FEATURE
Issue: 
Topic tags: 

Blink and you'll miss it

Industrial-grade femtosecond lasers are now available, but do they offer any significant advantages over their picosecond rivals? Greg Blackman finds out

A femtosecond is an incredibly short period of time; it’s a millionth of a billionth of a second, which is the same amount of time compared to a second as a second is compared to about 31.7 million years. Lasers that pulse at femtosecond durations have been around for a while, but up until recently have largely been confined to research laboratories. Today, however, a number of companies provide femtosecond systems for industrial processing – Raydiance in the US, Amplitude Systemes in France, and High Q Laser, which was recently bought out by Newport Spectra-Physics, to name a few.

In the world of ultrafast laser processing, the femtosecond laser is competing to a certain extent against its slightly longer-pulsed picosecond cousin, with the likes of Trumpf and Lumera supplying established picosecond systems. So what advantages do femtosecond lasers provide over picosecond devices? Before addressing this, let’s consider the benefits of ultrafast lasers in general.

This really comes down to precision. Ultrafast lasers in the low picosecond to the femtosecond range will ionise material, ablating it faster than the rate of energy transfer to neighbouring atoms. The result is so-called ‘cold ablation’, whereby the pulses of light remove material with little or no thermal effects on the part.

Machine tools or longer pulsed lasers will input heat into the part, melting surrounding material and generating debris, all of which means it is difficult to achieve the desired level of precision. ‘Any machining that requires high precision, down to 10µm or even 5µm, will benefit from ultrafast lasers, be that cutting, drilling, scribing, marking or texturing,’ states Richard Pierce, CEO of Raydiance.

Semiconductor-related sectors and micro-electronics are potentially big markets for ultrafast laser technology, because of the level of precision required. Dicing silicon wafers into constituent chips, for instance, is typically carried out mechanically with diamond scribes. ‘Cutting with a femtosecond laser would reduce the width of the scribes by a factor of 10,’ Pierce says, thereby increasing the yield of a silicon wafer.

Additionally, when processing brittle materials like silicon, ultrafast lasers have the advantage of producing a very clean cut compared to conventional processing methods, which can chip or crack the material. Similarly, display glass for flat panel displays, which is now around 300µm thick, is difficult to cut with conventional scribing wheels without causing damage. Ultrafast lasers can give the quality required to avoid micro-cracks.

Christof Siebert, senior manager of the Micro industry sector at Trumpf, sees processing scratch-resistant cover glass for mobile phones and other electronic devices as a potentially big market for ultrafast lasers. ‘There are approximately 300 million smart phones with cover glass today and this is expected to reach 1 billion by 2014,’ he says, adding that this is still only a small proportion of the total number of smart phones. The chemically-hardened glass is conventionally processed by chemical etching or milling, both of which require post-processing steps. ‘With a laser, you can reduce the post-processing or avoid it completely,’ he says.

Other sectors where ultrafast lasers are suited include patterning thin film solar cells and manufacturing medical devices like heart stents. There are also potential application areas in automotive and aerospace where precision machining is needed. For example, machining vehicle fuel injectors, in which the spray hole is less than 100µm wide and the precision associated with that hole directly relates to fuel mileage. ‘A femtolaser is able to make those holes much more precisely than traditional mechanical and EDM (electrical discharge machining) methods,’ explains Pierce, ‘and it doesn’t have the consumable aspect of mechanical machining, in which drill bits need to be replaced after so many cycles.’ He adds that in many precision-orientated applications there can be a degree of post-processing necessary with conventional methods. ‘The advantage with ultrafast lasers is that minimal post-processing is required,’ he says.

Power equals productivity

While you achieve high precision, the trade-off with ultrafast lasers is that they have lower power for the same price compared to traditional pulsed lasers and therefore throughput per unit cost is less. ‘Ultrafast lasers certainly provide an advantage in micromachining quality – however, one really needs an application that requires this high quality,’ comments Herman Chui, director of marketing at Newport Spectra-Physics. ‘Power is lower, but also the lack of thermal effect means that throughput is lower with a typical micromachining approach.

‘There are ways to make ultrafast laser processing effective,’ Chui continues, ‘but it typically requires a different approach.’ Because the short pulse duration results in shallow penetration depth, in many cases the lasers must be scanned quickly over the part and often making multiple passes over the material to achieve the same level of ablation as longer pulsed lasers.

Output power is one factor that some argue makes picosecond lasers more suited to industrial processing than femtolasers. ‘It is more difficult to generate shorter pulses and therefore the output power of femtosecond lasers is less and productivity is lower,’ comments Siebert of Trumpf. ‘If you don’t gain much in improvement in quality, then the sacrifice for the higher price and lower productivity of femtosecond lasers doesn’t make sense. That’s why 10ps is still the sweet spot in ultrafast laser processing.’

Trumpf’s TruMicro 5000 series is an ultrafast picosecond system based on the company’s disk laser technology. It provides up to 100W of average power for the infrared version, up to 60W in green, and 10W in UV.

Ten picoseconds is considered by some to be the point at which there is minimal heating of the surrounding material and therefore moving to shorter pulses has little effect. However, there is still a good deal of research into the effects of different pulse durations on different materials and the topic is less than clear-cut.

Amplitude Systemes manufactures lasers in the region of 500fs (0.5ps), which are competing against systems in the 10ps range. ‘The key question is “why would you use shorter pulse widths?”’ asks Vincent Rouffiange, sales and marketing manager at Amplitude Systemes. The company has worked with Alphanov, a research centre in Bordeaux specialising in micromachining and, with Bern University in Switzerland among other research centres, on micromachining with differing pulse durations. The work shows that shorter pulses provide a better etch rate, a measure of material removal. (The study investigated pulse durations from 500fs to 100ps.)

‘Up until now, lots of people were interested in 10ps lasers, but we can demonstrate that the etch rate is better with femtosecond lasers in the range of 500fs and the quality of the ablation is also improved,’ Rouffiange states. ‘We’re seeing a switch from longer ultrafast lasers in the picosecond range to femtoseconds in order to improve the etch rate and the productivity, and to also increase the ablation quality on all materials.’

The point at which heat transfer occurs varies with different materials; copper’s electron-phonon interaction time is tens of picoseconds, for instance, and below which there will be no heat dissipation to surrounding atoms, while iron’s is around 500fs. ‘It’s material-dependent and different materials will exhibit different properties,’ states Rouffiange.

The wavelength of the laser can also have an effect on processing, as Rouffiange explains: ‘People thought that ultrafast processing was not wavelength dependent. In practice, there are some differences using green or UV wavelengths. Femtosecond UV lasers give very good results, in terms of ablation quality, spot size, and heat dissipation in the material.’

Picosecond lasers can be frequency doubled to reach shorter wavelengths and the potential for second and third harmonic generation in femtosecond lasers is there, according to Rouffiange. The advantage of UV lasers is that they can be focused to a smaller spot size.

The other area of concern for micromachining using femtosecond lasers is their robustness in an industrial environment. Picosecond lasers have an advantage here, as they are the more mature technology. Colin Moorhouse, staff application engineer at Coherent, comments: ‘Picosecond lasers are mature and are quite robust now; they’re as reliable as established lasers such as Q-switched lasers.’

But Chui at Newport believes that femtosecond systems are now at a stage in terms of reliability where they can be run in an industrial environment. ‘High Q Laser [which Newport acquired last year] has products that are industrial grade in the femtosecond region,’ he states.

When ultrafast lasers were being developed, they were driven by bulk optics, which are sensitive to vibration and changes in temperature. These had to be eliminated to make them suitable for industry, explains Pierce at Raydiance. Raydiance femtosecond lasers are all fibre systems that are very stable.

In terms of providing the power levels for industry, Rouffiange says that there are two solutions being developed in research centres to increase productivity of femtosecond lasers. One is to have lasers with a very high repetition rate to move the beam rapidly over the workpiece; the second is to have a beam with a high energy and to split it into multiple beams for parallel processing. Solid-state lasers have higher energy per pulse (up to 2mJ, according to Rouffiange) suitable for splitting the beam, while fibre lasers provide high repetition rates resulting in high average power and faster processing. Amplitude Systemes is pursuing both avenues of development.

‘We are close to seeing more femtosecond systems being used in manufacturing, because of increases in throughput and stability,’ comments Rouffiange. Amplitude Systemes provides 20W 500fs lasers, which, for most applications Rouffiange says, is enough power. Reliability and stability in the field is also crucial, he says, adding that crystal-based femtosecond lasers are just as stable as fibre lasers. He believes the market for femtosecond lasers is growing, estimating that sales figures will increase 10-fold in five years.

Picosecond lasers are more prevalent in industry than femtosecond systems, according to Moorhouse at Coherent. ‘The rule of thumb is that for a given laser process, you use the longest pulse width you can,’ he says, adding that it’s pretty rare when a femtosecond laser can do something that a picosecond laser can’t in industry. ‘We’ve been competing against femtosecond lasers and we’ve won with the picosecond systems, because they have a higher power spec,’ as well as being able to focus to a small spot size, he says, which allows the customer to meet their quality and throughput needs.

Power, cost, and ease-of-use will all be important in the future. Chui at Newport quite rightly points out that if you have an application that doesn’t require the type of performance delivered from an ultrafast system, then why pay the premium for it? And a lot of applications do not need ultrafast lasers; the quality provided by traditional pulsed lasers is sufficient in many applications. But there are some that will benefit and in those cases there’ll be good adoption of ultrafast lasers, whether that’s picosecond or femtosecond varieties.

Ultrafast optics

When generating femtosecond pulses, it’s not just the source itself that becomes more complicated, but also the optics required to deliver the pulse to the workpiece. The exceptionally high peak power of the pulses means the optics have to have a high damage threshold.

There are also the problems of dispersion and other non-linear effects to contend with, as Dr Lukas Krainer, CEO of Swiss company Onefive, which produces both femtosecond and picosecond lasers, notes: ‘The higher peak power means non-linear effects play a much greater role. Also the spectral bandwidth, due to the short pulse duration, is much broader so any dispersion effects become more prominent.’

Group-delayed dispersion (GDD) is a phenomenon whereby the spectral components of a propagating femtosecond pulse reflect from a mirror at different speeds, resulting in a broadening of the pulse to longer time durations. Dr Krainer says these effects can be minimised by keeping the distances short between optical elements and working with mirrors instead of transmissive lenses where the beam would disperse through the glass.

Laser optics manufacturer, Laseroptik, produces dispersive dielectric mirrors, also called chirped mirrors, designed to compensate for GDD in broadband ultrafast laser systems. ‘Dispersion management for standard laser mirrors is not an issue,’ explains Dr Wolfgang Ebert, CEO of Laseroptik. ‘With femtosecond lasers, you have to throw away the old approaches for designing optical coatings. Coatings for femtosecond optics are designed with other priorities, i.e. to keep the pulse short.’

Laseroptik’s chirped mirrors are coated with a tailored stack of more than 70 non-quarterwave layers. Specialist simulation software is used for coating design to manage the dispersion. ‘The coating layers have no discernible pattern that could be produced easily,’ states Dr Ebert. Many of Laseroptik’s conventional coatings are based on quarterwave patterns, in that each layer is a quarter of the thickness of the incident wavelength. There is no pattern like that for chirped mirrors, according to Dr Ebert, and complicated algorithms are used to calculate the coating pattern.

Ion beam sputtering is used to deposit the coating layers, which provides the level of precision required. ‘Mirror coatings for very short pulses are extremely sensitive to coating thickness,’ says Dr Ebert. ‘The layers have to be deposited with a precision of a few atomic layers. An error that’s built into layer number three would propagate down to layer 70 and influence the whole system, so this cumulative error has to be controlled very precisely.’

Dr Ebert adds that the optics also have to be tested to ensure they operate correctly to control the dispersion. Laseroptik tests its dispersive dielectric mirrors using a custom spectrally-resolved whitelight interferometer.