International Journal of Fracture Mechamcs, Vol. 5, No. 3, September1969 Wolters-NoordhoffPublishing Groningen Printed in the Netherlands
179
Constitution and Strength of Glass Fibers G. M. B A R T E N E V Department of Phystcs of Solids, Lento State Teachers' Trainm9 Universtty, Moscow, USSR
(Received August 1, 1968)
ABSTRACT Five characteristicranges of glass strength are described wath respect to the sizes of its assocmted surfaceflaw. Particular attention is directed to the substantial differencein behavior between commercialand flawless glass fibers. It has been discoveredin the latter case that a surface layer one order of magmtude thinner ~an normally detected is present, probably as a consequenceof the drawing process. Distract differencesm behavior of the two kinds of layers was noted by using ultra-violet hght absorption and hydrofluoric acid etching. Basic strength of the flawless glass fibers in vacuum can reach one million pounds per square inch. Introduction Inorganic glasses are multi-component non-crystalline solids. The main components of glasses are glassformer-inorganic polymers with chain-like or network structures. Therefore, glasses possess certain features of a polymer structure and have corresponding physical properties. Nevertheless, the structure is complex. However, glasses may be characterized by three specific supramolecular structures. In ranges slightly exceeding the atomic (short-range order), the mutual arrangement of ions in glasses is relatively ordered and resembles the arrangement of ions in crystals. On a somewhat larger scale, glasses are observed to have chain-branched structures of glass formers, formed primarily by directed ionic-covalent bonds. As a whole, these structures form one or several frameworks throughout the total volume of the glass. When passing to microvolumes of the largest size, regions of different chemical composition are observed. These areas at microsegregation are sometimes so intense that the glass undergo phase transitions, lose their transparency, and then crystallize under certain conditions. All three types of these specific structural features can be modified by changing the chemical composition of the glass, the conditions of manufacture, or the conditions of heat treatment, respectively. Structural Ani~tropy in Glass Fibers
The polymeric structure of inorganic glasses must result in an anisotropy of the structure and properties of glass fibers. During drawing of glass fibers, orientation of Si-O and other strong bonds of chain structures of glass-formers must occur. However, some investigators have observed, and others have not observed, the expected anisotropy [,-1-8]. Similar contradictory results are possibly explained by the fact that the structural anisotropy of inorganic glass fibers is displayed less markedly than that of organic fibers. Structural anisotropy has been discussed in several papers. Optical anisotropy of glass fibers was observed by Merker [-3] who showed that it was not caused by internal elastic stresses. Using x-ray techniques, Brfickner [-4] observed an oriented crystallization and asymmetry of the density in glass fibers. The diamagnetic anisotropy discovered by Bannerjee [-6] in B203 glass fibers, and by Tarasov and Semenov [-7] in two-component B203" NazO glass fibers is further evidence of structural anisotropy. Shishkin [8] found a direct correlation between the strength and the birefringence of some glass fibers. Recently during an investigation of the nuclear gamma-ray-resonance (M6ssbauer effect) of massive glasses and glass fibers at our laboratory it was discovered that the curve of adsorption of ?-quanta by Sn-nuclei which were introduced into the glass, is asymmetrical for glass [nt Jow'n. of Fracture Mech., 5 (1969)179-186
180
G, M . Bartenev
fiber and symmetrical for massive glass (Fig. 1). These results also lead to the conclusion that glass fibers appear to have structural ahisotropy, but massive glasses are essentially isotropic. Because structural anisotropy in silicate glass fibers requires exceptionally careful techniques to detect it, it is likely that the structural anisotropy is weak. It therefore cannot be the main
i
•
O
0,5
Figure 1. Mossbauer's spectra of y-radiation absorption: (1) by massive alkah-silicate glass containing SnO; (2) by glass fiber made of the same glass, 10 # m diameter.
//.~
I00
Figure 2. Effectof chameteron the strength of flawlessalumina borosilicateglass fibers: (1) prior to chemicaletching, (2) after chemicaletching Given also are data on glass fiber of the same chemicalcompositionafter chemicaletching: (3) data by Sakka [9], (4) data by Bovkunenko [23]. reason for the high strength of glass fibers. This conclusion agrees with the observed weak dependence of the strength upon the diameter (degree of drawing of flawless glass fibers prior to and after chemical etching by hydrofluoric acid) (Fig. 2). It can be seen that the strength of flawless glass fibers prior to and after chemical etching is different. This weak dependence can be explained by the slight change of the structural anisotropy associated with changes in the degree of drawing during molding. The Strength of the Structure of Silicate Glasses and Glass Fibers The conclusion that structural orientation throughout the volume of the glass fiber is not great and therefore does not have a substantial effect on its strength is in agreement with the experimental results of Sakka [-9], Mrs. Izmailova [-10], Mrs. Motorina [-11], and Chernjakov [12] from long-time heat treatment of glass fibers and subsequent chemical etching. During heat treatment the structure of the glass is annealed, and the anisotropy of the structure disappears. However, for practical purposes this treatment has no effect on the strength of the structure Int. Joz~rn. of Fracture Mech., 5 (1969) 17%186
Constitution and strength of gIass fibers
181
TABLE 1
Strength data for alkali-sihcate masswe glasses (hydrofluoric aczd solution etch) Date
1960 1961 1962 1962 1962 1963 1965 1966
Source
Proctor Holloway Simmers Cornelissen and Zxjlstra Proctor I~tter and Cooper Witman Witman
Reference
Strength
Specimen
No.
(kgf/mm~)
type
24 25 26 27 28 29 14 14 Average : 220
200 250 190~200 200 235 220 240 230
8-mm rods 0.5-mm 0.5-ram 2-3-ram rods rods 0.7-mm Z5-mm sheet glass Z5-mm sheet glass
TABLE 2
Strength levels of commercial sthcate 9lasses and glass fibers (Atmospheric conditions, 20°C ; loading time approximately 1 minute) Strength levels
Approximate values of strength levels (kgf/mm 2) Massive glasses
Glass fibers
a, o0
5~ ~15
Absent 8-15
crI
az
Possible 20(0250
6~80 200-250
a3
Absent
300-350
Type of defects by typical size
Macrocracks occurred due to mechanical processing (ram) Microcracks occurred at heat treatment of glass fibers or during molding of massive glasses ~) Submlcrocracks occurred during molding of glass fibers, (0.01 #) Microruptures of surface layer occurred during drawing of glass fibers; Flawless massive glasses Flawless glass fibers
throughout the volume of the glass fiber, though it considerably decreases the strength of its surface layer. If the structure of the glass at high-temperature is of considerably greater strength than that of the annealed glass, then the strength of the glass fiber and massive glass after removing the imperfect surface layer by chemical etching would differ greatly. The facts do not confirm this prediction. For instance, rods and sheets of alkali-silicate glass, after careful selection of samples, chemical etching, and taking every precaution not to damage the glasses' surface, appear to be essentially of the same strength and in the range of 200-250 kgf/mm 2, according to data by many investigators (Table 1). The same level of strength is observed in glass fibers after chemical etching [10o13]. In Table 2 this level of strength, which characterizes the strength of the structure of silicate glasses, is designated by a2 and pertains to testing in the air environment. Its true value must be greater because humidity in the air, being a surface active medium, tends to decrease the strength. According to data by Witman et aI. [14], tests in vacuum lead to strength increases in a chemically etched sheet glass from 230 kgf/mm 2 to 485 kgf/mm 2, i.e. a factor of two. In our laboratory, similar results have been obtained for glass fibers. Therefore, the real strength of the structure of silicate glasses corresponds to the value of the strength level a 2 ---450-500 kgf/mm 2 in vacuum_ It is this value of strength which the interior layers of the glass fiber develop because the surface layer protects them from moisture in the air. Flawless Glass Fibers
Flawless glass fibers, flee of surface defects, can be produced by the drawing process proposed by Mrs. lzmailova and the author [15]. They possess several unusual properties [10, 16] : Int. Journ. of Fracture Mech., 5 (1969) 179-186
182
G. M. Bartenev
independence of strength on length (curve 1 in Fig. 3), extremely small scatter of strength (1-2~), and an explosive fracture mode when they suddenly split into fine particles (glass dust). The strength of flawless silicate glass fibers, designated as o 3 in Table 2, is 300-350 kgf/mm 2 t
eq
/
_o "
c~
o ,
o
),9 200
198
F~gure 3. Effect of length on the strength of alumina borosflicate glass fibers : (1) flawless, (2) commercaal, (3) after chemical etching.
in air, depending on the chemical composition of glass fibers, and again about twice as high in vacuo (600-700 kgf/mm2).
The Structural Surface Layer From general considerations it is suspected that the structure of the interior of the glass fiber and its surface layer differ because of different cooling conditions in its interior and on its surface. In this connection Griffith hypothesized the existence of a thin surface layer having a different structure than that of the interior of the fiber. There was no experimental confirmation of this hypothesis, and it was forgotten. Only now can we understand why this surface layer is so difficult to detect experimentally. Its thickness, which we discovered at our laboratory because our flawless fibers did not have the damaged surface layer usually observed in commercial glass fibers, is one order of magnitude less than customarily encountered. Our early experiments with chemical etching have already shown that flawless glass fibers possess an outstanding property. Whereas glasses and glass fibers increase their strength after chemical etching, the flawless fibers behave in an opposite manner; their strength decreases after chemical etching, and moreover it decreases greatly (by 100 kgf/mm2). The results of experiments with hydrofluoric acid solution etching are given in Figs 2-4. Using light absorption techniques in the ultra-violet range, it has been discovered [12, 16] that as thin surface layers are removed by chemical etching, the absorption coefficient, like the strength, sharply decreased. After the depth of etching reached 0.01~0.02 #, it becomes stabilized and if etching proceeds further, the light absorption and the strength do not change (Fig. 4). The behavior of sheet glass is- quite different. In this case the coefficient of light absorption prior to and after chemical etching is the same, but as the etching proceeds, its strength will slowly increase till the depth of the layer removed is 50/a The decrease of strength by chemical etching can be explained either by the removal of the strengthened surface layer or by damage caused to the natural surface due to the etchant. It is probable that both processes occur simultaneously, but that the second process is less important because in the case of the damaged natural surface of the fiber the coefficient of absorption would have to increase, not decrease. It is supposed that the surface layer forms when the fiber is molded, because the temperature of its surface is lower and its viscosity higher than its interior. The temperature gradient near the surface of the "bulb" is the greatest. The viscosity gradient of the surface layer is still greater because of the exponential dependence of the viscosity of the glass on temperature. As a result, the maximum stresses during the drawing of glass fibers are concentrated in the thin surface layer with a viscosity exceeding substantially the viscosity of the interior of the "bulb". Int. Journ. of Fracture Mech., 5 (1969) 179-186
Constitution and strength of glass fibers
183
~a
~
J00
L
t
o/-O
O
J
4,2
/ -2 o
/n
0
Figure 4. Change in the strength and the coefficient of light absorption m the ultra-violet range (2300 A) for alumina siltcate glass-fibers after chemical etching.
As a consequence, the surface layer which bears the main force of drawing is subject to higher degrees of drawing and structural orientation. This process can lead to an appreciable strengthening of the surface layer. It must be noted, however, that it is not experimentally easy to discover the structural anisotropy of this surface layer by using modem methods. By analogy with the properties of organic polymers [17] it can be supposed that the orientation of the glass structure in the surface layer does not lead to changes in Young's modulus. If then, the elastic constants of the surface layer and the interior of the glass fiber are essentially the same, then at fracture the stress throughout the cross-section of the glass fiber is the same. As the strength of the glass structure in the interior of the glass fiber is approximately 450-500 kgf/mm2, fracture of glass fibers in moist air, including flawless glass fibers, starts from the surface at stresses of 3130-350 kgf/mm2 or less. Levels of Strength Under ambient conditions, defects on surfaces of glasses and glass fibers are most severe, owing to the effect of moisture. Surface defects can be of different types, and each type can cause a more or less characteristic level of strength associated with a particular type of defect, or lack there of. In several publications it was established that glass fibers [10, 16, 18] and sheet glasses [-18-20] have five different strength levels (see Table 2), a,, ao, ~1, a2, and a3. The lowest level of strength a. is observed in sheet glasses after mechanical processing (cutting, grinding), during which macrocracks occur, comparable to the thickness of the glass. This level of strength is hardly affected by the size of macrocracks. It is not observed in glass fibers, as the size of macrocracks exceeds by far the diameter of glass fibers. Figure 5 represents the structure of glass fibers and different types of surface defects associated with the next levels of strength, ao, ai, and a2. The level of strength ~ro corresponds to the strength of the surface of commercial sheet glasses having microcracks, initiated during drawing under the effect of thermoelastic stresses. In glass fibers this level of strength is found to result from heat treatment, which leads to initiation of surface microcracks penetrating to a depth comparable to the radius of a glass fiber (Fig. 5). The depth of microcracks in glass fibers, as measured by the method of chemical etching, is 0.5~.6/~ This level of strength for both sheet glasses and glass fibers is very indistinct because of the presence of microcracks of different sizes in samples which are grouped round this level of strength~ Strength level ~1 is explained by the existence of finest surface submicrocracks generated during the drawing of commercial Int. Journ. of Fracture Mech., 5 (1969) 179-186
184
G. M. Bartenev
I
Figure 5. A schematic r~.'presentation of the axial section of glass fibers (R= 5 #) wath a structural surface layer (0,01 #) and different types of surface defects.
ao3~
0,02-
t 40O
Figure 6. Curves of strength distribution (t ogether with errors due to measurements) for al,,mina borosilicate glass fibers when passing from flawless glass fibers having the strength of 310 kgf/mm 2 (curve 1) to imperfect commercaal glass fibers having an average strength of 193 kgf/mm z (curve 2), 180 kgf/mm 2 (curve 3), 150 kgf/mm 2 (curve 4), and to heat-treated glass fibers having a strength of 80 kgf/mm 2 (curve 5). Int. Journ. of Fracture Mech., 5 (1969) 17%186
Constitution and strength of gIass fibers
185
glass fibers. Their depth is less than the thickness of the surface structural layer 0.01 # (Fig. 5). This level of strength also is indistinct, as there are samples having submicrocracks of different sizes. Submicrocracks are likely to occur also in massive glasses at junctions of microheterogeneities, associated with the fact that the structure of the glass contains fine microheterogeneities with linear sizes about 0.01 #. The level of strength az is observed in commercial high-strength glass fibers which were not chemically etched. Their surfaces are supposed to have microruptures of the structural surface layer which occur during the drawing of glass fibers and expose the interior lowstrength structure of the glass fiber (Fig. 5). This level is realized in glass fibers and rods and in sheet glasses after the removal of surface microcracks by chemical etching. Finally, the highest level of strength ~r3 characterizes the strength of flawless glass fibers which have the structural surface layer undamaged. The three highest levels of strength ~r1, a2, ~r3 are readily seen on the curves of distribution of the strength of glass fibers when passing from flawless glass fibers to very imperfect glass fibers (Fig. 6). It is seen how, when moving from high-strength to low-strength glass fibers, the maximum corresponding to the level of strength a3 at first gradually disappears, then the maximum corresponding to the level of strength az, and finally the lowest level of strength al remains. According to Sidorov [22], the same behavior is observed on the distribution curves of commercial glass fibers when moving from short to long lengths (Fig. 7).
I
I
I
,
Figure 7. Curves of strength distrtbution for alumina silicate glass fibers (10 ,u) when passing from short lengths to big lengths, (I) 3 turn; (2) 10 rnrn (3) 50 mm, (4) 150 ram, and (5) 400 mm Int. Journ. of Fracture Mech., 5 (1969) 179-186
186
G. M. Bartenev
From the data given above, it follows that the observed levels of strength are connected with different states of the structure of the surface and of the interior of glass fibers produced under different conditions of drawing and heat treatment.
REFERENCES [1] W. Otto and F. Preston, J. Soc. Glass Techn., 34 (1950) 65T. [2] G. Slayter, Amer. Cer. Soc. Bull., 31 (1952) 276. [3] L. Merker, Syrup. sur la r&istance rt~camque du verre, Florence (1961), (Compte rendu, Charleroi, 567-587 (1962)). [4] R. Brfickner, Zur Struktur der Glasfasern, VII Int. Congr. on Glass, BrusseLs (1965). [5] K. Oscar, Silikattechnik, 16 (1965) 281. [6] B. K. Bannerjee, GIastechra Bet., 33 (1960) 8. [7] V. V. Tarasov and L. V. Semenov, Glass State, IU (1962) 2, 52 54, Leningrad. [8] N. I. Shishkin, Glass State, III, (1%2) 2, 54-55, Leningrad. [9] S. Sakka, Bull. Inst. Chem. Res. Kyow, 34 (1957) 316. [10] G. M. Bartenev and Mrs. L. K Ismmli~va, Pro< Acad. Sci. (USSR), 146 (1962) 1136; Phys. SolMs (USSR), 6
(1964) 1192. [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29]
G. M. Bartenev and Mrs. L. I. Motorina, Proc. Aead. Sci. (USSR), 155 (1964) 1302. G. M. Bartenev and R. G. Chernjakov, Proc. Acad. Sci. (USSR), 174 (1967) 800. G. M. Bartenev, Glass and Ceramica (USSR), 8 (1967) 4. F. F. Wltman, G. S. Pougachev, and V. P. Poukh, Phys. Solids (USSR), 7 (1965) 2717; Inorganic Marls. (USSR), 2 (1966) 194 Mrs. L. tC Izmailova and G. M. Bartenev, Glass and Cerarn~a (USSR), 3 (1964) 12. G. IVL Bartenev, The Cher~ Enors., 182 (1964) CE249; Silikattechnik, 18 (1967) 315. G.M. Bartenev and Yu. S. Zuyev, The Strength and Fracture of Hioh-Elastic Materials, Publ. House ~Chemastry" Moscow (1964); Strength and Failure of Vtsco-Elastic Materials, Pergamon Press, Oxford (1968). G. M. Bartenev, The Structure and Mechanical Properties of Inorganic Materials, Publ. House on Construction, Moscow (1966); Wolters-Noordhol~ Groningen (1969). G. M. Bartenev and L. P. Tsepkov, Physico-Engr. J. (USSR), 2 (1959) 7, 20. G. M. Bartenev, Factors Lab. (USSR), 26 (1960) 1136. G. M. Bartenev and Mrs. A. I. Kolbasnikova, Glass and Ceramiea (USSR), 11 (1964) 10. G. M. Bartenev and A B. Sidorov, Slhkatteehnik, 16 (1965) 347 ; Mechanics of Polymers (USSR), 1 (1966) 74-81. G. M. Bartenev and A. N. Bovkunenko, Phys. Chem. (USSR), 29 (1955) 506. B. Proctor, Nature, 187 (1960) 492. D. Holloway and P. Hastilow, Nature, 189 (1961) 385. C. Symmers, Y. Ward and B. Sugarmem Phys. Chem. Glasses, 3 (1962) 76. J. Cornelissen and A. Zijlstra, C. R. Sympos~wm Rdsistanee Mdcanhtue du Verre, Charleroi, (1962) 337-358, B. Proctor, Phys. Chem. Glasses, 3 (1962) 7. J. PAtter and A. Cooper, Phys. Cherm Glasses, 4 (1963) 76.
R~SUMI~ On d4crit cinq cat4gones de fibres de verre et leur r6sistance en fonctaon des dimensions des d6fauts de surface qu'elles comportent. L'attentlon est particuli6rement attir~e sur les diffdrences substanfielles de comportement que prdsentent les fibres de verre courantes et les fibres de verre sans d6fauts. On a pu d6terminer que ces dernibres comportaient une couche superficielle dont l'4paisseur est d'un ordre de grandeur plus faible que celle que l'on detecte habimellement, cette couche 6tant probablement due au proc4de de fabrication par tinge. Des differences nettes de comportement de ces deux sortes de couches ont 6t6 observ6es, en mesurant l'absorption des rayons ultraviolets et en recourant ~ une attaque It l'acide hydrofluorique. La r~sLstance de base de fibres de verre sans d6fants peut atteindre daus le vide un milfion de psi.
ZUSAMMENFASSUNG Fllnf charakteristtsche GlasstSrkebereiche warden mit Riicksmht auf cite Gr613en ihrer dazugeh6rigen Oberfl~chenfehlerhaftigkeit beschrieben. Besondere Aufmerksamkeit wurde auf den wesentlichen Unterschied im Verhalten zwischen kommerziellen und fehlerfremn Glusfibern gelenkt. Es wurde im letzteren Fall festgestellt, dab eine Ober~ flfmhensctncht, ein Gr6BenverhS~ltnis dfinner als gew6hnlich beobachtet, gegenwgtrtig ist. Dieses ist wahrscheinlich eine Folge des Ziehungsvorganges. Deutliche Unterschiede im Verhalten der beaden Schichten wurden bemerkt. Man gebrauchte ultra-vlolette Lichtaufnahme und Fluorwasserstoffsiiuere ~tzung. Die fundamentale Stiirke der fehlerfreien Glasfibern im Vakuum kann eine Million Psi erreichen. Int Journ. of Fracture Mech., 5 (1969) 179 186