Prediction of the Flash Point Temperature of Organic Compounds with

Nov 28, 2012 - In the following study, the positional distributive contribution method is further extended for the prediction of the flash-point (FP) ...
0 downloads 0 Views 341KB Size
Article pubs.acs.org/jced

Prediction of the Flash Point Temperature of Organic Compounds with the Positional Distributive Contribution Method Qingzhu Jia,† Qiang Wang,*,† Peisheng Ma,‡ Shuqian Xia,‡ Fangyou Yan,‡ and Hongmei Tang† †

School of Material Science and Chemical Engineering, Tianjin University of Science and Technology, 13 St. TEDA, Tianjin, 300457, People’s Republic of China ‡ School of Chemical Engineering and Technology, Tianjin University, Tianjin 300072, People’s Republic of China S Supporting Information *

ABSTRACT: A new universal method to predict the properties of organic compounds named as the positional distributive contribution method has recently been proposed by our group, which has successfully been used for the prediction of critical properties (Tc, Pc, Vc, Zc), normal boiling point (Tb), normal melting point (Tm), enthalpy of vaporization at the normal boiling point (ΔvapHb), and acentric factor (ω). In the following study, the positional distributive contribution method is further extended for the prediction of the flash-point (FP) temperature of various pure organic compounds. Comparison results between experimental and predicted data provide very satisfactory results. The overall average absolute difference (AAD) for the FP prediction of 287 organic compounds is 3.77 K, and the overall average relative deviation (ARD) is 1.16 %. However, Stefanis, Constantinou, and Panayiotou’s second-order group contribution (SCP-GC) method for the prediction of FP temperature of 287 organic compounds leads to the AAD and ARD of 14.16 K and 4.64 %, respectively. In addition, 342 organics missing experimental flash-point data are predicted via the presented procedure. More importantly, the good prediction capability of the proposed method shown in our previous works and this work suggests that it could be realized to use the same universal framework to predict not only Tc, P c, Vc, Zc, Tb, T m, ΔvapHb, and ω, but also the FP of organic compounds containing various functional groups, which further demonstrates the universality and stability of our proposed method.



INTRODUCTION Flash-point (FP) temperature is one of the most widely used physical properties, used for the evaluation of the flammability hazard of combustible liquids. Above FP, a liquid is capable of producing enough vapors to form a flammable mixture with air. Thus, the determination of FPs of liquid chemicals is very important for the sake of safety purposes. In many countries, the regulations for safe handling, transportation, and storage of liquid chemicals mainly depend on their FPs, which is why experimental FP data are always desirable. However, there is often a significant gap between the demand and the availability of such data. Moreover, the experimental measurement of FP temperature is costly, laborious, and for some toxic, explosive, or radioactive compounds the experimental determination of FP can be very difficult or even impossible. Therefore, correlation methods for theoretical predictions of FPs are required. A number of studies have been carried out in the past to predict an unknown FP of pure chemicals, using, for instance, empirical correlations,1−6 quantitative structure−property relationship (QSPR) methodology,7−14 and group-contribution (GC) methods.17−23 Empirical correlation methods are one of the most widely used prediction methods.1−6 However, some physical properties © 2012 American Chemical Society

such as normal boiling point temperature, vapor pressure, and enthalpy of vaporization are required; that is, the accuracy of these methods mainly depends on the accuracy of the input parameters, and if any of the aforementioned properties’ experimental data is not available for a specific compound, these correlations cannot be used for accurate FP predictions. Flash point could also be predicted by using the QSPR methodology.7−14 This technique results in good prediction of the FPs, but its application is generally limited to a particular group of materials. For example, QSPR could predict FPs of special groups of compounds such as alkanes,7,8 organosilicons,9 esters,10 and alcohols.11 In addition, various linear and nonlinear approaches such as multiple linear regression (MLR) and neural networks were used for the prediction of FPs of pure compounds in QSPR studies.12−16 For instance, using geometrical, topological, quantum mechanical, and electronic descriptors calculated by CODESSA PRO software, Katritzky et al.13 developed a multilinear regression model for prediction of FPs of 758 organic compounds. Although their models give reasonable performances, Received: March 8, 2012 Accepted: November 5, 2012 Published: November 28, 2012 3357

dx.doi.org/10.1021/je301070f | J. Chem. Eng. Data 2012, 57, 3357−3367

Journal of Chemical & Engineering Data

Article

Table 1. Positional Distributive Group Contributions for the Prediction of the FP Temperature groupa

A/K

groupa

A/K

C−(CH2)(H)3 C−(CH)(H)3 C−(C)(H)3 C−(C)2(H)2 C−(C)3(H) C−(C)4 Cd−(H)(O) Cd−(H)2 Cd−(C)(H) C−(Cd)(C)(H)2 C−(Cd)(H)3 Cd−(C)2 C−(Cd)(C)2(H) Cd−(Cd)(H) C−(Cd)(C)3 C−(Cd)(0)(H)2 C−(O−C)(H)3 C−(O−CO)(H)3 C−(CO)(H)3 C−(C)(CO)(H)2 C−(C)2(CO)(H) C−(C)3(CO) C−(C)(O)(H)2 C−(C)2(O)(H) C−(C)3(O) C−(O)2(H)2 CO−(CH2)(O) CO−(CH)(O) CO−(C)(O) CO−(O)(H) CO−(C)(H) CO−(C)2 C−(C)(Cl)(H)2 C−(C)2(Cl)(H) C−(C)(Cl)2(H) Cb−(H) Cb−(C) C−(Cb)(H)3 C−(Cb)(C)(H)2 C−(Cb)(C)2(H) C−(Cb)(C)3 Cb−(O) Cb−(COOH) Cb−(Cb) C−(S)(H)3

−6.5126403084 −6.1334390236 −5.6825768030 9.0460547844 17.9777452335 30.0028245975 216.8651285221 −63.7322425130 100.1329219394 7.8258541494 −8.4065853094 229.1327112413 23.9508467398 160.0710688757 37.7572806360 11.5781156136 −3.7790934933 10.5454465915 14.1507180425 23.7953131313 30.5741538525 40.1518613082 5.1466835408 6.2463528096 20.8401604993 −4.7528152780 83.2486908299 57.9522411509 128.7389309004 80.9839947596 75.7957336252 62.6261536743 13.8187349774 23.1891347313 14.2822852433 61.0723788569 −192.0151512206 38.8524386955 49.0310818470 59.0803779965 64.6780839986 429.7690249931 52.2577411293 32.7931708189 33.8057453350

C−(C)(S)(H)2 C−(C)2(S)(H) C−(C)3(S) Cb−(N) C−(N)(H)3 C−(C)(N)(H)2 C−(C)2(N)(H) C−(C)(CN)(H)2 C−(C)2(CN)(H) O−(Cb)(H) O−(CH2)(H) O−(CH)(H) O−(C)(H) O−(C)2 O−(CO)(CH3) O−(CO)(CH2) O−(CO)(CH) O−(CO)(H) N−(CH2)(H)2 N−(CH)(H)2 N−(cyclopenty)(H)2 N−(cyclohexy)(H)2 N−(C)2(H) N−(C)3 N−(Cb)(H)2 N−(Cb)(C)(H) N−(Cb)(C)2 NI−(Cb)2 S−(C)(H) S−(C)2(H) Cl− ortho correctionb meta correctionb cyclopentane correction cyclohexane correction Cobc Cmbc Cpbc >(CH)-position factord >(C)