Beam sampling & process monitoring in laser material processing applications

High power lasers are used in a variety of industrial process, such as cutting, welding and surface treatment (cladding or transformation hardening) with different beam profiles needed for these three types of applications. Monitoring parameters associated either with the laser beam or with the process itself is fundamental to optimising the operating conditions, and with customers demanding better performance for more difficult to process parts, there is a genuine need to move towards closed-loop control for these processes. 

In order to better understand, and therefore control, the processes it is necessary to determine which process variables are fundamental to the success of an operation. In welding applications, for example, there are many defects that can arise. These include lack of contact, variation of weld depth, seam discontinuity, cutting, sagged weld and melt splattering. Not surprisingly, there are a number of variables that need to be controlled in order to be assured of producing results of satisfactory quality. These include laser power, beam quality parameter (M2/K/Q etc.), beam diameter, beam mode, beam pointing stability, shield/assist gas flow rate, beam focus, material homogeneity and component 'fit-up'. In order to be able to develop closed-loop control systems, it is necessary to monitor as many of the variables as possible and not just the outcome. For example, if the plasma generated in a welding operation overheats, there is a greater refraction of the laser beam leading to less energy absorption at the workpiece. This causes the working "keyhole" to close, resulting in porosity in the welds. However if the plasma region is inadequately sustained, then there is cooling at the workpiece that may also lead to porosity. Thus the same defect can result from different causes. There are two fundamental issues; the monitoring of the beam itself and the monitoring of the process. It is also possible to look at ways of monitoring the two simultaneously. 

With laser powers/energies reaching ever-higher levels it is becoming increasingly more difficult (and expensive!) to attempt to measure even a small proportion of the above variables. Most lasers from reputable suppliers offer (or claim to offer) closed-loop control over laser power etc. and assure users of the invariate quality of their laser beam. In reality the situation can often be clouded by confusing specifications. It is therefore important that beam sampling and process monitoring methods will utilise relatively inexpensive systems that will allow end-users to achieve lower failure rates during production and therefore better productivity and higher profits.

Laser Sampling Methods

Sampling and monitoring of high power Nd:YAG laser beams is relatively easy using a 'leaky' fold mirror and an appropriate sensor (a low cost CCD camera) and can be performed either at the cavity end of the fibre delivery optics or at the workpiece end of the system, as shown in Figure 1.

The situation with high power CO2 beams was more problematic until the development by Precision-Optical Engineering of a novel laser beam sampling system shown in Figure 2. This beam monitor, BM10.6, is based on the use of single point diamond turning to produce a very weak diffraction grating on a copper substrate (usually a fold mirror). The proportion of the beam sampled, and even the angle of the outcoupled sample can be varied to allow for customising to particular optical systems. The use of this sampling system does not affect the primary function of the beam and can be used continuously for beams up to 10 kW power. It is now also possible to utilise true diffractive optical elements for the same purpose, and these have the advantage of being more flexible and can be applied to a greater range of mirror substrates. At 10.6 microns the monitoring sensor is a key consideration, but the recent development of a low cost sensor, OCTAVO, by Laser Point of Italy will be sure to have a dramatic impact upon CO2 beam monitoring. OCTAVO is a multi-segmented thermopile sensor that provides real time monitoring of the power and diameter of an output beam from the laser source, the centroid position and the variation in profile of the intensity distribution. This data can show whether a malfunction has arisen from a deterioration in the generated beam or from a problem with the delivery optics. The system has already helped many industrial users better to schedule their service and maintenance by thorough logging of the laser beam characteristics in use.

Process Monitoring

The radiation emitted by a process is dependent on its temperature. Radiation is emitted by two mechanisms: energy transitions in the atoms giving rise to characteristic emission lines, e.g. the blue colour seen during welding with iron based materials, and black body radiation giving rise to a broad band background of light, where the intensity at a given wavelength is dependent on the temperature of the body (by Plank's equation) and the total energy radiated is proportional to temperature (Stefan-Boltzmann law) )1). The energy radiated is therefore a good subject for process monitoring since it is heavily temperature dependent and the temperature is crucial to the success of the process. Farson et al at Ohio State (2) have developed a system that uses signals from several sensors (optical and acoustic) to indicate whether or not a weld meets certain quality criteria. However the system is currently too slow (and expensive!) to be used in a production environment due to the complex multiple signal analysis requirement.

The usual method of outcoupling process radiation signals is to place a sensor 'off-axis' to the laser beam. This works equally well for both CO2 and Nd:YAG laser systems however it does have the effect of 'crowding' the workpiece. Viewing the process radiation off-axis makes careful set-up important. In both cutting and welding operations, the hottest point of the process is just below the top surface where the laser beam is incident on the process leading edge. As the hottest point contributes most to the radiation emitted and is the first point to undergo a temperature change should the process start to fail, it is best to monitor by viewing as close to this region as possible. Unless viewing directly along the path of the laser beam, viewing will be off-axis and this means that some of the radiation is effectively masked from the sensor. For Nd:YAG laser systems this problem can be overcome since the delivery optics can also be used as collecting optics for the process radiation which is then outcoupled at a fold mirror with an appropriate coating. It is also possible to both sample the laser and monitor the process radiation simultaneously. The diffractive mirror shown in Figure 2 for the sampling of high power CO2 laser beams can also be used to outcouple process radiation from CO2 laser welding operations. Figure 3 shows schematics of systems under development for the simultaneous analysis of the beam and the process radiation. Errors currently detectable by process monitoring systems vary from process to process (and in some circumstances can even cancel each other out!), but the more common ones used in industry are: 

• laser power

• beam focus (which with laser power variation leads to power density changes)

• workpiece fit-up

• processing speed (feed rate)

• shield gas pressure

• weld porosity

• holes/sags in welds

Beam sampling and process monitoring are today being used in a number of production environments. One example comes from the car roof seam welding operation in Ford, Cologne where process monitoring is helping to cut off-line quality assurance and rework. At Philips Components in Sittard, Netherlands, where 12, 000,000 spot welds are carried out daily, a system is being used in the manufacturing department to monitor spot welding of components on TV tube electron guns and has been shown to be effective in detecting laser power and laser focus variation and fit-up errors. A third example is in the manufacture of tailor-welded-blanks for the automotive industry where process monitoring ensures continuous welds.

In conclusion it is important to note that whilst the methodology and hardware are now available to monitor both the laser and the process itself in real-time, there is still a great deal of work to be done in understanding the signals obtained and assigning a degree of significance to their contribution to the overall process which is being monitored. Until this understanding has been achieved (and the task will need to be undertaken for each machine tool/process combination) it will not be possible to fully close-the-loop thereby producing truly flexible intelligent laser machine tools. 

References:

1. Zemansky, M.W., Dittman, R.H., Heat and thermodynamics, Published by McGraw-Hill

2. Personal communication.

Authors: Ian Johnstone, formerly of Precision-Optical Engineering, 42 Wilbury Way, Hitchin, Herts, UK, SG4 0TP, but now with Armstrong Optical Ltd, 44 Collingwood Road, Eaton Socon, St Neots, Cambs, UK, PE19 8JQ (ianjohnstone@yahoo.com), and Dr. Karen Williams, Department of Mechanical Engineering, Loughborough University, Loughborough, Leicestershire, LE11 3TU, UK, (k.williams@lboro.ac.uk) 

Press Enquiries: Denis Bulgin, In Press PR Ltd. PO Box 24, Royston, Herts, UK SG8 6TT.Tel: 01763 262621 Fax: 01763 262655 E-mail: denis@inpress.co.uk http://www.inpress.co.uk

This article is also available on the web pages of Industrial Laser Solutions at: www.ilr.com