Effect of Tire Rubber Particles on Crack Propagation in Cement Paste

Tire rubber �articles (NaOH-treated and untreated) were in�estigated as �ossible crack stabilizer and toughness enhancer when added to cement �aste through in situ crack �ro�agation measurements using two different ty�es of cement, ty�e I/II and an Interground �oly�ro�ylene Fiber �ement (IF�). �rack deflection and crack bridging were obser�ed in s�ecimens with untreated rubber in cement ty�e I/II. �rack ti� mechanisms associated with crack �inning and crack arrest were �resent in ty�e I/II cement and IF� with treated rubber �articles. �rack ti� mechanisms in IF� with treated rubber lead to the increase in �MOD at the ultimate load le�el. �rack wake mechanisms in IF� with untreated or treated rubber lead to strain hardening and strain softening beha�ior. �rack wake bridging mechanisms were re�laced by multi�le cracking mechanisms in the IF� s�ecimens with treated rubber. The IF� s�ecimens with untreated rubber inclusions �ro�ided the best results with res�ect to toughness enhancement.

The stress-strain cur�e of a quasi-brittle material generally consists of three regions corres�onding to the different mechanisms that control the beha�ior of the com�osite.The first region corres�onds to the linear elastic res�onse of the com�osite.The �oint at which the stress-strain cur�e becomes nonlinear (bend o�er �oint) corres�onds to the beginning of cracking extension.The second region, where the strength increases beyond the bend o�er �oint u� to a maximum, which is the ultimate strength of the com�osite, is associated with a strain hardening regime and characterizes the ability of the second-�hase to stabilize a growing cracking �rocess.The third region, where the com�osite gradually loses its load carrying ca�acity, is associated with the strain softening regime.Materials with this ty�e of quasibrittle beha�ior exhibit desirable �ro�erties such as enhanced flaw tolerance and toughness.
�rack ti� and crack wake mechanisms characterize toughening mechanisms in brittle com�osites.�rack ti� mechanisms commonly in�ol�e crack-inclusion interactions at the crack ti� such as crack �inning, crack bowing between inclusions and crack blunting.�rack wake mechanisms include bridging of crack surfaces by inclusions and in some instances �ullout of inclusions from the matrix.The crack wake mechanisms are considered more effecti�e as energy absorbing mechanisms due to their cumulati�e nature.
Most theoretical and ex�erimental studies on crack �ro�agation and crack/reinforcement interactions consider the reinforcements to be incor�orated into the crack wake by a crack front that has mo�ed in a stable manner �assed these reinforcements.The assum�tion of a stable crack interacting with reinforcements may be �alid for com�osites with high �olume fractions of reinforcements.Howe�er, at low �olume fractions (and hence large s�acing between reinforcements) such as in cement-based materials reinforced with rubber �articles, the crack may become unstable shortly after initiation before it has interacted with the rubber �articles or the interaction being limited to only a few �articles.An unstable crack front mo�ing �assed rubber �articles with high crack �elocity will require far more bridging ligaments behind the crack ti� to slow down its �ro�agation.S�ecial attention has to be gi�en to �re�ent the crack from becoming unstable before interacting with a certain critical number of inclusions.For toughness enhancement, both crack ti� and crack wake mechanisms may �lay an im�ortant role.These mechanisms ha�e to be o�timized in order to reduce the crack �elocity, allowing rubber �articles to be incor�orated into the crack wake by a crack front that has mo�ed in a stable manner when �assed these �articles.
In this study, rubber �articles were in�estigated as �ossible crack stabilizer and toughness enhancer when added to cement �aste.Some of the tire rubber �articles were surface treated with NaOH solution, which enhances the bonding of the rubber �articles to the cement matrix 1� .Zinc stearate is an additi�e �resent in tire formulations that migrates from the bulk and diffuses to the surface leading to �oor to the surface leading to �oor leading to �oor �oor adhesion characteristics.The NaOH treatment remo�es the stearate The NaOH treatment remo�es the stearate from the rubber surface, changing the surface chemistry, enhancing urface, changing the surface chemistry, enhancing , changing the surface chemistry, enhancing surface chemistry, enhancing , enhancing enhancing ing the surface homogeneity and, consequently, leading to a surface with and, consequently, leading to a surface with leading to a surface with more uniform free energy uniform free energy 19 .Influences of treated and untreated tire rubber �articles on crack ti� and crack wake mechanisms was studied through in situ crack �ro�agation measurements using a custom designed loading de�ice.The influence of the matrix on mechanical �ro�erties and crack �ro�agation characteristics was in�estigated using two different ty�es of cement, the common ASTM ty�e I/II �ortland cement and an Interground �oly�ro�ylene Fiber �ement (IF�).

Specimens preparation
�ntreated rubber tire �articles of �00 µm maximum size (density: 1.1�2 ± 0.001 g.cm -3 ) were su��lied by Borcol Industria de Borracha Ltda (Sorocaba/S�, Brazil).The �articles were surface treated with saturated NaOH aqueous solution for 20 minutes at room tem�erature whilst stirring.The mixture was filtered and the rubber was rinsed with water until neutral �H was achie�ed and allowed to dry at room tem�erature 1� .The Interground �oly�ro�ylene Fiber �ement (IF�) �roduction in�ol�es a new technique where �oly�ro�ylene (��) fibers (12 mm in maximum length and 1� µm in diameter) are added to the cement clinker during grinding.The intergrinding �rocess allows the fibers to be distributed homogeneously within the cement.The IF� contains 0.3% of �oly�ro�ylene fibers by weight of cement 20 .�ontrol s�ecimens and s�ecimens containing 10% (by weight of the �aste) of treated or untreated rubber �articles were �re�ared using both the IF� and the ASTM ty�e I/II cement.A water/cement ratio, w/c, of 0.36 was used for all s�ecimens.The �oly�ro�ylene fiber �olume fraction in the s�ecimens at the w/c ratio used is only 0.�%.In �olume, the amount of rubber in the s�ecimens re�resents 1�%.
The s�ecimens were cast using �0 x �0 x �0 mm steel molds.Two acrylic rods were �re-�ositioned inside the steel molds in order to obtain the two holes in the s�ecimens necessary to fix the sam�le in the loading de�ice.The mix was �laced in the formwork and com�acted on a �ibrating table for about 20 seconds.After 2� days of curing at 100% relati�e humidity, the cubes were cut into com�act tension s�ecimens with a width of �0 mm and a thickness of 6 mm.A notch length of 2� mm was cut in all s�ecimens using a saw blade of 0.3 mm thickness.The s�ecimens surface were successi�ely �olished with 3� µm Si� �owder, 12 µm, 9 µm and 3 µm Al 2 O 3 �owder to ensure uniform thickness and smooth surfaces to facilitate the crack �ro�agation obser�ations under the microsco�e.

In situ crack propagation measurements
The loading de�ice designed for crack �ro�agation measurements is shown in Figure 1.It uses a �iezoelectric transducer for load transfer.The �iezoelectric transducer is �laced between two �i�ot arms and, after being acti�ated by a high �oltage am�lifier, deli�ers o�ening forces to the com�act tension s�ecimen through the loading arms.The in�ut �oltage was com�uter controlled.A load cell is �laced in one of the loading arms and monitors the a��lied load 20,21 .
The crack mouth o�ening dis�lacement (�MOD) was monitored through a MTS cli� gage mounted on the s�ecimen.The loading de-�ice was staged under an o�tical microsco�e equi��ed with a �ideo camera connected to a TV screen and a �ideo recorder.
The loading fixture allows testing to be �erformed under dis-�lacement control.A small dis�lacement rate of 1 µm/s was chosen for all s�ecimens to be able to obser�e the crack ti� and crack wake mechanisms while loading.The crack �ro�agation is continuously monitored and recorded on �ideo u� to s�ecimen failure.

ASTM Type I/II cement paste containing tire rubber particles
�om�act tension tests were �erformed on Ty�e I/II cement �aste s�ecimens containing untreated or treated tire rubber �articles to gain information on their crack �ro�agation �rocesses.Load as a function of crack mouth o�ening dis�lacement (�MOD) for these s�ecimens and the control are shown in Figure 2. Beyond the ultimate load, a continuous strain softening cur�e could not be obtained.The de�iation from linearity in the load as a function of �MOD cur�es indicates the extension of the crack from the notch ti�.�rack initiation from the notch occurred, in a�erage, at similar load le�els for the control and s�ecimens with untreated rubber.For the s�ecimens with treated rubber, known for their better adhesion to the cement matrix 1� , a more �ronounced strain hardening regime was obser�ed in the load vs. �MOD �lots (Figure 2b).The load le�el at which the crack initiated from the notch ti� was on a�erage reduced com�ared to the control s�ecimens.The cement �aste matrix seemed to be weakened by the treated rubber �articles.
Box �lots were built to gain an o�er�iew of data from a statistical �ers�ecti�e.A box �lot is an effecti�e �isual re�resentation of both central tendency and dis�ersion.Box shows �0% of the data, the small square indicates the a�erage, the median is shown as a line across the box (or on its horizontal bounds) and the �ertical line contains the other �0% of the data and the minimum and maximum data �alues on its edges.The length of the �ertical lines indicates �isually how far from the middle of the distribution the extreme �alues are.
Figure 3 shows box �lots of ultimate load and toughness, which is �ro�ortional to the area under the entire load-�MOD cur�es, obtained for the s�ecimens tested.Variability in ultimate load, characterized by the boxes sizes and �ertical lines lengths, is common to all sets of s�ecimens, with or without rubber �article.No im�ro�ement in strength was obser�ed for the s�ecimens that contain rubber �articles, com�ared to the control.On the other hand, higher toughness is obser�ed for the s�ecimens containing treated rubber �articles and a tendency to higher toughness for the s�ecimens with untreated rubber, when com�ared to the control s�ecimens.��en though no statistical difference, in a�erage, for toughness is obser�ed between the s�ecimens containing rubber, the com�osites with treated rubber showed less scattered �alues when com�ared to the com�osites with untreated rubber.
The crack that initiated at the notch ti� remained the dominant crack until failure.All s�ecimens broke in two �arts after failure.As ex�ected, the crack in the control s�ecimens �ro�agated in a straight   For the s�ecimens with treated rubber, the in-situ crack �ro�agation measurements re�ealed that the crack was attracted by the rubber �articles.The crack deflected from its original �ath (Figure �a, b), where original �ath would ha�e been from the to� to the bottom of the microgra�h, to interact with the rubber �articles.This attraction allowed a more effecti�e crack/rubber interaction com�ared to the untreated rubber �articles.Tensile stresses in the matrix due to an elastic mismatch between the rubber �articles and the matrix may be res�onsible for the obser�ed load reduction at crack initiation and for attracting the crack towards the rubber �articles.The crack after being attracted by the rubber �articles incor�orates them in its wake.�rack �inning by rubber �articles was also obser�ed.The rubber �articles e�entually debond along both sides of the rubber/matrix interface (Figure �c, d) and may stretch between the crack surfaces before being �ulled out of the matrix.These crack/rubber �article interactions are res�onsible for the more �rominent strain hardening beha�ior.

IFC paste containing tire rubber particles
The load vs. �MOD �lots for the IF� �aste s�ecimens control and containing treated or untreated rubber �articles are shown in Figure 6.Strain hardening and �ronounced strain softening beha�ior was obser�ed for the treated and untreated rubber IF� s�ecimens, com�ared to the control.Similar to the ty�e I/II cement s�ecimens with treated rubber �articles the load required for crack initiation from the notch was reduced com�ared to that of the control and with untreated rubber s�ecimens.Howe�er, an increase in �MOD at the ultimate load le�el was obser�ed.Box �lots of ultimate load and toughness obtained for these s�ecimens are shown in Figure �.Although less scattered, lower �alues of ultimate load are obser�ed for s�ecimens containing treated rubber, com�ared to the control and to the s�ecimens with untreated rubber.A �rominent toughness enhancement is obser�ed for the s�ecimens with rubber, com�ared to the control.A tendency to higher toughness is obser�ed for the s�ecimens containing untreated rubber, when com�ared to the s�ecimens with treated rubber.On the other hand, a higher number of data �oints are obser�ed in the strain softening region of the cur�es obtained for the s�ecimens containing treated rubber (Figure 6b) indicating more stable crack �ro�agation.
After failure, all s�ecimens remained unbroken.The IF� control s�ecimens re�ealed intact �oly�ro�ylene fibers bridging the crack surfaces from the notch ti� to the end of the s�ecimen.For the rubber s�ecimens, no cracks could be detected with the naked eye or in the o�tical microsco�e unless the s�ecimens were reloaded.The untreated rubber �articles constituted bridging sites in the crack wake.A rubber-bridging site close to the crack ti� and in the wake of the crack is shown in Figure � at different a��lied load le�els.Due to stable crack �ro�agation this crack-rubber interaction could be obser�ed from their early stage of de�elo�ment in close �roximity to the crack ti� until their disintegration at a��lied dis�lacements close to failure.�rack �inning by treated rubber �articles was also obser�ed, as shown in Figure 9.For treated rubber �articles in IF� matrix, the majority of crack/rubber interactions lead to crack detention at the rubber �articles.The initial crack did not remain the dominant crack u� to failure as shown in Figure 10.Frequently secondary cracks initiated at the bridging sites at higher a��lied load le�els (Figure 10b) and one of these cracks became the dominant crack as shown in Figure 10c, which re�eals the bridging site after failure.On the left of the bridging site, an additional microcrack has formed (indicated by arrow in Figure 10c) �robably due to stress redistribution.
The crack is attracted by the rubber �articles (Figure 11a), like in the s�ecimens with treated rubber in cement ty�e I/II.The crack instead of �ro�agating from to� to bottom (i.e.�er�endicular to the a��lied tensile stress) it deflects to the left in order to interact with the rubber �article.Various ty�es of secondary cracks that form in close �roximity to the rubber bridging sites are shown in Figure 11.
The crack wake bridging mechanisms were re�laced by multi�le cracking mechanisms, which lead to a toughness enhancement at the ex�ense of weakening the matrix and reducing the hardness.The multi�le secondary microcrack formations are res�onsible for the increase in �MOD at ultimate load le�el obser�ed in Figure 6.
The multi�le cracking mechanisms may also be accountable for the reduced �ariability in ultimate load �alues shown in Figure �.All treated rubber s�ecimens failed in a �ery narrow range of load le�els, contrary to all other s�ecimens, which re�ealed a large �ariability in crack initiation and ultimate loads.

Discussion
To enhance toughness of cement based materials by rubber inclusions, crack ti� as well as crack wake mechanisms �lay an im-�ortant role and need to be o�timized in order to assure stable crack �ro�agation.�ffecti�e crack ti� mechanisms require high frictional stresses or chemical bonding at the rubber/matrix interface.If these are absent then a crack may sim�ly sur�ass the �article and the tendency to instability will be higher before a critical �olume fraction of �articles are incor�orated in its wake.In order for the crack to arrest or reduce its crack �elocity, it needs to incor�orate many of these rubber �articles in its wake.Since low �olume fractions of rubber is used the resulting crack length would be rather large and no strength and only limited toughness enhancement is to be ex�ected.This ty�e of cracking was obser�ed in s�ecimens with untreated tire rubber �articles in ty�e I/II cement �aste.The crack became unstable after some distance from the notch ti�, com�arable to the control s�ecimens.��en though crack/rubber interactions were obser�ed during crack �ro�agation, these interactions occurred while the crack was already unstable.�rack wake mechanisms, e�en though �resent (Figure �), are not effecti�e in increasing toughness significantly when crack ti� mechanisms are absent at high crack �elocities and low �olume fraction of reinforcements.Therefore, the com�osite �ro�erties are com�arable to the control s�ecimens.In the absence of effecti�e crack ti� mechanisms, stable crack �ro�agation can only be assured, and hence toughness increased, if additional bridging ligaments are incor�orated into the wake of the crack.This can be accom�lished by either increasing the �olume fraction of the rubber �articles or, as was done in this study, by using IF� as the matrix.The toughness enhancement obser�ed in IF�-rubber s�ecimens, far exceeding the toughness enhancement in ty�e I/II cement-rubber s�ecimens, is attributed to the �oly�ro�ylene, ��, fibers in IF�.The reduction in crack �elocity due to the ��-fibers allows the rubber �articles to be incor�orated into the crack wake by a crack front that has mo�ed in a stable manner when �assed these reinforcements, acti�ating the bridging mechanisms at relati�ely short crack lengths.It is worth noting here, that the IF� control s�ecimens did not exhibit toughness enhancement due to the ��-fibers com�ared to Ty�e I/II cement control s�ecimens (Figures 3 and �).
�rack ti� mechanisms were �resent in s�ecimens with treated rubber �articles, in ty�e I/II cement and IF�.The crack was attracted by the treated rubber �articles leading to crack �inning and crack arrest.�rack �inning in ty�e I/II cement-treated rubber s�ecimens slowed down the crack �elocity e�en before considerable crack wake mechanisms were acti�ated.The slow down in crack �ro�agation is due to the crack front bowing out between the �inning �ositions and thereby absorbing energy from the crack ti� due to the increase in crack front length and due to the creation of new surfaces 1� .This increase in energy contributes to toughness enhancement.The obser�ed strain hardening beha�ior enhancement in Figure 2b is associated with crack wake mechanisms such as successi�e debonding of the rubber/cement �aste interface, followed by bridging and stretching of the rubber �articles between the crack surfaces.
The toughening mechanisms in the treated rubber �articles in IF� were dominated by crack ti� mechanisms leading to crack arrest at the rubber �articles.�inning of the crack ti� could be obser�ed during load a��lication while watching the crack �ro�agation �rocess in situ.Instead, of un�inning and incor�orating the rubber �articles in the wake of the crack (as was obser�ed for the treated rubber �articles in ty�e I/II cement), the �inning lead to crack arrest and stress redistribution which caused secondary cracks to form in close �icinity to the crack �inning sites.In this case, crack ti� mechanisms �re�ented crack wake mechanisms such as rubber bridging and �ullout from de�elo�ing.Hence, less �rominent strain hardening could be obser�ed in these s�ecimens.The increase in �MOD at the ultimate load le�el (Figure 6b) is associated with the multi�le cracking mechanisms.The reduction in crack �elocity due to the �oly�ro�ylene fibers in IF� may ha�e modified the deformation characteristics of the treated rubber �articles.

Conclusions
In situ crack �ro�agation measurements were �erformed on cement �aste s�ecimens containing untreated or treated tire rubber to in�estigate the crack/rubber interactions and the underlying toughening mechanisms.The effect of matrix on crack �ro�agation was in�estigated using two different ty�es of cement, the common ty�e I/II �ortland cement and an Interground Fiber �ement (IF�).
The toughness of cement �aste with rubber inclusions was influenced by both crack ti� and crack wake mechanisms.The existence or absence of crack ti� and/or crack wake mechanisms de�ends on the adhesion between the rubber �articles and the cement �aste and the ty�e of cement used (ty�e I/II or IF�).�rack deflection and crack bridging were obser�ed in s�ecimens with untreated rubber in cement ty�e I/II.These mechanisms howe�er did not lead to significant toughening enhancement due to unstable crack �ro�agation �rior and during crack/rubber �article interaction.�rack ti� mechanisms were obser�ed for treated rubber-IF� and treated rubber-ty�e I/II cement s�ecimens.�rack ti� mechanisms in IF� with treated rubber lead to the increase in �MOD at the ultimate load le�el.�rack wake mechanisms in IF� with untreated or treated rubber lead to strain hardening and strain softening beha�ior, com�ared to the control.The effecti�eness of the crack wake mechanisms was de�endent on the crack �elocity encountered by the rubber �articles while interacting with the �ro�agating crack.�rack wake bridging mechanisms were re�laced by multi�le cracking mechanisms in the IF� s�ecimens with treated rubber.The IF� s�ecimens with treated rubber inclusions showed the most stable crack �ro�agation and the IF� s�ecimens with untreated rubber inclusions �ro�ided the best results with res�ect to toughness enhancement.

Figure 3 .
Figure 3. Box �lots for ultimate load and toughness for ASTM ty�e I/II �ortland cement �aste s�ecimens control and containing treated or untreated rubber �articles.

Figure 4 .
Figure 4. �rack/rubber interaction in ASTM ty�e I/II �ortland cement �aste s�ecimens with untreated rubber �articles.The crack �ro�agation occurs from to� to bottom of the microgra�hs.a) the crack is deflected by the untreated rubber �articles; b) the crack is bridged by the rubber �article; and c) �ull-out of a rubber �article, after s�ecimen failure.Letter R indicates a rubber �article.
Figure �a re�eals the bridging site in its early stage with the crack ti� just ahead of the rubber �article.With increasing dis�lacements, the crack o�ens and re�eals the �oly�ro�ylene fibers (indicated by arrows in Figure �c) bridging the crack surfaces.The microgra�h shown in Figure �e was taken after failure.

Figure 5 .
Figure5.�rack/rubber interaction in ASTM ty�e I/II �ortland cement �aste s�ecimens with treated rubber �articles.The crack �ro�agation occurs from to� to bottom of the microgra�hs.In a,b) the crack is attracted by the treated rubber �articles and incor�orates them in its wake.The crack is deflected from its original �ath (the original �ath being from to� to bottom of microgra�h) in order to interact with the rubber �articles; and c,d) rubber �article de-bonded along both sides of the rubber/matrix interface and is being stretched between crack surfaces.Microgra�hs c and d were taken after failure of the s�ecimen.Letter R indicates a rubber �article.

Figure 8 .Figure 9 .
Figure 8. �rack/rubber interaction in IF� �aste s�ecimens with untreated rubber �articles at different a��lied load le�els: a) bridging site in close �icinity to the crack ti� at �2% of ultimate load; b) bridging site at ��% of ultimate load; c) close to �eak load; arrows indicate the �oly�ro�ylene fibers bridging the crack surfaces in close �icinity to the rubber bridging site; d) after �eak load; and e) after failure.Letter R indicates a rubber �article.

Figure 10 .
Figure10.�rack/rubber interaction in IF� �aste s�ecimens with treated rubber �articles at different a��lied load le�els.a) rubber �article is bridging the crack surfaces; b) at higher a��lied load, a secondary crack initiates in close �icinity to the rubber �articles due to high frictional tractions at the rubber/matrix interface; and c) rubber �article after failure of the s�ecimen; the initial crack did not remain the dominant crack at failure; additional microcracks o�ened u� to the left of the rubber �article.Letter R indicates a rubber �article.

Figure 11 .
Figure 11.a) �rack being attracted by the treated rubber �article in IF� �aste matrix; b-d) Various ty�es of secondary cracks that form in close �icinity to bridging sites in IF� �aste s�ecimens with treated rubber �articles.Letter R indicates a rubber �article.