sign in sign up
<<
Home , Epoxy, Glass fiber, Glass transition, Optical path, Optics, Polyvinyl butyral

Article Optical and Mechanical Properties of Highly Transparent -Flake Composites

Benedikt Scharfe 1,2,*, Sebastian Lehmann 1 , Thorsten Gerdes 1 and Dieter Brüggemann 3

1 Chair of Materials , Keylab Glasstechnology, University Bayreuth, 95447 Bayreuth, Germany; [email protected] (S.L.); [email protected] (T.G.) 2 Deggendorf Institute of Technology, TAZ Spiegelau, 94518 Spiegelau, Germany 3 Department of Engineering and Transport Processes (LTTT), Centre of Energy Technology (ZET), University of Bayreuth, 95447 Bayreuth, Germany; [email protected] * Correspondence: [email protected]; Tel.: +49-8553-9799-611

 Received: 29 October 2019; Accepted: 18 November 2019; Published: 21 November 2019 

Abstract: In this paper, the dynamic mechanic and of composites made of Polyvinyl Butyral (PVB) and Micro Glass Flakes (MGF) with matching refractive indices (RIs) are investigated. The composite is produced by a slurry-based process using additional blade casting and . It can be shown that a high degree of ordering of the MGF in the can be achieved with this method. This ordering, combined with the platelet-like structure of the MGF, leads to very efficient strengthening of the PVB with increasing content of the MGF. By carefully adjusting the RIs of the polymer, it is shown that haze is reduced to below 2%, which has not been achieved with irregular fillers or glass fibers.

Keywords: Micro Glass Flakes; PVB; Glass Particles; Transparent Composite

1. Introduction Highly transparent have a very broad application . (PC) and polymethyl methacrylate (PMMA) are common examples for applications that require high stiffness like displays, windows or transparent electronic packaging [1]. Low- (LDPE) and polyethylene terephthalate (PET) is used in applications which require high flexibility like packaging and foils for food and beverages, as well as medical applications like tubing, pipes, and bottles [2]. Lesser known are Polyvinyl Butyral (PVB) and Ethylene Vinyl Acetate (EVA), which are utilized in laminated panes and windows as interlayer materials [3]. However, typical polymer properties like high , low stiffness, low , and low wear and scratch resistance have led to different approaches to using fillers to create a transparent composite with enhanced properties [4–8]. Employing glass particles as filler material for transparent polymers with similar refractive indices (RIs) shows promising properties for the application in transparent components [9]. Glass beads in a polymer matrix lead to increased stiffness [10,11] and reduced thermal expansion [12,13]. In theory, perfectly matched RIs between matrix polymer and glass filler should result in an unobstructed transparency [14]. However, in reality, RI matching is far from simple, as and cooling rates at production have a significant impact on the RIs. Also, the RIs of glass and polymers depend on temperature (thermo-optic coefficient) and measured (), which can be used for optical temperature sensors [15,16]. The RI of typical increases with increasing temperature with a thermo-optic coefficient ∆n /∆T between 6.7 and 24.1 10 6/K4 [17]. However, the RI of rel − × − the most common transparent polymers decreases with increasing temperature with thermo-optic coefficients which are nearly two orders of magnitude higher ( 0.9 to 3.1 10 6/K4)[18]. − − × −

J. Compos. Sci. 2019, 3, 101; doi:10.3390/jcs3040101 www.mdpi.com/journal/jcs J. Compos. Sci. 2019, 3, 101 2 of 17 J. Compos. Sci. 2019, 3, 101 2 of 17 with thermo-optic coefficients which are nearly two orders of magnitude higher (−0.9 to −3.1 × 10−6/K4) [18]. Extensive research has been conducted on optica opticall and mechanical properties of glass-reinforced polymers withwith similarsimilar RI RI using using PMMA PMMA and and glass glass fibers [19 ][19] or epoxyor filled filled with with irregular-shaped irregular-shaped glass particlesglass particles [20]. The [20]. influence The influence of different of particledifferent sizes, particle filler sizes, contents and contents influence and of RIinfluence mismatch of was RI studied.mismatch All was studies studied. show All that studies transmission show that decreases transm withission increasing decreases filler with content, increasing even withfiller identicalcontent, RI.even This with was identical traced back RI. This to the was traced behavior back to of the the wetting filler, particularly behavior of when the usingfiller, moldingparticularly processes when orusing inhomogeneity molding processes in the RIor inhomogeneity within the particles in the or RI matrix. within However,the particles using or matrix. refractive However, index oilsusing in cuvettesrefractive as index a model oils matrixin cuvettes with as perfect a model wetting matrix behavior, with perfect it could wetting be shown behavior, that highit could transparency be shown canthat alsohigh be transparency achieved with can high alsofiller be achieved contents with [21]. high filler contents [21]. The aim of this study is to create a new type of transparent composite using Micro Glass Flakes (MGF)—a platelet-likeplatelet-like glassglass particleparticle withwith aa high high aspect aspect ratio. ratio. Micro Micro glass glass particles particles such such as as MGF MGF are are a verya very recent recent innovation in glass in . production. Despite Despite that, they that, have they already have gained already a broadgained application a broad spectrumapplication due spectrum to their chemicallydue to their inert chemically properties combinedinert properties with platelet-like combined morphology with platelet-like [22,23]. Theymorphology are most [22,23]. commonly They used are most as reflective commonly and eusedffect as pigments reflective in cosmeticand effect applications pigments in [24 cosmetic] and as fillerapplications material [24] to increaseand as filler the barriermaterial properties to increase of protectivethe barrier coatingsproperties [25 of]. protective [25]. A method was developed to incorporate these glass flakesflakes in a model polymer via a slurry-based lab-scale castingcasting process process in in order order to increaseto increase the ordering the ordering of these of flakes. these Withflakes. this With combined this combined approach, usingapproach, platelets using in aplatelets highly orderedin a highly assembly, ordered the influenceassembly, ofthe residual influence RI mismatch of residual between RI mismatch particle andbetween matrix particle on the transparencyand matrix on and the haziness transparency of such aand structure haziness could of be such minimized, a structure as displayed could be in Figureminimized,1. At theas displayed same time, in glass Figure flakes 1. At with the a highsame aspect time, ratioglass couldflakes greatly with a improvehigh aspect the mechanicalratio could propertiesgreatly improve of the the composite mechanical in all properties in-plane directionsof the composite simultaneously. in all in-plane directions simultaneously.

Figure 1. Schematics of the influence influence of glass fibers fibers and glass flakesflakes on transmittedtransmitted in a matrix with a mismatch of refractive indices.indices.

2. Experimental 2.1. Materials 2.1. Materials 2.1.1. Micro Glass Flakes 2.1.1. Micro Glass Flakes Three different types of Micro Glass Flakes (MGF) were used in this study. Two of them are Three different types of Micro Glass Flakes (MGF) were used in this study. Two of them are commercially available from Glassflake Ltd® (GF100, GF001, Leeds, UK) [26] and one type was derived commercially available from Glassflake Ltd® (GF100, GF001, Leeds, UK) [26] and one type was by a laboratory flaker setup (GF-V8) with a -rich silicate glass composition (Figure2). They di ffer derived by a laboratory flaker setup (GF-V8) with a boron-rich silicate glass composition (Figure 2). in mean , thickness and thus, in the aspect ratio, as well as the RI due to a different glass They differ in mean diameter, thickness and thus, in the aspect ratio, as well as the RI due to a different composition between the commercial MGF and from the laboratory flaker setup (Table1). glass composition between the commercial MGF and from the laboratory flaker setup (Table 1).

J. Compos. Sci. 2019, 3, 101 3 of 17

J. Compos. Sci. 2019,, 33,, 101101 3 of 17

FigureFigure 2. Scanning 2. Scanning electron microscopy microscopy (SEM) (SEM) image of commercial of commercial Micro Micro Glass Glass Flakes Flakes (MGF). (MGF). Table 1. Properties of Micro Glass Flakes used in this study. Table 1. Properties of Micro Glass Flakes used in this study.

Type Mean Diameter (D50) Thickness Specific Density Type Mean Diameter (D50) Thickness Specific Density µ µ 3 GF100GF100 160 160 µmm 1.0–1.3 1.0–1.3 µm m 2.3 g/cm2.3 g3 /cm GF001 30 µm 0.9–1.3 µm 2.3 g/cm3 GF001 30 µm 0.9–1.3 µm 2.3 g/cm3 GF-V8 120 µm 2.0–5.0 µm 2.3 g/cm3 GF-V8 120 µm 2.0–5.0 µm 2.3 g/cm3

2.1.2. Polyvinyl Butyral (PVB) 2.1.2. Polyvinyl Butyral (PVB) Polyvinyl butyral (PVB) is considered to be an acetal and is formed from the reaction of an Polyvinyl butyral (PVB) is considered to be an acetal and is formed from the reaction of an aldehyde and alcohol. The general structure of PVB is shown in Figure3, whereby the shares of PVB, aldehyde and alcohol. The general structure of PVB is shownshown inin FigureFigure 3,3, wherebywhereby thethe sharesshares ofof PVB,PVB, (PVOH), and segments are variable. Depending on the supposed polyvinyl alcohol (PVOH), and polyvinyl acetate segments are variable. Depending on the supposed application of the PVB, the relative amounts of these segments are controlled but they are generally application of the PVB, the relative amounts of these segments are controlled but they are generally randomly distributed through the molecular chain. randomly distributed through the molecular chain.

Figure 3. Butyral (A), Alcohol (B) and Acetate (C) groups of commercial PVB compositions. The Figure 3. Butyral ( A),), Alcohol ( B) and AcetateAcetate ((C)) groupsgroups ofof commercialcommercial PVBPVB compositions.compositions. The relative amount of A:B:C varies with the targeted application and properties [27]. relative amount of A:B:C varies with the targeted applicationapplication andand propertiesproperties [[27].27].

For thisthis study,study, commerciallycommercially availableavailable PVBPVB powderpowder (Mowital(Mowital®® fromfromfrom kuraraykuraraykuraray®® EuropeEurope GmbH,GmbH, Hattersheim am Main, Germany)Germany) waswas used.used. Mowital® isisis offeredoffered offered inin aa rangerange ofof various various grades, grades, which which are mainlymainly determineddetermined by by their their molecular molecular weights weights and and their their degree degree of acetalization of acetalization [28]. For[28]. this For study, this study,B 30 H B was 30 usedH was as used a base as material a base material for all composites for all composites with an averagewith an molecularaverage molecular weight and weight degree and of degreeacetalisation of acetalisation (Table2). (Table 2).

J. Compos. Sci. 2019, 3, 101 4 of 17

J. Compos. Sci. 2019, 3, 101 4 of 17 Table 2. Properties of common types of Mowital® powder [28].

Properties B 14 S B 30 H B 60 HH B 75 H Table 2. ® Glass transitionProperties temperature of common (°C) types of 60 Mowital 68powder [ 6528]. 73 ContentProperties of polyvinyl alcohol (%) B 14 S 14–18 B 30 18–21 H B 12–16 60 HH 18–21 B 75 H Content of polyvinyl acetate (%) 5–8 1–4 1–4 0–4 temperature (◦C) 60 68 65 73 DynamicContent of polyvinyl 10% alcohol in EtOH (%) (mPa·s) 14–18 9–13 18–21 35–60 120–280 12–16 60–100 18–21 Content ofBulk polyvinyl density acetate (g/cm (%)3) 5–80.34 1–40.32 0.21 1–4 0.20 0–4 Dynamic viscosity 10% in EtOH (mPa3 s) 9–13 35–60 120–280 60–100 Specific density (g/cm ·) 1.1 1.1 1.1 1.1 Bulk density (g/cm3) 0.34 0.32 0.21 0.20 3 1.1 1.1 1.1 1.1 2.1.3. Softener 3G8Specific density (g/cm )

2.1.3.Softeners Softener 3G8are a standard ingredient in commercially available polymer products. They are added to increase the flexibility and workability of the polymers. In PVB applications like interlayer sheets for safetySofteners glass, are Triethylenglycol-di-(2-ethyl a standard ingredient in commercially hexanoate) availableor 3G8 is polymercommonly products. used, with They a aredensity added of to0.97 increase g/cm3 theand flexibility a molecular and weight workability of 402.57 of the g/mol polymers. with contents In PVB ranging applications from like20–27 interlayer wt %. [29] sheets The formolecular , structure Triethylenglycol-di-(2-ethyl is shown in Figure 4. hexanoate) or 3G8 is commonly used, with a density of 3 0.97 gIn/cm thisand study, a molecular softener weightwas added of 402.57 mainly g/mol to decrease with contents the RI ranging of PVB from to further 20–27 wtimprove %. [29 ]index The molecularmatching with structure the MGF. is shown in Figure4.

FigureFigure 4.4. MolecularMolecular structurestructure ofof softenersoftener 3G83G8 [[29].29].

2.2. PreparationIn this study, of Composite softener wasSheets added mainly to decrease the RI of PVB to further improve index matching with the MGF. 2.2.1. Slurry Preparation 2.2. Preparation of Composite Sheets The PVB–MGF composites were derived via a slurry-based preparation method. This process 2.2.1.was chosen Slurry Preparationas it allows to carefully control and adjust the composition and homogeneity of the composites as well as the orientation and distribution of the glass flakes. The PVB–MGF composites were derived via a slurry-based preparation method. This process was In the first step, Mowital® B 30 H powder was mixed with ethanol until a clear solution with chosen as it allows to carefully control and adjust the composition and homogeneity of the composites -like viscosity was achieved. For this, B 30 H powder was added in small portions to ethanol as well as the orientation and distribution of the glass flakes. in a rotary flask over a 48 h period. PVB powder is hygroscopic, with water not only influencing the In the first step, Mowital® B 30 H powder was mixed with ethanol until a clear solution with mechanical but also, the optical properties. Therefore, the flask was continuously purged with dry honey-like viscosity was achieved. For this, B 30 H powder was added in small portions to ethanol in nitrogen gas during the mixing period. The final solution was then kept in an air-tight glass drum to a rotary flask over a 48 h period. PVB powder is hygroscopic, with water not only influencing the prevent the evaporation of ethanol and contamination through . The final composition of mechanical but also, the optical properties. Therefore, the flask was continuously purged with dry the solution was 30 wt % B 30 H to 70 wt % ethanol. nitrogen gas during the mixing period. The final solution was then kept in an air-tight glass drum to In the second step, defined amounts of MGF were added to the solution and mixed using an AR- prevent the evaporation of ethanol and contamination through humidity. The final composition of the 100 Thinky Mixer (Thinky® U.S.A. INC., Laguna Hills, CA, USA). To ensure a homogeneous solution was 30 wt % B 30 H to 70 wt % ethanol. distribution of MGF in the slurry, a total of 6 min of mixing time at 2000 rpm was selected. To further In the second step, defined amounts of MGF were added to the solution and mixed using an prevent any kind of gas or bubble formation, an additional degassing step of 6 min was added after AR-100 Thinky Mixer (Thinky® U.S.A. INC., Laguna Hills, CA, USA). To ensure a homogeneous the initial mixing. The resulting slurry was a clear, homogeneous and bubble-free paste (Figure 5a,b). distribution of MGF in the slurry, a total of 6 min of mixing time at 2000 rpm was selected. To further prevent any kind of gas or bubble formation, an additional degassing step of 6 min was added after the initial mixing. The resulting slurry was a clear, homogeneous and bubble-free paste (Figure5a,b).

J. Compos. Sci. 2019, 3, 101 5 of 17 J. Compos. Sci. 2019, 3, 101 5 of 17

(a) (b)

FigureFigure 5. PolyvinylPolyvinyl Butyral Butyral (PVB)– (PVB)–MGFMGF slurry slurry before before ( aa)) and and after after ( (b)) mixing mixing in the planetary mixer.

In this study, the fraction of MGF and softenersoftener in the finalfinal composite are given as volume percentage (MGF) and weight percentage (3G8) relati relativeve to the PVB. To calculate the necessary weight of the raw components to achieve the desired composition, thethe followingfollowing calculationscalculations werewere used:used:

m𝑚==𝑚m y∙𝑦 (1) PVB solution· PVB 𝑚 mPVB VPVB𝑉== (2) ρPVB

uMGF = 1 uPVB (3) 𝑢 =1−𝑢− (3) uMGF VMGF = VPVB (4) · u𝑢PVB 𝑉 =𝑉 ∙ (4) m = V ρ𝑢 (5) MGF MGF· MGF

Using Equations (1), (2) and (3) with𝑚mPVB =𝑉as mass ∙ of PVB powder, msolution as mass of the PVB(5) ethanol solution, yPVB as the weight fraction of PVB in the solution, VPVB as volume of PVB, VMGF Using Equations (1), (2) and (3) with 𝑚 as mass of PVB powder, 𝑚 as mass of the as volume of the MGF, ρPVB and ρMGF as the specific density of PVB and MGF, respectively, and PVB ethanol solution, 𝑦 as the weight fraction of PVB in the solution, 𝑉 as volume of PVB, uMGF as the volume fraction of MGF in the final composite, the mass of MGF can be calculated with 𝑉 as volume of the MGF,  and  as the specific density of PVB and MGF, respectively, Equation (6). and 𝑢 as the volume fraction of MGF in the final composite,u ρ MGFthe mass of MGF can be calculated m = m y MGF (6) with Equation (6). MGF solution· PVB·1 u · ρ − MGF PVB 𝑢  When softener is added, the relative increased volume of the PVBVPVB,so f tener can be calculated 𝑚 =𝑚 ∙𝑦 ∙ ∙ (6) using Equations (7) and (8) with mso f tener as required mass1−𝑢 of softener, yso f tener as weight fraction of the softener and ρso f tener as the specific density of the softener. When softener is added, the relative increased volume of the PVB 𝑉, can be calculated 𝑚 𝑦 using Equations (7) and (8) with as required ymassso f tener of softener, as weight fraction  mso f tener = mPVB (7) of the softener and as the specific density of·1 they softener. − so f tener 𝑦 𝑚 =𝑚 ∙ m mPVB 1−𝑦so f tener (7) VPVB,so f tener = + (8) ρPVB ρso f tener 𝑚 𝑚 The weight content of MGF in the𝑉, final composite= can,+ therefore, be calculated using Equation (9).(8)     y The weight content of MGF in the finaluMGF composite can,1 therefore, beso calculated f tener using Equation (9). m = m y ρ  + . (9) MGF solution PVB u MGF    · ·1 MGF · ·ρPVB ρ 1 y  − so f tener − so f tener 𝑚 =𝑚 ∙𝑦 ∙ ∙  ∙ + . (9)  ()

2.2.2. Forming and Drying

J. Compos. Sci. 2019, 3, 101 6 of 17

To produce a composite with homogenous thickness and distribution of MGF, a blade casting J. Compos. Sci. 2019, 3, 101 6 of 17 setup, also known as doctor blade or knife coater, was used. As a substrate for the slurry, a standard 0.1 mm LDPE-foil was used. Due to the long chain of the LDPE with no hydrogen groups for 2.2.2.bonding, Forming no to andvery Drying little and sticking behavior to the PVB was observed. The substrate foil was thoroughly cleaned with ethanol before casting and was held in place by an array of holes To produce a composite with homogenous thickness and distribution of MGF, a blade casting combined with an applied . setup, also known as doctor blade or knife coater, was used. As a substrate for the slurry, a standard The sheets were cast using a blade with a width of 250 mm (Figure 6a). The gap between the 0.1 mm LDPE-foil was used. Due to the long carbon chain of the LDPE with no hydrogen groups for blade and the substrate foil was set to precisely 0.5 mm, which was found to be the smallest gap to bonding, no to very little adhesion and sticking behavior to the PVB was observed. The substrate still yield foils with an even thickness distribution, while at the same time, providing a geometric foil was thoroughly cleaned with ethanol before casting and was held in place by an array of holes constraint which is necessary to align the MGF (Figure 6b). The sheets were drawn with a speed of combined with an applied vacuum. 2.5 mm/s. The sheets were cast using a blade with a width of 250 mm (Figure6a). The gap between the blade After the casting step, the sheets were quickly transferred into a drying cabinet where they were and the substrate foil was set to precisely 0.5 mm, which was found to be the smallest gap to still yield kept for 5 to 7 days at room temperature and a relative humidity of below 5%. To ensure that no foils with an even thickness distribution, while at the same time, providing a geometric constraint ethanol or other form of moisture remains in the final composite, the sheets where then further dried which is necessary to align the MGF (Figure6b). The sheets were drawn with a speed of 2.5 mm /s. in a vacuum oven at up to 140 °C for 6 h.

(a) (b)

Figure 6. Blade casting setup ( a) and schematics of the blade casting process with MGF (b).

2.2.3.After Lamination the casting step, the sheets were quickly transferred into a drying cabinet where they were kept for 5 to 7 days at room temperature and a relative humidity of below 5%. To ensure that no After drying, the sheets have a mean thickness of 120 µm (PVB without MGF) to 300 µm (PVB ethanol or other form of moisture remains in the final composite, the sheets where then further dried with 26 vol % MGF). However, due to the inhomogeneous drying, small gas bubbles, as well as a in a vacuum oven at up to 140 ◦C for 6 h. rough surface, particularly at a high volume content of MGF, cannot be prevented. Therefore, an 2.2.3.additional Lamination lamination step is necessary. For this, the sheets were cut into 60 mm round blanks. Three blanks were stacked and placed into a forming mold with -finished surfaces. The After drying, the sheets have a mean thickness of 120 µm (PVB without MGF) to 300 µm (PVB forming mold was then placed into a sealed vacuum bag with a vacuum pump attached. The setup with 26 vol % MGF). However, due to the inhomogeneous drying, small gas bubbles, as well as a rough was then placed into a hot press. With the vacuum pump active, the press was heated to 100 °C and surface, particularly at a high volume content of MGF, cannot be prevented. Therefore, an additional a pressure of 2 kN was applied for a duration of 20 min. The setup was cooled and the composite was lamination step is necessary. For this, the sheets were cut into 60 mm round blanks. Three blanks were taken from the mold. stacked and placed into a stainless steel forming mold with mirror-finished surfaces. The forming It was found in mechanical characterization that the roughness and homogeneity of the samples mold was then placed into a sealed vacuum bag with a vacuum pump attached. The setup was then were sufficient. However, in optical measurements, the samples were found to be too rough. placed into a hot press. With the vacuum pump active, the press was heated to 100 C and a pressure Roughness greatly influences the optical properties, especially the haziness of a sample◦ due to the of 2 kN was applied for a duration of 20 min. The setup was cooled and the composite was taken from of light on an uneven surface. Therefore, all samples were laminated between two glass the mold. sheets for optical characterization. As glass panes, Optiwhite™ glass sheets from Pilkington® (St It was found in mechanical characterization that the roughness and homogeneity of the samples Helens, UK) with a thickness of 2 mm were used. The additional haziness and loss of transmission were sufficient. However, in optical measurements, the samples were found to be too rough. Roughness due to the addition of two glass panes on both sides can be neglected (Figure 7). greatly influences the optical properties, especially the haziness of a sample due to the scattering of light on an uneven surface. Therefore, all samples were laminated between two glass sheets for optical characterization. As glass panes, Optiwhite™ glass sheets from Pilkington® (St Helens, UK) with a thickness of 2 mm were used. The additional haziness and loss of transmission due to the addition of two glass panes on both sides can be neglected (Figure7).

J. Compos. Sci. 2019, 3, 101 7 of 17 J. Compos. Sci. 2019, 3, 101 7 of 17 J. Compos. Sci. 2019, 3, 101 7 of 17

Figure 7. Sample after lamination with two additional glass panes for optical characterization.

2.3. ExperimentalFigure 7. Sample Characterization after lamination Methods with two additional glass panes for optical characterization. 2.3. Experimental Characterization Methods 2.3.2.3.1. Experimental Transparency Characterization and Haze Measurement Methods

2.3.1. Transparency The transparency and Hazeand haze MeasurementMeasurement of the composite were measured according to standards ISO 13,468 and 14,782 [30,31] using a HAZE-GARD plus® haze meter from BYK-Gardener® (Geretsried, The transparencytransparency and and haze haze of of the the composite composite were were measured measured according according to standards to standards ISO ISO 13,468 13,468 and Germany). The total transmission of a ®sample is measured using an integrating® sphere. However, 14,782and 14,782 [30,31 ][30,31] using ausing HAZE-GARD a HAZE-GARD plus haze plus meter® haze from meter BYK-Gardener from BYK-Gardener(Geretsried,® (Geretsried, Germany). depending on the scattering behavior of the sample, transmitted either pass the sample TheGermany). total transmission The total transmission of a sample isof measured a sample usingis measured an integrating using an sphere. integrating However, sphere. depending However, on without disturbance (clarity) or get scattered away from the direct path and, therefore, lead to a thedepending scattering on behavior the scattering of the sample,behavior transmitted of the samp photonsle, transmitted either pass photons the sample either without pass disturbancethe sample phenomenon described as haziness. Haze is measured as relative transmission with an angle above (clarity)without ordisturbance get scattered (clarity) away fromor get the scattered direct path away and, from therefore, the direct lead path to a phenomenon and, therefore, described lead to asa 2.5° from the direct . Clarity is defined as relative transmission within 2.5° of the direct haziness.phenomenon Haze described is measured as haziness. as relative Haze transmission is measured with as anrelative angle transmission above 2.5◦ from with the an direct angle optical above optical path. The difference can be measured using a ring detector inside the integrating sphere path.2.5° from Clarity the isdirect defined optical as relative path. Clarity transmission is defined within as relative 2.5◦ of thetransmission direct optical within path. 2.5° The of dithefference direct (Figure 8). canoptical be measuredpath. The using difference a ring can detector be measured inside the usin integratingg a ring spheredetector (Figure inside8). the integrating sphere (Figure 8).

Figure 8. Measurement principle of a haze meter. Figure 8. Measurement principle of a haze meter.

2.3.2. Measuring of RI 2.3.2. Measuring of RI Figure 8. Measurement principle of a haze meter. The RI of the PVB sheets and the MGF was determined using an Abbemat MW® (Anton Paar®, ® ® Graz,2.3.2. MeasuringThe Austria) RI of usingthe of RIPVB a standard sheets and wavelength the MGF was of 589.3 determined nm (Na-D) using at an a fixedAbbemat temperature MW (Anton of 20.0 Paar◦C. , TheGraz, diff Austria)erence of using RI between a standard matrix wavelength and particle of 589.3 is commonly nm (Na-D) described at a fixed by temperature the ratio m, whichof 20.0 can°C. The be The RI of the PVB sheets and the MGF was determined using an Abbemat MW® (Anton Paar® , calculateddifference with of RI Equation between (10). matrix and particle is commonly described by the ratio m, which can be Graz, Austria) using a standard wavelength of 589.3 nm (Na-D) at a fixed temperature of 20.0 °C. The calculated with Equation (10). nParticle(λ) difference of RI between matrix and particlem(λ) is= commonly described by the ratio m, which can(10) be n (λ) Matrix𝑛(𝜆) calculated with Equation (10). 𝑚(𝜆) = (10) 𝑛(𝜆) 2.3.3. Morphology Characterization 𝑛(𝜆) 𝑚(𝜆) = (10) The morphology of the composite sheets was determined𝑛(𝜆) using light microscopy, as well as SEM 2.3.3. Morphology Characterization (Zeiss Leo 1530, Carl Zeiss Microscopy GmbH, Jena, Germany). To prepare the samples, 2.3.3. Morphology Characterization

J.J. Compos. Compos. Sci. Sci. 2019 2019, ,3 3, ,101 101 88 of of 17 17 J. Compos. Sci. 2019, 3, 101 8 of 17 TheThe morphology morphology of of the the composite composite sheets sheets was was determi determinedned using using light light microscopy, microscopy, as as well well as as SEM SEM imagingimaging (Zeiss(Zeiss LeoLeo 1530,1530, CarlCarl ZeissZeiss MicroscopyMicroscopy GmbH,GmbH, Jena,Jena, Germany).Germany). ToTo prepareprepare thethe samples,samples, theythey werewere cooledcooled downdown inin liquidliquid nitrogennitrogen andand thenthen brokenbroken toto receivereceive aa freshfresh areaarea ofofof fracturefracturefracture andandand cross-sectioncross-section ofof thethethe samplessamplessamples without withoutwithout any anyany preparation preparationpreparation artifacts. artifacts.artifacts. 2.3.4. Dynamic Mechanic Analysis (DMA) 2.3.4.2.3.4. DynamicDynamic MechanicMechanic AnalysisAnalysis (DMA)(DMA) Polymers deform either elastic, viscous or, when only a small deformation is applied, visco-elastic PolymersPolymers deformdeform eithereither elastic,elastic, viscousviscous or,or, whenwhen onlyonly aa smallsmall deformationdeformation isis applied,applied, visco-visco- as a mixture of both. is defined as a deformation that is completely reversible with the elasticelastic asas aa mixturemixture ofof both.both. ElasticityElasticity isis defineddefined asas aa deformationdeformation thatthat isis completelycompletely reversiblereversible withwith Elastic Modulus (Young’s Modulus) as a fixed ratio of applied strain and (Hook’s law). During thethe ElasticElastic ModulusModulus (Young’s(Young’s Modulus)Modulus) asas aa fixedfixed ratioratio ofof appliedapplied strainstrain andand stressstress (Hook’s(Hook’s law).law). viscous deformation, the polymer behaves more like a with the stress depending not only on the DuringDuring viscousviscous deformation,deformation, thethe popolymerlymer behavesbehaves moremore likelike aa liquiliquidd withwith thethe stressstress dependingdepending notnot applied strain but also on the speed of the applied strain (strain rate). Visco-elastic behavior, therefore, onlyonly onon thethe appliedapplied strainstrain butbut alsoalso onon thethe speedspeed ofof thethe appliedapplied strainstrain (strain(strain rate).rate). Visco-elasticVisco-elastic describes reversible deformation that is strain-rate- and thus time-dependent [32]. The stress–strain behavior,behavior, therefore,therefore, describesdescribes reversiblereversible deformatideformationon thatthat isis strain-rate-strain-rate- andand thusthus time-dependenttime-dependent curves of such a material usually show a hysteresis loop (Figure9). [32].[32]. TheThe stress–strainstress–strain curvescurves ofof suchsuch aa materialmaterial usuallyusually showshow aa hysteresishysteresis looploop (Figure(Figure 9).9).

(a)(a) (b)(b)

FigureFigure 9.9.9. Stress–strainStress–strain curvescurves ofof aa totaltotal elasticelastic ((aa)) andandand visco-elasticvisco-elasticvisco-elastic ((bb)) materialmaterialmaterial withwithwith hysteresishysteresishysteresis andandand time-dependenttime-dependent behavior.behavior. behavior.

TheThe dynamicdynamic mechanicalmechanical propertiesproperties ofof thethe compcompositecompositeosite werewere determineddetermined withwith aa TexasTexas InstrumentInstrument DMADMA 29802980 (Dallas,(Dallas, TX,TX, USA)USA) DynamicDynamicDynamic MechanicMechanicMechanic AnalyzerAnalyzerAnalyzer setup.setup.setup. UsingUsing this,this, aa slightslight deformationdeformation isis appliedapplied toto thethe samplesample andand thethe correspondingcorresponcorrespondingding forceforce is isis measured measuredmeasured (Figure (Figure(Figure 10 10).10).).

((aa)) ((bb))

FigureFigure 10.10.10. Stress–strainStress–strain behaviorbehavior of of elasticelastic ((aa)) andand visco-elasticvisco-elasticvisco-elastic ((bb)) materialsmaterialsmaterials withwith ananan appliedappliedapplied sinussinussinus deformationdeformation and and a a givengiven frequency.frequency. The The delay delay inin thethe measuredmeasured stressstress is is determineddetermined by by thethe phasephase shiftshift δδ [[32]. 32[32].].

TheThe raterate ofof thethe appliedapplied deformationdeformation can can be be adjusted adjusted as as deformations deformations per per second second or or frequency. frequency. UsingUsing thethe samplesample ,geometry, frequency-dependentfrequency-dependentfrequency-dependent stress–strainstrestress–strainss–strain behaviorbehavior cancan bebe measuredmeasured withwith thethe followingfollowing equationsequationsequations [ [33].[33].33].

J.J. Compos. Compos. Sci. Sci. 20192019,, 33,, 101 101 9 of 17

𝜎 |𝐸∗| 𝐸𝜔 𝐸𝜔. q (11) σ 𝜀 A = = [ ( )]2 + [ ( )]2 ∗ E∗ E0 ω E00 ω . (11) |𝐸 | is the complex modulus,ε A𝐸 the| |storage modulus and 𝐸 is the loss modulus.

E is the complex modulus, E the storage∗ modulus and E00 is the loss modulus. | ∗| 0 𝐸 𝜔 𝐸𝜔 𝑖𝐸 𝜔. (12) ( ) = ( ) + ( ) E∗ ω 𝐸 E0 |𝐸ω∗| ∙𝑐𝑜𝑠iE00 ω 𝛿 .(13) (12)

E0= E∗ ∗cos δ (13) 𝐸 | |𝐸|· | ∙sin𝛿 (14) E00 = E∗ sin δ (14) | |·𝐸𝜔 tan 𝛿E 00 (ω) . (15) tan δ = 𝐸 𝜔. (15) E0(ω) Using the complex modulus and the shift δ, the storage modulus (Equation (13)) and the loss modulusUsing the (Equation complex modulus(14)) can andbe calculated. the phase shiftThe δratio, the of storage loss to modulus storage (Equationmodulus is, (13)) therefore, and the describedloss modulus by tan (Equation δ. (14)) can be calculated. The ratio of loss to storage modulus is, therefore, described by tan δ. For this study, the was on elastic properties at room temperature, as well as on Tg of the T composite.For this The study, experimental the focus para wasmeters on elastic are propertiesshown in Table at room 3. temperature, as well as on g of the composite. The experimental parameters are shown in Table3. Table 3. Measurement parameters for Dynamic Mechanical Analysis (DMA). Table 3. Measurement parameters for Dynamic Mechanical Analysis (DMA). Measurement Parameter Symbol Settings Measurement Parameter Symbol Settings Sample length 𝑙 25 mm Sample length l 25 mm Sample length (clamping) 𝑙0 17.5 mm Sample length (clamping) lE 17.5 mm Sample width 𝑏 5 mm Sample width b0 5 mm Sample thickness 𝑑 0.5-0.7 mm Sample thickness d0 0.5–0.7 mm StrainStrain ε𝜀A 0.000850.00085 AmplitudeAmplitude ∆𝑙∆l 15 µm15 µ m FrequencyFrequency 𝑓f 1 Hz1 Hz Temperature range C 50–100 C Temperature range °C◦ -50–100− °C ◦ Heating/cooling rate 1 K/min Heating/cooling rate 1 K/min

3. Results and Discussion 3. Results and Discussion 3.1. Morphology of Composite Samples 3.1. Morphology of Composite Samples To determine the morphology and orientation of the MGF within the PVB matrix, samples To determine the morphology and orientation of the MGF within the PVB matrix, samples with with 9 vol %, 13 vol %, 16 vol %, 20 vol %, 23 vol % and 26 vol % of GF100 (d50 = 160 µm) and 9 vol %, 13 vol %, 16 vol %, 20 vol %, 23 vol % and 26 vol % of GF100 (d50 = 160 µm) and GF001 (d50 = GF001 (d50 = 30 µm) were produced. Their cross-sections were investigated with a light 30 µm) were produced. Their cross-sections were investigated with a light microscope (Figure 11) (Figure 11).

(a) (b) (c)

Figure 11. Cross-section of of PVB–MGF composites composites with with ( (aa)) 16, 16, ( (bb)) 23 23 and and ( (cc)) 26 26 vol % of GF100 glass flakesflakes and an average thickness of y ~120–150 µm.µm. The The alignment alignment of of the the flakes flakes parallel to the sample surface is increased with volume content.

J. Compos. Sci. 2019, 3, 101 10 of 17

As it can be clearly seen, the alignment of the glass flakes parallel to the surface is increased with increasing content of glass flakes. At the same time, this high degree of ordering does not lead to any

J.significant Compos. Sci. breakage2019, 3, 101 or any kind of failure of the interface between the matrix and glass surface10 with of 17 increasing volume content. J. Compos. Sci. 2019, 3, 101 10 of 17 Figure 12 shows a comparison of samples with the same volume content (16 vol %) but using glassAs flakes it can with be clearly different seen, mean the alignmentdiameters of the160 glassµm and flakes 30 parallelµm. A clear to the difference surface is in increased the degree with of As it can be clearly seen, the alignment of the glass flakes parallel to the surface is increased with increasingordering can content be seen. of glassOn average, flakes. Atthe the glass same flakes time, wi thisth the high bigger degree diameter of ordering show a does stronger not lead alignment to any increasing content of glass flakes. At the same time, this high degree of ordering does not lead to any significantparallel to breakagethe surface, or while any kind the ofsmaller failure glass of the flakes interface do not between show a theclear matrix alignment and glass or ordering. surface with significant breakage or any kind of failure of the interface between the matrix and glass surface with increasingUsing volume SEM imaging, content. the interface of glass and polymer was investigated. Although the increasing volume content. preparationFigure 12 was shows carefully a comparison carried out of using samples cryo-fracture with the same of the volume sample, content (16 vol artifacts %) but usingcould Figure 12 shows a comparison of samples with the same volume content (16 vol %) but using glassnot be flakes prevented, with di ffaserent displayed mean in Figure of 13. 160 Howeµm andver, 30it µcanm. Abe clear seen di thfferenceat the inglass the flakes degree are of glass flakes with different mean diameters of 160 µm and 30 µm. A clear difference in the degree of orderingcompletely can covered be seen. and On average,surrounded the glassby the flakes polymer with matrix. the bigger It can diameter be assumed showa that stronger the flakes alignment were ordering can be seen. On average, the glass flakes with the bigger diameter show a stronger alignment parallelwell dispersed to the surface, and homogeneously while the smaller distributed glass flakes in the do slurry not show with a a clear good alignment wetting behavior. or ordering. parallel to the surface, while the smaller glass flakes do not show a clear alignment or ordering. Using SEM imaging, the interface of glass and polymer was investigated. Although the preparation was carefully carried out using cryo-fracture of the sample, delamination artifacts could not be prevented, as displayed in Figure 13. However, it can be seen that the glass flakes are completely covered and surrounded by the polymer matrix. It can be assumed that the flakes were well dispersed and homogeneously distributed in the slurry with a good wetting behavior.

(a) (b)

Figure 12. Cross-section of PVB–MGF composites with 16 vol % of GF100 (d50 = 160 µm) (a) and 16 vol Figure 12. Cross-section of PVB–MGF composites with 16 vol % of GF100 (d50 = 160 µm) (a) and 16 vol 50 %% ofof GF001GF001 (d(d50 == 30 µµm)m) ((b)) glassglass flakes.flakes.

Using SEM imaging, the interface of glass and polymer was investigated. Although the preparation was carefully carried out using(a) cryo-fracture of the sample, delamination artifacts(b) could not be prevented, as displayed in Figure 13. However, it can be seen that the glass flakes are completely covered and Figure 12. Cross-section of PVB–MGF composites with 16 vol % of GF100 (d50 = 160 µm) (a) and 16 vol surrounded by the polymer matrix. It can be assumed that the flakes were well dispersed and % of GF001 (d50 = 30 µm) (b) glass flakes. homogeneously distributed in the slurry with a good wetting behavior.

(a) (b)

Figure 13. SEM (SEI) image using of PVB–MGF composites with 23 vol % (a) and 16 vol % (b) of

GF100 (d50 = 160 µm) with slight delamination and preparation artefacts.

(a) (b)

FigureFigure 13.13. SEMSEM (SEI)(SEI) image image using using of of PVB–MGF PVB–MGF composites composites with with 23 vol 23 %vol (a )% and (a) 16 and vol 16 % (volb) of% GF100(b) of GF100 (d50 = 160 µm) with slight delamination and preparation artefacts. (d50 = 160 µm) with slight delamination and preparation artefacts.

J. Compos. Sci. 2019, 3, 101 11 of 17

3.2. Optical Properties

3.2.1. Influence of Volume Content of MGF To determine the optical properties, three samples of each composition were produced and laminated and each sample was measured five times at slightly different positions on the surface. J.Figure Compos. Sci.14 2019shows, 3, 101 the results of glass laminates with a PVB–MGF interlayer containing different11 of 17 volume fractions of MGF as measured with the haze meter. The transmission of the Optiwhite™ soda-lime glass was measured at 93.5% which lies well within the typical range of flat glass of 92%– 3.2. Optical Properties 94%. The loss of transmission can be accounted for by on the surfaces of the glass pane and 3.2.1.can be Influence calculated of Volumewith a simplified Content of Fresnel MGF equation (Equation (16)). For glass (n = 1.5) and air (n = 1), this equals approximately 96% transmission for one surface and 92% for two surfaces (front and back).To determine the optical properties, three samples of each composition were produced and laminated and each sample was measured five times at slightly different positions on the surface. 4𝑛 ∙𝑛 Figure 14 shows the results of glass laminates𝑇= with a PVB–MGF. interlayer containing different volume(16) ( ) fractions of MGF as measured with the haze meter.𝑛 +𝑛 The transmission of the Optiwhite™ soda-lime glass wasAll other measured losses at in 93.5% transmission which lies can well be attributed within the to typical absorption, range scattering of flat glass or of reflection 92–94%. of The the lossparticles of transmission in the interlayer. can be accountedThe transmission for by reflection is reduced on thewith surfaces increasing of the volume glass panecontent and of can MGF. be calculatedHowever, with the slope a simplified of the reduction Fresnel equation is not linear, (Equation as the (16)). drop Forfrom glass 0 vol (n %= 1.5)to 9 andvol % air is ( nroughly= 1), this 1% equalsin transmission, approximately while 96% it further transmission only changes for one by surface another and 0.5% 92% at for 26 two vol surfaces%. A similar (front behavior and back). can be seen with the results of the haze measurement (Figure 14b). There is a steep increase of 4% of haze 4n n between pure PVB and 9 vol %, while itT does= notAir exceedGlass 9%. total at 26 vol % of MGF. (16) · 2 (nAir + nGlass)

(a) (b)

FigureFigure 14. 14.Transmission Transmission ( a()a) and and haze haze ( b(b) in) in relation relation to to volume volume content content and and size size of of MGF MGF with with GF100 GF100

(d(50d50= = 160160 µµm)m) andand GF001GF001 ((dd5050 = 30 µm).µm). Thickness Thickness of of composite composite was normalized to 200 µm.µm.

AllThis other non-linear losses in behavior transmission of the can transmission be attributed and to haze absorption, of these scattering PVB–MGF or reflectioncomposites of with the particlesrespect to in the the volume interlayer. content The contradicts transmission the is findin reducedgs of with most increasing other studie volumes that contentinvestigate of MGF. glass However,fibers or glass the slope particles of the as reduction fillers and is find not linear,the increase as the in drop haze from and 0 reduction vol % to9 transmission vol % is roughly to be 1% linear in transmission,with increasing while volume it further content. only changes It is, therefore, by another assumed 0.5% at 26that vol the %. Adifference similar behavior in behavior can be can seen be withexplained the results by the of thehigh haze aspect measurement ratio of the (Figure MGF and,14b). th Thereerefore, is a high steep ordering increase of 4%the offlakes haze parallel between to purethe surface, PVB and which 9 vol %,reduces while the it does effect not of exceedhaze caused 9% total by atan 26 RI vol mismatch. % of MGF. This non-linear behavior of the transmission and haze of these PVB–MGF composites with respect to3.2.2. the volumeInfluence content of Glass contradicts Flakes’ Size the and findings Ordering of most other studies that investigate glass fibers or glass particles as fillers and find the increase in haze and reduction transmission to be linear with To further evaluate the effect of ordering of the MGF in the PVB matrix, measurements of a increasing volume content. It is, therefore, assumed that the difference in behavior can be explained sample with 16 vol % of GF001 (d50 = 30 µm) were included in the comparison (Figure 14). While the by the high aspect ratio of the MGF and, therefore, high ordering of the flakes parallel to the surface, transmission does not seem to be influenced by the change of mean diameter of the MGF, the haze which reduces the effect of haze caused by an RI mismatch. doubles from 8.1% using GF100 to 18.7% with GF001 at 16 vol %. 3.2.2. Influence of Glass Flakes’ Size and Ordering

To further evaluate the effect of ordering of the MGF in the PVB matrix, measurements of a sample with 16 vol % of GF001 (d50 = 30 µm) were included in the comparison (Figure 14). While the transmission does not seem to be influenced by the change of mean diameter of the MGF, the haze doubles from 8.1% using GF100 to 18.7% with GF001 at 16 vol %. J. Compos. Sci. 2019, 3, 101 12 of 17 J. Compos. Sci. 2019, 3, 101 12 of 17

TheseThese findings findings were were compared compared to a studyto a bystudy Wildner by Wildner et al. They et investigatedal. They investigated the optical propertiesthe optical ofproperties irregularly of shaped irregularly glass particlesshaped glass with particles a mean size with of a 79 meanµm, 143sizeµ ofm 79 and µm, 202 143µm µm in a and PMMA 202 matrixµm in a usingPMMA injection matrix molding using injection [9]. The dimoldingfference [9]. of RIThe of difference the glass particles of RI of tothe PMMA glass particles was measured to PMMA as ∆n was= 0.0047.measured Figure as 15∆n shows = 0.0047. the absoluteFigure 15 haze shows values the ofabsolute these measurements haze values of when these corrected measurements for sample when thicknesscorrected (in for this sample case, thickness 500 µm). (in The this measured case, 500 hazeµm). valuesThe measured of the glass haze particles values of were the glass found particles to be betweenwere found 30% toto 50%be between at a volume 30% fraction to 50% ofat 11 a tovolume 13 vol fraction %. There of is 11 no to clear 13 dependencevol %. There on is particleno clear size.dependence When compared on particle to GF001 size. flakesWhen withcompared a low aspectto GF001 ratio, flakes the haze with is a comparable low aspect (45%).ratio, the However, haze is whencomparable compared (45%). to GF100 However, with awhen high compared aspect ratio, to theGF100 measured with a hazehigh isaspect two- ratio, to three-times the measured higher haze at similaris two- volume to three-times fractions. higher at similar volume fractions.

FigureFigure 15. 15.Haze Haze of of MGF MGF in in PVB PVB compared compared to to results results from from the the literature literature of of irregularly irregularly shaped shaped glass glass particlesparticles in in polymethyl polymethyl methacrylate methacrylate (PMMA) (PMMA) [9 ][9] with with respect respect to to volume volume content. content. Values Values were were normalizednormalized to to 500 500µ mµm sheet sheet thickness. thickness. 3.2.3. Influence of RI Mismatch of MGF and PVB Matrix 3.2.3. Influence of RI Mismatch of MGF and PVB Matrix Mismatch of the RI of filler to a matrix is the main contributor to haze as the degree of scattering Mismatch of the RI of filler to a matrix is the main contributor to haze as the degree of scattering increases with increasing difference of the RI at an interface (Snells’ Law, Equation (17)) between increases with increasing difference of the RI at an interface (Snells’ Law, Equation (17)) between material 1 with an incident angle δ1 and material 2 with a refractive angle δ2 material 1 with an incident angle 𝛿 and material 2 with a refractive angle 𝛿

n1 𝑛sin·sin(δ1)(𝛿=)n=𝑛2 sin·sin(𝛿(δ2) ) (17) · · (17) MeasurementsMeasurements show show that that pure pure B B 30 30 H H sheets sheets have have a a lower lower RI RI than than the the commercial commercial MGF MGF GF100 GF100 andand GF001 GF001 (Table (Table4). 4). The The di differencefference in in RI RI (Figure (Figure 14 14(a))a) was wasm = m1.005 = 1.005 with with∆n ∆=n0.007. = 0.007.

Table 4. n at 589.3 nm (nD) as measured with Abbemat MW. Table 4. Refractive index n at 589.3 nm (nD) as measured with Abbemat MW. Component nD @ 20 °C Component nD @ 20 ◦C GF100/GF001 1.500 GF100/GF001 1.500 GF-V8GF-V8 1.485 1.485 SheetSheet B 30 HB 30 H 1.493 1.493 SheetSheet B 30 HB 30+ 25 H wt + 25 % 3G8wt % 3G8 1.483 1.483 SoftenerSoftener 3G8 3G8 1.444* 1.444 * **from from TMDS TMDS (not (not measured). measured).

AsAs the the RI RI of of a a polymer polymer cannot cannot be be easily easily increased, increased, we, we, therefore, therefore, chose chose to to change change the the RI RI of of the the MGFMGF by by changing changing the the glass glass composition composition to to match match that that of of pure pure PVB PVB or, or, if if not not possible, possible, to to achieve achieve a a RI RI slightlyslightly below below that that of pureof pure PVB. PVB. While While increasing increasing the RI the of PVBRI of is PVB not easily is not possible, easily possible, it can be loweredit can be bylowered at least by∆n at= 0.010least ∆ byn adding= 0.010 Softener.by adding The Softener. Softener The itself Softener has an itself even lowerhas an RI even and lower by mixing, RI and the by RImixing, of the residualthe RI of PVB–softener the residual PVB–softener is therefore determined is therefore by determined a simple rule by ofa simple mixture. rule of mixture. MGF with different compositions were produced with the Sample “GF-V8” showing the most promising properties with a RI of 1.485 (+/−0.003), which was slightly below that of pure PVB. The

J. Compos. Sci. 2019, 3, 101 13 of 17

J. Compos. Sci. 2019, 3, 101 13 of 17 MGF with different compositions were produced with the Sample “GF-V8” showing the most promisingflakes were properties milled and with sieved a RI below of 1.485 250 (+/ µm0.003), with a which mean wasdiameter slightly of 120 below µm. that The of thickness pure PVB. of Thethese − flakesflakes were was milled measured and sievedto be between below 250 2 andµm 5 with µm. a It mean should diameter be noted of 120thatµ thesem. The laboratory-produced thickness of these flakesflakes was scatter measured more in to terms be between of thickness 2 and 5andµm. RI It when should compared be noted to that commercially these laboratory-produced produced flakes flakeslike GF100 scatter or more GF001. in terms However, of thickness the aspect and ratio RI when is comparable compared to to that commercially of commercial produced flakesflakes (~100-300). like GF100Using or GF001. these However, GF-V8 MGF, the aspect samples ratio with is comparable 15 vol % MGF to that and of 0 commercialwt %, 10 wt flakes %, 20 (~100–300). wt %, 30 wt % andUsing 40 wt these% of softeners GF-V8 MGF, were samples produced with and 15 volthe %haze MGF was and measured. 0 wt %, 10 In wtthis %, set, 20 the wt %,lowest 30 wt haze % and was 40found wt % ofwith softeners a composition were produced of 20 wt and %. the To haze furthe wasr determine measured. the In thisoptimal set, the softener lowest content haze was for found index withmatching a composition of the glass of 20 flakes, wt %. Toanot furtherher set determine of samples the with optimal 15 to softener 27.5 wt content % (in for2.5 indexwt % matchingsteps) was ofproduced the glass flakes,and measured. another set All of results samples were with normaliz 15 to 27.5ed wt using % (in a 2.5linear wt %regres steps)sion was to produced 200 µm andsheet measured.thickness.All results were normalized using a linear regression to 200 µm sheet thickness. FigureFigure 16 16 shows shows the the result result of of the the haze haze measurements. measurements. The The lowest lowest haze haze of of 1.5% 1.5% was was found found at at 22.5 22.5 wtwt % % 3G8. 3G8. This This is is only only marginally marginally above above the the haze haze produced produced by by pure pure PVB PVB (Figure (Figure 14 14b)b) and and can can be be consideredconsidered as as completely completely transparent. transparent.

FigureFigure 16. 16.Haze Haze values values of of composites composites with with 15 15 vol vol % of% GF-V8of GF-V8 MGF MGF with with respect respect to di toff erentdifferent contents contents of µ softenerof softener (3G8). (3G8). All resultsAll results were were normalized normalized to 200 to 200m sheetµm sheet thickness. thickness. As stated, these low haze values cannot simply be contbuted to the RI matching itself, as slight As stated, these low haze values cannot simply be contbuted to the RI matching itself, as slight variations in the RI of the flakes (+/ 0.003) are still present. However, the combination of the platelet-like variations in the RI of the flakes −(+/−0.003) are still present. However, the combination of the platelet- structure, high degree of ordering and low RI mismatch result in these highly transparent composites. like structure, high degree of ordering and low RI mismatch result in these highly transparent 3.3.composites. Dynamic Mechanical Analysis (DMA)

3.3.1.3.3. Dynamic Influence Mechanical of Volume Analysis Content (DMA) on MODULUS

3.3.1.Figure Influence 17 shows of Volume the DMA Content results on of MODULUS PVB B 30 H with 0, 5, 10 and 15 vol % of GF100 flakes. Pure PVB shows a storage modulus of about 2 GPa at room temperature when measured with 1 Hz, which increasesFigure to 3.5 17 GPashows at the 5 vol DMA %, 5.5results GPa of at PVB 10 vol B 30 % H and with 7.5 0, GPa 5, 10 at and 15 vol15 vol %. % The of storageGF100 flakes. modulus Pure decreasesPVB shows with a storage increasing modulus temperature, of about with 2 GPa a drop at room at around temperature 60 ◦C. Atwhen the measured same time, with tan δ1 increasesHz, which withincreases increasing to 3.5 temperatures—starting GPa at 5 vol %, 5.5 GPa at 60 at◦ 10C withvol % a maximumand 7.5 GPa at 76at ◦15C. vol This %. marks The storage the point modulus of the transitiondecreases between with increasing elastic and temperature, viscoelastic with behavior, a drop whichat around usually 60 °C. happens At the same at temperatures time, tan δ increases around thewith glass increasing transformation temperatures—starting point (Tg). In the at case60 °C of wi Bth 30 a H,maximum the glass at transformation 76 °C. This marks point the at point 68 ◦C of isthe 8 K transition below the between point of theelastic maximum and viscoelastic of tan δ. Forbehavior, simplification, which usually the point happens of tan δatmaximum temperatures is, therefore,around the considered glass transformation as Tg. Therefore, point with (Tg increasing). In the case an of volume B 30 H, content the glass of GF100, transformationTg does not point change. at 68 However,°C is 8 K thebelow maximum the point of of tan theδ changesmaximum from of tan 2.0 δ at. For 0 vol simplification, % to 0.8 at 15 the vol point %. This of tan can δ be maximum explained is, bytherefore, a higher sticonsideredffness and, as therefore, Tg. Therefore, withof increasing the composite an volume with increasing content of filler GF100, content. Tg does not change. However, the maximum of tan δ changes from 2.0 at 0 vol % to 0.8 at 15 vol %. This can be explained by a higher stiffness and, therefore, brittleness of the composite with increasing filler content.

J. Compos.J. Compos. Sci. 2019 Sci., 32019, 101, 3, 101 1414 of of 17 17 J. Compos. Sci. 2019, 3, 101 14 of 17

FigureFigure 17. Temperature-dependent 17. Temperature-dependent storage storag moduluse modulus E’ ofE’ PVBof PVB filled filled with with MGF MGF of of type type GF100. GF100. Figure 17. Temperature-dependent storage modulus E’ of PVB filled with MGF of type GF100. 3.3.2. Influence of Glass Flake Size on Modulus 3.3.2. Influence of Glass Flake Size on Modulus 3.3.2. Influence of Glass Flake Size on Modulus FigureFigure 18 shows 18 shows the storagethe storage modulus modulus E’ with E’ with diff differenterent volume volume fractions fractions of of MGF MGF of of types types GF100 GF100 and GF001atandFigure GF001at 2518 ◦showsC. 25 With°C. the With increasing storage increasi modulus volumeng volume E’ contentwith content different of theof thevolume MGF, MGF, the fractions the storage storage of modulusMGF modulus of types increases increases GF100 asas well.and However,well. GF001at However, the25 °C. increase the With increase inincreasi sti ffinness stiffnessng volume using using high-aspect content high-aspect of GF100the MGF,GF100 is aboutthe is aboutstorage two-times two-times modulus higher higher increases than than that as that of lower-aspectwell.of lower-aspectHowever, and smallerthe and increase smaller MGF in of stiffnessMGF type of GF001. typeusing GF001. high-aspect GF100 is about two-times higher than that of lower-aspect and smaller MGF of type GF001.

FigureFigure 18. Storage 18. Storage modulus modulus E’ at E’ 25 at◦ C25 with °C with diff erentdifferent volume volume fractions fractions of GF100of GF100 (d 50(d50= =160 160µ µm)m) and and Figure 18. Storage modulus E’ at 25 °C with different volume fractions of GF100 (d50 = 160 µm) and GF001GF001 (d50 = (d3050 µ= m).30 µm). GF001 (d50 = 30 µm). This canThis be can explained be explained using ausing model a ofmodel ideal serialof ideal and serial parallel and connections parallel connections of materials of sti materialsffnesses This can be explained using a model of ideal serial and parallel connections of materials usingstiffnesses Equations using (18) andEquations (19) [ 34(18)–36 and] with (19)E [34MGF−36]and withEMatrix 𝐸as and the 𝐸 Young’s as Modulusthe Young’s of Modulus the matrix of stiffnesses using Equations (18) and (19) [34−36] with 𝐸 and 𝐸 as the Young’s Modulus of materialthe andmatrix the material MGF, respectively, and the MGF, and respectively,uMGF as the volumeand 𝑢 fraction as the of volume MGF in fraction the matrix. of MGF in the thematrix. matrix material and the MGF, respectively, and 𝑢 as the volume fraction of MGF in the matrix. Ideal Parallel Connection Ec,parallel = uMGFEMGF + (1 uMGF)EMatrix. (18) Ideal Parallel Connection 𝐸, = 𝑢𝐸 +(1−𝑢− )𝐸. (18) Ideal Parallel Connection 𝐸, = 𝑢𝐸 +(1−𝑢)𝐸. (18) EMatrixEMGF Ideal Serial Connection E = (19) Ideal Serial Connectionc,serial 𝐸, = (19) uMGFEMatrix+(1 (uMGF)E)MGF Ideal Serial Connection 𝐸, = − (19) () Both ideal cases were included in Figure 18. As MGF have a finite size, they can be represented by a mixture of both serial and parallel connection in respect of the direction of the tensile force (in-plane). J. Compos. Sci. 2019, 3, 101 15 of 17

Both ideal cases were included in Figure 18. As MGF have a finite size, they can be represented by a mixture of both serial and parallel connection in respect of the direction of the tensile force (in- plane). Using this model, it can be explained why shorter flakes with lower aspect ratio show a behavior that is more similar to ideal serial connection, while longer flakes with higher aspect ratio show a behavior more towards ideal parallel connection. The effect of ordering thus cannot be easily observed. However, when again comparing to results from the literature, GF100 flakes show a nearly J. Compos.two-fold Sci. higher2019, 3, 101stiffness with volume content when compared to glass particles with similar mean15 of 17 diameter [9]. Using this model, it can be explained why shorter flakes with lower aspect ratio show a behavior 3.3.3. Influence of Softener on Modulus and Glass Transition of PVB (Tg) that is more similar to ideal serial connection, while longer flakes with higher aspect ratio show a behaviorAs more stated towards above, softener ideal parallel was used connection. in this study The emainlyffect of to ordering adjust the thus RI cannotof the polymer be easily to observed. match However,the RI of when the MGF. again The comparing main application to results fromof soften theer literature, is usually GF100 the decrease flakes show of stiffness a nearly and two-fold the higher“softening” stiffness of with the volumepolymer. content Figure when 19a shows compared the storage to glass modulus particles at with 25 similar°C as a mean function diameter of the [ 9]. softener and volume content of GF100 MGF. It can be seen that by adding 25 wt % of softener, the 3.3.3.storage Influence modulus of Softener is decreased on Modulus by two-orders and Glass of magnitude. Transition of PVB (Tg) At the same time, the stiffening effect of MGF is comparable in terms of relative values. As such E’ Asat 0 stated vol % above,GF100 softenerand 25 wt was % 3G8 used is in around this study 8 MPa, mainly while to it adjust is increased the RI ofto theabout polymer 40 MPa to at match 15 vol the RI% of GF100 the MGF. and The 25 wt main % 3G8. application Within ofthe softener margin isof usually error, this the five-fold decrease increase of stiffness is comparable and the “softening” to the offour-fold the polymer. increase Figure when 19 ano shows softener the is storage added. modulusThat means at 25that◦C using as a functionMGF in combination of the softener with and volumesoftener, content the storage of GF100 modulus MGF. Itof canthe beresulting seen that comp byosite adding can 25be wtadjusted % of softener, between the8 MPa storage and 8 modulus GPa. is decreased by two-orders of magnitude.

(a) (b)

FigureFigure 19. 19.Storage Storage modulus modulus E’ E’ in in respectrespect toto content of GF100 GF100 and and softener softener 3G8 3G8 (a ()a and) and glass glass transition transition Tg Tofg of PVB PVB in in respect respect to to softener softener ( b(b).).

AtAdding the same softener time, theto a stipolymerffening also effect changes of MGF the is glass comparable transition in temperature, terms of relative as shown values. in Figure As such E’19(b). at 0 vol Tg is % moved GF100 from and 76 25 °C wt to % room 3G8 temperature is around 8 MPa,(25 °C) while when itsoftener is increased is added to aboutfrom 0 40to 25 MPa wt at%. 15 volThis % GF100 is especially and 25 important wt % 3G8. when Within visco-elastic the margin behavior of error, is preferable this five-fold to the increaseapplication. is comparable to the four-fold increase when no softener is added. That means that using MGF in combination with 4. Summary softener, the storage modulus of the resulting composite can be adjusted between 8 MPa and 8 GPa. AddingThe results softener show to athat polymer it was alsopossible changes to produce the glass a transitiontransparent temperature, composite of as PVB shown with in different Figure 19 b. Tg volumeis moved contents from 76 of◦C MGF to room and temperaturesoftener using (25 a◦ C)slurry when process softener combined is added with from blade 0 to 25 casting. wt %. ThisThe is especiallystorage modulus important of whensuch composites visco-elastic increases behavior with is preferable the volume to content the application. of MGF and decreases with the addition of softener. MGF with higher mean diameter and, therefore, higher aspect ratio at a fixed 4. Summarythickness show a stronger stiffening behavior when compared to MGF with smaller diameter and, therefore, smaller aspect ratio. The results show that it was possible to produce a transparent composite of PVB with different The composites show increasing haze with the addition of MGF. The increase was found to be volume contents of MGF and softener using a slurry process combined with blade casting. The storage non-linear when adding MGF with a high aspect ratio. In comparison to irregular glass particles and modulusMGF with of such small composites aspect ratio, increases the haze with was the found volume to content be significantly of MGF and lower. decreases With the with addition the addition of of softener. MGF with higher mean diameter and, therefore, higher aspect ratio at a fixed thickness show a stronger stiffening behavior when compared to MGF with smaller diameter and, therefore, smaller aspect ratio. The composites show increasing haze with the addition of MGF. The increase was found to be non-linear when adding MGF with a high aspect ratio. In comparison to irregular glass particles and MGF with small aspect ratio, the haze was found to be significantly lower. With the addition of softener, the RIs of such composites can be matched, resulting in a residual haze below 2%, which is considered as highly transparent. J. Compos. Sci. 2019, 3, 101 16 of 17

It should be noted that using the above-stated approach, the mechanical properties due to the addition of softener cannot be independently changed without also influencing the RI mismatch and thus haze and vice versa. By further fine-tuning the glass composition to match the RI closer to 1.493, it is possible to achieve high stiffness and at the same time, high transparency, as explained above. Considering the possible industrial application of this method, the limiting factors are the size of the blade-casting setup and drying oven. However, combining smaller patches, composites of up to 0.5 m2 have already been produced. Instead of blade casting, printing the slurry on a substrate could, in theory, greatly increase the individual size of composite sheets and speed of production. A disadvantage of the current process for scaling is the roughness and waviness of the composite sheets after drying, which require an additional hot pressing step with a smooth surface or lamination, as shown in this study, to achieve high transmission and low haze as the surface properties greatly influence the optical properties.

Author Contributions: B.S. performed the experiments and wrote the paper. T.G., S.L. and D.B. provided guidance and helped in manuscript preparation. Funding: Financial support by the European Union project HarWin (FP7-CP-FP, Grant number: 314653) is gratefully acknowledged. Acknowledgments: The presented work is dedicated to Monika Willert-Porada, who passed away on 11 December 2016. She initiated the above-mentioned project. In addition, the authors would like to thank Angelika Kreis and Ingrid Otto for laboratory support; Catrin Chow, Mario Mösch, Tobias Schindhelm, Matthias Wurm, Jens Böckler, Nicole Bayer and Christoph Seifert for their support preparing the samples as well as Manfred Arnold for his expert input on the lamination process. This publication was funded by the German Research Foundation (DFG) and the University of Bayreuth in the funding program Open Access Publishing. Conflicts of Interest: The authors declare no conflicts of interest.

References

1. Bäumer, S. Handbook of , 2nd ed.; Wiley-VCH Verlag GmbH & Co. KGaA: Weinheim, Germany, 2010; ISBN 978-352-740-940-2. 2. Sastri, V.R. in Medical Devices: Properties, Requirements and Applications, 2nd ed.; William Andrew: New York, NY, USA, 2014; pp. 73–120, ISBN 978-145-573-201-2. 3. Willert-Porada, M. Harvesting Solar Energy with Multifunctional Glass-Polymer Windows. Available online: https://cordis.europa.eu/project/rcn/104545/reporting/en (accessed on 18 November 2019). 4. Iba, H.; Chang, T.; Kagawa, Y. Optically Transparent Continuous Glass Fibre-Reinforced Epoxy Matrix Composite: Fabrication, Optical and Mechanical Properties. Compos. Sci. Technol. 2002, 62, 2043–2052. [CrossRef] 5. Olson, J.R.; Day, D.E.; Stoffer, J.O. Fabrication and Mechanical Properties of an Optically Transparent Glass /. J. Compos. Mater. 1992, 26, 1181–1192. [CrossRef] 6. Khanna, S.K.; Ellingsen, M.D.; Winter, R.M. Investigation of Fracture in Transparent Reinforced Polymer Composites Using . ASME J. Eng. Mater. Technol. 2004, 126, 1–7. [CrossRef] 7. Moriwaki, T. Mechanical Property Enhancement of Glass Fibre-Reinforced Composite Made by Direct Process. Compos. Part A 1996, 27, 379–384. [CrossRef] 8. Zhu, H.; Khanna, S.K. A Novel Transparent Glass Fiber-Reinforced Polymer Composite Interlayer for Blast-Resistant Windows. J. Eng. Mater. Technol. 2016, 138.[CrossRef] 9. Wildner, W.; Drummer, D. The Mechanical and Optical Properties of Injection-moulded PMMA, Filled with Glass Particles of a Matching Refractive Index. Polym. Polym. Compos. 2017, 25, 453–462. [CrossRef] 10. Hashemi, S.; Din, K.J.; Low, P. Fracture behavior of glass bead-filled poly(oxymethylene) injection moldings. Polym. Eng. Sci. 1996, 36, 1807–1820. [CrossRef] 11. Lee, J.; Yee, A.F. Fracture of glass bead/epoxy composites: On micro-mechanical deformations. Polymer 2000, 41, 8363–8373. [CrossRef] 12. Naganuma, T.; Iba, H.; Kagawa, Y. Optothermal properties of glass particle-dispersed epoxy matrix composite. J. Mater. Sci. Lett. 1999, 18, 1587–1589. [CrossRef] 13. Kagawa, Y.; Iba, H.; Tanaka, M. Fabrication and optical/thermal properties of glass particle-epoxy optically transparent composites. Acta Mater. 1998, 46, 265–271. [CrossRef] J. Compos. Sci. 2019, 3, 101 17 of 17

14. Ajayan, P.M.; Schadler, L.S.; Braun, P.V. Nanocomposite and Technology; Wiley-VCH: Weinheim, Germany, 2003; ISBN 978-352-730-359-5. 15. Wildner, W.; Drummer, D. A fiber optic temperature sensor based on the combination of epoxy and glass particles with different thermo-optic coefficients. Photonic Sens. 2016, 6, 295–302. [CrossRef] 16. Wildner, W.; Drummer, D. A fiber optic temperature sensor based on the combination of two materials with different thermo-optic coefficients. IEEE Sens. J. 2015, 16, 688–692. [CrossRef] 17. Bach, H.; Neuroth, N. The Properties of Optical Glass; Springer: Berlin/Heidelberg, Germany, 1998; ISBN 978-364-257-769-7. 18. Zhang, Z.; Zhao, P.; Lin, P.; , F. Thermo-optic coefficients of polymers for optical applications. Polymer 2006, 47, 4893–4896. [CrossRef] 19. Breuer, H.; Grzesitza, J. Trübungserscheinungen in zweiphasigen Polymersystemen (glasfaserverstärkte Polymere). Die Angew. Makromol. Chem. 1975, 45, 1–19. [CrossRef] 20. Naganuma, T.; Kagawa, Y. Effect of particle size on light transmittance of glass particle dispersed epoxy matrix optical composites. Acta Mater. 1999, 47, 4321–4327. [CrossRef] 21. Wildner, W.; Drummer, D. Light scattering of glass-particle filled matrices with similar refractive index. J. Compos. Mater. 2018, 52, 4231–4246. [CrossRef] 22. Kyrgyzbaev, K.; , A.; Willert-Porada, M. Influence of temperature on the thickness of ultrathin particulate glass platelets. Eur. J. Glass Sci. Technol. Part A 2016, 57, 95–100. [CrossRef] 23. Kyrgyzbaev, K. Simulationsbasierte Prozessentwicklung zur Herstellung von Submikrometer-Glaspartikeln. Ph.D. Thesis, University of Bayreuth, Bayreuth, Germany, 25 November 2017. 24. ECKART EFFECT Pigments. Available online: https://www.eckart.net/markets/cosmetics/products.html (accessed on 30 June 2019). 25. González-Guzmán, J.; Santana, J.J.; González, S.; Souto, R.M. Resistance of metallic substrates protected by an organic containing glass flakes. Prog. Org. Coat. 2010, 68, 240–243. [CrossRef] 26. ECR Glassflakes Product Page. Available online: https://www.glassflake.com/products/ecr-glass-flake/ (accessed on 30 June 2019). 27. McKeen, L.W. Permeability Properties of Plastics and , 4th ed.; William Andrew: New York, NY, USA, 2017; ISBN 978-032-350-859-9. 28. Mowital®Technical Data Sheet. Available online: https://www.mowital.com/fileadmin/user_upload/ MOWITAL/downloads/technical_data_sheets/190410_TDS_Mowital_B_DE.pdf (accessed on 16 May 2019). 29. Chembook Triethylene Glycol Bis(2-Ethylhexanoate). Available online: https://www.chemicalbook.com/ CASEN_94-28-0.htm (accessed on 15 January 2019). 30. International Organization for Standardization. Plastics—Determination of the Total Luminous Transmittance of Transparent Materials—Part 1: Single-Beam Instrument; ISO 13468-1:2019; International Organization for Standardization: Genf, Switzerland, 2019. 31. International Organization for Standardization. Plastics—Determination of Haze for Transparent Materials; ISO 14782:1999; International Organization for Standardization: Genf, Switzerland, 1999. 32. Ehrenstein, G.W. Polymer-Werkstoffe, 2nd ed.; Carl Hanser Verlag: München, Germany, 1999; ISBN 978-344-621-162-1. 33. Ehrenstein, G.W.; Riedel, G.; Trawiel, P. Praxis Der Thermischen Analyse Von Kunststoffen, 2nd ed.; Carl Hanser Verlag: München, Germany, 2003; ISBN 978-344-622-340-0. 34. Thomason, J.L.; Vlug, M.A. The influence of fibre length and concentration on the properties of glass fibre reinforced : 1. Tensile and flexural modulus. Compos. Part A 1996, 27, 477–484. [CrossRef] 35. Thomason, J.L.; Groenewoud, W.M. The influence of fibre length and concentration on the properties of glass fibre reinforced polypropylene: 2. Thermal properties. Compos. Part A 1996, 27, 555–565. [CrossRef] 36. Thomason, J.L. The Influence of fibre length and concentration on the properties of glass fibre reinforced polypropylene: 5. Injection moulded long and short fibre PP. Compos. Part A 2002, 33, 1641–1652. [CrossRef]

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).