Cement & Concrete Composites 48 (2014) 2634

Contents lists available at ScienceDirect

Cement & Concrete Composites

journal homepage: www.elsevier.com/locate/cemconcomp

Effects of elevated temperatures on the interface properties of carbon textile-reinforced concrete

Flvio de Andrade Silva a,, Marko Butler b, Simone Hempel b, Romildo Dias Toledo Filho a, Viktor Mechtcherine b
a b

Department of Civil Engineering, COPPE, Universidade Federal do Rio de Janeiro, P.O. Box 68506, CEP 21941-972 Rio de Janeiro, RJ, Brazil Institute of Construction Materials, TU Dresden, 01062 Dresden, Germany

a r t i c l e

i n f o

a b s t r a c t
This work reports on an experimental investigation of the inuence of elevated temperatures on the interface between carbon yarns and a Portland cement based matrix. Polymer-coated and uncoated carbon yarns were tested at temperatures of 20 C under a double-sided pullout test conguration after being subjected to a heating regime at temperatures of 100, 150, 200, 400 and 600 C. The degradation mechanisms of the cementitious matrix were investigated by X-ray diffraction and thermal analysis. Using an environmental scanning electron microscope, micro-structural analysis was performed to evaluate the degradation of the carbon yarn and of its interface with the matrix. After preheating up to 150 C, samples with coated bres showed signicant increases in maximum pullout load and, correspondingly, work to pullout. On a micro-scale this is related to a polymer interlocking mechanism in the yarnmatrix interface, which is generated during the heating and cooling of the polymer yarn coating. Above 400 C no further typical bre pullout behaviour was observed; the reinforcing yarn failed suddenly after the cracking of the matrix. For uncoated yarn preheating up to 200 C had no signicant impact on the maximum pullout load. At 600 C the matrix and bres showed major signs of deterioration; no further typical bre pullout behaviour other than failure of the reinforcing yarn could be observed. 2014 Elsevier Ltd. All rights reserved.

Article history: Received 2 May 2013 Received in revised form 8 January 2014 Accepted 9 January 2014 Available online 24 January 2014 Keywords: Carbon bre Interface Textile reinforced concrete Pullout Temperature

1. Introduction The correct understanding of the bonding behaviour of textiles with cement-based matrices is an important step in comprehending the degrading mechanisms which a composite can suffer when subjected to elevated temperatures. Different to conventional reinforcement systems, the textiles are composed of yarns fabricated using a large number of laments. In the yarns themselves the outer laments in direct contact with the matrix show better bond performance, while neighbouring inner yarns are activated by friction [1]. When a coating is used, the laments are stressed more equally, resulting in more laments taking part in the loadbearing function [2]. To characterise brematrix interface pullout, experimental tests on yarns and laments were performed in single [35] and two-sided [68] congurations. Although two-sided pullout tests are more complicated and time-consuming in their designing and carrying out, they can predict material behaviour in a more realistic way since the bres bridge cracks during the test. Two Corresponding author. Tel.: +55 (21)2562 8493x48; fax: +55 (21)2562 8484.
E-mail address: fsilva@coc.ufrj.br (Flvio de Andrade Silva). http://dx.doi.org/10.1016/j.cemconcomp.2014.01.007 0958-9465/ 2014 Elsevier Ltd. All rights reserved.

dimensional fabrics can also be tested under a pullout-loading conguration [9,10]. Such tests take into account the fabric geometry and also the intersection of warp and weft yarns; which give rise to a further mechanical bond component. Bonding properties of carbon-bres at room temperatures have already been investigated in the past. Katz et al. [11] performed single-sided pullout tests in carbon laments with two different diameters: 10 and 46 lm. For the smaller bre the bond strength varied from 0.52 MPa, for a water-to-binder ratio of 0.50, to 1.29 MPa for a matrix with silica fume addition and w/b ratio of 0.35. Badanouiu and Holmgren [12] performed tests using carbon yarns. The authors tried to improve bond by treating the surface of the yarn with silane and also designing a matrix with polymer and silica fume additions. The results showed an improvement of the bonding properties of carbon bres in matrices containing silica fume and high amounts of polymer. The silane treatment of the bres led to an increase in the average bond strength of 368% for samples cured in air and by 140% for samples cured rst in a fog room. Schefer et al. [13] showed that tailoring sizings and compatible coatings provide a basis to achieve highest carbon roving performance in the concrete composite. In addition, a modication of coatings by nanoclay enables to tailor the morphology of the

Flvio de Andrade Silva et al. / Cement & Concrete Composites 48 (2014) 2634

27

interfase and contributes to reduce crack widths at maximum force by development of hydration products and thus enhancing the interface stiffness of the composite. The effect of high temperature on the ultimate tensile strength (UTS) of pitch-based carbon bres under a nitrogen atmosphere seems not to be affected up to 1300 C. Tanabe et al. [14] showed that in a nitrogen atmosphere the carbon bre UTS even increases slightly from 1000 to 1300 C. Sauder et al. [15] showed that for PAN and rayon-based carbon bres UTS increases slightly when the temperature increases to 1600 C. This phenomenon was related to a possible reduction in aw severity associated with internal stress relaxation. When exposed to oxygen atmosphere the carbon undergoes an oxidisation process and burning which starts from 300 to 400 C for uncoated carbon yarns and from 200 to 250 C for polymer-coated yarns [16]. When exposed to high temperatures Portland-cement-based concrete materials exhibit a loss in strength. Chen et al. [17] showed that the residual compressive strength of high strength concrete starts to decrease above 200 C, yielding only 10% of its original strength at 800 C. This can be traced back to the calcium hydroxide decomposition, which takes place between 370 and 470 C, and of calcium carbonate and of other carbonates, occurring between 600 and 730 C. Drchalov et al. [18] performed mechanical tests in PAN-carbon bre reinforced cement based composites after preheating the samples to temperatures up to 1000 C. For the material with Portland-cement matrix a drop in tensile strength from 4.86 to 3 MPa at 600 C was noticed, which was governed by the carbon bre pullout from the cement matrix. At higher temperatures this behaviour is dominated by the matrix. avdar [19] demonstrated that carbon bres as well as PP, glass, and PVA are active in cement-based matrices under bending and compressive loads only up to 450 C. In the present work the bond of polymer-coated and uncoated PAN-based carbon yarns with a Portland-cement matrix was investigated for temperatures ranging from 20 to 600 C. Double-sided pullout tests were performed on the pre-heated and reference specimens at room temperature (20 C). Micro-structural analysis using an environmental scanning electron microscope was carried out to evaluate the degradation of the carbon bre and of its interface with the matrix. X-ray diffraction and thermal analysis were used to quantify the degradation of the matrix when exposed to the different temperatures.

production of the carbon bre was Acrylonitrile. In a rst step, using a catalytic process together with co-monomers and solvents, a Polyacrylonitrile (PAN) bre is spun to obtain the so-called chemical precursor. In the second step the precursor is oxidised at temperatures ranging from 250 C to 300 C in air, and subsequently carbonated in nitrogen atmosphere at temperatures ranging from 1000 C to 1500 C to drive off non-carbon atoms. During this thermal treatment the bres are stretched in order to align the polymer molecules, thus enhancing the mechanical parameters, i.e., tensile strength and Youngs modulus. Before winding onto bobbins several surface treatments are performed to improve the handling quality and the chemical properties of the bre surface. Because the carbon bre under investigation was not produced for use in cementitious matrices, a subsequent polymer coating was applied. This coating is an aqueous dispersion based on self-crosslinking polymers. The polymer coating was applied as a suspension with a polymer content of 30% by weight by means of two counter-rotating rollers with a contact pressure of 1.7 bar between the rollers. The fresh polymer was dried and cross-linked by infra-red heating at 160 C for 1 min. The polymer lm remained stable at temperatures up to 200 C. At temperatures in excess of this, the polymer was decomposed stepwise up to 500 C. The carbon bre resisted oxidation up to temperatures in the range of 550600 C. 2.2. Matrix and specimen preparation In mixing the specimens for the pullout tests, a nely grained matrix was used, which consisted of cement, y ash, microsilica, and quartz sand [20]. Table 2 summarises the matrix composition. A superplasticizer with a basis of naphthalene-sulphonate was added in order to achieve sufcient owability. The average slump ow value measured with a small cone (bottom diameter 100 mm, top diameter 60 mm, height 70 mm) was 200 mm. Doubly symmetrical, narrowed prisms with a notch depth of 1 mm were used as specimens (cf. Fig. 1). In the vicinity of the notch each specimen was 5 mm thick, and at the ends the thickness increased to 10 mm. The width of the samples was 50 mm. Each specimen was reinforced with one multi-lament yarn extended over the entire length of the sample. Prior to concreting, the reinforcing yarn was xed on a stretching frame in order to imbed it properly into the ne-grained concrete. After casting the samples were demoulded at an age of two days and stored in water (20 C) until one day before testing. Afterwards the samples dried for one day at a controlled lab temperature (20 C). Steel plates were then glued to the loading areas of the specimens and xed between the clamping jaws of the testing machine. In this way a nearly unconstrained mounting of the specimens was possible. 3. Experimental testing procedure The double-sided pullout tests were performed using coated and uncoated multilament yarns embedded in the ne-grained matrix. Before testing, the specimens were heated up to 100, 150, 200, 400, and 600 C and subsequently cooled to room temperature (20 C). Additional experiments on specimens stored at

2. Materials and processing 2.1. Carbon bres The carbon bre was obtained from TohoTenax Europe GmbH. The bre used in the present study is the TohoTenax HTA (High Tenacity Fibre) with a nominal neness of 800 tex and density of 1.79 g/cm3. Table 1 presents the properties of coated and uncoated carbon yarn and lament. The basic material used in the

Table 1 Fibre lament and yarn: physical and mechanical properties. Properties Uncoated Filament Fineness (tex ) Diameter (lm) Number of laments Tensile strength (MPa) Strain to failure (%) Youngs modulus (GPa)
* *

Coated Yarn 803 12,000 1032 0.58 222.9 Yarn 903 12,000 2723 1.51 204.4 Table 2 Matrix composition [19]. Mix ingredients CEM I 32.5 R Fly ash Microsilica suspension (50% powder by mass, 50% water by mass) Sand 01 mm Water (kg/m3) 557 251 56 1114 251

0.067 6.96 3245 1.46 213.5

Mass in g of 1 km yarn or lament, respectively; (tex = g/km).

28

Flvio de Andrade Silva et al. / Cement & Concrete Composites 48 (2014) 2634

Fig. 2. Clamped specimen instrumented with extensometers.

Fig. 1. Dimensions of specimen and arrangements of load adapters and crack width sensors for double-sided pullout test.

room temperature were carried out. The heating regime was applied to the specimens at an age of 28 days subject to a heating rate of 2 C/min through the above-mentioned temperature sequence. The temperature level was kept constant for 120 min, followed by a natural cooling process. The mechanical testing was conducted under a displacementcontrol regime in a servo-hydraulic universal testing machine. The displacement rate was set to 0.05 mm/min until the negrained concrete matrix cracked at the notched cross-section. The test displacement rate was then increased to 1 mm/min and stopped when a crack opening of 2.5 mm had been reached. The values of load, crack width, and displacement of the crosshead of the testing machine were then measured and recorded. The measurement of crack width was accomplished by means of precision gauges xed in the vicinity of the crack. Fig. 2 shows the test arrangement. After testing, the bre and brematrix interface were investigated in an environmental scanning electron microscope (ESEM) model XL30 (FEI Philips Company, Netherlands). For this purpose the double sided pullout specimens were sectioned along the line of the multi-lament yarn in samples of 10 mm length. The ESEM was operated under a low vacuum. All the micrographs were taken under the gaseous secondary electron (GSE) detector mode at accelerating voltages of 20 and 30 kV and chamber pressure of 0.60.7 mBar. No coating with carbon or gold, as is usually done for high vacuum SEM, was required. Using X-ray diffraction (XRD) and thermo-gravimetrical analysis (TGA), the matrix was investigated after having being subjected to all temperatures under study. The XRD device used was a

Theta-Theta powder diffractometer model XRD 3003 TT (Seifert Company, Germany). For sample preparation the matrix material was pulverised to particle sizes of less than 20 lm in a micro-mill, which enables powder preparation without collapse of crystalline structure of micro-grains. Hence the content of amorphous material is practically capped in the pulverisation process. During pulverisation the matrix material was immersed in isopropanol. In order to enable a quantitative analysis of XRD data by means of the Rietveld method, ZnO was added to the matrix material as an internal standard, essential for quantication of amorphous phases. Before deposition on the sample holder the isopropanol was removed from the matrix-powder-isopropanol-suspension by drying in air. The XRD measurement was performed in the 2 (theta) range from 20 to 70 C. For TGA analyses a STA409 (Netzsch Company, Germany) was used. The sample mass of 60 mg was taken from the same powder as prepared for XRD analysis. The sample was deposited on the TGA crucible and heated from 20 to 1000 C at a heating rate of 10 C/min. During the heating process the loss in mass was noted for both processes, i.e., dehydration, the collapse of the ettringite, gypsum, CH, CSH, and deacidication, the collapse of the calcium carbonate. The samples were subjected to testing conditions in an Al2O3 crucible using 60 ml/min of helium as the purge gas. 4. Results and discussion Fig. 3 presents schematically a complete load-deformation curve as measured in the double-sided pullout tests. Three regions of the graph can be distinguished according to the degree of specimen deformation according to the theoretical background presented in [21]. Region 1 represents the straining of the uncracked samples. It ends with the formation of a macro-crack in the notched area of the specimen. In region 2 a sudden drop in force from Fu to Fs occurs. Caused by the specimens under-critical bre content, the reinforcement cannot bear the imposed, relatively high load Fu after matrix cracking. In region 2 the crack-bridging mechanisms of the matrix, e.g., aggregate locking and crack branching, occurs as well as the failure of sleeve yarn laments intensively bonded to the matrix. In region 3 nally the typical pullout of crack-bridging, multilament yarn takes place. The single laments are stretched in different manners during crack opening, according to their bond

Flvio de Andrade Silva et al. / Cement & Concrete Composites 48 (2014) 2634

29

Fig. 3. Schematic view of a typical pullout curve [20].

characteristics within the yarn or to the matrix. The laments fail successively with increasing crack opening. Thus, at rst, a rising of pullout force up to Fmax can be noticed, followed by a softening and leading to complete failure of the crack-bridging multilament yarn. In Fig. 4 the representative pullout force vs. crack opening curves for the uncoated yarn (Fig. 4a) as well as for the coated yarn (Fig. 4b) are presented instead of the complete load-deformation diagrams. Regions 1 and 2 of these curves are not displayed in order to simplify their graphic representation. Only when region 2 could not be distinguished from region 3 due to the continuous drop of the curve, e.g., in the case of the preheating up to 600 C, the graph was plotted starting with region 2. Each curve was selected from a collection of up to 10 individual curves as that representing typical behaviour as accurately as possible. To enable quantication and easy comparison from among the test series, three distinct parameters were dened to characterise the pullout curves, as is shown in Fig. 4: maximum bre pullout force Fmax and the work to bre pullout, W0.3 and W0.6 at crack widths of 0.3 and 0.6 mm, respectively (cf. Fig. 3). Whereas in Fig. 4 only characteristic curves are represented, the parameters Fmax, W0.3 and W0.6 are given for each measured curve so that the scatter of the values obtained can be observed as well. The specied crack widths were chosen as they can be correlated with the average Fmax. In Fig. 5 in the bottom graphs the maximum pullout load of each individual curve is plotted vs. the maximum temperature of preheating. In the top graphs for each individual test the

pullout work to a crack opening of 0.3 mm as well as up to a crack opening of 0.6 mm were indicated over the preheating temperature, respectively. The data for uncoated yarns were represented in Fig. 5a and the data for coated yarns in Fig. 5b. In addition, Table 3 gives the average values and the associated standard deviations. For uncoated yarn preheating up to 200 C has no signicant impact on the maximum pullout load. The slightly higher pullout load of the curve of the reference specimen (20 C) as presented in Fig. 4a is within the scattering of results (cf. Fig. 5a). However, the stiffness of yarn pullout (ascending branch of curve, cf. Fig. 4a) increases when the preheating temperature is increased to 100 C, but no further changes in stiffness can be observed when the preheating temperature is increased from 100 C to 200 C. While at larger crack openings the descending branches of the pullout curves become shallower with increasing temperature of the preheating, the work to pullout W0.3 and W0.6 (at relatively low crack openings) stays largely constant for the range of preheating temperatures between 100 C and 200 C. The rst sign of deterioration in material behaviour in the experiments with uncoated yarns could be observed at a preheating temperature of 400 C. The inclination of the rising branch of the pullout curve and the maximum pullout load were reduced. On the other hand, the descending branch of pullout curve became much shallower (cf. Fig. 4a). Again, due to the fact that the work to pullout was calculated only for relatively small crack openings, this phenomenon does not signicantly affect the results of work to pullout, W0.3 and W0.6, as represented in Fig. 5a. It must be noted, however, that there is a tendency to drops in these values when compared to samples subjected to preheating temperatures of 200 C or less. When preheating to 600 C no further typical bre pullout behaviour could be recorded. After exceeding the tensile strength of the concrete matrix, the samples failed catastrophically. The minor crack-bridging action of the bres could be neglected as compared to the results at the other preheating temperatures. After the pullout test was nished the half-parts of the specimens were separated completely by fast moving-down the hydraulic jaw of the testing machine so that the typical pullout length could be measured with the help of a caliper. The pullout length of the carbon multilament yarn (Table 4) for preheating up to 200 C was found to be 100 mm. Preheating up to 400 C reduced the pullout length to 42 mm due to yarn failure. After preheating up to 600 C only a characteristic length of 15 mm was left. The degradation in the matrix due to exposure to elevated temperatures was investigated by XRD and TGA. The results are shown in Figs. 6 and 7, respectively. The XRD analysis shows that at

Fig. 4. Inuence of temperature on the pullout resistance of carbon yarns from a cementitious matrix: (a) uncoated and (b) coated bres; selected, representative curves are shown.

30

Flvio de Andrade Silva et al. / Cement & Concrete Composites 48 (2014) 2634

Fig. 5. Inuence of temperature on the maximum pullout force and on the pullout work of (a) uncoated and (b) coated carbon yarns.

Table 3 Summary of pullout test results performed on coated and uncoated carbon bres (mean values); standard deviations in parentheses. Temperature (C) Uncoated bres Fmax (kN) 20 100 150 200 400 600 0.690 (0.109) 0.675 (0.061) 0.639 (0.155) 0.640 (0.074) 0.521 (0.069) 0.028 (0.025) W0.3 (J) 0.262 (0.032) 0.268 (0.024) 0.249 (0.052) 0.245 (0.021) 0.181 (0.015) 0.023 (0.006) W0.6 (J) 0.529 (0.083) 0.511 (0.039) 0.496 (0.106) 0.496 (0.057) 0.402 (0.039) 0.028 (0.007) Coated bres Fmax (kN) 0.566 (0.116) 0.726 (0.195) 0.729 (0.300) 0.534 (0.110) 0.101 (0.024) 0.050 (0.022) W0.3 (J) 0.171 (0.021) 0.196 (0.018) 0.181 (0.039) 0.154 (0.018) 0.040 (0.012) 0.036 (0.007) W0.6 (J) 0.388 (0.067) 0.445 (0.055) 0.434 (0.101) 0.363 (0.050) 0.084 (0.026) 0.042 (0.010)

Table 4 Filament pullout lengths after yarn pullout test. Temperature (C) Uncoated bres Lmax (mm) 20 100 150 200 400 600 100 100 100 100 42 15 Coated bres Lmax (mm) 100 100 100 99 93 28

temperatures over 100 C there were no traces of ettringite and that for temperatures of 500 C and 600 C Portlandite was absent from the matrix (Fig. 6). From TG analysis the results can be better visualised in the TG derivative (DTG) by the peaks, which correspond to the several steps of mass loss. It can be seen that between 35 and 200 C the DTG peaks indicate a loss of combined water, which comes initially from the calcium silicate hydrate (CSH) and ettringite, cf. Fig. 7. At higher temperatures the dehydroxylation of the calcium hydroxide (CH) can be observed, which occurs between 370 and 470 C, and nally between 600 and 730 C, the decomposition of calcium carbonate and other carbonates present in the initial cement composition occurs due to the loss of CO2. It can be seen that the amounts of ettringite and CSH gradually decrease with increasing temperatures up to 200 C, and for temperatures of 500 C and 600 C the calcium hydroxide is degraded from the matrix.

Fig. 6. X-ray diffractogram of the cementitious matrix after being exposed to the various elevated temperatures.

Based on the available data for the uncoated yarn pullout tests and for the matrix decomposition with increasing temperature the following conclusions can be drawn: Up to the preheating temperature of 200 C the bond behaviour between carbon laments and matrix is for all practical purposes

Flvio de Andrade Silva et al. / Cement & Concrete Composites 48 (2014) 2634

31

Fig. 7. Thermal analysis of the cement based matrix after exposure to the various elevated temperatures.

unaffected by thermal processing. The collapse of the hydration products at temperatures below 200 C has no impact on the interface performance. Moreover, the mechanical properties of carbon-bre are not inuenced. This is evidenced on the one hand by the mechanical behaviour of the carbon yarn [14] and on the other hand by the fact that the pullout length of laments stays constant, while the pullout curves do not change their shape signicantly. As discussed, the further increase of preheating temperature to 400 C leads to changes in pullout behaviour. At about 370 C the calcium hydroxide begins to collapse and is transformed to calcium oxide (CaO). As well, the dehydration of CSH and ettringite phases is intense at 400 C, cf. Figs. 6 and 7. The impact of the decomposition of these compounds on the morphology of the matrix leads to a loosening of the interface and consequently to a reduction of bond intensity between yarn and matrix. In addition, a slight reduction in yarn strength [14] can be noted as compared to its initial values. Furthermore, the reduction of yarn strength becomes obvious through the reduction in pullout lengths, cf. Table 4: due to the lower lament strength the probability of lament failure at lower stress levels increases, thus leading to the shorter pullout lengths. After preheating up to 600 C the calcium hydroxide collapse is complete. Also a large portion of the CSH phases are dehydrated. Therefore, the interface is essentially altered, becoming insufcient for effective bonding to the yarn, and the loss of yarn strength is very pronounced due to the oxidation of the carbons structure [14]. The loss of lament strength is also demonstrated by the minimal pullout length of laments (Table 4), which can be traced back to the failure of laments already at very low load levels. Finally, in case of uncoated multilament yarn both mechanisms, i.e., the intense matrix dehydration and the degeneration of multilament yarn laments above 400 C govern the decay of the pullout properties. The samples with coated yarn showed different brematrix interface mechanisms. Here the preheating up to 150 C led to a signicant increase in the maximum pullout load and corresponding crack opening (Figs. 4b and 5b). Due to the smaller amount of stiffening in the rising branches of pullout vs. crack opening curves, the work to pullout W0.3 and W0.6 for preheating up to 100 or 150 C rises only slightly. It must be emphasised that the work to pullout for larger crack openings, those above 0.6 mm, is consider-

ably higher for the specimens subjected to temperatures up to 150 C. At preheating up to 200 C a clear drop of pullout performance back to the same observed in the reference specimen (20 C) can be seen. Only at larger crack openings, above 0.6 mm, can a higher crack-bridging load transfer by means of the yarn be noticed. Thus, the maximum pullout force, Fmax, as well as the work to pullout W0.3 and W0.6 decrease to the reference level (Fig. 5b). Preheating to 400 C leads to a nearly complete loss of crack bridging ability in the multilament yarn. The pullout curve shows only a slight rise up to a mean pullout load of approximately 100 N (cf. Fig. 4b, Table 3); afterwards steady yarn pullout at low load levels takes place (Fig. 4b). After the preheating up to 600 C specimen behaviour is the same as in the case of uncoated yarn, i.e., after exceeding the tensile strength of concrete matrix the samples fail suddenly. The observed pullout length of the carbon multilament yarn when pre-heated up to 200 C (see Table 4) was 100 mm. Only a slight reduction of the mean value of the pullout length could be noticed after preheating up to 400 C. Preheating to 600 C causes a pronounced failure of the yarn leading to an average pullout length of 28 mm. In contrast to the samples reinforced with uncoated yarns, here chiey the coating of yarn governs the bonding behaviour between yarn and matrix. For the range of preheating temperatures of 100 C and 150 C the pronounced increase in the maximum pullout force and corresponding crack opening and therefore the total pullout energy might be related to the mechanisms described as follows: At a temperature of approximately 100 C the glass transition temperature of the yarn coating made of the copolymer styrol butadiene is reached. At this temperature the mechanical behaviour of the coating transforms from visco-elastic to plastic. At the same time a thermally induced volume expansion of the copolymer takes place. The volume expansion of copolymer generates a hydrostatic pressure on the plasticized polymer, thus pushing it into small pores of the surrounding cementitious matrix. When cooling the sample to the room temperature, the copolymer becomes stiffer and visco-elastic again. This volume reduction now generates tensile stresses inside the polymer. But a complete retraction of copolymer from the small pores is not possible due to the adhesive forces between the polymer and the hydration product surfaces in the matrix. Hence, the re-stiffened polymer becomes stretched and some of the newly generated polymer interlocks may fail. However, the predominant portion most likely remains intact, in this fashion forming effective cross links between the matrix and the coated multilament yarn. These mechanisms of matrixpolymer interlocking enable a high-performing bond between laments and matrix and consequently cause the high pullout maximum loads and energy absorptions. The slight steepening of the rising branch of the bre pullout curves can be attributed to the same mechanisms. Since there is no change in the visco-elastic material properties of re-cooled coating copolymer, only the improved mechanical interlocking can be the reason for these changes. Regarding yarn strength, preheating up to 150 C had no impact on coated yarn. The pullout lengths corresponded to the embedded yarn length. Thus, yarn strength was sufciently high to bear the crack bridging forces imposed by the improved matrixlament bond. Due to the preheating up to 200 C the rst thermo-chemical degradation was observable in the copolymer. During heating the same processes of softening and expansion of coating occur, but when exceeding a temperature of 150 C, some degradation in the polymer begins.

32

Flvio de Andrade Silva et al. / Cement & Concrete Composites 48 (2014) 2634

However, the damage to the coating is not very pronounced in this temperature range. So, a drop-back limited to the performance level of the reference specimens takes place. Damage to the yarn cannot be noted yet; see Table 4. Further preheating up to 400 C causes the nearly complete thermal destruction of the polymer. Now the yarn lies inside its matrix channel without effective cross links to the matrix, since the former cross-linking coating has disappeared. Consequently, effective load transfer is impossible. Corresponding to the ndings on uncoated yarn at this temperature level an initial, slight degeneration in the mechanical yarn properties could be observed. But here its impact on the characteristic pullout length of the laments (cf. Table 4) can be neglected since nearly no load can be transferred from the matrix to the weakened laments. A micro-structural investigation using the ESEM was carried out to conrm the mechanisms discussed above. Fig. 8 shows a strand of coated carbon yarn in its natural condition and after exposure to temperatures of 150 C and 200 C, respectively. It can be seen that the styrolbutadiene copolymer coating starts to melt after 150 C, which changes interface characteristics. The brematrix interface micrographs are depicted in Fig. 9 for the natural condition and the preheating temperatures of 150, 200 and 400 C. In the specimens pre-heated up-to 150 C there takes place a polymer interlocking mechanism which can be observed by a pronounced merging of polymer towards the matrix, cf. Fig. 9b. The matrix constituents seem now to be coated by thin

polymer lms. If compared to the reference (cf. Fig. 9a), the polymer lms on the bre surfaces are morphologically modied. On the other hand, for uncoated bres subjected to the same regime a typical interface is observed with CSH and CH crystals and partially hydrated y ash, cf. Fig. 10. As a result of the preheating up to

Fig. 10. Uncoated carbon-brematrix interface after exposure to a temperature of 150 C.

Fig. 8. Carbon bre (a) in its natural condition and (b) after exposure to 150 C and (c) 200 C.

Fig. 9. Coated carbon-brematrix interface: (a) in its natural condition and (b) after exposure to temperatures of (b) 150 C, (c) 200 C and (d) 400 C.

Flvio de Andrade Silva et al. / Cement & Concrete Composites 48 (2014) 2634

33

Fig. 11. Carbon-brematrix interface after exposure to a temperature of 600 C: (a) uncoated and (b) coated carbon yarns.

200 C the amount of coating of bre surface and matrix is reduced in comparison to the specimens pre-heated up to 150 C. In the gaps between bres, polymer is still present. However, on free bre surfaces and elevated matrix areas the polymer is largely removed by pyrolysis. At 400 C no polymer coating is present anymore. The laments remain without bonding particles in excavations vacated by the polymer. Indeed the matrix seems to be free of the polymer. After preheating up to 600 C the same mechanisms as for specimens with uncoated yarn take place since the coating is completely removed by pyrolysis. The matrix becomes damaged due to dehydration of the CH and CSH, hence affecting the chemical bond with the bre. The yarn strength is impacted by the heating as well, as evidenced by the pullout length (Table 4). Fig. 11 shows ESEM micrographs of the uncoated and coated carbon-brematrix interface after being subjected to a temperature regime of 600 C. It can be noticed that for uncoated bres the interface is degenerated with no signs of calcium hydroxide and CSH. Furthermore, the bres are now oxidised. The coated bres appear very similar: a complete destruction of the interface, pronounced damage to the bres, and cracked masses of the matrix. 5. Conclusions The following conclusions can be drawn from the present work on the interface properties of carbon textile reinforced concrete exposed to different temperatures before testing at room temperature: For uncoated yarn interfaces preheating up to 200 C has no signicant impact on the maximum pullout load, while the ascending branch of the force-crack opening curve after matrix cracking at elevated temperatures up to this temperature level is steeper in comparison to room temperature. No signs of bre and matrix degradation were noticed. The rst pronounced indications of material degradation could be observed upon heating the specimens up to a temperature of 400 C. At this temperature the calcium hydroxide begins to oxidise and is transformed to calcium oxide. Also the dehydration of CSH and ettringite phases are already in progress. The changes in matrix morphology lead to a loosening of the interface, while during dehydration the hydrate phases change their mechanical properties, the shape of the matrix, and the yarn matrix interface. After heating the specimens up to 600 C and subsequently cooling them, the matrix and bres showed major signs of degradation, and no typical bre pullout behaviour could be recorded due to the failure of the reinforcing yarn immediately after the cracking of the matrix.

For samples with polymer-coated yarns, the coating strongly affects the bonding properties at certain temperatures. After preheating up to 150 C signicant increases in the maximum pullout load and the corresponding crack opening were observed. This was traced back to a matrix-polymer interlocking mechanism of molten and re-stiffened polymer, which formed a high-performing bond between laments and matrix. After preheating up to 200 C a clear drop in pullout performance back to the same level was observed for the reference specimen (20 C). Preheating up to 400 C led to the polymer coatings thermal decomposition, which subsequently resulted in a nearly complete loss of the crack bridging ability of the multilament yarn. After preheating up to 600 C specimen behaviour was the same as in the case of the specimen with the uncoated yarn: after exceeding the tensile strength of the matrix the samples failed suddenly due to the failure of the reinforcing yarn.

References
[1] Brameshuber W. Textile reinforced concrete. State of the art report of Rilem Technical Committee 201-TRC: textile reinforced concrete. Rilem publications S.A.R.L.; 2006. [2] Kockritz U, Offermann P, Jesse F, Curbach M. Inuence of textile manufacturing technology on load bearing behavior of textile reinforced concrete. In: Proceedings of the 13th international techtextil-symposium 2005. Frankfurt/ Main, Paper n. 24; 2005. [3] Silva FA, Mobasher B, Soranakom C, Toledo Filho RD. Effect of ber shape and morphology on interfacial bond and cracking behaviors of sisal ber cement based composites. Cem Concr Compos 2011;33:81423. [4] Kim DJ, El-Tawil S, Naaman AE. Loading rate effect on pullout behavior of deformed steel bers. ACI Mater J 2008;105:57684. [5] Banholzer B, Brameshuber W, Jung W. Analytical evaluation of pull-out tests the inverse problem. Cem Concr Compos 2006;28:56471. [6] Butler M, Mechtcherine V, Hempel S. Experimental investigations on the durability of brematrix interfaces in textile-reinforced concrete. Cem Concr Compos 2009;31:22131. [7] Barhum R, Mechtcherine V. Effect of short, dispersed glass and carbon bres on the behaviour of textile-reinforced concrete under tensile loading. Eng Fract Mech 2012;92:5671. [8] Lorenz E, Ortlepp R. Bond behavior of textile reinforcements - development of a pull-out test and modeling of the respective bond versus slip relation. In: Parra-Montesinos GJ, Reinhardt HW, Naaman AE, editors. Proc. 6th international workshop on high performance ber reinforced cement composites (HPFRCC6), Ann Arbor; 2011. p. 46370 [19.22.6.11]. [9] Barhum R, Mechtcherine V. Inuence of short dispersed and short integral glass bres on the mechanical behaviour of textile-reinforced concrete. Mater Struct 2013;46:55772. [10] Sueki S, Soranakom C, Mobasher B, Peled A. Pullout-slip response of fabrics embedded in a cement paste matrix. J Mater Civ Eng 2007;19:71827. [11] Katz A, Li VC, Kazmer A. Bond properties of carbon bers in cementitious matrix. J Mater Civ Eng 1995;7:1258. [12] Badanoiu A, Holmgren J. Cementitious composites reinforced with continuous carbon bres for strengthening of concrete structures. Cem Concr Compos 2003;25:38794. [13] Schefer C, Gao SL, Plonka R, Mader E, Hempel S, Butler M, et al. Interphase modication of alkali-resistant glass bres and carbon bres for textile reinforced concrete II: water absorption and composite interphases (up to 2000 C). Compos Sci Technol 2009;69:90512.

34

Flvio de Andrade Silva et al. / Cement & Concrete Composites 48 (2014) 2634 [18] Drchalov J, Mnahoncakova E, Vejmelka R, Kolisko J, Bayer P, Cerny R. Hydric and mechanical properties of carbon ber reinforced cement composites subjected to thermal load. Constr Build Mater 2004;18:56778. [19] avdar A. A study on the effects of high temperature on mechanical properties of ber reinforced cementitious composites. Composites Part B 2012;43:245263. [20] Butler M, Hempel S, Mechtcherine V. Modelling of ageing effects on crackbridging behaviour of AR-glass multilament yarns embedded in cementbased matrix. Cem Concr Res 2011;41:40311. [21] Butler M, Mechtcherine V, Hempel S. Durability of textile-reinforced concrete made with AR glass bre effect of the matrix composition. Mater Struct 2010;43:135168.

[14] Tanabe Y, Yasuda E, Bunsell AR, Favry Y, Inagaki M, Sakai M. The strength of pitch-based carbon bre at high temperature. J Mater Sci 1991;26: 16014. [15] Sauder C, Lamon J, Pailler R. Thermomechanical properties of carbon bres at high temperatures. Compos Sci Technol 2002;62:499504. [16] Younes A, Seidel A, Engler T, Cherif C. Effects of high temperature and long term stress on the material bahaviour of high performance bres for composites. World J Eng 2010;7:30915. [17] Chen B, Liu J. Residual strength of hybrid-ber-reinforced high-strength concrete after exposure to high temperatures. Cem Concr Res 2004;34:10659.