result from contact with a holder or with the fingers is avoided. With this type of holder the fibers can be removed directly from a storage reel, thus eliminating extra handling steps.

Kirk [104,311] used two types of holders; one is a holder with multiple prongs on which a fiber is supported and shaped; the other is a Y-shaped holder on which a fiber is supported. Fibers are clipped or fastened to these holders at intervals along the usuable length and can then be held in any desired position. A predesigned assembly of these holders allows for repeated construction of fiber apparatus. However, where the holders are made of heavy silica fibers or other hard materials there is some danger of scratching resulting from the contact between holder and fiber.

In some applications where fairly heavy fibers are used for a beam or supporting member, metal, carbon, or asbestos blocks appropriately grooved are used as templates. The fibers are laid in the grooves, the joints fused or cemented and the fine fibers either drawn out from the assembly or fused on in the appropriate places [28, 32, 35 to 37, 39, 40]. Joints made in this way are frequently brittle and weak because of overheating and may be contaminated by the block itself [50]. Two methods using templates as patterns or guides in the construction process have been used. The templates used by O'Donnell [102] consist of thin glass plates equipped with pegs to locate the fibers and with silica fiber clips to hold the fibers in place for fusing. The plates are held in proper positions in jigs so that all the joints can be fused through the appropriate holes drilled in the plates. El-Badry and Wilson [50] used a series of patterns drawn on matt paper. Fibers attached to microscope slides are placed in position over the pattern and the joints cemented in successive steps until the assembly is completed. All the fibers can be attached in this manner.

method to be followed in individual applications.

In the construction of spiral or helical springs, long even silica fibers are needed. Although shorter fibers have been fused together to make a fiber long enough for a spring [106] this practice may affect the performance of the spring.

Silica springs are wound on a suitable mandrel by feeding the fiber in place and heating it to conform to the shape of the mandrel. The mandrel may be tapered for ease of removal and to suit the design characteristics of the spring [107]. Several materials have been used for mandrels including carbon and graphite [55, 95 to 98, 100, 101, 106], Pyrex [105], Vycor and fused-silica [99, 107] rods or tubes. Of these, fused silica and Vycor mandrels have been considered most suitable because they have the slight adherence necessary to prevent stretching and a coefficient of expansion the same as or similar to that of the fibers wound [107]. Carbon and graphite mandrels may damage the spring by overheating and by contamination. Springs wound on carbon or graphite mandrels may have squared rather than smooth coils.

Over the years many methods of winding silica springs have been developed. In the earliest successful method the mandrel was turned about its tilted axis by hand and, as the fiber coiled about the mandrel, a flame bent it to the mandrel [55, 95, 106]. method has been greatly improved and mechanized to produce more regular springs [97, 98, 100, 105]. In the latest and most completely mechanized method, as described by Ernsberger and Drew [107], the mandrel is turned in a lathe which has controls for the fiber and flame guides to allow for uniformity in the spacing and in the coils, and for proper heating of the fiber.

As with all procedures for fabricating apparatus from silica fibers, the precautions against contamination and damage resulting from handling and from the actual fabrication process should be observed.

For many applications a conducting surface of metal or of graphite [79] on silica fibers has proved very satisfactory. The re are several methods for putting this coating

ing, plating, or baking a painted coat [2, 5, 79, 91, 94]. A coating or plated surface on the ends of fibers has also been used to facilitate holding fibers in certain apparatus.

Additional equipment for fiber work includes a binocular microscope, micromanipulators and holders, torch holders, small torches made from fine fused-silica tubing or from hypodermic needles. Detailed descriptions of the techniques of construction of fiber devices and descriptions of the equipment used in their fabrication are given by Neher [9], O' Donnell [102], Haring [103], and Kirk [104]. Individual construction details and specialized design problems are discussed most completely by the workers who developed the apparatus (see applications).

In regular fusing operations involving quartz or fused silica [299, 300] the intense

odors cause fatigue and nausea among most workers. For these reasons and because of the possibility of silicosis most workers in this country work only 2 or 4 hours with quartz or fused silica. Special glasses suc: as those using Noviweld lenses should be use: to protect the eyes from the intense infrare radiations. A suitable exhaust system shoul be used to carry away the heat and the fumes. While these discomforts have been referred to in large scale operations there is little reference to such effects when workin with silica fibers. This could be because c the smaller masses involved and the consequer: smaller amounts of heat and fumes. It is well to note, however, that there are some precautions to be taken in quartz fusing operations.

4.1 Strength

Strength is defined as the resistance of material to fracture or quantitatively as he critical value of stress at which fracure occurs [223]. The mechanical strength of used silica can be considered in three cateories: compressive strength, bending trength, and tensile strength. The compres ive strength of fused silica is 2 or 3 times hat of ordinary glasses and is about one ifth less than that of quartz [7]. The ending strength, modulus of rupture, of silca fibers is generally higher than the tenile strength [7, 188 to 190, 241] and is afected by the same factors which cause uch scatter in the values of the tensile trength. The tensile strength of silica ibers is smaller than the crushing strength. lowever, the individual values vary much more han the uncertainty of measurement which akes comparison of values difficult.

SO

The tensile strength of silica fibers exceeds that of most metals and other materils. This high strength coupled with an alost perfect elasticity makes silica fibers lesirable for many applications. However the great range of individual values for strength ind the slight loss of strength due to cercain conditions of use prevent or limit those pplications needing more than the minimum alues of strength. For this reason there is need to investigate the strength of silica ibers with special attention given to the ethods of production and to the conditions nder which they are used. Most measurements of tensile and bending strength on any quanity of silica fibers [188 to 195] were made ith fibers produced by flame, hand, bow and rrow or other simple methods. The only vailable measurements on machine drawn fiers [11] as published are not very complete.

The scattering in the determinations of ensile and bending strength of fibers is so arge that the assignment of a definite strength is difficult. A graph showing ex

erimental values obtained for tensile strength and for bending strength might ap

pear to be of some help. However an analysis shows dissimilarities in individual testing procedures, in the specimens, handling, thermal treatment, temperature and humidity controls, in the test equipment and in the presentation of the results; all of which tend to affect the values determined for strength and thus make any strict comparison misleading. The results of the various investigators show trends in the strength such as an increase in strength with decreasing diameter and length, some variations due to the atmosphere, to thermal treatment, to age, to drawing methods, to handling methods. The possible errors that may result from taking values at random from experimental curves can be illustrated by a plot of all the previously determined values of tensile strength with respect to the diameter, on a log-log scale. Here where a value of the strength for a 100-micron fiber is about 35 to 40 kg/mm2, the tensile strength given for a 10-micron fiber may range from 2 to 12 times larger than this value, while those for a 3-micron fiber may range from 5 to 100 times larger than the values given for the 100-micron fiber. The trend on these curves for fibers over 100 microns is a slow approach to a constant value, while that for fibers less than 100 microns is a rapid increase in strength in a linear manner.

In the experimental curves showing the strength of silica fibers as determined by an investigator different methods of showing the results are used. Such curves may indicate maximum values [188 to 1911, values adjusted for the presence of flaws [195], average values [192, 193], values without the conditions specified [11], and in some cases all the values determined [1, 2, 188 to 190]. Without some standardization of the testing procedure, comparisons of such values are frequently meaningless. For example the diameter measurements can greatly affect the results [307, 309] especially with small diameter fibers. An interesting discussion of the problem of correlation of strength measurements on glass is presented by Bailey [207].

been termed a "structure sensitive" property [213]. The random structure with the resulting variations in bond strength and in the interatomic distances gives glasses and especially fused silica an initially high value for strength. However, the theoretical strength is somewhat modified by certain discontinuities in the structure and by the effects of various external factors on them [116, 213]. The differences between the theoretical and the experimental strength of glasses led Griffith [199, 200] to postulate the existence of certain concentrations of energy in the form of submicroscopic cracks or flaws throughout the material. These flaws could occur either during the manufacturing process or in subsequent treatment of a glass. The effect of flaws or discontinuities is to produce local stresses which exceed the average stress and as a result lead to failure. Calculations of the effects show that they could conceivably account for the differences between theoretical and actual strength [200, 208, 223]. Proof of the existence of flaws was demonstrated in experiments [163, 164] which showed fine cracks in places on annealed fused silica rods and tubes where no accidental scratching could occur. The Griffith flaws appeared to be about 10 microns long by about 0.01 microns deep.

a. Production

The drawing process for silica fibers has some effect on the final strength. In regard to the method of production fibers drawn by bow and arrow appear to have higher tensile strength and show less scattering of values than those blown in a flame which in turn appear to have higher tensile strength and show less scatter than those fibers drawn by hand [195]. Comparative tests with machine drawn silica fibers are not available in the current literature. It is entirely probable that such fibers could show greater strength and less scatter because of the controlled conditions under which they can be drawn.

The drawing process can be regulated with respect to drawing temperature, rate of drawing, and rate of cooling and thus determine to some extent the strength [103]. The

provides a fire polished surface which resists chemical reactions and decreases or delays the effects of cracks or flaws. Recent ly the importance of the rate of cooling on commercial glass fibers was brought out by Slater [227] who found little variation in strength between fibers 0.002 in. in diameter and those 0.0002 in. in diameter when both were drawn at the same rate of cooling. The rate of cooling can be regulated in the fiber drawing machines by adjusting the drawing temperature and the drawing speed within certain limits [11]. The thermal history of fi bers as determined by drawing conditions is one of the important single factors in the strength of silica fibers.

From further considerations of the drawing process two theories were developed in order to explain the difference in strength between fibers and the bulk product from which they are drawn. One theory was advanced by Murgatroyd [172] who suggested that in order to allow the continuity of the fibe as drawn the strongest bonds were selected from the melt and alined parallel to the direction of drawing while the weakest bonds were alined perpendicular to the direction of drawing. In the other theory Bickerman and Passmore [220] believed that the flaws in a fiber must be oriented favorably to enable the fiber to be drawn. These theories emphasize the importance of the drawing process rather than the reduction of size which accompanies it. Although there are many who support or have supported these ideas [143, 149, 173, 213, 215, 222], recent experiments [148, 161] on annealed glass fibers indicate that there is neither an orientation of bonds nor one of flaws in fibers which could be responsible for the high strength. Additional evidence showed that an oriented structure was not evidenced by X-ray studies [160] and that if an orientation at the surface existed, its influence on the strength could not be detected [193].

b. Size

Experiments on the breaking strength of silica fibers show an increase in strength with decreasing fiber diameter [1, 11, 188 to 195]. Similar results occur with other glass

all area is tested [236]. An additional crease in strength is found when the specin length is decreased [190,220]. Assuming at flaws exist in fibers, then the apparent pendence of strength on fiber size can be plained in a statistical manner by the deeasing probability of effective flaws ocrring as the volume or the surface of a fier is decreased [193, 213, 214, 220]. Alough there is some basis for assuming a retion between the probability of flaws ocrring in relation to the size of a specimen, exact expression for the distribution inction of the cracks or flaws is difficult > formulate [223].

Orowan [223] considered that the necesarily arbitrary assumptions as to the size f cracks and to the number of cracks in a iven volume or over a given area without acounting for the individual history of the racks led to results "more interesting mathnatically than correct physically." Howver, the thinnest silica fibers which have een investigated have diameters less than he supposed size of an ordinary Griffith rack and the strength of these fibers apears to approach the theoretical values deermined for the material. These fibers ould not contain flaws as large as those in arger specimens and unless the flaws were arge enough to weaken the fiber by reducing ts cross section, the strength, Orowan conluded, should be higher than that of a thick od. The consideration of the number and ize of flaws in a given length particularly ffects measurements of bending strength here the probability of flaws occurring at a ertain point of maximum stress is smaller han that of flaws occurring over a given ength, with the result that values for bendng strength are generally higher than those or tensile strength. In addition, the nonniformity in strength over the entire length. f a fiber [241] due to the existence of laws is a possible cause for the great range f values determined for any one size of fiA well-designed statistical test proceure might help to remedy the situation.

er.

c. Surface

The large surface area in comparison with the volume of fibers may account for the

treatment on the strength of silica fibers. The chemical and physical nature of the surface of glass has been described as different from that of the body of a glass [160]. Structurally, this influence of the surface can be explained by a broadening of the bondstrength distribution near the surface which tapers off toward the interior and leaves weak bonds at the surface. Although flaws are considered to be distributed throughout the material, the strength is determined by the most dangerous flaws on the surface [193, 199, 2001, for a surface flaw can exert twice the stress of the same sized inner flaw [221]. The stress variations in individual fibers resulting from the existence of flaws, scratches on the surface, and the effects of surrounding conditions on them, may cause fracture before the theoretical elastic limit is reached. Because of the effects on strength of these conditions it is difficult to bring values of individual fibers within a small range of error. It may be appropriate to mention here the often quoted idea of Little that one measures not the strength of a material but the weakness of its surface. A discussion of the chemical and physical aspects of the surface condition of glasses is presented in a series of works by Weyl [116, 175, 213].