The Generation of Autogeneous Surface Features using a Low Power Laser Beam
Jonathan Blackburn1,2, Paul Hilton1
2Laser Processing Research Centre, The University of Manchester
Paper presented at 2010 ICALEO Proceedings Anaheim, CA, USA, 26 - 30 Sept. 2010, Paper 1704
Surface features can be utilised to enhance functional properties of materials, for instance; to enhance the biocompatibility of orthopaedic implants, and to improve the thermal properties of heat exchangers. Electron beams, and more recently high-powered multi-mode laser beams, have been used to create customised surface features, using the Surfi-Sculpt® process, in which features are formed by repeated and cyclic sweeping of a focused power beam in a predetermined pattern. However, both electron beam machines and high-powered laser sources require significant capital investment, which may be cost prohibitive for certain industry sectors. The feasibility of generating surface features with a low-power Yb-fibre laser beam, manipulated with a relatively simple beam scanning system, has been explored here. The successful management of the temperature distribution in the workpiece was found to be critical to achieving an efficient and reproducible process. The effects of several key process parameters on the resulting feature shape have been investigated. High speed video observation of the Surfi-Sculpt process has been used to confirm initial thoughts regarding the key process mechanisms.
Surface features can be utilised to enhance the functional properties of materials, for instance; to improve the bond strength between fibrous composites and metallic materials, to increase the biocompatibility of orthopaedic implants, and to enhance the thermal properties of heat exchangers. The Surfi-Sculpt process is a surface modification technique, patented by TWI in the previous decade, which utilises a rapidly-deflected power beam to create customised autogeneous features on the surface of a material. In comparison with potential competing technologies, such as direct metal deposition, Surfi-Sculpt is not an additive process and does therefore not require complicated powder or wire-feed delivery systems. Surface features are produced by inducing a movement of material from the bulk of the substrate to points on the workpiece, resulting in arrays of protrusions (features) and corresponding arrays of intrusions. It is a non-contact process, in that a power beam induces the material movement, and consequently does not suffer from tool-wear difficulties.
Movement of the material is dependent upon both vapour pressure and surface tension forces. The process mechanisms can be fundamentally compared with keyhole welding; whereby a highly focused power beam produces a vapour cavity in a substrate which is translated through the workpiece according to the motion of the beam. Traversing the beam across the workpiece creates a trailing weld pool. The combination of surface tension variation along the weld pool and vapour pressure from the cavity, causes a displacement of material opposite to the beam's direction of travel. This is evidenced in keyhole laser welds by the presence of a crater at the end of the weld, and a corresponding bump at the weld initiation point. Relatively simple surface features can then be produced by repeating the beam traverse, or swipe, displacing more material from the bulk of the substrate to a common point. Integration of these features allows a multitude of potential features for different applications to be produced.
The Surfi-Sculpt process was invented using electron beams, which can be highly focused and quickly deflected using computer controlled electromagnetic coils. Ongoing development of the process with in-vacuum electron beams has allowed a wide variety of surface features to be produced which meet different functional requirements.[4,5] Recently, experimental trials have been performed out of vacuum using high-power focused laser-beams. In this variant of the process, the focused laser beam is deflected rapidly over workpiece using a galvanometer driven beam scanner. This technique requires the use of a high brightness laser beam source to give the required focusing length for use with scanning systems, while retaining the power density needed on the surface of the material. Comparable, macroscopic surface features were produced in several metallic materials using high-powered multi-mode solid-state lasers in combination with a beam scanner. Beam widths of >300µm in diameter were utilised with scanning speeds of <400mm/s to produce features several millimetres in height.
Solid-state lasers with an M2 value of <1.1 are commercially available at powers exceeding 4kW. The excellent beam quality of these laser sources allow focused spot sizes of <50µm to be obtained, whilst maintaining a working distance sufficient with modern scanning systems. Consequently, diffraction limited solid-state laser sources should be capable of producing features on a more microscopic scale than previously possible with laser Surfi-Sculpt. Furthermore, lower power (<500W) diffraction limited Yb-fibre lasers are significantly less expensive than the laser sources utilised by Hilton and Nguyen  and commercially available electron beam processing equipment. A compact, easily integrateable, laser Surfi-Sculpt machine, less expensive than its electron beam counterpart and not requiring the use of vacuum chambers could have may have noteworthy potential in a range of industry sectors.
This paper details recent work performed with a 200W Yb-fibre laser and a high accuracy galvanometer driven scan head to produce surface features using the Surfi-Sculpt process. Experimental trials have been performed systematically; initially establishing key process parameters for the simplest of features, followed by the integration of these features into more geometrically complex features, and finally arrays of these features. A range of metallic materials has been considered, and high speed video observation of the process has been used to observe key process mechanisms, and the effects of changes in key parameters on the process.
The materials used in this investigation were; grade 304 stainless steel plate, C-Mn steel plate, Inconel 718 plate, and Ti-6Al-4V plate.
The majority of trials were performed on Inconel 718 and 304 stainless steel, with selected processing conditions being transferred to C-Mn steel and Ti-6Al-4V. Plates were guillotine sheered into coupons of dimensions 50mm x 50mm, and edges deburred. The surface of all materials was prepared using a 2500 grit abrasive paper. Immediately prior to processing, the surfaces of all workpieces were cleaned with acetone to remove any grease or dirt deposits.
All trials were performed with a redPOWER® Yb-fibre laser (manufactured by SPI lasers), which had a beam parameter product (BPP) of <0.37mm.mrad. The laser source produced a rated maximum output power of 200W ±<0.5%, at an emission wavelength of 1070 ±10nm, which was fibre delivered to a Nutfield Technology XLR8-15-1064 galvanometer driven scan head. A 125mm focal length Jenoptic 03-90FT-125-1064 F-theta lens, protected by a gas-knife, gave a beam width of 38µm diameter, and a working field of ~60mm x 60mm. A linear stage with micrometer adjustment was utilised to adjust the focal plane of the workpiece. Unless stated otherwise, the minimum beam waist was positioned on the top surface of the workpiece. All processing was performed at 200W (requested laser power). WaveRunner, a Windows based software package, was used to control the beam scanner and the delivered power. Processing was performed at atmospheric pressure, either in air or in a chamber supplied with argon (99.998% purity). Argon was fed into the base of the chamber, at four points and through steel, wool to ensure a laminar flow of gas that did not interfere with the process other than to prevent oxidation. The typical experimental set-up is shown in Figure 1.
Figure 1. Typical experimental configuration with argon shielding
Scope of work
The most basic features that may be created with the Surfi-Sculpt process are those which are produced by repeating a swipe a number of times, in the same position on the workpiece (Figure 2a). The resultant feature is a simple protrusion. Significantly more complex features can be produced by incorporating several swipes into a motif (e.g. Figures 2b and 2c), where the expected protrusion of each swipe is expected to coalesce with the protrusions of other swipes from the same motif. The motif is then repeated a number of times and protrusions are grown simultaneously into a larger feature.
The experimental work reported here was performed in three distinct stages:
- Development of single protrusions.
- Development of relatively complex features with ≥8 swipes per motif.
- Production of an array of complex features.
It was envisaged that by understanding and optimising the single protrusion process through empirical investigation, then larger and more complex features could be more easily produced, through integration of single protrusions.
Figure 2. Examples of different swipe configurations investigated
a) single swipe
b) multiple single swipes integrated into a simple motif
c) multiple single swipes integrated into a complex motif
Table 1 details the process parameters investigated for the first stage of experiments, and the range of values assessed for each parameter. In order to assess a large number of variables, protrusions were produced in 1x8 arrays with sufficient separation that they did not coalesce into larger features. It should be noted that although a large number of trials were performed, not all permutations of parameter values possible were examined. After an Figure 2b and 2c, and of a more complex nature.
Table 1 Process parameters investigated
|Swipe speed, mm/s
||400 - 1000
|Number of swipe repeats
||80 - 400
|Swipe length, mm
||2.5 - 6.0
|Focal plane position, mm
||-1.5 - +2.5
||Inconel 718, Ti-6Al-4V, 304 steel, C-Mn steel
Single protrusions up to 3.2mm in height could be produced with the correct combination of process parameters. Macro images and scanning electron micrographs of typical protrusions, produced with optimised parameters, in Inconel 718 and Ti-6Al-4V are shown in Figure 3.
Figure 3. Typical protrusions produced with a 4mm swipe length, 600mm/s swipe speed, 400 swipe repetitions, at focus, 177ms swipe delay
a) Inconel 718 in air
b) Ti-6Al-4V in argon
The effect of swipe speed on the resultant protrusion height when processing Inconel 718 in air is shown in Figure 4. Swipe speeds of 400 to 1000mm/s were examined with 160 swipe repeats and a constant swipe delay. In general, higher swipe speeds and longer swipe lengths produced taller protrusions. At swipe speeds above 600mm/s, the protrusions were <1.0mm in height with significant variation noted between identical runs. Relatively small (<0.5mm) protrusions were produced at swipe lengths less than 2.5mm, independent of the swipe speed.
Figure 4. Effect of swipe length and swipe speed on protrusions produced in Inconel 718 in air, with 160 swipe repeats and a constant swipe delay
At the swipe speed producing the highest protrusions, (600mm/s) with a laser power of 200W and a constant swipe delay, protrusions with varying swipe lengths were produced using a different number of swipe repeats. The results, in terms of protrusion height, are detailed in Figure 5. Increasing the number of swipe repeats, and hence the work performed per protrusion, increased the height of the protrusion, although diminishing returns were seen at higher swipe repeat rates. Protrusions up to 3.2mm in height were produced with 400 swipe repeats and a swipe length of 6.0mm, compared to protrusions of height 1.5mm, when only 80 swipe repeats were used for a 6.0mm swipe length. As was observed when assessing the effect of swipe speed on protrusion height, longer swipe lengths produced higher protrusions, although the increase in protrusion height was relatively small at swipe lengths above 3.5mm.
Figure 5. Effect of swipe length and swipe repeats on protrusions produced in Inconel 718 in air, with a swipe speed of 600mm/s and a constant swipe delay
The Rayleigh length and depth of focus (defined here as a length over which the focused beam laser diameter increases by 5%) of the focused laser beam used here are approximately ±0.95 and ±0.61mm respectively. When operating within the depth of focus of the laser beam, up to 1.0mm difference in protrusion height is observed. Protrusions are still produced when operating outside the depth of focus, although their height and reproducibility decreases significantly when working at relatively large distances (i.e. 2.5mm) away from the focal plane.
Swipe conditions giving the largest build rate in Inconel 718 in air were transferred to Ti-6Al-4V, C-Mn steel and 304 stainless steel. The conditions were both in air and in an argon environment. The effect on the height of the protrusions is detailed in Figure 6. Protrusions produced in argon in all of the materials were free of oxidation. The protrusions produced in Ti-6Al-4V, C-Mn steel and 304 stainless steel in an argon environment are all of a similar height, independent of swipe length. It is evident that the argon environment aids the build process in Ti-6Al-4V. However, the argon atmosphere hinders the build process when producing protrusions in Inconel 718 when compared to air. Protrusions 3mm in height were produced in air compared to features up to 1.3mm in height in argon.
Figure 6. Protrusion height variation for 600mm/s linear swipes of lengths 2.5 to 6mm with 400 swipe repeats, performed in Inconel 718, Ti-6Al-4V, 304 stainless steel and C-Mn steel in air and Ar environments
The interaction of single protrusions, using parameters developed in the previous section, with other single protrusions was examined here. Linear wall features, such as those detailed in Figure 7, could be produced by reducing the separation between different swipes in the motif. As can be observed in Figure 7, a wall feature can also be produced using a triangular arrangement of swipes. Such a feature allows easy tessellation into numerous arrays of surface features. Although not reported here in detail, management of the heat input between successive swipes was critical in optimising the build rate. The heat input can be efficiently managed by altering the order each swipe is performed in the motif and the swipe delay.
Figure 7. Large wall features produced using several wall features joined in
a) Inconel 718 in air, and
b) Ti-6Al-4V in argon
Numerous experimental trials were also performed with the swipes arranged in a circular motif, with the feature intended to grow at the centre of the circle. Initially, eight swipes per feature were used (as shown in Figure 2c) and the swipe delay, swipe speed and swipe length, were all varied according to the values stated in Table 1. Further experimental trials were performed with an increasing number of swipes in the motif (up to 64 swipes) and with very large numbers of swipe repeats (up to 300). Figure 8 details typical conical Surfi-Sculpt features produced.
Figure 8. Typical conical shaped features produced in a) and b) Inconel 718, and c) Tt-6Al-4V. Scales in mm
Figure 9 shows the effects of swipe delay (1.5 to 400ms) and swipe speed (133, 267, 534 and 800mm/s) on the height of the resulting feature produced in grade 304 stainless steel with eight 1.5mm swipes in the motif. It should be recalled that the swipe delay refers to the time elapsed between a laser beam finishing a swipe and the beam revisiting that same swipe. However, as all the swipes begin at the centre of the motif, the laser beam will revisit this point on a much more frequent basis.
Figure 9. Effect of swipe speed and swipe delay on the feature height for motifs containing eight 1.5mm swipes, produced in grade 304 stainless steel
It can be seen from Figure 9 that for each swipe speed there is an optimum swipe delay with regards to the height of the feature. A similar trend was observed in features produced with swipe lengths of 2.5 and 3.5mm. In general, a higher swipe speed necessitated a shorter swipe delay in order to maximise the feature height. Little difference was observed between the maximum feature heights achieved with the different swipe speeds, despite the difference in energy input to the workpiece. With nominally identical parameters, features up to 60% taller could be produced when utilising a 2.5mm swipe length, compared with a 1.5mm length. However, the same increase was not observed when increasing the swipe length to 3.5mm and there was no advantage, in terms of feature height, for using this swipe length compared with the 2.5mm length, for the parameters examined here.
Figure 10 details a typical array of Surfi-Sculpt features produced on grade 316 stainless steel round bar, 32.0mm in diameter. A total of 170 nominally identical features were produced on the bar - which could be utilised for composite to metal joining, crushing of foodstuffs, or piercing thin films.
Figure 10. An array of 170 nominally identical Surfi-Sculpt features produced on 316 stainless steel bar
Currently the production of large arrays of surface features is not possible at high laser utilisation rates, since delays are built in to the program to optimise feature height. Ideally during this time the laser beam would be processing at another position on the workpiece and would return after the required delay. This is not easily achievable with the scan head software used since it was designed for laser marking applications. As a result, the process times observed have been undesirably long. Nevertheless, the optimised processing time (laser beam off time minimised) for producing the features shown in Figure 32 is estimated to be 4s per feature. The build time for the 170
High Speed Observation
Observation of the Surfi-Sculpt process when using an electron beam is particularly difficult as access is restricted due to the vacuum chamber. Hitherto, knowledge of the process mechanisms has been limited to a theoretical interpretation with little supporting evidence.
In selected cases, the process was observed with a high speed camera at sampling frequencies of 2-10kHz. Temporal and spatial filtering ensured the attenuation of the high intensity broadband process emissions. A fibre delivered copper vapour laser (primary wavelength of 510.6nm), manufactured by Oxford Lasers, of 20W average power and pulse duration ~20ns, provided illumination of the subject matter.
Figure 11. High magnification pictures of the tip of a single protrusion produced during the laser Surfi-Sculpt process
Figure 11 details a sequence of high speed pictures, taken at a frame rate of 10kHz during the production of a single protrusion in Inconel 718. The beam interacts with the protrusion between 0 and 0.1ms and the applied power density is capable of a creating a keyhole almost immediately. This is evidenced by the spatter ejected from the process in the frame at 0.1ms. The vapour cavity then progresses along the swipe causing a particularly unstable globule of
Thermal management of the workpiece
From an early point, the process was found to be particularly interactive, whereby changing the value of one parameter would also affect the values of several others and lead to anomalous results. A direct result of this, combined with the large number of experiments performed, is that it would be particularly challenging to draw direct comparisons between all of the features produced in the different substrates. However, it is apparent that a wide variety of surface features can be produced using low power, diffraction limited Yb-fibre laser beams. Process parameters for single protrusions or large integrated features, produced on a range of materials, can be optimised through empirical investigation.
From analysis of the results and high speed observation, it appears that correct thermal management of the workpiece is paramount to producing the largest features that do not suffer from overheating. For instance, nominally identical single protrusions could be produced in arrays but not individually, unless significant heat was input into either side of the workpiece. Furthermore, separating the protrusions in these arrays by distances of up to 5mm decreased the resultant height of the feature, compared with those produced with a swipe separation of 1mm. It is clear that a correct temperature distribution is required in order for the Surfi-Sculpt process to work at its optimum. This is most likely a result of the thermal properties of the materials and the absorptivity of the materials at the 1µm wavelength.
It is well known that, during laser processing, the absorptivity of the workpiece is dependent upon the wavelength of the incident laser radiation and the temperature of the workpiece. The majority of the work reported here was performed on Inconel 718. The absorptivity of Inconel 718 is known to increase linearly with a temperature increase. An absorptivity of 0.30 was calculated at 300K and this increased to 0.55 at 1180K. It was observed that the majority of the material, at least initially, in a feature is taken from an intrusion at the end point of the swipe. Only after a relatively large number of swipe delays is material removed from further along the swipe. Therefore, any variation in the temperature of the material at this point will affect the amount of energy absorbed by the workpiece and hence the volume of material moved. Furthermore, the increase in the thermal conductivity of Inconel 718 at increased temperatures will amplify the above effect.
However, it is clear that at workpiece temperatures above a certain undetermined value, the height of the protrusion is less than those produced at lower temperatures. This is most likely a result of the feature needing adequate time to solidify and/or reach a temperature so that, on the next swipe, only a small proportion is melted rather than a large volume, which would likely result in sagging of the feature. The temperature of the workpiece could be easily determined through the use of a thermal imaging camera and a process model could be developed to optimise the temperature of the workpiece and hence reduce the processing time. It is also worth considering that the workpiece could be pre-heated to reduce production times and help to quickly achieve a quasi steady-state temperature field in the workpiece.
Practical and economic implications
Compared with the high power, multi-mode lasers utilised by Hilton and Nguyen to produce Surfi-Sculpt features, the work reported here has used a significantly lower cost 200W diffraction limited Yb-fibre laser to similar effect. The current price of the laser source used for the work reported is approximately £25,000 (at 2010 prices).
A relatively low cost, but still high positional accuracy, XY scan head with an F-theta lens has been used to achieve a working field of 60 by 60mm (at focus). Other scan heads are available, such as a telecentric laser scan head (the laser beam is always delivered perpendicular to the workpiece) and an XYZ F-theta lens scan head (allowing manipulation of the focal plane of the laser beam), that may give added benefits to the process. Larger F-theta lenses can also be purchased which would increase the operating area of the system. An XYZ scan head would allow dynamic control of the focusing plane during the process, which may reduce build time. However, these scan heads are relatively expensive compared with the one used for the work reported here, which can be purchased for approximately £10,000 (at 2010 prices). It is worth noting that the features built at the extremes of the scan field are slightly irregular compared with those produced at the centre. An increasing tilt angle is observed as the feature is built at the extremities of the scan field. The use of a high accuracy two-axis table, under the laser beam, would eliminate this problem and significantly increase the processing area.
The laser source and scan head are the principal components of the system used in this report, although provision should be made for a suitable water chiller unit (although air cooled 200W diffraction limited Yb-fibre lasers are available), a suitable operating cabinet and system integration. Nevertheless, a working laser Surfi-Sculpt machine could be produced for <£50,000 which is relatively small amount compared with the higher power laser systems and the EB Surfi-Sculpt machines[4,5] used previously.
For comparison, a similar array of features to those produced on the 32mm diameter bar was also produced using the EB process. Parameters were developed using a modified Hawker-Siddeley 150kV 6kW rated machine, fitted with a 5-axis beam deflection system and running proprietary software. The software allowed a ~98-99% efficient beam utilisation which resulted in a process time, including all beam and workpiece manipulation, of 60s. A simple vacuum handling strategy would also be required and would add about 15s per part. This time is an order of magnitude less than achieved with the low power laser system, although it is worth noting that the cost of such an EB system is at least an order of magnitude greater.
The potential for using a low power Yb-fibre laser, in combination with an appropriate scan head, for producing features using the Surfi-Sculpt process has been assessed in this project. The following conclusions can be drawn:
- A 200W diffraction limited Yb-fibre laser, focused into a 38µm diameter beam width and used at focus, is sufficient to produce surface features, at swipe speeds up to 800mm/s, in Inconel 718, C-Mn steel, 304 stainless steel,and Ti-6Al-4V.
- The principal components of a nominally identical system to the one used here could be purchased and integrated into a fully functioning system for <£50,000.
- Successful management of the temperature distribution in the workpiece is critical to integrating single protrusions in to large surface features.
- A variety of surface features has been produced, including: single protrusions up to 3.2mm high, wall features, and conical features.
- High speed video observation has been utilised to observe the process and understand the process mechanisms.
This work was funded by the industrial member's of TWI, as part of its Core Research Programme. The authors would like to thank The Institute for Manufacturing at Cambridge University for use of the Yb-fibre laser and beam scanner equipment and in particular Martin Sparkes for his assistance with the system. Bruce Dance, from TWI, is acknowledged for numerous stimulating discussions, helpful suggestions and performing the electron beam comparison experiment. The high speed camera and copper vapour laser were loaned from the Engineering and Physical Sciences (EPSRC) Engineering Instrument Loan Pool.
- Smith, F. (2005): 'An innovation in composite to metal joining, Materials Technology', COMELD 20(2), 91-96.
- Islam, M.D., Oyakawa, K. & Kubo, I. (2009): 'Visualization of flow pattern and thermal image analysis of enhanced heat transfer surface, Heat Mass Transfer', 45, 511-517.
- Dance, B.G.I. & Keller, E.J.C. (2002): Workpiece Structure Modification. International patent publication number WO 2004/028731 A1. Applicant: The Welding Institute.
- Dance, B.G.I. & Buxton, A.L. (2006): 'Surfi-Sculpt®. A New Electron Beam Processing Technology', 8th Int. Conf. on Electron Beam technologies, June 6, Varna, Bulgaria.
- Dance, B.G.I. & Buxton, A.L. (2007): 'An introduction to Surfi-Sculpt technology - new opportunities, new challenges', Proc. 7th Int. Conf. on Beam Technology, April Halle, Germany.
- Hilton, P.A. & Nguyen, L. (2008): 'A New Method of Laser Beam Induced Surface Modification Using the Surfi-Sculpt® process', Proc. 3rd Pacific Conf. on Applications of Lasers and Optics,April, Beijing, China.
- Saint-Catherine, C., Jeandin, M., Kechemair, D., Ricaud, J.P. & Sabatier, L. (1991): 'Study of dynamic absorptivity at 10.6µm (CO2) and 1.06µm (Nd:YAG) wavelengths as a function of temperature', Journal De Physique IV, 1(C7), 151-157.
- Sweet, J.N., Roth, E.P. & Moss, M. (1987): 'Thermal Conductivity of Inconel 718 and 304 stainless steel', Int. Journal of Thermodynamics, 8(5), 593-606.
Meet the Authors
Mr Jonathan Blackburn is an Engineering Doctorate Research Engineer, sponsored by TWI Ltd, at The University of Manchester. He is funded by the EPSRC. His principal areas of research are laser welding of titanium alloys and laser surfacing techniques.
Dr Paul Hilton is the Technology Fellow for Laser Materials Processing at TWI Ltd and has over twenty years of laser processing experience. He has previously been conference chair for LMP at ICALEO and is the current President of the UK's Association of Industrial Laser Users.