MECHANICAL STUDY OF DIRECT LONG FIBER THERMOPLASTIC CARBON/POLYAMIDE 6 AND ITS RELATIONS TO PROCESSING PARAMETERS Kyle Rohan , T.J. McDonough Zoltek Corporation Vanja Ugresic, Eva Potyra, Frank Henning Fraunhofer Project Centre for Composites Research University of Western Ontario

Abstract Direct long fiber thermoplastic (DLFT) manufacturing using glass fibers has been in use for many years and provides a stable platform for a variety of automotive parts. With the continued goal of decreasing weight and increasing performance, switching to carbon-DLFT offers a viable alternative to the current technology. However, carbon as a reinforcement fiber is not currently commercialized due to a lack of optimization of processing, fiber length retention, and cost. As automotive technology advances and carbon fiber prices drop, carbon-DLFT will provide a route to increased performance. A joint effort between Zoltek Corporation and Fraunhofer Project Centre at Western studied processing responses of compression molded DLFT and mechanically quantified the effect of multiple manufacturing parameters for use in automotive semi-structural applications. Mechanical performance and physiochemical qualities of carbonpolyamide 6 (carbon/PA6) DLFT were investigated- particularly as they relate to fiber loading, plastificate placement, screw speed, and other manufacturing processing parameters. This research aims to provide a better understanding of carbon/PA6 DLFT by striving towards dropin replacement of glass fibers for increased performance and weight reduction utilizing currently available equipment with minor modifications.

Introduction and Background Materials Selection Utilization of polymer-fiber composites is an innovative strategy for weight reduction of automotive parts. Part weight reduction in turn decreases the fuel consumption and CO2 emissions required to comply with environmental regulations. Through different combinations of polymers and fibers, useful material combinations can be developed for a wide range of automotive applications (1) (2). Polypropylene is a common matrix material because of its good processability; resistance to organic solvents; and hydrophobic quality. However, it has relatively lower mechanical properties and service temperatures when compared to engineered thermoplastics (3). Substituting PP for an engineered thermoplastic such as a polyamide tends to improve the common desired design properties of the composite, strength, impact, and modulus. However, concerns with processing PA are the higher processing temperatures; higher viscosities; oxidation-sensitivity; higher water absorption; and increased fiber breakup (4). Carbon fibers are gaining momentum due to their good properties (particularly their high strength and stiffness) and low density (5). In order for carbon fibers to be readily utilized in automotive application, low cost and repeatable manufacturing technologies with high volume capability need to be developed and optimized (6). Furthermore, the design know-how needs to be strengthened where hybrid material and selective reinforcement concepts are utilized (5).

Page 1

Processing Technology Long fiber thermoplastics (LFT) are a type of composite material wherein thermoplastic polymers are mixed with fiber reinforcements. LFT material is well established in the automotive industry, mostly in a non- or semi-structural applications based on polypropylene (PP) resins and glass fibers for reinforcement. Recently the field of application broadened from covers and parts in unexposed areas, to instrumental panel supports and door modules, car seat shells, underbody panels and bumper supports (7). Processing technologies for manufacturing LFT parts strongly influence the final part properties (8) (9). LFT parts are typically manufactured using one of three different processing methods as seen in Figure 1. The primary difference between these methods is the initial materials used. Glass mat thermoplastics use randomly comingled glass-thermoplastic sheets. Indirect LFT processing uses semi-finished, pellet form materials (LFT-G) processed with an injection molding machine (6). Relatively new, direct LFT technology involves combining the raw materials (fiber and matrix) immediately before entering the molder and avoids the production of an intermediate (7). This gives the manufacturer the ability to enhance the properties of the resin as required for the application by adding fillers, fire retardants and additives ,while also significantly reducing material costs and only establishing a single heat history of the matrix (10) (2) (7). This direct process can be separated into injection molding compounding process (LFTD-IMC) and extruder compression molding process (LFT-D-ECM). In LFT-D-IMC an extruder, which melts the polymer and mixes it with the fiber, is attached to an injection molding machine. In contrast, LFT-D-ECM typically uses two extruders system for the polymer melting and fiber dosing coupled with hydraulic press for compression molding.

Glass Mat Thermoplastics

Long Fiber Thermoplastics (LFT)

Indirect LFT Methods

Injection Molding (LFT-G)

Direct LFT Methods

Injection Molding Compounding (LFT-D-IMC)

Extruder Compression Molding (LFT-D-ECM)

Figure 1: LFT manufacturing technologies There are two primary advantages to LFT-D-ECM over the other two technologies. First is that the compounding and fiber mixing steps are separated and therefore both extruders can be individually optimized for their specific functions. Second, the material is not subjected to the same high levels of stress as in injection molding, which helps to maintain fiber length. This is critical because the mechanical properties of fiber reinforced thermoplastics are strongly dependent on the fiber length (10) (11). Therefore, LFT-D-ECM process gives a maximum degree of freedom in terms of optimizing material choice, polymer modification, and process parameters, which is important for automotive applications (7) (6).

Page 2

Experimental Procedures Manufacturing Materials The engineered polymer used in this study was polyamide 6 (PA6), Ultramid® 8202 HS from BASF. It is a heat stabilized polymer has a density of 1.13 g/cm³ and a melting temperature of 220°C. The carbon fibers used were Zoltek’s PANEX35® (50K) Continuous Tow center-pull spools with PA compatible -11569 sizing. The carbon fibers have a density of 1.81 g/cm³ and a fiber diameter of 7.2 µm. The carbon fiber weight percentage in the composite was varied from 30 – 45% in 5% increments.

Machine Setup Long fiber reinforced panels were manufactured using a production size, Dieffenbacher, LFT-D-ECM line at the Fraunhofer Project Centre at Western. A schematic of the LFT-D-ECM process can be seen in Figure 2. The polymer was dosed via a granule dosing system from Motan into the first co-rotating twin-screw extruder, a Leistritz ZSE-60HP – 28D (2).

Figure 2: LFT-D-ECM process configuration The molten polymer was transferred from the ZSE via a film die in the open barrel section of the second extruder (3) along with continuous carbon fiber rovings (3a). The second extruder, ZSG Leistritz ZSG-75 P – 17D, is a 17 L/D, 75 mm diameter co-rotating twin-screw extruder. The carbon fibers were continuously pulled in by rotation of the extruder. In the ZSG, fiber rovings were wet out, distributed, and broken by stretch breakage. Fiber length was controlled by the screw design and screw speed. Well-mixed composite exited the ZSG through the servo die with a width of 100 mm onto the conveyer belt (4) where it was cut into 370 mm long plastificates using an automated cutting shear. The conveyor belt was heated and covered with an insulated tunnel to maintain plastificate temperature. The plastificate was then removed from the belt by an operator and transferred to the hydraulic press, a Dieffenbacher DCP-U 2500/2200 (5), where it was molded to its final shape on a 46 cm by 46 cm plaque mold with a thickness of 3mm.

Page 3

Processing conditions Processing conditions were varied through the trial for a total of 15 different conditions summarized in the Table I. The factors that were varied were the plastificate position, ZSG screw speed, fiber weight percent, and circumference setting (a measure of the feed rate). Table I: Trial overview of variable processing parameters Run Condition

FiberWt (%)

RPM

ChargePosition

Circ.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

30% 30% 30% 30% 35% 35% 35% 35% 35% 40% 40% 40% 45% 30% 30%

50 70 100 50 50 50 50 50 70 50 50 70 50 50 50

Edge Edge Edge Center Edge Edge Edge Center Edge Edge Center Edge Edge Edge Edge

0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.108

Tested Tensile Panels 1 2 2 2 2 1 2 2 2 2 2 2 1 1 2

Tested Flexural Panels 1 2 2 2 2 1 2 2 2 2 2 2 1 1 2

The constant processing conditions are summarized in Table II. Table II: Constant processing conditions for LFT-D PA6/CF Processing parameters Melt temperature [°C] Mold temperature [°C] Pressing pressure [MPa] Cooling time [s]

Set Value 270 125 33 30

To investigate the influence of the material flow and fiber orientation two plastificate positions were chosen, as shown in Figure 3. Some plastificates were placed at the edge to obtain a longer flow path and greater fiber orientation, while the other plastificates were placed at the center of the mold to increase fiber randomization.

Page 4

Figure 3: Relative plastificate placement

Mechanical Testing The test specimens were extracted from the molded plaques by abrasive waterjet and were tested in “as received/room temperature” (AR/RT) conditions by the Zoltek Corporation. A Tinius Olsen 150kU load frame was used with TestNavigator software installed.

Flexural testing The specimen geometry was selected using the recommendations of the European Alliance for Thermoplastic Composites (EATC) (12). Specimen dimensions were 80mm x 25mm and were tested in 3-pt flexure at a crosshead rate of 5mm/minute, per ISO 14125 guidelines. Span to depth ratio was set at 16 to 1. The relatively large specimen widths were used to avoid local variation due to the potential for fiber bundling and material inhomogeneity. Figure 4 shows the template used for specimen extraction. Note that the reference edge is the sprue outlet location. Plastificate location is indicated in blue for an edge charge and orange for a center charge. Each specimen was labeled, measured, and tested taking into account location and orientation. An Epsilon 50mm deflectometer was used to measure midpoint deflection of each specimen. Additionally, resin burnoff and acid digestion specimens were extracted from each panel for fiber weight content determination.

Flow

Figure 4: Flexural waterjet template A total of 26 panels with varying process parameters were tested in flexure. The number of

Page 5

coupons reserved from each panel for each type of test is shown in Table III. Table III: Number of specimens per flexural panel Specimen Type

Coupons Reserved

Flexure- Flow

16

Flexure- Crossflow

16

Acid Digestion

2

Resin Burnoff

4

Coupon IDs #-F-1 to #-T-5, #-F-11 to #-F-15, #-F-22 to #-F-27 #-F-6 to #-F-10, #-F-16 to #-F-21, #-F-28 to #-F-32 #-AF-7, #-AF-11 #-BF-1, #-BF-2, #-BF-3, #-BF-4

Failure Modes Observed failure modes of the flexural specimens were consistent with expected failure modes. The failure modes were similar to those proposed by Curtis (13).

Figure 5: Typical failure mode flexural testing

Tensile testing Tensile properties were measured in accordance with ISO 527-4, using 250mm long and 25mm wide specimens, tested at a rate of 2mm/min instead of the suggested 5mm/min. as the higher rate test specimens failed too quickly for data aquisition. The EATC guidelines were followed for the design of tensile specimens (12). Figure 6 shows the specific layout of each of the 26 tensile panels tested for strength and modulus. Note that the reference edge is the sprue outlet location. Each specimen was labeled, measured, and tested taking into account location and orientation. Additionally, resin burnoff and acid digestion specimens were extracted from each panel for fiber weight content determination. Specimens were tested using mechanical wedge grips. Modulus was measured using an Epsilon 2% clip-on extensometer with 3-point

Page 6

contact edges installed. Emory cloth was applied to each specimen at the tab area.

Flow

Figure 6: Tensile waterjet template A total of 26 panels with varying process parameters were tested in tension. The number of coupons reserved from each panel for each type of test is shown in Table IV. Table IV: Number of specimens per tensile panel Specimen Type

Coupons Reserved

Tensile- Flow

10

Tensile- Crossflow Acid Digestion Resin Burnoff

11 2 2

Coupon IDs #-T-1 to #-T-5, #-T-17 to #-T-22 #-T-6 to #-T-16 #-AT-1, #-AT-10 #-BT-1, #-BT-2

Failure Modes Specimens were initially tested at a 10mm gage width using an ASTM D638 dogbone shape, but these specimens were found to be too dependent on local bundling and anisotropy. This shape was modified to a 25mm wide flat sided specimen for all testing reported below as recommended by the EATC. The use of mechanical wedge grips may have contributed to tab failures. Data for tensile sets presented below failed in an appropriate manner according to the ISO 527-4 standard and EATC recommendations. Additionally, the specimen size may have contributed to variation as the relative length of the specimen to the molded panel length may have an effect on the homogeneity of the specimen. A typical specimen is shown in Figure 7 (top, middle). For each of these specimens it appears that local fiber bundling led to the initiation of failure within the gage section (13). Several specimens failed at the grips and were discarded from the data sets (bottom, Emory cloth removed).

Page 7

Figure 7: Acceptable failure modes top 2 specimens, unacceptable mode bottom Physiochemical Testing

Fiber weight fraction Fiber weight fraction was determined using two procedures- resin burn-off and acid digestion. During the burn-off procedure, the specimens were dried at 105°C for 3 hours in a desiccating oven and then burned off at 525°C for 35min in non-oxidative environment (14). The remaining fibers were weighed and fiber weight percent was calculated based on the total mass loss. Acid digestions were completed according to ASTM D3171 using Method I, Procedure B. The main concern of performing burnoffs with carbon fiber is oxidization of the carbon fiber (15). This was avoided by performing initial TGA (thermo-gravimetric analysis) studies into the weight loss versus time temperature and environment. The method showed no oxidization of carbon fiber under these test conditions.

Fiber Length Measurements The aspect ratio of fiber in a composite strongly influences the mechanical properties of the composite. The graph shown in Figure 8 is based upon the Halpin-Tsai equations (16) for the particular stiffnesses of the resin and fibers used in this trial. (Note: isotropic approximation 3 5 is 𝐸𝐼𝑠𝑜𝑡𝑟𝑜𝑝𝑖𝑐 = 𝐸𝐿 + 𝐸𝑇 , where EL & ET are the longitudinal and transverse moduli, 8 8 respectively). Therefore, fiber length measurements were taken in an attempt to quantify the typical fiber length. When viewing the fibers under a microscope, it was evident that there existed a broad range of fiber lengths. Therefore, a majority of samples were mechanically filtered to determine the weight percentage of the fibers greater than 2mm in length. This length was chosen because the difference in modulus in Figure 8 between the modulus at 7.2mm and at 2mm was less that 7% for all relevant fiber wt.%. Additionally, the fibers less than2mm in length from a small number of samples were then optically measured using an Olympus SZX-7 stereoscope, Paxit image capturing software, and ImageJ analysis software to count the lengths of the short fibers.

Page 8

100.00%

45

90.00%

40

80.00%

35

70.00%

30

60.00%

25

50.00%

20

40.00%

15

30.00%

10

20.00%

5

10.00%

Modulus % of max 45% E_Isotropic/E_7.2mm

E_Isotropic (GPa)

50

40% 35% 30% 25% 20% 15% 10% 5%

0.00%

0 0

1

2

3

4

5

6

7

8

Length (mm)

Figure 8: Tensile modulus of randomly oriented Panex 35 tows in PA6 resin vs length and at different fiber wt.% For the fiber filtering procedure, fibers were separated according to a method proposed by Dahl (17). Following matrix removal via burn-off, 0.9g of fibers were sampled and filtered through an ASTM no. 10 mesh (2.0mm mesh size). Care was taken not to break the fibers during sampling, mixing and filtering. After sieving, the samples were filtered through 4 micron filter paper and dried completely in a vacuum oven at 105°C. The mass of each filtered portion was taken and compared to the original sample size as a percentage of fibers less than 2.0mm. For the optical measurement method, short-fiber remnants from the previous step were placed in a water suspension and exposed to ultrasound to break up any clumped fibers. The fibers were then carefully extracted from the water using filter paper and oven-dried. Photomicrographs of the sample were used to obtain measurements of 250 randomly selected fibers. Figure 9 and Figure 10 show fibers from each method used to quantify fiber length.

Page 9

Figure 10: Micrograph of sampled fibers using the fiber count method

Figure 9: Fibers after drying, left2.0mm

Results and Discussion Manufacturing

Mass Temperature Findings During processing, the mass temperature in the ZSG increased depending on the fiber content and the screw speed Figure 11. Higher screw speeds resulted in a slight increase of mass temperature (2 °C). The increase of fiber content had a more pronounced effect on the mass temperature. With carbon fiber content of 45 wt.%, the mass temperature increased by approximately 8 °C. This was most likely due to higher fiber friction between the extruder and the abrasive fibers.

Figure 11: Mass temperature dependency

Fiber Feeding At the start of the trial, the carbon fibers were fed into the LFT-D-ECM line using the same fiber handling system as is typically used for fiberglass. However, the friction in the system produced unacceptable levels of carbon fuzz, as shown in Figure 12.

Page 10

Figure 12: Fuzz build-up in feeding section For the purposes of the rest of the trial, this fiber handling system was avoided entirely by using two rollers to guide the fibers into the ZSG with reduced friction. Refer to Figure 13 for details of the feed system used. This adjustment resulted in improved feeding of carbon fibers with reduced fuzz. There is still potential to optimize the line further by employing a typical carbon fiber tensioning system to better control the fiber feed rate.

ZSG entrance

Figure 13: Typical fiberglass feed system, left, and carbon feed system used, right

Plastificate Appearance Several different plastificates manufactured with varying fiber weight contents can be seen in Figure 14. Depending on the fiber content, the appearance of the plastificates changed. Additionally, there was increased die swelling with higher fiber content.

Page 11

Figure 14: Plastificate left to right; 30, 35, 40, and 45 wt.% carbon Part Warpage Figure 15 shows pictures of the plaques from trials 1-4 in the inverse position when compared to the actual placement inside of the mold. With increasing screw speed, the warpage appeared to increase. For edge-positioned plastificates, the warpage was greater on the bottom right side while the warpage of the center-positioned plastificates was symmetric from the middle.

Trial 1: 50 rpm, at the edge position

Trial 3: 100 rpm, at the edge position

Trial: 70 rpm, at the edge position

Trial 4: 50 rpm, center position

Figure 15: Images from plaques of trial 1-4 (30 wt.% carbon fiber)

Mechanical Testing

Data Processing Mechanical testing presented below was performed to quantify important factors for molding DLFT carbon fiber parts. These factors include fiber weight percent, screw speeds, plastificate placement and fiber feeding circumference. The location and orientation of each specimen within each process condition is investigated during the analysis to account for the observed anisotropy of the panels (18). Tested values are compared to their theoretical equivalent and PA6/Glass D-LFT (19). This data will serve as a baseline for future studies and investigations. To assist in data analysis, the specimens were broken into sets depending on their location and orientation for both tensile and flexural data. These data sets were then averaged across the relevant panels to produce the data in the following graphs. The error bars in the charts

Page 12

represent one standard deviation in either direction for each data set. Tensile panels (Figure 17) and flexural panels (Figure 16) were broken down into 4 and 6 sets respectively depending on their relative location and orientation. The first half of the sets corresponds to the flow direction and the second half to the cross-flow direction. The reference edge is the location of the outlet sprue.

Figure 16: Flexural set layout for specimen interpretation

Figure 17: Tensile set layout for specimen interpretation

Directionality & Fiber Weight Percentage The test data revealed that there was a strong dependence on the directionality of the tested coupons in both the flexural and tensile data sets. The cross-flow coupons failed consistently failed at lower loads than the flow-oriented coupons. Additionally, this difference became more pronounced as the specimens moved farther away from the initial plastificate position in the flow direction. This may be attributed to the alignment of the fibers as the flow front was increased (20). On the other hand, in the cross-flow direction there was a consistent decrease in the strength and modulus as more fibers were aligned in the flow direction. The flexural testing data, Figure 18 and Figure 20, revealed that increasing the fiber weight fraction increased the flexural strength and modulus. This trend was more evident in the flow direction than the cross-flow direction. The tensile testing data, Figure 19 and Figure 21, showed less of a pattern. In fact, strength appeared to decrease from 35% fiber weight to 40% fiber weight for several of the data sets. This may be an effect of increased fiber on fiber interaction leading to increased fiber degradation. A similar effect was observed by Thomason, namely, that a maximum peak exists for tensile specimens (21). Considering significant deviation in the results, tensile modulus appeared to increase with increasing fiber content.

Page 13

30% 35% 40% Set 1 Set 2 Set 3 Set 4 Set 5 Set 6

45%

Avg. Strength (MPa)

Avg. Strength (MPa)

600 500 400 300 200 100 0

300 250 200 150 100 50 0

30% 35% 40% Set 1

Coupon Group

35% 40% 45%

Coupon Group

Avg. Modulus (MPa)

Avg. Modulus (MPa)

Set 4

45%

Figure 19: Avg. Tensile Strength vs. Fiber Wt. %

30%

Set Set Set Set Set Set 1 2 3 4 5 6

Set 3

Coupon Group

Figure 18: Avg. Flexural Strength vs. Fiber Wt. %

50000 40000 30000 20000 10000 0

Set 2

40000 30000

30%

20000

35%

10000

40%

0 Set 1

Set 2

Set 3

Set 4

45%

Coupon Group

Figure 21: Avg. Tensile Modulus vs. Fiber Wt. %

Figure 20: Avg. Flexural Modulus vs. Fiber Wt. %

Page 14

Plastificate Positioning The distance the material flowed during pressing appeared to strongly influence the directionality and the resulting flexural strength (Figure 22). This was supported by both the center- and edge-positioned panels. The panels produced with an edge-positioned plastificate yielded increasing flexural strength in the flow direction and decreasing strength in the crossflow direction. For the panels produced with center-positioned plastificates, the strength of the flow-direction coupons was lowest at the center and increased with distance from the center of the panel. The cross-flow coupons had a higher flexural strength in the center than on the edges, suggesting the fibers became more oriented in the flow direction the further they traveled. The results for the tensile testing showed much greater scatter and made it difficult to see any trends emerging. It could be seen that there were greater fluctuations in the strength when the plastificate was placed at the edge rather than in the center.

500 400 300 200 100 0

Avg. Strength (MPa)

Avg. Strength (MPa)

It can be assumed that the increase in directionality leads to part warpage as observed during panel manufacture. Studies have been performed and models exist for prediction of the orientations of long fibers which can also predict part warpage of compression molded long fibers (22) (23) (24). For any part, plastificate placement should be strongly considered as a driving design constraint based on the part geometry. Methods for placement optimization of fiber filled compression molding solutions exist (20) (25).

30%-Edge 30%-Center 35%-Edge Set Set Set Set Set Set 1 2 3 4 5 6

35%-Center

300 250 200 150 100 50 0

35%-Center

40000 30000 20000

30%-Edge

10000

30%-Center 35%-Edge 35%-Center

Coupon Group

30000 25000 20000 15000 10000 5000 0

40%-Center

30%-Edge 30%-Center 35%-Edge 35%-Center Set 1 Set 2 Set 3 Set 4 Coupon Group

Figure 24: Avg. flexural modulus vs. plastificate position

40%-Edge

Figure 23: Avg. tensile strength vs. plastificate position Avg. Modulus (MPa)

Avg. Modulus (MPa)

35%-Edge

Coupon Group

Figure 22: Avg. flexural strength vs. plastificate position

Set Set Set Set Set Set 1 2 3 4 5 6

30%-Center

Set 1 Set 2 Set 3 Set 4

Coupon Group

0

30%-Edge

40%-Edge 40%-Center

Figure 25: Avg. tensile modulus vs. plastificate position

Page 15

Screw Speed In Figure 26 and Figure 28, the average flexural strength and modulus showed an increasing trend with increasing screw speeds for all sets. From the filtration test it did not appear that changing screw speed significantly impacted fiber length as seen in Figure 36. Based upon micrographs and observations during fiber filtering, it was evident that many of the larger fibers were not well-dispersed in the laminate, as shown in Figure 30, for all RPM settings. Fiber dispersion and breakup should be investigated further as a response to extruder RPM settings.

400 300 200

50

100

70 100

0 Set 1 Set 2 Set 3 Set 4 Set 5 Set 6

Avg. Strength (MPa)

Avg. Strength (MPa)

The tensile test data is inconclusive for the effect of screw speed within this range, and should be investigated further (26). However it appears that process stability may be increased with increasing screw speed. 300 250 200 150 100 50 0

50 70 100 Set 1

Coupon Group

50 70 100 Set 1 Set 2 Set 3 Set 4 Set 5 Set 6 Coupon Group

Avg. Modulus (MPa)

Avg. Modulus (MPa)

Set 3

Set 4

Figure 27: Avg. tensile strength vs. screw RPM at 30wt.%

Figure 26: Avg. flexural strength vs. screw RPM at 30wt.% 30000 25000 20000 15000 10000 5000 0

Set 2

Coupon Group

40000 30000 20000

50

10000

70 100

0 Set 1

Set 2

Set 3

Set 4

Coupon Group

Figure 29: Avg. tensile modulus vs. screw RPM at 30wt.%

Figure 28: Avg. flexural modulus vs. screw RPM at 30wt.%

Page 16

Figure 30: 5x Micrograph of 30wt.% coupon in flow direction showing fiber clumping Circumference

400 300 200

0.16

100

0.108

0 Set 1 Set 2 Set 3 Set 4 Set 5 Set 6

Avg. Strength (MPa)

Avg. Strength (MPa)

Circumference describes the length of one fiber roving drawn into the extruder per revolution. In order to maintain the desired screw speed with decreased circumference, a greater number of fiber rovings must be fed into the machine. This value is based on the geometry of the screw. These levels have previously been optimized by the machine provider for glass fiber only. In order to determine whether the values hold for carbon fiber, two different circumference values were selected. It was found that the reduction in circumference increases all levels of flexural strength and modulus while decreasing most tensile properties. This could also be attributed to increased fiber interactions or to the low sample size for this specific factor (only 80 total specimens).. 250 200 150 100

0.16

50

0.108

0 Set 1

Set 2

Set 3

Set 4

Figure 31: Avg. flexural strength vs. circumference

Figure 32: Avg. tensile strength vs. circumference

30000 25000 20000 15000 10000 5000 0

0.16 0.108 Set Set Set Set Set Set 1 2 3 4 5 6 Coupon Group

Avg. Modulus (MPa)

Coupon Group

Avg. Modulus (MPa)

Coupon Group

30000 25000 20000 15000 10000 5000 0

0.16 0.108 Set 1 Set 2 Set 3 Set 4 Coupon Group

Figure 34: Avg. tensile modulus vs. circumference

Figure 33: Avg. flexural modulus vs. circumference Page 17

Physiochemical Testing Fiber Weight Percentage The distribution of fiber weight percent within a panel remained relatively consistent for all weight percentages. This is in contrast to the findings of Lafranche for injection molded LFT material (26). There was some observed variation in the wt.% from panel to panel, possibly due to the variation of panel weight within each process condition. This was accounted for by normalizing the mechanical test results to the target wt.%. Average moisture content of all tested specimens was 0.472%, as determined by gravimetric analysis.

50%

45%

45%

40%

30%

40% 35% 30%

35%

35%

40%

30%

45%

25%

25%

Flow Direction

20%

20%

Figure 35: Fiber weight fraction vs. flow direction (Edge plastificate position)

30% 35% 40%

Flow Direction

Figure 36: Fiber weight fraction vs. flow direction (Center plastificate position)

Fiber Length Measurements From an initial optical analysis of the fibers, it was clear that a large variation existed in the fiber lengths. The combination of sieving and optical measurements sought to help determine the percentage of long fibers (>2mm) as well as the average length of the short fibers ( 2.0mm vs. RPM at 30 wt. % and Edge positioning

Page 18

Figure 38: Wt.% of fiber > 2.0mm vs. Fiber Wt.% at various RPM & positioning

The optical measurements of the short fibers revealed that the most common short fiber length was between 0.1-0.3mm, as shown in Figure 38. This corresponds to an aspect ratio of between 14 and 42. As shown in Table V, this aspect ratio is an accurate prediction of tensile modulus (13). It would appear that the typical short fiber length is a driving factor in the laminate performance rather than the fibers longer than 2mm. This may be due to observed bundling in the longer fibers, leading to a heterogeneous distribution of fibers, as shown in Figure 30 140 120

Counts

100 80

30%, 108circ

60

30%, 160circ 35%

40

40% 45%

20

1.9-2

1.8-1.9

1.7-1.8

1.6-1.7

1.5-1.6

1.4-1.5

1.3-1.4

1.2-1.3

1.1-1.2

1-1.1

0.9-1

0.8-0.9

0.7-0.8

0.6-0.7

0.5-0.6

0.4-0.5

0.3-0.4

0.2-0.3

0.1-0.2

0-0.1

0

Length Range (mm)

Figure 39: Optical length measurements of short fibers from 4 different samples Table V: Comparison of predicted tensile modulus to avg. test data 30%

35%

40%

45%

Lower Bound (l/d =14)

9.5

11.1

13.0

15.1

Upper Bound (l/d= 42)

14.4

17.0

20.0

23.0

Actual Avg. Tensile Modulus

14.0

15.8

15.3

20.2

Predicted Tensile Modulus (GPa)

The curves shown in Figure 38 illustrate the expected tensile modulus for different aspect ratios and weight percentages. They were derived in the same manner as Figure 8. It is evident that further optimization is required to achieve the maximum predicted stiffness for long fiber thermoplastics.

Page 19

35

Isotropic Modulus (GPa)

30 45%

25

40% 35%

20

30% 15

25% 20%

10

15% 10%

5

5%

0 1

10

100

1000

l/d ratio

Figure 40: Isotropic modulus vs fiber aspect ratios and various fiber wt.%

Comparison to Glass Fiber Comparison of data sets must always be carefully conducted, as many factors play a role in each test and it has been shown in this and other studies that there are many factors involved in processing long fiber thermoplastics. However, a comparison of current technologies to proposed technologies should be considered. In Table VI, a general comparison of Eglass/PA6,6 ,E-glass/PP and carbon/PA6 properties are compared from studies performed using similar equipment and similar data sets (19,6). It is important to note that the carbon/PA6 data values are overall averages of flexural and tensile strength from all runs and coupon orientations with a fiber weight of 35%. As such, the specific strength may be expected to increase with a better-optimized manufacturing process and more well-oriented fibers. Therefore, the actual benefits of a carbon/PA6 material system are likely greater than the values reported in the table.

Page 20

Table VI: Comparison of Strength of Carbon/PA6,6 D-LFT vs. Glass/PA6,6 D-LFT Tested Weight %

Specific Strength (σ/ρ) (Mpa/(g/cm3))

Comparison Vs. Carbon/PA6

Tensile: Carbon/PA6 35 wt.%

35

104.0

N/A

Tensile: E-Glass/PA66 30 wt.% (19)

30

71.7 (83.65)

-19.5%

Tensile: E-Glass/PP 40 wt.% (6)

40

54.0 (47.25)

-54.5%

Tensile: OEM Approved 12.5mm Eglass/PP pellets 40 wt.% (6)

40

40.0 (35)

-66.3%

Flexural: Carbon/PA6 35 wt.%

35

174.5

N/A

Flexural: E-Glass/PP 40 wt.% (6)

40

92.4 (80.9)

-53.6%

Flexural: OEM Approved 12.5mm Eglass/PP pellets 40 wt% (6)

40

63.3 (55.41)

-68.2%

Material

• • •

Items in parenthesis have been linearly normalized to 35% wt. for comparison purposes. All specimens other than the OEM pellets were manufactured using the D-LFT ILC or EMC process, tensile E-glass/PA6 was not reported by Krause (19). Densities; PP 0.946g/cm3, PA6 1.13g/cm3, PA66 1.14g/cm3, Carbon 1.81g/cm3, E-Glass 2.55 g/cm3 Table VII: Comparison of Modulus of Carbon/PA6,6 D-LFT vs. Glass/PA6,6 D-LFT Material

Tested Weight % 35 40

Specific Modulus (E/ρ) (Gpa/(g/cm3)) 11.3 4.62 (4.04)

Comparison Vs. Carbon/PA6 N/A -94.6%

Tensile: Carbon/PA6 35 wt.% Tensile: E-Glass/PP 40 wt.% (6) Tensile: OEM Approved 12.5mm E40 3.26 (2.85) -128.1% glass/PP pellets 40 wt.% (6) Flexural: Carbon/PA6 35 wt.% 35 13.0 N/A Flexural: E-Glass/PP 40 wt.% (6) 40 3.35 (2.93) -126.4% Flexural: OEM Approved 12.5mm 40 3.50 (3.06) -123.8% E-glass/PP pellets 40 wt.% (6) • Items in parenthesis have been linearly normalized to 35% wt. for comparison purposes. • All specimens other than the OEM pellets were manufactured using the D-LFT ILC or EMC process. • Densities; PP 0.946g/cm3, PA6 1.13g/cm3, PA66 1.14g/cm3, Carbon 1.81g/cm3, E-Glass 2.55 g/cm3

Page 21

Conclusion: Summary and Next Steps Processing factors of direct compounding of polyamide carbon have been studied. The results show promise for improvements and further study. Although performed as an initial study, the carbon/PA6 DLFT material system performs well against commercialized products and provides immediate improvement in mechanical properties. The maximum processable limit of carbon is 45% by weight, with this matrix and fiber configuration. Additionally carbon specific fiber guiding would reduce fuzz generation and part homogeneity. Certain trends in the processing conditions may be concluded from this study. First, the mechanical performance is better in the flow direction than the cross-flow direction, due to fiber orientation effects. Second, this directionality becomes stronger with the further that the material flows during the pressing operation. Third, although the directionality increases with flow distance, the fiber weight percentage is relatively consistent across the panel. Fourth, the shorter fibers appear to influence performance more than the longer fibers, possibly due to fiber bundling of the longer fibers. The impact of the circumference is not clear from the data collected and requires further investigation; although it appears that decreasing the circumference improves flexural properties, the impact on tensile performance is less clear. Further investigation of the tensile strength shall be performed to investigate the response of the specimen shape. The response of the material to both Charpy and Falling Dart impact testing would provide a better understanding of fiber length. Measurements of interfacial shear strength of this system will be performed. Additionally, testing at elevated and reduced environmental conditions could be performed to better understand the response of this system to varying environmental effects. LFT-D technology can be further enhanced by application of local continuous reinforcements in forms of fabrics, profiles and preforms which lead to production of tailored made parts (19). Additionally fiber clumping, rheology, and dispersion could be studied through varying non-destructive techniques to ensure that the plastificate remains consistent from part to part within the same processing conditions.

Page 22

Bibliography 1. Cheah, Cars on a Diet: The Material and Energy Impacts of Passenger Vehicle Weight Reduction in the U.S., in an Unpublished Disseration; MIT Sloan Automotive Laboratory (2010). 2. Weber, C.D., Ledebuhr, S. & Barum, G., One-Piece DLFT Automotive Running Boards , presented at 6th Annual ACCE Conference, Troy, MI (2006). 3. Brast, K. & Michaeli, W., Processing of Long-Fibre Reinforced Thermoplastics Using the Direct Strand-Deposition Process, in Plastics Additives & Compounding, vol. June, pg. 22-29 (2001). 4. Thomason, J.L., Influence of Fiber Properties of the Performance of Glass-Fibre-Reinforced Polyamide 6,6, in Composites Science and Technology, vol. 59, pg. 2315-2328 (1999). 5. Husman, G., Carbon Fiber Composites, Low Cost Materials & Manufacturing Options, presented at SPE Automotive Composites Conference & Exhibition, Troy, MI, (2012). 6. Ernst, H., Henning, F. & Robbins, J. R., Long Fiber Reinforced Thermoplastic LFT-D and Thermosetting D-SMC Processes for Lightweight Parts Production-Trends and Recent Applications, presented at ACCE Conference, Troy, MI, (2009). 7. Krause, W., Geiger, O., Henning, F. & Eyerer, P., Development of a Technology for Large Scale Production of Continuous Fiber Reinforced Thermoplastic Composites, in ANTEC, pg. 2782-2786 (2005). 8. Thomason, J.L. & Vlug, M.A., Influence of Fibre Length and Concentration on the Properties of Glass Fibre-Reinforced Polypropylene; 4. Impact Properties, in Composites: Part A, vol. 28: A, pg. 277-288 (1997). 9. Hassan, A., Hornsby, P. R. & Folkes, M. J., Structure–Property Relationship of Injection-Molded Carbon Fibre-Reinforced Polyamide 6,6 Composites: the Effect of Compounding Routes, in Polymer Testing, vol. 22, pg. 185-189 (2003). 10. Vaidya, U. Composites for Automotive and Truck Applications. DEStech Publications, Lancaster; Pennsylvnia, pg. 99-103, (2011). 11. Thattaiparthasarthy, K.B., Pillay, S., Ning, H. & Vaidya, U.K., Process Simulation, Design, and Manufacutring of a Long Fiber Thermoplastic for Mass Transit ApplIcation, in Composites: Part A, vol. 39, pg. 1512-1521 (2008). 12. Schivje, W. & Ruegg, A., Properties and Test Methods for LFT Materials, in European Alliance for Thermoplastic Composites, preseneted at 10th International AVK Conference, Stuggart, Germany, (2007). 13. Curtis, P.T., Bader, M. G. & Bailey, J. E., The Stiffness and Strength of a Polyamide Thermoplastic Reinforced with Glass and Carbon Fibres, in Journal of Materials Science, vol. 1: 3, pg. 377-390 (1978). 14. McDonough, W.G., Hunston, D.L. & Dunkers, J.P., Test Method for Constituent Content of Composite Materials J. ASTM Int. 1 (3, March), PaperID JAI12076, pg. 1-15 (2004). 15. Kunc, V., Frame, B., Nguyen, B.N., Tucker III, C.L. & Velez-Garcia, G., Fiber Length Distribution Measurement for Long GF and CF Reinforced Injection Molded Thermoplastics (2008). 16. Halpin, J.C., Effects of Environmental Factors on Composite Materials, in US Air Force Material Laboratory, Technical Report AFML-TR-67-423 (1969). 17. Dahl, J., Blanchard, P., Latimer, T., Sudrla, J. & Henshaw, J., A Method For Characterizing Fiber Length Disribution in Random Fiber Composites (2009). 18. Nguyen, B.N., Holbery, J.D., Johnson, K.I. & Smith, M.T., Long Fiber Thermoplastic Injection Molded Composites: From Process Modeling to Property Prediction, in Confernce Proceedings of the 5th Annual Automotive Composites Conference and Exhibition (ACCE) Troy, MI, pg. 27-35 (2005) 19. Krause, W., Henning, F., Troster, S., Geiger, O. & Eyerer, P., LFT-D A Process Technology for Large Scale Production of Fiber Reinforced Thermoplastic Components, in Journal of Thermoplastic Composite Materials, vol. 16: July, pg. 289-302 (2003). 20. Nguyen, B.N. et al., Fiber Length Orientation in Long-Fiber Injection-Molded Thermoplastics- Part I: Modeling Microstructure and Elastic Properties, in Journal of Composite Materials, vol. 42: 10, pg.

Page 23

21.

22. 23.

24. 25.

26.

27.

28.

1003-1029 (2008). Thomason, J.L., Influence of Fiber Length and Concentration on the Properties of Glass Fiber Reinforced Polypropylene 6. Properties of Injection Molded Long Fiber PP at High Fiber Content, in Composites: Part A, vol. 36, pg. 995-1003 (2005). Hamada, H., Futamata, K. & Naito, H., Curing and Deformation Analysis, in ICCM-Proceedings-; vol. 4, pg.190-199 (1997). Zheng, R., Kennedy, P., Phan-Thien, N. & Fan, X.-J., Thermoviscoelastic Simulation of Thermally and Pressure-Induced Stresses in Injection Moulding for the Prediction of Shrinkage and Warpage for Fibre-Reinforced Thermoplastics, in Journal of Non-Newtonian Fluid Mechanics, vol. 84, pg. 159-190 (1999). Tornqvist, R., Sunderland, P. & Manson, J.A.E., Non-Isothermal Process Rheology of Thermoplastic Composites for Compression Flow Modeling, in Composites: Part A, vol. 31, pg. 917-927 (2000). Kim, M.S., Lee, W.I., Han, W.S. & Vautrin, A., Optimisation of Location and Dimension of SMC Precharge in Compression Moulding Process, in Computers and Structures, vol. 89, pg. 1523-1534 (2011). Lafranche, E., Krawczak, P., Ciolczyk, J.P. & Maugey, J., Injection Moulding of Long Glass Fiber Reinforced Polyamide 66: Processing Conditions/Microstructure/ Flexural Properties Relationship, in Advances in Polymer Technology, vol. 24: 2, pg. 114-131 (2005). Park, C.H., Lee, W.I., Yoo, Y.E. & Kim, E.G., A Study on Fiber Orientation in the Compression Molding of Fiber Reinforced Polymer Composite Materials, in Journal of Materials Processing Technology, vol. 111, pg. 233-239 (2001). Thomason, J.L. & Vlug, M.A., Influence of Fiber Length and Concentration on the Properties of Glass Fiber Reinforced Polypropylene 1. Tensile and Flexural Modulus, in Composites: Part A, vol. 27: A, pg. 477-484 (1996).

Page 24