A Novel System for the Production of World's First Micro Ball Grid Array (µBGA) Spheres for enabling the EU Electronics Industry to produce smaller electronics goods

Project Information

µBGAS

Grant agreement ID: 243653

Project closed

Start date 1 June 2010

End date 31 May 2013

Funded under

• Specific Programme "Capacities": Research for the benefit of SMEs

Total cost € 3 924 714,02

EU contribution € 3 118 500,00

Coordinated by

KINGSTON COMPUTER CONSULTANCY LIMITED
United Kingdom

Final Report Summary - µBGAS (A Novel System for the Production of World's First Micro Ball Grid Array (µBGA) Spheres for enabling the EU Electronics Industry to produce smaller electronics goods)

Executive Summary:
The Silicon electronics industry's appetite for increased density and increased I/O pins (more functions) has led to the demand for smaller and smaller electronic packages e.g. integrated circuits (ICs) within our mobiles. This needs further miniaturisation of the die, packages and PCB modules. In this context µBGA or Ball-Grid-Packages (BGA) are ideal solutions for improving the ratio of pin counts to the board area. In BGAs the I/O connections are on the underside of the device and the BGAs themselves are shrinking in size thus allowing for miniaturisation and added functionality for the same or smaller ‘real estate’.
To facilitate this and ensure security of supply to the European manufactures at a competitive and affordable price; the development of a high volume µBGA production system with high material utilisation and cost minimisation to manufacture solder spheres of 100µm diameter or less (typically 75 and 50µm) is the main objective of this project.
This is an European research collaboration for the benefit of the SME-AGs, especially those associated with the European Electronic Packaging and Chip manufacturers. The realisation of such a manufacturing system for the creation of solder balls; utilises number of technologies that the RTD partners (KCC, IKH and Tyndall) and SMEs (ABIS, CIT and SEMICON) of this consortium have developed.
The consortium has successfully completed in full the tasks and the objectives of the DOW in meeting the above stated objective. This consisted of bringing together (integrating) the solder feed tank, solder jetting tank, the ultrasonic breaking up of the solder stream into solder spheres, the imaging and analysis of the solder sphere images, the prevention of solder spheres from conjoining and the separation of the spheres into specific sizes, thereby having an integrated laboratory prototype for the manufacture of solder spheres of <100µm down to 50µm in diameter from molten solder. Having assembled the individual units (hardware and software) as one laboratory prototype we demonstrated the system capability in the manufacture of solder spheres from molten solder and validated this at a partner premises, where the dissembled (for shipping) was reassembled using the written step-by-step instructions and run for a few days to demonstrate the capability of the system and also to assess the ease of use by engineers (other than those involved in the development, though they were there to assist in the training). While the objective of producing the said sphere sizes were successful some minor issues remain to be addressed, this will be addressed outside the project during a commercialising phase; it is our intention to make solder spheres available for sale commercially.
In conclusion we have demonstrated a laboratory prototype capable of manufacturing of solder spheres of typically 100, 75 and 50µm from molten solder using ultrasonics and thus facilitate security of availability of BGAs at a competitive price to especially European manufacturers. We have in this regard outlined in Deliverable D15.1 and D15.2 a production facility setup plan (PFSP) and sources of funding to make this production facility a reality.

Project Context and Objectives:
The microelectronics industry has been the engine of the modern information revolution for almost 50 years. In our everlasting quest to process more and more data faster and faster, the $100 billion Silicon industry's appetite for increased density and increased I/O pins has led to the demand for smaller and smaller packages (small footprint) with the miniaturisation of the die, packages and PCB modules. In this context µBGA or Ball-Grid-Packages (BGA) are ideal solutions for improving the ratio of pin counts to the board area. In BGAs the I/O connections are on the underside of the device and the BGAs themselves are shrinking in size.
The historical progression from chunky discrete devices with connections wired externally to yesterday’s Dual-in-Line packages to today’s BGA configuration is shown in figure 1 (taken from October 2010 µBGA Kick-Off Meeting presentation).

Figure 1 (Left) Historical progression from discrete devices to dual-in-line packages to BGAs (Left) and figure 2 (Right) the laboratory prototype with the µBGA sub systems and components integrated to form the solder sphere production system (Integration and acoustic separation is described fully in Deliverable D7.1 the acoustic separator is also described in Deliverable D6.1)
To facilitate this development and continued miniaturisation, the development of a high volume µBGA production system to manufacture solder spheres of 100µm diameter or less (typically 75 and 50µm) is the main objective of this project. This is an European research collaboration for the benefit of the SME-AGs, especially those associated with the European Electronic Packaging and Chip manufacturers.
In conventional packages (chip carriers with lead pins), leads are placed in the periphery of the package; making the footprint larger. In the case of the BGAs these leads are replaced with solder balls arranged in a matrix across the bottom of the package, these are then soldered onto a PCB.
There are number of advantages of BGA packages, namely; with the solder balls at the edge of the package and with decreased pitch for µBGAs the ratio of the I/O 'leads' to area is greatly increased. Furthermore with BGAs being solder balls, rather than leads, damage during handling and soldering is reduced and BGAs are amiable to high frequency operation (by custom design of ground planes, ground rings and power rings in the package construction). BGAs also have improved heat dissipation since the GND and VCC 'pins' are located at the centre of the package and under the package, the heat generation can be transferred out easily, with the GND and VCC acting as a heat sink. Finally the BGA packages can tolerate slight misalignment during mounting (placement is relative to the pad area), requiring less expensive surface mount equipment. This is so since the solder balls self-align during solder reflow.
It is clearly evident that µBGAs will play an important role in the miniaturisation of electronic packaging and our challenge is to develop a system to manufacture 75µm with a target of 50µm solder balls (beyond the DOW requirement) with high material utilisation and cost minimisation.
Today much of the solder spheres in use are 300µm in diameter except for some niche applications where smaller spheres are available from some USA and Far-Eastern producers.
Production of smaller spheres are ONLY one side of the coin, most current ‘pick and place’ systems are not geared to handle sub 100µm spheres and especially 50µm spheres that the µBGA system has and is capable of producing. This necessitated in addition to the system developed for sub 100µm spheres, to develop a ‘pick and place’ technology and moreover demonstrate it with the use of solder spheres produced from our prototype system. This too has been successfully demonstrated, along with a novel acoustic separation to prevent and minimise co-joining (Patent was applied) and a centrifugal sorter for size separation.
In Figure 2 we show the integrated system, figures 3 shows another important aspect of the system i.e. the feedback system to control mono size sphere production (illustrated with an image of the graphical user interface (GUI)), figure 4 shows the SEM images of randomly selected spheres produced from the system in figure 2 (spheres are nominally 50µm), figure 5 shows the novel acoustic separator to prevent/minimise co-joining of spheres, figure 6 shows the centrifugal separator which separates spheres into different sizes and figure 7 shows an illustration of the µBGA sphere pick up tool, the test PCB assembly using the tool developed ‘for pick and place’ the µBGA spheres and finally the Euro-Millions working demonstrator.

Figure 3 (Left) The GUI of the feedback control system for producing mono size solder spheres and Figure 4 (Right)shows the SEM image of cluster of spheres from sample A run (700 X Magnification, fuller details in deliverable D10.1)

(a)
(b)
(c)
Figure 5 The acoustic separator (a) with the audio speaker connected to the acrylic cooling column and in intimate contact with the cooling column through the conical pot (b) The acoustic separator in operation (c) The separation can be seen at first with the spheres in a near straight trajectory, once the acoustic signal is introduced the solder spheres take a sinusoidal trajectory

Figure 6 Centrifugal Separation System – original system (left), modified rotating element (centre) and image of separator plates & separated spheres (right)

Figure 7 Illustration of the operation of the novel Silicon MEMs pick up head for use with flip-chip alignment & bonding systems (Left), View of assembled structure on test PCB (Middle, µBGAs on the underside), Final miniaturised version of Euro-Millions number generator device using µBGA solder sphere interconnect (Right)
The above illustrates not only the makeup of the system but also then end product meeting the DOW objective and more. The realisation of such a manufacturing system for the creation of µBGA solder balls; utilised number of technologies that the RTD partners (KCC, IKH and Tyndall) and SMEs (ABIS, CIT and SEMICON) of this consortium developed. What have we achieved? The following are the main Technical achievements:
1. A system for producing 100µm spheres (DOW requirement), as well as 75 and 50µm spheres (beyond DOW)
2. A system capable of running without the necessity to clean beyond the ‘As –received’ solder pellets (no special cleaning, no filtering). Quality of our spheres as good or better than commercial sphere, this has been reported in deliverables
3. A system capable of running for a week or more producing 500 million solder spheres daily without the need to refill i.e. system has sufficient capacity
4. A system which does not require the Jetting tank to be ‘opened’ apart from occasional cleaning i.e. continuous jetting without opening jetting tank to atmosphere
5. A Magnetostrictive transducer instead of conventional piezoelectric transducer for breaking up the solder stream
6. Shown the breakup of solder stream to produce the required size spheres can be achieved with a ‘fixed’ size orifice but changing frequency and velocity in the main (other parameters have a secondary effect)
7. A Neural Network feedback system to control the size of the produced spheres to achieve greater yield
8. Minimisation of solder sphere conjoining by ensuring exit spheres move in a ‘straight line’
9. A novel acoustic system to separate spheres whilst molten thereby minimising co-joining
10. An In-Line centrifugal system to separate spheres of different sizes where this occur
11. A new ‘Pick and Place’ tool for 50µm solder sphere, this can also handle larger spheres
12. Demonstrated a working electronic device (Euro-Million Device) with µBGA spheres
13. Moreover with the use of lead-free solder in this project the manufacture of these balls are environmentally friendly; a green technology
14. A system ready for commercialisation after some optimisation outside this project timeline
Much has been achieved despite many problems; replacing non-performing partner and on-going temporary cessation of funding.

Project Results:
The system envisaged and developed (with small variations, which were justified in the various deliverables) is shown in figure 1. In this section we discuss the S&T in the work package (WP) starting with WP2 to WP12.
Work Package 2: Solder Alloy Feed System
The heart of the system is the solder jetting tank shown in the middle, with liquid solder, jetting aperture and a transducer.

Figure8 Schematic of the µBGA laboratory prototype
As one would expect, to jet out a solder stream and break it up, first and foremost liquid solder need to be fed via a liquid solder reservoir (shown on top RHS) into the jetting tank.
The option of feeding the solder from the solder feed tank into the solder jetting tank invariably would mean at the very least duplicating (with the exception of the orifice/nozzle) the solder jetting tank to perform as the solder feed tank wherein the solder could be melted. Added functionalities is also required, as the molten solder in the solder feed tank now needs to be transported into the solder jetting tank in a controlled manner, adding more complexity. To transport a molten liquid such as solder in a controlled manner require a high temperature valve.
The calculation (shown in Table 2 of deliverable D2.1) indicated that the volume of solder used in 1 minute for 50 and 100µm spheres are 3.8E-8 and 3.14E-7m3 respectively. Thus the volume per minute that the high temperature valve would need to be able to cope is very small less than a micro-litre (μl) this will of course be a 60 times smaller flow rate if solder feed is per second. Realistically this is difficult to regulate and further more unnecessary as the calculations in Table 3 of D2.2 show.
The intended purpose of the solder feed tank is to replenish solder as it is used up in the jetting tank and to maintain the pressure head. The pressure head as can be seen from the far right column of table 3 in D2.1 has only a small influence on the overall pressure and the added complexities (heated flow valves, heated solder feed tank etc.) does not justify, nor is it beneficial (see later) to feed liquid solder into the solder jetting tank. Our design ensures that the total pressure is always maintained at the set value, within the ‘reaction time’ of the Horiba pressure regulator (see deliverable D2.1) that we use to maintain a constant Nitrogen pressure ‘head’.
Our final and preferred option is to use solder pellets (solid feed) in the solder feed tank. In Table 3 of D2.1 we showed that the loss in pressure head due to the production of 50 and 75μm diameter solder spheres over an 8 hour shift is negligible compared to a typical operating pressure of 1.2bars. This being the case and since we always maintain the applied pressure via the use of a pressure regulator (Horiba), the most critical parameter that influences size as given by d = dj(1.5ν/djf)1/3 is the velocity which is a function of pressure and this we maintain ensuring the velocity and thus sphere size does not change.
With the loss of solder over an 8 hour production run being negligible (Table 3 od D2.1 volume used in 1 day) and the Horiba pressure regulator maintaining the set pressure, thereby minimising or negating the influence of pressure/velocity on the sphere diameter, it is evident that the use of solder pellets in the solder feed tank with the option of feeding the required amount of solder in the form of solid pellets into the jetting tank is a simpler solution for ‘indefinite’ production without opening the jetting tank. The amount of solder required to load into the jetting tank and thus pre-loaded into the solder feed tank can easily be determined from Table 3 of D2.1 (Mass (Kg) of Solder used for production in 1 hour) and with real data in use. This route of using solid solder pellets in the solder feed tank and replenishing every day or every few days is a much simpler solution than the solder melt feed (liquid).
We thus justify not using a liquid solder feed into the solder jetting tank as required by the DOW. Moreover introducing liquid solder into the solder jetting tank disturbs the equilibrium and thus changes the jetting and the sphere sizes. Furthermore in using ‘solder pellets’ of the good quality and in the ‘As Received’ state without further cleaning or passing molten solder through a ‘filter’ system, the question to answer is does this introduce an inferior solder sphere. We show you that this is not the case.
We demonstrate (see deliverable D2.1) with Scanning Electron Microscopy (SEM) images, cross-sections, Energy Dispersive X-ray (EDX) that the spheres produced have no additional elemental metals and oxidation has been minimised with the native oxide (10-20Å) forming a protection barrier and the use of Nitrogen in pre-loading, ‘storage’ and during production runs ensures high quality spheres as good as or even slightly better than commercial sphere. We have also shown that these spheres perform well in real devices (Euro-Million device, D12.1).
‘Real-time’ liquid molten solder feed is not only unnecessary due to the small volumes and mass of solder used but may have disadvantages in maintaining the homogeneity of a production run. We also have set out in deliverable D2.1 the case for not using solder wire. This leads us to the system of solid solder pellet feed into the solder jetting tank (a full discussion and system development is given in D2.1). For convenience the system developed is shown below in figure 9.
(a)
(b)
(c)
Figure9. Shows an image of the (a) solder feed tank with the rotator (Black Knob) and the (b) transporter (clear tube) and (c) the system with the solder feed tank, jetting tank, Nitrogen control panel and Camera. The cooling column and centrifugal separator are not shown
Work Package 3: Development of techniques and prototype solder jetting techniques
The main Science and Technology work in this WP3 was to:
1. Establish the technique to be used for solder jetting
2. Investigate materials for orifice and jetting tank that would allow continuous jetting at high velocities and throughput
3. Design and develop a prototype for solder jetting
With regard to the techniques for solder jetting we considered:
1. Pressure Driven Valves: In essence molten solder can be ejected as a solder droplet of nanoliter volume with the use of a fast switching pressure valves (Micro Valves).
2. Pneumatic Drive Valves: Micro valves supplying small volumes of Nitrogen at fast switching speed and low power consumption can be used to eject the molten solder contained in a heated container e.g. glass, stainless steel.
3. Syringe Technology: In the previous two technologies an external pressure regulator is used to create the pressure boundary conditions, however in the syringe technology the external controller can be unregulated and is only used to hold the solder in equilibrium with surface tension and the atmospheric pressure, the syringe supplies the required additional pressure (overpressure) to release the solder droplet. This technology can be thought of as equivalent to pipetting.
4. Ink-Jet Technology: This technology is akin to the ink-jet printer technology. In this type of system, a change in the volume of a fluid (solder) is induced by the application of a pressure.
5. Glass Capillary Technology: Another technique is glass capillary dispensers. The glass container which holds the fluid (solder) is enclosed by a piezoelectric material which generates a pressure wave (an acoustic wave) initiated electrically. The pressure wave travels to the nozzle and a high local acceleration helps to eject the droplet through the orifice.
6. Ultrasonic Technology: In here an ultrasonic acoustic energy wave generated with the use of a piezoelectric or a magnetostrictive transducer is focused onto the fluid near the ejecting orifice. The pressure of the acoustic wave creates a local pressure gradient and a droplet is ejected. The ejected droplet volume can range from femtolitres to nanoliters and is applicable to our solder ejection of 50-100μm diameter spheres.
The breakup of a liquid (Solder, Ink etc.) jet stream into droplets by the application of a periodic perturbation has been successfully used in the manufacture of solder balls, ink jet printing etc. In the early 1990s, researchers at MicroFab Technologies took out a Patent (No. 5229016) on a methodology and an apparatus for ejecting and producing solder balls. In 1992, researchers at MIT labs developed a system to produce uniformly sized metal (solder) droplets (Patent No. 5266098). These systems comprised of a droplet generator i.e. molten metal container, a vibrator, charging system and a monitoring system. Since these early developments a number of researchers have developed and refined the technique for drop-on-demand (single droplet) and continuous mode (multi-drop) liquid droplet generation.
In the light of this investigation, we adapted to use ultrasonic technology for solder jetting, furthermore we decided to use a Magnetostrictive transducer to generate the ultrasonics deviating from most of the conventional approaches reported in the literature (piezoelectric was still an option if the Magnetostrictive approach failed). The use and success of Magenetostrictive transducer is justified and shown in deliverable D4.1 and the physics of jetting given in D3.1.
The choice of materials for orifice and jetting tank that would allow continuous jetting at high velocities and throughput was investigated with a study of quartz, sapphire, stainless steel, molybdenum and platinum as nozzle materials and stainless steel for the μBGA solder ball production system. We investigated the orifice materials to ensure that the chosen material does not wet the solder and withstands in excess of 100 hours as required by the DOW (page 42). Our chosen material Molybdenum readily withstands 300hours. Full details of this investigation is given in D3.1 here we present in figure 3 the two extreme cases (One very good and the other high wear and tear).

Figure 10 (Left) A 70μm Molybdenum orifice showing oxidation on the periphery of the aperture/orifice where the solder is in contact but otherwise no erosion (Middle) An optical image (looking from the solder side) of a 70μm Molybdenum orifice after many months of use. The orifice itself appears to be reasonably well defined and clear (Not blocked). The measured diameter was 80μm (possibly due to measurement error).The outer area appears to be oxidised (Right) A 100μm Platinum orifice after 8 hours of use, the 100μm orifice was eroded and the diameter after use was measured to be around 408μm and the edges appeared rugged due erosion
In this work, we investigated the wear and tear of five materials quartz, sapphire, stainless steel, molybdenum and platinum when subjected to a molten solder flow to ascertain the suitability as a nozzle material in a solder ball production system. 600 µm orifices were drilled in quartz, sapphire, stainless steel samples and used as an orifice in a solder tank. Using the heaters of an Inkjet printer system, a solder flow was produced at a rate of 1.6×10-4 l/s through the orifice to form solder drops. The observation by SEM of the orifices and their cross-sections after completion of the series of experiments showed no sign of wear, regardless of the material. However, sapphire and quartz were found to be non-ideal candidates due to their brittle nature, making them harder to drill, and easily shattered. Moreover, the significant CTE mismatch between these materials and solder alloys may induce potentially high stress on the orifices with temperature changes (heating up and cooling down phases), which may lead to crack propagation and failure of the sample. While cracking can be mitigated with gradual heating and cooling phases, a ductile material with CTE better matched to ‘solder alloys’ such as stainless steel would be an alternative to form a wear resistant orifice in a solder ball jetting tank system. ‘Real’ life assessment with platinum and molybdenum, clearly indicated that platinum is unsuitable as an orifice material due to the formation of intermetallic PtSn4. However molybdenum was seen to be an excellent material for solder jetting with very little if any erosion after 288hours of usage. It is recommended that Molybdenum orifices be used and commercially procured. It is also recommended that the orifice be changed every 250-300 hours of usage.
Having established the orifice material, we looked at materials for the jetting tank and established that Stainless Steel is best (See Deliverable D2.1 and D3.1). To investigate what material to be used for the solder jetting tank; we carried out a comparative assessment (a paper exercise) to determine how the prospective material, meet our requirements i.e. the μBGA solder jetting systems to be designed is to operate in the temperature range of 200-400°C with a typical operating temperature of 250°C. This requires the jetting tank, nozzle and other materials such as the ‘O’ rings and gaskets to withstand this temperature range and a maximum pressure of approximately 3 bars. Stainless steel is able to operate at this temperature range (i.e. ≤400°C) and withstand pressure. Thus, the proposed stainless steel type would be 316 which can present an increased corrosion resistance and strength at elevated temperatures. Moreover, stainless steel 316 presents important machining properties, thus determining our choice.
A prototype solder jetting system was designed and manufactured. The solder jetting system consisted of the following main items:
1. A stainless steel tank to contain the solder. The tank consisted of the main body and a top and bottom flange. The top flange was provided with a number of entry and exits ports;
a. An inlet port to introduce Nitrogen into the tank and an exit port for the nitrogen outlet. The outlet is connect to a safety valve which would open up if an excess pressure was to exist
b. A locating port with two ‘O’ rings to guide the titanium coupling rod which transfers the acoustic (ultrasonic) wave from the transducer to the orifice at the bottom flange
c. An observation port to direct a laser beam from a laser distance meter to monitor the level of the solder in the jetting tank
d. A thermocouple to monitor the temperature of the molten solder
The bottom flange is designed to take the orifice/nozzle through which the molten solder is ejected
2. Wrap round heating elements (A high and a low power element) to melt the solid solder (pellets) into a molten liquid
3. A low wear and tear orifice to discharge the solder as a stream. The choice of the orifice is based on our inference of the orifice designs and material assessments.
4. An ultrasonic transducer to break up the solder stream into solder spheres. The chosen transducer was a magnetostrictive transducer not a piezoelectric. This choice and the solder stream break up was reported in Deliverable D4.1
5. A triangular bracket to locate the ultrasonic transducer away from the heated tank
6. A XY micrometre stage attached to the transducer, whereby the precise distance of the titanium rod from the orifice can be controlled
7. A titanium rod coupled to the transducer to deliver the acoustic wave generated in the transducer to the vicinity of the solder at the orifice
8. A Mass Flow Controller (MFC) to introduce a controlled and stable Nitrogen flow to maintain a set pressure level to facilitate the ejection of the solder at the orifice
9. A Nitrogen cooling column of sufficient length for the molten solder spheres to travel on breaking up and cool down to form solid spheres
Full details are given in Deliverable D3.1
Work Package 4: Development of ultrasonic techniques and prototype for breaking up solder jet
Techniques for the Production of µBGAs
There exist several technologies by which µBGAs can be manufactured. The technologies used are: Gas atomisation where the molten solder is atomized with the use of a high velocity gas into spheres with a wide size distribution (solder powder), Solder foil or solder wire is stamped or cut respectively, then melted and solidified in a temperature controlled bath such as an oil bath. This technique is suitable for large solder spheres > 300µm and Solder jettison using ultrasonics techniques. This latter technology is best for mono-sized spheres and for spheres with diameters less than 300µm.
Solder jetting is defined as the process by which a solder stream is broken up into individual solder spheres, in a uniform and controlled manner. This process is initiated by the solder exiting through a micron size nozzle as a solder stream. The breaking of the stream into solder spheres by large is due to an external instability being imposed to the solder stream and due to surface tension drawing the solder into spherical solder balls.
The external instability (see description of WP6 below) in the form of a vibration generates ‘necking’ in the solder stream (See deliverable D4.1). This is crucial to the break-up of the stream and along with other factors.
There are number of factors that influence the solder breakup. The solder droplet formation depends on the fluid (molten solder) properties, the orifice shape, length and width, the drive voltage shape, amplitude and frequency, back pressure and many others. All of these need optimisation to obtain droplets in a controlled manner. The most important of these factors are:
1. Viscosity
Viscosity plays an important role in the solder breakup and the creation of a solder droplet (sphere). Increasing solder viscosity (temperature has a strong function on viscosity) dampens the acoustic wave used to create the droplets. In one respect this has a positive impact in that it dampens the instabilities that lead to satellite formation. Furthermore for a given velocity, increased viscosity requires a greater drive voltage (voltage amplitude). It also negates the effect of the nozzle diameter, the droplet size decreases when the viscosity is high (provided velocity is kept constant).
2. Density
Temperature has a weak dependence on density and thus does not have a direct effect on the solder breakup but would as mentioned above affect acoustic speeds and/or bulk modulus of the molten solder. These have a minor effect on the optimum waveform timing and voltage amplitude requirements.
3. Surface Tension
An increase in the surface tension has again a small impact in that an increased drive voltage will be required (for a given droplet velocity), also high surface tension would require special consideration of the material used for the orifice as the wear-out may be high. On the other hand, a very low surface tension may result in air ingestion, particular at high drop velocities.
4. Drive Waveform (frequency, pulse width and amplitude)
Frequency: The types of drive waveforms that can be used are many: sinusoidal, square, triangular and trapezoidal. The drive waveform ‘pushes’ the transducer and then returns it to a rest state and this is repeated over and over in a continuous or a drop-on-demand jetting system. In the case of the latter as and when required. Given a finite rise, fall and dwell times of the waveforms the molten solder (fluid) has acoustic resonance due to compressibility effects, initially a negative pressure is created and then a pressure rise on the return to the ‘rest’ state. Even after the droplet ejection, there are residual acoustic oscillations; to cancel these out a bipolar pulse may be used. Hence in order to obtain a steady and controlled droplet formation, the next pulse has to wait until the last waves are fully decayed. In other words the optimum frequency is limited by the decay speed of the acoustic wave, which will depend on a number of factors to do with the acoustic cavity. The formation of the jet stream with ‘necking’ nodes depends on both the orifice diameter and the vibrational frequency as illustrated in figure4.

Figure11. Illustrates the effect of frequency and orifice diameter (a: No vibration, b: f=20.32KHz d=90µm, c: f=20.32 KHz, d=180µm) (deliverable D4.1)
Pulse width: The droplet volume is a linear function of the pulse width for a given drive voltage.
Drive voltage (Amplitude): For a given pulse frequency, the droplet volume increases linearly with increasing amplitude once the minimum drop velocity is exceeded. The non-linearity at drop velocities below the minimum is due to the viscous and surface tension restraining forces.
5. Orifice Diameter
As the orifice diameter increases, the amplitude needs to be increased to maintain the same drop velocity due to acoustic impedance effects and the increase in the volume of fluid (molten solder) is less than the diameter squared relationship due again to the acoustic impedance effects.
6. Pressure
Pressure is an important factor in ejecting a solder jet stream. With no differential (excess) pressure, no solder jet stream can be expected. Once the atmospheric pressure (~1 bar) is exceeded a solder stream can be initiated.
This leads us conveniently to the type of transducer to be used for our application.
There are two fundamental transducer designs used for power ultrasonic applications, magnetostrictive and piezoelectric. In the case of the latter, application of an electrical voltage creates a mechanical strain in the material (PZT) or vice versa. In a magnetostrictive material (such as Terfenol-D) the application of a magnetic field induces a mechanical strain in the material. An advantage of the piezoelectric transducer is that it is not limited in frequency in the typical ultrasonic frequency range of 20 to 200 KHz. However the piezoelectric transducers require a much higher voltage (usually a factor of 10) than magnetostrictive transducer which can only operate to a maximum frequency of approximately 30 KHz. Our choice of Magnetostrictive Terfenol-D transducers have more displacement and force and unlike others this property does not degrade or change with time and number of times used. Moreover it has a microsecond response time. These attributes make Terfenol-D magnetostrictive transducers an attractive choice for our application despite the need to house the transducers away from the hot molten metal; the operating temperature of the transducer is around 200°C (the Curie temperature is 380°C). Another attribute that was considered to be positive is that magnetostrictive transducers operating at a maximum rated frequency of 30 KHz will have their first sub-harmonic at ½ operating frequency which is within the adult human hearing range (typically 18 KHz maximum). From an experimentalist’s perspective able to hear audio would be regarded as a benefit (provided it is not harmful to once ears). For these technical reasons and the novelty of using magnetostrictive transducers in solder jetting, our preferred choice was the ‘Etrema Terfenol-D’ magnetostrictive transducer.
Full details of the experimental work carried out in WP4 are given in deliverable D4.1 and subsequent deliverables where the break up and roundness etc. was optimised. The main outcomes of this WP were the establishment of; a frequency range for different size of spheres with a fixed orifice diameter, a pressure just above 1bar, an operating solder temperature range, a working amplifier setting, drive amplitudes (there is a clear indication that the effect of the drive voltage amplitude is that as amplitude decreases the droplet size also decreases for any given frequency. This is an important result in that along with the nozzle diameter and frequency, this gives another parameter to control the solder droplet volume and thus the droplet diameter) and the important need for Nitrogen during the ejection and solidification process (more and full details in deliverable D4.1 and other deliverables). Another important result is that the established rule of the sphere diameter being at least twice the orifice diameter is shown not to be the case with orifice diameter and sphere diameter being generally equal; we believe this can be controlled by waveform frequency (and amplitude) and velocity (pressure).
We have demonstrated a solder jet stream being ejected from a nozzle of 100µm in diameter at a pressure of 1-1.5bar and that it can be broken up into droplets.
Work Package 5: Development of techniques and prototype for measuring solder ball sizes
This WP5 is to develop imaging solutions to image at least 30% of the fast moving (up to 10m/s) solder spheres and analyse the images in real-time to determine the diameter and roundness of the spheres with micron resolution.
In this regard to develop techniques for observing and measuring solder ball sizes; methods were investigated that included strobe lighting and constant light with fast shutter. We conclude that the constant light with high-speed camera has better performance as the strobe illumination cannot be fully calibrated or controlled. The main reason is that the speed of spheres is not constant and therefore we cannot tune the strobe's frequency with high accuracy. By using a high-speed global shutter we can capture images at maximum time of 1μs. In later work the strobing is adapted instead of continuous for image capture. Moreover, apart from the methods mentioned above a laser beam generator was utilized to further evaluate the acquisition properties of the system. By directing the stable ray of the laser light source near the nozzle of the system we created an environment of high contrast thus allowing the camera module to capture images of sharper contrast and thus allowing an improvement in the measurement procedures.
In addition to the new laser source a blur reduction algorithm was also developed to allow the clearer acquisition of images. Blur occurs when the stream of spheres is out of focus. The phenomenon is mainly linked to voltage and pressure miss-calibrations as well as calibrations relevant to the size of the produced spheres. In our system the use of the telecentric lens provides a constant size for the out of focus spheres. This is determined to be about 10% of the measured diameter. When a sphere is located in the frame multiple cross edge lines are generated crossing the sphere and measure the variation of pixel gain. That variation is used to determine the blur phenomenon allowing the algorithm to compensate.
In order to handle the demanded application of inspecting spheres with up to 10m/s velocity and window of processing under 100μs, an optical system as described in detail in deliverable D5.1 using a CMOS camera and a telecentric lens is used for image capture. The system consisted of a Optronics CamRecord CL600x2 camera, a telecentric lens LENSAGON TC-60-70C, a camera link cable, a LED backlight with a brightness of 5500cd/m2 and a X64 Xcelera-CL PX4 frame grabber, along with other sub components such as X-Y-Z micromanipulator stage and the camera/lens platform.
The physical sphere diameter measurements from a sample of collected spheres (it is not possible to relate the optical measurement analysis to the precise sphere) have been co-related to optical measurements. The real-time measurement and analysis is embedded in WP9 (D9.1) and is also discussed in deliverable D10.1.
In addition to this optical imaging, the possibility of using radiographic imaging was looked at using a test on an experimental test bench; the results indicated the radiography technique needs further refinements for imaging in real time but is acceptable for inspection. During the trials it was established that the existing detector as having a slow response to meet the requirements of UBGA. Following this is investigation; a high speed detector which would meet the requirement was evaluated. The results of this are given in the report ‘µBGA Inspection Trials May 2013’ in Appendix The conclusion as with the earlier radiographic attempt is that radiography while being a useful technique for inspection needs further testing and improvement for use in real time sphere imaging at high speeds.
Work Package 6: Development of techniques and prototype for separation of solder balls
Prevention of Co-Joining of Solder Spheres
Lord Rayleigh showed that when a waveform is applied to a jet stream, provided it’s wavelength is greater than the jet stream’s circumference, the jet stream will become unstable and exponentially grow in time as exp(βt). Our calculations given in table 1 of deliverable D6.1 shows with the exception of the wavelengths (and the corresponding frequencies) highlighted in blue, the frequencies and velocities that can be used. This exponential growth, ultimately results in the breakup of the jet stream into solder spheres. Having said this, in our work we aim to achieve this breakup with the application of a periodic ultrasonic waveform rather than leave to ‘nature’, thereby avoiding as far as possible spheres (droplets) of different sizes. Nevertheless, as the solder sphere traverses through the cooling column they experience a drag force, this affect their dynamic behaviour. Unless each sphere is identical in size and follows one another in a ‘straight line’, different spheres will have different velocities and there is a strong likelihood of the spheres co-joining. We discuss this in detail in D6.1.
To prevent this co-joining, the most common technique is to induce an electrostatic charge (electrostatic induction) of the same polarity onto the molten solder spheres whereby, by virtue of like charges repelling each other, the spheres are prevented from colliding and merging together. This can be accomplished by having a high positive voltage ring with respect to the grounded molten solder stainless steel tank; other configurations of electrodes such as charging plates (positive and negative plates) are also possible. We have explored this; again full details are in deliverable D6.1. The results presented in D6.1 indicate that ~15% of the spheres could be co-joined.
We developed a novel ‘Acoustic Separation’ technique (figure 5 above) for the prevention of co-joining. We have applied for IPR protection with respect to this acoustic technique. When an acoustic waveform is imposed the solder spheres experiences a time-averaged force that drives the spheres to the pressure nodes or the anti-nodes of the wave or somewhere in between, this dispersion makes the probability of solder sphere co-joining low. We use an acoustic force that has an axial (longitudinal) and transverse component. The axial component which acts in the direction of propagation of the acoustic wave, moves the particles (solder spheres) to the node or anti node if a standing wave is set up, if not it disperses the solder spheres throughout the waveform. The latter is what is being used by us for the separation of solder spheres to prevent co-joining. The acoustic force is proportional to the third power of the solder spheres radius, that is to say the size of the solder sphere has a strong influence, larger the diameter, larger is the force. This acoustic force (F) on the solder sphere is always counter balanced by the Stokes drag force (scales with the spheres diameter). The net result is that the acoustic separation is linearly proportional to the solder sphere diameter. Thus in combination with an acoustic waveform and nitrogen flow (laminar flow, not turbulent), the spheres can be separated in a size dependent and a continuous manner. In this invention, the acoustic frequency is arranged to provide a wavelength which is twice the diameter of the chamber and a wavelength range of 100 – 500Hz was used. The working of this was demonstrated with a video at the Year 1 Review Meeting held on the 4th November at REA in Brussels.
The acoustic separation was successfully implemented on solder spheres of diameter that were ejected from the orifice. The spheres travelled in a nitrogen atmosphere at velocities of around 4ms and were collected at the bottom of the cooling column. This is discussed in deliverable D6.1.
Separation of Solder spheres of different sizes
The work done in WP9 is to adjust system ‘run parameters’ to produce mono size spheres; however in the event this does not work or other additional techniques are needed, we explored the separation of spheres produced in a single run into different sizes using:
(1) the charge induced on a sphere. The methodology uses the concept of charge to mass ratio, as the mass increases the charge increases and vice versa. This gives us the possibility to use a second high voltage electrode to deflect the solder spheres according to the induced charge the spheres carry; at least in principle this is possible. Having done the ‘sums’ we believe electrostatic separation of spheres of different sizes is indeed very demanding and due to the very small displacement is not very practical with low to medium voltages. This is fully detailed in D6.1.
(2) A second method we developed and used for separation solder spheres of different sizes was centrifugal separation. A centrifugal separator was built (shown in figure6 above).This centrifugal system has shown much promise in separating solder spheres of different sizes though some further optimisation is necessary.
(3) A third option was manual sieving, used in today’s industry.
A number of different sieve sizes are used, one of which is shown in figure12 (a) and (b).

Figure12a. A mechanical sieve which will allow 50μm diameter and under to be filtered out leaving those spheres larger than 50μm in diameter
Figure12b. An image of the inner ’mesh’ of the sieve
(4) We also considered the systems shown in figure 13

Figure 13 Inclined and horizontal mechanical sieves (LHS) Inclined (RHS) Horizontal
The electromechanical sieves of figure 13 have the potential for separating spheres with a very good yield (see deliverable D6.1 for calculations). This scenario was not tested experimentally as the ‘Novel’ centrifugal separation was promising in real experiments and field trials. This work is reported in Deliverable D6.1
Work Package 7: Integration of prototype system (WP2 to WP6)
The integrated system is shown in figure 14 and fully discussed in D7.1 and the assembly and running instructions given in deliverable D14.1

Work Package 8: Development of man-machine interface for the control of the lab based prototype
This work package targeted the development of a Human Machine Interface (HMI) for providing the necessary control functions over the system parameters. The HMI was designed in a way that can incorporate all the control functions implemented by the linear control algorithms and the artificial neural network through a straight forward GUI (figure 15). Furthermore the control algorithm and its respective libraries were optimized allowing an increase in the performance of the HMI. Finally the heater control algorithm was updated and improved to support and provide control over 3 thermocouples

Figure 15: Control GUI
More on this is discussed in deliverable D8.1 also in deliverable D9.1 which uses the HMI.
Work Package 9: Development of self compensation mechanisms for the elimination of variations in sphere size (Feedback control system)
WP9 was the development of the control algorithms that would allow the automatic management of the functions of the system. Based on the work done two types of algorithms were developed. One followed a linear approach and the other a non-linear one. The algorithms are initialized through a GUI that prompts the user to provide a production target as well as the type of control algorithm and the system would then operate autonomously. Figure 16 provides a snapshot of the ANN control tab that is incorporated in the HMI developed in WP8.

Figure16: GUI – ANN setup TAB
Based on the test results we observed that the system reached a 96% percent accuracy level when no control was activated, a 98.2% accuracy with the linear control algorithm activated and an accuracy exceeding 99.4% with the ANN activated. Roundness levels were determined to be between 95 to 100% and in all cases the tolerance level was no more than ±5%. The testing proved that the linear control algorithm but more importantly the ANN could provide stable production rates. The linear control algorithm provided fast adjustment in a sort amount of time but with problems in stability after long operation periods whereas the ANN required time to train but provided accurate and stable production rates for long periods of time.
During the internal validation procedure the control algorithms were evaluated for their production capabilities by comparing their performance with actual samples that were taken from the camera module and samples that were examined under a microscope. The target of the evaluation was to confirm that the results provided by the system were co aligning with manual measurements. Tables 3 and 4 provide the average values of these measurements both for linear control and the neural network. Also figures 17 and 18 provide a graphical representation of systems performance in terms of diameter.
Table 3: Validation results linear control algorithm
Average values Logged Results Snapshot Images Microscope
Diameter 151.66 μm 146μm 141μm
Roundness 94.89% 97% 96%

Table 4: Validation results ANN
Average values Logged Results Snapshot Images Microscope
Diameter 144.52 μm 146μm 143μm
Roundness 95.11% 98% 97%

Figure 17: sphere distribution – Linear Control
Figure 18: sphere distribution – ANN

Based on the validation results the stability of the control algorithms was confirmed. Furthermore the validation process proved that the ANN was able to provide a stable production of spheres with even smaller diameter. The algorithms proved that based on the orifice used they can provide a stable production rate that is within the expected target.
A web application was also tested for its imaging and remote control functions. The web application was successfully able to provide a remote user control over the system and store all relevant data to an OPC format as dictated by the implemented OPC server running the application. The idea behind this connection was for the software to automatically input data to that database and through that database the data can be accessed via the website. This work was reported in deliverable D9.1.
Work Package 10: Field Trials of the µBGA prototype
The field trials of the prototype originated with the installation of the system (assembling the system) at one of project partner premises (ABIS in Krakow, Poland) with the aim to determine the quality of its operation and the output (solder spheres) and following this trial to modify the system where necessary. The first trials conducted were aimed at performance validation with the closed control loop feedback and the second set of trials aimed to benchmark performance of the open loop system. Prior to the field trials, extensive trials were carried out at the RTD performer KCC to optimize the system thereby minimizing the modifications needed post Krakow field trials.
The first of the two field trials took place between Monday 22nd of April and Friday 26th of April 2013 and were conducted by Michał Calik and Piotr Bistron (ABIS), Ioannis Friganiotis (IKH) and Aurel Duhalm (KCC). The second field trials were between 3rd and 5th of July 2013 at ABIS and were conducted by Rafał Łopatka (KCC) and Michał Calik (ABIS).
The assembly and the running of the system followed the step-by-step instructions given in Appendices of deliverable D14.1. On running the system, spheres were collected and subsequently samples from each run were randomly selected and distributed in vials (the vials were marked A, B, C, D etc. so as not to prejudice the findings) for analysis to Tyndall National Institute in Cork, Ireland and to Bob Willis at SMART GROUP UK. Also samples were sent to CIT for radiographic inspection, the results were inconclusive in accurately determining the size and shape (Appendix 1).
The results of their (Tyndall and Smart Group UK) analysis and operational problems encountered in running the system in Poland are reported in deliverable D10.1 and the appendices therein. The main conclusions are:
Spheres of nominally 100µm and 50µm can be achieved with the current system though the yield is not high (the highest yield achieved at KCC was 85%).
Results of the field trials points to further optimization as been necessary to address the following problems and achieve better productivity for commercializing the system, the issues to address are:
1. Nozzle needs further redesigning to minimize blockages (A design is proposed in Appendix D of D10.1)
2. The camera needs to have an ‘Auto-focus’ capability with a deeper depth of field and a larger field of view
3. The software analysis of images needs to be corrected
4. Incorporation of the ‘crap or non-compliant’ percentage prediction into neural network (NN) system and adjustment of the image processing algorithm in order to obtain more reliable online inspection of the output is desirable
5. Provision for user to input target diameter on the front GUI needs to be provided
6. Lower ejection velocities and lower pressures in the cooling column needs to be used to prevent scatter of spheres and ensure ejected spheres move in a straight line (will lower co-joining, help with optical focus and lower sphere diameter)
It is strongly believed once the above are implemented prior to commercializing the product for the manufacture of spheres 100-50µm at a high volume, the system will truly be ‘Fit for Purpose’. Due to financial and time constraints resulting from outstanding issues with regard to the project, it has not been possible to implement the above desirable modifications before end of May 2013, the project end.
Work Package 11: Development of µBGA placement prototype
A comprehensive discussion is given in deliverable D11.1 and a good overview was given in the 3rd year periodic report, this is reproduced below.
During the µBGA project, solder spheres as small as 50µm in diameter were produced. The production of such small spheres present many technical challenges and more so in handling and assembling devices using such small solder spheres. In this regard techniques for solder ball pick and placement were developed. The work involved 3 stages:
1. Design and evaluation of vacuum pick up heads.
2. Design and evaluation of solder ball sorting trays.
3. Direct solder ball transfer technology.
The system developed is for placing the solder balls (spheres) such that they could be used in high density interconnects applications. The requirements for the placement method developed here is that it be quick & repeatable to carry out and that it could be readily adaptable to high volume manufacturing situations.
It is necessary to assemble these small spheres in array format. The placement in this dimensional range exceeds the capabilities of current PCB technology (typically limited to 60µm track & gap) and so is effectively the developed done in this WP is a technology for next-generation of high-density interconnects. The placement techniques developed here however are also applicable to solder of larger sizes and thus can be used with current PCB technologies where solder of suitable size (i.e. generally larger than 100µm in diameter) are used. Because of the limitations of PCB manufacturing technology, at the smaller dimensions, it was decided to carry out much of the process development work on silicon demonstrators manufactured in a CMOS facility where µm-scale track and gap dimensions are easily achievable.
The approach employed involved the development and fabrication of a tool set for use with a commercial flip-chip alignment & bonding system (which is a standard high-volume manufacturing system). The tool set consisted of a set of vacuum pick-up heads & placement holders (fabricated using silicon MEMS technology) and interfaced with the flip-chip system. The technique developed is described in detail in Deliverable D11.1 while the samples produced to demonstrate the use of the technique are described in deliverable D12.1.
The achievements of this work are:
1. A placement technique for solder spheres down to 50µm in diameter has been developed and is compatible with the requirements of a high volume manufacturing environment has been developed and has been successfully demonstrated (refer to D12.1).
2. The developed technique exceeds the requirements of current advanced PCB manufacturing technology, but can still be used with the larger current PCB dimensions. It is therefore an approach which can be used with the current generation of high-density packaging and will be an enabling technology, for future generations of high-density interconnect which will eventually be used when further advances are made in printed circuit board manufacturing. Currently, at the smaller dimensions, it is best suited to silicon devices and substrates which can be readily fabricated with much smaller track and gap dimensions than is possible with PCBs.
3. The technique makes use of a standard manufacturing tool (a flip-chip alignment & bonding system) which is interfaced with a specially developed tool set for handling the solder balls. The tool set can be fabricated in a very cost effective manner (and in high volumes) using silicon MEMS processing.
The approach is illustrated in figure 19 below while an example of one of the pick-up heads is shown in figure 20.

Figure 19:- Illustration of the operation of the novel Silicon MEMs pick up head for use with flip-chip alignment & bonding systems.

Figure 20:- Vacuum hole array - 60μm holes - 150μm pitch.
Once established, the placement technique was used in WP12 to produce a range of test structures and a functional demonstrator which verified and demonstrated the process successfully.
Work Package 12: Demonstration of the capability of the prototype µBGA system by producing the World’s first PCB with µBGAs of 75µm
A full description is given in the deliverable D12.1 we reproduce here an overview from that reported in the year 3 periodic report.
To address the tasks of WP12, test structures were designed and fabricated to test these small spheres in an assembled device format. Due to the small nature of these spheres standard printed circuit board technologies could not be used in there assembly. Therefore it was necessary to use silicon device processing for both the test structures and the substrate board. These were then die attached to standard PCB and wire bonded out to tracking onto which wire could be soldered, to allow for testing.
Three solder ball sizes were selected for the evaluation. Spheres of 50, 75 and 100μm’s diameter were selected to be assembled. This meant that pad sizes appropriate to these ball sizes were required on the test structures. In addition to this, two different pitch sizes for each ball size were used to evaluate assembly (please see deliverable D12.1 Table 1). The design of the daisy chain structures means that electrically the spheres form 2 chains. For a fully functioning device there should be continuity within each chain and one chain should be isolated from the other. This will mean that all the ball joints are in contact and there is no shorting between spheres. It is also possible to measure at different points along the chain to pinpoint any failures that may occur. The recorded resistance values (Table 2 in Deliverable D12.1) were low (indicating good quality joints were formed) and there was low variability between the measurements recorded for the individual daisy chain structures. Furthermore cross sectioning (above figure 7, Middle) of a number of devices was carried out during the evaluation of the demonstrator technologies. From this it appears that the UBM is as expected and the solder has wetted to the pads in all cases.
Finally a functional demonstrator (above figure 7, RHS) incorporating µBGA interconnect was also developed. It was agreed in the early stages of the project that this would be a small battery powered device which would generate a set of numbers for the “Euro millions” lottery. The requirement for the demonstrator was to design a circuit which would generate 7 random numbers in the range 1 to 50 and two additional (lucky star) numbers in the range 1 to 11 when a button was pressed. The device would then remain active for a period of 30 seconds during which the numbers could be marked on the lottery play slip and then return to sleep mode to conserve battery power.
Prior to the manufacture of the ‘Euro Million’ device, the below was carried out:
1. Silicon test structures onto silicon substrates using the placed spheres were assembled. These were then subjected to a programme of tests to verify their integrity and their quality of assembly. Examples of the test structure samples assembled are given in figure 3 below.
2. The development and fabrication of a functional demonstrator – this was a Euromillions lottery number generator which was designed to incorporate the µBGA spheres as interconnect between a motherboard & daughterboard, the latter carrying the main IC in the device. The prototype functional demonstrator device is shown in figure 4 below.

Figure 21:- Final miniaturised version of Euro Millions number generator device using µBGA solder sphere interconnect.
Full details can be found in deliverable D12.1.
Achievements of this work are:
1. The placement techniques developed in WP11 and WP12 (refer to D11.1 for details of the placement techniques) were used to build demonstrator circuits in a reliable and repeatable manner
2. A successful assembly & verification of test structures using 50, 75 and 100μm spheres has been demonstrated.
3. A functional circuit (Euro Million) using 75µm µBGA solder spheres has also been successfully demonstrated
WP No. 13: Training and Dissemination
The training and dissemination which was mainly entrusted to the SME-AGs has been promoted via:
1. The project website www.ubga.eu
2. the SME-AG website (http://www.smartgroup.org/) has news and progress
3. webinars/meetings in UK, Ireland, Greece (various will be given in Deliverable D13.2 and D13.3)
•Surface Mount Technology Failure Analysis Workshop – Tyndall National Institute, 20th Sept. 2012.
•Surface Mount Rework of BGA, PoP, CSP & QFN components, Enterprise Ireland, 11th July 2013.
4. You Tube videos
(µBGA - "What are the biggest challenges in reworking uBGA with underfill?", http://www.youtube.com/watch?v=NmnfWdHgqOc
µBGA - "What are the challenges for placement of uBGA during assembly?", http://www.youtube.com/watch?v=ADkVKk29L9Q
Steve Dowds in "What are the common defects with fine pitch/flip chip reflow" http://www.ubga.eu/node/3177
Peter Marshall in "What is important when setting up the printing process for fine pitch?" http://www.ubga.eu/node/3177
Keith Bryant in "What are the limitations for x-ray inspection?" http://www.ubga.eu/node/3177
5. training material developed as part of Deliverable D14.1
6. attending meetings, exhibitions and personal contact
7. preparation of technical papers, posters
• Journal paper prepared for submission to the WSEAS Transactions on Systems and Controls(http://wseas.org/wseas/cms.action?id=4073 ) entitled “On the control of micro Ball Grid Array (μBGA) production systems
8. the coordinator has been in touch with interested parties on exploiting the know-how on completion of the project
9. a patent was applied for on acoustic sorting but has now lapsed due to financial constraints
10. the consortium has carefully looked at viability of production technology both from a technical and commercial/financial view point and has come to the conclusion that the technology will be embraced by the semiconductor packaging industry and the market is substantial and with our higher expected yield will challenge the current far-eastern manufacturers. This view has been reaffirmed recently after further investigation and discussion with experts in the field
While much has been achieved by the SME-AGs and the RTD’s, this work which was seen as secondary to the technical challenges and was greatly hampered by the very long process of resolving the issues related to the project which has caused a temporary suspension of funds and has not been resolved to date.
The Deliverable D13.2 and D13.3 is in preparation at the time of writing this final report.

WP No. 14: Development of Guidelines for the application of developed technology
In the deliverable D14.1 which is associated with the tasks of this WP 14, we give full details of:
1. Guidelines (requirements) for equipment (task 14.1) details of the system components are given
2. Guidelines for the use of the system, in-service operation (task 14.2) here we give two operational routes i.e. how to run the system
3. Guidelines for training and certification (task 14.3) detailed step by step instructions on assembly and running of system are given for training of personnel
4. Dissemination of results to national and European standards committee, we elaborate on the steps which we have undertaken in dissemination, namely workshops, newsletters, webinars, websites, e-Shop and others but not national and European committees, though at European level we are planning for a special session in late 2013 or early 2014.
Please refer to the deliverable D14.1 for a comprehensive discussion on the above.
WP No. 15: Production facility set-up plan (PFSP)
This WP involved: Task 15.1 (Alloy preparation for high volume production): In this regard we have moved away from conventional alloy preparation at the production facility to procuring quality assured solder pellets manufactured to industrial standards thereby saving considerable costs which allows us to be competitive with global producers. Task 15.2( Solder sphere production for high volume production): In Deliverable D2.1 (Solder Feed tank), D3.1 (solder Jetting tank), D4.1 (Ultrasonics for solder jetting), D5.1 (Imaging of Spheres), D6.1 (Separation, collection and sorting of spheres in liquid free state i.e. as solid spheres), D7.1 (The integration, assembly and running of system) and D9.1 (the feedback system) we outlined the key elements of producing of solder spheres at a high volume. Task 15.3(Solder sphere inspection for high volume production): In figures 8 to 12 (of Deliverable D15.1) we gave a snapshot of the imaging and image processing. Figure 8 showed the camera image capture and figure 9 a random selection of spheres measured with an optical microscope. The image analysis was shown in figure 10, with the results (sphere diameter and roundness) indicated at the bottom left hand corner of the GUI panel. The imaging has a dual camera operation and a frame of the captured of the images was shown in figure 11. Also shown in figure 12 was the video capability. Use of a second camera is optional and software allows for this. Task 15.4 (Man-power requirements for high volume production): With the µBGA system, many of the conventional operations associated with solder alloy ‘in-house’ preparation has been dispensed and replaced with the procurement of solder pellets, pre-cleaned and with very low oxygen content and impurities. The cost of using solder pellets ‘ready to use’ without further cleaning etc., far outweighs the cost of conventional route. It is estimated that the man power requirement is 4 persons running two parallel system on a 8 hour shift per day; two for the running of the system and two for sifting of spheres which have already being centrifugally separated and batch analysis (optical microscope analysis) and packaging, against 39 personnel per day with conventional production facilities. Task 15.5 (Factory space and power requirements for high volume production): In Figure 15 of Deliverable D15.1 we showed a floor plan for a production facility with a manufacturing area housing two complete systems, a laboratory (and quality control) area, a packaging and sales stock area. In addition a Goods-In and dispatch area (Loading/Exit) is shown, this allows for current and future needs of a µBGA production facility. For the production system a class 10,000 clean room of 20 square metres was considered more than adequate. The total factory space was in the region of 150-200 square metres. The requirement in terms of space is very modest and was planned with future expansion in mind. For each system the minimum space required is 2m x 2m x2.5m (height) and the power is a mains supply of 240V ac, single phase. Nitrogen required for production will be in cylinders (bottles) within the production room or piped from outside e.g. cylinders in an outside corridor. The production room/s will have Nitrogen/Oxygen sensors installed for safety. Task 15.6 (Equipment and consumables requirement for high volume production): The main consumables in terms of solder usage were estimated to be 2Kg (approximately €58 per Kg)for the daily production of 500 million 75µm diameter solder sphere and a Nitrogen cylinder (200 bar, 1.89m3) to last about a 0.5-1 month (cylinder Cost ~€26). In addition to the above the other main consumable is the molybdenum orifice at a cost of €58-€70 and our experience suggests that we change the orifice every 250-300 hours of usage. Full details of equipment were given in Deliverables D2.1 to D9.1inclusive. Task 15.7 (Guidelines for training and certification for high volume production): The operational and building of the equipment along with training and certification were given in great detail in Deliverable D14.1. The deliverable D15.1 along with deliverable D2.1 to D9.1 gives a full and comprehensive account of the system, consumables, production facility set-up floor plan, equipment design, building of system and operational and training guidelines. Task 15.8 (Sources of funding for the implementation of a production facility for high volume solder sphere production): A number of financial sources are available these are discussed in full in the deliverable D15.2 and listed below. A word of caution, in sourcing funding the ‘first rule’ is to be aware that the slogan ANY MONEY IS GOOD MONEY is NOT CLEARLY THE CASE and where and how you finance a small business might be the difference between success and failure. Of the many financial sources, we list 10 of the most likely sources that would fund a production facility of this nature: Venture Capitalists (VCs), Banks, Enterprise Guarantee Scheme, Local, Regional, National and EU Development funding, European Regional Development Funding (ERDF), Leasing, Angel Investment/Equity Funding, Friends and family, Customers or Competitors, Vendors, Post ‘Proof of Concept’ Funding Grants.Please see the deliverable D15.2 for detail discussion on the 10 financial sources
WP No. 16: Project Management and Coordination. This WP16 has two associated tasks: Management and Coordination activities (Task 16.1) and Periodic progress reports, Technical deliverables and final report (Task 16.2). In this regard, the activity has involved regular online meeting, consortium meetings, RTD and Exploitation meetings, progress reports and telephone and written discussions, along with visits to partners and reading, financial control, checking deliverables and finally uploading same to SESAM/ECAS. In addition it has two deliverables (D16.1): A draft PUDK (an interim PUDK which has been submitted and accepted) and a final PUDF. The final PUDF, at the time of writing this final report is in preparation.

Potential Impact:
The semiconductor industry is the driving force behind major advances in computing, electronics and man’s quest for advancement in many fields ranging from medicine, biosciences and others. Advances in chip functionality, the speed of processing power have enabled individuals to hold computing power unthinkable even a generation ago. With this came portability of telephone communication (mobiles), enormous computing power in the form of laptops to name a few. Portability requires compressing electronics to smaller and smaller ‘real estate’. It is this that this µBGA project addresses in developing smaller solder spheres for connections of the high density of chip functionality to the outside world by its use in electronic packaging. This facilitates companies to create, access, and analyse data rapidly, improving individual and business efficiency and developing new markets within the national and global economies.
With the semiconductor industry unable to keep up with the so called “Moore’s Law,” the prediction made by Gordon Moore, cofounder of Intel, that the number of transistors per chip (transistors per logic chip increased from 3.1 million in 1994 to 1.7 billion in 2005) in a semiconductor device would double every 2 years, it soon became apparent that the way forward was rooted in exploiting the potential of nanotechnology. This has driven circuits within chips to submicron features with regard to the fabrication of integrated circuits (ICs) and thus higher functionality and smaller foot-prints, whereby smaller packaging and thus smaller µBGAs are needed.
Many factors helped the industry realize the achievements we see today, but without the strategic work done under the US NTRS and its successors, the International Technology Roadmap for Semiconductors (ITRS), many of these achievements would not have been possible. Similarly, the key to having a high impact here in Europe, is European investment in a component (µBGA sphere production) taken together with the packages that the chips go on and the PCBs would create a multimillion packaging industry in Europe (currently much of this work is being done in the Far-East) The µBGA solder sphere manufacturing system is capable of delivering spheres of 50 and 75µm in diameter and larger.
The potential impacts are:
1. security of supply to the European chip and packaging industry
2. package miniaturisation with the use of 75 and 50µm balls as our placement trials indicates
3. importantly availability of a European manufacturer along with some other trends that has in recent times begin to appear, is likely that the packaging of devices will come back to Europe (currently this is outsourced to Asia). Create a multimillion packaging industry with many socio-economic benefit
a) though the direct employment in producing µBGA spheres is small the employment in chip packaging industry is very high
b) more test houses and equipment manufacturers will also be a secondary positive impact
c) new ‘Place and Pick’ equipment to handle 50µm spheres would have secondary impact in the development on these equipment
d) more research as miniaturisation and associated ‘problems’ need addressing, for example heat dissipation, interconnections
4. miniaturisation is also expected to lead to further higher functionality, higher reliability and a cost reduction
5. imaging and analysis software with neural networks used here may have wider application in vision systems
6. the novel ‘Acoustic Separation’ technique has a potential impact as a retro-fit system for sphere separation and the centrifugal separation system has the potential as an automated sphere size separator which again will replace the manual sieving and can be retro-fitted
7. Acoustic separation has application in other industries such as bio-medical
8. the technology of micron size droplet dispenser has potential use in other areas of application e.g. Bio-medical arena
9. it is also expected to be in a position to manufacture high temperature solder spheres using Sn10Pb90 or Sn5.0Pb93.5Ag1.5 this high temperature solder is important in automobiles, defence and space

List of Websites:
www.ubga.eu

Company
short name Company full name and location Web site Contact Persons Contact details
KCC Kingston Computer Consultancy Ltd, UK www.kccltd.com
Aasim Khalid
Simon-Peter Santospirito
Ravin Ginige [e-mail]
[e-mail]
[e-mail]

SGUK (SMART-UK) Surface Mount Technologies Limited, UK www.smartgroup.org
Tony Gordon
Bob Willis [e-mail]
[e-mail]

SEPE Federation of Hellenic Information Technology and communication Enterprises, Greece www.sepe.gr
Yannis Sirros
Pantelis Nikolaidis
Anna-Maria Tsakalaki [e-mail];
[e-mail]
[e-mail]

APEMETA Associacao Portuguesa De Empresas De Technologias Ambientais, Portugal www.apemeta.pt
Ana Cunha
Jose Costas
Carlos Iglezias [e-mail]
[e-mail]
[e-mail]

Tyndall Tyndall National Institute, Ireland www.tyndall.ie
Finbarr Waldron
Cian O’Mathuna
Frank Stam [e-mail]
[e-mail]
[e-mail]

IKH Ainoouchaou Pliroforikis AE www.iknowhow.gr
Xenia Bania
Iwannis Friganiotis
Gregory Kotsikaris [e-mail]
[e-mail]
[e-mail]

SEMICON Semicon Sp.Zo.o Poland http://www.semicon.com.pl/
Piotr Ciszewski
Mariusz Sochacki
Jacek Tomaszewski [e-mail]
[e-mail]
[e-mail]

ABIS Abis Spolkaz Ograniczona Odpowiedzialnoscia Spk, Poland www.abis.krakow.pl
Piotr Bistron
Łukasz Stec [e-mail]
[e-mail]

SGI (SMART-Ireland) Surface Mount Technologies Ireland Limited, Ireland www.smartgroup.org
Philip O’Rourke [e-mail]

SEPVE Association of Information Technology companies of Northern Greece, Greece www.sepve.org
Grigoris Chatzikostas
Spyros Ignatiadis [e-mail] [e-mail]
[e-mail]

CIT Computerised information Technology Limited, UK www.cituk-online.com
Subash Sood [e-mail]