Laser cutting of carbon fiber reinforced plastic using a 30 kW fiber laser

Conventional machining of carbon fiber reinforced plastic (CFRP) results in a high tool wear and short tool life of regularly less than 50 m cut contour, which in turn produces high costs as well as machining quality issues.^1–3 Due to its wear-free mode of operation, laser machining is a promising alternative. However, thermal loading by laser processing has proved to be a major concern with potential industrial users. Although having a negligible influence on the static strength,⁴ visible damage to the material is still unacceptable for most users. Therefore, the reduction of the thermal damage during laser cutting of CFRP has lately been the focus of a number of researchers.^5–7 

One approach to reduce the thermal damage is to use short and ultra-short pulsed laser systems. It has been found that CFRP may be cut by picosecond^5,8,9 and femtosecond lasers,¹⁰ successfully limiting the dimension of the heat affected zone (HAZ) to a range between 10 and 100 μm, compared to a HAZ of typically ≫100 μm when continuous wave (cw) laser systems are used.^4,6,7 This however comes at the price of a slow effective machining speed in the range of v_eff ≪ 1 m/min due to the comparably low average power of ultra-short pulsed laser systems, which is often insufficient in terms of productivity.

Therefore, another approach to reduce the thermal damage is to cut CFRP at high intensities and high processing speeds, thus allowing minimum time for heat conduction. Bluemel et al.¹¹ investigated the HAZ in dependence of laser power of a fiber laser system. It was shown that by increasing laser power from P_L = 1 kW up to P_L = 4 kW, the cutting speed could be increased from v_eff = 5 m/min to approx. v_eff = 22 m/min, and the dimension of the HAZ is in general decreasing from approx. 500 μm to 250 μm. No further increase in cutting speed was detected when the maximum power of 5 kW was used. In contrast, experiments of the same authors¹¹ with a 16 kW disk laser using a thermoplastic matrix CFRP proved that even higher laser powers can yield higher cutting velocities of up to v_eff = 80 m/min, and the HAZ was further reduced to values of approx. 180 μm.

In Ref. 12, Weber et al. introduced a perpendicular heat flow model. For this model, the interaction of the laser beam with an infinite plane of carbon material and the thickness of a single carbon fiber has been studied. The process was divided in three phases:

1. heating phase, characterized by the energy input needed to heat the carbon fiber to its sublimation temperature of T_sub = 3,900 K, followed by

2. sublimation phase, characterized by the energy input needed to overcome the material's latent heat, followed by

3. cooling phase, characterized by no further energy input but by a heat wave propagating along the fiber into the material.

Weber et al. found that the amount of energy needed for sublimation is by far dominating the energy balance and that the time needed for sublimation can only be influenced by the laser intensity, and thus the total energy absorbed by the fiber is a function of laser intensity as well. The model then was used to calculate the lateral distance to the point up to where the matrix evaporation temperature of T_ev = 800 K is exceeded over time. The same was done for the structure damage temperature of T_dam = 450 K. The maximum distance from the interaction zone for both temperatures occurred in the cooling phase, yielding theoretical values for minimum achievable matrix damage, shown in Fig. 1 (black line: evaporation temperature and dashed line: structure damage temperature). The model was validated by inserting experimental values for HAZ obtained from literature,^4–8,13 which are in good agreement with the calculations. Hence, a further increase in laser power and thus intensity is a promising approach to further reduce the HAZ for the cw-approach investigated in this paper.

In industrial applications, a kerf width of approx. w = 200–250 μm is needed so that the cut out parts can be easily removed. Using a 30 kW laser system, a focal diameter of d = 250 μm can be obtained, while maintaining a high intensity of I = 6.11 × 10⁷W/cm². According to Fig. 1, an extent of the HAZ of ∼10–50 μm can be expected. In theory, a single laser pass strategy may be used with this setup for laminates of a thickness of up to t = 2.5 mm, still keeping the aspect ratio of 10 between laminate thickness and kerf width supposed in Ref. 14.

However, the increase in intensity also requests further increasing the feed rates. The energy needed for sublimation of the carbon fibers is approximately E_sub,fiber = 85 J/mm³, while the energy needed for sublimation of the matrix is negligible.¹⁴ The theoretically achievable feed rate then is given by

For a given laminate of the thickness t = 1.4 mm, a fiber volume content of φ = 37.8%, the focal spot diameter of d = 250 μm, a laser power of P_L = 30 kW, and the expected feed rate is v_th = 2.7 m/s. If a multipass strategy is used to further reduce the amount of energy input per pass, a further increase in feed rate is necessary, with a linear dependency of the feed rate v_th on the number of laser passes n. At the moment, no commercial scanner system is available that could provide such feed rates and handle the high laser power at the same time. Therefore, a specific experimental setup was designed, using a high-speed rotational movement of the specimens under the fixed laser beam.

Experiments were conducted using the following material and setup:

The material for the process was chosen to specifications from typical automotive applications of CFRP. 2D specimens were manufactured using vacuum infusion technology, consisting of a total of 6 layers of noncrimp carbon fabrics and an epoxy resin. Table I gives an overview over the resulting characteristics of the CFRP laminate.

A high power Yt:YAG fiber laser was used in combination with a special optical system that can handle the high intensities needed for the experiments. The laser beam was directed to the manufacturing cell in a 300 μm process fiber and then collimated and focused by an optical system with an image scale of S = 1:1.2 to achieve a theoretical focal diameter of d = 250 μm. To determine the actual beam characteristics in the focal plane, the intensity profile and beam parameters were measured with a focus monitoring system, Primes FocusMonitor. A homogenous top-hat intensity profile was detected.

The laser and optical system used have the specifications shown in Table II.

Normally, in laser materials processing, relative movement is realized either by movement of the workpiece or the beam with the means of a linear axis or alternatively by scanner systems. Commercially available scanner systems operate already with comparably high feed rates of several m/s.¹⁵ For simple geometries, such as a line or an oval, which may be considered ideal from a kinematics point of view, these systems may reach maximum feed rates of v = 10–20 m/s. However, no scanner systems are available allowing a laser power of P_L = 30.5 kW in combination with the desired extremely high feed rates. Therefore, a setup was developed, which uses a rotational axis to move small CFRP specimens while the laser beam is delivered through a fixed optical setup.

In order to allow for a detailed investigation on the influence of the number of passes in a multipass cutting strategy, the achievable feed rate should be high enough to enable a cutting process with n ≥ 30 passes at highest possible intensity. With Eq. (1), the theoretical feed rate per pass v_n for a multipass strategy is

This sets the requirement for the relative movement system to a maximum speed of v = 80 m/s for n = 30.

To achieve these feed rates, a rotary plate with a diameter of d_p = 0.715 m was manufactured from aluminum and connected to a motor, which was operated with a rotational speed of up to U = 2,250 min⁻¹. Stress analysis for the rotary plate using von Mises yield criterion showed that a simple rotary plate would theoretically withstand the occurring stresses with little reserve. The design of the rotary plate was therefore altered including several cut-outs to reduce the mass, compare Fig. 2.

The rotary plate possesses three flanges, each serving as a mount for a CFRP specimen. The specimens are fixed by holders, which partially also act as protection against the laser radiation for the flanges, and can easily be replaced (Fig. 3).

For the experiments, the movement system was enclosed by a case that serves as a mechanical protection and features a connection to the emissions extraction system. It was mounted to a precision two-axis-positioning table. A beam trap was positioned below the specimens to take care of the excess laser energy when the laser beam is not interacting with one of the three specimens. The optical system was attached to a fixed stand which allowed for a precise positioning in beam propagation direction by an additional movement system with an accuracy of Δ _z < 100 μm. The complete setup is shown in Fig. 4.

A one-pass strategy was used to define the process parameters, particularly the necessary energy per unit length and the achievable cut quality with the 30 kW system. Afterwards, a multipass strategy was investigated, and the influence of the number of passes on the extension of the HAZ was determined.

First experiments were conducted with the laser power fixed to P_L = 30.5 kW and varying the feed rate in order to define the maximum speed for which a cut through the full thickness can be obtained with one laser pass. The maximum achievable feed rate was v_(n=1) = 1.2 m/s, which is considerably lower than the expected feed rate from theory of v_th = 2.7 m/s. Thus, a 2.25 times higher amount of energy was deposited in the specimen than theoretically necessary. The energy per unit length, defined as

can be calculated for the one-pass strategy to E_s = 25.4 kJ/m.

Figure 5 shows a cross section of the obtained cut with a single laser pass at full laser power and maximum achievable cutting speed.

The lateral extension of the heat affected zone was measured from the cross section as a mean value over the thickness of the laminate. For each of the three specimens positioned on the flanges of the rotary plate, left and right side of the cut have been investigated, yielding six separate values for the HAZ. The mean value of the HAZ for n = 1 was found to be 139 μm.

This is considerably higher than the 10–50 μm expected from the diagram in Fig. 1 and could well be a result of the 2.25 times higher energy input than necessary for sublimation. A possible explanation is that losses occur due to shielding of the laser power by process emissions. In addition, fissures are occurring especially close to the top surface yielding a maximum extent of the HAZ of 472 μm. Single carbon fibers are bulging from the surface, indicating high process pressures. As the matrix material is highly transparent for the wavelength of the laser, part of the laser beam may penetrate deeper into the material until being absorbed by the first layer of carbon fiber, which may lead to a heat source below the top surface. Thus, evaporation may partially start from within the material, which could be the origin of the process pressure.

A multipass strategy was investigated to further reduce the interaction time between laser beam and material per pass. For the experiments, the laser power was again kept constant at P_L = 30.5 kW. It was the aim of the investigations to define the actual amount of energy input necessary to achieve a full cut for a given number of passes n. According to Eq. (2) and with the actual value for the feed rate in a one-pass strategy, the expected feed rate in a multipass strategy can be calculated and was used as a starting value, e.g., for n = 2, the feed rate was initially set to v_n = 2.4 m/s. If the specimen was fully cut through under these conditions, the feed rate was increased by a value of 0.1 m/s. If not, the feed rate was reduced by 0.1 m/s. The experiment was iterated until a maximum feed rate for a given number of passes was defined.

Neglecting energy losses, the same amount of total energy for sublimation of the material would be necessary, indifferent of the number of passes. Therefore, a cumulative energy per unit length E_s,c is proposed, to compare the process efficiency in a multipass strategy

with E_s,i being the energy per unit length of each pass.

Figure 6 shows the minimum cumulative energy per unit length that is necessary to cut the material in a defined number of passes n. It can be concluded that the necessary energy input for a multipass strategy is at first decreasing with increasing number of passes n. After reaching a minimum, the energy input sharply increases again with n > 20.

Furthermore, the effective feed rate, given by the feed rate per pass v_n multiplied by the number of passes n, is shown in Fig. 6. Solving Eq. (3) for the feed rate v, the effective feed rate is

Less energy needed to achieve a full cut is therefore directly related to a higher possible process speed. Optimum values were achieved for n = 16, with E_s,c = 18.8 kJ/m and v_eff = 1.63 m/s. The actual amount of energy needed is therefore reduced to approx. 1.6 times the theoretical energy input needed for evaporation, compared to a factor of 2.25 identified for the one-pass strategy. Obviously, the energy input was used more efficiently for the evaporation process. In a multipass strategy with n = 16, only 1/16th of the total volume is sublimated per pass, which may be favorable in terms of less emissions that could shield the laser beam, reducing energy losses.

The HAZ was derived from the cross sections of the cut specimens and is shown in Fig. 7 in dependence of the number of passes n.

As shown in Fig. 7, the HAZ is slowly decreasing with the number of passes n up to n = 12 passes, finding its minimum at slightly lower n than the minimum of the cumulative energy per unit length. With further increasing number of passes n, a steep increase in the dimension of HAZ was observed. In general, a good correlation exists between the overall energy input (Fig. 6) and the HAZ (Fig. 7), whereas the minimum of the curve is more pronounced in the case of the cumulative energy per unit length.

To further investigate the phenomenon of increasing HAZ for a number of passes n > 12, the HAZ is plotted in dependence of interaction time per pass t_int and delay Δt between two passes (Fig. 8). The interaction time between the laser beam and the material per pass can be described as

The delay is defined by the time the laser spot needs to travel the remaining circular pass after leaving the specimen with the given feed rate until it reaches the same position on the specimen again and is therefore also a function of the feed rate per pass v_n.

It can be concluded that the HAZ sharply increases when the delay time falls below approx. Δt < 100 ms. A similar heat accumulation effect for delay values below Δt = 200 ms was also observed in literature.¹⁶ Taking into account that the heat accumulation will be influenced by specimen and cut geometry, the order of magnitude of the critical delay is in good agreement.

Figure 9 shows the cross section of a specimen cut with n = 12, at the minimum of the HAZ curve. Over the central part of the 1.4 mm thick specimen, the HAZ is in a range < 50 μm, which is in good agreement with the theoretical values from Fig. 1. However, close to the top and to the rear surface, some fissures and chipping can be observed, same as was seen in Fig. 5 in the case of the one-pass strategy. This again may be a result from high process pressures evolving from partial matrix evaporation within the material.

Figure 10 shows the cross section of a specimen cut with n = 16, at the minimum required energy input. In general, the behavior seems to be similar to n = 12, with a slightly larger variation in HAZ due to imperfections in the laminate. It can be concluded that regarding the quality of the cut no significant difference is observed between n = 12 and n = 16. Thus, the cutting operation can be optimally performed at n = 16 yielding the highest effective feed rate.

Using 30.5 kW of laser power focused down to a spot of 244 μm in diameter, it was possible to cut a 1.4 mm thick CFRP laminate in a one-pass strategy at v = 1.2 m/s and with a mean HAZ of 139 μm. With a multipass strategy, finding its optimum in a range between n = 12–16 passes, the HAZ was further reduced down to 78 μm at n = 12, and the energy input needed was reduced by approx. 26% to E_s,c = 18.8 kJ/m. The energy input needed was approx. 1.6 times the energy needed for sublimation, and the HAZ was in good agreement with the model of Weber.¹² 

The effective feed rate can be increased up to v_eff = 1.63 m/s with the ultra-high power laser system used, which makes it an attractive option for industrial applications with high production volumes and medium quality requirements. However, fissures and chipping behavior were observed especially close to the surface of the specimens, probably resulting from high process pressures. It is assumed that this behavior results from an insufficient absorption of the laser wavelength by the matrix, leading to fast energy deposition within the material and finally to partial evaporation below the surface. Further investigations, e.g., cutting experiments with a laminate including absorbing particles, such as carbon-black in the matrix, are needed to prove or disprove this theory. A simulation of the behavior could be helpful to define the optimum absorption characteristics of the material.

Part of this work was supported by the German Federal Ministry for Economic Affairs and Energy (BMWi) within the frame of the Project No. 01 MX 12049. The authors would like to thank the BMWi as well as the PT-DLR project management agency for their support. Furthermore, the authors express their gratitude to the companies Precitec Optronik GmbH for providing the optical system and Rhein Composite GmbH for providing the carbon fiber fabrics.

Dr.-Ing. Dirk Herzog has studied Mechanical Engineering in Hanover, Germany. He started his career with the Laser Zentrum Hannover e.V. in 2002, leading the group “Cutting, Safety & Special Processes” from 2005, and the department “Materials and Processes” from 2008. Also in 2008, he received the degree “Dr.-Ing.” from the Leibniz Universität Hannover. After working for the Grant Advisory of Ernst & Young, he joined the LZN Laser Zentrum Nord GmbH in 2011, where he is currently Head of R&D and also Senior Engineer with the Institute of Laser and System Technologies of Hamburg University of Technology (TUHH).

Dipl.-Ing. Matthias Schmidt-Lehr has studied Mechanical Engineering at TUHH and is a research associate at the Institute for Laser and Systems Technology within the competence field of synthetic materials processing since 2013.

Dipl.-Ing. Marten Canisius has studied Mechanical Engineering at TUHH and is a research associate at the Institute for Laser and Systems Technology since 2011. He leads the competence field of synthetic materials processing.

Dipl.-Ing. Max Oberlander has studied Mechanical Engineering at Friedrich-Alexander-Universität Erlangen-Nürnberg, Germany. He is a research associate at the Institute for Laser and Systems Technology within the competence field of synthetic materials processing since 2014.

B.Sc. Jan-Philipp Tasche is a student of mechanical engineering at TUHH. He conducted his Bachelor Thesis “Investigations on the influence of the interaction time during laser cutting of CFRP” at iLAS.

Professor Dr.-Ing. Claus Emmelmann is managing director of LZN Laser Zentrum Nord GmbH as well as director of the Institute for Laser and Systems Technology.