Sage Journals: Discover world-class research

Abstract

Carbon fiber reinforced thermoplastics (CFRTP) have increasing use in aerospace structures due to improved process-ability and weldability. In this study, lap joints between carbon fiber reinforced low-meltpolyaryletherketone (LMPAEK) are formed by friction stir welding (FSW). This study presents novelty by applying FSW to continuous carbon fiber composites in woven laminate form. FSW disrupts the fibers in the weld zone and distributes fragments as small as several microns in length. Thermal analysis shows that the weld zones degrade at 40°C cooler temperatures than the base laminate material due to enhanced polymer mobility surrounding the disrupted carbon fibers. Though optimized joints have regions of over 9% porosity, tensile strengths of up to 73.8 MPa retains up to 50% joint efficiency of a comparable short carbon fiber reinforced composite. CFRTP also requires lower processing forces during FSW than metals, and the power consumption of 67 W during the traverse period for strength optimized welds retains energy efficient characteristics.

Keywords

Friction stir welding carbon fiber reinforced thermoplastics low-melt polyaryletherketone thermoplastic welding

Introduction

Implementing lightweight structures into aerospace vehicles is imperative to increase cargo capacity and improve efficiency. Carbon fiber reinforced plastics have increasing popularity due to improved specific strength, corrosion resistance, and damping capacity¹ that allows for a 20%- 40% weight reduction compared to conventional metallic materials.² Carbon fiber reinforced thermoplastics (CFRTP) have growing attention in aerospace industries due to non-definite shelf life, rapid cure times, high tensile strength, high fracture toughness, and good chemical resitance.^3,4 Unlike thermosetting polymer matrices, the molecular arrangement in thermoplastics allows them to re-melted, re-processed, and ultimately, welded. Aerospace structures regularly use polymer matrices from the family of polyaryletherketones (PAEK) such as PEEK and PEKK due to their attractive mechanical properties, chemical resistance, and glass transition temperatures suitable for aerospace applications. Most PAEK’s have a ratio between the melting and glass transition temperature of 1.5. Their high melting temperatures make them difficult to process and weld. Recently, low-melt polyaryletherketone (LMPAEK) was developed with a reduced ratio between the melting and glass transition temperature of 1.35.⁵ LMPAEK is easier to process and weld than PEEK with little compromise to mechanical properties and rigidity at high temperatures. The Clean Sky 2 Multifunctional Fuselage Deomstrator⁶ and PROCOMP’s upper stage LH² propellant tank⁷ are examples where LMPAEK serves as the matrix in welded composite structures.

Recently, extensive research has explored joining various fiber reinforced thermoplastics using friction stir welding (FSW). FSW utilizes a rotating tool plunged into the joining materials to heat and soften them through friction and plastic deformation.⁸ The rotating tool sweeps and mixes the joining materials in the weld zone. Because intertwined polymer chains require them to be melted during thermoplastic welding processes, FSW is a liquid state joining process for polymer composites.⁹

A typical FSW joint contains several zones including the base material, heat affected zone (HAZ), thermo-mechanically affected zone (TMAZ), and the nugget. Most polymer weld joints do not have a heat affected zone extending beyond the TMAZ due to their extremely low thermal conductivity. Without consideration of finely tuned welding parameters, inadequate heat transfer through the weld thickness can induce poor material flow and poor polymer diffusion, resulting in tunneling, porosity, root defects, and kissing bonds that hinder mechanical strength.^9,10

The tool rotation speed and traverse speed affects the size of the material flow zone in a polymeric FSW joint which has been shown to correlate with weld strength.¹¹ Though many parameters ultimately affect joint integrity, a general consensus is that the tool rotation and traverse speeds are of greatest influence.¹² Tool rotation speed is responsible for heat generation.¹³ Kordestani et al. joined both 30% wt. glass fiber and 30% wt. carbon fiber reinforced polypropylene in butt weld configuration, citing large tunneling defects when using low rotation speeds due to inadequate processing temperatures.¹⁴ However, too high of a rotation speed can overheat and eject materials from the weld zone.¹⁵ Not only can overheating the polymer oversoften it, but it can also thermally degrade the polymer in the form of oxidation, cross-linking, and chain rupture that ultimately leads to losses in mechanical properties and changes in appearance.¹⁶

The tool traverse speed is the most influential in controlling the heat transferred through the material thickness.¹¹ Slow traverse speeds maintain a molten polymer pool behind the tool and further promote interdiffusion across the joint.¹⁰ Ahmadi et al. joined 20% wt. short carbon fiber (SCF) reinforced polypropylene in lap joint configuration.¹⁷ A maximum tensile shear strength of 6.06 MPa was achieved through parameter optimization, with the traverse speed as the greatest contributor to strength. Other influential welding parameters include the tool tilt angle and plunge depth. The tool tilt angle is usually considered when utilizing shouldered tools as it influences vertical and horizontal material flows.^14,18 The plunge depth dictates the degree of pin interaction with the joining material as well as the forging pressure of the shoulder at the surface.¹⁹ Adequate axial pressure promotes material consolidation necessary to prevent formation of voids and other defects. The tool shoulder and pin geometry influence heat generation and material flow. Required spindle torque and peak temperatures have been shown to be proportional to shoulder diameter.²⁰ Additional shoulder features, such as spiral scrolls, can act as wipers to smooth the weld surface. They can serve the purpose of directing material towards the center pin and reduce expulsion of material.²¹ Researchers have excessively studied the influence of center pin geometry during polymer FSW. Of all pin geometries, threaded probes generally result in the best weld quality since they promote vertical flow of material that adequately mixes the weld zone. Threaded pins also have a large surface area to generate frictional heat and ultimately produce the least porosity of other pin geometries.¹² Kordestani et al. studied the effects of tool pin geometry on butt welded joints between similar 30% wt. SCF reinforced polypropylene. Tools with threaded tapered pins produce the maximum tensile strength, achieving up to 34% of the base composite’s strength.²²

Reinforcing fibers add complexity to polymeric FSW due to fiber disruption. Carbon fiber is an extremely abrasive material²³ and causes premature wear on standard FSW tools made of H13 tool steel.²⁴ Fiber disruption during FSW and related processes also reduces fiber size and changes their reinforcing properties.^25–27 For example, General Motors Global R&D report 50% reduced fiber length in the weld zone of friction stir scribe joined SCF 45-polyamide 6 and aluminum.

The critical fiber length L_c is the minimum length of reinforcing fiber that improves the mechanical strength over the neat polymer. L_c can be calculated by equation (1):²⁶

L_{c} = \frac{σ_{F b} d_{f}}{2 T_{b}}

(1)

σ_F _b is the fiber breakage resistance, d_f is the diameter of the fiber, and τ_b is the shear strength of the polymer. According to Meyer et al. fragmentation and reduction of glass fiber length below the critical fiber length in the weld zone limits the joint strength to that of the neat polymer.²⁶ However, fiber disruption in FSW joints can be beneficial to mechanical properties due to realignment of fibers. Czigany et al. observed random deposited fiber orientation and interlacing across the joint in glass fiber reinforced polypropylene. The unique material transfer during FSW was attributed to improved joint strength compared to conventional welding methods.²⁵

There are a limited number of studies concerning FSW joints between similar CFRTP. Additionally, the majority of studies joining CFRTP use SCF reinforcement. However, most large aerospace structures manufactured with CFRTP utilize continuous fibers, both in laminate and unidirectional tape form. Thus, a greater focus on welding technologies to join CFRTP composed of continuous carbon fibers is necessary to meet the needs of future aerospace industries. FSW lags behind other welding technologies for thermoplastic applications, though it may prove to be highly beneficial in joining thermoplastics reinforced with continuous carbon fibers. This study is the advent of such research with the formation of woven carbon fiber reinforced LMPAEK lap joints using FSW. Parameter optimization of tensile strength, morphology characterization, and thermal analysis lay the groundwork for future study and identify areas of necessary improvement.

Material and methods

Materials

Carbon fiber reinforced low-melt polyaryletherketone (LMPAEK) sheet supplied by Toray Advanced Composites served as the joining material in this study. Four layers of Toray Cetex TC1225 2/1 twill weave in [(0.90)]2S layup composed the laminate sheets with a thickness of 1.2 mm. Figure 1 illustrates the orientation of the fabric weave with respect to the 0^◦ and 90^◦ directions.

Figure 1.

Cut laminate sheet dimensions and respective fiber orientations.

Specimens were cut to dimensions of 30 mm × 70 mm using a wet tile saw with a diamond tipped blade. Due to the abrasive nature of carbon fibers and risk of delamination, the tile saw provided the highest quality cuts over other cutting methods available. The cut sheets were lapped along the 0^◦ direction with a 15.75 mm overlap. FSW was conducted using a Simulink controlled modified Kearney and Trecker Milwaukee Model K milling machine at the Vanderbilt University Welding Automation Lab (VUWAL). The tool consisted of a 25.4 mm scrolled parabolic shoulder and removable 6.35 mm diameter threaded pin set to 1.14 mm in length. The threads had a pitch of 0.79 mm (32 TPI) in the left-hand orientation, facilitating downward material flow during a clockwise rotation.²⁸

Preliminary experimentation

The material novelty of in this study required preliminary experiments to establish general welding parameters. All preliminary and further studied welding experiments in this manuscript used a weld sequence following the VUWAL’s default weld program, which utilizes a 5.08 mm/min plunge rate, a four second dwell at maximum plunge depth, tool traverse, and a four second dwell at the end of the weld followed by tool retraction. Tool pin length and overall plunge depth was established through incremental adjustments using a 1500 RPM rotation speed and 25.4 mm/min traverse speed. A 1.52 mm plunge depth was found optimal for the tool used in this study by visual inspection as shallower plunge depths did not consolidate the surface and deeper plunges cause further thinning of the welded joint. A 1.5^◦ rearward tool tilt proved adequate across a variety of material types with this tool shoulder geometry, justifying its implementation in this study. Further preliminary experiments focused on determining the effect of weld traverse direction, since one direction places the tool’s advancing side along the lapped edge, and opposite welding direction places the tool’s retreating side along the lapped edge. The preliminary study was conducted using all combinations of levels 3 and 5 for rotation speed and levels 1 and 3 for traverse speed listed in Table 1. The welding direction orienting the advancing side along the lapped edge produced greater average tensile strengths and was used for the remainder of the study.

Table 1.

CFRTP welding parameters.

Level	1	2	3	4	5
Rotation A Speed (RPM)	600	800	1000	1400	1800
Traverse B Speed (mm/min)	10.2	30.5	50.8	n/a	n/a

Parameter experimentation

The effects of tool rotation speed (A) and traverse speed (B) were tested in a 2 parameter, full-factorial design of experiments. Table 1 lists the ranges of A and B. Levels 3-5 were originally of interest for parameter A based on a similar range used by other researchers.²⁷ However, levels 1 and 2 were later added to capture the effects of lower rotation speeds. A minimum of three replicate welds for each parameter combination performed in separate blocks accounted for unintended environmental variations.

Evaluation

Mechanical tests consisted of tensile testing on an Instron load frame. Specimens were cut 14 mm wide across the lap welded sheets using a wet diamond saw with a 10 mm minimum distance from the plunge location. Two tensile specimens were cut from each weld. Emory cloth bonded to carbon fiber pads compensated for the lapped sheet offset during tensile testing. The load frame crosshead travelled at a constant 2 mm/min according to the ASTM designation D3039-08. Microstructure characterization consisted of optical microscopy and scanning electron microscopy (SEM). Weld cross sections were sanded with up to 600 grit sand paper. Polishing with 5 µm and 1 µm particle slurries concluded the surface preparation. Particle sizing and dimensional analysis was performed using ImageJ. Thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC) was used for thermal analysis on a TA Instruments Q600 SDT. Neat laminate and weld zone specimens were clipped using diagonal pliers. Analysis used heating curves sweeping from 20°C to 600°C at a rate of 10°C/min. Welding forces in the axial, lateral, and traverse directions and spindle torque were measured using a Kistler type 9123C dynamometer mounted to the machine spindle. Figure 2 references the welding directions with respect to the weld setup.

Figure 2.

Schematic of welding setup. The dynamometer measures forces in the axial (Z), lateral (Y), and traverse (X) directions in addition to the torque around the rotation axis. The weld direction goes from right to left.

Parameter effects on mechanical properties and welding forces were analyzed using Minitab Statistical Software, where a parameter’s statistical significance was evaluated on a 95% confidence interval (p ≤ .05).

Results

Surface Integrity

Figures 3(a), 3(b), and 3(c) display the surface appearance of multiple welds with varying parameters. Generally, FSW created joints with smooth and consistent surfaces for most parameter combinations. Severe surface delamination was characteristic of 600 RPM and 800 RPM welds and welds with 50.8 mm/min traverse speeds. The delamination results from cold welding parameters that fail to provide the thermal input necessary to promote polymer diffusion. Welds with the parameter combination of 1000 RPM and 30.5 mm/min had the most consistent and highest quality top surface appearance. Excessive flash occurred during 1800 RPM welds due to over softening of the material. This material expulsion reduced shoulder contact with the top surface and formed voids and inconsistent surface appearance from the lack of forging pressure. Figure 3(d) shows the underside of the welded sheets. The weld zone does not penetrate through the entire thickness of both sheets, leaving a completely unaltered surface appearance under the weld zone. Figure 3(e) shows an edge on view of the lap welded sheets.

Figure 3.

Typical surface appearance of selected welds. (a) 800 RPM (b) 1000 RPM (c) 1800 RPM (d) Bottom view (e) End view.

Microstructure

FSW influences non-homogeneous material morphology throughout the width and thickness of the weld as a result of mechanical material interaction with the tool and variable processing temperatures. Figure 4 reveals the cross section of a welded specimen with 1000 RPM and 30.5 mm/min parameters. All weld parameter combinations present similar morphological characteristics as those presented in Figure 4 from a wide view, though further discussion highlights parameter influence at microscopic scales.

Figure 4.

1000 RPM, 30.5 mm/min weld specimen cross section.

The weld cross section can be characterized into five distinct zones. Zone I represents the base material (BM) outside of the thermomechanically affected zone (TMAZ). Regions II and III represent the boundary between the TMAZ and BM, where region II lies at the advancing side (AS) interface and region III lies at the retreating side (RS) interface. The TMAZ comprises regions IV and V, where the shoulder affected zone (SAZ) lies in region IV and the pin affected zone (PAZ) extends into region V. Region V primarily lies within the bottom laminate sheet.

The composite laminate undergoes heavy transformation in the TMAZ since the FSW process must disturb the fibers in order for mass material flow and mixing to occur. Material flow is complex as the polymer has a sharp temperature dependent gradient between a solidus state, a highly ductile material, and a liquid, while the continuous carbon fibers are processed into short particles. Shearing effects further reduce fiber sizes post fragmentation from the long fibers. Figures 5 and 6 show the transformation of continuous fibers to small fragments. In Figure 5, the continuous fibers have relatively uniform circularity with a 7µm diameter. In Figure 6, chopped fibers show splintered surfaces of variable sizes encapsulated in the matrix.

Figure 5.

SEM cross section of carbon fiber laminate. The fibers are oriented normal to the page.

Figure 6.

SEM micrograph of the stir zone cross section. Broken fiber ends show splintering.

Flow stresses orient the fibers to the flow field.²⁷ In region IV, fibers up to 150 µm long orient primarily parallel to to the weld cross section. The shoulder sweeps the fibers in region IV and deposits them tangent to the shoulder’s aft end, thus orienting many of the fibers across the weld. Fiber orientation in the weld is analogous to injection molding of SCF composites. It has been shown that SCF’s located near part/mold edges tend to orient parallel to mold walls in the flow direction, resulting in the part’s mechanical properties favoring that of the fiber orientation.²⁹ Thus, particles in the center of the weld zone orient mostly across the joint, while particles near the edges of the weld zone favor orientation to the weld seam.

Particles in region V have further refined size and greater sphericity compared to those in region IV as depicted in Table 2. The average particle size in region V is 68% of that in region IV. Of important consideration, though, is that these cross section measurements do not account for fiber orientation and long particles angled towards the weld path direction will appear small. The maximum observed particle lengths in region V are approximately one third of those in region IV, indicating significant reduction in particle size. The particle refinement in region V results from contact with the threaded tool pin. Greater complexity in the tool pin, especially in terms of the number of edges, increases particle fragmentation.²⁵ The pin threads also promote particle mixing in the PAZ.³⁰ Thus, enhanced particle-particle interaction in region V increases fragmentation.

Table 2.

Void content and carbon fiber particle dimensions for regions IV and V.

Region	Void (%)	Sphericity	Avg. Size (µm)	Max. Fiber Length (µm)
IV	9.3	0.67	1	150
V	2.4	0.88	0.68	54

From equation (1) the critical fiber length L_c is estimated to be 81µm, where σ_F _b = 3530 MPa,³¹ d_f = 7 µm (measured from the fiber diameter in Figure 5), and τ_b = 152 MPa.³² These results suggest that the longer fibers in zone IV provide greater reinforcement than zone V.

Figure 4 shows an asymmetry between the weld advancing side (AS) and retreating side (RS). In this instance, the RS presents a hooking effect around fiber clusters along region III, resulting in region III having approximately a 20% longer interface from weld center to the weld surface than region II. The weld asymmetry results from greater stresses on the AS than the RS, contributing to higher temperatures.^10,33 The higher temperatures on the AS in turn lead to greater polymer mobility¹¹ and allows for improved sweeping of carbon fibers along the weld zone interface.

The weld shown in Figure 4 has the greatest ultimate strength of all tested parameter combinations, which is further discussed in section Mechanical Properties. However, microscopic images reveal substantial defects and voids in the polymer matrix that vary in size and prevalence depending on the location in the weld. Voids in the matrix occur from poor matrix mobility due to insufficient heating²⁷ and uneven cooling rates.³⁴ Figure 7 shows a large void that occurs along region II. A fiber cluster frozen in the solid base material has partially exposed ends in the SAZ. Material flow in the SAZ induces tugging on the exposed fibers, but low temperatures prohibit the polymer and chopped fibers from replacing the displaced fiber cluster.

Figure 7.

Large void visible along the AS from fiber tugging.

Large voids are especially common along region III which is characteristic of the RS in polymer FSW. These voids develop due to lower processing temperatures and lack of swept material forging around the back of the tool compared to the AS.⁹ Tool rotation and traverse speeds have pronounced contribution to the prevalence of voids along region III. Figure 8 compares region III defects between two sets of parameters. Figure 8(a) demonstrates a cold weld fabricated using the lowest tested rotation speed and fastest traverse speed. It is well understood that the rotation speed affects heat input, while the traverse speed dictates the thickness of heat diffusion.^11,35 Thus, too fast of a traverse speed outpaces the thermal conductivity of the composite, resulting in enhanced void formation. Figure 8(b) shows the same region fabricated from a faster rotation speed and slower traverse speed, allowing for more bulk melting and consequently a reduction in void size.¹¹ The void locations shown in Figure 8 correspond to the fracture location described in section Fracture Surface, indicating these defects are the main source of joint failure initiation.

Figure 8.

Fracture initiation locations. Scale bar applies for both sub figures. (a) 600 RPM, 50.8 mm/min weld (b) 1000 RPM, 30.5 mm/min weld.

Microvoid content typically increases with material depth during polymer FSW with shouldered tools due to the inverse relationship between processing temperatures and weld depth that promotes less polymer mobility.^11,35 However, Figure 9 shows a conflicting trend compared to those typically observed in polymer FSW. White highlights improve the void visibility for the reader. Region IV (Figure 9(a)) has a 9.3% void content, while region V (Figure 9(b)) has a 2.4% void content. The discrepancy in void distribution compared to conventional polymer FSW is the advent of variable fiber size fabricated by in-situ processing. Reduction in fiber size increases particle sphericity. Typically, processing of particles with greater sphericity results in a lower fraction of voids in a composite since the flowing polymer can more easily displace voids and encapsulate particles.³⁶ Thus, higher average particle sphericity in region V contributes to the lower void fraction.

Figure 9.

a) Region IV (void content 9% b) Region V (void content 2%) (c) Bottom of stir zone. Void regions are artificially colored with white overlays to highlight them. Scale bar applies for all three subfigures.

Figure 9(c) shows a trend common to conventional polymer FSW and polymer composite FSW with a lack of material consolidation at the bottom of the PAZ. This region has the coolest processing temperatures, limiting bulk material melting between the PAZ and ultimately limiting interdiffusion across the interface.³⁵

Mechanical properties

Fracture Surface All specimens failed under stress via tensile fracture along the RS, shown in Figure 10. This region is essentially an SCF composite bridging the continuous fiber laminate and the cross-sheet boundary region. Since the disruption of fibers inherently weakens the laminate, this region is the weakest on a lapped carbon fiber FSW joint non-exclusive to the advancing or retreating side configuration. Therefore, the joint tensile strength with the sheet thickness used in this study depends on the integrity of the weld in the top sheet rather than the shear strength between the lapped sheets.

Figure 10.

a) Top view of fractured specimen. (b) Cross section of fractured specimen.

Tensile Load Table 3 summarizes the results from mechanical testing. Run refers to a specific combination between parameters A (rotation speed) and B (traverse speed). UTS and σ report the effective average ultimate tensile strength and standard deviation from each set of parameters based on the base material sheet thickness of 1.2 mm. FSW thins the materials in the weld zone when implementing a rotating shoulder, ultimately reducing the load bearing cross section. Therefore, UTS (adj.) and σ (adj.) consider reduced material thickness at the fracture location to better represent the weld zone’s strength. To make thickness adjustments, the material thickness was measured in unstressed weld specimens. The measurement location was mapped from the fracture location in each weld’s respective tensile specimens.

Table 3.

Average UTS and weld thickness adjusted UTS for each welding parameter combination along with their respective standard deviations σ. Units for for UTS and σ are MPa.

Run	A (RPM)	B (mm/min)	UTS (MPa)	σ	UTS (adj.) (MPa)	σ (adj.)
1	600	10.2	33.0	2.0	41.8	3.1
2	600	30.5	40.1	0.1	45.5	0.1
3	600	50.8	34.7	1.8	41.4	4.3
4	800	10.2	33.4	8.8	39.9	10.1
5	800	30.5	41.5	6.4	48.9	9.3
6	800	50.8	45.6	9.3	52.8	11.3
7	1000	10.2	46.5	4.3	57.6	6.0
8	1000	30.5	56.0	10.4	73.8	10.3
9	1000	50.8	39.8	7.8	49.0	13.2
10	1400	10.2	35.6	3.4	45.0	8.3
11	1400	30.5	52.4	16.2	56.7	17.3
12	1400	50.8	39.2	6.0	46.0	7.4
13	1800	10.2	36.5	7.4	39.7	5.8
14	1800	30.5	37.4	6.0	39.0	6.2
15	1800	50.8	39.2	3.7	46.0	1.8

Parameters A and B have quadratic effects on the ultimate tensile strength, making the necessity of testing a minimum of three parameter levels apparent. Figure 11 plots the average effects of each parameter on UTS and Figure 12 does similarly for UTS adjusted.

Figure 11.

Fitted mean effects on the effective ultimate tensile strength from (a) rotation speed and (b) traverse speed.

Figure 12.

Fitted mean effects on the thickness adjusted ultimate tensile strength from (a) rotation speed and (b) traverse speed.

The three originally tested tool rotation speeds did not produce quadratic effects on the UTS as hypothesized. 600 RPM and 800 RPM runs added curvature to the mean effects on UTS, placing 1000 RPM at the maxima. These results differ from those in a similar study,²⁷ but the rotating shoulder in this work generates more heat at lower RPM’s.

Weld parameters of 1000 RPM and 30.5 mm/min consistently formed the strongest joints with 56.0 MPa average effective UTS and 73.8 MPa average thickness adjusted UTS. Tool rotation and traverse speeds often produce the greatest contribution to weld strength in polymer FSW¹⁵ due to their influence on thermal input, mechanical interaction, and and material consolidation. 600 RPM and 800 RPM rotation speeds lacked sufficient thermal input from friction and plastic deformation to promote polymer diffusion across zones II and III. 1400 RPM and 1800 RPM rotation speeds have greatly accelerated work piece velocities that promote excessive void formation and material expulsion. A 1000 RPM rotation speed provided sufficient thermal input without overly agitating the weld pool. Likewise, the traverse speeds influence curved behavior on the UTS. A low pressure zone follows the tool during FSW, which can limit polymer adhesion in the weld zone with fast traverse speeds.³⁷ Slow traverse speeds allows for more contact time between the molten polymer pool and joining surfaces. However, traverse speeds also influence the amount of thermomechanical stress, and it has been shown that a faster traverse speed can use the enhanced thermomechanical stress to further distribute particles in composites.³⁸ Therefore, it appears that the 30.5 mm/min traverse speed balances the effects of thermomechanical stress induced particle distribution and polymer weld pool contact time.

The tensile data presented considerable variation between replicate welds and between replicate tensile specimens within the same weld. Runs 8 and 11 had the greatest non-adjusted variation with standard deviations of 10.4 MPa and 16.2 MPa, respectively. Runs 8 and 11 also had the two highest non-adjusted average tensile strengths at 56.0 MPa and 52.4 MPa but had tensile specimens with strengths of up to 68.0 MPa and 80.2 MPa, respectively. Adjusted UTS values contributed to additional variation within parameter sets since the small tensile cross sections have high sensitivity to calculations. There was no correlation to the tensile specimen strength with respect to its location along a weld. In other words, tensile specimens cut closer to the initial weld plunge did not have consistently higher tensile strengths than specimens cut further from the initial plunge. CFRTP’s inherently present variations in mechanical performance due to their non-homogeneous structure. Mentioned in section Microscructure, the woven structure greatly affects the composite morphology in the weld zone and contributes to highly localized mechanical characteristics. Since weld specimens were cut from a larger laminate sheet, the fiber sequence at the edge of each cut specimen is unique and affects the alignment of fiber clusters for each weld. Runs with large σ likely have instances where the fiber clusters local to the weld promote exemplary strength. 600 RPM and 1800 RPM welds have the least variability, though they have fewer replicate runs since it was clear early in the experimentation that the optimal welding parameters laid somewhere in between them.

Thermal properties

Thermogravimetric Analysis Figure 13 shows thermogravi-metric analysis (TGA) curves for the base carbon fiber laminate and the weld zones of 1000 RPM, 30.5 mm/min specimens. The base material runs are denoted as B1, B2, and B3. The welded material runs are denoted as W1, W2, and W3. T₉₅ refers to the temperature with 5% weight loss. Table 4 reports the T₉₅ values for each TGA run. The base material has an average T₉₅ of 583.3°C while welded specimens have an average T₉₅ of 543.4°C.

Figure 13.

TGA curves for (a) base carbon fiber laminate and (b) 1000 RPM, 30.5 mm/min weld zones.

Table 4.

^T95 values for base material and weld zones.

Base Specimen	T₉₅ (^◦C)	Weld Specimen	T₉₅ (^◦C)
B1	579.5	W1	526.6
B2	583.8	W2	551.2
B3	586.5	W3	552.3
Average	583.3	Average	543.4

The lower thermal stability in welded specimens is a result of the short fiber length in the weld zone. The addition of carbon fiber into a polymer matrix increases the overall absorption capacity of the composite, and longer fibers allow the composite to withstand greater temperatures since they absorb more heat.³⁹ Highly localized temperatures around the tool/composite interface may also contribute to slight reduction in thermal stability, but the low thermal conductivity further limits widespread thermal degradation throughout the weld. Confirmation of decreased thermal stability due to overheating of the polymer requires the capability to accurately measure the temperature at the interface between the tool and polymer during welding.

Differential Scanning Calorimetry Figure 14 shows the differential scanning calorimetry (DSC) melting endotherms obtained for specimens while undergoing TGA, as the TA Instruments Q600 SDT performs these measurements simultaneously. Table 5 summarizes the DSC results. Toray reports the melting temperature for the base composite to be 305°C.³² The onset melt temperatures (T_onset) for both the base material and weld zones are in good agreement with the manufacturer’s reported melting temperature. The average peak melting temperature T_m for the base laminate material is 312.5°C, reported in Table 5. Specimens cut from weld zones exhibit a broader melting curve with two inflection points indicating two peaks. The double-melting behavior results from reorganization of the crystal structure as the temperature increases.⁴⁰ The chopped fibers in the weld zone improve chain mobility and allow for this double-melting behavior.

Figure 14.

DSC curves for (a) base carbon fiber laminate and (b) weld zones of 1000 RPM, 30.5 mm/min specimens. Endothermic curves point upwards.

Table 5.

Summary of DSC results. B1, B2, B3 refer to base specimens. W1, W2, W3 refer to specimens from 1000 RPM, 30.5 mm/min welds.

Base Specimen	^Tonset (^◦C)	T_m (^◦C)	Weld Specimen	^Tonset (^◦C)	T_m1 (^◦C)	T_m2 (^◦C)
B1	303.3	312.6	W1	302.6	310.6	315.9
B2	303.3	312.9	W2	302.6	309.9	316.3
B3	305.6	312.0	W3	302.5	310.6	317.0
Average	304.1	312.5	Average	302.6	310.4	316.4

Welder Inputs

Welding Forces The rotation and traverse speeds have statistically significant contribution to the average axial and lateral welding forces during the weld traversing period, with a greater contribution from the rotation speed in both cases. Only the rotation speed has statistically significant contribution to traverse forces, and there does not appear to be interaction effects between the two parameters. Table 6 displays the average axial, lateral, and traverse forces measured by the dynamometer during weld traverse periods. Generally, faster tool rotation speeds reduce all forces due to its improved heat generation. Figure 15 shows the average axial force response for each respective parameter, illustrating rotation speed’s greater contribution from the steeper slope. The axial forces range from approximately 900 N-4700 N during the weld traversing period, whereas the traverse and lateral forces are significantly less by an order of magnitude. Figure 16 shows the effect of tool rotation speed on traverse force. The rotation speeds reduce in magnitude between 600 RPM and 1000 RPM due to improved material softening on the leading edge of the tool. At 1400 RPM and 1800 RPM, the traverse forces measured at the spindle slightly dominate towards the welding direction. The positive measured traverse forces are attributed to the resultant component of axial force induced by the rearward tool tilt. For welds producing the optimal tensile strengths with 1000 RPM, 30.5 mm/min parameters, the required forces during the traversing period for the tool used in this study are approximately 2200 N in the axial direction and 190 N in planar directions. Thus, the tool forces during FSW of laminate sheets are on the order of polymer FSW in comparison to metal FSW.^41–43

Table 6.

Average welding forces during traverse period for each parameter combination. A and B refer to rotation speed and traverse speed, respectively. All forces are reported in Newtons (N).

A	B	Axial	Lateral	Traverse
600	10.2	3543	233	−207
600	30.5	4345	377	−330
600	50.8	4742	458	−351
800	10.2	2348	220	−82
800	30.5	2490	310	−139
800	50.8	2808	387	−182
1000	10.2	1844	154	−32
1000	30.5	2222	186	−42
1000	50.8	2612	247	−20
1400	10.2	1344	51	81
1400	30.5	1290	152	117
1400	50.8	1886	83	83
1800	10.2	1449	7	81
1800	30.5	928	−9	73
1800	50.8	1667	21	23

Figure 15.

Fitted mean effects for tool rotation speed (a) and traverse speed (b) on axial force.

Figure 16.

Fitted mean effects for tool rotation speed on traverse force.

Power Consumption Equation (2) calculates the power consumption by the tool during welding:

P = ω \times τ

(2)

where ω is the tool spindle angular velocity and τ is the spindle torque measured by the dynamometer. Calculating the power from the spindle torque eliminates mechanical losses from the welder and gives a pure power input to the weld piece. Strong correlations were not present between the welding parameter values and spindle torque. It is hypothesized that small carbon fiber particles act as a lubricant under the tool shoulder and limit spindle torque. Any correlations in torque between parameter inputs are within the baseline signal noise. The average power across all parameter combinations is 98.2 W during the traversing period. For the 1000 RPM, 30.5 mm/min welds, the average processing power was 67.4 W, making FSW an energy efficient method to join composites. Of course, the actual value is higher when considering the total power to run the machine from on board control systems and mechanical losses.

Discussion

Until now, extensive study pertaining to FSW of CFRTP laminate sheets have yet to be explored. The feasibility and applicability of this process over other joining methods depends on joint strength, processing time, joint configuration, and tooling requirements. Processing of CFRTP inherently has challenges due to its non-homogenous structure. FSW of carbon fiber laminate sheets has complexity due to the disruption of carbon fibers in the weld zone and poor through-sheet thermal conductivity. This study shows promising results with instances of high strength, but also clear indications of improvements needing attention.

Some tool wear was evident throughout the course of the study, shown in Figure 17. Tool wear influence likely has little significance on studied parameter effects contributing to weld strength in this study, as the order of blocked welding replicates were randomized. However, tools in large volume production of CFRTP FSW joints should be constructed of materials with higher abrasion resilience such as A11 tool steel.²⁴

Figure 17.

Worn tool shoulder and pin from carbon fiber abrasion.

It is imperative to recognize the mode of fracture in this study. Carbon fiber laminates obtain their high specific strength from continuous carbon fibers, and any disruption or discontinuity along a fiber length requires the polymer matrix to distribute loads to other fibers, inherently providing weaker mechanical properties over continuous fibers. Thus, even welding methods that do not disrupt the fibers in the adjoining sheets are limited to the shear strength of the polymer matrix if the fibers do not cross he joint interface. Reported UTS values for the base laminate TC1225 woven sheet are 805 MPa in the 0^◦ direction and 739 MPa in the 90^◦ direction.³² These strength values are vastly superior to the 73.8 MPa UTS observed in this study’s FSW joints. However, discontinuity and chopped carbon fiber in the joint requires joint comparison to the UTS of a polymer reinforced with SCF’s. Existence of studies reporting the mechanical properties of LMPAEK reinforced with SCF are unknown to the authors at this time. However, LMPAEK and PEEK have similar mechanical strength, shown in Table 7. Tensile strengths for injection molded SCF/PEEK composites range between 152.5 MPa and 211.9 MPa, though the higher strengths result from alignment of fibers in the tensile direction at the edge of the mold.²⁹ Thus, the optimal FSW parameters in this study produce joints with approximately 50% joint efficiency in relation to a comparable SCF composite that has variable fiber orientation. The low joint strength in the reinforced composite weld compared to the neat polymer results from sub-critical fiber lengths²⁶ and favored fiber orientation to the weld zone walls.²⁵ Ultimately, matrix porosity in the weld zone likely has the greatest contribution to detriments in weld performance. Future study to further optimize joint strength in friction stir welded CFRTP laminates requires consideration of avenues for both higher processing temperatures and retainment of fiber length. Auxilliary heating such as heated tools,⁴⁴ hot shoes at the weld surface,³⁸ and heat sources below the welding materials³⁵ are likely to improve processing temperatures and reduce porostiy. Optimizing fiber length may be accomplished with studies in the tool geometry, though balancing fiber size with trade-offs in matrix porosity may be especially challenging unless significantly improved processing temperatures counteract low mobility around longer fibers.

Table 7.

Comparison of PAEK tensile properties.

Reinforcement Type	UTS	Reference
Neat LMPAEK	95 MPa	45
Neat PEEK	97.2 MPa	45
FSW LMPAEK Weld Zone	73.8 MPa	Current Study
SCF PEEK	152.5-211.9 MPa	29
LMPAEK Laminate 42% RC, 0^◦	805 MPa	32
LMPAEK Laminate 42% RC,90^◦	739 MPa	32

Though complications persist for carbon fiber laminate FSW, there are several attributes that may qualify it as an attractive process to join CFRTP laminates in the future. Since the fracture mode for thin sheets in CFRTP laminate FSW is purely tensile in the top sheet, structural strength can be further improved by increasing the thickness of the top adjoining sheet. Whether or not the load bearing capacity of friction stir welded laminate sheets scales linearly with top sheet thickness is reserved for future study. The lap shear strength between the weld zone and bottom sheet is also a mechanical feature to consider when increasing the top sheet thickness since excessive thickening of the top sheet may encroach on the lap shear load bearing capacity.

CFRTP laminate FSW has an advantage over ultrasonic welding since it does not require integration of energy directors that makes consistent continuous welds and in-situ monitoring challenging tasks to achieve.⁴⁶ Further, FSW does not use consumables such as gas in hot gas welding, and does not leave embedded wires such as those in resistive welding. A drawback to FSW in CFRTP joints is the welding speed. Continuous ultrasonic welders can join CFRTP on the order of several meters per minute,⁴⁷ while FSW traverses on the order of several centimeters per minute. However, the simplicity of FSW improves the overall process efficiency when considering material preparation and post-weld processing.⁴⁸ Current FSW equipment designed to join metal alloys used in components such as rocket fuel tanks and pressure vessels can be easily adapted to weld newer composite versions of these components since metal and polymeric FSW share the same basic tooling principles. Thus, improvements in future study of continuous CFRTP FSW will make it a viable and competitive process in the manufacture of composite structures.

Conclusions

This work uses FSW to join LMPAEK CFRTP laminate sheets in lap joint configuration. Joining two laminate carbon fiber sheets via FSW has novelty in the field and presents challenges since it requires disruption of the woven fibers in the laminate. A parameter study was conducted to determine the effects of tool rotation speed and traverse speed on weld integrity. The results indicate the following:

• Fiber disruption in the weld zone forms distinct zones with variable fiber size and orientation.

• Parameters of 1000 RPM rotation speed and 30.5 mm/min traverse speed produce tensile strengths of 73.8 MPa, which is approximately 50% the strength of a comparable short carbon fiber reinforced composite.

• Weld porosity appears to have the greatest influence on strength reduction, with failure initiating defects located along the base material/weld zone interface.

• Thermal analysis shows slightly reduced thermal stability in the weld zone at high temperatures due to greater polymer mobility for discontinuous fibers with thermal degradation occurring at approximately 40°C cooler temperatures than the base laminate material.

• Optimized welding parameters require 2200 N of axial force and 67.4 W of power during traversing periods requiring lower machine requirements than FSW of metal alloys.

Advancements in aerospace technology requires increased used of high-performance materials and solutions to join them that better harness their capabilities in space systems. Though pre-existing welding methods exist to join CFRTP laminates, FSW offers additional versatility and competitive joint strengths to other joining methods.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the NASA Tennessee Space Grant Consortium (SFP_300530).

ORCID iD

Lucas T Wilkins

Appendix

References

Ghossein

Hassen

Kim

, et al. Characterization of mechanical performance of composites fabricated using innovative carbon fiber wet laid process. Journal of Composites Science 2020; 4: 4030124.

Mohammad

Mohamed

. The Role of advanced polymer materials in aerospace. Elsevier, 2018.

Sudhin

Remanan

Ajeesh

, et al. Comparison of Properties of Carbon Fiber Reinforced Thermoplastic and Thermosetting Composites for Aerospace Applications. Mater Today Proc. 2020; 24: 453–462. DOI: 10.1016/j.matpr.2020.04.297.

Ajeesh

Bhowmik

Sivakumar

, et al. Investigation on polyetheretherketone composite for long term storage of nuclear waste. J Nucl Mater 2015; 467: 855–862.

VICTREX . A comprehensive review of the processing guidelines of VICTREX ® PEEK ™ high performance polymer. Available from: www.victrex.com.

Gardiner

. Thermoplastic composites welding advances for more sustainable airframes. CompositesWorld. 2022; 8: 50–60.

Gardiner

. CFRTP upper stage propellant tank. Compos- itesWorld 2022; 8: 42–46.

Thomas

. Friction stir butt welding. Great Britain. International Patent Application PCT/GB92, 1991. Patent Application GB9125978.8.

Simões

Rodrigues

. Material flow and thermo-mechanical conditions during Friction Stir Welding of polymers: Literature review, experimental results and empirical analysis. Mater Des 2014; 59: 344–351.

10.

Huang

Meng

Xie

, et al. Friction stir welding/processing of polymers and polymer matrix composites. Compos Appl Sci Manuf 2018; 105: 235–257. DOI: 10.1016/j.compositesa.2017.12.005.

11.

Sheikh-Ahmad

Deveci

Almaskari

, et al. Effect of process temperatures on material flow and weld quality in the friction stir welding of high density polyethylene. J Mater Res Technol. 2022;5:1692–1703.

12.

Pereira

MAR

Galva˜o

Costa

, et al. Joining of Fibre-Reinforced Thermoplastic Polymer Composites by Friction Stir Welding—A Review. Appl Sci 2022; 3: 12.

13.

Hossein

Alipour

Mostafapour

. Heat-assisted friction stir welding of polymeric nanocomposite. Sci Technol Weld Join 2020; 25: 56–65.

14.

Kordestani

Ghasemi

Arab

NBM

. An investigation of FSW process parameters effects on mechanical properties of PP composites. Mechanics and Industry 2016; 17: 8.

15.

Zafar

Awang

Khan

. Friction stir welding of polymers: An overview. In: 2nd International Conference on Mechanical, Manufacturing and Process Plant Engineering (pp. 19–36), 2017.

16.

Georg

Osswald

. Materials Science of polymers for Engineers. 2nd ed. Carl Hanser Verlag, 2003.

17.

Ahmadi

Mostafa Arab

NBM

Ghasemi

. Optimization of process parameters for friction stir lap welding of carbon fibre reinforced thermoplastic composites by Taguchi method. J Mech Sci Technol 2014; 1: 279–284.

18.

Payganeh

Arab

dadgar asl

, et al. Effects of friction stir welding process parametrs on appearance and strength of polypropylene composte welds. Int J Phys Sci 2011; 6: 4595–4601.

19.

Mendes

Loureiro

. Effect of friction stir welding parameters on morphology and strength of acrylontril butadine styrene plate welds. Mater Des 2014; 58: 457–464.

20.

Mehta

Arora

, et al. Tool geometry for friction stir welding-optimum shoulder diameter. Metall Mater Trans 2011; 42: 2716–2722.

21.

Trueba

Heredia

Rybicki

, et al. Effect of tool shoulder features on defects and tensile properties of friction stir welded aluminum 6061-T6. J Mater Process Technol 2015; 219: 271–277. Available from: https://www.sciencedirect.com/.science/article/pii/S0924013614005238

22.

Kordestani

Ashenai Ghasemi

Arab

NBM

. Effect of pin geometry on the mechanical strength of friction-stir-welded polypropylene composite plates. Mech Compos Mater. 2017;9:525–532.

23.

Khairusshima

MKN

Sharifah

ISS

. Study on Tool Wear during Milling CFRP under Dry and Chilled Air Machining. Procedia Eng. 2017;184:506–517. DOI: 10.1016/j.proeng.2017.04.121.

24.

Carlson

Ollett

Kleinbaum

. Final Technical report- friction stir scribe joining of carbon fiber reinforced polymer (CFRP) to aluminum (General Motors). 2016.

25.

Czigany

Kiss

. Friction Stir Welding of Fiber Reinforced Polymer Composites. In Proceedings of the ICCM International Conferences on Composite Materials, 2011.

26.

Meyer

Jaeger

Wunderling

, et al. Friction stir welding of glass fiber-reinforced polyamide 6: Analysis of the tensile strength and fiber length distribution of friction stir welded PA6-GF30, vol. 480. Institute of Physics Publishing, 2019.

27.

Huang

Meng

Xie

, et al. Joining of carbon fiber reinforced thermoplastic and metal via friction stir welding with co-controlling shape and performance. Compos Appl Sci Manuf 2018; 112: 328–336. DOI: 10.1016/j.compositesa.2018.06.027.

28.

Panneerselvam

Lenin

. Joining of Nylon 6 plate by friction stir welding process using threaded pin profile. Mater Des 2014; 53: 302–307. Available from: DOI: 10.1016/j.matdes.2013.07.017.

29.

Chang

Long

, et al. Effects of temperature and fiber orientation on the tensile behavior of short carbon fiber reinforced PEEK composites. Polym Compos 2021; 42: 597–607.

30.

Sun

. Influence of tool thread pitch on material flow and thermal process in friction stir welding. J Mater Process Technol 2020; 2020: 275.

31.

T300-Technical-Data-Sheet-1.pdf. Toray Advanced Compos- Ites; 2018.

32.

Toray Cetex TC1225 LMPAEK Product data sheet. Toray Advanced Composites; 2020.

33.

Eslami

Tavares

Moreira

PMGP

. Friction stir welding tooling for polymers: review and prospects. Int J Adv Des Manuf Technol. 2017; 89: 1677–1690. DOI: 10.1007/s00170-016-9205-0.

34.

Hamza

Jalal

. A review on manufacturing the polymer composites by friction stir processing. Eur Polym J 2022; 9: 178.

35.

Rehman

Sheikh-Ahmad

Deveci

. Effect of preheating on joint quality in the friction stir welding of bimodal high density polyethylene. Int J Adv Des Manuf Technol. 2021; 117: 455–468. DOI: 10.1007/s00170-021-07740-w.

36.

Kalman

. Effect of particle shape on void fraction. Powder Technol 2022; 7: 407.

37.

Derazkola

Simchi

. Processing and characterizations of polycarbonate/alumina nanocomposites by additive powder fed friction stir processing. Thin-Walled Struct 2020; 157: 107086. DOI: 10.1016/j.tws.2020.107086.

38.

Barmouz

Seyfi

Givi

MKB

, et al. A novel approach for producing polymer nanocomposites by in-situ dispersion of clay particles via friction stir processing. Materials Science and Engineering: A. 2011; 528: 3003–3006. DOI: 10.1016/j.msea.2010.12.083.

39.

Rezaei

Yunus

Ibrahim

. Effect of fiber length on thermomechanical properties of short carbon fiber reinforced polypropylene composites. Mater Des 2009; 30: 260–263.

40.

Youngchul

Roger

Lin

. On the double-Melting Behavior of Poly(ether ether ketone). Macromolecules 1989; 22: 1756–1760.

41.

Mendes

Neto

Simão

, et al. A novel friction stir welding robotic platform: welding polymeric materials. Int J Adv Des Manuf Technol 2016; 85: 37–46. DOI: 10.1007/s00170-014-6024-z.

42.

Panneerselvam

Lenin

. Investigation on effect of tool forces and joint defects during FSW of polypropylene plate. Procedia Eng. 2012; 38: 3927–3940.

43.

Trimble

Monaghan

O’Donnell

. Force generation during friction stir welding of AA2024-T3. CIRP Annals 2012; 61: 9–12.

44.

Vijendra

Sharma

. Induction heated tool assisted friction-stir welding (i-FSW): A novel hybrid process for joining of thermoplastics. J Manuf Process 2015; 20: 234–244.

45.

Hron

Kadlec

Ru˚zˇek

. Effect of the test procedure and thermoplastic composite resin type on the curved beam strength. Materials 2021; 14: 1–13.

46.

Kemmish

. Update on the technology and applications of polyaryletherketones. iSmithers, a Smithers Group Co., 2010.

47.

Jongbloed

Teuwen

Palardy

, et al. Continuous ultrasonic welding of thermoplastic composites: Enhancing the weld uniformity by changing the energy director. J Compos Mater 2020; 54: 2023–2035.

48.

Strand

. Joining plastics - Can Friction Stir Welding compete? In: Electrical Insulation Conference and Electrical Manufacturing and Coil Winding Conference and Exhibition, 2003, pp. 321–326.

Friction stir welding of lapped low-melt polyaryletherketone carbon fiber reinforced thermoplastic laminate

Abstract

Keywords

Introduction

Material and methods

Materials

Preliminary experimentation

Parameter experimentation

Evaluation

Results

Surface Integrity

Microstructure

Mechanical properties

Thermal properties

Welder Inputs

Discussion

Conclusions

Footnotes

Declaration of conflicting interests

Funding

ORCID iD

Appendix

References