Advanced laser welding of high-performance thermoplastic composites

Owing to prominent strength and stiffness properties, continuous glass and carbon fiber reinforced composite structures are recognized as having a significant lightweight construction potential for a wide variety of industrial applications. Outside of the aerospace sector, carbon fiber reinforced plastic (CFRP) is gaining more and more importance in the automotive and energy sectors, including the renewable energy market. Furthermore, in the fields of transport, sports, and leisure, the amount of parts and components based on CFRP are increasing continuously.^1,2

Several technical and economic studies are stating a significant increase of the global demand of CFRP. CFRP production started at 75 000 ton in 2013 and is predicted to increase to over 100 000 ton in 2015 and even 200 000 ton in 2020.^2,3 The expected high growth rates for the next years are to be found in the start of serial productions of a new generation of civil aircrafts, e.g., the 787 (Boeing) or the A350 XWB (Airbus Industries). Furthermore, lightweight construction in the automotive sector, e.g., the i3 (BMW) or the XL1 (Volkswagen) as well as the realization of huge offshore wind turbines will be main reasons for an increased CFRP demand.² 

Nevertheless, in order to cope with high material costs of CFRP components compared to metallic structures, the production processes as well as the semifinished products of composites have been comprehensively analyzed.⁴ In this context, the three essential categories matrices, carbon fibers, and manufacturing processes were defined and each of them was analyzed with respect to the potential of respective possible savings. It is expected that the overall costs of CFRP parts can be reduced by up to 30% until the year 2020. Whereas the potential cost reduction for matrix materials is negligible, a reduction in production costs for carbon fibers of approx. 20% is assumed. However, the highest possible savings are expected in the field of production of CFRP. This comprises both the production costs during the manufacturing of the composite itself and the processing of components, e.g., the cutting, drilling, and joining steps. Here, an overall cost reduction of 40% is predicted.⁴ 

Compared to thermoset polymers, continuous carbon fiber reinforced plastics based on thermoplastic matrix materials offer several advantages, e.g., higher impact tolerances, uncritical and nearly unlimited storage times of semifinished products, high ultimate strains, and excellent chemical resistance as well as the possibility of quick-forming processing.⁵ The biggest difference between these two common matrices is the possibility of realizing welding connections while using thermoplastics, whereas mainly riveting and adhesive bonding are used for thermoset systems.⁶ For the manufacturing of structural elements providing high class technical functions, components of basic geometries have to be joined to complex assemblies. In consequence of these results, the development of appropriate welding technologies for fiber reinforced thermoplastics becomes more and more important.⁷ 

For the joining of fiber reinforced thermoplastics, different techniques are available, e.g., adhesive bonding, mechanical fastening, and welding.⁸ Adhesive bonding is characterized by a good fatigue resistance, corrosion resistance, and a high strength to weight ratio compared to mechanically fastened joints. The main drawbacks compared to other joining methods are a long curing time until the joint reaches the final mechanical load capacity and the pretreatment of the surface.⁹ For mechanical fastening, only a little surface pretreatment is needed and this technique provides the possibility to disassemble the parts. Disadvantages of mechanical fastening are local stress concentration, a lower strength to weight ratio, and the damage of fibers for continuous reinforced plastics.^9,10 In order to weld fiber reinforced plastics, techniques, such as resistance, ultrasonic, vibration, induction, and laser transmission welding, can be applied.⁸ 

The laser transmission welding technique is based on the optical transparency of thermoplastics for the wavelength of the laser radiation. This allows the laser radiation to pass the transparent part (LT) and generate heat in the absorbing part (LA). Laser radiation will be absorbed when the thermoplastic material contains appropriate additives, e.g., carbon black (c.b.) or carbon fibers. Due to heat conduction between the joining parts, the transparent part heats up and melts. In order to generate a uniform and reliable weld seam, it is necessary that the parts to be joined are pressed together gap-free. The laser transmission welding can be divided into simultaneous, mask, contour (C) and quasisimultaneous (QS) welding. For contour welding, which is typically utilized to generate three-dimensional weld seams, the laser beam is guided once over the work piece.^11,12 With this technique, the energy to melt the thermoplastic is brought into the material all at once. During the quasisimultaneous welding, the laser beam is guided over the work piece several times at high speed, introducing a slower heat development inside the welding zone.^13,14 To compare these two welding methods, the reference value energy per unit length E_S is introduced,

This reference value consists of the number of laser beam passes over the weld seam (n), the laser power (P), and welding speed (v).¹² 

The laser transmission welding process is affected by the kind of matrix material and the existence of fiber reinforcements. Glass fibers in the transparent part will scatter the laser radiation. Increasing glass fiber content will increase the possibility for the laser beam being scattered.¹⁵ Compared to polymeric matrix materials, carbon fibers are good heat conductors. The heat conductivity in the parallel direction is much higher than that perpendicular to the fiber orientation. Therefore, the heat conduction in CFRP is a combination of the heat conductivity of matrix material and carbon fibers as well as the carbon fiber orientation relative to the welding direction.¹⁶ The weld seam development is based on the process heat during the welding, which is affected by the local fiber orientation in the welding zone.^17,18

The welding experiments were performed using a Nd:YAG and a diode laser. The main specifications of both laser sources are summarized in Table I. For the investigations, bead on plate welding and transmission welding were performed. Therefore, the laser radiation is guided by mirrors or by optical fibers to a galvoscanner system, which focusses the laser beam by means of F-Theta-optics and deflects the beam across the CFRP surface and along the welding line, respectively (Fig. 1).

For bead on plate welding, no laser transparent part is used, so the laser beam directly hits the laser absorbing part. In order to characterize the influence of different reinforcements within the LA part on the melt pool, corresponding bead on plate weldings were analyzed with respect to the width of the resulting weld seam. The average weld seam width was determined over the width of three weld seams. In this context, the overall laser power as well as the energy per unit length were varied.

During the transmission welding process, the LT- and the LA parts were kept in contact by means of a joining pressure, provided by a pneumatic cylinder. The overlap arrangement consisting of LT- and LA parts is pressed against a glass substrate which is highly transparent for the NIR laser radiation and does therefore not influence the welding process. By movement of this overlap geometry relative to the laser beam, a contour welding and a quasisimultaneous welding process are realized.

In Table II, the main specifications of the materials together with the respective nomenclature are summarized.

The carbon fibers on the surface of the CF-PPS (polyphenylene sulfide) UD were orientated parallel (CF-PPS UD 0) or perpendicular (CF-PPS UD 90) to the welding direction. The laser transparent materials used in these investigations were reinforced by short and continuous glass fibers, respectively. These fibers scatter the laser beam and reduce the amount of radiation which will be available for the melting process of the matrix material inside the welding zone. In Fig. 2, the transmissivity of the laser transparent plastics used is depicted. The data provide a reference of how much laser radiation will pass through the transparent part and will be available for the actual welding process.

Within a defined range of energy per unit length E_s, laser welded samples were generated for static strength measurements. The specimens were prepared in overlap geometry. As a testing method, lap shear tests according to the standard DIN EN 14869-2 were chosen. Each test result was verified by testing four identical specimens which were prepared to be tested with the force along the warp direction of the composite material. All tests have been performed at room temperature. In order to visualize the welding zone, fracture patterns were prepared.

For bead on plate welding, the laser radiation is applied directly to the absorbing part without passing the transparent part. This provides the possibility to observe the direct effect of the heat conduction on the weld seam width of the CFRP. In bead on plate welding on unidirectional CFRP, the weld seam width was mainly affected by four effects:

Effect A: With increasing laser power and constant welding speed, more matrix material will be molten and the weld seam will get wider due to a change of the intensity distribution in the focal point.

Effect B: With increasing welding speed and constant laser power, the weld seam becomes smaller due to a shorter laser-material interaction time and due to a lower energy per unit length.

Effect C: The applied energy is higher than needed to melt the matrix material in the area of the focal point (d_F) (Fig. 3). For the CF-PPS UD 90, the heat conducts mainly out of the weld seam and the weld seam width increases by 2d_W so that the weld seam width is given by d_S = d_F + 2d_W. For CF-PPS UD 0, the heat mainly conducts ahead and behind the laser radiated area. The heat conduction area, which increases the weld seam width d_S by 2d_W, is lower compared to those for CF-PPS UD 90.

Effect D: The quasisimultaneous welding has a certain time span between each laser beam pass relative to a point on the surface of the CF-PPS UD. In this time span, the material cools down and the heat conducts along the carbon fibers. If the laser radiation provides just enough energy to melt the matrix material in the laser radiated area on CF-PPS UD 90, part of the applied energy is conducted out of the laser irradiated area and does not contribute to the melting of the matrix material. This is repeated every time the laser beam passes.

In Fig. 4, the dependency between the laser power and the weld seam width for CF-PPS UD 0 is depicted. For both welding speeds, the weld seam width increases with an increasing laser power, which is mainly dedicated to effect A. Furthermore, the progression of the weld seam width for both welding speeds has an almost constant offset (Δd₁ ≈ Δd₂). This indicates that the effect B has a higher influence on the weld seam width development than the effect C, which is due to the low heat conductivity perpendicular to the carbon fibers.

This is different for the weld seam width development on the CF-PPS UD 90 (Fig. 5), where the offset between both welding speed progressions increases with increasing laser power (Δd₃ < Δd₄). The effects A and B should have similar influences on the weld seam width as for the CF-PPS UD 0. This indicates that the weld seam width is strongly affected by effect C for higher energy per unit lengths. For the laser power of P = 2 W, the difference in the energy per unit length for the two speeds is ΔE_S = 0.15 J/mm. This increases for the laser power of P = 10 W to ΔE_s = 0.75 J/mm.

For the laser transmission welding of thermoplastics providing low transmissivity, it is necessary to generate the heat slowly in the weld seam to avoid degradation of the laser transparent part. Therefore, the welding technique must be changed from contour to quasisimultaneous welding. The welding technique influences the geometry of the weld seam, especially for the welding of fiber reinforced plastics. In Fig. 6, the influence of the applied energy per unit length on the average weld seam width for contour and quasisimultaneous bead on plate welding of CF-PPS UD 0 is depicted. Independent of E_s, the weld seam width for quasisimultaneous welding is lower than for the contour welding. As described in effect D, the CFRP cools down after each pass of the laser beam, but due to the fiber orientation parallel to the weld seam orientation, the generated heat remains in the weld seam. Furthermore, the offset between the progressions of the weld seam width is almost constant (Δd₅ ≈ Δd₆) as it was observed in Fig. 4.

For the quasisimultaneous welding of CF-PPS UD 90 (Fig. 7), the offset between the progressions increases for increasing energy per unit length (Δd₇ < Δd₈). This is due to the effects described for the contour welding of CF-PPS UD 90, but effect D also has an influence on the weld seam development. Between each laser beam pass, the material cools down and some of the generated heat is conducted out of the weld seam. A certain amount of the heat is lost for the melting process of the matrix material so the weld seam is smaller than for contour welding. As an example, for an energy per unit length of E_s = 0.2 J/mm, the weld seam width for contour welding is d = 1.19 mm but for quasisimultaneous welding (v = 1000 mm/s) an energy per unit length of E_s = 1 J/mm is needed to generate a weld seam width of d = 1.16 mm. So in this case, a five times higher energy input is needed in order to generate almost the same weld seam width.

Especially, for welding combinations based on glass fiber reinforced thermoplastics with high fiber volume content, contour welding becomes critical due to a drastic decrease of the transmissivity for the laser wavelength. Therefore, higher energy per unit lengths have to be applied in order to generate a reliable weld seam, which could lead to thermal damage of the surface of the LT partner. In addition, for LA partners based on continuous carbon fiber reinforced composites, this effect will be emphasized due to the high thermal conductivity of the reinforcement compared to unreinforced a short fiber reinforced thermoplastics.

In this context, a change from contour welding to quasisimultaneous welding seems reasonable. In order to evaluate significant differences in the process behavior and in the resulting weld seam strengths, overlap samples have been prepared, consisting of PA 6.6 GF 50 as LT part and CF-PA 6.6 as LA part. In Fig. 8, the resulting seam strengths as a function of energy per unit length are presented for both joining strategies. The measured data represent the respective mean value and the corresponding standard deviation of four independently welded and tested lap shear specimens. The investigations were performed for a constant laser power of P = 100 W and varying welding speed in order to vary the energy per unit length for contour welding and varying number of cycles for quasisimultaneous welding, respectively. For contour welding, the preparation of the strength curve was aborted for energy per unit lengths E_s > 125 J/cm due to initiating thermal damage at the surface of the PA 6.6 GF 50, caused by the low transparency of this material for the process wavelength.

Obviously, independent of the welding technique, for composites on the basis of a PA 6.6-matrix in conjunction with a carbon fiber reinforcement, the typical curve structure, which is known from the laser welding of plastics, can be observed. The maximum achievable seam strengths are on a similar level, whereas for quasisimultaneous welding the strength curve is shifted to higher energies per unit length. A similar offset in E_s is known for the development of the weld seam width from the bead on plate welding investigations by applying contour and quasisimultaneous welding (cf. Figs. 6 and 7). In order to achieve the same temperature in the weld seam and to compensate for losses due to thermal conductivity, for quasisimultaneous welding, much higher E_s values must be applied. In consequence, differences in lap shear strength should be explainable by a varying formation of the weld seam.

In Fig. 9, fracture patterns, showing the connection areas of PA 6.6 GF 50 and CF-PA 6.6 for different energy per unit lengths, are visualized. For contour welding with E_s = 50 J/cm, a distinctive weld seam becomes visible, revealing adhering carbon fibers at the bottom of the laser transparent part. In contrast, for quasisimultaneous welding, the weld seam only consists of slightly fused surfaces. Differences in the resulting shear strengths can be directly attributed to differences in the extent of the respective connection areas. For higher energy per unit lengths (E_s = 125 J/cm), contour welding only results in a slightly wider weld seam compared to quasisimultaneous welding. However, beginning decomposition in the surface of the CF-PA 6.6 can be observed, resulting in comparable maximum lap shear strengths for both process strategies.

Referring to the transmission spectra in Fig. 2, it can be seen that the transmissivity of the short glass fiber reinforced PA 6.6 GF 50 at the welding wavelength of λ = 940 nm reaches a value of T_M = 26%. If the reinforcement is changed from short to continuous glass fibers, as it is typical for high-performance composites, the transmissivity decreases even more, making a laser transmission welding process much more difficult. For continuous glass fiber reinforced semicrystalline PPS with a thickness of s = 1 mm, the transmissivity drops down to T_M = 17%. In order to evaluate the weldability of glass fabric reinforced laminates as the laser transparent part, GFRP-CFRP welding was performed using the quasisimultaneous welding strategy.

In Fig. 10, the mean value as well as the corresponding standard deviation of the measured lap shear strength is shown as a function of the applied energy per unit length for the joining combinations GF-PPS–PPS c.b. and GF-PPS-CF-PPS FAB. Again, the data represent the results of four independent measurements per E_s. On the basis of PPS c.b. as the unreinforced, laser absorbing part, a similar shift of the strength curve to higher E_s can be observed analogous to contour welding while changing from unreinforced to continuous carbon fiber reinforced CF-PPS.¹⁷ Both experimental series were performed by varying the number of cycles at a constant laser power of P = 100 W and a scanning velocity of v = 2000 mm/s. Due to the low transmissivity of the GF-PPS neither of the two joining combinations could be realized by contour welding.

While for PPS c.b. as the laser absorbing part, the region of maximum seam strengths could be clearly determined, the recording of the test series for CF-PPS was aborted for energy per unit lengths of E_s > 350 J/cm due an extensive decomposition of the polymeric PPS-matrix on top of the GF-PPS. In order to compensate for the heat losses in the welding zone induced by the high thermal conductivity of the carbon fibers within the CF-PPS, a higher energy per unit length was applied, resulting in high radiant energy exposures. The significant higher lap shear strengths in the case of CF-PPS as the LA part are based on a broadening of the weld seam in consequence of the higher thermal conductivity of the carbon fibers compared to the PPS c.b.

In order to avoid thermal damages at the surface of the laser transparent part, the type of energy input during the quasisimultaneous welding process was modified. Therefore, the laser power was reduced to P = 80 W and the scanning velocity was increased to v = 4000 mm/s. Due to a slower energy input and the corresponding heat development inside the welding zone, the applied E_s had to be increased again in order to weld the combination GF-PPS–CF-PPS. As a consequence, the strength curve is shifted another time along the energy per unit length axis. The development of the lap shear strength as a function of E_s for the modified quasisimultaneous welding method is also depicted in Fig. 10. The area of maximum achievable seam strengths can be clearly determined. In addition, the GF-PPS surface remains intact.

Fundamentally speaking, it can be stated that by applying adapted quasisimultaneous welding techniques, it is possible to increase the required energy per unit length without damaging the surface of the LT part. This allows the welding of high-performance composites, e.g., thermoplastics based on continuous glass and carbon fiber reinforcements, which is not possible when using conventional contour welding strategies.

In this paper, fundamental investigations on the influence of laser transmission welding techniques on the weld seam characteristics, while using CFRP as an absorbing part, are presented. It has been demonstrated that the carbon fiber orientation at the surface of the CFRP largely effects the weld seam formation due to high heat conductivity along these fibers.

Bead on plate welding was performed as contour and quasisimultaneous welding and was conducted while varying laser power as well as energy per unit length, using unidirectional CFRP with a PPS matrix. Four main effects were determined, which influence the weld seam width. For the composite materials and the chosen welding parameters investigated in this paper, it was found that for quasisimultaneous welding an almost five times higher energy input is necessary in order to generate the same weld seam width as for contour welding.

Furthermore, the laser transmission welding process is affected by the transmissivity of the laser transparent part at the welding wavelength. Therefore, experiments were performed based on short and continuous glass fiber reinforced plastics, which were welded to fabric reinforced CFRP. It was demonstrated that by adapting the welding technique from contour to quasisimultaneous welding, low transparent composites can be joined to CFRP, which is not possible using conventional contour welding strategies. In this case, the required energy per unit length can be increased without damaging the surface of the transparent part. Due to the different thermal conductivity of the carbon fibers and the polymer matrix, for quasisimultaneous welding, higher energy per unit lengths have to be applied in order to generate a reliable weld seam.

The authors would like to thank the Federal Ministry of Economics and Technology (BMWi) for funding parts of these investigations within the EraSME project A'QUILACO (FKZ: KF2186414AB3) and the AIF Projekt GmbH for their support. Furthermore, the authors would like to express their gratitude to TENCATE Advanced Composites bv, The Netherlands, for the supply of the consolidated CETEX^® PPS composite laminates. Special thanks go to Sebastiaan Wijskamp.

Dr. Peter Jaeschke studied Physics at the University of Oldenburg, Germany. Since 1999, he has been with Laser Zentrum Hannover e.V., working in the field of laser-material processing. Key activities are the laser based cutting and welding of continuous carbon and glass fiber reinforced plastics. Currently, he is head of the Composites Group.

Verena Wippo works in the Composite Group at the Laser Zentrum Hannover e.V. Her main field of activity is the laser transmission welding. Mrs. Wippo has a diploma in Mechanical Engineering with the main focus on Machines, Systems and Automation in Production Engineering and Mechanisms and Robotics.

Dr. Oliver Suttmann has graduated in mechanical engineering at the University of Hannover in 2007. Since then, he worked at Laser Zentrum Hannover on ablation with short and ultra-short pulsed laser sources. He received his Ph.D. in 2013 and heads the Production & Systems Department at Laser Zentrum Hannover today.

Professor Dr.-Ing. Ludger Overmeyer studied electrical engineering at the University of Hannover. In 1996, he has diploma in electrical engineering, Ph.D. in mechanical engineering (production technology) at the Leibniz Universität Hannover. He heads the Institute of Transport and Automation Technology (ITA) at the Leibniz Universität Hannover and is member of the management board of the Institute of Integrated Production Hannover (IPH). Since November 2009, Professor Dr.-Ing. Overmeyer is member of the Executive Board of the Laser Zentrum Hannover e.V.