Coupling agent and glass fibers in polyester mortar.

INTRODUCTION

Polymer mortar is a mixture of aggregates and a polymerizing monomer. The high-strength, rapid-setting, and corrosive resistance makes the mortar a potential material for use in structural repairs and in new constructions (1). Polymer mortar is also used in overlays (bridge decks, floors), precast products, machine tools, grouting, and waste encapsulation (2, 3).

Polyester is one of the most common polymer used for repairs and in constructions. The wide spectrum of applications of polyester mortar requires further improvement in its mechanical properties. Also, polyester mortar exhibits brittle failure under normal working conditions (4, 5). To improve the post-peak behavior and toughness, glass fibers can be added to the polyester mortar.

There is increasing interest in the quantification of the effects of glass and steel fiber addition on the behavior of polymer mortar (4, 6, 7). Brockenbrough and Patterson (6) used both glass and steel fibers in an acrylic resin-based mortar and observed that while steel fibers increased the compressive, flexural and splitting tensile strength. But it was observed that the glass fibers decreased the compressive strength, but increased both flexural and splitting tensile strength. Ohama and Nishimura (7) noted an increase in compressive, flexural, and impact strength of polyester polymer mortar with steel fibers. Vipulanandan, et al. (4), have concluded that the addition of glass fibers increased the strength (made with polyester resin and Ottawa sand with uniform size aggregates) in compression, flexure, and splitting tension.

The interfacial bond in the polyester mortar may be enhanced by using coupling agents (or adhesion promoters). Many types of coupling agents are available and the choice is dictated by the compatibility between the aggregate and the type of polymer matrix used. The organofunctional silanes are noted for their abilit to bond organic polymer systems to inorganic substrates. The silane coupling agent ([gamma]-methacryloxypropyltrimethoxysilane, [gamma]-MPS) was used for enhancing the polyester mortar strength. Vipulanandan and Paul (8) reported a 14% increase in compressive strength from 72.0 to 82.1 MPa (10,500 to 11,900 psi) and 35% in splitting tension for polyester mortar with uniform Ottawa 20-3 sand with an average particle diameter of 0.6 mm. Using the [gamma]-MPS in the polyester mortar systems Vipulanandan and Dharmarajan (9) reported an increase of 53% in flexural strength for unreinforced mortar and 38% increase for reinforced mortar with glass fibers, but the study was limited to a uniform san mainly composed of quartz.

In this study, the effect of polymer content, type of aggregates, glass fibers, and silane coupling agent on the compressive and flexural strength of polymer mortar are investigated at room temperature. A crushed sand (mixture of sandstone and limestone) and blasting sand (mainly quartz) were used as aggregates. The polymer content was varied between 10% and 18% and glass fibers were added up to 6% by weight of the unreinforced polymer mortar. The failure patterns on the specimen surface after compression tests are mapped and the typ of failure (shear, splitting or bulging) are quantified using a simple relationship. Also a stress-strain relationship is proposed to represent the complete stress-strain curves.

EXPERIMENTAL PROGRAM

Four systems of polyester mortar were prepared in this study: unreinforced polyester mortar (PM) containing polyester and either blasting sand or crushed sand, glass fiber reinforced polyester mortar (GFRPM), silane treated polyester mortar (PM-S), and silane treated reinforced polyester mortar (GFRPM-S). The crushed sand (mixture of sandstone and limestone) particle size varied from 0.1 to 5 mm (0.004 to 0.2 inch) with a coefficient of uniformity (Cu) of 3.7. The blasting sand (mainly quartz) with sub-angular particles, had a coefficient of uniformity of 5.8, and was obtained by mixing various sizes of blasting sand which are commercially available. The sand particles had a specific gravity of 2.65. The average diameter, [d.sub.50] (50% of sand are less than this size), o this blasting sand mix was 0.60 mm (0.024 inch). Chopped glass fibers were 0.01 mm in diameter and 13 mm (0.5 inch) long. The glass fiber had up to about 800 bonded glass strands. The polyester resin had a viscosity between 40 and 50 poise at room temperature and a specific gravity of 1.07. In the case of PM-S o GFPM-S, the sand and glass fibers were treated by wetting them with 2% aqueous solution of the silane coupling agent. The treated sand and fibers were allowed to dry at 100 [degrees] C for one day prior to mixing with polyester. The initiator used with the polyester was methyl ethyl ketone peroxide and promoter was cobalt naphthenate. In preparing PM specimens, cobalt naphtenate was first added to the polyester resin and the solution was mixed for at least 2 min before adding methyl ethyl ketone peroxide. After further mixing, the dry sand mixture or sand-fiber mixture was slowly added to the polyester and mixed long enough to obtain a uniform mixture. The mixture was then poured slowly into the cylindrical teflon molds to prepare specimens 38 mm (1.5 inch) in diameter and 75 mm (3 inch) long. Beam specimens (50 x 50 x 230 mm or 2 x 2 x 9 inches) were also cast for flexure tests. The specimens were allowed to cure for a day at room temperature and then at 75 [degrees] C for another day. From earlier studies on similar aggregate-polyester mix (8), it was demonstrated that this curing condition gave the best results. Before testing, the end of the cylindrical specimens were trimmed using a diamond saw to ensure smooth and parallel surfaces. For all monotonic loading, displacement-controlled, compression and flexure tests, a closed-loop servo hydraulic testing machine with an axial capacity of 22,680 kg (50 kips) was used. Load-axial displacement responses were recorded using X-Y recorders and a computer. The deflection of the beams was measured using a LVDT (linear variable differential transducer) with an accuracy of 2.5 x [10.sup.-3] mm ([10.sup.-4] inch). The axial strain i the cylinders during the compression test was measured using two extensometers, of 25 mm (1 inch) gage length, with strain accuracy of 3.5 x [10.sup.-5]. Several systems have been considered in this study where the polyester content varied from 10% to 18% and the fiber content varied from 0% to 6% (by weight of PM).

Table 1. Polymer Mortar Composition.

1)Polymer Matrix                               Content (by weight)
Resin: polyester (Dion Iso-6315)               10%-18% (PM)
Promoter: cobalt naphtenate                    0.3% (Resin)
Initiator: methyl ethyl ketone peroxide        1.5% (Resin)

2)Aggregates
Blasting sand (mainly quartz)                  82%-90% (PM)
Crushed sand (sandstones and limestone)        82%-90% (PM)

3)Fibers
Chopped glass strands                          0%-6% (PM)
(13 mm long and 0.013 mm in diameter)

4)Silane Coupling Agent
[gamma]-methacryloxypropyltrimethoxysilane     2% (Solution)

Notations
Polymer mortar                                 PM = 1) + 2)
Glass fiber reinforced polymer mortar          GFRPM = 1) + 2) + 3)
Silane treated polymer mortar                  PM-S = 1) + 2) + 4)
Silane treated reinforced polymer mortar       GFRPM-S = 1) + 2) + 3) + 4)

RESULTS AND DISCUSSION

The 14% mortar system with blasting sand was the best system (based on the compressive strength) and about 100 specimens were tested to quantify the statistical variation in the results. The effects of glass fiber, silane treatment, size, and type of aggregates on some properties of PM are discussed below. The type of failure and the stress-strain predictions are also discussed

Statistical Evaluation

Compressive Strength

The distribution of the compressive strength of 75 specimens (14% PM) is represented graphically by a histogram shown in Fig. 2a. The data are also depicted on a normal probability plot. Linear relationship with coefficient of correlation of 0.98 indicates that a normal distribution adequately represents of the test data. The mean is estimated as the compressive strength at the 50% level on the probability scale, which is 63.2 MPa (9170 psi). In Fig. 2b one standard deviation in strength is defined by the increment between the 50% and the 84% probability levels. The standard deviation is estimated to be 2.6 MPa (377 psi) and the coefficient of variation is 4%. With 95% confidence it may be predicted that the mean compressive strength is in the range of 62.6 MPa (9080 psi) to 63.8 MPa (9260 psi).

Flexural Strength

The distribution of the flexural (four-point bending tests) strength of 24 specimens is represented graphically by a histogram in Fig. 3a. The data are also plotted on a normal probability scale in Fig. 3b and may adequately be represented by a normal distribution. The mean flexural strength was 16.45 MPa (2390 psi) and the standard deviation was 0.84 MPa (120 psi). The coefficient o variation (C.O.V.) of flexural strength is 5%, compared to 4% for compressive strength.

Effect of Glass Fibers

Compressive Properties

The use of low-weight fraction of fibers was essential to ensure a workable mixture and to achieve a homogeneous distribution of fiber within the PM composites. Figure 4 shows that the compressive properties of GFRPM are very much dependent on the fiber and polyester contents. Test results suggest that there is an optimum fiber content (based on maximum strength) for each polyeste content. It can be observed from Fig. 4 that for the PM the optimum polyester content is 14%. This optimum also had the lowest porosity as observed in Fig. 5 The failure mode for most of the GFRPM under uniaxial compression was shear. However, local failures were observed when the amount of fibers in the PM was 6 for the 10% PM. For the fiber and polyester contents investigated, the compressive strength varied from 33 MPa (4800 psi) for the 10% PM with 6% fiber content to 84 MPa (12,200 psi) for the 18% PM with 4% fiber content.

Glass fibers were added to improve the toughness or energy absorption capacity of the PM. ASTM C1018 (10) procedure was followed to determine the toughness indices [I.sub.5] and [I.sub.10] for compression. [I.sub.5] represents the rati between the area under the curve up to three times the first crack to the area up to the first crack deflection. In the case of [I.sub.10], the endpoint is 5. instead of 3 times the first crack. For elastic (up to first crack) plastic material [I.sub.5] is equal to 5 and [I.sub.10] is equal to 10. Variation of these indices with fiber content is shown in Fig. 6. As can be seen, the toughness was substantially increased with the addition of fibers. [I.sub.5] wa greater than 5 and [I.sub.10] varied between 7 and 13. The ratio [I.sub.10]/[I.sub.5] varied from 1 to 2.2, and for an elastic-perfectly plastic material [I.sub.10]/[I.sub.5] is equal to 2. Figure 7 shows that the fracture stresses for the 14% and 18% PM systems remains almost unchanged. However, the 10% PM system shows a significant decrease because of the relative high fiber content, which also decreased the strength of this system as was shown by Fig. 4.

Flexural Properties

Figures 8 through 12 are the results of specimens tested in four-point bending. As can be seen from Fig. 8a, the flexural strength increases with the increase in the amount of fibers. Addition of 6% fiber into the 18% polyester PM resulte in flexural strength of 28 MPa (4060 psi) and flexural modulus of 10.5 GPa (152 ksi). This is more than 60% increase in strength over the PM. As compared with the PM, the strain at failure improved by about 300%. Figure 8c shows that fibers do not have any effect initially. This is confined by the insensitivity of the initial modulus on the fiber content. The surface failure patterns were mapped as shown in Fig. 9. Inspection of the failure surface showed that failur occurred mostly by pullout of the fibers. Increasing the amount of glass fibers increased the pullout failure and hence increased nonlinearity of the mapped curve representing failure cracks. The flexural stress-strain curves are shown in Fig. 10. It can be observed that the toughness increases with the increase o glass fibers. Nonlinearity is observed for the ascending part of the curves representing the GFRPM.

The toughness indices in flexure for the 18% PM system are shown in Fig. 11 and the first crack stresses are shown in Fig. 12. The trends in either compression or flexure is similar for the 18% PM system. Addition of fibers did not change the first crack stresses.

Effect of Silane Coupling Agent

The results shown in Figs. 13 to 15 are averages of two to four tests. As seen in Fig. 13, there is an overall increase in the flexural (three-point bending) strength, due to the silane treatment of the blasting sand. The 10% PM system shows the highest percentage increase in flexural strength (62%). As shown in Fig. 13 the addition of silane to 10% PM-S showed higher flexural strength (25. MPa (3720 psi)) as compared to the untreated 14% or 18% PM systems (20.8 MPa (3020 psi)). Use of less polyester in the 10% PM-S will result in reduced cost-to-strength ratio. The role of silane was not only to strengthen the polyester adjacent to the filler (sand) by forming interpenetrating networks that promote adhesion, but also to increase the consistency of the mix by actin therefore as a wetting agent to uniformly spread the polyester during mixing. The fiber pullout was the major failure mode for the untreated beam PM specimens. In contrast, in coupling agent containing systems (PM-S), cohesive resin failure was a more predominant feature. The effect of silane on the flexural (three-point bending) strength of glass fiber reinforced PM (GFRPM-S) is shown in Fig. 14. The following were observed:

1) All PM systems showed an increase in flexural strength when the blasting san and glass fibers were treated with the silane coupling agent.

2) In general, silane treatment resulted in higher relative flexural strength for the 10% PM system.

3) The relative increase in flexural strength varied between about 2% to 70% depending on the polyester and fiber contents. The lowest increase was for the strongest system of 41 MPa (5930 psi) and the highest for the weakest system of 12.6 MPa (1830 psi).

Silane coupling agent had a greater effect in compression than in flexure. Figure 15 showed an increase of 66% in compressive strength for the 14% PM system as compared to the 28% increase in flexure. The importance of silane is evident when comparing the treated PM with the compressive strength of the pure polyester. Using only 14% polyester with 86% treated aggregates, it was possibl to achieve a compressive strength almost equal to the pure polyester polymer (110 MPa or 16,000 psi). Compressive strength of 103 MPa (15,000 psi) was recorded for the 14% PM-S system, which also had the lowest porosity (2%). The standard deviation of compressive strength was 2.7 MPa (400 psi).

Effect of Aggregates

The flexural (three-point bending) strength and modulus were increased by using the crushed sand (a mixture of sandstones and limestone), and the increase was dependent on the polyester content. Compared to the blasting sand systems a maximum strength increase of 50% was noted. An increase of 266% in the modulus for the 18% PM was observed and the modulus was 31 GPa (4500 ksi). Changing the aggregate system to crushed sand had a higher increase in flexural strength of the 18% PM than silane treatment of the blasting sand. Figure 17 shows the effect of aggregate type on the compressive and flexural strength. An increase of 20% is observed for compressive strength compared to 50% for flexural strength.

Failure Modes in Compression

Changing the fiber and polymer contents resulted in shear, splitting, or a combination of failure making it difficult to visually identify the type of failure. A simple method is used to investigate the failure patterns on the surface of the cylindrical specimens. It consists of mapping the failure patter on the specimen surface as shown in Fig. 18. The abscissa x is the length of th arc ox and the ordinate y is the corresponding height to the failure surface (xy). The origin is chosen to be on the bottom of the specimen close to the lowest point (of ordinate [y.sub.o]) of the curve. After testing the specimen, the failure pattern was traced on a plane. The pattern is represented by

y - [y.sub.o] = d tan [Alpha][sin.sup.2](x/d)] (1a)

where d the diameter of the specimen, [Alpha] is the angle of the failure plane to the horizontal. The above equation may be written in dimensionless form as

y - [y.sub.o]/h = tan [Alpha]/h/d[sin.sup.2][Pi](x/[Pi]d]) (1b)

where h is the height of the specimen. The angle [Alpha] may be obtained by using the least square method. Figure 19a gives a qualitative description of th three different types of failure for all the specimens tested.

In Fig. 19b, the angle of failure [Alpha] was plotted vs. the compressive strength. The estimation of this angle leads us to quantitatively define these failures: a cone failure if the angle [Alpha] is between 0 [degrees] and less than 45 [degrees], a shear failure if [Alpha] is between 45 [degrees] and 70 [degrees] (selected based on experimental observations), and a split failure if [Alpha] is between 70 [degrees] and 90 [degrees]. From the shear failure data shown in Fig. 19b, the average material friction angle and cohesion may be obtained.

Stress-Strain Relationships

The complete stress-strain curve for PM is essential to better understand the material's behavior. The post-peak behavior becomes important to characterize when fibers are added to the PM. Three categories of empirical relationships have been used to fit the experimental stress-strain curves. The first category consists of two separate relations, one for the ascending part and the other fo the descending part. For example the following equations were used to describe the tensile stress-strain curve of cement concrete (11):

Y = x/q + (6q -5/4q) [X.sup.2] + (1 - 2q/4q) [X.sup.6] (ascending) (2a)

Y = X/[Beta][(X - 1).sup.2] + X (descending) (2b)

Y, X and q are given by the following:

Y = [Sigma]/[[Sigma].sub.f] (2c)

X = [Epsilon]/[[Epsilon].sub.f] (2d)

q = [E.sub.f]/[E.sub.o] (2e)

where [Sigma] is the stress and [[Sigma].sub.f] the strength, [Epsilon] is the

strain and [[Epsilon].sub.f] the failure strain. The parameter q is the ratio o secant modulus at peak stress ([E.sub.f]) to initial modulus ([E.sub.o]) and reflects the nonlinear pre-peak behavior of the stress-strain relationship. The constant [Beta] is determined using the experimental data. The second category consists of one equation but with two sets of constants for the ascending and the descending curves. For example, Fanella and Naaman (12) used the following relationship:

Y = X + [m.sub.1][X.sup.2]/[q.sub.1] + (1 - 2[q.sub.1]) X + ([q.sub.1] + [m.sub.1]) [X.sup.2] (ascending) (3a)

Y = X + [m.sub.2][X.sup.2]/[q.sub.2] + (1 - 2[q.sub.2]) X + ([q.sub.2] + [m.sub.2]) [X.sup.2] (descending) (3b)

where Y and X are as defined by Eq 2 and [q.sub.1], [m.sub.1], [q.sub.2], and [m.sub.2] are constants. For the ascending part of the curve [q.sub.1] = q, as defined by Eq 2e. The third category is represented by one equation with the constants for the complete stress-strain curve. b An example of such equation may be as follows (13):

Y = X/q + (1 - q) [X.sup.1/(1 - q)] (4)

where q as defined by Eq 2e. The above equation has one parameter q. A two-parameter relationship was proposed by Vipulanandan and Paul (5). This relationship may be written as

Y= X/q + (1 - q - p) X + p [X.sup.(q + p)/p] (5)

where Y and X are as defined by Eq 4 and p and q (p, q vary from 0 to 1) are th material parameters to be determined from experimental data. The parameter q is defined by Eq 2e and controls the pre-peak curve. The parameter p controls the post-peak curve and is determined by the least square fit of the predicted to the experimental data points, The steeper the descending stress-strain curve, the smaller the p value (p = 0 for brittle material). Figure 20 shows a comparison of the above relationships to the experimental test data. The experimental test data constants of Eqs 3, 4, and 5 are shown in Table 2. As ca be seen from Fig. 20, the two-parameter stress-strain relationship gave the bes fit to the experimental stress-strain data.

Table 2. Parameters for the Stress-Strain Relationships for the 18% PM
Reinforced With 6% Fiber.

Equation         Parameters         Values

Eq 2                q                0.52
                  [Beta]             1.40
Eq 3            [q.sub.1]            0.52
                [m.sub.1]            0.18
                [q.sub.2]            2.22
                [m.sub.2]            0.24
Eq 4                q                0.52
Eq 5                q                0.52
                    p                0.23

In order to quantify the toughness, the integration of Eqs 2a, 2b, 3a, 3b, 4, and 5 is evaluated as

[Mathematical Expression Omitted] (6)

where T([Epsilon]) is the toughness at the end point [Epsilon]. Figure 21 shows the toughness-strain curves obtained from Eq 6. The comparison with the experimental data shows a good agreement when using the two-parameter relationship (Eq 5) to evaluate Eq 6.

CONCLUSION

Increasing uses of polyester mortar (PM) in varying applications dictated the need for further improvement of its properties in a systematic way. This study dealt with the effect of glass fibers (up to 6%), silane coupling agents, and type of aggregates on the performance of PM. Within the range of variables investigated, the following can be concluded:

1) Addition of 6% of fibers to the 18% PM system improved the flexural strength by 60%, the failure strain by 300% (without change in the initial modulus). The compressive strength increased by 30%. The toughness index ratio ([I.sub.10]/[I.sub.5]) in both flexure and compression increased to 2.2.

2) Silane coupling agent improved the compressive and flexural strengths. For the best system (14% polyester), an increase of 35% in flexural strength and 66 in compressive strength was recorded. A compressive strength of 103 MPa (15,000 psi) was obtained for the treated 14% PM system.

3) Aggregates have varying effect on the strength and modulus. When the blastin sand (mainly quartz) was replaced by the crushed sand (sandstone and limestone) in the PM, an increase of 266% in the flexural modulus of the 18% PM system was observed. Also, the compressive strength increased by 20% and the flexural strength by 50%.

4) A simple relationship can be used to investigate the failure patterns on the surface of the cylindrical specimens. The two parameter stress-strain relationship gave the best fit to the experimental stress-strain data.

ACKNOWLEDGMENT 

This study was partly supported by a grant (No. 0036521262ARP) from the Texas Higher Education Coordinating Board to the University of Houston under the Texa Advanced Research Program (ARP).

REFERENCES

1. ACI Committee 548, ACI Journal, 83, 798 (1986).

2. International Congress on Polymers in Concrete, First American Workshop, ACI San Francisco (1991).

3. C. Vipulanandan and S. Krishnan, International Congress on Polymer in Concrete, Special Applications and New Developments, ACI, San Francisco (1991).

4. C. Vipulanandan, N. Dharmaraian and E. Ching, Materials and Structures, 21, 268 (1988).

5. C. Vipulanandan and E. Paul, ACI Materials Journal 87, 241 (1990).

6. T. W. Brockenbrough and D. N. Patterson, ACI Journal, 79, 322 (1982).

7. Y. Ohama and T. Nishimura, 22nd Congress on Materials Research, 364 (1982).

8. C. Vipulanandan and E. Paul, UHCE 88-13, 8, University of Houston, Houston (1988).

9. C. Vipulanandan and N. Dharmarajan, ACI SP104-5, 89 (1988).

10. ASTM C 1018, ASTM Standards for Concrete and Mineral Aggregates, 4, 637 (1985).

11. G. Zen-hai and Z. Xiu-qin, ACI Materials Journal, 84, 278 (1987),

12. D. A. Fanella and A. E. Naaman, Journal of ACI, 82, 475 (1985).

13. D. J. Carreira, and K. H. Chu, Journal of ACI, 82, 797 (1985).