Construction and Building Materials 69 (2014) 221–231

Contents lists available at ScienceDirect

Construction and Building Materials

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

Characterization of permeable pavement materials based on recycled

rubber and chitosan
Christopher A. Murray ⇑, Kayla S. Snyder, Brooke A. Marion
Department of Interdisciplinary Studies, Lakehead University, 500 University Avenue, Orillia, Ontario L3V 0B9, Canada

h i g h l i g h t s

 Composite materials for use as permeable pavers were prepared from chitosan and tire crumb.
 Compressive strength and hydraulic conductivity similar to conventional permeable pavers was achieved.
 Sorption capacity of dissolved zinc from water was as high as 0.63 mg per gram of chitosan.
 Unlike other permeable pavements, properties were not strongly dependent on binder content.
 Material properties and morphology were dominated by the concentration of chitosan solution used.

a r t i c l e i n f o a b s t r a c t

Article history: A variety of composite permeable pavement materials were prepared from crumb rubber embedded in a
Received 13 March 2014 matrix of the biopolymer chitosan, which is a waste product of the seafood industry. We have character-
Received in revised form 11 July 2014 ized the hydraulic conductivity, mechanical properties, and the capability of these materials to remove
Accepted 17 July 2014
particulate and dissolved pollutants (including zinc) from water. The dependence of material properties
Available online 12 August 2014
on process parameters such as binding polymer content differs from what is typical of binder-based per-
meable pavement, due to the mechanism by which chitosan is introduced, and in many cases the stability
Keywords:
of the composite material increases with decreased binding polymer content.
Tire crumb
Chitosan
Ó 2014 Elsevier Ltd. All rights reserved.
Permeable pavement
Hydraulic conductivity
Compressive strength
Zinc

1. Introduction carried by stormwater, and when existing stormwater systems

One of the most important areas of focus where water quality is
concerned is management of stormwater and runoff. As develop-
ment increases and natural vegetative groundcover is replaced by
pavement and roofing, rainfall that would have otherwise been
absorbed and slowly released by plants and soil is concentrated
into fast-moving streams in pipes and gutters, dramatically
increasing its ability to erode soil, carry pollutants and overflow
into other water management systems such as wastewater sys-
tems. In addition to increasing the fraction of impermeable space,
development is typically associated with increased soil vulnerabil-
ity – removing the vegetative groundcover for development makes
soil susceptible to erosion. Developing countries and regions
responding to disaster are similarly vulnerable to pollutants
⇑ Corresponding author. Tel.: +1 705 330 4008x2651; fax: +1 705 329 4035.
E-mail address: cmurray1@lakeheadu.ca (C.A. Murray).
are overtaxed or destroyed, a major health risk is caused by uncon-
trolled flood water mixing with sewage. A simple solution to these
problems involves increasing the capacity of developed areas to
infiltrate rainwater rather than conveying it over impermeable sur-
faces to rivers, lakes and streams.
Permeable or porous pavement is a family of passive technolo-
gies that address the problem of runoff by providing pathways for
water to infiltrate down through walkways, parking lots and roads
[1–3]. Rainwater may be directed over an expanse of permeable
pavement, which (if large enough) can allow all of the runoff to
infiltrate down into groundwater. This reduces the buildup of ice
in cold conditions, and may achieve physical filtration and pollu-
tant removal [4,5], but its most important role is to reduce the
impact on other stormwater conveyance systems by reducing the
volume of runoff and its capacity for erosion.
Two major categories of permeable pavement that include a
large fraction of aggregate materials are (1) permeable cement or

http://dx.doi.org/10.1016/j.conbuildmat.2014.07.047
0950-0618/Ó 2014 Elsevier Ltd. All rights reserved.
222 C.A. Murray et al. / Construction and Building Materials 69 (2014) 221–231
asphalt and (2) resin- or elastomer-bound pavement. Materials
falling into the former category are typically prepared by omitting
or limiting a void-filling component of what would normally be
impermeable concrete or asphalt [1,6] (Fig. 1) and those falling into
the latter category are made by bonding granular and/or aggregate
material within a polymeric matrix that inherently leaves spaces
[7]. Cement- or asphalt-based permeable pavement is usually
heavy and capable of supporting similar compression as their
impermeable counterparts [6]. Binder-based permeable pavers
are light, often flexible and are more likely found in walkways than
driveways because of their lesser resistance to compression.
Crumb rubber is a common filler material for both types of per-
meable paver [7,8], as it is low in cost and can often impart elastic-
ity and resilience that is desirable in walkways. Crumb rubber is
generally made from waste car and truck tires that have failed or
have exceeded their usable lifetime [9]. More than 300 million
waste tires accumulate each year in the USA and the majority of
the approximately 80% that are reused are used as fuel. Less than
5% of the tire is typically eroded from the surface before the tread
depth becomes insufficient to ensure safe use and the tire is dis-
carded. Because the rubber used in tires is chemically crosslinked,
it cannot be reformed into new rubber without significant energy
input that generally makes such recycling uneconomical. Tires
brought to a tire recycling facility are cryogenically or mechani-
cally crumbled and ground, and the steel belting and fiber is
removed magnetically and through the flow of air, respectively.
The remaining crumb rubber is sized for the appropriate applica-
tion and most often there is no washing or cleaning step involved
in the preparation of the final crumb product. Primary applications
for tire crumb include filler, athletic and equestrian running sur-
faces, road resurfacing [6,10,11], playing surfaces [12] and garden
mulch. In some of these applications, significant cleaning of the
rubber is required after it has been purchased from the tire
recycling facility, because of the contaminants present in tires-
most notably organic compounds and heavy metals such as zinc,
lead and cadmium [12,13]. These metals eventually leach from
untreated crumb rubber into water, and similarly pose a threat
to water along roadways, where the tread has been abraded off
in the form of tiny, high-surface area particles that eventually get
swept from the road into ditches and other stormwater
conveyances.
Because of the environmental impact of tire waste, significant
interest exists in developing new applications for crumb rubber.
While most existing applications treat tire crumb as a commodity
filling material, some researchers have pursued more exotic appli-
cations that might increase its value and indirectly motivate more
progressive approaches to tire recycling. Crumb rubber has been
used as an experimental filter media [14–16] that becomes more
compressed with increased depth in the filter, leading to decreas-
ing pore sizes ideal for physical filtration (where large particles
are stopped first, smaller particles make their way further into
the filter and ultimate filter clogging involves a full filter, rather
Fig. 1. Schematic representation of cement-based permeable pavement, which differs from conventional concrete (left) in that the cement content is reduced, leaving a
continuous network of pores between aggregate materials (right).
than one blocked by a small amount of material at the top). Crumb
rubber has also been considered as an absorbent of the very heavy
metals and other contaminants it is known to contribute to surface
water.
In this paper we present results describing the preparation of
stormwater filter materials suitable for permeable pavement from
crumb rubber that is untreated following collection from the tire
recycling facility. The binding material we have used to hold the
matrix of rubber particles together is the biopolymer chitosan,
which is derived from chitin. Chitin is the second-most abundant
polysaccharide found on Earth, and plays a structural role in most
invertebrate exoskeletons. Chitosan can degrade in a landfill (with
suitable temperatures, moisture and enzyme content) and corre-
spondingly is a candidate material for many single-use packaging
applications that promises to relieve stress on overfilling landfills.
Chitosan is a waste product of the seafood industry and, like tire
rubber, it is underutilized because of difficulties associated with
its processing. Unlike common thermoplastics, the high degree of
inter- and intra-molecular bonding in chitosan prevents it from
melting, and only solution-based processing methods are available.
Dilute solutions of chitosan (dissolved in weakly acidic, aqueous
solutions) can be dried to form films, coatings and membranes
but chitosan is rarely used in a structural capacity (in spite of this
being one of its roles as a biomolecule in nature). Instead, chitosan
currently finds application in specialty products requiring antimi-
crobial properties, biocompatibility (chitosan can be formed into
implants that resist rejection from the body) and fat-binding
capacity: chitosan can be eaten but not digested, and it can bind
to fats and prevent their digestion. Because of hydroxyl and amine
side groups on the chitosan monomer, it is reactive and readily
binds with metals [17].
In developed countries, rubber and chitosan are both industrial
waste products for which much-needed recycling technologies
may be encouraged by higher value applications. We have used lar-
gely untreated rubber and chitosan as it is available from large-
quantity manufacturers to demonstrate that this type of technol-
ogy can be made available not only to the developed world where
stormwater is increasingly of concern, but also to developing
nations and areas recovering from disaster. Cleanliness of water
is of primary concern in such areas, and it is our hope that because
both chitosan and rubber are easily available and because the pro-
cessing techniques described below are simple and inexpensive
this type of technology may find use in such areas where water
quality and pollution are of more critical concern.
2. Materials and methods
We have prepared a variety of candidate permeable paving materials by mixing
crumb rubber and weakly acidic aqueous chitosan solutions, using various solution
concentrations and ratios of solution to rubber. A subset of these samples were
exposed to sodium tripolyphosphate solution to achieve chemical crosslinking
(but without the toxic consequences of typical crosslinkers such as glutaraldehyde),
as some form of crosslinking of chitosan is necessary to prevent dissolution should
 C.A. Murray et al. / Construction and Building Materials 69 (2014) 221–231
these materials encounter acidic environments. The ability of these materials to
perform as permeable pavers was characterized by measuring their hydraulic
conductivity, their ability to remove fine sediment from water, their mechanical
compressibility and their ability to remove zinc from aqueous solution.
2.1. Sample preparation
Crumb rubber (with average particle diameter of approximately 2 mm) was
donated by Emterra Rubber (Mississauga, Ontario). The crumb rubber was prepared
from a typical mix of used car and truck tires, and was used in all preparations with-
out any cleaning or other treatment beyond the mechanical and magnetic sizing
and separation carried out at the plant. As such, fine particles, impurities such as
fiber and wire were still present.
A high-molecular weight sample Chitosan was provided by Imtex membranes
Corp., and had an average molecular weight (as measured through viscometry of
weakly acidic solutions [18]) of 450,000 g/mol, and a degree of acetylation (which
ranges from 0% to 100% for pure chitosan and chitin, respectively) of DA = 81%, as
measured using potentiometric titration [19] and UV–vis spectroscopy [20]. Before
mixing with crumb rubber, chitosan was dissolved in solutions of 5% acetic acid in
water, at concentrations between 1% and 8% by mass of chitosan.
Each sample was prepared by mixing 10 g of crumb rubber with a well-defined
amount of chitosan solution (ranging from 5 to 40 g) in a cylindrical mold with
inner diameter of 3.8 cm (see Fig. 2). The mixture was left to dry in a dehydrator
and sample mass was periodically measured until no change within a 5 hour period
was observed, at which point drying of the sample was judged to be complete. Once
dried, the samples were removed from the molds, photographed and qualitatively
examined. Depending on the ratio of chitosan to rubber used and the initial chito-
san concentration the initial sample height ranged from 1.5 cm to 3 cm. As
described below, initial compression testing was performed before further treat-
ment and characterization.
Following drying and initial compression testing, half of the samples were neu-
tralized by replacing them in their molds and exposing them to an excess amount of
0.1% NaOH solution for 48 h. The samples were thoroughly rinsed with deionized
water to remove any precipitated salts and then dried. The other half of the samples
were simultaneously neutralized/insolubilized and crosslinked through exposure to
an excess amount of 5% sodium tripolyphosphate (STPP) in water for 48 h, after
which samples were thoroughly rinsed with deionized water and dried.
A small number of ‘‘conventional’’ permeable paver material samples were pre-
pared by mixing the same crumb rubber with polyurethane adhesive at various
ratios and allowing these similarly sized coupons to cure.
2.2. Hydraulic conductivity measurement
Hydraulic conductivity of the samples was measured using a modified falling-
head type permeability test [21] in which the sample was secured in a pipe (sealed
around the edge using silicone caulking to prevent water from bypassing the sam-
ple). The apparatus allowed measurement of pressure drop across the coupon
(through a clear manometer tube in communication with the upstream side of
the sample chamber) as water was allowed to flow into the top of the pipe and
through the sample. Flow rate was measured using a stopwatch and a large beaker
to collect water on the downstream side of the sample. The mass of water collected
in a given amount of time was measured using a balance and flow rate was
determined for each sample as a function of headloss (the height of water above
the sample).
Fig. 2. Digital images of the sample coupons resulting from varying the concentration of chitosan solution used (vertical axis) and the amount of chitosan solution (horizontal
axis) in combination with 10 g of crumb rubber.
223
2.3. Filtration performance
Following hydraulic testing, the filtration performance of a subset of samples
was measured. Influent water with concentration of finely ground silica particles
(sil-co-sil 106) [22] (average particle size of 22 lm) of 200 mg/l was allowed to flow
into the hydraulic testing cell, and inlet and outlet samples were collected and mea-
sured using a modified Total Suspended Solids (TSS) removal test procedure [23].
Trans-sample pressure drop and flow rate were recorded (as described above).
The samples were not tested for sediment loading capacity, and no pressure change
that might be associated with clogging was observed over the course of the
measurements.
Each collected sample was filtered through a pre-weighed glass fiber filter (with
nominal pore size of 1 lm) under light vacuum, and the sample collection beaker
was rinsed with deionized water into the filter until no trace of the sediment was
observed. The filter was dried in an oven at 85 °C and weighed to obtain a measure
of the amount of sediment in the sample. Control samples with well-defined sedi-
ment concentration were interspersed with the experimental samples to confirm
accuracy of the measurement.
2.4. Mechanical (compressive) strength
The mechanical response of the samples was evaluated by measuring stress and
strain of samples with a procedure modeled after ASTM D575 [24], using a custom-
built device consisting of a screw and force transducer, which was capable of apply-
ing in excess of 1330 N of force to the samples with a resolution of 4 N, and allowed
simultaneous measurement of compression with a resolution of 0.03 mm. The
apparatus was calibrated by applying compression to the limits of the device with-
out sample, and the stress/strain behavior of the force transducer was subtracted
from the measurements made with samples in place.
Following release from the molds and examination, preliminary measurements
of the mechanical response of the samples was made, so that any change accompa-
nying crosslinking, hydraulic or filter testing could be discerned. Compression was
kept to a minimum to avoid any destruction of the samples.
After neutralizing/crosslinking and the hydraulic and filter testing described
above, samples were replaced in the compression testing apparatus and exposed
to progressively higher values of compressive force while the displacement of the
samples was measured. The samples were thus cycled in pressure until either the
sample disintegrated or the limits of the testing device were reached. As described
in standardized compression testing procedures [24], every value of displacement
was held for approximately 3 s before pressure was measured, to allow for a
consistent amount of stress relaxation in the samples.
2.5. Zinc removal
Samples were tested for their ability to remove dissolved zinc from water using
UV–vis spectroscopy. Both neutralized and crosslinked samples were left to soak
under solutions (100 g of solution for each test, which covered the sample in small
beakers) of zinc at concentration of 3 mg/l chosen to be high but appropriate for
accelerated capacity testing. After times of 1, 2, 4 and 8 days, aliquots of the
solution surrounding the various samples were collected and measured using a
Dithizone-based spectroscopic method [25], and removal (or leaching) was
determined by comparison with the original concentration of zinc in the solution.
224 C.A. Murray et al. / Construction and Building Materials 69 (2014) 221–231
3. Results and discussion
3.1. Qualitative observations
Dramatic differences were observed in the samples as a func-
tion of the amount and concentration of chitosan solution added
to the rubber (see Fig. 2). Because of the means by which chitosan
solution dries, high ratios of solution to rubber caused crumb rub-
ber particles to be widely dispersed in the final dried material,
leading to a more porous sample with less mechanical integrity
than when a smaller amount of solution was used. On average,
the void space formed by the gaps between 10 g of crumb rubber
particles is 12–13 ml. As such, when the volume of solution used
in preparing the sample, Vsol, is greater than 12 ml, rubber particles
are forced to separate. As the amount of solution used increases
beyond this void space amount, the resulting structure becomes
increasingly fragile and includes larger gaps between rubber parti-
cles. Because the chitosan solution at the surface of the sample
dries and becomes immobile before the solution in the bulk of
the sample has time to dry and contract, in samples prepared using
large amounts of chitosan solution rubber particles become immo-
bilized near the surface (where a crust formed) and at the bottom
(where chitosan remains mobile the longest), leaving a large open
space in the center of the sample (Fig. 3a). If smaller amounts of
more concentrated solution are used, a much denser, stronger sam-
ple resulted (see Fig. 3b). Though only one sample prepared using a
chitosan solution concentration of 1% was sufficiently stable for
mechanical testing, several samples prepared using solutions of
higher concentration but smaller absolute amounts of chitosan
were stable. This result is contrary to what is expected of conven-
tionally prepared ‘‘binder-based’’ permeable materials, in which
more binding polymer generally leads to a stronger material that
is less porous.
Clearly, there must be a lower limit to the amount of chitosan
that can be used and still maintain sufficient sample integrity. In
the samples prepared for this study, the smallest amount of
chitosan that when used resulted in a cohesive sample was 0.1 g
Fig. 3. Schematic diagram indicating how composite samples prepared using large
amounts of dilute solution can lead to a fragile, open structure upon drying (a),
while for samples prepared using smaller amounts of more concentrated solution, a
denser sample may be obtained (b).
(5 g of a 2% solution), but twice this amount, when introduced as
20 g of 1% solution, resulted in a sample with too little stability
for any kind of testing. It is likely that an even smaller amount of
chitosan than 0.1 g could be used to prepare a stable sample, if
solutions of higher concentrations were used.
This qualitatively different mechanism of composite material
preparation leads to a much richer behavior than what is observed
with traditional resin-bound permeable pavement. Because the
extra degree of freedom afforded by variable solution concentra-
tion strongly affects the final sample structure, much more
variation can be achieved through adjustment of a few process
parameters.
In the ‘‘conventional’’ samples prepared by mixing polyure-
thane adhesive with crumb rubber, the stiffness and durability of
the sample increased with polyurethane content, and there was
very little change in the mass (<1%) of the sample upon drying/cur-
ing. The highest amount of polyurethane added was only barely
sufficient to fill the void space between the rubber particles, and
as such overall sample dimensions were not strongly affected by
the amount of polyurethane used. The thickness of the polyure-
thane/rubber samples were all nominally identical (16 ± 1 mm).
Because there is little change in resin volume associated with cur-
ing, the properties of the polyurethane/rubber composite material
are a relatively simple function of binder content: porosity
decreases as more binder is added, and when the void spaces
between the rubber particles are completely occupied by the
binder, a solid composite material without pores is formed upon
curing.
There was no qualitative difference observed between the
chitosan/rubber samples that had been neutralized with NaOH
exposure and those that had been simultaneously neutralized
and crosslinked through exposure to STPP.
3.2. Hydraulic conductivity
Results obtained from chitosan/rubber samples using the mod-
ified falling head method [21] are shown in Fig. 4, where flow rate
through a cross-sectional area of 11.3 cm2 is shown as a function of
pressure drop across the sample. Consistent with the qualitative
observations made above, the amount of chitosan in the sample
did not strongly affect the ease with which water passed through
the sample. The data shown in Fig. 4 was analyzed using two meth-
ods: first, it was used to calculate an average value of hydraulic
conductivity K (as is typical when characterizing the flow of water
Fig. 4. Flow rate of water measured through various chitosan/rubber samples as a
function of the height of driving head (in cm).
 C.A. Murray et al. / Construction and Building Materials 69 (2014) 221–231 225

through granular media, such as gravel, sand, and other materials used to calculate an equivalent hole diameter using the orifice
used for stormwater filtration) through the use of Eq. (1) [21]: equation [26], Eq. (2):
pffiffiffiffiffiffiffiffi
K ¼ QL=Ah; ð1Þ Q ¼ C d A 2gh; ð2Þ

where Q is the flow rate, L is the thickness of the sample, A is the
cross-sectional area of the sample and h is the height of the driving
head of water. The resulting values of hydraulic conductivity are
shown as a function of the amount of chitosan in the sample in
Fig. 5a, as a function of the amount of solution used in preparing
the sample in Fig. 5b, and as a function of solution concentration
in Fig. 5c (a dashed line is added to indicate hydraulic conductivity
values typical of sand or gravel). Clearly, the hydraulic properties of
the sample depend more strongly on the amount of solution and
solution concentration than on amount of chitosan in the sample.
This is consistent with the qualitative observations that porosity
of the sample varies strongly with the amount of chitosan solution
used, though for Vsol < 12 ml the porosity of the sample should
decrease as the amount of chitosan increases. In Fig. 5a the hydrau-
lic conductivity for samples for which Vsol < 12 ml are differentiated
from those for which Vsol > 12 ml, and when either of these subsets
of data is considered separately, the dependence of hydraulic con-
ductivity on the amount of chitosan is clear.
To facilitate comparison with other devices used for manage-
ment of stormwater quantity, the data shown in Fig. 4 were also
where Q is flow rate, Cd is the coefficient of discharge, A is the cross-
sectional area of the hole, g is the acceleration due to gravity and h
is the height of driving water. In this way, an estimate of the
hydraulic behavior of the sample was made by calculating the size
of a sharp-rimmed orifice (Cd = 0.62) that would demonstrate the
same dependence of flow rate on trans-sample pressure. A sample
curve from Fig. 4 is shown in Fig. 6, with a fit to the orifice model
(Eq. (2)) shown. The results of this calculation are shown in Fig. 7,
where Fig. 7a shows this calculated equivalent hole size as a func-
tion of the amount of chitosan in the sample, Fig. 7b shows the
equivalent hole size as a function of the amount of chitosan solution
used in preparation of the sample and Fig. 7c shows what is the
strongest correlation – effective hole size as a function of solution
concentration.
For comparison, a representative dataset of flow rate versus
trans-sample pressure for a polyurethane/rubber sample (where
2 g of polyurethane is combined with 10 g of rubber) is shown with
that of a chitosan/rubber sample (in which 10 g of 2% chitosan
solution is combined with 10 g of rubber) in Fig. 8. These two
samples have quantitatively similar hydraulic behavior but very

Fig. 5. Hydraulic conductivity, K, calculated for chitosan/rubber composite samples by application of Eq. (1) to the data shown in Fig. 4, as a function of the amount of
chitosan in the resulting sample (a), the amount of chitosan solution used in the preparation of the sample (b) and the concentration of the chitosan solution used in preparing
the sample (c). In (a) the results obtained using samples prepared using volume of solution Vsol less than the void space between rubber particles of approximately 12 ml are
differentiated from those measured for samples with Vsol > 12 ml. The dashed line indicates a hydraulic conductivity of K = 0.001 m/s, typical of sand or gravel used for
filtration.
226 C.A. Murray et al. / Construction and Building Materials 69 (2014) 221–231

different polymer contents, indicating that the same hydraulic

Fig. 6. One example data set from those shown in Fig. 4, with a fit to Eq. (2) shown,
indicating the degree to which the hydraulic behavior of the composite sample can
be described by that of a sharp orifice with 6.7 mm diameter.
behavior can be achieved with an order of magnitude less chitosan
than what would be required of from a similar sample containing
polyurethane.
The two samples shown together in Fig. 8 demonstrate that
similar values of hydraulic conductivity and porosity are possible
to achieve in these two types of permeable materials, but the very
different mechanisms by which they are prepared gives rise to a
more complex dependence of hydraulic conductivity on process
parameters for the chitosan/rubber sample than what exists for
traditional resin-bound samples. When the volatile fraction of
the binder resin and associated binder resin shrinkage with curing
are low (as is the case with polyurethane resin), the porosity of the
sample is a relatively smooth function of binder content: once the
void space formed by the rubber particles is completely filled by
the binder, the sample is completely impermeable. In contrast,
because most of the chitosan solution volume is lost during curing,
it is not trivial to achieve a completely impermeable sample using
the chitosan/rubber combination of materials. Because pore size in
the chitosan/rubber samples is not strictly a function of polymer
content, a relatively small amount of polymer can yield a dense
sample with low hydraulic conductivity, whereas a large
amount of polymer can (if introduced as a large volume of low

Fig. 7. Diameter of an orifice with equivalent hydraulic properties to that of the chitosan/rubber samples, made by fitting Eq. (2) to the data shown in Fig. 4, as a function of
the amount of chitosan in the resulting sample (a), the amount of chitosan solution used in the preparation of the sample (b) and the concentration of the chitosan solution
used in preparing the sample (c).
 C.A. Murray et al. / Construction and Building Materials 69 (2014) 221–231
Fig. 8. Comparison of flow rate, Q, versus driving head height of water, for a
chitosan/rubber composite sample and a conventional polyurethane/rubber
sample.
concentration solution) give rise to high porosity and high
hydraulic conductivity.
Consistent with qualitative observations, there was no system-
atic dependence of hydraulic conductivity on whether the samples
were neutralized with NaOH or simultaneously neutralized and
crosslinked using STPP.
3.3. Filtration testing
The fraction of fine particles removed from 200 mg/l inlet water
by passing through the chitosan/rubber samples is shown in Fig. 9,
as a function of the solution concentration used in the preparation
of the sample. The TSS removal efficiency [23] of the samples is
calculated by comparing the inlet concentration to that measured
in the outlet samples (Eq. (3)):
% TSS Removal ¼ 100  ð½C i   ½C o Þ=½C i ; ð3Þ
where [Ci] and [Co] are the concentrations of the inlet and outlet
samples, respectively.
Fig. 9. Total Suspended Solids (TSS) removal efficiency measured for chitosan/
rubber samples as a function of the concentration of chitosan solution used in their
preparation. Inset: TSS removal efficiency as a function of hydraulic conductivity K,
calculated using the data shown in Fig. 4.
227
Since there is no reason to expect these samples to remove sed-
iment by a mechanism beyond physical filtration, it is perhaps not
surprising that there is a strong correlation between hydraulic con-
ductivity and filter performance (inset to Fig. 9). It is worth noting
that the removal efficiency of these materials is comparable to that
of typical stormwater treatment practices [27]. A typical commer-
cially-available stormwater filter can achieve approximately 80%
removal of similar influent sediment (i.e. sediment of similar parti-
cle size and concentration) after passing through 20 cm of granular
filter media at a specific flow rate of under 1 l/sm2. The ‘‘slowest’’
samples tested demonstrated a removal efficiency of approxi-
mately 65% for an equivalent specific flow rate of 0.88 l/sm2 after
passing through less than 3 cm of sample. It must be noted that
no measurements were performed which might characterize the
capacity for sediment loading in these composite samples, and that
commercial stormwater treatment filters must not only achieve
adequate filtration and hydraulic conductivity, but must maintain
this performance over long periods of time without maintenance.
As such, the measurements presented here only provide prelimin-
ary support for the use of these chitosan/crumb rubber composite
materials as candidate for permeable pavers, and do not necessar-
ily indicate that they would perform comparably to commercial
products over long periods of use.
3.4. Mechanical testing
The results of compression testing of samples immediately after
preparation are shown in Fig. 10, for maximum strains between 7%
and 25%. For comparison, a line is shown on Fig. 10 indicating
200 kPa, a pressure typical of foot traffic that a permeable paving
material might be expected to withstand without significant dete-
rioration. An average compressive modulus was determined for
each sample and, similarly to the hydraulic and filtration measure-
ments, the mechanical response of the chitosan/rubber composite
materials (when all samples are considered together) displayed a
much clearer dependence on the concentration of the chitosan
solution used in preparing the sample than on the amount of chito-
san contained in the sample. Initial compressive bulk modulus val-
ues ranging from 502 kPa to 2.6 MPa are shown in Fig. 11 as a
function of the total amount of chitosan used (Fig. 11a), the
amount of solution used to prepare the sample (Fig. 11b) and the
solution concentration used to prepare the sample (Fig. 11c). As
was demonstrated in measurements of hydraulic conductivity,
Fig. 10. Initial measurements of pressure applied versus strain for a subset of
chitosan/rubber samples. The dashed line shows a pressure of 200 kPa, typical of
foot traffic.
228 C.A. Murray et al. / Construction and Building Materials 69 (2014) 221–231

Fig. 11. Compressive modulus, calculated by linear regression of the data shown in Fig. 10, as a function of the amount of chitosan in the resulting sample (a), the amount of
chitosan solution used in the preparation of the sample (b) and the concentration of the chitosan solution used in preparing the sample (c). In (a) the results obtained using
samples prepared using Vsol < 12 ml are differentiated from those measured for samples with Vsol > 12 ml.

for samples prepared using more than 12 ml of chitosan solution a

more porous final structure results, in which (on average) cross-
sections of the sample decreasingly intersect rubber particles as
pore size increases. While more concentrated solution generally
leads to a stronger sample, a greater amount of solution does
not. In Fig. 11a modulus values measured for samples prepared
with Vsol < 12 ml are differentiated from those obtained for samples
with Vsol > 12 ml, and it is clear that above and below this critical
solution volume the dependence of sample strength on the amount
of chitosan is consistent with the overall trend shown in Fig. 11c.
Comparison between preliminary mechanical testing and the
first cycles of mechanical testing after hydraulic and filtration
performance measurement showed no significant change in the
samples. There was little or no difference in the results of compres-
sion tests performed on samples that had been neutralized and
that of those that had been simultaneously crosslinked and
neutralized using STPP.
As the samples were cycled to progressively higher displace-
Fig. 12. One representative set of compression/decompression cycles for a chito-
ments and pressures, the hysteresis typical of rubber materials
san/rubber sample. The first compression cycle (shown by open triangles) and final
was observed, with higher pressures for a given displacement as compression cycle (shown by open circles) are highlighted.
pressure was increased than was measured when pressure was
decreased. A representative compression/decompression curve is
shown in Fig. 12, showing nine compression/decompression cycles, On any individual cycle, initial compressive modulus was lower
with strains ranging from 9% to 40% at the limit of the cycle. In than what was measured as the sample became increasingly com-
Fig. 12, the initial and final cycles are highlighted. pressed, an observation qualitatively consistent with the porous
 C.A. Murray et al. / Construction and Building Materials 69 (2014) 221–231 229

nature of the sample, which becomes more dense with compres-

sion. As the maximum pressure achieved on each cycle was
increased, the sample was observed to become softer (as the chito-
san matrix began to break down), such that the initial compressive
modulus decreased with each cycle. In spite of this deterioration,
the maximum modulus increased as a function of cycling, due to
the increasing compression.
Cyclic mechanical testing of many different samples was consis-
tent with the abovementioned observation that the properties of
the sample are dominated more strongly by the concentration of
chitosan solution used during preparation than the overall amount
of solution or of chitosan in the sample.
A comparison of the mechanical testing results of four samples
with identical amounts of chitosan (but varying amounts of solu-
tion and concentration of solution) is shown in Fig. 13, and shows
that sample strength increases as a function of solution concentra-
tion (or decreases as a function of amount of solution used).
A comparison of the mechanical test results of samples pre-
pared with the same amount of solution but different concentra- Fig. 14. The results of compression testing of three chitosan/rubber samples
prepared with the same amount of chitosan solution but different solution
tion is shown in Fig. 14, and indicates that the strength of the
concentrations, indicating greater compressive strength for samples prepared with
material increases with increased solution concentration. In con- higher concentration solution.
trast, comparison of results obtained for samples prepared using
solution of the same concentration (2%) but different amounts (5,
10, and 20 g) does not show any systematic dependence of
samples for which less than 12 ml of chitosan solution was used
strength on solution amount (data not shown), as the greatest
are differentiated from those samples where more solution was
amount of chitosan solution corresponded to the most porous
used. For the samples where less solution was used and the rubber
and fragile sample.
particles did not, on average, become separated by large pores
As was the case for the initial mechanical measurements before
upon curing, an expected positive correlation between mechanical
hydraulic or filter testing the multi-cycle compression testing
strength and the amount of chitosan in the sample is observed. For
showed a strong dependence on the concentration of the chitosan
samples in which more solution was used the behavior is less
solution used and on the amount of solution used, but not on the
straightforward: all have relatively low compressive strength, as
amount of chitosan in the sample. In Fig. 15, average compressive
is expected for a more fragile structure, and the mechanical
modulus values are shown (calculated as an overall average of all
strength does not seem to be strongly affected by the amount of
stress/strain measurements over all cycles) as a function of the
chitosan in the sample.
amount of chitosan in the final sample (Fig. 15a), the amount of
As a function of binding polymer used, the polyurethane/rubber
solution used in preparing the sample (Fig. 15b) and solution
samples used for comparison were all much stiffer than the rubber/
concentration (Fig. 15c).
chitosan samples. For comparison, a sample prepared using 0.5 g of
As was observed when considering hydraulic conductivity
polyurethane and 10 g of rubber demonstrated an average modu-
(shown in Fig. 5) and the initial measurement of compressive
lus of 2.3 MPa over eight cycles, and a similar average modulus
strength (shown in Fig. 11), the dependence of ultimate mechani-
of 2.3 MPa was measured for a sample prepared using 10 g of
cal strength on the amount of chitosan in the sample and on the
chitosan solution at a concentration of 2% (for a total polymer of
amount of solution used in preparing the sample is not straightfor-
0.2 g) over five cycles. With the above hydraulic conductivity
ward when all the data is considered together. In Fig. 15a the
results in mind, this indicates that while mechanical properties
similar to conventional permeable pavers can be achieved using
similar amounts of polymer, where hydraulic conductivity and fil-
tration are of primary interest much less polymer need be used to
achieve the same effect. Conversely, a chitosan/rubber sample that
has similar mechanical properties as a polyurethane/rubber sam-
ple has a much lower hydraulic conductivity and a correspondingly
higher sediment removal capability.
There was no systematic difference in mechanical strength
observed between the samples neutralized by exposure to NaOH
and those simultaneously neutralized and crosslinked with STPP.

3.5. Zinc absorption capacity

Two varieties of the chitosan/rubber composite samples

Fig. 13. The results of compression testing of four chitosan/rubber samples with
equivalent amount of chitosan, prepared using different solution concentrations
and different amounts of solution, indicating greater compressive strength for
samples prepared with smaller amounts of higher concentration solution.
described above were used to measure zinc absorption capacity:
samples prepared using 5 g of 2% solution and 10 g of 2% solution,
with 10 g of rubber. Half of this set of samples were crosslinked
using STPP and half were neutralized through NaOH exposure.
The absorption capacity of the media is shown in Fig. 16, and
indicates a capacity of up to 0.63 mg of Zn absorbed per gram of
chitosan, after 96 h of exposure. For longer absorption times,
however, the Zn concentration after exposure is much higher than
230 C.A. Murray et al. / Construction and Building Materials 69 (2014) 221–231

Fig. 15. Compressive modulus calculated by averaging all pressure/strain data for each sample, over several compressive cycles, as a function of the amount of chitosan in the
sample (a), the amount of chitosan solution used in the preparation of the sample (b) and the concentration of the chitosan solution used in preparing the sample (c). In (a) the
results obtained using samples prepared using Vsol < 12 ml are differentiated from those measured for samples with Vsol > 12 ml.

prepared with amounts of chitosan that differed by a factor of 2,

suggesting that all the chitosan in the sample was similarly
involved in adsorption of zinc (as opposed to some chitosan being
screened in the samples with higher polymer concentration).
Measurement of zinc removal by chitosan/rubber samples
crosslinked using STPP were inconclusive, possibly due to interfer-
ence of the colorimetric analysis and complexes formed in the
crosslinking process.

4. Conclusions

The results of hydraulic conductivity, TSS removal, compressive

Fig. 16. Sorption capacity of dissolved zinc, shown as mg of zinc removed for each
gram of chitosan in the chitosan/rubber sample used. Capacity is shown as a
function of time spend soaking in a zinc solution of 3 mg/l concentration, and
indicates that after 96 h the capacity of the sample has been reached and zinc
leaching from the crumb rubber can be observed.
the initial concentration, suggesting that the chitosan has reached
its sorptive capacity and zinc from the rubber continues to leach
from the rubber. Similar capacity values were obtained for samples
strength and metals removal testing of chitosan and crumb rubber
composites described here suggest that this type of composite may
be a promising candidate material for permeable pavement.
Because of the mechanism by which the chitosan solution binder
cures, the dependence of hydraulic and mechanical performance
of these materials does not follow a dependence on polymer con-
tent that is typical of binder-based permeable pavement. Instead,
much more varied final material properties can be achieved and
much more complex dependence of those properties on a small
number of process parameters is observed. When large amounts
of solution are used in the preparation of a sample (beyond the
void space in a compact sample of rubber particles) the final sam-
ple becomes highly porous and brittle, and exhibits associated high
 C.A. Murray et al. / Construction and Building Materials 69 (2014) 221–231
hydraulic conductivity and mechanical fragility. When solution
volume below this void space amount are used, expected depen-
dence of material properties on the amount of chitosan is observed.
In general, stronger materials capable of improved physical filtra-
tion can be achieved with a small amount of polymer introduced
as high concentration solution than when large amounts of chito-
san are introduced in the form of low concentration solution.
Removal of dissolved zinc by these materials suggests that appro-
priate sizing of devices incorporating this combination of waste
products may present a viable option for of pollution removal from
stormwater. Considering the ease with which these composite
materials may be prepared, the ubiquity of the source materials
and the impressive properties of the resulting cured materials,
the combination of chitosan and crumb rubber offers a potentially
low-cost, effective means of treating surface water in a wide vari-
ety of scenarios. The following conclusions can be drawn from the
results presented here:
 Compressive strength, hydraulic conductivity and physical fil-
tration efficiency quantitatively similar to that of accepted per-
meable pavers and stormwater treatment practices can be
achieved in composite materials prepared from these two ubiq-
uitous waste products.
 Unlike more traditional binder-based permeable pavement
materials, the use of a polymer binder in dilute solution pro-
vides a new degree of control in sample preparation that
strongly impacts the resulting material behavior and enables a
wide range of material properties without qualitative change
to the preparation process.
 The material properties are dominated by the porosity of the
sample, which discontinuously increases when enough polymer
solution is added to more than fill the void space between the
compact rubber particles. Above and below this critical solution
volume the hydraulic properties of the samples exhibit
expected dependence on the amount of chitosan in the sample,
though there is a large variation between results obtained from
samples made with solution above and below this critical
volume.
 Mechanical properties are similarly a discontinuous function of
the amount of chitosan in the sample, but only below this crit-
ical solution volume are they a clear function of the amount of
chitosan. For larger amounts of solution, the fragility of the
sample due to large pores dominates the mechanical response.
 Removal of dissolved zinc from water was demonstrated, and
results suggest a capacity similar to other metals removal
media may be achieved in chitosan/crumb rubber composite
materials. Beyond that capacity for absorption, the composite
materials leach zinc into the water.
 While there was no systematic dependence of the hydraulic or
mechanical properties of the samples on crosslinking through
STPP exposure, zinc removal values measured using these sam-
ples were inconclusive, suggesting possible interference with
the measurement.
231
References
[1] Imran HM, Akib S, Karim MR. Permeable pavement and stormwater
management systems: a review. Environ Technol 2010;34(18):2649–56.
[2] Scholz M, Grabowiecki P. Review of permeable pavement systems. Build
Environ 2007;42:3830–6.
[3] Brattebo B, Booth DB. Long-term stormwater quantity and quality
performance of permeable pavement systems. Water Res 2003;37:4369–76.
[4] Pratt CJ, Mantle JDG, Schofield PA. Urban stormwater reduction and quality
improvement through the use of permeable pavements. Water Sci Technol
1989;21(8–9):769–78.
[5] Newman AP, Aitken D, Antizar-Ladislao B. Stormwater quality performance of
a macro-pervious pavement car park installation equipped with channel drain
based oil and silt retention devices. Water Res 2013;47:7327–36.
[6] Maguesvari MU, Narasimha VL. Studies on characterization of pervious
concrete for pavement applications. Proc Soc Behav Sci 2013;104:198–207.
[7] Meiarashi S. Porous elastic road surface as urban highway noise measure.
Transport Res Rec 2004;1880:151–7.
[8] Cetin A. Effects of crumb rubber size and concentration on performance of
porous asphalt mixtures. Int J Polym Sci 2013:1–10.
[9] Sunthonpagasit N, Duffey MR. Scrap tires to crumb rubber: feasibility analysis
for processing facilities. Resour Conserv Recycl 2004;40:281–99.
[10] Paje SE, Luong J, Vázquez VF, Bueno M, Miró R. Road pavement rehabilitation
using a binder with a high content of crumb rubber: influence on noise
reduction. Constr Build Mater 2013;47:789–98.
[11] Richardson AE, Coventry KA, Ward G. Freeze/thaw protection of concrete with
optimum rubber crumb content. J Clean Prod 2012;23:96–103.
[12] Menichini E et al. Artificial-turf playing fields: contents of metals, PAHs, PCBs,
PCDDs and PCDFs, inhalation exposure to PAHs and related preliminary risk
assessment. Sci Total Environ 2011;409:4950–7.
[13] Llompart M, Sanchez-Prado L, Lamas JP, Garcia-Jares C, Roca E, Dagnac T.
Hazardous organic chemicals in rubber recycled tire playgrounds and pavers.
Chemosphere 2013;90:423–31.
[14] Tang Z, Butkus MA, Xie YF. Crumb rubber filtration: a potential technology for
ballast water treatment. Mar Environ Res 2006;61:410–23.
[15] Valdes JR, Liang SH. Stress-controlled filtration with compressible particles. J
Geotech Geoenviron 2006;132:861–8.
[16] Xie Y, Killian BA, Gaul AS. Filter media: crumb rubber for wastewater filtration.
Filtr Sep 2007;44:30–2Xie Y. Method of using waste tires as a filter media. US
Patent No. US6969469; 2005.
[17] Muzzarelli RAA. Selective collection of trace metal ions by precipitation of
chitosan, and new derivatives of chitosan. Anal Chim Acta 1971;54(1):133–42.
[18] Knaul JZ, Kasaai MR, Bui VT, Creber KAM. Characterization of deacetylated
chitosan and chitosan molecular weight review. Can J Chem
1998;76:1699–706.
[19] Broussignac P. Un polymere natural pecu cannu dans 1’ industrie e chitosane.
Chim Ind Genie Chim 1970;99:1241–7.
[20] Tan SC, Khor E, Tan TK, Wong SM. The degree of deacetylation of chitosan:
advocating the first derivative UV-spectrophotometry method of
determination. Talanta 1998;45:713–9.
[21] Erlingsson S. Measurement techniques for water flow. In: Dawson A, editor.
Water in road structures: movement, drainage and effects. New
York: Springer; 2009. p. 45–67.
[22] Ash M, Ash I. Handbook of fillers, extenders, and diluents. 2nd ed. Synapse
Information Resources Inc.; 2007.
[23] USEPA. Method 160.2: residue, non-filterable, gravimetric, dried at 103–105 °C
(TSS). In: Methods for chemical analysis of water and wastes, EPA-600/4-79-
020; 1983.
[24] ASTM standard D575-91. Standard test methods for rubber properties in
compression. West Conshohocken: ASTM international; 2007.
[25] American Public Health Association (APHA). Standard methods of water and
wastewater, 18th ed. Washington (DC): American Public Health Association,
American Water Works Association, Water Environment Federation
publication; 1992.
[26] Swamee PK, Swamee N. Discharge equation of a circular sharp-crested orifice. J
Hydraul Res 2010;48:106–7.
[27] Winer R. National pollutant removal performance database for stormwater
treatment practices. Ellicott City (MD): Center for Watershed Protection; 2000.