_k AUTOMATED FIBER PLACEMENT OF PEEK/IM7 COMPOSITES WITH FILM INTERLEAF LAYERS' A. Bruce Hulcher 2,William I. Banks III, R. Byron Pipes and Surendra N. Tiwari Old Dominion University, Norfolk, VA Roberto J. Cano and Norman J. Johnston NASA Langley Research Center, Hampton, VA ABSTRACT The incorporation of thin discrete layers of resin between plies (interleafing) has been shown to improve fatigue and impact properties of structural composite materials. Furthermore, interleafing could be used to increase the barrier properties of composites used as structural materials for cryogenic propellant storage. In this work, robotic heated-head tape placement of PEEK/IM7 composites containing a PEEK polymer film interleaf was investigated. These experiments were carried out at the NASA Langley Research Center automated fiber placement facility. Using the robotic equipment, an optimal fabrication process was developed for the composite without the interleaf. Preliminary interleaf processing trials indicated that a two-stage process was necessary; the film had to be tacked to the partially-placed laminate then fully melted in a separate operation. Screening experiments determined the relative influence of the various robotic process variables on the peel strength of the film-composite interface. Optimization studies were performed in which peel specimens were fabricated at various compaction loads and roller temperatures at each of three film melt processing rates. The resulting data were fitted with quadratic response surfaces. Additional specimens were fabricated at placement parameters predicted by the response surface models to yield high peel strength in an attempt to gage the accuracy of the predicted response and assess the repeatability of the process. The overall results indicate that quality PEEK/IM7 laminates having film interleaves can be successfully and repeatability fabricated by heated head automated fiber placement [1]. Key Words: Thermoplastics, Fiber placement, Interleaving, Barrier films This paper isdeclared a work of the U.S. Government and isnot subject tocopyright protection inthe United States. 2Currently employed at the NASA George C.Marshall Space Flight Center, Huntsville, AL, and towhom all correspondence should be addressed. 4 1.0 INTRODUCTION The enhancement of the mechanical properties of composite materials through various methods has been investigated in some detail over the past several years. Most of these methods have focused on improving damage tolerance, interlaminar fracture toughness, and fatigue life. Design approaches for optimization of composite material mechanical properties include laminate stacking sequence, fiber orientation, z-axis reinforcement, modifications to the matrix resin, and hybrid laminate concepts including resin and metal interleaves [2]. The most important material property governing the fatigue behavior of composite materials is the toughness of the matrix resin component. Most composite material failures occur as the result of ply delamination, especially in mechanical fatigue. Attention has been directed toward the toughening of this inter-ply region. This may be achieved by modification of the matrix resin or by the incorporation of thin discrete layers of tough, ductile resin between plies. The addition of such interleaves has been shown to improve fatigue and impact properties by increasing the interlaminar fracture toughness. Ideal materials for use as interleaves are tough, high strain-to- failure thermoplastics that reduce shear stress concentrations at the ply interfaces. The current study attempts to determine the feasibility of fabricating composite structure having film interleaf layers by heated-head fiber placement technology. Preliminary process development was carried out on the composite without the interleaf layer prior to conducting trials on the fabrication of interleaved laminates. A summary of the study and suggestions for future work are given. 2.0 EXPERIMENTAL 2.1 General 2.1.1 Fiber Placement Facility The NASA Langley Research Center automated fiber placement facility consists of an Asea Brown Boveri 3 (ABB) Model IRB 6000, six-axis, fully-articulating robot [4]. Attached to this platform is the robot heated-head placement end-effector, designed and fabricated by Automated Dynamics Corporation. A fully-programmable rotating spindle is included which provides a seventh axis for fabrication of cylindrical components. The facility is also equipped with a heated flat tool for fabrication of open-section panels. The head consists of a tape dispensing system that feeds material from the spool creel, through a guide chute, and onto the placement surface. The head is capable of placement of up to five 0.635 cm wide composite ribbons or one 3.17 cm wide composite tape. A combination of hot gas and radiant heat sources are available to preheat both the incoming and substrate materials prior to laminate compaction and consolidation [4]. The hot gas source consists of two nitrogen torches that are capable of heating the gas to 900°C. The heated gas is directed into the nip-point region during placement via a single nozzle shared by the torches. A 1000 W tungsten halogen lamp serves as the radiant heat source [5]. The preheating sources may be used either individually or in tandem during processing depending on the thermal properties of the composite and other placement parameters. Theincomingtapeisfedbeneathacompactionrollerthatmaybeheatedto500°Cbyaninternal cartridge resistanceheater. Gas torch and compaction roller temperaturecontrol are accomplishedby closed-loopfeedbackcontrolsystems.Rollertemperatureis monitoredby an IRsensorlocatedaftoftheroller•Torchtemperatureismonitoredby twothermocouplesthatare exposedtothegasstreamwithin thesharedgasexit nozzle.Laminateconsolidationis effected byapneumatically-actuatecdompactionrollerwhichiscapableofproducingconsolidationloads upto 1.47kN. 2.1.2 Material The composite material used in the study was supplied by Cytec Fiberite 3 and was manufactured by their proprietary 'TIFF' process• This material consisted of PEEK resin (Tg:145°C, Tm:342°C) and IM7 fibers in a fully-consolidated and dimensionally accurate tape form. The width and thickness of the as-received tape were 3.17 cm and 0.014 cm, respectively. Tape void content was determined by both acid digestion (ASTM D3171-76) and optical image analysis. Results from digestions indicate a void content of less than 1%. Optical image analysis of void content was performed on an Olympus BH-2 laboratory microscope using the Olympus • • • 2 Cue 2 image acquisition and analysis system. Thirty screen _mages totahng 0.06 cm of area were analyzed• The mean void content as determined by image analysis was found to be 0.45%. A photomicrograph of five samples of the as-received composite tape is shown in Figure l(a). The micrograph reveals smooth, flat ribbon surfaces on the well-consolidated tape form. 2.2 Summary of Composite Process Development Prior to initiating process development with interleaf films, optimization of the process for fabricating non-interleaved laminates was conducted. Unidirectional four-ply wedge peel specimens were fabricated and tested and the results, together with void content data, were used as process quality indicators [6]. Preliminary processing trials were conducted to determine the approximate ranges of hot gas torch temperature and compaction roller load that should be more fully investigated in a process optimization study. A Box-Wilson designed experiment was chosen for process optimization. Thirteen experiments were conducted with roller temperature and compaction load ranges of 600°C - 800°C and 0.67 kN - 1.20 kN, respectively• Held constant were IR lamp output (100%), placement speed (2.54 cm/s), and compaction roller temperature (475°C). Based upon both peak and average peel strength results, the optimal settings for the compaction load and torch temperature were found to be 1.33 kN and 700°C. Void content by optical image analysis was obtained for each of the thirteen experiments and was found to range from 0.18% to 3.72%. Analysis of the void content and strength data revealed no statistical correlation between the two sets; final optimization was based solely upon the peel strength data [1]. A photomicrograph of a cross-section of a four-ply peel specimen (0.18%voids) is shown in Figure l(b). 2.3 Interleaf Film Process Development 2.3.1 Preliminary Processing Trials Initial placement trials were conducted with 0.076 cm PEEK film. Single strips of composite tape 45.7 cm long and 3.17 cm wide were placed onto the tool surface at the conditions deemed optimal in the composite process development 3The use of trademarks or names of manufacturers in this report is for accurate reporting and does not constitute an official endorsement, either expressed or implied, of such products or manufacturers by the NASA. _•'i, "r'=lir i o,_-_% ........ ;.,......... • '/'1' ..." ) 7 ; Figure 1. (a) Photomicrograph of as-received PEEFUIM7 composite tape. (b) Photomicrograph of cross-section of 4-ply composite peel specimen. portion of the study. Strips of film were taped to the beginning end of the composite ply and draped along their length. These strips were manually constrained and placed in minimal tension at the free end during processing. This served to prevent the film from buckling upward from the substrate composite ply surface and thus from contacting the heated roller prior to the moment of adhesion. An initial set of experiments was conducted using the heated compaction roller as the sole heat source. The roller compaction force and placement rate were held constant at 0.44 kN and 2.54 cm/s. The roller temperature was varied from 270°C to 520°C. This range was used such that actual film temperatures from above Tg to above Tm could be attained at the fixed placement speed of 2.54 cm/s. The results for roller temperatures of from 410°C to 420°C indicate that the films could be made to adhere lightly, or to be 'tacked', to the composite ply surface. The best results for light adhesion as indicated from visual inspection were obtained with roller temperatures of between 320°C and 400°C. These films were observed to be of high integrity and uniformity along the specimen length and could be readily peeled from the ply surface. This was an indication that neither the film nor composite matrix resin had fully melted and thus interfacial healing of the two components had not been achieved: measurement of the film thickness and width indicated no change of dimension, a fact that supported this conclusion. A series of 4-ply peel specimens was fabricated to determine the peel strengths at the film/composite interface using only a film 'tack' processing stage. Although it was determined that the film had not fully bonded to the substrate ply after the tack pass, it was thought that placement of the subsequent composite ply might effect intimate healing of the film to both the lower and upper composite plies. Roller temperatures for the tack stage were varied between 3000C and 390°C. The compaction force during film processing was held constant at 0.56 kN. The composite plies were again processed at the conditions deemed optimal in the earlier composite processing experiments. No generaltrend was foundbetweenthe strengthdataandincreasesin roller temperature, indicatingthatfilm tacktemperaturesin thisrangedonotsignificantlyinfluencethequality of theinterfacialbond.Althoughseveralofthepeakstrengthvalueswerehigh,theaveragestrength valuesarewell belowthosefoundpreviouslyfor thecompositealone.Visual inspectionof the fracturesurfacesaftertestingrevealedresin-richregionson the upperply surfacesandbare compositesurfaceson thelower plies.This would indicatesignificantadhesivefailure of the film atthelowercompositeply surface.Composite cohesive failure was observed as evidenced by fiber pullout, however was considered to be minimal for this specimen group. These results led to the conclusion that a second processing pass would have to be executed in order to more fully melt and bond the film to the lower composite ply surface. A second set of peel specimens was fabricated in an attempt to increase film adhesion to the composite substrate ply. Films were lightly tacked to the substrate as previously accomplished and a second film 'melt' processing pass was performed. These melt staging trials were conducted at higher compaction roller temperatures and with additional thermal energy supplied by the IR radiant heat source. Earlier melt stage trials using a rotating compaction roller resulted in severe film de-bonding from the lower composite ply as the result of film pull-up and adhesion to the roller surface. This difficulty was effectively eliminated by constraining roller rotation during melt stage processing. The use of a 'sliding' roller was continued throughout the remainder of the study during the film melt processing pass. Only marginal increases in peel strength were achieved for these initial tack/melt processing trials. The averages of both the peak and average peel strengths from the specimens having no melt stage were 9.34 kN/m and 3.62 kN/m, respectively. The averages for the results with melt stage processing for the same data were 9.64 kN/m and 4.25 kN/m, respectively. The fracture surfaces were visually inspected and adhesive failure of the film at the upper surface of the substrate composite ply was again observed, though to a lesser extent than for the previous experiment set. These results indicate that the use of a second processing pass is beneficial in terms of increasing film adhesion to the substrate ply. A final set of preliminary experiments using the gas torches was conducted in an attempt to achieve complete film/composite adhesion. The first, second, and fourth composite plies were fabricated at the same processing conditions determined to be optimal in the composite process development portion of the study. Due to the possible influence of the processing parameters of the ply immediately preceding the film layer on specimen peel strength, the compaction load and torch temperatures for this ply were varied in the experimental set. Held constant for the film processing passes were the film tack stage roller temperature (340°C), the film melt stage roller temperature (480°C), the compaction load (0.56 kN) during both film tack and melt stage processing, and the lamp percent output during the melt processing stage (100%). Processing parameters that were varied for the experiments were the melt stage processing rate, the torch temperature during film melt staging and the 'upper' composite ply compaction load and torch temperature. The results of the study are presented below in Table 1. The data represent a 43% increase in average peel strength and an 18% increase' in peak peel strength as compared to the previous experiment set. In contrast to the previous studies, examination of the peel surfaces revealed the absenceof a distinctfilm layerontheupperply fracturesurfaces,indicativeof anincreasein adhesionofthefilm tothelowercompositeply. Additionally,inspectionofbothlowerandupper specimenfracturesurfacesrevealedfiberpulloutandthuscohesivefailurewithin thecomposite plies. Table1.Resultsof4-plypeeltestingofspecimensprocessedatincreasedmeltstageandupper compositeplyprocessingtemperatureandcompactionforce.All compositepliesplacedat2.54 cm/s. Note: Torch temperatures and loads li•sted for composite. are for the 3rtl, or uppe, r composite ply only. Specimen Load Torch Temp. Speed Peak Ave. Strength Number Composite Melt/Composite @ Melt Strength (kN/m) (kN) (°C) (cm/s) (kN/m) 1 1.33 - / 700 1.27 10.73 7.84 2 1.33 700 / 700 1.27 12.18 8.19 3 1.11 - / 700 1.27 11.94 4.89 4 1.11 - / 850 1.27 12.34 9.92 5 1.11 700 / 850 2.54 10.93 7.79 6 1.11 - / 850 2.54 11.42 8.33 Several observations regarding the results of these trials may be made. Specimens 1 and 3 of Table 1 were placed at the same conditions with the exception of the upper-ply composite compaction load. This force was 1.33 kN for specimen 1 and 1.11 kN for specimen 3. The average peel strength for specimen 1 (7.84 kN/m) was 38% higher than for specimen 3 (4.89 kN/m). The mean value of the average strength data of the specimens fabricated with a 700°C torch during upper composite ply processing was found to be 20% lower than the mean of those fabricated with a torch temperature of 850°C. Each of these findings would indicate that the processing parameters used during placement of the third composite ply are a significant factor in the process. The specimens having the highest average peel strengths overall, specimens 4 and 6, were processed without the gas torches during the melt stage. The only difference in the processing of these two specimens was the processing rate during the film melt stage. Specimen 4 was placed at half the rate of specimen 6 and shows an increase in both peak, and average peel strength. This would be expected from placement at lower rates; the amount of thermal energy available is a function of both thermal energy source temperatures and placement processing rates. It was also noted that three out of the top four specimens in terms of average peel strength were fabricated without gas torches during the film melt stage. This could be due to stagnation flow of the hot gas in the nip-point region; only small increases in actual material pre-heat temperatures result from relatively large changes in torch temperature. The results of the three preliminary film processing trials are presented in Table 2. Table 2. Results ofpreliminal 7 film processin_ trials. Processing Method Peak Peel Strength Average Peel Strength (kN/m) (kN/m) Tack Stage Only 9.35 3.62 Melt and Tack I 9.64 4.25 Melt and Tack II 11.59 7.83 2.3.2 Screening Experiments A relatively large number of process variables could possibly influence the adhesion quality and the resultant peel strengths of the film specimens. In an attempt to determine which, if any, of these variables have little or no measurable impact on the quality of the film-composite bond, a Plackett-Burman screening experiment set was performed. This design is specifically intended to screen a large number of potentially important factors that may affect the desired quality characteristic(s). The disadvantage with this design is that, while the main effects of a large number of factors may be determined, knowledge of any non-linear effects is forfeited. A screening experiment worksheet was generated having 16 experimental runs. The process parameters for the first, second, and fourth composite plies were held constant as follows: placement rate, 2.54 cm/s; compaction load, 1.33 kN; compaction roller temperature, 480°C; torch temperature, 700°C; IR lamp output, 100%. Film tack stage processing parameters held constant were processing speed, 2.54 cm/s; compaction load, 0.56 kN; compaction roller temperature, 360°C. Specimens were peel tested and values for both the peak and average peel strengths were recorded. The results of the screening experiments are listed in Table 3 in order of their relative importance as determined by the magnitude of their confidence coefficients. Clearly the most significant factors influencing the peel strengths are the melt stage roller temperature, compaction load, and use of the IR lamp during melt staging. It is assumed that the majority of thermal energy available to melt and bond the film to the lower composite ply is transferred via conduction by the heated compaction roller. The relatively minor influence of the torch temperature during the melt stage is again thought to be due to gas flow stagnation in the nip point region. The IR lamp was originally added to the placement machine to serve as a supplemental preheat source due to the inability to elevate the nip point temperatures sufficiently with heated gas alone. The high ranking of the IR heat source is therefore confirmation of the benefit of this additional heat source for material pre-heating. The importance of the compaction load during the film melt stage is thought to be due to an increase in melt flow and interfacial healing at higher compaction pressures. The processing variables for placement of the upper composite ply were investigated in the study to determine if the presence of the film layer on the substrate would significantly alter the optimal conditions as found for the composite alone. The torch temperature and compaction load were varied for placement of this ply. The torch temperature was found to be more significant during placement of the upper composite ply than during the film melt processing stage. The compaction load for placement of the upper ply was found to be the least important of the screened parameters. This may be explained by the presence of the film layer; the film provides a smooth, resin-rich surface for the upper composite ply to adhere to. The processing speed during the film melt stage was found to be of borderline importance in the screening experiment. Due to the uncertainty in the significance of this parameter in the process, and to the relatively low values of peel strength obtained in the screening experiments, further experiments designed to optimize the process were carried out at lower placement speeds. Table 3. Confidence coefficients obtained from process variable screening experiments based on average peel strength data. Process Variable Confidence Coefficient Melt Stage Roller Temperature 0.89 Melt Stage Compaction Load 0.85 Melt Stage IR Lamp 0.80 Melt Stage Processing Speed 0.70 Upper Composite Ply Torch Temperature 0.61 Melt Stage Torch Temperature 0.38 0.22 Upper Composite Ply Compaction Load 2.3.3 Optimization by Design of Experiments and Response Surface Methodology The results of the Plackett-Burman screening suggest that the three most significant parameters for processing film interleaves, in terms of average peel strength, are the melt stage roller temperature, the melt stage compaction load, and the IR heat source during melt staging. The most insignificant parameters were determined to be the torch temperatures and the compaction load during upper composite ply placement and the torch temperatures during the melt stage. As previously stated, the processing rate during the film melt stage was of borderline importance. Further investigations to determine the bounds of an optimal process were performed [1]. A series of experiments utilizing a Box-Behnken design were conducted at film melt rates of 0.64 cm/s, 0.95 cm/s, and 1.27 cm/s. Torch temperature values for placement of the upper composite ply and for the melt stage processing pass were held constant at 700°C. The compaction load for placement of the upper composite ply was also held constant at 1.33 kN. The processing rate for all of the composite plies was held constant at 2.54 cm/s. The compaction roller temperature, the compaction load, and the lamp percent output during the film melt stage were varied during the experiments. The melt stage roller temperature range investigated was 400°C to 480°C, the compaction load range was 0.56 kN to 1.22 kN, and the lamp output power range was set from 0% to 100%. 3.0 RESULTS AND DISCUSSION 3.1 Interleaf Processing The peel data obtained from the Box-Behnken optimization experiments of Section 2.3.3 were fitted using a quadratic regression model. The resultant response surfaces were generated using JMP statistical software. An analysis of the results was undertaken to determine the goodness-of-fit for the response surface models at each of the three film melt processing rates. The indicators used to determine the quality of fit of the quadratic model to the data were the R-value, the sum-of-squares of the model (SSm), and the sum-of- squares of the error (SSe). An R-value of 1 signifies a perfect fit and complete confidence in the predictive capability of the model. Conversely, low R-values signify a poor fit, and hence an inability of the model to make predictions regarding the dependent variable. Additionally, the goodness-of-fit may be determined from the sum of the squares of the model and the error. Good correlations have SSe values much less than the SSm. Thequalityof fit ofthemodelfortheexperimentsperformedatamelt speedof0.64cm/swas foundto bethebestof thethreeexperimentgroups.TheR-valueforthefit of theexperiments performedatthisratewasfoundtobe0.93.The SSmandthe SSewerefoundtobe63.75and 10.02,respectively.It isconcludedthatarelativelyhighdegreeof confidencemaybeattributed tothemodelforthisdataset. TheR-valueforthemodelfor theexperimentsperformedat0.95 cm/smeltprocessingspeedwasfoundtobe0.88andtheSSmandSSewerefoundto be56.38 and16.62,respectively,indicatingjustificationintheconfidenceofthismodelaswell. Analysis ofthefit ofthemodelfortheexperimentsperformedat1.27cm/sgiveanR-valueof0.61andan SSm andSSc of 21.1and35.1,respectively.Thesevalueswould indicatethatvirtually no confidencecanbeattributedtothefit ofthemodelforthisdataset. Inspectionof theresponsesurfaceplotsrevealedageneralupwardtrendin peelstrengthwith increasesin compactionroller temperatureandload ateachlamp outputsetting.An upward trendin maximumstrengthvalueswith increasesin IR lampoutputpowerwasobserved.Also notedwasageneralincreasein themaximumpeelstrengthasplacementratesdecreasedfrom 1.27cm/sto0.64cm/s.It maybeconcludedfromthecontourplotsthatthewedgepeelstrengths aremaximizedatan IR lampoutputof 100%anda placementrateof 0.64cm/s,conditions whichprovidethemaximumthermalenergyflux to thematerial. Photomicrographsof the peelspecimenhavingthe highestpeelstrengthsat eachof the film processingrates are presentedin Figures3, 4, and 5. All three specimensexhibit well- consolidatedvoid-freeregionsatboththeupperandlowerfilm/compositeinterfaces.A general decreasein film thickness,however,wasobservedwith decreasingfilm melt processingrates. Measurementsoftheresultantfilm thicknessforeachofthespecimenswasperformedusingan opticalmicroscopefittedwith a BoeckelerInstrumentsMicroCodeII digital positionreadout. Themicroscopemagnificationpowerusedwas500X.Forty film thicknessmeasurementwsere recordedforeachofthethreespecimens. Figure3.Photomicrographof4-plypeelspecimenplacedatmeltstageconditionsof 440°C rollertemperature1,.22kNcompactionloadandwithoutsupplementaIlR lampenergy.Shown isspecimenhavingthehighestpeelstrength(8.30kN/m)ofthosefabricatedat1.27cm/s. Averagefilm thicknessafterprocessing:0.0069cm. Figure4. Photomicrograph of 4-ply peel specimen placed at melt stage conditions of 440°C roller temperature, 0.56 kN compaction load, and at 100% lamp power. Shown is specimen having the highest peel strength (8.00 kN/m) of those fabricated at 0.95 cm/s. Average film thickness after processing: 0.0049 cm. Figure 5. Photomicrograph of 4-ply peel specimen placed at melt stage conditions of 480°C roller temperature, 0.89 kN compaction load and 100% lamp power. Shown is specimen having the highest peel strength (8.60 kN/m) of those fabricated at 0.64 cm/s. Average film thickness after processing: 0.0024cm. The decrease in film thickness may be primarily attributed to the increased melt flow of the film due to the higher energy fluxes achieved at the lower placement rates. As a result of this increase in melt flow, the amount of resin that adheres to the roller surface also increases. This leads to a 'skimming' of the resin from the specimen during the melt processing stage and to the observed decrease in film thickness. The results of the film thickness measurements are presented in Table 4. Note that the ratio of the standard deviation to the mean thickness increases as film thickness decreases; the thickness of the films processed at higher placement rates is less variable than for those processed at lower rates.