent required for a monomolecular layer, assuming that the molecules are long chains with a “sticky head” group like NH2 or COO. The diameter of such groupings at the end of organic molecules will obviously be rather larger than that of a single oxygen atom, say twice as large. So this means that the area “covered” by each molecule will be 4 times larger. Adapting Eq 3 above, we can write; The weight of a reagent in a monolayer is given by:
44 x 1018 x MW/6.0 x 1023 gm = MW x 7.33 x 10-5 gm = MW x 0.0733 mg/m2 but this is assuming the same molecular diameter as oxygen. For a molecule with 4x the area, this drops to MW x 0.0733 mg/m2/4 = MW x 0.0183 mg/m2.

Now for a moderately large long chain hydrocarbon, like a C10, the molecular weight is of the order of 100-200gm. Thus a monolayer of such a compound needs a loading of 0.0183 x 100-200 = 1.8-3.6mg/m2. If we take the case of a 4 micron particle with 236m2/kg this implies an addition of (1.8-3.6) x 236 mg/kg = 420-840mg/kg.
This implies, at 50kg/hour, an addition of 50 x 420-840mg/hour = 21-42 grams/hour

So I think we can suggest that the metering system for our coating substance needs to be able to work down to as low as 10grams/hour and up to at least 100gram/hour. Clearly injecting less coating than needed for a monolayer cannot be effective. On the other hand, we may need more than a monolayer, and the proportion of the injected substance that actually bonds to the powder surface may be as low as 10% or even less. Thus a very wide range of injection rates must be designed for.

Discussions at the kick-off meeting showed that there is little hope of achieving a really impermeable coating that will not allow any oxygen or water molecules through. However, in theory we might hope that, if we ensure that all metal atoms on the surface are chemically bonded to a suitable molecule, then the driving force for oxide formation may be reduced or eliminated. In any case, a shelf life of 4-8weeks is probably sufficient to allow normal transit to the paste.

Injection system design/specification

We plan to convey the atomised solder spray in a gas flow of ~10-15a.m3/min. This implies that, at 50kg/hour, the spray concentration is (50/60)/10 = ~0.1kg/m3
Just for interest, let us estimate the number of particles in a cubic metre. The number mean is finer than the median or Sauter mean, and so we can take a 3micron sphere as our case. The weight of each particle is thus π d3/6 x 7400 kg = (3.142 x 3 x 3 x 3 x 10-18/6) x 7400 kg. = 104 x 10-15 kg = 1 x 10-13kg. Thus at a concentration of 0.1kg/m3 we have 0.1/(10-13) particles per m3 or 1012 /m3. Incidentally it also means that we are creating particles at the rate of about 1013/min (ten trillion per minute!)

It may be interesting to estimate the distance between particles in such a system. If they are uniformly spaced, the distance between them will be one over the cube root of the number/m3 or 10-4m, which is 100microns. Thus the particles are some 30 diameters apart and collisions will not be very frequent (or will they? – I do not know a simple way to estimate this).

If we were to inject sufficient vapour of the coating substance into the conveying gas flow to coat as calculated above, we would be injecting at a rate of 10-100g/hr into a flow of ~10m3/min. This is a rate of (10-100)/(60 x 10) = 0.016-0.16g/m3 This could be put in ppm as around 16-160ppm.

If we base the design of the injector on taking a proportion of the flow in the recycle gas, say 1-10% (100-1000l/min) and adding the substance to this to mix with the main flow, then these concentrations need to rise by 1-2 orders of magnitude to 200ppm – 2%. This, of course, implies a vapour pressure of the species of the order of 0.2-20mbar divided by the ratio of the molecular weight of the species to that of nitrogen. The obvious way to do this is by saturating a controlled gas flow with liquid substance at a controlled temperature (both gas and liquid).


1 ASL experience of ultrasonic atomisation indicates that sintering of powder can be prevented by a less than 20% coverage of the surface with oxygen.
2 To achieve a powder with 100ppm or less of oxygen content, we must coat with a chemical that will bond to the metal itself, not an oxide layer
3 The injection system needs to be able to add the surface protection agent at rates ranging very widely from 10g/hour to well over 100g/hour

Potential Impact:
Partial sucess in terms of final achievements. Main dissemination through
5th Internation Conference on Spray Deposition and Melt Atomization
SDMA 2013 , Bremen Germany
23 – 25th September 2013

Challenges of making finer solder powders by atomization
Authors: Dr. J. J. Dunkley, Dipl-Ing. D. Aderhold

Trends in miniaturisation of electronic circuitry are driving a demand for ever-finer powders. From typical sizes of 25-45 microns (Type 3) to as fine as 2-11microns (Type 7). This his paper discusses the limitations of current main stream atomising methods, namely gas and centrifugal atomisation and describes the challenges in the powder handling of super-fine oxidation prone solder powders.
Gas atomisation can make fine powder, but broadly distributed and suffers from “satelliting” with adherent superfine particles giving non-spherical shape. Centrifugal atomisation give a narrower distribution, but with a 30micron powder needing a cup speed of over 50,000rpm it is hard to believe that such fine powders could be practically produced by cups spinning in excess of 120.000 rpm. Then the question of surface oxidation arises; surface area increases directly with reducing diameter. It is also almost impossible to get such fine powders to flow. Thus a method that suspends the powder in a liquid to protect and transport it is needed. Finally the challenge is to find methods to separate fines and oversize particles in the size range below 30 microns. This cannot involve sieves so must be aerodynamic.

List of Websites:
No longer supported site - Continuation usage of the project equipment into project 680718