Induction‑based joining of titanium with thermoplastics Creation and examination of titanium‑thermoplastic connections

The importance of hybrid composites is increasing, particularly with regard to structural lightweight construction. Therefore, a joining process for the production of hybrid bonds using induction technology is examined in this paper. An appropriate test rig was developed, assembled, commissioned and characterized for this purpose. Heating tests with titanium Ti6Al4V were carried out with this device to determine the influence of the process parameters on the temperature level and the temperature field. This applies in particular to the feed rate, the sample geometry, the relative positioning between sample and inductor as well as to the inductor setting like induction frequency and pulse width modulation. Based on the results of these examinations, a two-stage heating process was developed, which allows the formation of a homogeneous temperature field in the range of the melting and degradation temperature of the used polymer (PPS; Ten-Cate Cetex TC1100) and thus to achieve optimized bond strengths. More specifically, a short phase with a high energy input was used for quick and intensive warming and a longer time period was deployed to reach an even distribution of temperature. Subsequently, joints with CFRP were produced with a specimen geometry suitable for tensile shear tests according to DIN EN 1465 (Adhesives—determination of tensile lap-shear strength of bonded assemblies; German version EN 1465:2009, 2009). The analysis of the tensile shear strength and the fractured surfaces shows that the metal is much more difficult to heat up due to the melting enthalpy of the plastic than preliminary tests indicated. Further tests were carried out, which revealed a threshold value for heating where the temperature was sufficiently high to achieve a complete coating of the plastic on the metal can be achieved. In summary, the investigations showed that the presented method as well as the design of the test rig is generally suitable for bonding metal to thermoplastic polymers. The influence of various parameters on heating and bond strength could be identified. Moreover a process has been established, which allows industrially useful bonding forces. This not only proves that further investigations are useful, but also laid the foundation for further experiments.


Introduction
Lightweight construction is a major concern of the transport sector in order to increase energy efficiency and reduce emissions. One approach is to use different materials as required. However, this leads to heterogeneous compounds, such as metals and fiber reinforced composites (FRC), which have to be joined.
Thermal joining is an appropriate process in this case. The advantages are short cycle times, the elimination of additional material and less fibre damage compared to conventional processes such as screwing, riveting or adhesive bonding. Recent tests on adhesive joints have also shown that laser-assisted pretreatment of the metal surface can significantly increase the strength of the joint [2,3]. For this paper, the heat supply for thermal joining was realized by electromagnetic induction to enable a contactless and flexible joining process.

Objective and approach
The aim of the work presented here was to identify a process window for the inductive joining of titanium with thermoplastics. A test rig was designed and constructed for this purpose (Sect. 4). Subsequently, a test design was developed and corresponding tests were carried out for heating titanium samples (Sect. 5) and for joining titanium with a carbon fiber reinforced polymer (CFRP) (Sect. 6). The results of the test showed the basic suitability of inductive heating for thermal joining (Sect. 7).

Examined materials
The titanium alloy Ti6Al4V was used as a metallic joining partner due to the diverse fields of application in aerospace, marine technology and motor sports. This alloy is characterized by its low density, good ductility and high strength compared to other titanium alloys [13]. Table 1 contains selected property information for this alloy. Table 1 Properties of Ti6Al4V according to [13] Chemical The CFRP Cetex TC1100 from TenCate Advanced Composites B.V., Nijverdal, Netherlands, consists of a semi-crystalline thermoplastic polyphenylene sulfide (PPS) matrix with a fiber to volume ratio of 50% and a fiber to weight ratio of 43% . The laminate, Carbon T300 3K 5HS 280 GSM, consists of 3000 fibres per thread in 5HS-woven fabric structure, as shown in Fig. 1, and has a grammage of 280 g∕m 2 . Table 2 gives an overview of the characteristics of the CFRP and its matrix material.

Requirements for the test rig
The tensile shear strength according to DIN EN 1465 has to be determined to evaluate the strength of the joint [1]. This approach is suitable considering the comparability of adhesive connections with direct thermal bonding and the fact that this test has been successfully used in several investigations [3,11,17]. The geometric dimensions of the specimens were determined based on this standard to a length of 100 ± 0.25 mm and a width of 25 ± 0.25 mm , with an overlap of the joining partners of 12.5 ± 0.25 mm [1]. For later industry-relevant investigations, a connection of samples with a width of up to 400 mm should be possible. To achieve a certain mechanical strength, the range of material thickness was defined to be 0.3-5.0 mm. The sample dimensions are shown in Fig. 2.
The temperature range at the surface of the metal to be achieved by induction can be derived from the thermal properties of the thermoplastic joining partner. On the one hand, the thermoplast must be heated above the melting temperature in order to achieve wetting of the metal; on the other hand, the degradation temperature must not be reached. A low viscosity is advantageous for wetting the metal surface with the plastic melt. To achieve this goal and prevent thermal degradation of the polymer, an operating point just below the degradation temperature is advantageous. Frequently used representatives of high-performance plastics with the highest operating, melting and degradation temperatures are polyether ether ether ketones (PEEK), which melt at 343 • C and decompose at 450 • C [18]. For this reason, the maximum temperature to be reached with the test rig has been set to 450 • C.
Another parameter that was specified was the operating pressure. It must be high enough to bond the surface structures of the metal completely with plastic, but on the other Fig. 1 Schematic drawing of the 5HS-woven fabric structure, distinguishing the warp and fill direction. The non-transparent yarns consist of the repetitive unit cells (based on [14]) fill direction warp direction hand, the polymer must not be pressed out of the joining zone by too high pressure. Therefore, an operating pressure range of 0.1-1 bar was chosen.

Setup of the test rig
In order to fulfill the requirements mentioned in Sect. 4.2, an 'IT-0069' linear inductor in coaxial design from IFF GmbH, Ismaning, Germany, was selected for heating tests with titanium. It is controlled by a frequency generator 'EW050W' from the same company and cooled by a circulating air heat exchanger 'ProfiCool Novus PCNO-20.03-NED' from National Lab GmbH, Moelln, Germany.
The inductor was mounted on a device comprising two spindle modules of type 'SET-25-AWM-30-HK-PA-H' and 'SHTP-01-06-AWM-20-PA-HR' of IBUS GmbH, Cologne, Germany. These allow a positioning accuracy of the inductor of 0.1 mm relative to the joining zone.
The remaining degrees of freedom, the assembly and the parameters d I and p I , which can be varied by this module, are shown in Fig. 3.
The test specimens were mounted on an single fixture with quick-release. Figure 4 gives an overview of this module. The fixture for the titanium was made of steel due to the high requirements on temperature resistance. An adjustable quick-release 'K0660.006001' from Heinrich Kipp AG, Sulz am Neckar, Germany, was attached to it. This entire part can be moved in y-direction to vary the overlap length of the specimens in the range of 2.5-22.5 mm. The holder of the thermoplastic was made of aluminum, it can be tilted via hinges and it is height-adjustable by means of screws. This allows different sample thicknesses to be compensated for and to ensure the parallelism of the joining surfaces. An additional balancing mechanism has been installed to guarantee freedom from force. To avoid distortion, the thermoplastic specimen can be clamped down with two small quick-releases 'K0660.004001'. The assembly of the two parts to be joined is shown in Fig. 5.
To realize the feed velocity, the above-mentioned modules were mounted on two toothed belt axes 'ZLW-1040-02-B' of Igus GmbH, Cologne, with a traverse path of 1300 mm each (see Fig. 6). They are controlled by an 'EMMS-ST-57-M-SE-G2' stepper motor from Festo Vertriebs GmbH & Co. KG, Munich, Germany. To ensure synchronous movement, the motor and the axes are connected to each other via a shaft coupling. In this way, line feeds in the range of 0-2134 mm∕s can be realized.
In order to apply the required joining pressure, a device was constructed consisting of two components (see Fig. 7). The upper one is designed to apply the joining force and contains a guide unit with two guide shafts 'SFUL4-38-B8-S8' from MiSUMi Europa GmbH, Schwalbach am Taunus, Germany, six pneumatic cylinders 'ADN-12-50-l-P-A-S11' from Festo Vertriebs GmbH, Esslingen-Berkheim, Germany, eight return springs 'Z-060EI' from Gutekunst+Co KG, Metzingen, Germany, and a joining head. While the add-on parts ensure clearance-free positioning of the entire upper part, the guide shafts ensure movement only in z-direction and absorb forces and moments in the other spatial directions.
The pneumatic cylinders provide the necessary power transmission to the joining head.
The joining head is equipped with 17 rows of three ball bearings SS 604 ZZ 4 mm × 12 mm × 44 mm from Kugellager-Express, Schloss Holte-Stukenbrock, which touch the top of the thermoplastic and apply the joining pressure. It covers an area of 200 mm × 13 mm . The return springs have two functions: First, they pull the joining head back into its initial position in the event of an emergency stop. At the same time, they provide a preload and thus compensate for the frictional forces of the guide and the cylinders during pressure adjustment. The lower part serves as support for the metallic joining partner in order to avoid an undesirable deviation of the sample. For this purpose, a steel guide rail is supported over the joining area.
By taking into account the specified joining pressure P j of 0.1-1 bar (see Sect. 4.2), the overlap O j of 12.5 ± 0.25 mm according to DIN EN 1465 [1] and the width of the sample W j of 25 mm (according to standard [1]) to 200 mm (length of the joining head) the required joining force F j can be calculated.
With these boundary conditions the minimum force ( F j,min ) and the maximum force ( F j,max ) can be calculated. The results are given by Eq. 1: To cover the specified force range and to minimize the influence of friction caused by the piston rod movements, two operational areas with two corresponding configurations of the joining unit were specified. The first one, concerning smaller joining forces between 3.1 and 87.5 N, which are relevant for specimens from 25 to 70 mm, uses only two cylinders and four return springs. In the second one, used for larger forces from 87.5 to 250 N, concerning samples of a width above 70 mm, four cylinders and eight return springs were used. These configurations were realized by plugging and unplugging of (1) (2) F j,min = 0.01 N/mm ⋅ 12.5 mm ⋅ 25 mm = 3.1 N, single cylinders and via mounting or demounting of the particular springs. It is important to note that the piston rods have no fixed connection to the joining head and are not moved without the aid of compressed air.
Due to the interaction of the pneumatic cylinders and the return springs, there is a linear relationship between the joining force F j and the operating pressure P w . This situation is visualized in the force-pressure-graph in Fig. 8. Due to the spring constant and the invariable cylinder faces, the gradient is steady for each configuration. However, the vertical position of the curve depends on the working distance z w . Relating to this, the upper surface of the guide rail is defined as the point of origin for the working distance z w0 with the associated initial joining force F j,0,w0 .
The change in the joining force F j,z w in relation to the variation of the working distance z w can be expressed by a further gradient k z A : Observing the conditions mentioned above, the joining force F j can be generally described as To determine the configuration-specific relationship between joining force and operating pressure, the variables of Eq. 7 had to be identified. For this purpose, the characteristic curves of each configuration were recorded at a working distance of 4.0 mm and 11.5 mm. Additionally, a regression analysis was carried out to estimate the values of F j,0,w0 and m w . This was done with a force sensor 'KD-40s' and the associated measuring technology of ME-Messsysteme GmbH, Henningsdorf, Germany. The results are given in Figs. 9 and 10.
On the basis of these findings, the gradient k z A could be calculated by of Eq. 5. Together with an averaging of the slopes and by inserting the particular y-intercepts into Eq. 6, the joining force F j , in dependence on the operation pressure P w and the working distance z w , could be estimated. The results for configuration 1 with two cylinders and four springs and for configuration 2, containing six cylinders and eight springs, are given in Eqs. 8 and 9: Operating pressure P w Joining force F j measuring points for z w = 4.0 mm measuring points for z w = 11.5 mm F j (4.0 mm, P w ) = 13 m 2 · P w − 48 N, R 2 = 0.99 F j (11.5 mm, P w ) = 13 m 2 · P w − 44 N, R 2 = 0.99 Fig. 9 Joining force F j as a function of the operating pressure P w for the joining device configuration with two cylinders and four springs For monitoring and documentation of the joining process, three infrared thermometers 'DM301 D' of the B+B Thermo-Technik GmbH, Donaueschingen, Germany, with a measurement accuracy of 1.5% were installed. They are located in the guide rail of the lower component of the joining device (cf. Fig. 7). The first one is positioned at the beginning of the rail to estimate the temperature right after the heating process, whereas the second one is in the middle of the rail. The last one measures the final temperature at the end of the guide rail. The enlargement ratio of the measuring beam is 15:1 and the initial diameter at a measuring distance of 0 mm is 7 mm. Since the sensors are located 19.0 mm beyond the metallic joining partner, the diameter at the surface is 7.19 mm and the results are averaged over an area of 40.6 mm 2 .
For automation purposes, a cDAQ-Chassis '9172' of National Instruments Germany GmbH, Munich, Germany, was installed. Together with the corresponding software 'LabView 2012' and a standard desktop PC, the process parameter can be set and documented.
All above mentioned devices and units are installed in a framework made of Bosch Rexroth aluminium profiles ( 45 mm × 45 mm ) (see Fig. 11). The frame construction is positioned on a worktable and enclosed on four sides with acrylic glass to only allow front access. To prevent hazards to people due to thermal and mechanical effects as well as by induction fields, a safety light curtain MLC510 of Leuze electronic GmbH + Co. KG, Owen, Germany was implemented. The latter shuts down the system when accessed during operation.

Inductive heating of titanium
As described in Sect. 4.2, it is necessary to exceed a certain temperature to bond metal and thermoplastics. For this reason, heating tests were carried out with titanium before the joining tests.
To evaluate the heating, an infrared thermographic camera 'IR-TCM 640' from InfraTec GmbH, Dresden, Germany, with the corresponding software 'IRBIS 3 plus' was applied. In order to enable a valid measurement, the titanium specimens were pretreated with graphite spray 'GRAPHIT 33' from CRC Industries GmbH, Iffezheim, Germany, to ensure an emissivity of 0.95 [19].
For a preliminary investigation, the manufacturer's recommended settings with an induction frequency of f p = 20 kHz , an induction time of t p = 20 s , and a pulse width modulation value of PMW p = 350 ‰ were used. The joining zone was positioned in y-direction in the center of the inductor ( p I = 6.25 mm , cf. Fig. 3) and at a vertical distance of d I = 0.5 mm . The result is displayed in Fig. 12.
The thermographic image shows a inhomogeneous temperature distribution with two distinctive heat sources. The maximum temperature T max was 398 • C , the minimum temperature T min = 241 • C , and the average temperature T avg = 312 • C over the joining area. This posed the following challenge: While the maximum temperature exceeded the degradation temperature T d = 370 • C of the used PPSmatrix, T min did not even reach the melting temperature T m = 280 • C (cf. Sect. 4.1). Therefore, it was expected that a strong connection cannot be achieved in this way.
Starting from this first experiment it was necessary to investigate the influence of different process parameters  The feed rate v p was examined within a region from 0 up to 9 mm/s in steps of 3 mm/s . Figure 13 shows an example of a static specimen in comparison to a moved specimen with the corresponding temperatures. A difference between a stationary and a moving sample in the formation of the temperature field is not recognizable. Since no influence of the feed rate could be determined, the next tests were carried out with a constant feed rate of v p = 3 mm/s.
The effect of the sample width was estimated by heating the titanium with constant parameters of f p = 20 kHz , t p = 20 s , PMW p = 350 ‰ , p I = 3 mm , d I = 0.5 mm , and dimensions of b = 25 mm, 80 mm, 130 mm . The results are displayed in Fig. 14.
A direct comparison of these three cases shows that the thermal inhomogeneity is caused by the sample edges directed transversely to the inductor axis. These edges lead to a lateral displacement of the heating and to less heating of the sample corners. It can therefore be concluded that homogeneous heating is possible for large samples, but critical for small samples. On the basis of this insight, the influence of positioning can now be investigated.
The lateral sample position p I was varied between 0 and 20 mm in steps of 2 mm to determine the formation of the temperature field by using the same parameters for the inductor and a sample with a width of b = 25 mm . The result after a process time of 5 s , at which the difference could be observed best, is given in Fig. 15.
The figures show correlations of the number of heat spots (in the following called heat sources), their position, and the intensity on the lateral deviation p I . An analysis of the achieved temperature difference T = T max − T min in the joining area indicates a minimal temperature disparity of T = 78 • C at p I = 16 mm . This corresponds with the symmetrical and dense distribution of the heat sources in the corresponding thermographic image.
On the basis of these findings, the influence of the distance between inductor and sample d I (cf. Fig. 3) was investigated at the lateral shift of p I = 16 mm . The value of d I was successively increased to 0.5 mm, 1.0 mm, 2.0 mm, 3.0 mm, 4.0 mm, 5.0 mm, 6.0 mm. The pulse width modulation PMW p was adjusted so that T max was approximately 400 • C and the results therefore comparable. Since the PMW p has its maximum at 750 ‰ , which was reached at d I = 4 mm , the specified T max could not be achieved at  Fig. 16.
From the necessary increase of the PWM p value to get the same temperature, it can be concluded that the transmission efficiency decreases with increasing distance d I . Furthermore, it can be seen that the distance y between the heat sources' centers and the central axis of the inductor grows with rising d I . As a consequence of this, the temperature difference within the process zone takes the lowest value of T = 96 • C at the smallest distance d I = 0.5 mm . Therefore, the smallest distance of 0.5 mm was maintained in further investigations.
The influence of the induction frequency f p was determined by varying the value in a range between 15 and 20 kHz in steps of 2.5 kHz . The remaining parameters were set to t p = 20 s , PMW p = 350 ‰ , p I = 16 mm , and d I = 0.5 mm , according to the above mentioned results. The observations after a process time of 7 s , at which the difference could be observed best, are shown in Fig. 17.
It could be shown that the warming depends on the frequency, with the strongest warming at f P = 15 kHz , but is independent of the distance of the heat source and the central axis y . The smallest temperature difference of In all investigations described so far, the temperature difference T was larger than 78 • C . This is seen critically in the light of the specified process window from T m = 280 • C to T d = 370 • C [16], whereas a typical process temperature is defined to be T p = 330 • C . This means that the achieved temperature range is just below the size of the process window or exceeds the optimum process window significantly and therefore it is presumably not possible to create a stable joint connection in this way.
To meet this challenge, the process was divided into two parts: a short sequence with a high energy input for a rapid and strong heating of the material and a longer sequence with low energy input to achieve a homogenization of the temperature field. Based on the previous parametric studies, the settings presented in Table 3 have been made. The results of the corresponding experiments are displayed in Figs. 18 and 19.
As can be seen from the graphs, it is possible to comply with the specified conditions with the applied parameter settings. Both the minimum and the maximum temperature were within the desired range over a certain period and the difference between them was much smaller than the process window. Moreover, T was decreasing in the course of the process. This continued until a time of t p = 80 s , when the temperature difference no longer decreased. Shortly thereafter, the maximum temperature T max exceeds the upper limit of the process range. As a conclusion of these findings, the further experiments were executed with an adapted process time t p = 70 s , as shown in Table 4. For reasons of statistical validation, the heating experiments were repeated nine times with the adapted parameter set. While the temperatures at the end of the process are given in Fig. 20, Table 5 lists the global temperature values. Both show that all deviations are Fig. 17 Influence of the induction frequency f p on the formation of the temperature field after 7 s of heating

Fig. 18
Maximum temperature T max , average temperature T avg and minimum temperature T min as a function of the process time t p during the heating with the process parameters given in Table 3 within the specified process window. Furthermore, it can be seen that the fluctuations are below or close to the measuring accuracy of the thermal camera of 2% . It can therefore be stated that sufficient repeatability is achieved for the following joining tests.

Induction based joining of titanium and thermoplastics
The surfaces of the metallic joining partners were modified in order to carry out the joining tests. For this purpose, a nano-structure according to Ref. [20] was applied by the aid of a pulsed laser source 'Powerline E Air 25' from ROFIN-SINAR Laser GmbH, Hamburg, Germany. This procedure proved to be useful for the improvement of the adhesion behavior in the past. The parameters used are given in Table 6. Whereas the titanium specimens were purchased from Airbus Group, Leiden, Netherlands, the CFRP samples were delivered from the same manufacturer in plates and cut with a water jet.
To realize a joining, the joining partners have to go through the two-stage warming process described in Sect. 5.
The specified process pressure was adjusted according to Eqs. 8 and 9 respectively and applied by using the joining device. Since the investigations of Sect. 5 showed that it takes about 65 s for the metal to cool down below the glass transition temperature of 90 • C , the feed rate for the pressure application was set to 3 mm/s . In this way, the complete strength is reached at the end of the joining process.    For a first exploratory study, the experimental design shown in Table 7 was used. The numbering reflects the execution order; each measurement was repeated once. The tensile shear strength of the compound was measured by means of a tensile/compression testing machine 'Z020' of Zwick GmbH & Co. KG, Ulm, Germany. The test procedure was force controlled and a force increase of 125 N/s was applied.
The results are displayed in Figs. 21 and 22. The data show that tensile strengths between 21.1 and 25.3 MPa could be achieved. The maximum value resulted from the combination of P J = 0.6 N/m 2 and PWM 2 = 193 ‰ . However, the optical examination of the fracture surfaces revealed that only partial bonding was accomplished. Figure 23 shows this situation: The areas of cohesion and adhesion fracture as well as a section without connection can be clearly differentiated.
The damage pattern leads to the conclusion that the necessary heating could not be achieved over the whole area, despite the use of the parameters determined in the preliminary tests. This is probably caused by the heat transfer between the joining partners and the heat conduction in the fiber composite material. Based on these results, the further procedure was designed as follows: In order to investigate the relationship between the process temperature T p , the melting process, and the tensile shear strength , only the titanium sample without a plastic sample was initially heated and the temperature T 1 was measured at the end of the heating by means of the built-in pyrometer. The pyrometer position is 6 mm behind the inductor and 19 mm below the metal specimen. Subsequently, the procedure was repeated with applied CFRP sample and the temperature T 1 at the end of the process was determined. The value combinations shown in Table 8 were tested twice. The first three results were taken from the previous experiments. The joining pressure P J was left constant at the best value of 0.6 N/m 2 of the first series of tests.  The thermal behavior of the sample bodies is shown in Fig. 24; the tensile strengths in each case as a function of the temperature are displayed in Fig. 25.
The results show an increase in temperature T 1 of titanium with an upturn in pulse width modulation PWM 2 . The temperature of the material combination T 1 had a smaller increase. This leads to a rise in the difference between the two values from 32 • C at PWM 2 = 197 ‰ to a maximum of 57 • C at PWM 2 = 217 ‰ , which corresponds to an enlargement of 80% . The heating was always in the specified process range, but the heat input in the thermoplastics increased considerably with increasing pulse width modulation PWM 2 . This should lead to an enlarged and more extensive melting of the matrix and thus also to an improved connection.
This assumption is confirmed by a considention of the tensile shear tests. The tensile shear strength shows an expansion with rising temperature T 1 and therefore with increasing temperature difference T 1 -T 1 . However, a significant increase only from 21.1 MPa at PWM 2 = 191‰ to 29.8 MPa at PWM 2 = 201 ‰ can be observed, which corresponds to a gain of 41.2% . As a result, the value remains stable in a range of 28. 4-30.9 MPa.
This behavior can be explained by considering the failure process. As Fig. 26 shows, the bonding surface increases with increased heating until the complete area is connected at PWM 2 = 205 ‰.

Summary and outlook
In the context of the work presented in this paper, a device for the creation of titanium-thermoplastic connections was developed and assembled. The functionality of the test rig could be proven and different heating experiments with titanium were conducted. The latter showed that a homogeneous warming of titanium within the required temperature range is possible with the chosen components by using a two-stage heating process. Furthermore, the influence of several parameters was examined.
Joining experiments were carried out based on these findings. They showed that the titanium has to be warmed up much more than originally anticipated to gain a sustainable connection. It was possible to determine a threshold value, above which a complete connection could be achieved.
In summary, the basic proof of functionality was provided and the influence of several parameters was shown. Further experiments should focus on the effect of the joining pressure. Moreover, the transferability from standardized tensile lap shear samples to other geometries should be investigated. The production and testing of actual industrial components and the transfer into practice should be the final objective.