CHARACTERIZATION OF SILICA MODIFIED WITH SILANES BY USING THERMOGRAVIMETRIC ANALYSIS COMBINED WITH INFRARED DETECTION
ABSTRACT
Tire treads with reduced rolling resistance and increased wet grip can be achieved by coupling hydrophilic silica to hydrocarbon rubber by using an alkoxysilane. The silica surface was modified by reaction with a wide range of coupling and non-coupling silanes. The chemistry and extent of these silanizations were elucidated using thermogravimetric analysis (TGA) combined with infrared detection. The silane grafting efficiencies were typically 52–72%, but efficiencies were lower with the bulkier 3-(di-(tridecyloxypenta(ethyleneoxy))ethoxysilyl)propyl mercaptan. However, the silica surface coverage increases with increasing size of the silane. Grafting efficiencies were lower with higher silane loadings. In the TGA, ethoxy and methoxy groups are displaced from the grafted silanes mainly at moderate temperatures (up to about 495 °C) to form siloxane bridges. Over a similar temperature range, the weaker S–S bonds present in bis(3-triethoxysilylpropyl) tetrasulfide (TESPT)- or bis[3-(triethoxysilyl)propyl] disulfide (TESPD)-modified silica are cleaved, leading to weight losses from TESPT or TESPD bound at one end to the silica and from TESPT bound at both ends. The remaining weight losses from bound silanes occurred mainly at higher temperatures. In the commercial silanized silica Coupsil 8113, TGA indicates that about two of three ethoxy groups in each triethoxysilane were lost during the silanization process.
INTRODUCTION
The introduction of tire labeling and legislation in Japan, Europe, Korea, China, and elsewhere is increasing the demand for tires with both reduced rolling resistance and increased wet grip. These features can be accomplished using tire treads reinforced with silica, provided that the silica is strongly bound to the rubber. In the commonly used solution SBR/BR blend, this bonding can be achieved by coupling the hydrophilic silica to the hydrocarbon rubber by using a bifunctional alkoxysilane, such as bis(3-triethoxysilylpropyl) tetrasulfide (TESPT).1–3 The silane provides not only efficient reinforcement by bonding the silica to the rubber but also improves dispersion of the silica4 through reducing the silica–silica interactions5 and increases the compatibility of the silica with the rubber.
Usually the silica surface is silanized during mixing. This process is referred to as reactive mixing, and typically a temperature of 140–150 °C is required.6 The coupling silanes, such as TESPT, should then form covalent bonds with the rubber chains through their sulfur-containing groups during the subsequent sulfur vulcanization. However, in the more reactive silanes containing –S–S– or –SH groups, a limited amount of premature bonding to the rubber may occur during the mixing, especially at higher than recommended mixing temperatures.7 Alternatively, the silica may be modified by reacting with silane before mixing with the rubber.8 This is the approach followed in this study.
The silanol groups on the surface of the silica react with the silane, thereby generating an alcohol, normally ethanol. The reaction may be a direct displacement of alcohol from the silane by a silanol group on the silica,9 as shown in Scheme 1 with TESPT, or stepwise by hydrolysis of an alkoxy group followed by condensation of water,10 as shown in Scheme 2 with TESPT. A second, or even a third, ethoxy group on a bound silane group may be displaced by a silica silanol, resulting in the silane becoming doubly, or triply, bound to the silica at one end or at both ends, as shown in Scheme 3.11 The likelihood of a second link being made to the silica depends primarily on the availability of an appropriately sited silanol group on the silica surface. If there is an OH on an adjacent silicon atom on the silica surface, then this OH group could be well placed to react to form a six-membered cyclic siloxane structure.12Alternatively, the silanol might be on an adjacent silica particle in the silica aggregate, in which case the silane would bridge the particles. A third link to the silane would be unlikely, because of the high unlikelihood of there being a third appropriately placed silanol group. Hydrolyzed silane molecules, either bound or not bound to the silica, may also react with a second silane molecule, either not bound or bound to an adjacent position on the silica surface, leading to oligomeric silane structures on the silica surface.10 Examples of these reactions are shown in Scheme 4. Thus, during silanization with triethoxysilanes, such as TESPT, more than one ethoxy group can be displaced, and previous studies10,11,13 have indicated that about two ethoxys in each triethoxysilyl group are normally displaced.
Citation: Rubber Chemistry and Technology 92, 2; 10.5254/rct.18.82626
Citation: Rubber Chemistry and Technology 92, 2; 10.5254/rct.18.82626
Citation: Rubber Chemistry and Technology 92, 2; 10.5254/rct.18.82626
Citation: Rubber Chemistry and Technology 92, 2; 10.5254/rct.18.82626
Silanized silicas have been characterized by infrared (IR) spectroscopy,9,11,14–16 NMR spectroscopy,17–19 and inverse gas chromatography,20–24 and extensively by thermogravimetric analysis (TGA).18,22,25–33 TGA has been widely used because it is the only technique that provides an easy evaluation of the extent of silanization in a sample of silanized silica. However, apart from in the work of Law et al.25 none of the studies have considered the chemistry occurring during the TGA process in any detail.
In their investigation of silica modified with a bis(trialkoxysilylalkyl) disulfide, Law et al.25 observed two peaks in the derivative-TGA (D-TGA) plots attributed to weight losses from the silane groups bound to the silica. The first weight loss they assigned to cleavage of the –S–S– bond from silane groups bound only at one end to the silica, leading to loss of a trialkoxysilylalkylthio group. The second weight loss they assigned to the remaining weight losses from silanes grafted at one end, together with all weight losses from silane groups bound to the silica at both ends. Thus, they were able to calculate the amounts bound, differentiating grafting at one and both ends, and how they evolved with reaction time. Law et al.25 assumed that, in the silanization process, only the initial one or two alkoxy groups had been displaced in grafting at one or both ends, respectively, and that the remaining alkoxy groups only dissociated in the TGA during the second weight loss at a much higher temperature. However, as mentioned, previous studies of silica silanization10–13,30 have provided good evidence that a considerable amount of additional ethanol is displaced during the silanization reaction. In addition, it seems likely that the remaining alkoxy groups would be lost at lower temperatures in the TGA, closer to the temperatures used in the silanization reaction. Furthermore, Law et al.25 base their prediction of the temperatures at which losses will occur on a limited number of bond dissociation energies, thus assuming that only homolysis is occurring. Other types of reactions are expected to occur during the TGA process, in particular those leading to loss of the ethoxy groups.
The current study sought to address these concerns by combining TGA with IR detection (TGA-IR), to better identify what groups are fragmenting at the different temperatures, and by considering in more detail the chemistry expected in the TGA, based on an extensive survey of relevant background literature. In addition, a wide range of both coupling (Scheme 5) and non-coupling silanes (Scheme 6) was used to silanize the silica, as the current work forms the basis for a larger study comparing the properties of these silanized silicas and their performance in rubber. The TGA study was intended to enable allowance to be made, if needed, for any variation of the efficiency of these silanization reactions.
Citation: Rubber Chemistry and Technology 92, 2; 10.5254/rct.18.82626
Citation: Rubber Chemistry and Technology 92, 2; 10.5254/rct.18.82626
EXPERIMENTAL
Materials
The silanes TESPT (Si 69®), bis[3-(triethoxysilyl)propyl] disulfide (VP Si 266®, TESPD), 3-(triethoxysilyl)propyl mercaptan (VP Si 263®, TESPM), and 3-(di-(tridecyloxypenta(ethyleneoxy)) ethoxysilyl)propyl mercaptan (VP Si 363®, DTSPM) in Scheme 5 were obtained from Evonik labs in Wesseling, Germany. TESPT is actually a mixture of di- and polysulfides; an average sulfur chain length of 3.8 has been quoted,34 and this figure was assumed in the current study. The silanes 3-(triethoxysilyl)propyl thio-octanoate (NXT®, TESPO) and a reaction product of TESPO, TESPM and 2-methyl-1,3-propanediol (NXT® Z45, TESPO/M) in Scheme 5 were kindly provided by Momentive Performance Materials, Inc. (Tarrytown, NY, USA). TESPO/M is a co-oligomer combining the mercapto-silane TESPM with the blocked mercapto-silane TESPO in the ratio 55:4535; in Scheme 5, R2 is a –CH2CHMeCH2– group and R1 is –CH2CHMeCH2– or residual –Et.35,36 The silane octyltriethoxysilane (Dynasylan® OCTEO, OTES) in Scheme 6 was kindly supplied by Evonik Industries AG, and silanes methyltrimethoxysilane (MTMS), methyltriethoxysilane (MTES), trimethylchlorosilane (TMCS), and dimethyldichlorosilane (DMDCS) in Scheme 6 were obtained from Sigma-Aldrich Co. Ltd. (Poole, UK).
The silicas investigated in this study were the precipitated silicas Zeosil® 1165 MP, 1115 MP, and Premium 200 MP (provided by Solvay SA, Collonges, France) and Ultrasil® VN3 GR (provided by Evonik Industries AG), but only Zeosil 1165 MP was silanized in this work. The reported characteristic properties of these silicas are presented in Table I.14,37–40 The commercial silanized silica Coupsil 8113 (provided by Evonik Industries AG) was also compared with the silanized silicas prepared in the study. Coupsil 8113 is understood to be Ultrasil VN3 GR grafted with TESPT (12.7%, w/w).41
Silanizing silica
The grafting of Zeosil 1165 MP silica was performed using a Dean–Stark apparatus. Toluene (600 mL) containing the silica (120 g) was stirred and refluxed for 45 min to separate and collect the loosely bound physisorbed water from the silica surface. The silane solution [15% (v/v) in toluene] was then added to the reactor, and the reaction mixture was refluxed for a further 1 h. However, the temperature of the oil bath was lowered to 55 °C for silanizing with MTMS, MTES, TMCS, and DCDMS, as these silanes have lower boiling points than toluene. The silanized silica was filtered off, washed with toluene to remove any unreacted silane, and dried to constant weight under vacuum.
Two loadings of the silane TESPT were used: 8% (w/w) silica (including the physisorbed water), corresponding to the standard amount used in rubber compounding for tire applications, and 12% (w/w) (including the physisorbed water). The loadings of the other silanes were normalized to the 8% (w/w) TESPT loading to have the same number of silane groups available for silanization, by taking account of both the molecular weight and the number of silane groups in each molecule. Silica was also silanized with twice the loading of TMCS. The experiments were carried out three times to study the reproducibility of the silanization, and the reaction time with 8% TESPT was also varied from 10 min to 24 h.
TGA coupled to a fourier transform infrared spectrometer
The amount of physisorbed water and silanol groups on the untreated silica, as well as the silane on the modified silica surface, was measured with a Pyris 1 thermogravimetric analyzer connected to a Spectrum 100 FTIR spectrometer through a balanced flow FTIR EGA System TL 8000 (PerkinElmer Inc., Waltham, MA, USA). The untreated silica and silanized silica samples were heated from a room temperature (RT) of 16–22 °C to 800 °C at a heating rate of 30 °C/min in an inert gas (nitrogen, flow rate 300 mL/min) environment and then in oxygen at 800 °C for 15 min. The evolved volatiles and decomposition products were examined by IR spectroscopy from RT to 800 °C by using Spectrum 10™ software (PerkinElmer Inc.). Three replicates of TGA of silica silanized with 8% (w/w) TESPT for 1 h were carried out to confirm the reproducibility of the measurements; a 0.04% (w/w) standard deviation in weight loss between 200 and 800 °C was found. A slower heating rate of 10 °C/min was also used with silica silanized with 8% (w/w) TESPT; the TGA was very similar.
The concentrations of the fragmentation products in the TGA experiments were often too low to be clearly distinguished in the IR spectra. Consequently, larger-scale isothermal experiments were also carried out to provide better spectra, in particular at the peak temperatures in the D-TGA (rate of weight loss) plots.
For convenience in comparing the TGA curve shapes, the TGA weight loss curves were normalized at 200 °C, the temperature by which it was assumed that all the weakly bound physisorbed water was removed. This normalization avoids concerns that the original moisture content in the sample may vary depending on the ambient humidity in the laboratory, which was not controlled. Indeed, in repeated TGA measurements on the same batch of silanized silica, weight losses up to 200 °C were found to vary by up to about 30%, a variation in weight loss equivalent to about 1% of the original TGA sample weight. The amount of physisorbed water and silanol groups in the untreated silicas was first calculated, the former based on the weight loss up to 200 °C and the latter based on the weight loss above 200 °C, which was assigned to condensation of silanols, forming siloxane (–Si–O–Si–) bridges and releasing water. Then the measured weight losses from the silanized silicas were determined relative to “dry” silanized silica, that is, the weight at 200 °C after loss of the physisorbed water. The weight losses above 200 °C were corrected for the weight losses observed over the same temperature ranges from the corresponding untreated silicas, due to dehydration of the silanols, and further corrected to allow for the silanols that had been silanized, although these silanols accounted for only 3–13% of the calculated original silanol concentrations.
RESULTS AND DISCUSSION
TGA-IR of untreated silicas
The TGAs of the four silicas are compared in Figure 1. IR detection revealed that water evolved throughout the temperature range. This evolution occurred in two stages: quite rapidly below about 200 °C and more slowly above 200 °C. The former is assigned to loss of loosely bound physisorbed water, hydrogen bonded to the silanols on the silica; the latter is assigned to dehydration of the silanols, forming siloxanes and generating one molecule of water from two silanols. The amount of physisorbed water and the silanol concentration (the number of silanols per square nanometer of surface area) could be calculated and are presented in Table II, compared with reported data.9,14,22,38,39,40,42 The amounts of physisorbed water are in good agreement with the reported values, bearing in mind that this could vary with the ambient temperature and humidity. The silanol concentrations are also in good agreement with the literature values that are based on TGA or IR analysis. Lower values were reported by other analytical methods, such as 3.7 and 7.5 OH/nm2 for Zeosil 1165 MP, by hydroxide treatment (the Sears method),39,43 and by esterification with methanol,39 respectively. It is believed that these methods are only measuring silanols on the surface, whereas TGA and IR measure all silanols, including those within pores.39 No method was described by Majeste and Vincent42 for their figures for Zeosil 1115 and 1165 MP in Table II, but these figures are significantly lower than the other reported figures; so, probably their method is also only measuring silanols on the surface. Roughly one molecule of water is physisorbed on the silica per each silanol group.
Citation: Rubber Chemistry and Technology 92, 2; 10.5254/rct.18.82626
Results of TGA-IR characterization of silanized silicas
The TGA (weight percent) and D-TGA (rate of weight loss) plots for the different untreated and silanized silicas are compared in Figures 2–6. The IR bands observed are listed in Table III, and an example of the spectra is shown in Figure 7. The different chemical structures identified by IR from the different silanized silicas as the TGA temperature was increased up to 800 °C are collected and compared in Figure 8.
Citation: Rubber Chemistry and Technology 92, 2; 10.5254/rct.18.82626
Citation: Rubber Chemistry and Technology 92, 2; 10.5254/rct.18.82626
Citation: Rubber Chemistry and Technology 92, 2; 10.5254/rct.18.82626
Citation: Rubber Chemistry and Technology 92, 2; 10.5254/rct.18.82626
Citation: Rubber Chemistry and Technology 92, 2; 10.5254/rct.18.82626
Citation: Rubber Chemistry and Technology 92, 2; 10.5254/rct.18.82626
Citation: Rubber Chemistry and Technology 92, 2; 10.5254/rct.18.82626
In the silicas silanized with the bis-silanes TESPD and TESPT, the derivative plots in Figures 2 and 6 suggest that the TGA may be divided into three main periods of weight loss: up to 200 °C, between 200 and 495 °C, and above 495 °C. After all the other silane treatments, there was either only one main peak in the D-TGA plot above 200 °C, or the different weight losses in this range could not be clearly separated; thus, in the following analysis their TGA weight losses are divided into two regions: below 200 °C and above 200 °C. Following the analysis used with the untreated silicas, the TGA weight losses up to 200 °C are assigned to loss of physisorbed water, although this assignment neglects low levels of alcohol loss detected by IR in most of the TGAs below 200 °C and thus the assigned levels of physisorbed water should be considered as maximum values. Weight losses above 200 °C are assigned to fragmentation of the bound silane, corrected for water loss arising from condensation of the remaining silanols on the silica.
The IR spectra demonstrate that ethanol and methanol displacement from the alkoxysilane-treated silicas occurs at much lower temperatures than assumed by Law et al.25 Based on the IR evidence, it is assumed in the following analysis of the TGA data that ethanol and methanol displacement from the trialkoxysilane-treated silicas, leading to siloxanes, occurs mostly in the 200–495 °C range. However, as mentioned, there is IR evidence of low levels of alcohol loss below 200 °C in all of the silica samples silanized with trialkoxysilanes, apart from Coupsil 8113 (those silanized with TESPT, TESPD, TESPM, DTSPM, TESPO, OTES, MTMS, and MTES) and also above 495 °C from the samples silanized with TESPT, MTMS, and TMCS, although as discussed below the methanol generated from MTMS and TMCS at high temperatures probably arises through hydrolysis of the methyl groups on the silane. In general it is assumed that the ethanol and methanol are displaced by reaction with adjacent silanol groups or through hydrolysis by the released water molecules. In analyzing the TGA data, it is assumed that each alkoxy group is converted to a siloxane and consequently that the net molecular weight loss per alkoxy is that of the alkoxy group less half an oxygen atom.
Perhaps not surprisingly, the IR spectra have indicated that the different fragmentations occur over quite wide temperature ranges and overlap with one another. Consequently, it is not possible to fully differentiate between fragmentations based on weight losses over specific temperature ranges; assumptions have to be made in estimating the silane grafting efficiency, the alcohol loss during silanization, and, for TESPT and TESPD, the proportions bonded at both ends to silica.
Fragmentation of bound silanes in TGA-IR
A good starting point for assessing the temperatures at which different groups could be split from the silanized silica in the TGA is the relevant bond dissociation energies, as there is much published data, and the bond energies will relate to the temperatures at which different homolytic bond cleavages could occur. The relevant bond dissociation energies reported in the literature are collected in Table IV.44–59 However, other degradation processes are expected to occur, including nucleophilic displacements, such as hydrolysis of alkoxy or other groups, condensation of silanols, and concerted processes. Potential degradation reactions are discussed below for each silanized silica, roughly in the order of the temperature at which they start to fragment. It is generally assumed that at the end of the TGA process (at 800 °C), of the original silane, only the silicon atoms bonded to the silica via oxygen and any other siloxane linkages to these silicon atoms remain.
TESPT-Modified Silica
TESPT itself was found to be relatively stable in the TGA-IR up to about 274 °C, when it vaporizes. The IR spectrum of the vapor indicates that it is largely unchanged TESPT. However, poly(ethylene tetrasulfide) was reported to undergo degradation in TGA at 274 °C.60 The TESPT-modified silicas and Coupsil 8113 have peaks in the D-TGA in the 310–350 °C region and at about 560 °C. It is evident from Table IV that the polysulfide S–S bonds have the lowest dissociation energies of all the bonds in the silanes studied. Thus, these would be expected to dissociate homolytically during the lower second weight loss period up to 495 °C, leading to a loss from TESPT bound at one end to silica of half the original TESPT molecule, plus, on average, 0.9 sulfur atom (as mentioned, an average sulfur chain length of 3.8 has been assumed for TESPT). There is IR evidence for these losses with the observation of carbon disulfide (CS2) from TESPT-modified silica in isothermal measurements at 333 and 344 °C. In TESPT bound at both ends, loss of 1.8 sulfur atoms is assumed. However, the possibility that these homolytic weight losses are still occurring above 495 °C cannot be ruled out.
Formation of relatively stable diatomic sulfur (S2) may also be envisaged. It is generated at very high temperatures (above 720 °C or at 530 °C at low pressures), and its transitory presence has been demonstrated at lower temperatures by trapping,61 but we have found no reported evidence that S2 is generated in thermolysis of alkyl polysulfides. However, it is possible that S2 may be generated during TGA of TESPT-silanized silica, for instance from the tetrasulfide through the concerted mechanism shown in Scheme 7, by analogy with the reported S2 generation from dialkoxy disulfides.62 This would be an alternative to the simple homolysis suggested above.
Citation: Rubber Chemistry and Technology 92, 2; 10.5254/rct.18.82626
Based on the above-mentioned information, in interpreting the TGAs, it is assumed that during the second period of weight loss (200–495 °C) from the TESPT-modified silicas, all the alkoxy groups that remained after the silanization process are lost; that on average –S1.8SR′ is lost from silica–RS–SxSR′ in TESPT grafted at one end (where R is bound to silica, but R′ is not); and that on average 1.8 sulfur atoms are lost from TESPT grafted at both ends. All the remaining groups are then lost in the third period, from 495 to 800 °C.
TESPD-Modified Silica
TESPD itself was found to be relatively stable in the TGA-IR up to about 274 °C, when it vaporizes. The IR spectrum of the vapor was similar to that of TESPD. The TESPD-modified silicas have peaks in the D-TGA at about 400 and 560 °C. Based on the bond dissociation energies in Table IV, C-SSR homolysis would be expected to be favored over CS–SC homolysis. However, in studies of dimethyl and diethyl disulfide pyrolysis at 316 °C and above, S–S cleavage was reported to predominate over C–S cleavage, leading to thiol and thioaldehyde.63,64 It is likely that the mechanism of S–S cleavage is not simple homolysis; free-radical chain63 or concerted mechanisms are possibilities. Thus, S–S cleavage was assumed in interpreting the TGA data and also that this occurred during the second period of weight loss (200–495 °C), although some continuation above 495 °C cannot be ruled out. There is IR evidence for these losses with the observation of CO, SO2, OCS, CS2, and –COOH groups from TESPD-modified silica in an isothermal measurement at 341 °C. Consequently, it is assumed that during the second period of weight loss (200–495 °C), all the alkoxy groups that remained after the silanization process are lost and that –SR′ is lost from silica–RS–SR′ in TESPD grafted at one end (where R is bound to silica, but R′ is not). All the remaining groups are then lost in the third period, from 495 to 800 °C.
DTSPM- and TESPM-Modified Silica
TESPM itself was found to vaporize at 160 °C in the TGA; the IR spectrum of the vapor was similar to that of TESPM. DTSPM itself fragmented around 392 °C. Both DTSPM- and TESPM-modified silicas have single broad peaks in the D-TGA above 200 °C, around 395 and 440 °C, respectively.
DTSPM contains polyethyleneoxy groups. Degradation, pyrolysis, pyrolysis–mass spectrometry (MS), and TGA of polyethylene oxides (PEOs) and polyethylene glycols have been widely studied. D-TGA peaks at about 420 °C65,66 and also lower67 have been reported for PEOs, whereas degradation of PEOs in pyrolysis–MS has been reported peaking as low as about 250 °C.68 CO, MeOH, EtOH, HCHO, MeCHO, and EtOEt were prominent among the products arising from chain scission.65,66,68,69,70 Similarly, CO, alcohol, ether, and aldehyde or ketone carbonyls were observed in the TGA-IR spectrum of DTSPM-modified silica at 385 °C.
Ethanethiol has been reported to undergo pyrolysis at about 500 °C, breaking down to give H2S and ethene, by elimination of H2S.71,72 Alkenes were observed in the TGA-IR spectra of TESPM-modified silica at 414 °C (as well as alcohol, CO, OCS, CS2, and SO2) and in TESPO/M-modified silica at 447 °C (as well as CO, OCS, and RCOR′). H2S has a weak broad peak at about 2700 cm−1 and so probably would not be seen.
MTMS-, MTES-, and OTES-Modified Silica
These have broad peaks or shoulders in the D-TGA in the 400–440 °C region, corresponding in the IR to loss of ethanol or methanol. At higher temperatures (700 °C), CO and alkenes are observed, presumably associated with dissociation of the alkyl groups. It was considered that there was not a clear enough differentiation of the weight losses in the TGA to separate the losses of the alkoxy and alkyl groups.
TESPO- and TESPO/M-Modified Silica
These silicas have a broad peak in the D-TGA centered at 450–470 °C, corresponding in the IR to peaks assigned to CO, OCS, alkenes, and RCOR′ with TESPO/M and to CO, OCS, alkenes, alcohols, SO2, alkyl, and –SCO– with TESPO. These products may be associated mainly with breakdown of the alkyl chains, probably including hydrolysis of the thioester, as well as ethanol displacement from TESPO. TESPO itself seemed to mainly vaporize in the TGA at 229 °C, with the vapor having an IR spectrum similar to that of TESPO. TESPO/M showed water loss at 156 °C and degraded at 337 °C, giving a rather different IR spectrum.
TMCS-, DMDCS-, MTMS-, and MTES-Modified Silica
These silicas all have a broad peak in the D-TGA in the 430–550 °C region and a sharper peak at 680–710 °C. Trimethylsilylated silicas have been reported to lose methane above about 400 °C, based on thermal desorption–MS.73,74 The TMCS-, DMDCS-, MTMS-, and MTES-modified silicas would be expected to undergo similar displacement of the methyl groups in the TGA, and indeed methane was observed in the IR spectra from TGA of all four modified silicas at 670–700 °C and in the TMCS- and DMDCS-modified silicas also at 470 and 458 °C, respectively. In addition, methanol was observed in the spectra from TMCS-modified silica at 470 and 700 °C and in the spectrum from MTMS-modified silica at 600 °C. This presumably arises through hydrolysis of the methylsilyl linkages.
The decomposition products identified by IR, the reported bond dissociation energies, and the reported temperatures for other potential transformations were used to assess the temperature ranges over which different weight losses are believed to occur in the TGA. These are brought together in Table V and form the basis for the interpretation and analysis of the TGA data.
Grafting efficiency of silanes based on TGA-IR
Grafting efficiencies were calculated based on the following equation:
where Wm is the measured weight loss in the TGA between 200 and 800 °C and Wc is the calculated weight loss assuming all of the silane used in the silanization was chemically bound to the silica. Wm values were corrected to allow for the weight losses observed over the same temperature range from the corresponding untreated silicas, due to dehydration of the silanols, and further corrected to allow for the silanols that had been silanized, although these accounted for only 3–13% of the calculated original silanol concentrations.
Without knowing the extent of alkoxy or chloro group loss during silanization, it is not possible to unambiguously determine from the TGAs the grafting efficiency of the trialkoxysilanes and DMDCS, and also the relative amounts of the bis-silanes, TESPD and TESPT, grafted at one or both ends. Thus, in Tables VI–VIII, it was assumed in calculating the grafting efficiencies that 1, 1.5, or 2 ethoxy/methoxy groups in each trialkoxysilyl group, and 1 or 1.5 chloro groups in DMDCS, were displaced during silanization. At the same time, in the bis-silanes TESPT and TESPD, the percentage grafted at both ends of the bis-silane molecule was also calculated. This was achieved using the calculated and measured ratios of the corrected weight loss from 200 to 495 °C, relative to the total corrected weight loss from 200 to 800 °C (Wc). Using Solver in Excel (Microsoft, Redmond, WA, USA), the percentage grafted at both ends was adjusted until the calculated ratio equaled the measured ratio.
Tables VI–VIII reveal how much the calculated grafting efficiency varies with assumed alcohol displacement during silanization. Grafting efficiencies decrease with reduced alcohol displacement (from two to one out of three), but not greatly, whereas the percentages of TESPT and TESPD grafted at both ends increase to what seems to be unreasonably high levels.
The commercial silanized silica Coupsil 8113 is specified to contain 12.7% TESPT.41 Assuming that this percentage is relative to the total weight of the silica sample, that is, containing the measured 1.7% (w/w) physisorbed water, the extent of ethoxy group loss during silanization and of grafting at both ends can be calculated to match the observed TGA weight losses. This was achieved by adjusting both the percent grafted at both ends and the proportion of ethoxys lost during silanization until both the grafting efficiency equaled 100% and the calculated ratio of weight loss from 200 to 495 °C, relative to the total weight loss from 200 to 800 °C, equaled the measured ratio. On this basis, 3.80 of 6 ethoxys were lost during silanization and 49% of the TESPT was grafted at both ends (Table V); 1.90 of 3 ethoxys fits very well with the 2 of 3 alkoxy groups converted to siloxane reported in other studies.10,11,13
Thus, it seems reasonable in the current study to generally assume that two of three alkoxy groups were lost during the silanization. However, the MTMS and MTES silanizations were carried out at a much lower temperature (55 °C), so it seems reasonable in these cases to assume that only 1.5 alkoxy groups would be displaced during the silanization. The DMDCS silanization was also carried out at 55 °C, so it seems likely that only one of the two chloro groups would have been displaced. Based on the above-mentioned assumptions, the most likely grafting efficiencies and percent grafted at both ends in the TESPD and TESPT silanizations are indicated in bold in Tables VI–VIII, and these results are brought together in Figure 9.
Citation: Rubber Chemistry and Technology 92, 2; 10.5254/rct.18.82626
Setting aside for now the silanizations with the chlorosilanes, 12% (w/w) TESPT and DTSPM, the TGA results indicate that the grafting efficiency of the silica silanization process after 1 h varied with the different silanes between 52 and 72%. The bulkier silane, DTSPM, shows a lower grafting efficiency. This presumably reflects the limited silica surface available for grafting, and indeed the loading of DTSPM recommended by the manufacturer is much lower than would be needed to have the same number of silane groups as 8% (w/w) TESPT. The ether groups in bound DTSPM will be coating the silica surface through hydrogen bonding with unreacted silanols and thus preventing them from reacting with another DTSPM molecule. The grafting efficiency with 12% (w/w) TESPT is also lower than that with 8% (w/w). Again this seems to reflect the limited silica surface available for grafting and is consistent with the view that there are diminishing returns when using more than the standard 8% (w/w). A similar result was found with TMCS when double the amount was used in the silanization (Figure 9).
Following the discussion in the previous section it was assumed that in the TGA, TMCS and DMDSC lose only their methyl groups and in DMDSC, any remaining chlorine atoms. However, the obtained grafting efficiencies in Table VIII are significantly higher, especially the 107% for TMSC-modified silica, which is clearly untenable. In this case it seems that there must be some additional weight loss occurring. Trimethylsilyloxy groups bonded to carbon are readily hydrolyzed in acidic conditions.75 Something similar might be occurring at high temperatures with trimethylsilyloxy groups bonded to silicon, and calculations were carried out assuming loss of trimethylsilyl alcohol from TMCS-modified silica. The results are given in Table VIII and displayed in Figure 9 and seem more reasonable. However, it should be borne in mind that the IR results concurred with loss of the methyl groups. Perhaps both processes are occurring.
The calculated values of percentage of bis-silane grafted at both ends initially seem somewhat higher than anticipated, and they also are higher than those reported by Law et al.25 For example, if reaction at either end is statistically random and assuming that during silanization only one of three ethoxys is displaced, then 64% grafted would correspond to 25% grafted at both ends. However, it seems likely that the silica would react more readily at the other end of an already bound silane than with a new silane molecule that is dispersed within the solvent. Two other factors may also significantly increase the observed percent grafted at both ends. First, conversion of alkoxy groups at the other end of singly grafted silanes to siloxanes during the TGA process will in effect convert at least some of them to silanes grafted at both ends. If this conversion occurs below the temperature at which the silanes grafted at one end fragment, the apparent proportion grafted at both ends would be greatly increased. Second, the fragmentation of the silanes grafted at one end (CSS–SSC, CS–SS, CS–SC dissociation) may not be completed at 495 °C, especially cleavage of the stronger CS–SC bonds in grafted TESPD. After TESPT silanization, conversion of alkoxy to siloxane during TGA seems likely to be the main factor, whereas after TESPD silanization, the stronger CS–SC bond may also be very significant.
Although varying the amount of TESPT used was only briefly investigated, the evidence suggests that using more reduces the proportion grafted at both ends, as might be anticipated.
Physisorbed water on silanized silicas
Following the analysis used with the untreated silicas, TGA weight losses up to 200 °C are assigned to the loss of physisorbed water. The amount in the sample depends on the ambient humidity, which was not controlled in the laboratory where the TGA measurements were carried out. The analysis also neglects the alcohols detected by IR in most of the TGAs below 200 °C; thus, the assigned levels of physisorbed water should be considered as maximum values. In Figures 2–6 it is clear that silanization reduces the physisorbed water, by reducing the surface area available to the water and the number of silanol groups on the surface available for hydrogen bonding with the water molecules. The bulkier silanes are likely to physically block access by water to the silica. In Figure 10 the physisorbed water is plotted against estimated surface coverage. Calculation of surface coverage used the grafting efficiencies in Figure 9 and published silane group areas.15 Where silane group areas were not available, they were approximately estimated from the published values for larger or smaller groups; these data points are labeled with error arrows in the figure, although the errors involved are probably smaller than those alluded to above. Despite these uncertainties, the figure does indeed show the expected trend of physisorbed water decreasing with increasing surface coverage, although there is a large initial decrease, suggesting that a relatively small amount of silane restricts most of the water adsorption. By far the bulkiest silane, DTSPM, which can also hydrogen bond to several silanol groups, shows the smallest water content and the greatest surface coverage, even though its grafting efficiency is significantly lower (Figure 9). This demonstrates that this silane can be successfully used at a lower molar level, as recommended by the manufacturer.
Citation: Rubber Chemistry and Technology 92, 2; 10.5254/rct.18.82626
Reduction by silanization of the surface area and of the silanol groups on the silica available for hydrogen bonding might be counterbalanced to some extent by water bonding to the attached silane groups, through hydrogen bonding to hydroxy or ethoxy groups. In this case, one might expect bound TESPT and TESPD, with more ethoxy groups, to have less effect in reducing the amount of physisorbed water, especially when grafted only at one end, compared, for example, with OTES or TESPO, which have long hydrocarbon chains and no additional ethoxy groups at the other end. However, in Figure 10, there is no evidence of this; if anything the reverse is the case, with TESPD and TESPT having less physisorbed water.
Variation of reaction time with TESPT
The effect of reaction time was investigated using TESPT [8% (w/w)], with times ranging from 10 min up to 24 h. As shown in Figure 11, the TGA plots are very similar with similar weight losses above 200 °C, indicating that the silanization was largely completed after only 10 min. There is no evidence of the third weight loss (495–800 °C) increasing with reaction time, which would have indicated either additional silanization or conversion of singly grafted silane to silane grafted at both ends, but there is evidence of a small drop in the second weight loss (200–495 °C) over the first hour, indicating a small increase in loss of ethoxy groups between 10 min and 1 h reaction time, but thereafter negligible changes.
![Fig. 11. — . TGA plots of Zeosil 1165 MP silica silanized with TESPT [8% (w/w)] for 10 min to 24 h.](/view/journals/rcat/92/2/i0035-9475-92-2-237-f11.png)
![Fig. 11. — . TGA plots of Zeosil 1165 MP silica silanized with TESPT [8% (w/w)] for 10 min to 24 h.](/view/journals/rcat/92/2/full-i0035-9475-92-2-237-f11.png)
![Fig. 11. — . TGA plots of Zeosil 1165 MP silica silanized with TESPT [8% (w/w)] for 10 min to 24 h.](/view/journals/rcat/92/2/inline-i0035-9475-92-2-237-f11.png)
Citation: Rubber Chemistry and Technology 92, 2; 10.5254/rct.18.82626
The TGA data were analyzed, assuming that two of three ethoxy groups were displaced after the standard 1 h reaction time, and the percent grafted at both ends was thus calculated as 59.3% after the 1 h reaction. It was then assumed that the 59.3% grafted at both ends remained constant with varying reaction time, and the percent alkoxy loss was calculated for the other reaction times. This seems to increase a little from 59 to 67%, with reaction time increasing from 10 min to 1 h, but thereafter remains roughly constant.
Blume13 has compared the reactivity of different silanes toward silica. TESPM, TESPD, and MTES were found to have reactivity similar to that of TESPT, whereas OTES was significantly less reactive, DTSPM more reactive, and MTMS the most reactive. The other silanes in the current study (TESPO, TESPO/M, TMCS, and DMDCS) were not included in Blume's investigation, but the chlorosilanes are likely to be more reactive than the alkoxysilanes. Because TESPT silanization seems largely completed after only 10 min, it seems likely that the 1 h reaction time was sufficient for all of the silanes investigated here. Bearing in mind the lower reaction temperature used with MTES, MTMS, TMCS, and DMDCS, of the silanes studied, only OTES and MTES might conceivably have not completely reacted after 1 h. However, the OTES and MTES grafting efficiencies shown in Figure 9 are a little greater than that of TESPT [8% (w/w)], indicating that reaction with these two silanes was indeed largely completed after 1 h.
CONCLUSIONS
The chemistry and extent of silanization of silica with a wide range of coupling and non-coupling silanes were elucidated using TGA-IR. The chemistry occurring during TGA was assessed based on IR identification of the fragments released at different temperatures, together with an extensive survey of relevant background literature. The silane grafting efficiencies were mainly in the range 52–72%, although the bulkier silane DTSPM shows a lower grafting efficiency. However, the silica surface coverage seems to increase with increasing size of the silane. This presumably reflects the limited silica surface available for grafting, and indeed the loading of DTSPM recommended by the manufacturer is much lower than that used here, which provides the same number of silane groups as the standard 8% (w/w) TESPT. The grafting efficiency with 12% (w/w) TESPT is also lower than that with 8% (w/w). Again this seems to reflect the limited silica surface available for grafting and is consistent with the view that there are diminishing returns when using more than the standard 8% (w/w).
In the commercial silanized silica Coupsil 8113, the TGA results indicate that about two of three ethoxy groups in each triethoxysilane group were displaced during the silanization process, and other studies of silanization using TESPT have come to a similar conclusion.10,11,13 Thus, loss of two of three ethoxy or methoxy groups during silanization was generally assumed when interpreting the TGA results in the current study. In the TGA, ethoxy and methoxy groups are displaced from the grafted silanes at moderate temperatures (up to about 495 °C) to form siloxane bridges. Over a similar temperature range, the weaker S–S bonds present in TESPT or TESPD bound to silica also undergo homolysis, leading to weight losses from TESPT or TESPD bound only at one end to the silica, and from TESPT bound at both ends. All the remaining weight losses from the bound silanes occur at higher temperatures, from about 400 °C up to 800 °C.
For the bis-silanes TESPT and TESPD, there are essentially three unknowns in interpreting TGA data: the percentage of the silane used that has become bound to the silica (the grafting efficiency), the percentage of this that is bound at both ends, and the amount of additional ethanol that is displaced during silanization. With the other silanes there are only two unknowns, but the TGA weight losses could not be clearly differentiated into the two regions used in the TESPT- and TESPD-silanized silica TGA analysis (200–495 and 495–800 °C). Consequently, assumptions have to be made in interpreting TGA analyses of silanized silicas, and these should be borne in mind. However, TGA is still a very useful and easily applied technique for comparing silanizations and assessing their efficiencies.
This study forms part of a larger PhD study20 in which the surface energies of the silanized silicas were characterized using inverse gas chromatography, and the effects of the silanized silicas in rubber were evaluated. The properties compared included silica dispersion76 and the mechanical properties of silica-filled vulcanizates. This work is being reported separately.
Direct reaction between silica silanol and alkoxy silane (TESPT).
Hydrolysis of an alkoxy group in TESPT followed by condensation with silica silanol.
Second grafting of TESPT to silica.
Oligomerization of alkoxy silane (TESPT).
Molecular structures of bifunctional coupling silanes.
Molecular structures of non-coupling silanes.
TGA of untreated Zeosil 1115 MP, 1165 MP, and Premium 200 MP and Ultrasil VN3 GR silicas.
TGA and D-TGA plots for (a) untreated Zeosil 1165 MP silica and Zeosil 1165 MP silanized with (b) TESPD, (c) 8% TESPT, and (d) 12% TESPT.
TGA and D-TGA plots for Zeosil 1165 MP silica silanized with (a) TESPM, (b) DTSPM, (c) TESPO, and (d) TESPO/M.
TGA and D-TGA plots for Zeosil 1165 MP silica silanized with (a) OTES, (b) MTMS, and (c) MTES.
TGA and D-TGA plots for Zeosil 1165 MP silica silanized with (a) TMCS and (b) DMDCS.
TGA and D-TGA plots for (a) untreated Ultrasil VN3 GR silica and (b) Coupsil 8113-silanized silica.
IR spectrum of evolved vapor from TGA at 536 °C of silica silanized with 8% TESPT.
Chemical structures identified by IR spectroscopy in TGA of silanized silicas.
Concerted elimination of S2 from tetrasulfide.
Grafting efficiency and percentage of bis-silanes grafted at both ends.
Physisorbed water on silanized silicas, compared with estimated surface coverage (silane group areas of data points labeled with error arrows were estimated from reported values for other silanes; those without error bars used reported areas for those silanes15).
TGA plots of Zeosil 1165 MP silica silanized with TESPT [8% (w/w)] for 10 min to 24 h.
Contributor Notes