International Journal of Applied Glass Science 3/2/ 107-121(2012) pplied glass sC丨ENCE Strength of High Performance Glass Reinforcement Fib lichelle L. Korwin-Edson, Douglas A. Hofmann, and Peter B. McGinnis Owens Corning Science d Technology, 2790 Columbus Rd, Granville, Ohio The practical strength of glass is highly dependent on the amount and type of damage that a glass article has experienced in its lifetime and can be 50% less than its theoretical strength. Glass reinforcement fibers in the pristine state exhibit some of the highest failure strengths of any glass form. Strength degradation is a sequential process the further from the point of formation a glass travels. Individual filament strength is important in the manufacturing process as the fiber interacts with water,HVAC, sizing applicators, contact shoes, and guide eyes and ultimately this combination impacts productivity. a dis. cussion of glass fiber strength- pristine versus usable, and the effects of temperature, humidity, and composition on glass Length follows in this manuscript. New data collected in Owens Cornings Glass Properties Laboratory on the effect of tem- erature and relative humidity on strength and modulus for Advantex glass, Owens Cornings S-glass (XStrand"S, Flite- Strand"S, and ShieldStrand"S)and H-glass (WindStrandH) are presented. Owens Cornings understanding of the effect of composition on strength and modulus, and particularly how individual oxides contribute to these properties are shared. Introduction al damage and can often be caused only by chemical damage or the presence of water near strained bonds. To It is well known in the field that the practical understand the effects of composition, temperature or strength of a given glass is dependent on that glass arti- relative humidity on failure strength, the glass fiber must cle's surface condition. The presence of Griffith faws on be collected in a careful and consistent manner and kept a glass surface govern the strength at which the glass will in a pristine state until testing. Glass strength is impor- fail. For bulk glass, such as container glass these Aaws tant not only in the end use application for the fiber but are typically created from mechanical damage during the also during the time between fiberizing and installation manufacturing process. For glass fibers these Aaws can and certainly during the manufacturing of said fibers as be made from a combination of mechanical and chemi- it can influence productivity. For a complete discussion on glass fiber applications the reader is referred to the Advanter',XStrand",ShieldStrand'WindSerand"and FliteStrand"are registered made- following manuscript entitled"Glass Fiber-Reinforced Composites: From Formulation to Application"found e 2012 The American Ceramic Sociery and Wiley Periodicals, Inc following this article
Strength of High Performance Glass Reinforcement Fiber Michelle L. Korwin-Edson,* Douglas A. Hofmann, and Peter B. McGinnis Owens Corning Science & Technology, 2790 Columbus Rd, Granville, Ohio The practical strength of glass is highly dependent on the amount and type of damage that a glass article has experienced in its lifetime and can be 50% less than its theoretical strength. Glass reinforcement fibers in the pristine state exhibit some of the highest failure strengths of any glass form. Strength degradation is a sequential process the further from the point of formation a glass travels. Individual filament strength is important in the manufacturing process as the fiber interacts with water, HVAC, sizing applicators, contact shoes, and guide eyes and ultimately this combination impacts productivity. A discussion of glass fiber strength — pristine versus usable, and the effects of temperature, humidity, and composition on glass strength follows in this manuscript. New data collected in Owens Corning’s Glass Properties Laboratory on the effect of temperature and relative humidity on strength and modulus for Advantex® glass, Owens Corning’s S-glass (XStrand®S, FliteStrand®S, and ShieldStrand®S) and H-glass (WindStrand®H) are presented. Owens Corning’s understanding of the effect of composition on strength and modulus, and particularly how individual oxides contribute to these properties are shared. Introduction It is well known in the field that the practical strength of a given glass is dependent on that glass article’s surface condition. The presence of Griffith flaws on a glass surface govern the strength at which the glass will fail. For bulk glass, such as container glass these flaws are typically created from mechanical damage during the manufacturing process. For glass fibers these flaws can be made from a combination of mechanical and chemical damage and can often be caused only by chemical damage or the presence of water near strained bonds. To understand the effects of composition, temperature or relative humidity on failure strength, the glass fiber must be collected in a careful and consistent manner and kept in a pristine state until testing. Glass strength is important not only in the end use application for the fiber but also during the time between fiberizing and installation and certainly during the manufacturing of said fibers as it can influence productivity. For a complete discussion on glass fiber applications the reader is referred to the following manuscript entitled “Glass Fiber-Reinforced Composites: From Formulation to Application” found following this article. Advantex®, XStrand®, ShieldStrand® WindStrand® and FliteStrand® are registered trademarks of Owens Corning. *michelle.korwin-edson@owenscorning.com © 2012 The American Ceramic Society and Wiley Periodicals, Inc International Journal of Applied Glass Science 3 [2] 107–121 (2012) DOI:10.1111/j.2041-1294.2012.00089.x
International Journal of Applied Glass Science-Korwin-Edson, Hofmann, and McGinnis Vol. 3, No. 2, 201 Review of Strength Theory as it Pertains to Glass Fiber ture. Surface faws in a reinforcement glass fiber made ng settin g can be created by contact There are already several fiber and glass strength damage with water, sizing applicators, sizing solids reviews worth noting which contain a wealth of histori- guide eyes, shoes, other fibers, cots, cutters and collets cal references. At its very core, glass strength and These Aaws may be considered physical or chemical in modulus deals with the bond strengths present in a glass nature, but can still be considered a Griffith Aaw in structure, the connectivity and the dimensionality of the either case since we are dealing with fibers on the order d work. Generally speaking, weaker bonds have greater of 10 um in diameter and a faw resulting in a signifi nces between atoms which lead to weaker glas cant stress concentration may only be a few nanometers and glasses with lower connectivity have lower stiffne deep. In 1920, Griffith stated that: Theory of Condon-Morse Eγ The Condon-Morse curve describes the attractive where or is the failure stress(strength), E is the Young's and repulsive forces at play inside the network of an modulus, y is the fracture surface energy, and a is the critical crack depth for crack growth. There is a stress rium separation distance between neighboring atoms concentration at the tip of the critical crack and the where these two forces are in harmony. The interatomic failure stress is the stress at which the strained bonds at force is F=-dUldr, where U is the thermodynamic the crack tip break and propagate causing failure and internal energy and r is the separation distance. Stiffness is s=dU/dr and the Elastic modulus is E- Sr minimizing the energy of the system. Griffith stated and is also the slope of the Condon-Morse force curve the rupture of the solid has occurred if the system can pass from the unbroken to the broken condition by a r un separation distanc process involving a continuous decrease in potential energy". It is believed that the presence of water mole Tbeory of Field Strengi cules at this crack tip reduces the stress or energy Field strength(proposed by Dietzel) is another the required to separate the glass network bonds oretical principle to describe the bond interactions fith performed two point bend experiments on glass between cations and anions(usually oxygen). The field hers, where two ends of a fiber are brought closer trength F of a cation is given by fiber, since he theorized that the critical faws must be (1) smaller due to the smaller diameter of the fiber ze is the valency of the cation and re and ra are the Orowvan's Theoretical Strength radii of the cation and the anion(oxygen).The While griffith was ng the effects of cracks force exerted by a cation on a point charge similar to an on strength, others were working on calculating the oxygen anion is Force=Re, where e is the clectronic maximum theoretical cohesive strength of solids.In arge of the cation. The equilibrium separation distance 1949, Orowan proposed the following for the theoreti useful when considering Young's modulus as it cal strength of glass depends on the relationship between an applied force and the magnitude of change in this separation distance. The field strength is useful when considering strength as high field strength cations are glass formers and low field strength cations are network modifiers where om is the Orowan stress (or Cohesive stress) and To is the interatomic equilibrium separation dis- Grifith Flaw Theory The Orowan stress is considered the stress or red to create two new surfaces. The work The practical strength of a glass fiber is determined in stressing the glass to om must at least equal by surface flaws, bulk flaws, composition, and tempera- the energy required to create the two new surfaces
Review of Strength Theory as it Pertains to Glass Fiber There are already several fiber and glass strength reviews worth noting which contain a wealth of historical references.1–5 At its very core, glass strength and modulus deals with the bond strengths present in a glass structure, the connectivity and the dimensionality of the network. Generally speaking, weaker bonds have greater distances between atoms which lead to weaker glasses and glasses with lower connectivity have lower stiffness. Theory of Condon-Morse The Condon-Morse curve describes the attractive and repulsive forces at play inside the network of an ionic material such as glass.6,7 There exists an equilibrium separation distance between neighboring atoms where these two forces are in harmony. The interatomic force is F = dU/dr, where U is the thermodynamic internal energy and r is the separation distance. Stiffness is S = dU2 /dr 2 and the Elastic modulus is E = S/ro and is also the slope of the Condon-Morse force curve at ro, the equilibrium separation distance. Theory of Field Strength Field strength (proposed by Dietzel) is another theoretical principle to describe the bond interactions between cations and anions (usually oxygen).8 The field strength F of a cation is given by: F ¼ Zc ðrc raÞ 2 ð1Þ where Zc is the valency of the cation and rc and ra are the ionic radii of the cation and the anion (oxygen). The force exerted by a cation on a point charge similar to an oxygen anion is Force = F(e) 2 , where e is the electronic charge of the cation. The equilibrium separation distance is useful when considering Young’s modulus as it depends on the relationship between an applied force and the magnitude of change in this separation distance. The field strength is useful when considering strength as high field strength cations are glass formers and low field strength cations are network modifiers. Griffith Flaw Theory The practical strength of a glass fiber is determined by surface flaws, bulk flaws, composition, and temperature. Surface flaws in a reinforcement glass fiber made in a manufacturing setting can be created by contact damage with water, sizing applicators, sizing solids, guide eyes, shoes, other fibers, cots, cutters and collets. These flaws may be considered physical or chemical in nature, but can still be considered a Griffith flaw in either case since we are dealing with fibers on the order of 10 µm in diameter and a flaw resulting in a signifi- cant stress concentration may only be a few nanometers deep. In 1920, Griffith stated that: rf ¼ ffiffiffiffiffiffiffiffi 2Ec pc r ð2Þ where rf is the failure stress (strength), E is the Young’s modulus, c is the fracture surface energy, and c* is the critical crack depth for crack growth.9 There is a stress concentration at the tip of the critical crack and the failure stress is the stress at which the strained bonds at the crack tip break and propagate causing failure and minimizing the energy of the system. Griffith stated “the rupture of the solid has occurred if the system can pass from the unbroken to the broken condition by a process involving a continuous decrease in potential energy”. It is believed that the presence of water molecules at this crack tip reduces the stress or energy required to separate the glass network bonds.10,11 Grif- fith performed two point bend experiments on glass fibers, where two ends of a fiber are brought closer together creating a curve in the center section of the fiber, since he theorized that the critical flaws must be smaller due to the smaller diameter of the fiber. Orowan’s Theoretical Strength While Griffith was considering the effects of cracks on strength, others were working on calculating the maximum theoretical cohesive strength of solids. In 1949, Orowan proposed the following for the theoretical strength of glass: rm ¼ ffiffiffiffiffiffi Ec ro r ð3Þ where rm is the Orowan stress (or Cohesive stress), and ro is the interatomic equilibrium separation distance.12 The Orowan stress is considered the stress or energy required to create two new surfaces. The work done in stressing the glass to rm must at least equal the energy required to create the two new surfaces. 108 International Journal of Applied Glass Science—Korwin-Edson, Hofmann, and McGinnis Vol. 3, No. 2, 2012
amics.org/lAGS High Performance Glass Reinforcement Fiber Using this equation the theoretical strength of a silicate a higher probability of occurring on the surface com glass has been calculated to be approximately 30 GPa. pared with bulkier samples. As opposed to samples Reinforcement fibers have very small diameters (-10- tested in bending, samples tested in tension have their d their strength limiting entire volume between the test grips exposed to equal undetectable. The prefiberization protocol(dwell tem- stress assuming pure tension. Pristine fibers that have eratures), treatment of the fibers(pristine or whether been created in the laboratory under highly controlled they have undergone stress corrosion/fatigue), the conditions have a low possibility of containing bulk or method used to measure the strength of fibers(includ- surface flaws. For the tensile strength data collected on hinah emperature, humidity, speed of loading), and pristine fibers in Owens Corning Glass Properties Lab- ly whether the data are diameter dependent dete oratory, the effective volume ranges between 0.00383 mines whether the resulting data are considered extrin- and 0.00415 mmand the effective surface area ranges ic, intrinsic, inert, fatigue-free etc. as defined by between 0.782 and 0.814 mm. It is assumed that any Kurkjian et al. In this article, we discuss strength in Baws albeit rare would occur on the surface of the context and only compare data that were collected in a fibers, but as this range represents 4% of the average, similar fashion on equally treated samples unless stated size-strength scaling considerations were not taken into otherwise. For the purposes of determining the compe sitional effe gth, all samples were tre Bulk faws are those that exist on the nterior of a equally, considered to be pristine, having undergone no fiber and act as stress concentrators which may lead to fatigue, diameters were kept as equivalent as possible at reductions in strength. Examples of these types of faws 10 um, and tested at room temperature under tension are: devitrification, seeds, unmelted batch particles, het with a gauge length of two inches. The authors con- erogeneities, cord, and other inclusions such as refrac sider this treatment and testing of the samples to be tory or precious metals. One example is shown in pristine tensile strength. It does not fall into any of Fig. 1. Bulk flaws are common in manufacturing and the categories mentioned above as they are defined in lead to lower realized strengths, as has been confirmed Kurkjian et al's 2003 paper, but it is nearly the same by researchers. 4. I It is critical therefore to keep glass temperatures well above the liquidus during manufac- turing. For the purposes of research and the data pre- Weibull statistics sented in this paper bulk Aaws are rare and are not onsidered an is As fractures seek the weakest spot, there tends to Composition plays an important role in the theo be an inherent distribution of failure strengths in any retical strength of a glass and to a great extent in the set of samples. In 1939, Weibull created a statistical treatment of strength data which has become a com- mon way to deal with brittle fracture data, such that: Pa here Mo)is the probability of fracture under a uni form tension a and g. is the reference stress. 5 The traditional formula includes a V(sample volume before the stress ratio. He assumed that the risk of rup- ture of an elemental volume was dependent upon the stress raised to a positive power m, where m is the Weibull modulus. When m is high there is less varia- tion or distribution in the data. tensile fiber data and two point bend fiber data can have representative Wei- bull distributions if the experiments are carried out Fig. 1. An example of a bulk faw collected during produc properly. For glass fibers, the surface area to volume as a broken"bead "In this case the faw is an agglor high and therefore strength limiting Aaws have melter refractory material
Using this equation the theoretical strength of a silicate glass has been calculated to be approximately 30 GPa. Reinforcement fibers have very small diameters (~10– 20 µm) and their strength limiting flaws are practically undetectable. The prefiberization protocol (dwell temperatures), treatment of the fibers (pristine or whether they have undergone stress corrosion/fatigue), the method used to measure the strength of fibers (including temperature, humidity, speed of loading), and finally whether the data are diameter dependent determines whether the resulting data are considered extrinsic, intrinsic, inert, fatigue-free etc. as defined by Kurkjian et al.4 In this article, we discuss strength in context and only compare data that were collected in a similar fashion on equally treated samples unless stated otherwise. For the purposes of determining the compositional effect on strength, all samples were treated equally, considered to be pristine, having undergone no fatigue, diameters were kept as equivalent as possible at 10 µm, and tested at room temperature under tension with a gauge length of two inches. The authors consider this treatment and testing of the samples to be “pristine tensile strength.” It does not fall into any of the categories mentioned above as they are defined in Kurkjian et al.’s 2003 paper, but it is nearly the same as intrinsic strength. Weibull Statistics As fractures seek the weakest spot, there tends to be an inherent distribution of failure strengths in any set of samples. In 1939, Weibull created a statistical treatment of strength data which has become a common way to deal with brittle fracture data, such that: PðrÞ ¼ 1 exp r ro m ð4Þ where P(r) is the probability of fracture under a uniform tension r and ro is the reference stress.13 The traditional formula includes a V (sample volume) before the stress ratio. He assumed that the risk of rupture of an elemental volume was dependent upon the stress raised to a positive power m, where m is the Weibull modulus. When m is high there is less variation or distribution in the data. Tensile fiber data and two point bend fiber data can have representative Weibull distributions if the experiments are carried out properly. For glass fibers, the surface area to volume ratio is high and therefore strength limiting flaws have a higher probability of occurring on the surface compared with bulkier samples. As opposed to samples tested in bending, samples tested in tension have their entire volume between the test grips exposed to equal stress assuming pure tension. Pristine fibers that have been created in the laboratory under highly controlled conditions have a low possibility of containing bulk or surface flaws. For the tensile strength data collected on pristine fibers in Owens Corning Glass Properties Laboratory, the effective volume ranges between 0.00383 and 0.00415 mm3 and the effective surface area ranges between 0.782 and 0.814 mm2 . It is assumed that any flaws albeit rare would occur on the surface of the fibers, but as this range represents 4% of the average, size-strength scaling considerations were not taken into account. Bulk flaws are those that exist on the interior of a fiber and act as stress concentrators which may lead to reductions in strength. Examples of these types of flaws are: devitrification, seeds, unmelted batch particles, heterogeneities, cord, and other inclusions such as refractory or precious metals. One example is shown in Fig. 1. Bulk flaws are common in manufacturing and lead to lower realized strengths, as has been confirmed by researchers.14,15 It is critical therefore to keep glass temperatures well above the liquidus during manufacturing. For the purposes of research and the data presented in this paper bulk flaws are rare and are not considered an issue. Composition plays an important role in the theoretical strength of a glass and to a great extent in the Fig. 1. An example of a bulk flaw collected during production as a broken “bead.” In this case the flaw is an agglomerate of melter refractory material. www.ceramics.org/IJAGS High Performance Glass Reinforcement Fiber 109
International Journal of Applied Glass Science-Korwin-Edson, Hofmann, and McGinnis Vol. 3, No. 2, 201 Table I. Comparison of Four Glass Types and their Pristine Tensile Strength Along with the mol% of Glass Former Pristine tensile strength SiO2+ Al2o Glass name Glass type mol (kpsi)(MPa) ASTM D578 E for general applications -75 500 ≈3450 Advantex(Owens Corning) Boron Free EC-R glass 77 ≈3800 XStrand S(Owens Corning) Alumino silicate glas 740 ≈5100 Ref. 16 820 ≈5700 Pristine tensile strength is measured as described in Experimental Method. practical strength as well. Higher intrinsically strong The graph in Fig. 3 shows a set of Owens Corning glasses have higher hardness and tend to be more dam data and the effect of fiber diameter on measured pris- age resistant to strength limiting flaws in the first place tine tensile strength for an E-glass. As Otto showed in In comparing just four glass types in Table L, a trend is 1955, there is still no appreciable effect of fiber diame seen in the pristine tensile strength with compositional ter on strength in this size range. These diameters changes, specifically the overall amount of Al2O3 and shown (6-24 um) are typical of those sold in a major- SiO2 combined. The Sio2 data came from Kurkji ity of reinforcement fber products. The strengths are and paek somewhat lower for the smaller diameter fibers. this The chart in Fig. 2 shows the effect of sample could be a result of the increased tension found in form, test conditions, composition, and surface condi- forming finer fibers if the temperature and composition tion on measured strength. Clearly it is seen that there of the glass remains the same a large discrepancy between the theoretical strength and the actual data collected. the trends observed are Efects of Temperature 1. The smaller the sample form the larger the One simple example of the effect of measured strength as shown in going from bulk emperature on strength is seen in Fig. 2 as the Pristine glass to rods to fibers. This is due to the fact that there is a higher probability for larger grif- fith flaws as the sample size increases and the effects of Weibull strength-size scaling and 2. Samples tested under conditions such as liquid 8 nitrogen show drastically improved strength(30 -50% increase). This occurs because the main reactive species water is frozen at 77 K and kept from being mobile on the glass surface 00sa7152536a5 and participating in the bond breaking reaction 3. Going from alkali silicate to a traditional E-glass (containing Boron) to Owens Corning S-glass ncreasing streng removal of alkali ions and borate and additions of silica, alumina, and magnesia. The effects of Fig. 2. Tensile strength as it relates to sample form, composition rength will be discussed in of glass, and surface conditions(S-Glass source was Owens Corn- more detail later on in this article ing's XStrand" S and LN2 means submerged in liquid nitrogen)
practical strength as well. Higher intrinsically strong glasses have higher hardness and tend to be more damage resistant to strength limiting flaws in the first place. In comparing just four glass types in Table I, a trend is seen in the pristine tensile strength with compositional changes, specifically the overall amount of Al2O3 and SiO2 combined. The SiO2 data came from Kurkjian and Paek.16 The chart in Fig. 2 shows the effect of sample form, test conditions, composition, and surface condition on measured strength. Clearly it is seen that there is a large discrepancy between the theoretical strength and the actual data collected. The trends observed are such that: 1. The smaller the sample form the larger the measured strength as shown in going from bulk glass to rods to fibers. This is due to the fact that there is a higher probability for larger Grif- fith flaws as the sample size increases and the effects of Weibull strength-size scaling and effective area or volume are predominant. 2. Samples tested under conditions such as liquid nitrogen show drastically improved strength (30 –50% increase). This occurs because the main reactive species water is frozen at 77 K and kept from being mobile on the glass surface and participating in the bond breaking reaction. 3. Going from alkali silicate to a traditional E-glass (containing Boron) to Owens Corning S-glass “handled” fibers show increasing strength with removal of alkali ions and borate and additions of silica, alumina, and magnesia. The effects of composition on strength will be discussed in more detail later on in this article. The graph in Fig. 3 shows a set of Owens Corning data and the effect of fiber diameter on measured pristine tensile strength for an E-glass. As Otto showed in 1955, there is still no appreciable effect of fiber diameter on strength in this size range.17 These diameters shown (6–24 lm) are typical of those sold in a majority of reinforcement fiber products. The strengths are somewhat lower for the smaller diameter fibers, this could be a result of the increased tension found in forming finer fibers if the temperature and composition of the glass remains the same. Effects of Temperature One simple example of the effect of measurement temperature on strength is seen in Fig. 2 as the Pristine Table I. Comparison of Four Glass Types and their Pristine Tensile Strength Along with the mol% of Glass Former Glass name Glass type SiO2 + Al2O3 Measured pristine tensile strength* mol % (kpsi) (MPa) E ASTM D578 E for general applications ~75 500 3450 Advantex® (Owens Corning) Boron Free EC-R glass ~77 550 3800 XStrand® S (Owens Corning) Alumino silicate glass ~92 740 5100 SiO2 Ref. 16 100 820 5700 *Pristine tensile strength is measured as described in Experimental Method. Fig. 2. Tensile strength as it relates to sample form, composition of glass, and surface conditions (S-Glass source was Owens Corning’s XStrand®S and LN2 means submerged in liquid nitrogen). 110 International Journal of Applied Glass Science—Korwin-Edson, Hofmann, and McGinnis Vol. 3, No. 2, 2012
amics.org/lAGS High pe nce Glass Reinforcement Fiber tion brin. rgy present to conduct the well known reac- kinetic ene een adsorbed water and stressed siloxane O≡SiO-Si≡→2Si-OH Equation (5)is the standard reaction which oc curs during stress corrosion or environmental fatigue. It is thought that water molecules are present at the tip of Avg actual Avg measured tengu图r a crack on a glass surface where the water reacts with the siloxane bond at the crack tip leading to slow growth or propagation of the crack. However, the glass has to be under some form of tensile stress for this to occur. In a pristine fiber, however, there are no classi- Fig. 3. E-glass strength as a fiunction of fiber diameter for cal Aaws on the surface and if this fiber is not put orcement fiber p I2. under an applied stress there should be no reduction in strength, but there is. Gupta et al stated that the mechanism of fatigue in the absence of cracks is not Owens Corning S-Glass Fiber tested in liquid nitrogen. clear Some work has been done in this area from liquid There may not be cracks present on a pristine fiber Helium temperatures of 4 K to temperatures of surface, but there must be irregularities at the very least -900K, primarily for silica at the lower tempera- in the structure of the glass itself to cause stress concen- tures Proctor showed increasing strength of silica trations thus targeting bonds for destruction. Afiber fibers measured under decreasing temperature with 12- that is not under an applied load is still under internal 13 GPa at 77 K and 14 GPa at 4 K 8 Cameron stress because of the nature in which the fiber is showed that the strength of E-glass went from 820 kpsi formed. The fiber is elongated at high speeds and high at -190oC to approximately 580 kpsi at 29C with temperature which freeze in a slightly oriented and many measurements in between at various humidities. 2 strained structure. This is known because of experi- Duncan measured both silica and sodium borosilicate ments done on annealing fibers which show changes in glass in two point bend and showed that the strain of properties such as density and refractive index. So silica was 18% at -196C and was 5% at 100.C, for can be said that even pristine fibers under no load can sodium borosilicate the strain was 15% at C and still undergo stress corrosion in the presence of was2.5%at100°C, with sever adsorbed water on the surface. Gy estimated the size of between showing both the effect of temperature and a surface crack for a fiber which exhibited a room temper- humidity. Otto collected data at much higher tem- ature tensile strength of 3000 MPa to be approximately eratures for E-glass and showed the strength to be 200 A and at 77 K the strength being twice as high 580 kpsi at room temperature and then 200 kspi at would suggest a surface crack of only 50 A.Cameron 620 C both at 0% RH (relative humidity). These calculated the Aaw to be 6 A at 77 K20 Gy goes on to data are shown graphically later on in the article along- say that". the fact that the fracture originates from side more the surface, has not been directly evidenced: high speed It is well established that measurements of strength images recording of the fracture process shows that the at 77 K under liquid nitrogen are free of environmental fiber is fully pulverized by the release of the stored elas- corrosion or fatigue as the reaction kinetics of water tic energy, making hopeless any attempt to obtain an and the silica structure are minute at this temperature. insight into this question from an examination of the As such, the strength values obtained at this tempera- fracture surfaces. "0 Tomozawa has even said that water ture are much higher owing to the fact that the energy will diffuse into the structure of a silica glass under a required to create two new surfaces is greater when the tensile load removing even the requirement for surface surfaces do not have readily available species with which water to cause strength reductions. Gy and guillemet to react. On the other hand, at elevated temperatures tested fibers under high temperature and were able to there is a reduction in strength due to the increased preserve the fracture surfaces. Analysis of these
Owens Corning S-Glass Fiber tested in liquid nitrogen. Some work has been done in this area from liquid Helium temperatures of 4 K to temperatures of ~900 K, primarily for silica at the lower temperatures.18–22 Proctor showed increasing strength of silica fibers measured under decreasing temperature with 12– 13 GPa at 77 K and 14 GPa at 4 K.18 Cameron showed that the strength of E-glass went from 820 kpsi at 190°C to approximately 580 kpsi at 29°C with many measurements in between at various humidities.20 Duncan measured both silica and sodium borosilicate glass in two point bend and showed that the strain of silica was 18% at 196°C and was 5% at 100°C, for sodium borosilicate the strain was 15% at 196°C and was 2.5% at 100°C, again with several points in between showing both the effect of temperature and humidity.19 Otto collected data at much higher temperatures for E-glass and showed the strength to be 580 kpsi at room temperature and then 200 kspi at 620°C both at 0% RH (relative humidity).22 These data are shown graphically later on in the article alongside more recent data. It is well established that measurements of strength at 77 K under liquid nitrogen are free of environmental corrosion or fatigue as the reaction kinetics of water and the silica structure are minute at this temperature. As such, the strength values obtained at this temperature are much higher owing to the fact that the energy required to create two new surfaces is greater when the surfaces do not have readily available species with which to react. On the other hand, at elevated temperatures there is a reduction in strength due to the increased kinetic energy present to conduct the well known reaction between adsorbed water and stressed siloxane bonds:23,24 H2O Si-O-Si ! 2Si-OH ð5Þ Equation (5) is the standard reaction which occurs during stress corrosion or environmental fatigue. It is thought that water molecules are present at the tip of a crack on a glass surface where the water reacts with the siloxane bond at the crack tip leading to slow growth or propagation of the crack. However, the glass has to be under some form of tensile stress for this to occur. In a pristine fiber, however, there are no classical flaws on the surface and if this fiber is not put under an applied stress there should be no reduction in strength, but there is. Gupta et al.25 stated that the mechanism of fatigue in the absence of cracks is not clear. There may not be cracks present on a pristine fiber surface, but there must be irregularities at the very least in the structure of the glass itself to cause stress concentrations thus targeting bonds for destruction. A fiber that is not under an applied load is still under internal stress because of the nature in which the fiber is formed. The fiber is elongated at high speeds and high temperature which freeze in a slightly oriented and strained structure. This is known because of experiments done on annealing fibers which show changes in properties such as density and refractive index. So it can be said that even pristine fibers under no load can still undergo stress corrosion in the presence of adsorbed water on the surface. Gy estimated the size of a surface crack for a fiber which exhibited a room temperature tensile strength of 3000 MPa to be approximately 200 A˚ and at 77 K the strength being twice as high would suggest a surface crack of only 50 A˚. 10 Cameron calculated the flaw to be 6 A˚ at 77 K.20 Gy goes on to say that “… the fact that the fracture originates from the surface, has not been directly evidenced: high speed images recording of the fracture process shows that the fiber is fully pulverized by the release of the stored elastic energy, making hopeless any attempt to obtain an insight into this question from an examination of the fracture surfaces.”10 Tomozawa has even said that water will diffuse into the structure of a silica glass under a tensile load removing even the requirement for surface water to cause strength reductions.26 Gy and Guillemet tested fibers under high temperature and were able to preserve the fracture surfaces.27 Analysis of these surFig. 3. E-glass strength as a function of fiber diameter for typical reinforcement fiber product sizes. www.ceramics.org/IJAGS High Performance Glass Reinforcement Fiber 111
International Journal of Applied Glass Science-Korwin-Edson, Hofmann, and McGinnis Vol. 3, No. 2, 201 faces showed that the fractures originated from within Thomas exposed E-glass fibers to either 0% or the bulk 100%RH at 20 C for varying amounts of time.2He The effect of temperature treatments on the room showed that the breaking stress decreases with an temperature strength of E-glass fibers were studied by increase of RH and with an increase of time of expo Thomas. Great care was taken such that no load was sure. He then offered two data points (7 and 28 days) laced on the fibers during the heat treatment. He on exposure of the fiber in a vacuum and showed that showed that the strength was greatly reduced with it produced no change in the breaking stress. Thomas increasing temperatures of treatment and that the scatter indicated that he expected the 0% RH samples to show in the data also increased with increasing temperature. no change in breaking stress and the 100% RH sam- There was a gradual decrease in strength from 0C to ples to show an immediate and severe drop on breaking 450C and then a fattening of strength at 1/3 the origi- stress. The disconnect between expected results and nal at 550%C and 600oC. He claimed that the scatter was obtained results for Thomas is somewhat explained by due to the heating and not due to atmospheric attack. Cameron with his low and high vacuum studies which He proposed that "the scatter was caused by molecular suggest that adsorbed water plays the key role in corro- nd structural rearrangements in the glass during the heat sion not relative humidity. Feih performed a zero-stress treatment, causing stresses to be set up which result in study, but used exposures in ambient air, dry air, and the production of flaws and cracks. In a very recent dry nitrogen at elevated temperatures for very short report, Feih et al reported the effects of thermal recy- times compared with Thomas. All treatments showed cling processing conditions on the tensile strength of tensile strength reductions due to the elevated tempera- commercial E-glass fibers and bundles. The fibers were ture alone, but at 450@C for 30 min, the dry nitrogen exposed to temperatures between 150C and 600C for strengths were still considerably higher(quite close to various times at -50%RH and then tested at room tem- the original strengths) than the dry air or ambient air perature in tension. Similar to Thomas, Feih showed that strengths measured. At 450.C and 2 h, strengths were the individual fiber and bundle strengths drop rapidly equally degraded independent of the atmosphere and at over a temperature range 250-550oC. The strength ini- higher temperatures also. Possible explanations give tially decreases rapidly with time and then reaches a low for the mechanism of zero-stress corrosion were: struc- steady-state value that is dictated by the temperature. tural relaxation, increased diffusion of molecular water, The time taken to reach steady-state strength lessens with ion exchange induced tensile stress, embrittlement, and re. They provide mathematical of course growth of pre-existing fay dels for predicting strength loss and rate thereof. Strength degradation takes place under aging with Efects of Humidity slightly elevated temperatures and especially when there is excess water supplied. Several studies on silica optical Significant work has been done on the effects of fiber and their susceptibility to static and dynamic fati- humidity and adsorbed water on strength degradation gue were undertaken to support the telecommunica and is extensively reported in the literature. This is in tions application of fused silica fibers. . Static fatigue part because water adsorption is practically unavoidable is when a low constant load is applied to a fiber and nd because of the well-known water and siloxane bond failure occurs with time, dynamic fatigue is when a reaction and understanding it is critical to understand- constant rate of loading is applied to a fiber until fail essence of glass strength. Cameron stated it elo- ure occurs which happens within seconds. Armstrong in his paper when he said, "A bond in a glass et al. showed that the strength is reduced with which is highly strained by forces applied exter- increasing test humidity at a constant temperature of ally, undoubtedly attracts molecular water. At some 25"C and that it is a nonlinear relationship(exponen- degree of bond strain during a tensile test, a water tial). The results were also independent of whether the molecule will attach itself to a bridging oxygen in the r was bare or coated with silicone imide,or network, dissociate, and break a network bond. Neigh- polyurethane-acrylate. This study was mainly concerned boring bonds will be severed in the same way until with determining the proper kinetic model for predict- stress to ing the reliability of optical fibers in application. Dun develop and lead to spontaneous fracture can also showed a similar dependence of strength on
faces showed that the fractures originated from within the bulk. The effect of temperature treatments on the room temperature strength of E-glass fibers were studied by Thomas.28 Great care was taken such that no load was placed on the fibers during the heat treatment. He showed that the strength was greatly reduced with increasing temperatures of treatment and that the scatter in the data also increased with increasing temperature. There was a gradual decrease in strength from 0°C to 450°C and then a flattening of strength at 1/3 the original at 550°C and 600°C. He claimed that the scatter was due to the heating and not due to atmospheric attack. He proposed that “the scatter was caused by molecular and structural rearrangements in the glass during the heat treatment, causing stresses to be set up which result in the production of flaws and cracks.” In a very recent report, Feih et al.29 reported the effects of thermal recycling processing conditions on the tensile strength of commercial E-glass fibers and bundles. The fibers were exposed to temperatures between 150°C and 600°C for various times at ~50% RH and then tested at room temperature in tension. Similar to Thomas, Feih showed that the individual fiber and bundle strengths drop rapidly over a temperature range 250–550°C. The strength initially decreases rapidly with time and then reaches a low steady-state value that is dictated by the temperature. The time taken to reach steady-state strength lessens with increasing temperature. They provide mathematical models for predicting strength loss and rate thereof. Effects of Humidity Significant work has been done on the effects of humidity and adsorbed water on strength degradation and is extensively reported in the literature. This is in part because water adsorption is practically unavoidable and because of the well-known water and siloxane bond reaction and understanding it is critical to understanding the essence of glass strength. Cameron stated it eloquently in his paper when he said, “A bond in a glass surface which is highly strained by forces applied externally, undoubtedly attracts molecular water. At some degree of bond strain during a tensile test, a water molecule will attach itself to a bridging oxygen in the network, dissociate, and break a network bond. Neighboring bonds will be severed in the same way until there is a flaw large enough for a critical stress to develop and lead to spontaneous fracture.”20 Thomas exposed E-glass fibers to either 0% or 100% RH at 20°C for varying amounts of time.28 He showed that the breaking stress decreases with an increase of RH and with an increase of time of exposure. He then offered two data points (7 and 28 days) on exposure of the fiber in a vacuum and showed that it produced no change in the breaking stress. Thomas indicated that he expected the 0% RH samples to show no change in breaking stress and the 100% RH samples to show an immediate and severe drop on breaking stress. The disconnect between expected results and obtained results for Thomas is somewhat explained by Cameron with his low and high vacuum studies which suggest that adsorbed water plays the key role in corrosion not relative humidity. Feih performed a zero-stress study, but used exposures in ambient air, dry air, and dry nitrogen at elevated temperatures for very short times compared with Thomas.29 All treatments showed tensile strength reductions due to the elevated temperature alone, but at 450°C for 30 min, the dry nitrogen strengths were still considerably higher (quite close to the original strengths) than the dry air or ambient air strengths measured. At 450°C and 2 h, strengths were equally degraded independent of the atmosphere and at higher temperatures also. Possible explanations given for the mechanism of zero-stress corrosion were: structural relaxation, increased diffusion of molecular water, ion exchange induced tensile stress, embrittlement, and of course growth of pre-existing flaws. Strength degradation takes place under aging with constant load (static fatigue) applied even at only slightly elevated temperatures and especially when there is excess water supplied. Several studies on silica optical fiber and their susceptibility to static and dynamic fatigue were undertaken to support the telecommunications application of fused silica fibers.30,31 Static fatigue is when a low constant load is applied to a fiber and failure occurs with time, dynamic fatigue is when a constant rate of loading is applied to a fiber until failure occurs which happens within seconds. Armstrong et al.30 showed that the strength is reduced with increasing test humidity at a constant temperature of 25°C and that it is a nonlinear relationship (exponential). The results were also independent of whether the fiber was bare or coated with silicone, polyimide, or polyurethane-acrylate. This study was mainly concerned with determining the proper kinetic model for predicting the reliability of optical fibers in application. Duncan also showed a similar dependence of strength on 112 International Journal of Applied Glass Science—Korwin-Edson, Hofmann, and McGinnis Vol. 3, No. 2, 2012
amics.org/lAGS High Performance Glass Reinforcement Fiber relative humidity for silica and borosilicate fibers. on the glass, the mobility of those molecules and the Duncan studied a broad range of humidity levels and amount of energy present to conduct the OH and related his data to those of Sakaguchi at higher siloxane reaction. Then there are researchers who have humidities(30-90%RH for silica fiber) who found a time and again shown the effect of increasing humidity linear relationship of strength to water vapor pressure on the strength degradation during the testing. Clearly and Wiederhorn3(0.02-100% RH for soda lime bulk the mechanisms for strength degradation, stress corro- glass) who focused on determination of the reaction sion, subcritical crack growth, fatigue, and failure are order of subcritical crack growth models to help explain highly complex and there is not a single mechanism to the mechanism for the strength degradation. Wieder- describe these phenomena. Luckily molecular dynamics horn's slow crack growth data showed that the reaction and experimental techniques are improving and taking order was -1 for relative humidities between 10% and great strides in advancing our knowledge of 100%and that it decreased to 0.5 for relative humidi ties <1%. However, there is some question as to whether slow crack growth data on bulk samples are Test metbods accurate enough to predict failure in high strength fibers. Gy commented on work by Matthewson et aL. 10 There are several ways to measure the fracture/fail that "the dependency of strength of silica fibers on ure strength of glass articles including tensile, two point humidity is not consistent with a first order kinetics of bend, three point bend, and four point bend. For pris- the chemical reaction between H2O and glass, except at tine fibers the most practical tests are tensile and two sistent with the experimental data, raising the possibil- surement of the ensile testing provides a direct mea- ity that in this case, water has to be dissociated first stand before fracture. The two point bend technique, and that hydroxyl species are actually reacting with the recently improved upon by Lower and Brow is an indi- silicon-oxygen bond. Mrotek et al re-iterated the rect method of measuring fracture strength, but gives same conclusion for silica fibers where they said that very repeatable results nevertheless. 4. 4.15 For the two the reaction order was -2 between RHs of 15-100% point bend the strain at failure is measured and but that it changed to first order at low RHs of 0.1 must independently know Youngs modulus to then According to Cameron and works that he has calculate the strength at failure. The Youngs modulus cited, there appears to be a very small effect of relative has been shown to decrease in a material under applied humidity during testing on the failure strength(<5%) stress so that the exact e during failure is hard to deter at relative humidities between 10% and 100% 20 22.35 mine which then creates uncertainty in the calculated However,at relative humidities between 0% and 10% failure strength 4.37 In addition, the two point bend is there is an appreciable effect of RH on strength. Cam- limited to fibers with diameters greater than 100 um eron focused on vacuum treatments of E-glass fibers We at Owens Corning utilize the tensile test technique and testing in the treatment environment to distinguish for direct determination of fracture strengt between the effect of free and adsorbed water 20 All of Camerons specimens showed increased strength after treatment in vacuo. The low vacuum samples showed Experimental Method an 8-10% strength increase and no time dependence of holding time suggesting that adsorbed water was not Several Owens Corning internal standard operating being removed from the glass surface. The higher vac- procedures (SOPs) are utilized in creating tensile uum samples showed significant holding time depen- strength data and reference standard test methods, such dence suggesting removal of bound water with time. as ASTMs D76 and D3822. For creation of pristine Higher vacuums did not produce any more strength single filaments the glass is pulled through a single-hole increase beyond the impressive 25% increase at 50 to bushing in an environmentally a controlled labora ator 1×10-3torr (22C and 35% RH)at a controlled speed to create Some researchers claim that it appears that strength fibers of the desired diameter. The glass was precondi degradation is dependent not on the amount of free tioned at the melting ter moisture in the atmosphere, but on adsorbed moisture before the temperature was reduced to the forming
relative humidity for silica and borosilicate fibers.19 Duncan studied a broad range of humidity levels and related his data to those of Sakaguchi32 at higher humidities (30–90% RH for silica fiber) who found a linear relationship of strength to water vapor pressure and Wiederhorn33 (0.02–100% RH for soda lime bulk glass) who focused on determination of the reaction order of subcritical crack growth models to help explain the mechanism for the strength degradation. Wiederhorn’s slow crack growth data showed that the reaction order was ~1 for relative humidities between 10% and 100% and that it decreased to 0.5 for relative humidities <1%. However, there is some question as to whether slow crack growth data on bulk samples are accurate enough to predict failure in high strength fibers. Gy commented on work by Matthewson et al.10 that “the dependency of strength of silica fibers on humidity is not consistent with a first order kinetics of the chemical reaction between H2O and glass, except at low RH. A second order reaction would be more consistent with the experimental data, raising the possibility that in this case, water has to be dissociated first and that hydroxyl species are actually reacting with the silicon-oxygen bond.” Mrotek et al.34 re-iterated the same conclusion for silica fibers where they said that “the reaction order was ~2 between RHs of 15–100% but that it changed to first order at low RHs of 0.1%.” According to Cameron and works that he has cited, there appears to be a very small effect of relative humidity during testing on the failure strength (<5%) at relative humidities between 10% and 100%.20,22,35 However, at relative humidities between 0% and 10% there is an appreciable effect of RH on strength. Cameron focused on vacuum treatments of E-glass fibers and testing in the treatment environment to distinguish between the effect of free and adsorbed water.20 All of Cameron’s specimens showed increased strength after treatment in vacuo. The low vacuum samples showed an 8–10% strength increase and no time dependence of holding time suggesting that adsorbed water was not being removed from the glass surface. The higher vacuum samples showed significant holding time dependence suggesting removal of bound water with time. Higher vacuums did not produce any more strength increase beyond the impressive 25% increase at 50 to 1 9 103 torr. Some researchers claim that it appears that strength degradation is dependent not on the amount of free moisture in the atmosphere, but on adsorbed moisture on the glass, the mobility of those molecules and the amount of energy present to conduct the OH and siloxane reaction. Then there are researchers who have time and again shown the effect of increasing humidity on the strength degradation during the testing. Clearly the mechanisms for strength degradation, stress corrosion, subcritical crack growth, fatigue, and failure are highly complex and there is not a single mechanism to describe these phenomena. Luckily molecular dynamics and experimental techniques are improving and taking great strides in advancing our knowledge of the strength of glasses.36 Test Methods There are several ways to measure the fracture/failure strength of glass articles including tensile, two point bend, three point bend, and four point bend. For pristine fibers the most practical tests are tensile and two point bending. Tensile testing provides a direct measurement of the maximum pull force a fiber can withstand before fracture. The two point bend technique, recently improved upon by Lower and Brow is an indirect method of measuring fracture strength, but gives very repeatable results nevertheless.4,14,15 For the two point bend the strain at failure is measured and one must independently know Young’s modulus to then calculate the strength at failure. The Young’s modulus has been shown to decrease in a material under applied stress so that the exact E during failure is hard to determine which then creates uncertainty in the calculated failure strength.4,37 In addition, the two point bend is limited to fibers with diameters greater than 100 µm. We at Owens Corning utilize the tensile test technique for direct determination of fracture strength. Experimental Method Several Owens Corning internal standard operating procedures (SOPs) are utilized in creating tensile strength data and reference standard test methods, such as ASTMs D76 and D3822. For creation of pristine single filaments the glass is pulled through a single-hole bushing in an environmentally a controlled laboratory (22°C and 35% RH) at a controlled speed to create fibers of the desired diameter. The glass was preconditioned at the melting temperature for at least 2 h before the temperature was reduced to the forming www.ceramics.org/IJAGS High Performance Glass Reinforcement Fiber 113
International Journal of Applied Glass Science-Korwin-Edson, Hofmann, and McGinnis Vol. 3, No. 2, 201 temperature for fiber pulling. The fiber is then collected that were initially formed from either lab melts,our onto forks such that the fiber never comes in contact pilot facility or our manufacturing plant were tested with anything but the laboratory air. These fibers are both in ambient air and liquid nitrogen. The results are then placed in a dessicator until they are tested which shown as a Weibull plot in Fig. 5. As previously men is always the same day. Fibers are meticulously trans- tioned, the liquid nitrogen strengths are much higher ferred to tensile cards using wax to hold the fibers in in the linear region, but there is much more variation place which are then loaded into an Instron machine (m of -7)or a higher COV(coefficient of variation with a 500 g load cell. If an elevated temperature is for the liquid nitrogen data. The ambient air tensile desired for the test a custom furnace designed to heat data has a very high Weibull modulus (70)or low single filaments is slid into place around the fiber after COV. A unimodal distribution was observed for both it has been secured in the test rig grips. The fiber gauge of the laboratory prepared sample sets and a few of the lengths in this case are 10"instead of the usual 2" to sample sets made in the plant. A bimodal distribution allow room for the furnace. For the liquid nitrogen can be seen in a couple of the plant made sample sets che ing the entire gauge length of the fber as well as and the fibers made in the pilot facility. This could be btai iy s test grip is immersed in a dewar of liquid due to the presence of some internal bulk faws. All of ogen tensile test program is then run using a speed of I mm/min and the maximum load obtained. fiber diameters are then measured on the fractured fibers and this gives the fracture strength according to o- FA, where F is the load at failure and A is the cross-sectional area of the specimen. Results and discussion Part I- Glass Fiber Strength Glass fiber strength is subject to the application environment, exposure history, and composition. Here, re discuss some new data and how the data can be interpreted by using thermodynamic relationships and understanding the mechanism for strength de In(strength( kpsi)) Fig. 4. Weibull plot of Aduantex glass pristine tensile strengt tima tested in ambient air and liquid nitrogen. Efect of Testing Tem ture Advantex glass fibers were created according to the method described above and tested in both ambient air conditions and liquid nitrogen. The Weibull plots liquid nitrogen was 860.8 kpsi (5935 MPa). The Increase In s gth Is seen une conditions along with a small decrease in the slope m or Weibull modulus from 77 in air to 22 in liquid nitrogen. This is the showed more variation in the data for samples tested In(strength(kpsi Owens Corning S-glass fibers(developed for aero- Fig. 5. Measurements of Pristine tensile strength for Owens pace, military, and specific industrial applications), Coming S-glass fibers in liquid nitrogen and ambient air
temperature for fiber pulling. The fiber is then collected onto forks such that the fiber never comes in contact with anything but the laboratory air. These fibers are then placed in a dessicator until they are tested which is always the same day. Fibers are meticulously transferred to tensile cards using wax to hold the fibers in place which are then loaded into an Instron machine with a 500 g load cell. If an elevated temperature is desired for the test a custom furnace designed to heat single filaments is slid into place around the fiber after it has been secured in the test rig grips. The fiber gauge lengths in this case are 10” instead of the usual 2” to allow room for the furnace. For the liquid nitrogen testing the entire gauge length of the fiber as well as the bottom test grip is immersed in a dewar of liquid nitrogen. The tensile test program is then run using a cross head speed of 1 mm/min and the maximum load is obtained. Fiber diameters are then measured on the fractured fibers and this gives the fracture strength according to r = F/A, where F is the load at failure and A is the cross-sectional area of the specimen. Results and Discussion Part 1 — Glass Fiber Strength Glass fiber strength is subject to the application environment, exposure history, and composition. Here, we discuss some new data and how the data can be interpreted by using thermodynamic relationships and understanding the mechanism for strength degradation and ultimate failure. Effect of Testing Temperature Advantex® glass fibers were created according to the method described above and tested in both ambient air conditions and liquid nitrogen. The Weibull plots of the strength data are shown in Fig. 4. The median strength in air was 550 kpsi (3790 MPa) and that in liquid nitrogen was 860.8 kpsi (5935 MPa). The increase in strength is seen under the liquid nitrogen conditions along with a small decrease in the slope m or Weibull modulus from 77 in air to 22 in liquid nitrogen. This is the opposite of Kurkjian et al.4 who showed more variation in the data for samples tested in ambient. Owens Corning S-glass fibers (developed for aerospace, military, and specific industrial applications), that were initially formed from either lab melts, our pilot facility or our manufacturing plant were tested both in ambient air and liquid nitrogen. The results are shown as a Weibull plot in Fig. 5. As previously mentioned, the liquid nitrogen strengths are much higher in the linear region, but there is much more variation (m of ~7) or a higher COV (coefficient of variation) for the liquid nitrogen data. The ambient air tensile data has a very high Weibull modulus (~70) or low COV. A unimodal distribution was observed for both of the laboratory prepared sample sets and a few of the sample sets made in the plant. A bimodal distribution can be seen in a couple of the plant made sample sets and the fibers made in the pilot facility. This could be due to the presence of some internal bulk flaws. All of Fig. 4. Weibull plot of Advantex® glass pristine tensile strength tested in ambient air and liquid nitrogen. Fig. 5. Measurements of pristine tensile strength for Owens Corning S-glass fibers in liquid nitrogen and ambient air. 114 International Journal of Applied Glass Science—Korwin-Edson, Hofmann, and McGinnis Vol. 3, No. 2, 2012
amics.org/lAGS High Performance Glass Reinforcement Fiber the sample sets tested in liquid nitrogen showed some- 20125G5,35%阳H4 what of a bimodal distribution where one distribution →2012Ega55,35%阳 falls close to the ambient air data. This shows that even Egs5,Ott50%阳 when the water is immobilized there is a failure mecha -E-glass, otto, O% RH nism where the energy required to create two new sur- faces is met at a lower threshold (close to ambient data) and is independent of mobile water. The median strength for these Owens Corning S-glass specimens at ambient was around 5050 MPa and for liquid was 7820 MPa At higher temperatures, again similarly to previous studies mentioned in the earlier portions of this article, the strength of fibers is reduced with increasing test Shown in Fig. 6 are data from 1959 and Test Temperature(deg c) recently on the same glass composition (traditional Fig. 6. Median pristine tensile strength of traditional E-glass boron containing E-glass) show only minor differences data sets taken from premo iner of test temperature. Two and OC S-glass fibers as a fincti al work by Orto probably due to increased standardization of e test methods used to create, collect, and test fibers in tensile or perhaps small differences in the E-glass chemistry. Also shown is the strength for Owens Corning S-glass tested under tension at some of the same temperatures. In addition to the decreased strengths with increas- ing temperature there is an interesting effect on the Weibull modulus in that it actually increases with increasing temperature (or the COv decreases). The Weibull plots of the E-glass data are shown in Fig. 7 with the values of m shown in the inset table. A toi ble explanation for the decreased variation in data wit temperature is that at higher temperatures, the activa- tion energy for the water and siloxane reaction (as shown in Eq. (5)is met and therefore the reaction occur readily at any stressed bond thus increasing the Fig. 7. Weibull plots of traditional E-glass strength data as a According to several researchers the strength finction of temperature dependence on temperature for most glasses can be described with an equation of the form: vation energy equal to [ Q/(N+ 1). This decrease is caused by the increased slow crack growth rate at In S(T) Q t constant (N+DRT (6) higher temperatures. This equation assumes a constant relative humidity of the testing environment. where S(T) is the temperature dependent strength, Q is the activation energy of the reaction and N is the stress Eect of Temperature Treatments corrosion susceptibility described as: Figure 8 shows some tensile strength data mea- bKc N (7) sured on single filaments taken from strands of fibers that were treated at various temperatures for I h and then tested in air. Even though these samples came where Kc is the critical stress intensity factor and b is a from strands where each filament would have come in Int. Strength decreases with increase in Approxi- contact with other fibers, the data clearly shows the nately in an Arrhenius manner with an apparent effect of composition over a broad range of glasses. The
the sample sets tested in liquid nitrogen showed somewhat of a bimodal distribution where one distribution falls close to the ambient air data. This shows that even when the water is immobilized there is a failure mechanism where the energy required to create two new surfaces is met at a lower threshold (close to ambient data) and is independent of mobile water. The median strength for these Owens Corning S-glass specimens at ambient was around 5050 MPa and for liquid nitrogen was 7820 MPa. At higher temperatures, again similarly to previous studies mentioned in the earlier portions of this article, the strength of fibers is reduced with increasing test temperature. Shown in Fig. 6 are data from 1959 and recently on the same glass composition (traditional boron containing E-glass) show only minor differences probably due to increased standardization of the test methods used to create, collect, and test fibers in tensile or perhaps small differences in the E-glass chemistry. Also shown is the strength for Owens Corning S-glass tested under tension at some of the same temperatures. In addition to the decreased strengths with increasing temperature there is an interesting effect on the Weibull modulus in that it actually increases with increasing temperature (or the COV decreases). The Weibull plots of the E-glass data are shown in Fig. 7 with the values of m shown in the inset table. A possible explanation for the decreased variation in data with temperature is that at higher temperatures, the activation energy for the water and siloxane reaction (as shown in Eq. (5) is met and therefore the reaction can occur readily at any stressed bond thus increasing the probability for failure at a constant load. According to several researchers the strength dependence on temperature for most glasses can be described with an equation of the form: ln SðT Þ ¼ Q ðN þ 1ÞRT þ constant ð6Þ where S(T) is the temperature dependent strength, Q is the activation energy of the reaction and N is the stress corrosion susceptibility described as: N ffi bK c RT ð7Þ where Kc is the critical stress intensity factor and b is a constant. Strength decreases with increase in T approximately in an Arrhenius manner with an apparent activation energy equal to [Q/(N + 1)]. This decrease is caused by the increased slow crack growth rate at higher temperatures.5 This equation assumes a constant relative humidity of the testing environment. Effect of Temperature Treatments Figure 8 shows some tensile strength data measured on single filaments taken from strands of fibers that were treated at various temperatures for 1 h and then tested in air. Even though these samples came from strands where each filament would have come in contact with other fibers, the data clearly shows the effect of composition over a broad range of glasses. The Fig. 6. Median pristine tensile strength of traditional E-glass and OC S-glass fibers as a function of test temperature. Two data sets taken from previous internal work by Otto.22 Fig. 7. Weibull plots of traditional E-glass strength data as a function of temperature. www.ceramics.org/IJAGS High Performance Glass Reinforcement Fiber 115
International Journal of Applied Glass Science-Korwin-Edson, Hofmann, and McGinnis Vol. 3, No. 2, 201 10000 △Sica 。。 Basalt D ARantes 1000 Treatment Temperature"C Several ypes of glasses in the form of strands were Fig. 9. Weibull plots of the effect of direct humidity during In(strength) to high temperatures and their tensile strength measured. basalt fibers had the lowest strength and e most strength decrease over the range of temperatures studied, due to their high iron and sodium content. The data for silica is lower than expected perhaps because it does not have as good damage resistance as glasses containing alu mina or magnesium oxide(damage resistance is different than pristine strength). This is in conjunction with a study by aslanova where they showed that the most dam- age resistant glasses were those containing alumina and magnesium oxide and that vitreous silica showed severe drops in strength(50%) with minimal amount of force compared with the magnesium aluminosilicate glass fect of Humidity In (strength(kpsi) Advantexglass fibers were formed under a direct Fig. 10. Weibull plots of OC S-glass strength as a finction of humid environment(steam)and tested in the normal lab- testing humidity oratory air(35% RH)and then compared directly to fibers formed in air and tested in air using the same bush ing. The results are shown in Fig. 9. It is observed that interesting point is that the Weibull moduli shows the humidity during fiber formation has no effect on the eral trend that increases with increasing humidity different from one other set of published data sufficient humidity to cause a rapid uptake of an adsorbed showed no change in the distribution of the data with pristine tensile strength. This is because even a 35%RH is monolayer of water on the surface of the glass fiber humidity level. Proctor made no mention of it, but did show that his strength versus humidi as has been discussed in the review section humidity variation during the test itself, however, does ation as humidity increased as well l data had less vari- have an effect on the measured pristine strength. Sev- From this data a simple linear relationship ermined to predict strength of Owens Corning S- eral pristine Owens Corning S-glass fiber samples were glass based on testing%RH levels. The results are plotted in Fig. 10 with a much Strength(kpsi)=755-103 X Test %RH(8) smaller scale on the x-axis to show the difference in strength values. It is clear that the fibers show an increas- he regression had a P-value of 0.000 indicating ing strength with decreasing test humidity. The other that it is statistically significant with a confidence interval
basalt fibers had the lowest strength and had the most strength decrease over the range of temperatures studied, due to their high iron and sodium content. The data for silica is lower than expected perhaps because it does not have as good damage resistance as glasses containing alumina or magnesium oxide (damage resistance is different than pristine strength). This is in conjunction with a study by Aslanova where they showed that the most damage resistant glasses were those containing alumina and magnesium oxide and that vitreous silica showed severe drops in strength (50%) with minimal amount of force compared with the magnesium aluminosilicate glass.38 Effect of Humidity Advantex® glass fibers were formed under a direct humid environment (steam) and tested in the normal laboratory air (~35% RH) and then compared directly to fibers formed in air and tested in air using the same bushing. The results are shown in Fig. 9. It is observed that the humidity during fiber formation has no effect on the pristine tensile strength. This is because even a 35% RH is sufficient humidity to cause a rapid uptake of an adsorbed monolayer of water on the surface of the glass fiber. As has been discussed in the review section, humidity variation during the test itself, however, does have an effect on the measured pristine strength. Several pristine Owens Corning S-glass fiber samples were measured in tensile under increasing relative humidity levels. The results are plotted in Fig. 10 with a much smaller scale on the x-axis to show the difference in strength values. It is clear that the fibers show an increasing strength with decreasing test humidity. The other interesting point is that the Weibull moduli shows a general trend that increases with increasing humidity. This is different from one other set of published data which showed no change in the distribution of the data with humidity level.15 Proctor made no mention of it, but did show that his strength versus humidity data had less variation as humidity increased as well.18 From this data a simple linear relationship was determined to predict strength of Owens Corning Sglass based on testing%RH. Strength(kpsi) ¼ 755 1:03 Test % RH ð8Þ The regression had a P-value of 0.000 indicating that it is statistically significant with a confidence interval Fig. 8. Several types of glasses in the form of strands were exposed to high temperatures and their tensile strength measured. Fig. 9. Weibull plots of the effect of direct humidity during forming of Advantex® glass on “pristine” tensile strength. Fig. 10. Weibull plots of OC S-glass strength as a function of testing humidity. 116 International Journal of Applied Glass Science—Korwin-Edson, Hofmann, and McGinnis Vol. 3, No. 2, 2012