Improved material processing achieved using high brightness electron beam power deposition

TWI Bulletin, September/October 2007

This year's Richard Weck paper was authored by Colin Ribton. In this abridged version he describes his award winning work.

Colin Ribton graduated from The University of Nottingham in 1984 and joined TWI's Electron Beam group in 1985 and since that time he has worked extensively on the research and development of high power electron beam equipment. In 2001 he moved to a start-up company developing novel antenna technologies, where he became Vice President of Application Engineering. He returned to TWI in 2003 to lead the Electron Beam group. He received the Richard Weck Award for his Core Research Programme work and Industrial Members Report 858/2006, entitled 'High brightness electron beam power deposition for improved material processing.'

High intensity electron beams can be used for a variety of material processing requirements including drilling, cutting, surface texturing, welding, and machining. As Colin Ribton reports the processes can be applied at macro (over 100mm by 100mm) and micro (presently limited to 100µm by 100µm) levels, on both metallic and non-metallic materials. Here he reviews previous work, and examines processing trials and equipment optimisation for better processing. He also looks at theoretical understanding and modelling and verification trials of the electron-material interaction. In part, this work sought to develop and validate design tools to optimise the accelerating potential through understanding the effect of this parameter on the power deposition.

Electron beam (EB) equipment offers a number of potential advantages over laser beam equipment in this context. First, the high speed of beam manipulation possible with electro-magnetic deflection offers the potential for rapid production. Movement can be reprogrammed readily from one part to another, enabling the possibility of bespoke, flexible manufacturing. Secondly, it is possible to achieve higher power densities in combination with relatively high average power levels with an electron beam rather than with a laser beam.

Objectives

• To develop gun column optics design tools further to enable increased beam intensity.
• To characterise equipment capabilities in fine scale processing of a range of relevant materials.

Background

The project has applied high intensity EB equipment, developed in previous work, to a range of materials. The processing carried out has aimed to be relevant to a number of product applications. The products being targeted are:

• Medical devices - particularly those coated with drug impregnated polymers. Texturing of the metallic device could be used to ensure the polymer is mechanically bonded to the surface.
• Wire-polymer composites. Texturing of wire could be used to enhance the bond between the wire and the polymer. Such composites are of interest in the automotive industry, for example, for car bumpers.
• Micro-fluidic devices. 'Lab on-chip' devices and fluid/gas analysers being developed for a range of applications require micro channels to be produced of dimensions from 1µm to 10µm. These channels may be directly machined using high intensity electron beams or may be impressed or moulded using a tool produced by high intensity electron beam processing.

There are other possibilities for high intensity EB processing. The above examples are seen as relatively near-to-market opportunities that may be achieved in the future through further, mainly incremental, developments of the process and equipment developed using the results of this work.

To determine the design of the high intensity gun and electron optics to carry out some or all of these processes, it is useful to understand the mechanisms by which the beam interacts with materials. This allows the necessary beam qualities to be determined for the particular processing required, and provides design targets for the gun and electron optics.

Review of Literature and Electron Interaction Modelling

Electron penetration

The potential range of applications of high intensity electron beams is extended by further reducing the beam diameter and increasing the intensity of the beam. Part of this work has examined understanding interactions of the electron beam with materials. This is different from the interaction of laser beams with materials. Laser beams interact with a few surface layers of atoms. Electrons penetrate much deeper into the material, typically in the range 1µm to 100µm, but the depth is dependent on the accelerating potential ( ie energy) of the electrons. At higher energies the depth of penetration can exceed the beam diameter (at present the beam diameter is approximately 40µm). To understand further the beam-material interaction, a review has been carried out of relevant work in the electron microscopy area.

The models developed in this work, that predict the deposition of power into the material, have been compared with published empirical models for electron beam welding that describe the effect of the power deposition and also with some trials in materials of interest for future applications.

To optimise the intensity through the cross section of the electron beam, the highest acceleration potential is required to produce the smallest diameter focal spot. This is a consequence of minimising the effects of space charge spreading of the beam and minimising the effect of thermal velocity spread. These issues have been addressed in previous work. However, electron beams (unlike laser beams) penetrate into the material surface, to depths significant when compared to the high intensity electron beam diameter. The maximum penetration depth is dependent on the material and the beam energy. For example, a 150keV electron beam will have a penetration depth of 137µm into silicon, but only 47µm into iron. In considering depositing the maximum power into the smallest volume of material this penetration depth must be taken into account. The depth of beam penetration is shown in Fig.1.

Electron interaction modelling

A number of articles have examined electron interaction with matter and aimed to provide models for this interaction. Most relevant to this work have been those papers that have examined the effects that occur in electron microscopy and materials analysis. The accelerating potentials used tend to be similar to those used in the high intensity EB equipment ( ie from 10 to 200kV), and the formulae and models produced tend to be limited in the range of electron energies for which they are applicable.

Two texts have been identified as providing the most relevant modelling of electron-material interactions and related experimental data. Schiller proposes empirical formulae to fit the measured results in a range of materials. Kanaya proposes a model based upon the cumulative effect of all the electron scattering events that can occur within the material.

One of the most significant scattering effects occurs close to the surface of the material and results in electrons being backscattered. The proportion of electrons backscattered varies with the atomic number and two results are compared in Fig.2. The results shown are measured results by Schiller, and values calculated using the Kanaya model. The proportion of current backscattered is approximately independent of the incident beam energy in the range 10keV to 100keV. In electron beam processing (where the surface is melted or where a keyhole is formed) the backscatter would be expected to be less, due to the reduced material density and (in the case of keyholes) the increased likelihood of backscattered electrons being re-absorbed by the material.

In considering the distribution of power dissipation in the volume of interaction, we need to use an electron diffusion model. Kanaya provides a review of models and then consolidates these into a model that compares well with measured results.

As the electrons penetrate into the material they undergo multiple collisions that gradually reduce their energy. The material absorbs most of this energy. The energy absorbed by the material at a fraction of the range has been modelled and experimentally verified by Kanaya and others and is shown in Fig.3 for a 150kV beam impinging upon copper. This energy distribution as a function of penetration can then be used with Archard's 'diffusion model' to produce a model of the heat distribution over the volume of interaction. The diffusion model uses a parameter called the diffusion depth defined as the depth at which the transmitted fraction of the current is 1/e ( ie 36.8%). From the Kanaya model this can be calculated as shown in [1].

where R is the electron range ( Fig.1), and Z is the atomic number of the target material, [2]. 

E ₀ is the electron energy (keV)
and K is defined in [3]

where A is the atomic weight of the target material.

The diffusion model assumes that the electrons move equally in all directions in a sphere centred at XD depth, in such a way that all their paths are R long. This has been used to calculate the bulk average energy dissipation in the material, shown schematically in Fig.4. This is readily converted to a power distribution by multiplying the lost potential with the beam current.

The diffusion model predicts that the shape of the interaction volume is dependent on the material and is independent of the accelerating potential, for beams of negligible diameter. Although the diffusion depth and range are each dependent on the accelerating potential, the ratio (XD/R) of the diffusion depth (XD) to the overall range (R) remains constant over the accelerating potentials for which the model is applicable. In defining the shape of the volume of interaction, for this material, and referring to Fig.4, this volume is always bowl shaped, ie a truncated sphere.

Electron gun design model

The diameter of the beam at the surface (which in these examples is not negligible) is dependent upon gun and electron optic design. However, there are physical limits to the minimum beam diameter defined by the gun geometry, the accelerating potential, and the beam power. The beam radii achievable for a generic gun design have been calculated in this work and are shown in Fig.5 for a 100W beam. The diameter of the beam generated is related to the accelerating potential. The calculations have assumed the following:

• The cathode diameter is optimised such that it emits 80mAmm-2 for each of the designs at the different accelerating potentials.
• The diameter of the beam in the lens is fixed.
• The primary crossover of the beam is at a fixed position relative to the lens.
• Lens internal diameter is 52mm.
• Working distance from the end of the column is 25mm (112.5mm from the lens centre plane).

Figure 5 also shows the contribution to the overall beam diameter that is made by three aberration elements within the model. The contribution of space charge to the beam radius is significant only at lower accelerating voltages, where the electrons' mutual repulsion leads to a greater divergence. For a 100W beam this is seen to be the factor that dictates the beam radius at accelerating potentials of less than 20kV. Above this potential, the thermal velocity spread determines radius. The effect of electrons having thermal energy as they leave the surface is to introduce a range of non-laminar angled rays into the beam.

This effect is reduced gradually as the accelerating potential is increased ie as the thermal energy becomes less significant relative to the energy of acceleration. Ultimately, at high accelerating potentials, the beam radius may be limited by lens spherical aberration as shown in Fig.5 from 60kV and more increasingly up to 150kV. Lens spherical aberration occurs as an electro-magnetic lens will not focus a beam perfectly. This degree of aberration will increase as the diameter of the beam increases relative to the lens diameter - but this parameter has not been varied within this work, although the model is capable of predicting on this variable. This was to allow a like-for-like comparison of beams at different accelerating potentials and powers.

There is a trade-off in designing a gun column to balance the above three aberration effects. For example, through gun design the angle of the beam may be increased in order to reduce the effect of space charge, but this will carry a penalty of increased lens spherical aberration. The parameters used for the gun column design in the model match the current configuration of the high intensity development equipment.

The model has also been used to find the beam intensity for a range of different beam powers over a range of accelerating potentials from 10 to 150kV. The output of the model is shown in Fig.6. Below 20kV space charge most severely diminishes the intensity of high power beams. This is because space charge is directly related to the electron number density and this is proportional to the beam current. At higher voltages, the effect of lens aberration is greater on the intensity of lower power beams. The degree of lens aberration is independent of power level of the beam or accelerating potential, but is more significant on beams focused to a smaller radius, ie low power but high intensity beams.

At intermediate accelerating potentials where neither the space charge aberration nor the lens spherical aberration is significant, the beam intensity is independent of power but proportional to the accelerating potential. In this region the beam intensity is limited by the source brightness - ie the emission density of the cathode and primary beam angle from the gun. As the cathode emission density is constant, and the beam angle fixed for all beams modelled, the source brightness is fixed in this model.

Combined model to predict power density

By combining the modelling of the interaction volume with the modelled beam intensity it is possible to predict the power density produced in a target material. The volume of interaction is derived from the electron diffusion model.

The function is shown in Fig.7 for copper and for powers from 50W to 2kW. Two observations are made from this graph:

• The optimum accelerating potential for achieving maximum beam power density increases as the beam power increases.
• As the beam power increases, the peak power density increases

For low beam power processing in copper ( eg 50W) the optimum accelerating potential to maximise the power density (at 32kWmm ^-3 ) is 22kV. For higher power beams ( eg 2kW) the optimum accelerating potential is 60kV, giving a power density of 40kWmm ^-3 .

Changing the target material has a large effect on the maximum power density that can be achieved, but has a much smaller effect on the accelerating potential at which the maximum power density is obtained. Fig.8 shows plots of the beam power density for 100W beams for a range of materials (copper, iron, titanium, aluminium and silicon) over the range of accelerating potential of interest. It can be seen that the peak power density in copper of 34kWmm ^-3 is achieved at 24kV. In contrast, for silicon the peak power density is obtained at a similar beam accelerating potential of 20kV, but is less than one third the magnitude of the peak power density in copper.

From this work we can summarise that the optimum accelerating potential for a high intensity electron beam system should be selected taking into account the target material and the processing power required. It is to be expected that as more processing power is required (which may be necessary for higher speed processing for increased productivity) then the optimum accelerating potential will increase. It is also the case (as seen in Fig.6) that at optimum voltages, the beam intensity will be independent of power level - consequently higher power beams will have a focal spot increased in proportion with the increase in beam power. This will limit the maximum production rate of the process in cases where the beam diameter must be constrained to a limit for processing to be acceptable - eg in machining small features.

Experimental work

Trial approach

Trials were carried out on a range of materials. Trials aimed to investigate the viability of high intensity electron beam processing on representative materials. They also provided an opportunity to observe the electron beam interaction with the materials.

In stainless steel (316L) wire it has been possible to produce a range of textures (see Table 1 and Fig.9). Although this could be achieved with mechanical techniques - for example threaded dies, compared with power beam processing, this would have the disadvantages of tool wear and generation of swarf that could detach from the surface when the product was in use. For medical applications this represents a significant risk, which amplifies the advantage of using high intensity electron beam processing.

Table 1 Process parameters applied to stainless steel wire

c) Sample St15, magnified

In processing silicon (see Table 2 and Fig.10) it has not yet been possible to ablate the surface to cut channels. However, narrow melt runs of some 20µm have been produced at 30kV. The wider melt runs produced tended to crack, but this is of no consequence when aiming to ablate the material. In stainless steel wire, notches are produced at higher power densities and, with higher power, and higher intensity beams, it is probable that channels could be cut in silicon. At 30kV the total electron range in silicon is approximately 10µm. As a consequence, further reduction in beam diameter will be unlikely to enable significant reduction in feature size. The feature size currently obtainable is suitable for application in the area of micro-fluidics.

Table 2 Process parameters applied to silicon samples

Copper has been processed for comparison with the beam interaction with silicon (see Table 3 and Fig.11). It has been possible to produce narrow tracks of width less than 30µm using 150kV beams on copper.

Table 3 Process parameters applied to copper samples

It is anticipated that with further development of the high intensity electron beam design, it will be possible to EB machine and otherwise produce features at dimensions of 10mm. Although smaller diameter beams can be produced, at powers of 50W and above, the accelerating potential must be increased. Consequently the depth of penetration will exceed the beam width. Therefore it can be stated that the scale of features that can be produced is proportional to the power required in the beam.

Conclusions

• A model has been produced that can predict the accelerating voltage for a beam of a chosen power that will maximise the power density deposited in a specified target material.
• It has been shown that maximum power densities for beams from 50W to 2kW are obtained at corresponding accelerating potentials from 22kV to 60kV.
• High intensity beams can be applied to a range of materials to produce features as small as 20µm.