Experimental investigations into extrusion-based 3D printing of PCL/CIP composites for microwave shielding applications

Abstract

In the present work, a solvent-based extrusion 3D printing technique has been utilized to fabricate polymer composites comprising of polycaprolactone (PCL) polymer and carbonyl iron particles. A homogenous composite ink containing PCL and carbonyl iron filler particles with suitable solvent was synthesized with required viscosity for the 3D printing operation. Rectangular samples were successfully fabricated using the extrusion 3D printing technology. A response surface methodology was utilized for planning the set of fabrication experiments so as to estimate the effect of process parameters, namely infill density, printing speed and filler concentration on the printed composite density, percentage of shrinkage and compressive strength. Shrinkage was found to reduce with an increase in infill density and filler concentration. However, density was found to shoot up with an increase in infill density and filler concentration. Similarly, compressive yield strength was improved with an increase in infill density and filler concentration. Increase in shrinkage and density values and decrease in compressive yield strength value were noticed with an increase in the printing speed. Furthermore, a genetic algorithm multiobjective optimization tool was utilized to obtain optimum process parameters to minimize shrinkage, minimize density and maximize compressive yield strength for the electronics and microwave applications of the fabricated composite. A microwave shielding performance test of the developed composite was also carried out as a case study. The shielding performance test indicated the efficacy of the polymer composite fabricated using solvent-based extrusion 3D printing technique.

Keywords

3D printing filler concentration genetic algorithm infill density microwave polymer composites printing speed shrinkage

Introduction

3D printing technology is extensively utilized and commercialized across several industries such as automobiles, aerospace, electronics and biomedical to name a few. These additive manufacturing technologies are transformational and futuristic owing to the prominent attributes they encompass, such as, potential to fabricate intricate designed parts, accurate and repeatable products with minimal cost, automated processing and zero material wastage and so on. A tessellated computer aided design (CAD) model of the desired physical part is utilized as an input to the 3D printers to fabricate the parts using layer by layer approach. Printability of a gamut of materials like polymers, metals, even composites and ceramics is possible today due to alacritous advancement in 3D printing technology frontiers. Currently available 3D printing technology to print composites include stereolithography, fused deposition modelling (FDM), laser engineered net shaping, ultrasonic consolidation, selective laser melting and laminated object manufacturing.^1–5 A considerable size of research explorations have been reported on the composite 3D printing using such techniques. Kalsoom et al.⁶ presented a spectrum of 3D printable composite materials recently developed. They further suggested that various composites with improved properties can be obtained by incorporating different fillers in to the 3D printable polymer matrix. Moreover, this technology can sway several electrochemical, optical and physical behaviour of the polymer composites.⁶ Torrado et al.⁷ investigated the impact of additives on the properties of acrylonitrile butadiene styrene (ABS) parts produced in FDM process. They found that the mechanical property anisotropy along the build direction was reduced for a blend of two materials such as ethylene butadiene styrene and ultra-high molecular weight polyethylene with ABS. The reduction in the anisotropy was 22% as compared to 47% in the case of pure ABS specimens. Ferreira et al.⁸ carried out the material characterization and micrography studies of a composite consisting of polylactic acid (PLA) reinforced with carbon fibre and 3D printed using fused filament fabrication. It was found that the stiffness of the composite increased to 2.2 times along the print direction and 1.25 times along transverse to the print direction as compared to the pure PLA specimen. Panda et al.⁹ developed a fly ash-based geopolymer composite and investigated its rheology and strength for extrusion-based 3D printing. Authors observed the 3D printed specimen with higher compressive strength than the mould casted ones. Zheng et al.¹⁰ presented an extrusion-based 3D printing of ink with apparent low viscosity using silicone elastomers of varying modulus. They suggested a spectrum of low viscosity and readily mixable silicone inks utilizing a combination of base polymers, cross linkers, chain extenders and photoinitiators. They further demonstrated the extrusion printing of overhanging multimaterial structure using single nozzle printer. Kim et al.¹¹ optimized the 3D printing parameters for screw-type extrusion-based 3D printing of ceramics. They utilized L9 orthogonal array to determine the effect of independent 3D printing variables such as water content in ceramics ink, revolutions per minute (RPM), diameter of nozzle tip on the discharge volume of material. It was found that RPM was the most significant parameter with a contribution of 70%. Yang et al.¹² investigated the ABS polymer, reinforced with carbon fibre in an FDM process. When compared with the tensile and flexural strength of the composite with that of ABS, the results indicated impressive increase in both accounts. Duty et al.¹³ suggested a model for the evaluation of the candidature of different polymeric materials for various extrusion-based 3D printing processes.

Owing to the swift growth on the frontiers of polymer composites, these composites can be utilized as a potential candidature for absorption of electromagnetic radiation.¹⁴ These absorbers are the unique class of materials which are used to minimize or absorb the harmful electromagnetic radiations. The shielding materials are selected considering the design, processing, functional and economical aspects. These materials are as diverse as metals to polymers and more recently the polymer composites.¹⁴ Metals have been considered as the foremost choice for researchers owing to the high electrical conductivity. However, the challenges associated with metal microwave absorbers such as high density, poor corrosion resistance, less ductility, difficulty in processing and high cost compelled the researchers to hunt for other materials.¹⁴ In this context, lightweight, ductility, cost-effectiveness and tailorable functional properties make a polymer composite material as a viable substitute for metal-based absorbers.^14–17

Survey of some of the aforementioned prior research work suggested that the extrusion 3D printing process has been used to 3D print composites with improved mechanical properties. However, the capability of this disruptive technology is curtailed owing to the restricted spectrum of 3D printable materials with narrow range of physiochemical properties. Furthermore, there is scanty literature on the solvent-based extrusion 3D printing process, and most of the literature on extrusion 3D printing process are focused on thermal-based extrusion process such as FDM. Hence, there is a contemporary relevance of developing novel polymer-based composites with efficient electromagnetic shielding potential and improved mechanical properties utilizing solvent-based extrusion 3D printing technique.

Thus, the present article is focused on the fabrication of a composite material consisting of polycaprolactone (PCL) polymer and carbonyl iron particle (CIP) filler using a customized solvent-based extrusion 3D printing technique. A design of experiment technique, namely response surface methodology (RSM) was utilized to plan the experiments for the fabrication of prior mentioned composites samples.

The basic idea behind RSM was to capture the effect and interrelationship between process parameters such as printing speed, infill density and filler concentration on the properties of the fabricated composite sample such as shrinkage, part density and the final compressive strength. The properties of the fabricated samples were measured by following specific ASTM standards. Furthermore, a multiobjective optimization technique based on metal heuristic search algorithm such as genetic algorithm was used to optimize the printing parameters for 3D printing of polymer metal composites using solvent-based extrusion 3D printing technique. Additionally, a case study was also conducted to investigate the potential of such 3D printed composites as microwave shielding materials.

Materials and methodology

To 3D print the polymer composite using solvent-based extrusion 3D printing technique, the composite ink was synthesized following a specific procedure. The details of synthesis of ink and fabrication of composite specimens are described in the subsequent segments.

Synthesis of ink

The raw materials used for the preparation of the composite ink were PCL pallets (glass transition temperature (T_g) = −60°C, melting temperature = 60°C, supplier: Sigma Aldrich, Germany), carbonyl iron powder (CIP) (CS grade, supplier: BASF, Germany) and dichloromethane (DCM) (boiling point= 40°C, supplier: Sigma Aldrich, Germany). PCL pallets (size: 3 mm, density: 1.145 g/ml) and CIPs with particle size ranging from 1 µm to 10 µm were mixed separately with DCM solvent in a proportion of 0.5 g/ml and 0.05 g/ml, respectively. The CIP solution was ultrasonicated for 2 h to eliminate agglomerations. After 5 h, the PCL pallets were observed to be completely dissolved in the DCM, and the CIP solution was mixed to the dissolved solution containing PCL. Ultrasonication of this mixture was carried out for 1 h and kept for 12 h to obtain the final homogenous PCL/CIPs composite ink. The detailed specification and composition of CIP is presented in Table 1, and scanning electron microscopic image of CIP is shown in Figure 1(a). A histogram depicting the particle size distribution of CIP is also presented in Figure 1(b). Furthermore, the rheology of the composite ink is measured with the help of rheometer and presented in Figure 2. It can be observed from Figure 2 that the composite ink undergoes shear thinning which is the characteristic of non-Newtonian fluids. Thus, the viscosity of the fabricated composite ink reduces with applied load suggesting better flowability of the ink during extrusion 3D printing process.

Table 1.

Detailed specification and composition of carbonyl iron powder.

Metal powder	Iron (%)	Carbon (%)	Nitrogen (%)	Oxygen (%)
Carbonyl iron CS grade	99.5	0.03	0.01	0.12–0.3

Figure 1.

(a) Scanning electron microscopic image of CIP (at 20KX) and (b) particle size distribution histogram.

Figure 2.

(a) Viscosity versus shear rate for composite ink consisting of PCL and 2% CIP; (b) PCL/CIP composite ink.

Fabrication of the PCL/CIP composite

The CAD model of the polymer composite with a rectangular shape and dimension of $12.7 \times 12.7 \times 20 {mm}^{3}$ was designed using a commercial software tool Solidworks 2017 as per ASTM D695-15 standard. The CAD model was further converted to stereolithography format (STL) which was considered as the input to another software tool CURA 15.04.6. CURA 15.04.6 was utilized to generate the G-codes from the STL file for the movement of the nozzle and platform of the solvent-based extrusion 3D printer to print the PCL/CIPs composite using layer by layer approach. Figure 3(a) shows the flow chart of the detailed fabrication procedure of the 3D printed PCL/CIP composite using solvent-based extrusion. The computer-aided design model of the composite sample along with dimensions is presented in Figure 3(b).

Figure 3.

(a) Flow chart of the fabrication route of the composite specimen and (b) CAD model of the composite specimen.

Experimentation

Selection of process parameters

The printing parameters for the fabrication of the aforementioned composite using solvent-based extrusion 3D printing process were selected based on the design of experiments technique. Three input parameters, namely infill density, printing speed and filler concentration were considered with five levels for each parameter. The parameters and levels are illustrated in Table 2.

Table 2.

Factors and their levels.

Factors	Description	Levels
Factors	Description	−2	−1	0	1	2
X ₁	Infill density (%)	20	40	60	80	100
X ₂	Printing speed (mm/s)	1	3	5	7	9
X ₃	Filler concentration (%)	2	6	10	14	18

In solvent-based extrusion printing, infill density is the amount of material consumed for printing a part. A higher infill density makes a high-density 3D printed part where as a low infill density makes a lightweight part with larger void spaces. Generally, an infill density of greater than 20% is recommended for a good durability of the 3D printed part.¹⁸ Thus, the infill density range was selected to be between 20% and 100%. Figure 4 demonstrates the infill densities between 20% and 100%. The speed of the nozzle movement on the build platform is defined as the printing speed. At higher speed, the deposited ink does not get sufficient time to solidify, and the previous layer deposited gets disoriented by the next layer. Furthermore, very low speed tends to reduce the adherence between the consecutive layers. Hence, range of the printing speed was selected between 1 mm/s and 9 mm/s. Similarly, the filler (CIPs) concentration range was fixed between 2% and 18% as higher quantity of CIPs tends to increase the density of composite.

Figure 4.

Cross section of the 3D printed specimens with varying infill densities.

Experimental set-up

The experimental set-up for the customized solvent-based extrusion 3D printing is shown in Figure 5. The composite ink was loaded in a syringe with 10-ml capacity, and the syringe was inserted in the 3D printer. The movement of the syringe can be done along the X and Z axes, whereas the platform is responsible for the Y-axis movement. The G-code of the CAD model was utilized as an input to the 3D printer software. Finally, the nozzle was driven to the desired position where the printing can be started. The detailed specification of the experimental set-up is given in Table 3.

Figure 5.

Experimental set-up for solvent-based extrusion 3D printing.

Table 3.

Specification of experimental set-up.

3D printer	Constant parameters
3D printer	Feed (mm)	Layer height (mm)	Nozzle size (mm)	Raster angle (degree)	Printbed temperature (°C)
Solvent-based extrusion printer	10	0.2	0.41	45	28

Experimental design methodology

For the fabrication experiments of the PCL/CIP composite, the experimental design approach employed was central composite design (CCD). This method was suitable to capture the effect of process parameters such as infill density, printing speed and filler concentration on the process responses, namely shrinkage, part density and compressive yield strength. The CCD sampling method is very useful in this context, as it builds a robust polynomial function with statistical integrity to couple the responses with the process variables.¹⁹

The second-order response equation satisfies the following form

Y = β_{0} + \sum_{i = 1}^{n} β_{i} X_{i} + \sum_{i = 1}^{n} β_{i i} X_{i}^{2} + \sum_{j > i} \sum β_{i j} X_{i} X_{j} + ε_{​}

where Y suggests the response variable; n describes the number of variables; $β_{0}, β_{i}, β_{i i} and β_{i j}$ describe the constants of the above-mentioned polynomial; $X_{i}, X_{i}^{2} and X_{i} X_{j}$ symbolize the first order, second order and interaction terms, respectively, of the concerned process variables. In the equation (1), ε represents the random error.

A total number of 20 experiments were carried out with CCD design using a value of ∝ as 2. The value of ∝ in a CCD is based on the criteria of orthogonal blocking. To yield accurate results, rotatable CCD designs are preferred. However, rotatable CCD design resulted in impractical values of process variables which were difficult to achieve at the existing experimental conditions. Thus, value of ∝ was chosen to be 2. Standards and procedures followed to measure response variables are illustrated in the following segments.

Measurement of shrinkage

The shrinkage of the 3D printed samples were calculated using ASTM B610-13 standard. As per the standard, the longer side dimension of the sample should be measured to calculate the shrinkage. Thus, the shrinkage of the 3D printed sample was calculated using the following mathematical relation

S = (\frac{L_{C} - L_{m}}{L_{c}}) \times 100

where, ‘S’ is the shrinkage of the printed sample in percentage; ‘L_C’ is the CAD dimension in mm of the longer side of the sample; and ‘L_m’ is the measured dimension in mm of the longer side of the printed sample using micrometre.

Measurement of density

Density of the 3D printed composite samples were obtained as per Archimedes principle utilizing ASTM B962-15 standard. Equation (3) is used to estimate the density as per the aforesaid standard

D = \frac{M_{a} ρ_{w}}{M_{o} - (M_{t} - M_{w})}

where ‘D’ depicts measured density of composite sample in g/cm³; ‘M_a’ describes mass of the composite sample measured in air in g; ‘M_o’ suggests mass of the composite sample submerged in oil in g; ‘M_t’ suggests the total mass containing both the oil dipped composite sample and the water dipped support in g; ‘M_w’ describes mass of the water submerged support in g; and ‘ρ_w’ is the density of water.

Measurement of compressive yield strength

Compressive yield strength of all the rectangular 3D printed PCL/CIPs composites were measured using a universal testing machine (Instron, Mechanical Engineering Department, IIT Delhi, India) having 10-kN capacity and a traverse speed of 1 mm/min. The sample size used for the uniaxial compression test was $12.7 \times 12.7 \times 20 {mm}^{3}$ . The 0.2% offset method was utilized to compute the compressive yield strength of the printed samples.

Experimental details along with the responses are compiled in Table 4. Figure 6(a) represents the 3D printed samples fabricated as per the experimental conditions listed in Table 4. The crystallographic planes of the 3D printed composite samples with varying filler concentration is shown in Figure 6(b) utilizing X-ray diffraction (XRD) plots. The XRD plot suggests absence of any foreign material in the composite sample.

Table 4.

Experimental details.

Experimental run order	Infill density, X₁ (%)	Printing speed, X₂ (mm/s)	Filler concentration, X₃ (%)	Shrinkage, S (%)	Density, D (g/cm³)	Compressive yield strength, Y_S (MPa)
1	40	3	6	6.50	1.66	1.64
2	80	3	6	3.98	1.85	2.44
3	40	7	6	7.57	1.38	1.40
4	80	7	6	5.87	2.01	1.49
5	40	3	14	4.30	2.42	1.89
6	80	3	14	2.74	2.62	2.66
7	40	7	14	4.20	1.45	2.04
8	80	7	14	3.43	2.37	2.40
9	20	5	10	7.33	1.34	1.31
10	100	5	10	3.73	2.67	1.97
11	60	1	10	4.04	2.47	2.47
12	60	9	10	5.85	1.71	1.71
13	60	5	2	7.90	0.84	1.44
14	60	5	18	2.95	2.18	2.67
15	60	5	10	4.18	1.76	2.63
16	60	5	10	3.71	1.67	2.47
17	60	5	10	3.90	1.62	2.52
18	60	5	10	3.30	1.75	2.29
19	60	5	10	3.52	2.11	2.39
20	60	5	10	3.14	1.81	2.50

Figure 6.

(a) 3D printed PCL/CIPs composite samples; (b) XRD plot of pure PCL sample and composite samples with 2%, 6%, 10%, 14% and 18% CIP filler.

Statistical modelling of process responses

The regression equations for all the responses such as shrinkage, density and compressive yield strength were formulated using CCD methodology. The regression equations for the aforementioned responses after eliminating insignificant terms are presented using equations (4) to (6). A 95% confidence interval was considered while formulating the regression equations

S = 18.54 - 0.2390 \times X_{1} - 0.519 \times X_{2} - 0.847 \times X_{3} + 0.001178 \times X_{1}^{2} + 0.0812 \times X_{2}^{2} + 0.0278 \times X_{3}^{2} + 0.00295 \times X_{1} \times X_{3} - 0.037 \times X_{2} \times X_{3}

D = 1.949 - 0.0281 \times X_{1} - 0.352 \times X_{2} + 0.2029 \times X_{3} + 0.000164 \times X_{1}^{2} + 0.02173 \times X_{2}^{2} + 0.00363 \times X_{1} \times X_{2} - 0.01719 \times X_{2} \times X_{3}

Y_{S} = - 1.637 + 0.0880 \times X_{1} + 0.204 \times X_{2} + 0.0995 \times X_{3} - 0.000532 \times X_{1}^{2} - 0.02509 \times X_{2}^{2} - 0.00682 \times X_{3}^{2} - 0.00350 \times X_{1} \times X_{2} + 0.01688 \times X_{2} \times X_{3}

The analysis of variance (ANOVA) tables for the aforementioned responses were obtained using commercially available design of experiment software package (MINITAB 17). After removing insignificant parameters, the final ANOVA table for all the responses are presented in Tables 5 to 7.

Table 5.

ANOVA for shrinkage.

Source	DF	Adj. SS	Adj. MS	F	P	R ²	Remark
Regression	8	48.52	6.06	58.82	0.000	97.72%	From F-distribution table, $F_{0.05, 8, 11} = 2.95$ . $F_{regression} = 58.82 > F_{0.05, 8, 11} .$ Thus, model is adequate. Again, from F-distribution table, $F_{0.05, 6, 11} = 3.10$ . $F_{lack of fit} = 0.63 < F_{0.05, 6, 11} .$ Hence, lack of fit is insignificant
Linear	3	37.95
Square	3	9.42
Interaction	2	1.15
Residual error	11	1.13	0.10
Lack of fit	6	0.39	0.07	0.63	0.990
Pure error	5	0.74
Total	19	49.65

ANOVA: analysis of variance; DF: degree of freedom; Adj. SS: adjusted sequential sum of squares; Adj. MS: adjusted mean sum of squares; F: Fisher’s value; P: level of significance; R²: coefficient of determination.

Table 6.

ANOVA for density.

Source	DF	Adj. SS	Adj. MS	F	P	R ²	Remark
Regression	7	3.80	0.54	17.02	0.000	91.0%	From F-distribution table, $F_{0.05, 7, 12} = 2.91$ . $F_{regression} = 17.02 > F_{0.05, 7, 12} .$ Thus, model is adequate. $F_{lack of fit} = 1.04 < F_{0.05, 7, 12} .$ Hence, lack of fit is insignificant
Linear	3	3.18
Square	2	0.30
Interaction	2	0.32
Residual error	12	0.38	0.03
Lack of fit	7	0.23	0.03	1.04	0.520
Pure error	5	0.15
Total	19	4.25

Table 7.

ANOVA for compressive yield strength.

Source	DF	Adj. SS	Adj. MS	F	P		Remark
Regression	8	4.05	0.51	36.61	0.000	96.38%	From F-distribution table, $F_{0.05, 8, 11} = 2.95$ . $F_{regression} = 36.61 > F_{0.05, 8, 11}$ . Thus, model is adequate.Again, from F-distribution table, $F_{0.05, 6, 11} = 3.10$ . $F_{lack of fit} = 1.02 < F_{0.05, 6, 11}$ .Hence, lack of fit is insignificant
Linear	3	2.45
Square	3	1.30
Interaction	2	0.30
Residual error	11	0.15	0.01
Lack of fit	6	0.08	0.01	1.02	0.445
Pure error	5	0.07
Total	19	4.20

Statistical uncertainty is inevitable owing to experimental errors.¹⁹ Thus, the prediction of responses using the aforementioned regression equations (4) to (6) may not yield the exact values. However, the spectrum in which the predicted responses may vary can be obtained using the following equation¹⁹

{\hat{Y}}_{range} = \hat{Y} \pm t_{α / 2, DF} \times \sqrt{V_{e}}

where ${\hat{Y}}_{range}$ is the range of predicted value of responses; $\hat{Y}$ is the responses prediction from the regression equations formulated in the prior text; $t_{α / 2, DF}$ is the two tailed t-distribution value at a particular level of significance (α) and total degree of freedom; V_e is the Adj. MS of the residual error. To confirm the validity of the regression equations, a couple of more experiments were carried out and the experimental values of responses were found to be within the predicted range from the regression equations. Table 8 presents the confirmatory experiments and the comparison with predicted value.

Table 8.

Confirmatory experiments.

Serial no.	Factors			Shrinkage (%)		Density (g/cm³)		Compressive yield strength (MPa)
Serial no.	X₁ (%)	X₂ (mm/s)	X₃ (%)	Experiment	Prediction $({\hat{Y}}_{range})$	Experiment	Prediction $({\hat{Y}}_{range})$	Experiment	Prediction $({\hat{Y}}_{range})$
1	40	5	10	4.51	3.94 ± 0.77	1.49	1.77 ± 0.42	1.89	1.88 ± 0.24
2	80	5	10	1.65	1.21 ± 0.77	1.82	2.16 ± 0.42	1.92	2.15 ± 0.24
3	100	3	6	2.49	2.62 ± 0.77	1.67	1.92 ± 0.42	1.76	1.83 S± 0.24

X₁: infill density; X₂: printing speed; X₃: filler concentration.

Results and discussion

Shrinkage

Effect of printing process variables

The main effect and contribution plot of process variables for shrinkage are depicted in Figure 7. From Figure 7(a), it can be deduced that shrinkage was reduced with increase in infill density and filler concentration, whereas increase in printing speed increased the shrinkage. Higher infill density ensures lesser gap between two consecutive printed lines thereby prohibiting the further reduction in dimension of the printed part. Furthermore, higher printing speed does not render sufficient time for the proper flow of the extruded filament thereby producing samples with non-uniform material at different regions of the sample, which in turn increases the shrinkage. From Figure 7(b), it was observed that filler concentration was the most contributing factor affecting shrinkage with a contribution percentage of 46%. The reason for the same may be attributed to the fact that increased filler concentration reduces the polymer percentage and the volume of solvent required to prepare the composite ink. Reduced volume of solvent ensures lesser evaporation to obtain solid samples thereby reducing the shrinkage.

Figure 7.

(a) Main effect plot and (b) shrinkage contribution plot.

Effect of interaction

The interaction plot comprising of surface plot and 2D plot between infill density and filler concentration on shrinkage is shown in Figure 8. It can be inferred from the interaction plot that minimum printing shrinkage was obtained at a higher filler concentration and medium to high value of infill density (60%). Furthermore, Figure 9 represents the interaction plot between printing speed and filler concentration on shrinkage. It can be observed from the interaction plot that higher filler concentration and minimum printing speed result in minimum shrinkage in printed parts. The reason for the same has already been narrated in the previous section.

Figure 8.

(a) Surface plot and (b) interaction plot of infill density and filler concentration on shrinkage.

Figure 9.

(a) Surface plot and (b) interaction plot of printing speed and filler concentration on shrinkage.

Density

Effect of printing process variables

The main effect and contribution plot of process variables for density is depicted in Figure 10. It can be inferred from Figure 10(a) that density of the 3D printed parts increased with increase in infill density and filler concentration. However, the effect of printing speed on density was contradictory. Higher infill density suggests more amount of material present in the fixed volume of sample thereby increasing the part density. Furthermore, higher filler concentration renders more amount of carbonyl iron particles in a specific volume of sample. Since the density of carbonyl iron is much higher (7.8 g/cm³) than the density of PCL polymer (1.145 g/cm³), thus increased filler concentration suggests rise in part density. Figure 10(b) represents contribution pot of significant process parameters on density. It was observed that both infill density and filler concentration contribute the most on density.

Figure 10.

(a) Main effect plot and (b) density contribution plot.

Effect of interaction

The interaction plot between infill density and printing speed is presented in Figure 11. It can be observed from the interaction plot that lowest density was observed for lowest infill density and lower printing speed. However, with increase in infill density, the density of the printed part increased exponentially at a higher printing speed. This increase is on expected lines as higher infill density suggests higher amount of composite material in the same volume. Furthermore, the interaction plot between printing speed and filler concentration is shown in Figure 12. Lower filler concentration and higher printing speed suggested lower density for the printed samples. The reason for the same may be accredited to the cause that lower infill amount renders higher polymer concentration and thus lesser density. Moreover, higher printing speed produces thin filaments of non-uniform cross section thereby reducing the density of printed parts.

Figure 11.

(a) Surface plot and (b) interaction plot of infill density and printing speed on density.

Figure 12.

(a) Surface plot and (b) interaction plot of printing speed and filler concentration on density.

Compressive yield strength

Effect of printing process variables

The main effect and contribution plot of process variables for compressive yield strength are depicted in Figure 13. Figure 13(a) illustrates increase in the yield strength with corresponding rise in filler concentration and infill density. Furthermore, a reduction in the yield strength was observed with rising printing speed. Compressive yield strength is higher for brittle materials.²⁰ Increased metal filler concentration induces brittleness in the material thereby increasing the compressive yield strength of the 3D printed composite. Furthermore, increasing infill density till 60% increases the yield strength due to higher intramolecular adhesion. However, beyond 60% infill density, the compressive yield strength reduces as the ductility of material overcomes the adhesion force. From the contribution plot in Figure 13(b), it is noticed that filler concentration is the highest contributing factor suggesting the increased brittleness due to higher amount of metal fillers, resulting in higher compressive yield strength.

Figure 13.

(a) Main effect plot and (b) compressive yield strength contribution plot.

Effect of interaction

Figure 14 depicts the interaction plot between process variables, infill density and printing speed on the response, compressive yield strength. It may be inferred that minimum printing speed and moderate infill density (60%) produced higher compressive yield strength. Furthermore, the interaction plot between printing speed and filler concentration is shown in Figure 15. It can be inferred from the interaction plot that higher filler concentration and lower printing speed resulted largest compressive yield strength. The reason for the same has already been described in the prior section.

Figure 14.

(a) Surface plot and (b) interaction plot of infill density and printing speed on compressive yield strength.

Figure 15.

(a) Surface plot and (b) interaction plot of printing speed and filler concentration on compressive yield strength.

Multiobjective optimization

Optimizing the printing process variables for solvent-based extrusion 3D printing process on the development of PCL/carbonyl iron particle composite, multiobjective optimization was performed using genetic algorithm. MATLAB 2015b was utilized to carry out the optimization process using a population size of 200. The objective of the present work was formulated to utilize the 3D printed composite in electronics and microwave applications. Hence, the composite samples obtained using solvent-based extrusion 3D printing technique is supposed to have minimum shrinkage, minimum density and maximum strength. Thus, in the present context, the objective of the present work was to minimize printing shrinkage, minimize density and maximize the yield strength of the printed part. The details of the multiobjective optimization embracing the objective functions and the input arguments are narrated subsequently

Minimize f_{1} = S (x)

Minimize f_{2} = D (x)

Maximize f_{3} = Y_{s} (x)

x = \{x_{1} x_{2} x_{3}\}

where the input argument x is subjected to the following constraints

20 \leq x_{1} \leq 100

1 \leq x_{2} \leq 9

2 \leq x_{3} \leq 18

The Pareto front encompassing all the Pareto solutions obtained from the multiobjective optimization is represented in Figure 16. Another confirmatory experiment was also carried out using one optimum set of input parameters and compared with the solutions from the Pareto front. Table 9 depicts the confirmatory experiment results.

Figure 16.

Pareto front with optimum set of solutions.

Table 9.

Confirmatory experiment with optimum set of parameters.

Factors			Shrinkage (%)		Density (g/cm³)		Compressive yield strength (MPa)
X₁ (%)	X₂ (mm/s)	X₃ (%)	Experiment	Prediction	Experiment	Prediction	Experiment	Prediction
50	6	12	4.51	3.84 ± 0.77	1.63	1.90 ± 0.42	1.90	2.13 ± 0.24

X₁: infill density; X₂: printing speed; X₃: filler concentration.

Case study

To evaluate the potential of the aforementioned 3D printed PCL/CIPs composite for an effective microwave shielding application, the absorption properties in the microwave spectrum of the 3D printed composite were measured. The facility utilized for this purpose was the vector network analyzer via the rectangular wave guide method. The 4-mm-thick fabricated samples were used for the measurement of microwave properties. The measurements were performed at X-band frequency range from 8 GHz to 13 GHz. The microwave shielding properties of the aforementioned polymer composite, 3D printed using the optimum parameters from Table 9, in terms of the reflection loss and transmission loss are presented in Figure 17. Furthermore, a comparative assessment of the reflection loss of the composite fabricated using solvent-based extrusion 3D printing with some of the previous literature is presented in Table 10.

Figure 17.

Microwave shielding parameters of PCL/CIP (12%) composite in X band frequency range.

Table 10.

Comparative evaluation of reflection loss.

Serial no.	Prior work		Present work
Serial no.	Literature	Reflection loss (dB)	Reflection loss (dB)	% reduction
1	Ahmed et al.²¹	−12	−27	125
2	Quing et al.²²	−17	−27	58
3	Sunny et al.²³	−18	−27	50

It can be depicted from Figure 17 that the PCL/CIP composite sample fabricated, utilizing the optimum printing parameters in solvent-based extrusion 3D printing, exhibits a reflection loss value as low as −27 dB and transmission loss value as low as −18 dB. The reflection loss value in decibels suggest that 3.7% of the incident electromagnetic energy is reflected back from the composite sample, whereas the transmission loss value suggests that 5.6% of the electromagnetic energy is transmitted forward from the composite sample. Thus, PCL/CIP composites fabricated using solvent-based extrusion 3D printing process can be recommended as a potential candidature for microwave shielding material. Moreover, Table 10 renders a comparison of the reflection loss value of the present composite with some of the prior literature. A conclusive reduction in the reflection loss of the PCL + CIP composite fabricated using solvent-based extrusion 3D printing technique was realized.

Conclusion

In the present work, the experimental investigations followed by statistical modelling and multiobjective optimization of the solvent-based extrusion 3D printing of PCL/CIP composite yielded the following concluding remarks.

Polymer composites comprising of PCL polymer and carbonyl iron particles were successfully fabricated using solvent-based extrusion 3D printing technique. The XRD plots confirmed the absence of any other foreign material in the composite samples.

An RSM-based statistical model was used to study the relative effect and interrelationships between printing process variables such infill density, printing speed and filler concentration on the process response parameters such as the shrinkage, density and compressive yield of the printed composite parts.

A reduction in shrinkage was observed with increase in infill density and filler concentration, whereas the printing speed increase was found to increase shrinkage. On the other hand, density and compressive yield strength were observed to increase with increase in infill density and filler concentration and a decrease in printing speed.

A set of optimum process parameters were obtained using multiobjective optimization, genetic algorithm. The optimum process parameters such as 50% infill density, 6 mm/s printing speed and 12% filler concentration yielded 3.84% shrinkage, 1.9 g/cm³ density and 2.13 MPa compressive yield strength for the composite part.

The case study involving microwave shielding performance test strongly suggested the efficacy of the solvent-based extrusion 3D printing technique in developing lightweight microwave shielding material with reasonable compressive strength.

The reflection loss and transmission loss values of the 3D printed PCL/CIPs composite with 50% infill density, 6 mm/s printing speed and 12% filler concentration was found to be −27 dB and −18 dB, respectively.

The losses in the electromagnetic energy in terms of reflection and transmission were found to be 3.7% and 5.6%, respectively, thereby suggesting the rest energy that is 90.7% to be absorbed by the 3D printed PCL/CIP composite fabricated using solvent-based extrusion 3D printing technique.

Footnotes

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Pulak Mohan Pandey

References

Kumar

Kruth

. Composites by rapid prototyping technology. Mater Des 2010; 31: 850–856.

Parandoush

Lin

. A review on additive manufacturing of polymer-fiber composites. Compos Struct 2017; 182: 36–53.

Wang

Jiang

Zhou

, et al. 3D printing of polymer matrix composites: a review and prospective. Compos B: Eng 2017; 110: 442–458.

Quan

Keefe

, et al. Additive manufacturing of multidirectional preforms for composites: opportunities and challenges. Mater Today 2015; 18: 503–512.

Ngo

Kashani

Imbalzano

, et al. Additive manufacturing (3D printing): a review of materials, methods, applications and challenges. Compos B: Eng 2018; 143: 172–196.

Kalsoom

Nesterenko

Paull

. Recent developments in 3D printable composite materials. RSC Adv 2016; 6: 60355–60371.

Torrado

Shemelya

English

, et al. Characterizing the effect of additives to ABS on the mechanical property anisotropy of specimens fabricated by material extrusion 3D printing. Addit Manuf 2015; 6: 16–29.

Ferreira

RTL

Amatte

Dutra

, et al. Experimental characterization and micrography of 3D printed PLA and PLA reinforced with short carbon fibers. Compos B: Eng 2017; 124: 88–100.

Panda

Unluer

Tan

. Investigation of the rheology and strength of geopolymer mixtures for extrusion-based 3D printing. Cem Concr Compos 2018; 94: 307–314.

10.

Zheng

Zlatin

Selvaganapathy

, et al. Multiple modulus silicone elastomers using 3D extrusion printing of low viscosity inks. Addit Manuf 2018; 24: 86–92.

11.

Kim

Cho

Zielewski

. Optimization of 3D printing parameters of Screw Type Extrusion (STE) for ceramics using the Taguchi method. Ceram Int 2019; 45: 2351–2360.

12.

Yang

Liu

XTT

Cao

, et al. 3D printing for continuous fiber reinforced thermoplastic composites: mechanism and performance. Rap Prototyp J 2017; 23(1): 209–215.

13.

Duty

Ajinjeru

Kishore

, et al. What makes a material printable? A viscoelastic model for extrusion-based 3D printing of polymers. J Manuf Process 2018; 35: 526–537.

14.

Pawar

Biswas

. High frequency millimetre wave absorbers derived from polymeric nanocomposites. Polymer 2016; 84: 398–419.

15.

Idris

Hashim

. Recent developments of smart electromagnetic absorbers based polymer-composites at gigahertz frequencies. J Magn Magn Mater 2016; 405: 197–208.

16.

Meng

Wang

Huang

, et al. Graphene-based microwave absorbing composites: a review and prospective. Comp B: Eng 2018; 137: 260–277.

17.

Bakir

Karaaslan

Unal

, et al. Microwave metamaterial absorber for sensing applications. Opto-Electr Rev 2017; 25: 318–25.

18.

Mohanty

Larsen

Trifol

, et al. Fabrication of scalable and structured tissue engineering scaffolds using water dissolvable sacrificial 3D printed moulds. Mater Sci Eng: C 2015; 55: 569–578.

19.

Montgomery

. Introduction to statistical quality control. 6th edn. Hoboken, NJ: John Wiley & Sons, 2009.

20.

Srivastava

Verma

. Mechanical behaviour of copper and aluminium particles reinforced epoxy resin composites. Am J Mater Sci 2015; 5(4): 84–89.

21.

Ahmad

Abbas

Aziz

, et al. Synthesis and characterisation of nickel oxide reinforced with polycaprolactone composite for dielectric applications by controlling nickel oxide as a filler. Results Phys 2018; 11: 427–435.

22.

Qing

Zhou

Luo

, et al. Epoxy-silicone filled with multi-walled carbon nanotubes and carbonyl iron particles as a microwave absorber. Carbon 2010; 48: 4074–4080.

23.

Sunny

Kurian

Mohanan

, et al. A flexible microwave absorber based on nickel ferrite nanocomposites. J Alloys Compd 2010; 489: 297–303.