Flash point of organic binary mixtures containing alcohols: experiment and prediction

Cent. Eur. J. Chem. 11(3) 2013 388-393 DOI: 10.2478/s11532-012-0171-6 Central European Journal of Chemistry Flash point of organic binary mixtures containing alcohols: experiment and prediction Research Article Mariana Hristova 1*, Dimitar Damgaliev University of Chemical Technology and Metallurgy, 1756 Sofia, Bulgaria Received 11 August 2012; Accepted 1 November 2012 Abstract: The flash points of three organic binary mixtures containing alcohols were measured in the present work. The experimental data was obtained using the Pensky-Martens closed cup tester. The experimental data were compared with the values calculated by the Liaw model. Activity coefficients were calculated by the Wilson equation and NRTL equation. The accuracy of predicted flash point values is dependent on the thermodynamic model used for activity coefficient. Keywords: Flash point Binary mixture Pensky-Martens Prediction Versita Sp. z o.o. 1. Introduction The flash point (FP) is one of the most important flammability characteristics of liquids and low-melting substances. Knowledge of the flash points is important for classification of materials according to the classes defined in each particular regulation [1,2] and has great practical significance in handling, transport, storage and packaging of these materials. The flash point is defined as the lowest temperature (corrected to a pressure of 101.3 kpa) at which the application of an ignition source causes the vapors of a sample specimen to ignite under specified testing conditions [3]. Flash points are determined experimentally by heating the liquid in a container and then introducing a small flame just above the liquid surface. The temperature at which there is a flash/ignition is recorded as the flash point. Two general methods are called closed-cup and opencup [4,5]. The closed-cup method prevents vapors from escaping and therefore usually results in a flash point that is a few degrees lower than in an open cup. Because the two methods give different results, one must always list the testing method when listing the flash point. Flash points of common pure chemical substances are widely reported, but very limited data are available for mixtures. Since the experimental measurement of flash point is expensive and time consuming, predictive theoretical methods are required to estimate the flash points of both pure components and mixtures. Several prediction models are presented in the literature for the prediction of mixture flash point. Wickey et al. [6] reported a method for calculating the flash point of miscible, ideal solutions of petroleum blends. Catoire and Paulmer [7] proposed a model for total miscible combustible solvent blends. McGovern studied a method [8] for estimating the flash points of mixtures of oxygenated and hydrocarbon solvents and petroleum distillates. Affens and McLaren [9] suggested the model for ideal solution by the lower flammability limit (LFL) temperature dependence assumption; White [10] simplified the Affens model by ignoring the temperature dependence of LFL. Liaw et al. [11-15] have reported a series of models, which could be used for predicting the flash points for ideal and non ideal solutions. The basic assumption in these models is that the liquid phase is in equilibrium with the vapor. The non-ideality of the liquid phase is accounted by liquid-phase activity coefficients by means of thermodynamic models. The activity coefficient is a dimensionless parameter that measures the deviation from ideality in a mixture. Some of the models that can be used to obtain the activity coefficients are: Margules 388 * E-mail: mariana_hristova@abv.bg

M. Hristova, D. Damgaliev Corrected flash point = T 0 +0.25(101.3 P) Figure 1. Photograph of the experimental apparatus. [16], Van Laar [17], Wilson [18], NRTL [19], UNIFAC [20] and UNIQUAC [21]. The first four models for calculating the activity coefficients depend on experimental binary interaction parameters (BIPs). The UNIQUAC model requires only pure component molar volumes as well as surface area and volume parameters. By contrast, predictive models such as the UNIFAC do not need experimental BIPs. The contributions due to molecular interactions are obtained from a database using a wide range of experimental results. In general, Liaw et al. s model is the most frequently used, but accuracy of predicted mixture flash points depends on the reliability of input data. The flash points of three binary mixtures, 1-propanol + 1-pentanol, 4-methyl-2-pentanone (MIBK)+1-butanol and ethanol + aniline, were measured by Pensky- Martens closed cup tester, and compared with the values calculating by using Liaw`s model. The activity coefficients were estimated by using the Raoult s law, Wilson and NRTL equations. 2. Experimental procedure The experimental data was obtained using the Pensky- Martens closed cup tester (Fig. 1) model PM 1, SUB (Berlin, Germany).The closed cup tester was operated according to the standard test method, EN ISO 2719 [22]. The mixture was heated at a rate of 1.5ºC min -1 with continual stirring. The temperature control was sustained by electrical heating. Tester thermometer having a range from -7º to +110º C was used. The ambient barometric pressure was observed and recorded at the time of the test. When the pressure differed from 760 mm Hg (101.3 kpa), the flash point was corrected as follows: where T 0 is the observed flash point (ºC); P is barometric pressure (kpa). The mole fraction of each component was determined by measuring the mass using a Sartorius digital balance (sensitivity 0.0001 g, maximum load 100 g). The sample was prepared and transferred to the cup of the apparatus at least 10ºC below the expected flash point. The sample was not stirred while the flame was lowered into the cup. The flash point was the temperature at which the test flame application caused a distinct flash in the interior of the cup. The measured value was the mean of two measurements which do not differ by more than 2ºC. All materials used in this study were purchased from Merck and Fluka. Purities were at least 99.5% (analytical grade) for all compounds used for these experimental flash point determinations. 3. Results and discussion The flash point of a binary mixture can be estimated by the model developed by Liaw et al. [11]: sat sat where x i,, P and i P i, are the mole fraction, fp activity coefficient, vapour pressure at temperature T, and vapour pressure at the flash point temperature of the mixture components, respectively. If the mixture is an ideal, Eq. 1 becomes: The temperature that satisfies Eqs. 1 or 2 is the flash point temperature of the mixture. sat The vapour pressure, P, can be estimated from i an equation, such as Antoine s equation, if the required constants are known: sat Bi log Pi = Ai (3) T + C i where A i, B i and C i are the parameters of compound i. This correlation should not be used outside the temperature range at which the parameters were obtained. The parameters for the Antoine equation can be obtained from different collections [23,24]. The activity coefficients,, were estimated by the Wilson equation and the NRTL equation. The estimated activity coefficients were subsequently used to predict the corresponding flash points. In addition, it is necessary to input the flash points of the pure components into such (1) (2) 389

Flash point of organic binary mixtures containing alcohols: experiment and prediction Table 1. Antoine coefficients, flash points and molecular volume for pure components. Substance CAS number Antoine coefficients* [23] A B C V i L (cm 3 mol -1 ) FP (ºC) Exp. Aniline 62-53-3 7.2418 1675.30 200.01 91.60 70±0.5 Ethanol 64-17-5 8.2133 1652.05 231.48 58.60 13±0.7 1-Butanol 71-36-3 7.4768 1362.39 178.73 91.51 37±0.5 1-Propanol 71-23-8 7.6192 1375.14 193.01 75.09 23±0.9 1-Pentanol 71-41-0 7.1776 1314.56 168.16 108.29 49±0.5 MIBK 108-10-1 * B log P( mmhg) = A 0 T ( C) + C 6.9920 1365.03 215.90 124.89 14±0.8 Table 2. VLE parameters of the Wilson and NRTL equations for the studied systems*. Mixtures Wilson NRTL Ref. A 12 A 21 A 12 A 21 α 4-methyl-2-pentanone (1)+1-butanol (2) -323.39 1666.30 1118.87 203.72 0.30 [26] 1-propanol (1)+1-pentanol (2) -1472.74 4557.2 6478.94-3523.07 0.30 [27] Ethanol (1)+aniline (2) 862.9016 598.5729 679.8036 538.0489 0.29 [25] Wilson: ; NRTL: Figure 2. Comparison of the flash point prediction curves with experimentally derived data for 1-propanol (1) + 1-pentanol mixture. a model to predict the flash point of a mixture. The pure compound data are listed in Table 1. The parameters of the Wilson and NRTL equations were also from the literature [25-27]. It is well known that these parameters are obtained by regression of the experimental data for such binary mixtures and are listed in Table 2. The measured flash points of studied binary mixtures and those predicted by Liaw s model are presented in Tables 3-5 respectively, where ΔT fp = T experimental T predicted. In Figs. 2-4, the flash point variation between the model predictive curves and the experimentallyderived data for the binary solutions are compared. Most liquid mixtures made of members of homologous series are practically ideal. The propanol-pentanol mixture exhibits no deviation from ideal behavior and no azeotropes are present [27]. This mixture has properties that can be predicted with a simple mixing rule that ignores interactions among the individual components because these chemicals are very similar. Fig. 2 shows that predicted results by the Wilson equation and as ideal solution (Raoult s law; activity coefficients equal unity) are in excellent agreement with the experimental data. On the other hand, the NRTL model predicts flash points which differ considerably from experimental data even if the mixture exhibits minimum flash point behavior. Similar results can also be seen for ethanol-aniline mixture (Fig. 3). Experimental data for the ethanolaniline mixture were taken from the literature [28]. The Wilson and NRTL predicted values are lower than the experimental values measured. As indicated in Fig. 3, the values in the complete set of flash point experimental data are lower than those calculated for the corresponding ideal mixture. This indicates the positive deviation of the mixture from ideal behavior. In other words, the volatility of this mixture is higher and its boiling point lower than the corresponding values estimated for ideal mixtures of the same components. The formation of this mixture is associated with the predominance of repulsive interactions. Nevertheless, 390

M. Hristova, D. Damgaliev Table 3. Experimental flash points and predictions for 1-propanol (1) + 1-pentanol mixture. X 1 Exp. (ºC) Ideal DT fp / o C Wilson DT fp / o C NRTL DT fp / o C 0 49.5 49.5 0.0 49.5 0.0 49.5 0.0 0.1 42.5 43.8 1.3 43.9 1.4 23.3 19.2 0.2 39.0 39.7 0.7 39.8 0.8 19.8 19.2 0.3 35.5 36.3 0.8 36.3 0.8 19.2 16.3 0.4 33.0 33.6 0.6 33.4 0.4 19.4 13.6 0.5 29.5 31.2 1.7 30.8 1.3 19.9 9.6 0.6 28.0 29.2 1.2 28.6 0.6 20.5 7.5 0.7 25.5 27.4 1.9 26.8 1.3 20.9 4.6 0.8 25.0 25.8 0.8 25.2 0.2 21.2 3.8 0.9 23.5 24.3 0.8 24.0 0.5 21.2 2.3 1.0 23.0 23.0 0.0 23.0 0.0 23.0 0.0 Table 4. Experimental flash points and predictions for Ethanol (1) Aniline (2). X 1 Exp. (ºC) Ideal DT fp / o C Wilson DT fp / o C NRTL DT fp / o C 0 70.0 70.0 0.0 70.0 0.0 70.0 0.0 0.1 32.7 48.9 16.2 27.2 5.5 11.3 21.4 0.2-38.7-20.8-8.9-0.3 22.3 32.2 9.9 18.3 4.0 9.2 13.1 0.4 19.8 27.5 7.7 17.0 2.8 10.5 9.3 0.5-23.9-16.2-11.3-0.6 17.5 21.0 3.5 15.7 1.8 12.4 5.1 0.7-18.6-15.2-13.1-0.8 15.5 16.5 1.0 14.7 0.8 13.5 2.0 0.9 13.3 14.6 1.3 14.0 0.7 13.6 0.3 1.0 13.0 13.0 0.0 13.0 0.0 13.0 0.0 Table 5. Experimental flash points and predictions for 4-methyl-2-pentanone (1) +1-butanol (2). X 1 Exp. (ºC) Ideal DT fp / o C Wilson DT fp / o C NRTL DT fp / o C 0 37.0 37.0 0.0 37.0 0.0 37.0 0.0 0.1 31.0 33.5 2.5 31.4 0.4 31.4 0.4 0.2 28.0 30.4 2.4 27.5 0.5 27.5 0.5 0.3 25.5 27.5 2.0 24.6 0.9 24.6 0.9 0.4 23.0 24.9 1.9 22.3 0.7 22.3 0.7 0.5 19.5 22.7 3.2 20.5 1.0 20.5 1.0 0.6 19.0 20.6 1.6 18.5 0.5 18.9 0.1 0.7 17.0 18.7 1.7 17.6 0.6 17.6 0.6 0.8 16.5 17.0 0.5 16.4 0.1 16.3 0.2 0.9 15.0 15.4 0.4 15.2 0.2 15.2 0.2 1.0 14.0 14.0 0.0 14.0 0.0 14.0 0.0 391

Flash point of organic binary mixtures containing alcohols: experiment and prediction Table 6. Average absolute deviation (A.A.D.) between calculated and experimental flash points. Solution Ideal Wilson NRTL 1-propanol (1) + 1-pentanol 0.61 0.51 8.73 Ethanol (1) Aniline (2) 6.6 2.6 8.53 4-methyl-2-pentanone (1) +1-butanol (2) 1.47 0.44 0.51 point values from experimental data and predicted by the Wilson equation is approximately 5ºC. The flash point predictions for 4-methyl-2- pentanone +1-butanol mixture are presented in Fig. 4. All thermodynamic models agree in their flash point predictions. The ideal solution model predicts higher but acceptable flash point values. Table 6 includes average absolute deviation (A.A.D.) for three binary solutions: (4) Figure 3. Comparison of the flash point prediction curves with experimentally derived data for Ethanol (1) Aniline (2). A.A.D. is a measure of agreement between the experimental data and the calculated values. In the prediction model, it was assumed that the vapour phase and liquid phase of a solution are in equilibrium. The predicted data was only adequate for the data determined by the closed cup test method, and may not be appropriate to apply to the data obtained from the open cup test method because of its condition of having deviated from the vapour-liquid equilibrium. 4. Conclusions Figure 4. Comparison of the flash point prediction curves with experimentally derived data for 4-methyl-2-pentanone (1) +1-butanol (2). this effect is not strong enough to cause flash point values to be lower than the value obtained for the pure light component. The Wilson equation better represents the experimental data. The largest difference in the flash The flash points of binary mixtures containing alcohols, 1-propanol + 1-pentanol, 4-methyl-2-pentanone (MIBK) +1-butanol and ethanol + aniline, were measured by Pensky-Martens closed cup tester. The experimental data were compared with values calculated by using Liaw`s model. The activity coefficients were estimated by the Wilson equation and the NRTL equation. For the 4-methyl-2-pentanone + 1-butanol mixture, all the predictions agree with the experimental data. Significant deviations were observed for the other two mixtures when Wilson and NRTL models are used. However, the calculated values based on the Wilson equation were found to be better than those based on the NRTL equation. 392

M. Hristova, D. Damgaliev References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] Regulation (EC) No 1907/2006 of the European Parliament and of the Council concerning the Registration, Evaluation, Authorization and Restriction of Chemicals (REACH) Regulation (EC) No 1272/2008 of the European Parliament and of the Council of 16 December 2008 on classification, labeling and packaging of substances and mixtures ASTM. Annual book of standards. Philadelphia, PA: American Society for Testing and Materials, (2002) American Society for Testing and Materials, ASTM E 502-84: Selection and Use of ASTM Standards for the Determination of Flash Point of Chemicals by Closed Cup Methods (ASTM International, West Conshohocken, PA, 2000) American Society for Testing and Materials, ASTM 1310: Standard Test Method for Flash Point and Fire Point of Liquids by Tag Open-Cup Apparatus (ASTM International, West Conshohocken, PA, 2001) R.O. Wickey, D.H. Chittenden, Hydrocarb. Process, 42(6), 157 (1963) L. Catoire, S. Paulmier, J. Phys. Chem.Ref. Data 35, 9 (2006) J.L. McGovern, J. Coats Technol. 64, 810, 39 (1992) W.A. Affens, G.W. McLaren, J. Chem. Eng. Data 17, 482 (1972) D. White, C.L. Beyler, C. Fulper, J. Leonard, Fire Saf. J. 28, 1 (1997) H.-J. Liaw, Y.H. Lee, C.L. Tang, H.H. Hsu, J.H. Liu, J. Loss Prev. Process Ind. 15, 429 (2002) H.-J. Liaw, Y.Y. Chiu, J. Hazard. Mater. 101, 83 (2003) H.-J. Liaw, Y.Y. Chiu, J. Hazard. Mater. 137, 38 (2006) [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] H.-J. Liaw, Y.H. Lee, Chien-Tsun, V.Gerbaud, Chem. Eng. Sci. 63, 4543 (2008) H.-J. Liaw, Y.H. Lee, V. Gerbaud, Y.H. Li, Fluid Phase Equilib. 300, 70 (2011) K. Whol, Chem. Eng. Prog. 49, 218 (1953) R.C. Reid, J.M. Prausnitz, J.M. Poling, The Properties of Gases and Liquids (McGraw-Hill, New York, USA, 1987) G.M. Wilson, J. Am. Chem. Soc. 86, 127 (1964) H. Renon, J.M. Prausnitz, AIChe Journal, 14, 135 (1968) A. Fredenslund, R.L. Jones, J.M. Prausnitz, AIChE Journal 21, 1086 (1975) D.S.Abrams, J.M. Prausnitz, AIChE Journal, 21, 116 (1975) EN ISO 2719. Standard Test Methods for Flash Point by Pensky-Martens Closed Cup Tester T. Boublik, V. Fried, E. Hala, Vapour Pressure of Pure Substances: Selected Values of the Temperature Dependence of the Vapour Pressures of Some Pure Substances in the Normal and Low Pressure Region, 2nd edition (Elsevier Science Ltd, Amsterdam, New York, 1984) B.E. Poling, J.M. Prausnitz, J.P. O Çonnell, The Properties of Gases and Liquids, 5th edition (McGraw-Hill, New York, 2001) S. Munjal, O. Muthu, J.R. Khurma, B.D. Smith, Fluid Phase Equilibria 12, 51 (1983) E. Lladosa, J.B. Montón, M.C. Burguet, R. Muñoz, Fluid Phase Equilibria 247, 175 (2006) N.F. Martinez, E. Lladosa, M.C. Burguet, J.B. Montón, M. Yazimon, Fluid Phase Equilibria 277, 49 (2009) M. Vidal, PhD Thesis (Texas A&M University, College Station, TX, 2006) 393