Masonry units bound with waste vegetable oil – Chemical analysis and evaluation of engineering properties

Construction and Building Materials 64 (2014) 460–472

Contents lists available at ScienceDirect

Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat

Masonry units bound with waste vegetable oil – Chemical analysis and evaluation of engineering properties T. Heaton a, C. Sammon b, J. Ault c, L. Black a, J.P. Forth a,⇑ a

School of Civil Engineering, University of Leeds, UK Materials Engineering and Research Institute, Sheffield Hallam University, UK c Astbury Centre for Structural Molecular Biology, Faculty of Biological Sciences, University of Leeds, UK b

h i g h l i g h t s  Provides further confirmation of the positive use of recycled binders in construction materials.  Improves the understanding of the mechanism of the vegetable oil based binder.  Generates further clarification of the chemical changes that occur during the curing process.  Examines the significance of the chemical changes on some of the physical properties.  Apparently the stiffness of the oil is not down to cross-linking and increased molecular weight.

a r t i c l e

i n f o

Article history: Received 22 January 2014 Received in revised form 3 April 2014 Accepted 4 April 2014 Available online 5 May 2014 Keywords: Masonry units Vegetable oil binder Oxidation Infra-red spectroscopy Mass spectrometry Liquid chromatography Physical stability Compressive strength

a b s t r a c t Masonry units with attractive environmental credentials can be produced from waste aggregate materials and vegetable oils. Heat curing at low temperatures induces chemical changes in the binder which stiffen the block and afford them a compressive strength which can be compared to existing blocks and bricks. This method allows use of 100% waste materials, which far exceeds the level of replacement possible in traditional concrete and clay matrices. To better understand the chemistry of the vegetable oil binder at different stages of the curing process a range of experiments have been performed including infrared spectroscopy, liquid chromatography and mass spectrometry. Results show production of chain-shortened and oxygenated derivatives and the nature of reactions based on curing time and physical environment of the binder. Compressive strength and mass loss are dependent on curing time and the altered molecular architecture of the oil, but other physical properties are independent of the chemistry and reliant on physical concerns such as aggregate selection. Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction The promotion and development of green construction materials has been increasing in recent years, in line with efforts by governments and industry to attempt to reduce energy consumption and carbon emissions. Attempts to improve the environmental credentials of ubiquitous construction materials such as concrete blocks and clay bricks tend to involve the incorporation of wastes or recycled materials either as part of the binder matrix or aggregate portion. A substantial field of research concerning supplementary cementitious materials exists to examine the pozzolanic ⇑ Corresponding author. Tel.: +44 (0)113 3432270. E-mail address: [email protected] (J.P. Forth). http://dx.doi.org/10.1016/j.conbuildmat.2014.04.079 0950-0618/Ó 2014 Elsevier Ltd. All rights reserved.

properties of incinerator wastes and industrial by-products [1,2]. Such efforts make valuable use of chemically active or inert wastes, thus diverting them from landfill. However, the upper limit of material that can be included without significant detriment to performance tends to be low. An alternative approach to green construction is the revival of traditional building techniques and vernacular architecture; the use of locally available, sustainable materials typically applied to low-rise construction. Examples include adobe, rammed earth, timber, straw bales and unfired clay bricks [3]. Some of the materials have low strengths, but many have very attractive thermal and acoustic insulation alongside their low embodied CO2 values. This study enhances previous work utilising waste aggregate containing building products which used bitumen as a binder [4].

461

T. Heaton et al. / Construction and Building Materials 64 (2014) 460–472

Manufacture of these units required a step in which the bitumen and aggregates were pre-heated to facilitate mixing. Subsequent modifications employed the addition of waste vegetable oils to the bitumen and aggregate mixture, removing the requirement for the pre-heating step. The results presented in this paper were then the next logical step, with the vegetable oils themselves trialled as the sole binder. The waste vegetable oil binder and waste aggregates can be mixed, compacted and heat cured at low temperatures to form a masonry unit with a compressive strength of 8–10 MPa [5]. The change in strength upon curing was indicative of chemical changes in the oil and therefore a driver for the current investigation. Many assumptions can be and were drawn about the nature of the reactions occurring in the oil during the curing process based on the complex process of free radical oxidation that occurs at the carbon–carbon double bonds in the fatty acid chains present in triacylglycerols (TAGs). Background information is provided by the wealth of literature in food science publishing [6–10] and papers concerning oil-based paints [11,12]. There was, however, a requirement for specific chemical investigation of these units due to the unique physical environment in which the oil is heated and the nature of the aggregates used. The groundwork for durability studies and the stability of the binding matrix will require an understanding of this chemistry, even if full characterisation of the vast range of species is an unrealistic or unnecessary goal. Fourier Transform Infrared Spectroscopy (FTIR) with an Attenuated Total Reflectance (ATR) is a tool that permits the collection of ‘chemical fingerprint’ from a material providing detailed information about its composition and structure. When coupled with a heated stage, it facilitates the analysis of in situ measurements of changes that occur during thermal treatment, such as oxidation [13] or gelation [14]. Some products of isomerisation and secondary oxidation not previously noted were highlighted by the timeresolved spectroscopy and spectral processing. Analysis of early stage products via High Performance Liquid Chromatography (HPLC) and offline Electrospray Ionisation Mass Spectrometry (ESI–MS) provided insight into the oxidation mechanisms in the first 4 h of the curing process, as well as some new information about the concentrations and nature of chain-shortened products. Chemical analysis was complemented by selected tests of engineering properties such as compressive strength testing, mass loss during curing, volumetric and gravimetric stability after curing. The work provides an early paradigm and insight into the structure of the system, which is a valuable addition to the canon of green construction. It also supplements and relates to chemical information obtained from studies by Nuclear Magnetic Resonance (NMR) spectroscopy [15], in which consumption of unsaturated hydrocarbons was monitored and signals attributable to secondary oxidation products including aldehydes and epoxides were found. 2. Materials and methods Clean rapeseed oil (KTC, Wednesbury, UK) and the corresponding post-frying waste product were supplied by the University of Leeds catering services. The oil is a standard culinary vegetable oil and the initial approximate fatty acid composition is 62% monounsaturated, 30% polyunsaturated and 8% saturated [16]. The waste rapeseed oil is discarded at discretion based on visual inspection and odour. The waste oil was not processed except for filtering to remove solids and was stored under ambient conditions away from direct sunlight. Thin films of oil cured in the absence of aggregate were prepared by deposition of oil in a petri dish subject to curing in a convection oven at 160 ± 5 °C and samples were taken at various time intervals, being cut out of the films and stored in Eppendorf tubes at room temperature 24–48 h prior to analysis. This treatment shows how the viscosity of the oil increases and how it changes from a liquid to a gummy, sticky solid. Block sample preparation was undertaken as described previously [4,5]. In brief, a mixture of waste or model aggregate and oil were mechanically mixed and transferred to 100  100 mm steel moulds, compacted at 4 MPa in a hydraulic press, demoulded and cured in a convection oven at 160 ± 5 °C. Sample details are shown in Table 1. The incinerator bottom ash (IBA) was received from a municipal solid

Table 1 Sample properties (aggregate percentages are described in terms of the dry weight total). Sample type

Coarse aggregate Fine aggregate

90% IBA < 10 mm Glass powder block None

Oil (wt.%)

IBA block

10% IBA < 0.5 mm

12 30

Thin film

100% Glass powder < 0.5 mm None

None

100

waste incinerator in Rainham, courtesy of Veolia environmental services (UK), and was screened through a 10 mm sieve. A fine filler material was also produced from it by passing through a solid waste grinder with a 0.5 mm aperture sieve screen. Aggregates were oven dried at 105 ± 5 °C prior to storage. For the HPLC and ESI–MS analysis, block samples were prepared with a soda lime glass powder as a clean model aggregate in lieu of incinerator ashes. Samples for analysis were chiselled from the arrises of the cured blocks. All infrared spectra were recorded on a Thermo Nicolet Nexus FTIR spectrophotometer using a Graseby Specac Golden Gate temperature controlled single reflection diamond ATR accessory and a liquid nitrogen cooled Mercury Cadmium Telluride (MCT) detector. Spectra of the pre-formed samples were acquired using a resolution of 4 cm 1 and 64 scans using the clean ATR crystal as a background reference. For the in situ curing study the number of scans was reduced to 32 to improve temporal resolution. ATR-FTIR requires intimate contact between the sample and the ATR crystal as it collects the spectrum from the few microns at the sample/ATR crystal interface [17]. The pre-formed samples were clamped between the ATR crystal and a sapphire plate using a calibrated torque driver set at 90 cNm to ensure intimate contact. For the in situ experiments, the temperature of the ATR accessory was set to 160 °C and 50 ll of waste oil was deposited directly on the ATR crystal surface. For these data both the clean ATR crystal and the sample at t = 0 min were used as background references. Chromatography was performed on a Dionex (Camberley, UK) system with a P680 quarternary pump and a membrane degasser, ASI-100 sample injector and PDA-100 photodiode array detector set to a primary wavelength of 215 nm and a secondary range of 210–230 nm. A C18 Zorbax ODS analytical column (length = 250 mm, diameter = 46 mm, particle size = 5 lm, pore size = 300 Å) with a guard cartridge of the same type (Agilent) was used. Acetonitrile, propan-2-ol (Fisher, Loughborough, UK) and hexane (Acros, Geel, Belgium) were all HPLC grade. The acetonitrile and propan-2-ol made up the mobile phase and the bulk of the injection mixture and were vacuum filtered prior to use. The injection mixture was 58% acetonitrile, 37% propan-2-ol, 5% hexane (v/v/v), which has been used as a mobile phase composition [18]. Clean and waste oil samples were dissolved in the injection solvent to a concentration of 10 mg/ml. The absolute concentration of cured samples could not be measured, since solubility varies upon curing. However, a decrease in solubility was observed as curing progressed, due to formation of cross-links. These samples were washed and vortexed thoroughly with an amount of the injection solvent proportional to a concentration of 10–15 mg/ml; where the latter value assumes a yield of 100% solubility. All samples were spun at 13,000 rpm for five minutes in a benchtop centrifuge and the supernatant retained. The samples recovered from blocks were spun down twice; once to pellet the bulk of the solid material and the supernatant was spun again to ensure no glass powder fines which could cause damage to the equipment remained. A gradient of acetonitrile into propan-2-ol was used throughout. The gradient programme was 0–5 min, 0–30% propan-2-ol linear, 5–60 min 30–80% propan-2ol linear, held for 10 min and returned to 100% acetonitrile and held for 15 min re-equilibration. The flow rate for all analyses was 0.5 ml/min and the column oven temperature was 25 °C. Injection volume was 20 ll. All samples collected were stored in varying proportions of the mobile phase solvent mixture from the HPLC analysis, i.e. directly transferred from fraction tubes to Eppendorf tubes and stored at 20 °C. Fractions obtained from the oxidised oil samples were diluted 1:1 with 20 mM ammonium acetate but were also diluted with the injection solvent mixture from the HPLC analysis (acetonitrile/propan-2-ol/hexane, 58/37/5) where necessary to obtain a suitable mass spectrum for ion selection for MS/MS. MS and MS/MS analysis was performed by Z-spray nanoelectrospray ionisation mass spectrometry on a quadrupole-IMS-orthogonal time-of-flight mass spectrometer (Synapt HDMS; Waters UK Ltd., Manchester, UK). The mass spectrometer was operated in positive ion TOF–MS mode with a capillary voltage of 1.9 kV, cone voltage of 25 V, nanoelectrospray nitrogen gas pressure of 0.1 bar, backing pressure of 2.34 mbar and a trap bias of 4 V. The source and desolvation temperatures were set at 80 °C and 350 °C respectively. During TOF–MS acquisition, nitrogen was used as the collision gas in the Triwave, with a pressure of 2.4  10 3 mbar in the trap and transfer regions and 2.64  10 4 mbar in the IMS cell. For MS/MS analysis the precursor ion was selected by the quadrupole and dissociation of the precursor was induced in the trap region of the Triwave device by increasing the potential difference between the source and trap ion guides. Data were acquired over the mass range for a minimum of one minute. Mass calibration was performed by a separate injection of sodium iodide at a concentration of 2 lg lL 1.

462

T. Heaton et al. / Construction and Building Materials 64 (2014) 460–472

Compressive strength testing was carried out on block samples composed of IBA aggregate (12% oil wt.%) using a Denison hydraulic press. The ramp rate was selected based on reaching the expected failure load of the sample within 30 s. Recorded failure loads were modified by the appropriate shape factor from the European standard for testing of equivalent masonry units [19], which in the case of 65 mm high and 100 mm wide units is 0.85. To examine gravimetric and volumetric stability, samples were weighed as soon as they had been cured and were suitably cool to handle. As soon as possible after curing, Demec points were attached on two opposite faces of the samples at 50 mm from one another vertically and horizontally. The samples were kept in a temperature and humidity controlled environment (65 ± 3% RH, 21 ± 0.5 °C) and monitored to evaluate weight gain and expansion. All measurements were recorded in triplicate and the mean values and standard deviations reported here.

3. Results and discussion 3.1. Fourier Transform Infrared Spectroscopy The FTIR spectra of a 50 ml deposit of vegetable oil collected over a period of 12 h (at 0, 40, 240 and 720 min) whilst being held at 160 °C are shown in Figs. 1–4. van de Voort et al. [20] have shown the application of the ATR technique to study oxidised oils, where they used standard compounds to verify peak positions of model primary and secondary oxidation products [20] and their work gives some insight into the nature of spectral features in this region. Referenced band assignments of the main features in these data are provided in Table 2. Fig. 1 shows an expansion of the m(OH) and m(CH) region between 3800 and 2750 cm 1 which contains the most obvious and predictable feature of oil oxidation, namely the disappearance of the band at 3006 cm 1, attributable to the m(@CH) of cis double bonds. The region between 3700 and 3100 cm 1 is associated with m(OH) of hydroxyl containing compounds. Generally sharp, discrete peaks in this region indicate isolated OH groups, whereas broad features would indicate a range of different hydroxyl containing species involved in hydrogen bonding. In this system we anticipate that the broad features of the m(OH) region are due to hydroperoxides or, more likely, carboxylic acids, involved in hydrogen bonding. Carboxylic acids seem more favourable as hydroperoxides are known to decompose readily at high temperatures giving rise to secondary oxidation products. The peaks observed here indicate the presence of a secondary alcohol of some sort and although it is not easy to unambiguously identify, we propose that it is most likely due to the hydroxyl group in a sn-1,3 diacylglycerol, a product of hydrolysis and free fatty acid release.

A discrete band at 3730 cm 1 also appears as curing progresses, this is as yet unassigned, but the shape and frequency of this band would suggest an OH group not involved in hydrogen bonding. Fig. 2 shows the carbonyl region of the waste oil heated on the ATR crystal. A strong band at 1744 cm 1 is one of the defining features of the infrared spectra of virgin vegetable oils and is due to the stretching vibration of the carbonyl group in the ester linkage in triacylglycerols. A broadening of this band is an expected oxidation feature as the concentration of carbonyl containing secondary oxidation products such as aldehydes and ketones increases. The vibrational frequency of aldehydes and ketones are lower than those of esters, but they all contribute to the complex m(C@O) between 1700 and 1800 cm 1. Processing of the data, for example performing second derivatisation and/or Fourier self-deconvolution analysis (trace (ii)), better facilitates observation of the minor peaks and ‘shoulders’ in the original data and highlight the position of new features formed in this region. A band at 1700 cm 1, as shown in Fig. 3a, indicates the presence of unsaturated aldehydes [20], the saturated counterparts absorbing at 1730 cm 1, this is overlapped by the ester contribution to the peak. A peak at 1710 cm 1 is present in the cured sample (Fig. 3b, trace (ii)) and the peak at 1700 cm 1 (Fig. 3a, trace (ii)) diminishes when curing progresses. The logical assumption is that the peak at 1700 cm 1 which reduces in intensity earlier in the curing process is attributable to the unsaturated aldehydes with two double bonds, and the more stable peak at 1710 cm 1 is attributable to the trans-2-alkenal form aldehydes. This is supported by previously published 1H NMR assignments in similar samples [15]. Aldehydic termini are formed as the hydrocarbon chains undergo scission and volatile formation. The oxygenated function can be retained either on the core portion of the molecule, which is still attached to the glycerol backbone, or on the short-chain portion [23]; their infrared absorption would not be distinguishable from one another. Changes in the ester band at 1744 cm 1 include a shift to a lower wavenumber and increase in intensity. These changes are likely attributable to changes in the type and number of intermolecular interactions possible in the system as more polar oxygenated species are produced. Systematic studies on mixtures of water and acetone [24–27] clearly show that the effects of hydrogen bonding on position and intensity of the carbonyl band. The appearance of a band at the high wavenumber side of the ester peak (1785 cm 1) is not a typical feature of vegetable oil

Fig. 1. ATR-FTIR spectrum of waste oil cured at 160 °C, taken in situ over 12 h curing time (3800–2800 cm

1

). Arrows indicate change in time.

T. Heaton et al. / Construction and Building Materials 64 (2014) 460–472

463

Fig. 2. Carbonyl region of the ATR-FTIR spectrum of waste oil cured at 160 °C taken in situ at (i) t = 0 min, (ii) t = 40 min, (iii) t = 120 min and (iv) t = 720 min.

Fig. 3. Carbonyl region at (a) t = 40 min, (b) t = 720 min, (i) shows the original absorbance spectrum, (ii) shows the Fourier self-deconvolution trace and (iii) the second derivative.

464

T. Heaton et al. / Construction and Building Materials 64 (2014) 460–472

Fig. 4. Fingerprint region of the ATR-FTIR spectrum of waste oil cured at 160 °C taken in situ at (i) t = 0 min, (ii) t = 40 min, (iii) t = 120 min and (iv) t = 720 min. Inset are shown the ratio of the region at 1500–1400 cm 1 (I) and the ratio of 5–35 min in the region of 1000–950 cm 1 (II).

Table 2 Assignment of signals in ATR-FTIR spectra of edible oils [21,22]. Wavenumber (cm 1)

Assignment

3006 2953 2927 2870 2856 1744 1657 1465–1458

@CAH (cis) stretch CAH antisymmetric stretch (CH3) CAH antisymmetric stretch (CH2) CAH symmetric stretch (CH3) CAH symmetric stretch (CH2) C@O stretch (ester) C@C (cis) stretch CAH antisymmetric bend (CH3)/CAH scissor bend (CH2) CAH symmetric ‘umbrella’ bend (CH3) CACAO stretch (ester) CAO stretch (ester, antisymmetric) (OACAC stretch, ester) CAO stretch (ester, symmetric) @CAH (cis) out of plane bend CH2 (>4) rocking vibration

1377 1238 1160 1119 1096 915 722

oxidation but is observed in oil based paints [12] and may be due to species with multiple oxygenated functions in close proximity to one another such as organic carbonates, lactones or acid anhydrides [28–30]. The production of mixed multiply oxygenated species may be of interest as they will be amongst the more polar derivatives formed in the mixture and therefore may influence further reactions. The low intensity absorbance at 1657 cm 1 is due to C@C stretching in cis double bonds. As might be expected, this absorbance disappears upon curing in the same way as the aforementioned band at 3006 cm 1 does; however, over time, three new very small peaks appear in this region, of a clearly different shape and origin. These peaks are attributed to the double bonds present in the already identified a,b-unsaturated aldehydes. The so-called fingerprint region of the infrared spectrum is shown in Fig. 4. The overall increase in absorbance in the complex series of bands between 1400 and 800 cm 1 can be attributed to an increase in CAO containing species such as alcohols or ethers [29,30], the latter being found in intermolecular cross-links or the mixed multiply oxidised species described previously. The inset (I) in Fig. 4 shows the same region between 1465 and 1450 cm 1 as shown in the main part of Fig. 4, using the oil at t = 0 as a reference background. This highlights only the changes

to the sample as a function of time and bands remaining unchanged should not appear in the spectrum. Methyl and methylene groups are responsible for these peaks and therefore this is strong evidence that the scission of hydrocarbon chains is occurring indicating the formation of aldehydes and other short-chain volatile molecules. This is corroborated by the decrease in absorbance of the band at 722 cm 1, a peak that appears only due to the rocking vibration of four successive methylene groups in a hydrocarbon chain [29]. The breakdown and cross-linking of hydrocarbon chains is therefore taking place concurrently during the curing process. Evidence for carbon–carbon linkage formation is not seen, though CAC links would not absorb strongly and this is therefore not unexpected. Another feature of note is the development of bands at 985 cm 1 and 970 cm 1 highlighted in the inset (II) in Fig. 4. The former appears within 5 min but is absent by 35 min. The latter also develops within 5 min but remains until around 200 min. The absorbance at 985 cm 1 is assigned to the CAH out of plane bending vibration produced by a mixture of conjugated cis, trans and trans, trans dienes and the peak at 970 cm 1 is due to the same vibration mode from isolated trans double bonds [20,31]. These signals indicate a rapid formation and consumption of conjugated dienes, with a more prolonged production and consumption of trans isomers from monounsaturates. These species probably disappear as a result of secondary product formation in which the double bond system is removed, and the order in which this occurs is in agreement with what is known about the relative stability of mono- and poly-unsaturates [32]. There is also evidence for the formation of epoxides, namely an antisymmetric vibration from trans epoxides at 885 cm 1 [28,33], though the cis vibration at 830 cm 1 is barely notable. These secondary oxidation products can be detected at low levels in heated frying oils but are expected to be relatively unstable due to the strained nature of the epoxide ring, the cis form especially. Epoxides have also been detected by proton NMR [15] where it was also observed that the cis isomer was transient. The in situ ATR-FTIR spectroscopy measurements can be compared to the measurements obtained from pre-cured films, shown in Fig. 5. Obvious similarities include the reduction in the intensity of the C@C band attributable to cis unsaturation at 3006 cm 1 and increase and subsequent decrease of CAH bending vibrations from trans double bonds at 970 cm 1, though in these

T. Heaton et al. / Construction and Building Materials 64 (2014) 460–472

spectra the absorbance due to the cis, trans polyunsaturated isomers is not observed, which is not unexpected given its short-term presence in the in situ data. The full width at half height of the carbonyl band broadens due to the formation of the new species including aldehydes and ketones, as expected. The hydroxyl stretching absorption at 3400 cm 1 which was attributed to a DAG in the in situ data survives heating, cooling and storage. Some indications of hydrolytic or thermolytic cleavage of the ester linkages in TAG molecules might be expected; the intensity of the carbonyl peak cannot be used as a measurement of this process, however, since there are numerous carbonyl containing compounds which may contribute to it. As previously mentioned, there may be overlap with saturated aldehydes at around 1730 cm 1. Comparison of the 6 and 24 h samples in the spectra in Fig. 5 indicates the disappearance of a peak at 1119 cm 1 and the appearance of a peak at 1222 cm 1. The former is assigned as an OACAC stretching vibration in the literature [21,22] and we suggest that it is specific to the sn-2 ester linkage, which undergoes lysis more readily. The peak at 1222 cm 1, therefore, may be related to the OACAC stretch of the hydroxyl bearing position in a 1,3-diacylglycerol (DAG). This would also agree with the previous comments about the presence of DAGs that are responsible for the absorbance at 3400 cm 1. Fig. 6 shows spectra obtained from block samples made with IBA aggregate at different curing times and a spectrum of the ground IBA to show the main peaks attributable to the aggregate, shown in Table 3. Most evident is the dominance of these spectra by the m (SiAO) peaks at 1000 cm 1 in the aggregate. The m(CH) associated with alkene functional groups, which should be observed at 3006 cm 1, is barely notable even in the uncured sample (i) but some broadening of the carbonyl peak (as explained previously) can be observed. The aggregate also contains water (broad m (OH) band at ca. 3400 cm 1 and d (OAH) at 1624 cm 1) despite having been oven dried, indicating some hygroscopic behaviour during storage. The water content reduces upon curing as the water evaporates from the cured block. The presence of water in the aggregates may have implications for reactions within the aggregate and binder mixture such as the promotion of hydrolytic scission of FA chains from the glycerol core. Another notable feature of the infrared spectrum of the aggregate is its variable absorption, particularly in the region 1200–900 cm 1, where the signal from SiAO stretching occurs.

465

The incinerator ash is a highly heterogeneous mixture of ceramic and glassy materials. The silicate stretching vibration will of course be representative of the composition within the sample, therefore, is often variable and can be a broad, intense peak, or show resolved peak maxima dependent on the contact with the ATR crystal; see Fig. 6(i and ii) for an example. This variability is less noticeable in the ground material, different samples of which tend to produce more consistent absorbance data. There may be a mixture of alkali metal or alkali earth metal hydroxides present in bottom ash which result in hydroxyl stretching vibrations (in the 3600– 3100 cm 1 region) in addition to the water that the ash contains. The corresponding oxides are slaked when the ash is quenched after incineration and the development of these alkali species and their subsequent carbonation when subjected to weathering [35,36] is of interest since it affects the possible leaching of heavy metals from the material. The presence of calcium hydroxide in fresh bottom ash samples gives rise to a sharp peak at 3650 cm 1 which is absent in these samples [34], indicating that their alkali content has been reduced by carbonation during weathering. Although extensive characterisation of the aggregate is not the purpose of these analyses, it is of interest since any potential influence of reactive species in the aggregate are of interest such as alkaline conditions which could promote reactions in the intact material such as saponification or further reaction of the secondary oxidation products. The weathering of the samples and low hydroxide content was later confirmed by simultaneous thermal analysis of the material (not shown) and given the strength properties produced using different alternative aggregates [4] it is thought at this time that any reactions which could take place between the aggregate and the binder have no effect on the performance characteristics measured. Analysis of the binder itself via FTIR when mixed with the aggregate is of limited use and subtraction of the signal from the aggregate is not feasible given its variable composition. 3.2. Mass spectrometry and liquid chromatography Fig. 7 shows the chromatograms of eluate recovered from four samples: thin films cured for 2 or 4 h and from a block made with glass powder then cured for 2 or 4 h. This shows the similarities in elution behaviour and similarity in the product mixture between the model thin films and samples prepared in block form. The

Fig. 5. ATR-FTIR spectra of a pre-cured thin film of 90 lm at (i) 6 h, (ii) 12 h, (iii) 18 h and (iv) 24 h curing time.

466

T. Heaton et al. / Construction and Building Materials 64 (2014) 460–472

1.2 1.1 1.0 (i)

Absorbance

0.9 0.8 (ii) 0.7 0.6 (iii) 0.5 (iv)

0.4 0.3 (v) 0.2 0.1 (vi) 3500

3000

2500

2000

1500

1000

Wavenumbers (cm-1) Fig. 6. ATR-FTIR spectra of pre-cured aggregate block samples at (i) 0 h, (ii) 6 h, (iii) 12 h, (iv) 18 h and (v) 24 h curing time. The spectrum from the aggregate powder is shown in (vi).

Table 3 Assignment of signals in ATR-FTIR spectra of block samples containing IBA aggregate [29,34]. Wavenumber (cm 3700–3000 2922 2853 1640 1410 1200–900 873 740

1

)

Assignment OAH stretch CAH antisymmetric stretch (CH2) CAH symmetric stretch (CH2) OAH scissor CH bend SiAOASi stretching CAO out of plane bend CAO in plane bend

crucial difference exhibited by the two samples in their curing behaviour is shown in the region immediately after the void volume (6 min), in which species with short-chains and/or of high polarity elute, such as hydrocarbons formed from the scission of FA chains. The thin film eluate contains some of these species but the glass powder sample shows a much higher response, particularly after 4 h curing. The actual concentration of this material cannot be measured by UV detection without its firm identity and an absorption coefficient to utilise the Beer–Lambert law; however, it is obvious from the chromatography that short-chain polar volatiles are present at higher concentrations in the block sample than in the thin film. This raises a further point of interest. It may be the case that due to the physical environment, volatiles are retained by the block matrix, effectively trapping them from escaping during the heating process in contrast to the thin films in which they can readily escape as convection occurs in the oven. Alternatively, it may be that the relative rates of reaction are actually different in the samples and volatile formation is promoted more readily in the block sample resulting in a higher concentration, perhaps due to temperature or diffusion of oxygen. No oligomeric products were identified by HPLC or MS of samples from the glass powder samples; this is an artefact of the sample preparation which involves two steps of centrifugation, pelleting out the higher molecular weight material. Several peaks which were collected from the 4 h glass powder sample as fractions for analysis by mass spectrometry are marked in Fig. 7iv. Fig. 8 shows the initial TOF–MS spectrum of a fraction collected at 26 min. It shows a range of high molecular weight products, chain-shortened TAGs and DAGs as well as some

ubiquitous contaminants. This mixture of material was contained within one fraction and highlights the difficulty in separating such a range of products as may be expected from the high levels of thermal stress to which the oils are subjected. Further examination of selected ions present in the mass spectrum (Fig. 8) can confirm that they are oxidised TAG molecules. Work regarding the behaviour of TAG molecules during tandem mass spectrometry has been published [37,38] and there are several key behaviours to consider. One is the charge driven fragmentation which produces characteristic daughter ions analogous to diacylglycerols (DAGs), monoacylglycerols (MAGs) and free fatty acids (FFA) which can confirm the identity of an isolated TAG molecule and even provide insight into the stereochemistry due to the labile sn-2 hydrogens promoting formation of sn-1,2 or sn-2,3 fragments and the less likely production of a sn-1,3 DAG [39]. With specific reference to oxidised TAGs, there are a number of artefacts that can be formed from the oxygenated function, and Byrdwell and Neff in particular have published in this area [40–42]. Interpretation of MS/MS spectra of oxidised TAGs becomes more difficult when one considers that the oxygen atom from, for example, an epoxide, a common secondary oxidation product, can be removed and result in unsaturation at the site of oxidation. Such artefacts can also form from primary oxidation products. The oxygenated functions can serve as breaking points for charge remote fragmentations, resulting in truncated daughter ions; though this serves as a useful indicator of the position of the alteration, it also adds to the plethora of fragments. Considering the MS/MS spectrum of the species at m/z 948 as shown in Fig. 9, the exact nature of this species is challenging to determine due to the complex fragmentation pattern, and there are a range of products to consider. A detailed list of rationalisations is beyond the scope of this article, but an appraisal of the main features shows the nature of the oxidation that occurs in polyunsaturated TAGs in the cured blocks. Considering first the high mass region, the pattern shows an ammoniated ion (m/z 948), a protonated near-molecular ion (931) and three subsequent peaks at 18 Da (913, 895 and 877). The loss of 18 Da is indicative of dehydration and this can occur in MS/MS of intact TAG molecules, though typically only once. A loss of 18 Da from an oxidised TAG indicates loss of an oxygen atom and two hydrogen atoms to form additional unsaturation and two new double bonds. The DAG region shows the high mass fragment at m/z 649 which

467

Absorbance at 215nm (mAu)

T. Heaton et al. / Construction and Building Materials 64 (2014) 460–472

300

300

250

250

200

200

150

150

100

100

50

50

(i)

0 0

20

(ii)

0

40

0

20

40

700

250

600

200

500

150

400 [

Absorbance at 215nm (mAu)

800 300

300

100

200 50

100

(iii)

0 0

20

40

Time (mins)

(iv)

0 0

20

40

Time (mins)

Fig. 7. Chromatograms of solvent wash from (i) thin film cured for 2 h, (ii) thin film cured for 4 h, (iii) glass powder block cured for 2 h and (iv) glass powder block cured for 4 h.

Fig. 8. TOF–MS spectrum of fractions recovered from a HPLC run of glass powder eluate (Fig. 7iv).

corresponds to loss of a single, unoxidised oleic chain and the ammonium adduct ( 299 Da). The derivatives of the highest mass DAG fragment do not readily reveal the nature of the oxidations since they could be produced from de-oxygenation of the oxidised chains and/or retention of the intact oleic chain. Examining the charge remote fragments shows loss of C8 fragments indicating oxidation across C9, C10 which is to be expected and likely identifies an epoxide and another charge remote fragment which occurs closer to the methyl terminus.

The low mass region of this spectrum proves the presence of intact oleic acid (MAG ion, m/z 339) and indicates two linoleic chains, one with a single oxidation (acylium ion, m/z 277, epoxidised across a double bond) and one with a net addition of +34 Da (acylium ion, m/z 295) which may be the result of one +16 addition (epoxide or perhaps a keto-function across a double bond) and one +18 addition such as an alcohol. The patterns in the mass spectrum afford this rationalisation and highlight how complex interpretation of these species can be. There is no doubt

468

T. Heaton et al. / Construction and Building Materials 64 (2014) 460–472

Fig. 9. MS/MS spectrum of a parent ion selected from the TOF–MS spectrum shown in Fig. 8, m/z 948.

given the daughter ion formation that this is an oxidised TAG molecule and the data provides insights into their possible structures as the mixed multiply oxidised species OL(2ox)L(ox); though multiple oxidation make firm identification difficult as already discussed, they are likely keto and/or epoxy groups. Regarding a chain-shortened, core TAG molecule, interpretation is considerably simpler. Consider the MS/MS spectrum of a species at m/z 802 with a retention time to similar to that of the species previously described, shown in Fig. 10. The high mass pattern does not indicate the presence of an epoxide, hydroperoxy or keto function, since it only exhibits the 18, 18, 18 pattern which is common to intact TAG molecules [40]. Its identification is facile when one considers the base peak at m/z 503 which represents loss of an oleic chain and retention of the truncated FA chain, the peak at m/z 505 which represents loss of a linoleic chain and retention of the truncated FA chain and the peak at 601 in which the truncated FA chain has been lost and an OL + DAG ion remains. The presence of intact oleic and linoleic chains is confirmed by examination of the low mass region (MAG ion and acylium ion respectively). The proposed structure for the molecular ion is shown inset in Fig. 10, and shows the truncated C10 chain and retained unsaturated aldehyde. Given what is known about reactivity of FA chains it is likely that the chain which underwent scission was originally linoleic. The MS data indicates that at least some of the aldehydes identified via both 1H NMR [15] and FTIR spectroscopy are truncated TAGs in which the aldehydic carbonyl group has been retained; though with what has been discovered about short chain volatile retention in block samples it cannot be conclusively stated that they are solely responsible for those signals. 3.3. Physical stability and strength properties Fig. 11i shows the increase in compressive strength with respect to curing time of samples made with IBA aggregate. Notably, there is a ‘lag’ phase between 12 and 18 h during which time the strength barely increases at all. As described in the introduction, it is clear that as curing progresses from 0 to 24 h and the consumption of unsaturated material occurs, the compressive strength of the material increases. The mass loss from both the thin films and IBA block samples is shown in Fig. 11ii. What is most immediately apparent is the

difference between the thin films and block samples. This simple test confirms the theory that the binder in the blocks undergoes volatilisation more readily than the thin films and undergoes more extensive volatilisation. However, blocks also retain a higher concentration of short-chain species as evidenced by the chromatograms. Another note may be made about the fact that the aggregates, despite oven drying to remove the bulk of water, absorb some moisture from the atmosphere during storage, as was noted in their FTIR spectra. Oven drying the material, weighing it and then allowing it to equilibrate in the laboratory conditions in order to oven dry and weigh it again has shown this is typically in the region of 2–3 wt.% and has been considered in the calculation of the mass loss from the blocks. This moisture, escaping as steam, may carry some small molecules with it. Finally, the thin films might form a ‘skin’ of sorts that retards volatilisation, an effect which is expected to occur in bitumen samples subjected to a ‘thin film oven test’ [43], though the thickness of bitumen films prepared in such tests is considerably greater than the samples produced in this work. One striking feature of the mass loss curves is the similarity in shape to the compressive strength data, particularly in mass loss from the blocks, i.e. the rate of volatilisation is higher in the 0– 12 h period and slows during the 12–18 h period. This likely reflects the rate of reaction overall, including both volatile production and polymerisation. Diffusion of oxygen into the centre of the block may be limited and the material at the surface mostly reacted by this time, which may mean that the strength at early curing stages is determined more by stiffness at the surfaces. It may also be noted that the signals due to unsaturation in the infrared spectra diminish during this period which shows that the reactions are continuing. Bearing in mind that the blocks not only lose mass as a result of volatilisation but also desiccation of the aggregate, the gravimetric and volumetric stability of cured samples were examined under conditions of constant temperature and humidity to follow postcuring stability. The mass of samples over time under controlled conditions is shown in Fig. 12. The initial measurements were made immediately after cooling and over time upon storage, the samples clearly gain weight. The results are similar for all the samples, indicating that this behaviour is not specific to a particular stage of curing, and therefore can be attributed to the aggregate

469

T. Heaton et al. / Construction and Building Materials 64 (2014) 460–472

H+ HNH H O O

O

O

O O

O

Fig. 10. MS/MS spectrum of a parent ion selected from the TOF–MS spectrum shown in Fig. 8, m/z 802. The proposed structure of the molecule is shown inset.

9

16

8

14

Mass loss of binder (%)

Compressive strength (MPa)

10

7 6 5 4 3 2 1 0

Films

Blocks

12

18

12 10 8 6 4 2 0

6

12

18

Curing time (hours)

24

6

(i)

Curing time (hours)

24

(ii)

Fig. 11. Compressive strength (i) and mass loss and (ii) values calculated from blocks made with IBA aggregate.

6h cure

12h cure

18h cure

24h cure

985

980

Mass (g)

which, as previously mentioned, adsorbs moisture from the atmosphere during storage. The total gain in weight in the sample cured for 24 h represents 1.04% of the sample weight, which is lower than that of the dry material (2–3%, as previously mentioned). This is greater than the reduction which would be expected by dilution alone. This is not surprising when considering the aggregate has been coated in a hydrophobic material. The fact that some of the IBA-based aggregate adsorbs moisture indicates that it is not completely coated; addition of a higher wt.% of binder to these IBA-based samples, however, may affect the stability of the samples, increasing their liability to sag after demoulding then prior to and during curing. The corresponding volumetric expansion of the samples described in Fig. 12 is shown in Fig. 13(i) (horizontal) and (ii) (vertical). Similarly to the gravimetric data, curing time has no influence on expansion. Of note is the difference in the horizontal and vertical expansion, which is related to the anisotropy common in pressed manufacturing, i.e. the direction in which the material is compacted into the mould. Expansion and weight gain have to be considered in terms of manufacturing practice, however, in a similar fashion, it is recommended not to use kiln-fresh clay bricks in construction [44].

975

970

965

960 0

5

10

15

Storage duration (days) Fig. 12. Gravimetric stability of differently cured samples over 2 weeks under controlled conditions.

470

T. Heaton et al. / Construction and Building Materials 64 (2014) 460–472 6h cure

12h cure

18h cure

24h cure

6h cure

18h cure

24h cure

1400

1000

Length change (µs)

Length change (µs)

12h cure

1600

1200

800 600 400

1200 1000 800 600 400

200

200 0

0 0

5

10

15

Storage duration (days)

0

5

10

Storage duration (days)

(i)

15

(ii)

Fig. 13. Volumetric stability of differently cured samples over 2 weeks under controlled conditions. Horizontal (i) and vertical (ii) expansion are shown.

DAG

Oxidised monomeric TAG

Cross-linked TAGs Aggregate O OH

Short chain molecule O

Chain-shortened core aldehyde TAG

O O

O

O

O

OH

Cross-linked oxidised TAGs O O O

O

Oxidised monomeric TAG O

O OH

Intact TAG Oxidised, H-bonded TAGs Fig. 14. Schematic diagram indicating the major components of the masonry matrix at early curing stages.

4. Conclusions Changes in the vegetable oil binder during thermal curing have been monitored using various instrumental techniques which have provided complementary information. The chemical reactions occurring appear to be related principally to the strength properties of the unit; other properties are independent of curing time and thus of the changes in the binder. This study confirms the relative insignificance of the change in binder chemistry over curing time on particular properties such as gravimetric and volumetric stability, which are likely more related to aggregate selection, grading and porosity. Vibrational spectroscopy indicates rapid disappearance of cis double bonds and the isomerisation can be monitored in situ. Hydroperoxide formation and subsequent decomposition is difficult to highlight due to spectral overlap,

but must play a role in the initial stages of the reactions according to the currently established dogma on lipid oxidation. Common to all samples is the appearance of secondary oxidation products such as saturated and unsaturated aldehydes, epoxides and ketones and the loss of fatty acid chains from whole TAGs. Fig. 14 shows a schematic of the products both detected and expected within the block. It is clear that the initial hypothesis that the main effect that would be observed in the oil would be simply the production of high molecular weight, cross-linked TAG molecules was insufficient and indeed it now seems counter-intuitive considering the scission of the hydrocarbon chains and the mass loss which are observed. Mass loss and volatile formation from the oil when used with the IBA-based aggregate occurs more than originally thought. Gravimetric and chromatographic analyses show that volatile loss from the oils and retention by the blocks are both higher in the

T. Heaton et al. / Construction and Building Materials 64 (2014) 460–472

blocks than in the thin film model samples. The former effect has implications for manufacturing as previously noted [15] but the latter could potentially affect the aging of the cured binder and the chains that retain oxygenated functions may well participate in cross-linking or chain-branching reactions. It may be worthwhile to draw several divisions in the chemistry underlying the manufacture process.  Reactions that occur at the double bonds during curing to form oxygenated monomeric, cross-linked or volatile products.  Reactions which occur in the secondary oxidation products themselves in the short term which may or may not be promoted by the thermal and oxidative conditions during and immediately after curing.  Reactions which might occur under ambient conditions during storage or in-service use. This work demonstrates the complexity of analysing waste vegetable oils which have been mixed with waste IBA-based aggregate materials and heat cured, due to the range of free radical oxidation reactions that occur. Model samples also contain a wide range of products, but analysis indicates that the predominant alterations that occur in the hydrocarbon chains introduce polar substituents and/or alter chain length. Cross-linking reactions are more difficult to demonstrate beyond the increase in absorbance in the fingerprint region which could be attributed to ether cross-linkages. However, both types of chemical changes are occurring during heating; upon which the altered binder provides strength to the unit. Whether the binding mechanism is principally due to the formation of cross-links and formation of sticky material in the interstitial spaces or by some adherence to the aggregate molecules is unclear, but the incorporation of polar substituents may influence intermolecular interactions between TAG molecules and indeed may result in interactions with surface groups on the aggregate molecules. The masonry units can be considered particulate composites in which both coarse and fine aggregate particles are coated in a complex organic matrix that changes as a result of the curing process and contributes to the final compressive strength of the unit. Acknowledgements The authors acknowledge the Engineering and Physical Sciences Research Council (EPSRC) for their financial support in the form of a doctoral training grant. References [1] Lothenbach B, Scrivener K, Hooton RD. Supplementary cementitious materials. Cem Concr Res 2011;41(12):1244–56. http://dx.doi.org/10.1016/j.cemconres. 2010.12.001. [2] Pera J, Coutaz L, Ambroise J, Chababbet M. Use of incinerator bottom ash in concrete. Cem Concr Res 1997;27(1):1–5. http://dx.doi.org/10.1016/S00088846(96)00193-7. [3] Walker P. The good old ways. Mater World 2007;15:27–9. [4] Forth JP. Non-traditional binders for construction materials. Paper presented at the IABSE Henderson Colloquium, Cambridge, 10–12 July 2006. [5] Zoorob S, Forth JP, Bailey HK. Vegeblock: masonry units from recycled waste and vegetable oil. In: 21st international conference on solid waste technology and management, Philadelphia, USA; 2006. p. 511–20. [6] Chan HWS. Autoxidation of unsaturated lipids. 1st ed. London: Academic; 1987. [7] Frankel EN. Lipid oxidation. 1st ed. Bridgwater: The Oily Press; 1998. [8] Holman R, Elmer O. The rates of oxidation of unsaturated fatty acids and esters. J Am Oil Chem Soc 1947;24(4):127–9. [9] Holman RT. Autoxidation of fats and related substances. Prog Chem Fats Other Lipids 1954;2:51–98. [10] Gardner HW. Oxygen radical chemistry of polyunsaturated fatty acids. Free Radical Bio Med 1989;7(1):65–86. [11] Marshall GL, Lander JA. The characterization of alkyd paint binders using 13C NMR spectroscopy. Eur Polym J 1985;21(11):949–58.

471

[12] Meilunas RJ, Bentsen JG, Steinberg A. Analysis of aged paint binders by FTIR spectroscopy. Stud Conserv 1990;35(1):33–51. [13] Celina MC, Giron NH, Rojo MR. An overview of high temperature micro-ATR IR spectroscopy to monitor polymer reactions. Polymer 2012;53(20):4461–71. http://dx.doi.org/10.1016/j.polymer.2012.07.051. [14] Bajwa GS, Sammon C, Timmins P, Melia CD. Molecular and mechanical properties of hydroxypropyl methylcellulose solutions during the sol:gel transition. Polymer 2009;50(19):4571–6. http://dx.doi.org/10.1016/j.polymer. 2009.06.075. [15] Heaton T, Fisher J, Forth J. Waste rapeseed oil used as a binder for masonry units: NMR spectroscopic analysis. J Am Oil Chem Soc 2012;89(6):1101–11. http://dx.doi.org/10.1007/s11746-011-1984-8. [16] KTC. Product specification. KTC Rapeseed Oil 2008. [17] Mirabella FM. In: Mirabella FM, editor. Internal reflection spectroscopy theory and applications. New York: Marcel Dekker; 1993. [18] Segall SD, Artz WE, Raslan DS, Ferraz VP, Takahashi JA. Analysis of triacylglycerol isomers in Malaysian cocoa butter using HPLC–mass spectrometry. Food Res Int 2005;38(2):167–74. http://dx.doi.org/10.1016/ j.foodres.2004.09.008. [19] BSI. BS EN 772-1:2011. Methods of test for masonry units. Determination of compressive strength; 2011. [20] van de Voort F, Ismail A, Sedman J, Emo G. Monitoring the oxidation of edible oils by Fourier transform infrared spectroscopy. J Am Oil Chem Soc 1994;71(3):243–53. http://dx.doi.org/10.1007/bf02638049. [21] Dobson G. Spectroscopy and spectrometry of lipids – Part 1 – Lipid analysis by vibrational spectroscopy. Eur J Lipid Sci Technol 2001;103(12):815–26. [22] Guillen MD, Cabo N. Infrared spectroscopy in the study of edible oils and fats. J Sci Food Agric 1997;75(1):1–11. [23] Grosch W. Reactions of hydroperoxides – products of low molecular weight. In: Autoxidation of unsaturated lipids. Academic Press; 1987. p. 95–139. [24] Paul SO, Ford TA. Infrared spectroscopic studies of hydrogen bonded complexes of water with oxygen bases. Part 1. Aliphatic and alicyclic ketones. J Chem Soc; 1981. Faraday Transactions 2: Molecular and Chemical Physics 77 (1). p. 33–46. http://dx.doi.org/10.1039/ f29817700033. [25] Paul SO, Ford TA. Infrared spectroscopic studies of hydrogen bonded complexes of water with oxygen bases: Part IV. The infrared spectra in the OH-stretching region of HDO molecules dissolved in some ketones and ethers. J Mol Struct 1987;160(1–2):67–79. http://dx.doi.org/10.1016/00222860(87)87005-9. [26] Paul SO, Ford TA. Infrared spectroscopic studies of hydrogen-bonded complexes of water with oxygen bases—V. The effect of temperature on the OH-stretching bands of water molecules dissolved in some ketones. Spectrochim Acta Part A: Mol Spectrosc 1988;44(6):587–600. http:// dx.doi.org/10.1016/0584-8539(88)80112-0. [27] Paul SO, Ford TA. Analysis of the infrared absorption spectra of solutions of water in some organic solvents: Part 2. Measurement of band intensities. J Mol Struct 1989;198:65–75. http://dx.doi.org/10.1016/0022-2860(89)80029-8. [28] Mallegol J, Gonon L, Lemaire J, Gardette JL. Long-term behaviour of oil-based varnishes and paints 4. Influence of film thickness on the photooxidation. Polym Degrad Stabil 2001;72(2):191–7. [29] Smith B. Infrared spectral interpretation: a systematic approach. 1st ed. CRC Press; 1999. [30] Socrates G. Infrared and Raman characteristic group frequencies. Tables and charts. 3rd ed. Wiley; 2001. [31] Guillen MD, Cabo N. Some of the most significant changes in the Fourier transform infrared spectra of edible oils under oxidative conditions. J Sci Food Agric 2000;80(14):2028–36. [32] Cosgrove JP, Church DF, Pryor WA. The kinetics of the autoxidation of polyunsaturated fatty-acids. Lipids 1987;22(5):299–304. [33] Shreve OD, Heether MR, Knight HB, Swern D. Infrared absorption spectra of some epoxy compounds. Anal Chem 1951;23(2):277–82. http://dx.doi.org/ 10.1021/ac60050a014. [34] Smidt E, Meissl K, Tintner J, Ottner F. Interferences of carbonate quantification in municipal solid waste incinerator bottom ash: evaluation of different methods. Environ Chem Lett 2009. [35] Bertos MF, Li X, Simons SJR, Hills CD, Carey PJ. Investigation of accelerated carbonation for the stabilisation of MSW incinerator ashes and the sequestration of CO2. Green Chem 2004;6(8):428–36. http://dx.doi.org/ 10.1039/b401872a. [36] Chimenos JM, Fernandez AI, Nadal R, Espiell F. Short-term natural weathering of MSWI bottom ash. J Hazard Mater 2000;79(3):287–99. [37] Cheng C, Gross ML, Pittenauer E. Complete structural elucidation of triacylglycerols by tandem sector mass spectrometry. Anal Chem 1998;70(20): 4417–26. [38] Hsu FF, Turk J. Structural characterization of triacylglycerols as lithiated adducts by electrospray ionization mass spectrometry using low-energy collisionally activated dissociation on a triple stage quadrupole instrument. J Am Soc Mass Spectrom 1999;10(7):587–99. http://dx.doi.org/10.1016/S10440305(99)00035-5. pii:S1044-0305(99)00035-5. [39] Hsu F-F, Turk J. Electrospray ionization multiple-stage linear ion-trap mass spectrometry for structural elucidation of triacylglycerols: assignment of fatty acyl groups on the glycerol backbone and location of double bonds. J Am Soc Mass Spectrom 2010;21(4):657–69. http://dx.doi.org/10.1016/j.jasms.2010. 01.007.

472

T. Heaton et al. / Construction and Building Materials 64 (2014) 460–472

[40] Byrdwell WC, Neff WE. Non-volatile products of triolein produced at frying temperatures characterized using liquid chromatography with online mass spectrometric detection. J Chromatogr A 1999;852(2):417–32. [41] Byrdwell WC, Neff WE. Dual parallel electrospray ionization and atmospheric pressure chemical ionization mass spectrometry (MS), MS/MS and MS/MS/MS for the analysis of triacylglycerols and triacylglycerol oxidation products. Rapid Commun Mass Spectrom 2002;16(4):300–19.

[42] Byrdwell WC, Neff WE. Electrospray ionization MS of high MW TAG oligomers. J Am Oil Chem Soc 2004;81(1):13–26. [43] Read J, Whiteoak D. The shell bitumen handbook. 5th ed. Thomas Telford; 2003. [44] Forth J. Movement in masonry. In: Movement in masonry: ICE manual of construction materials. Thomas Telford; 2009. p. 431–42.