Methyl Linoleate and Methyl Oleate Bond Dissociation Energies

Jan 9, 2018 - Bond dissociation in a 44-reaction database set for which experimental energies are known is used to evaluate methods. We find that the ...
0 downloads 12 Views 2MB Size
Subscriber access provided by READING UNIV

Article

Methyl Linoleate and Methyl Oleate Bond Dissociation Energies: Electronic Structure Fishing for Wise Crack Products Zachary R. Wilson, and Matthew R Siebert Energy Fuels, Just Accepted Manuscript • DOI: 10.1021/acs.energyfuels.7b02798 • Publication Date (Web): 09 Jan 2018 Downloaded from http://pubs.acs.org on January 13, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Energy & Fuels is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Methyl Linoleate and Methyl Oleate Bond Dissociation Energies: Electronic Structure Fishing for Wise Crack Products Zachary R. Wilson and Matthew R. Siebert* Department of Chemistry, Missouri State University, Springfield, Missouri 65897, United States

ABSTRACT

The world depends on petroleum for everything from the plastics that contain our food, to the natural gas that heats our homes, to the gasoline that feed our cars’ engines. With rising prices of petroleum reflecting demand for this finite resource, attention has been turned to alternative sources of energy. Biodiesel, which exhibits many of the same properties as conventional diesel but is derived from biological sources, is an attractive alternative. Fats and oils are converted to biodiesel, fatty acid methyl esters (FAMEs), by transesterification. FAMEs are subsequently thermally cracked to form more light-weight transportation fuels such as natural gas, kerosene, and possibly gasoline. We aim to further understand the thermal cracking procedure, at an atomic-level, in hopes that this may aid in future engineering of viable fuels. We will present our study on the effective computational modeling of bond dissociations in the FAMEs methyl linoleate and methyl oleate, which are the most common biodiesel product of soybeans and

ACS Paragon Plus Environment

1

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 33

rapeseeds (also known as canola seeds). We have employed quantum chemical methods, including: the density functionals B3LYP, M06-2X, and B97D; the wavefunction-based MP2; and the composite CBS-QB3 method. Bond dissociation in a 44-reaction database set for which experimental energies are known are used to evaluate methods. We find that the M06-2X/631+G(d,p) model chemistry provides results comparable to the composite CBS-QB3 method at a much reduced cost. Lastly, data are compiled for possible bond dissociations in FAMEs methyl oleate and methyl linoleate.

ACS Paragon Plus Environment

2

Page 3 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

INTRODUCTION Crude oil, and the products derived from it, drive much of our modern society. The refining process provides us with, e.g., plastics used in everyday life, natural gas used to heat homes and cook food, and the fuel used for transportation. However, the current supply of crude oil is both finite and in high-demand as the population grows and develops.1-11 The process of extracting crude oil from the Earth is a very costly endeavor as it involves locating, drilling, transporting, and refining.2, 4 The environmental impact of utilizing crude oil has also proven to be very costly. Among its many impacts fossil-based fuels are extracted from the ground, resulting in previously-sequestered carbon entering the atmosphere to contribute to global warming effects.5, 6, 8-10, 12-18

All of the above motivates society to find an alternate source of transportation fuel. One of these possible solutions may be biofuels;1,

3, 5-10, 12-14, 16-23

the biofuel with, perhaps, the most

promise is biodiesel. Biodiesels are a mixture of fatty acid methyl esters (FAMEs) derived from naturally occurring fats and/or oils.1, 3, 6-10, 12-25 These FAMEs are characterized by containing monoalkyl chains 12-20 carbons in length.23 Biodiesel compounds have many favorable qualities as a transportation fuel. First, synthesis of biodiesel is a renewable resource.8,

10, 12-14, 17, 19

Given that the triglycerides used to produce

biodiesel are often formed by photosynthesis,1, 3, 6, 24 the process of synthesizing biodiesel is a carbon-neutral process.6,

18

Biodiesel is, reportedly, less hazardous to the environment than

traditional diesel fuel.5, 7-10, 12-18 Further, the emissions generated by biodiesel contain less SOX compounds and soot particles.5, 8-10, 12-15, 17, 18 Synthesis of FAMEs occurs via transesterification of a triglyceride with methanol and a catalyst1, 3, 5-10, 12-21, 23, 24 liberating glycerol (a commodity chemical itself).3 Although either of an

ACS Paragon Plus Environment

3

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 33

acid or a base can be used to catalyze transesterifaction, industrially catalysis is affected by bubbling hydrochloric acid followed by sodium hydroxide.7,

18

The identity of the FAMEs

produced by transesterification is dependent upon the triglyceride.12, 15, 18 O H

O

H

H C O C R O H C O C R' O H C O C R'' H triglyceride (fat or oil)

H3C O C R

H C OH O transesterification CH3OH (3 eq)

H C OH

H3C O C R'

H C OH H glycerol

O H3C O C R'' fatty acid methyl esters (FAMEs)

Figure 1. Fats and/or oils undergo transesterification in the presence of methanol and a catalyst to form the commodity chemical glycerol and fatty acid methyl esters (FAMEs; biodiesel). The exact identities of the FAMEs depends on the identity of the fat or oil used. In the United States, the most common source of triglyceride stock is soybean oil.3, 10, 16, 20, 21 Linoleic acid appears in the largest proportion in soybean oil at 52 – 56 wt % (23.0 – 25.0 wt % oleic acid).15, 19 In Europe, the main biodiesel feedstock is canola oil and its largest constituent is oleic acid (60.0 – 64.3 wt %) while linoleic acid comprises 19.1 – 20.0 wt % of canola oil.3, 10, 16, 20, 21

As linoleic acid and oleic acid appear in the largest proportion in the most commonly used

stock in the United States and Europe, this study will focus on the methyl esters of these fatty acids (Figure 2).

ACS Paragon Plus Environment

4

Page 5 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

O H3C O methyl oleate

O H3C O

methyl linoleate

Figure 2. The structure of methyl linoleate and methyl oleate, the FAMEs most prevalent in the most common biodiesel stock for the United States and Europe, respectively. Despite all the benefits of biodiesel listed above, there are several detriments. For example, biodiesel combustion is observed to produce higher proportions of NOx compared to fossil fuel.5, 9, 12-18

Additionally, the typical pour point (the point where fuel becomes a gel) is -9 to 15 °C,

which hinders its use in cold climates.1, 12, 15, 26 The amphiphilic nature of biodiesel leads to its action as a surfactant that can dislodge engine deposits resulting in possible damage. Retrofitting of engines to avoid this can be cost prohibitive, which may be one of the hindrances to biodiesel availability in our current infrastructure.3, 9, 13 Many of these detriments to the use of biodiesel might be alleviated by thermal cracking of the FAMEs to lower molecular-weight products, including transportation fuels, plastic precursors, natural gas, kerosene, and possibly gasoline.5, 6, 9, 14, 16, 17, 19, 20

ACS Paragon Plus Environment

5

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Thermal cracking, also known as pyrolysis,

5, 11, 22, 25, 27-29

Page 6 of 33

uses thermal energy to break down

larger molecules into smaller hydrocarbons.16, 17, 19-22, 29-32 This process is already utilized in the petroleum refining industry. It is dominated by radical mechanisms as thermal energy propagates through the molecular system causing homolytic bond cleavages.14, 16, 20-22, 25, 29, 31-33 This will produce a wide array of low- to medium-weight hydrocarbon products that resemble those derived from processing of fossil oil, some of which resemble the constituents of gasoline (hydrocarbons containing 4 – 12 carbons).3 Experimental pyrolysis of soybean and canola seed-derived triglycerides have been reported.34 During this procedure, the triglycerides were thermally cracked at 430–440 °C to produce a range of hydrocarbon products. The procedure produced 4–7 wt % of gas-phase products. 15–20 wt % was produced as a tar-like solid product. The polar phase formed approximately 5 wt % of the mixture. The remaining 68-76 wt % of the pyrolysis products were organic liquid products (OLP). GC analysis was performed on the OLP layer to determine the products formed during the thermal cracking process. Of these OLPs, approximately 20 mol % are alicyclic, aromatic, or polycyclic aromatic hydrocarbons (PAH). In the OLP layer, are compounds associated with low to middle-weight transportation fuels. Along with these transportation compounds, there are commodity chemicals present that can be used as plastic precursors and other organic reagents. Chart 1 demonstrates the variety of compounds in the OLP layer.

ACS Paragon Plus Environment

6

Page 7 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

H

O

H C O C R O H C O C R' O H C O C R'' H triglyceride (fat or oil) ∆T ∆P

Alkenes Alkanes Linear Branched Cyclopentanes Cyclohexanes

Linear Branched Linear (non-terminal) Branched (non-terminal) Cyclopentenes Cyclohexenes

Polyaromatics Aromatics Indanes and indenes Naphthalenes and fluorenes

BTEX Alkylbenzenes

Figure 3. A wide variety of compounds are found in the OLP Layer when soybean and canola oil are pyrolyzed at 430 – 440 °C.34 This study aims to determine an accurate yet cost-effective model chemistry to describe the thermochemical properties that govern the thermal cracking process for the most prevalent FAMEs in the most common biodiesel stock for the United States and Europe (methyl linoleate and methyl oleate, respectively). Computational cost is of the utmost importance here as in subsequent studies the attention will be turned to chemical dynamics simulations. This will lead to further understanding of the thermal cracking procedure on an atomic level, which may aid in future engineering of viable fuels.

ACS Paragon Plus Environment

7

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 33

METHODS Herein, the thermochemical properties of bond dissociation pathways for methyl linoleate and methyl oleate were computed with an eye towards direct chemical dynamics simulations. Dynamics simulations involve propagation of Newton’s equation of motion:  = . Although

parameterized dynamics simulations of FAMEs have been reported utilizing the ReaxFF method. The ReaxFF method is a purely classical trajectory simulation that includes a parameterized threshold for bond distance and bond order to predict homolytic bond scissions for hydrocarbon systems.22,

25, 32, 33, 35

The data used to parameterize ReaxFF was generated using B3LYP/6-

311G** (vide infra).32, 35 We aim to carry out direct dynamics simulations, in which the forces used to numerically integrate Newton’s equation of motion are derived from linked electronic structure calculations carried out “on-the-fly.” It is important to note that such “on-the-fly” calculation of forces has important implications for the utility of the electronic structure model chemistry, namely, that post hoc correction of the potential energy surface is not possible; this includes scaling of zero-point energy correction (ZPEC) as well as enthalpy correction. For this reason, herein, reaction energy values are computed as the change in electronic energy with unscaled zero-point energy added (∆ in kcal/mol). This should be viewed as a pragmatic assessment of candidate model chemistry,

where experimentally derived ∆  is to be reproduced with computational ∆ . ZPECs are

specific to the model chemistry, i.e., they depend on both the operator and the basis set chosen. There are many possible combinations of operator and basis set that do not have their ZPEC

determined (including most of those discussed herein: vide infra), and in such cases it is common to incorporate the ZPEC uncorrected (as is done herein). Scaling factors tend to modify ZPECs

by less than 5%36 and, since herein ∆ are computed, the error introduced applies to the

ACS Paragon Plus Environment

8

Page 9 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

difference in zero-point energy between reactant(s) and product(s) (i.e., there is substantial cancellation of error). The impact (error) from use of uncorrected ZPEC in computing ∆

herein is minimized insofar as a model chemistry is sought that produces a ∆ most similar to the experimentally derived ∆  (i.e. fortuitous cancelation of error).

Direct dynamics require a large number of “on-the-fly” electronic structure calculations per

trajectory (as well as an ensemble of trajectories), therefore, economy of the model chemistry is a strong consideration along with accuracy. The B97-D, B3LYP, and M06-2X density functionals were tested; additionally, the wavefunction theory method MP2 was included. Basis sets were also tested; 6-31G(d), 6-31+G(d,p), 6-311++G(2d,p), cc-pVDZ, and aug-cc-pVDZ are discussed herein. All calculations were completed with the Gaussian09 suite.37 Both geometry optimization and frequency analyses were performed in the gas phase. Ideal geometries would represent the global conformation minimum. Geometry optimization algorithms locate the nearest local minimum on the potential energy surface which may be, but is not necessarily, the global conformational minimum. Herein, chemical intuition was utilized to generate conformers believed to be reasonable candidates of global conformational minima. To otherwise locate the global conformational minimum one must either conduct: 1) exhaustive conformational searches, which entail systematic enumeration of every conformational isomer followed by optimization and frequency analysis with each model chemistry to be evaluated; or 2) dynamics simulations with sufficient energy in the system for trajectories to sample conformational space (which we are currently carrying out for methyl oleate and methyl linoleate to be discussed in our future report on the cracking of said FAMEs). Frequency analysis allows for identification of the stationary points as minima or transition state structures, and provides zero-point energy correction (vide supra). Ball and stick structures were produced using CYLview.38

ACS Paragon Plus Environment

9

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 33

The B97-D density functional is a hybrid meta-generalized gradient approximation (HGGA) constructed from Becke’s 1997 exchange and correlation functionals with an additional semiempirical dispersion correction factor of Grimme (G2).39, 40 B3LYP is a hyper-GGA DFT hybrid functional. B3LYP was parameterized against the G2 database where it achieved an absolute error of 2.0 kcal/mol for describing atomization energies.41-43 Currently, most reports computationally predicting bond dissociation energies (BDEs) in an attempt to describe the thermochemical properties of FAMEs make use of the B3LYP density functional.16, 17, 20, 22, 25, 29, 32, 35, 44-46 Mounting evidence is bringing light to the limitations of B3LYP, which include deficiencies in describing 1,3-dialkyl interactions,47-51 πbonded conjugated systems, van der Waals forces, and large systems.42,

43, 52

In order for a

functional to accurately model methyl linoleate, it must be able to describe these factors but, given that we aim to conduct direct dynamics simulations of FAME pyrolysis, it must do so while maintaining a relatively low computational cost. M06-2X is a hyper-GGA DFT hybrid functional commonly employed in describing organic molecular systems.11, 14, 27, 28, 42, 43, 53-55 Truhlar and Zhao parameterized M06-2X specifically to describe the thermochemistry, kinetics, and noncovalent interactions of main-block elements.11, 14, 53-55

Because of the ability to describe π-π interactions, hydrogen bonding, and other long-

range forces, M06-2X has been recently used to describe bond dissociations in methyl linolenate (as compared to methyl linoleate and methyl oleate described herein).14 The wavefunction theory based Møller-Plesset second-order perturbation theory method (MP2; MBPT2) was included in testing due to its relative success in describing longer-range effects that may be important in molecules as large as those discussed herein.42, 43, 56 This method has been used in thermochemical calculations of main-block elements due to its accuracy in describing

ACS Paragon Plus Environment

10

Page 11 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

bond lengths, electron correlation, and geometry optimizations of minima.29,

42, 43, 56

An

additional advantage to using MP2 is that it is not size dependent (while the DFT methods herein are).29, 42, 43, 56 There has also been literature published describing the thermal cracking process using the MP2/6-31G(d) method.29 The CBS-QB3 composite method is also used herein.57 The CBS-QB3 composite method is relatively inexpensive (compared to other composite methods) and has been utilized to accurately reproduce radical thermochemistry and bond dissociation, albeit with some error due to correction of spin contamination.58-64 The Gaussian-n (e.g., G4)65 and Weizmann-n (e.g., W4)66 composite methods are found to be more accurate and precise at reproducing thermochemical tables than the CBS-QB3 approach,67,

68

however, attempts to optimize

structures herein (that contain upwards of 21 non-hydrogen atoms) were found to be computationally intractable. RESULTS AND DISCUSSION Homolytic Bond Cleavage of Chemical Moiety Representative Compounds. The primary mechanism of thermal cracking is dominated by homolytic bond scission. Therefore, the most effective method to describe the thermal cracking of FAMEs must be able to describe bond dissociation accurately. To evaluate the best method for describing bond dissociations, a list of reactions was compiled that have experimental BDE data.67, 69-76 These reactions (1 – 44, Chart 1) were chosen for structural resemblance to FAMEs methyl linoleate and methyl oleate (chemical moiety representatives; CMRs). The CMR reactions can be separated into different classes describing the bond breaking: Reactions 1 – 24 are carbon-hydrogen bond scission; Reactions 25 – 35 are carbon-carbon bond scissions; Reactions 36 – 42 are mostly carbonoxygen bond scissions where Reaction 39 (formally a C–C bond scission) was included due to a

ACS Paragon Plus Environment

11

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 33

direct electronic effect of the oxygen atoms; and Reactions 43 – 44 are double bond scissions. The products of Reactions 43 and 44 could exist in different spin states. For both of Reactions 43 and 44 the lowest energy spin state was used in the BDE calculation (for Reaction 43 this is triplet, for Reaction 44 CO is a singlet while O is a triplet).

ACS Paragon Plus Environment

12

Page 13 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Scheme 1. Chemical Moiety Representative (CMR) Reactions with Structural Resemblance to Fatty Acid Methyl Esters (FAMEs) Methyl Linoleate and Methyl Oleate that have Experimental Bond Dissociation Energies (BDEs) Reported.67, 69-76 C-H Bond Dissociation

C-C Bond Dissociation

CH4

CH3

CH3

(25)

CH3

(26)

H

(1)

H

(2)

H

(3)

(27)

H

(4)

(28)

H

(5)

(29)

H

(6)

H

(7)

H

(8)

H

(9)

H

(10)

H

(11)

(34)

H

(12)

(35)

H

(13)

H

(14)

H

(15)

H

(16)

H

(17)

H

(18)

H

(19)

H

(20)

H

(21)

CH3

(30) CH3

(31) (32) (33)

Other Bond Dissociation

O

O

O

O O O

O H O

(23)

O

CH3

(39)

CH3

(40)

O

O

O O (41) O

O CH3O

O

(38)

O

O O

CH3

O

O (22)

(36)

(37)

CH3O

O

H

CH3

CH3O

O

(42) O

O H

O

(24)

O

C=X Bond Dissociation

O

C

O

CH2

CH2

(43)

CO

O

(44)

ACS Paragon Plus Environment

13

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 33

The BDEs were calculated for the CMR reactions 1 – 44 (Chart 1) and compared to experimentally determined values by computing their deviation:

   = ∆ , − ∆,

the mean signed deviation (MSD):



     !    = " #

and the population standard deviation ($%&' ):





(()* +!,! !  $%&'  = 2." 

 −  /# 

The MSD and $%&' for the collection of CMR reactions 1 – 44 (Chart 1) appear in Table 1,

reported as MSD ± $%&' .68 This approach was chosen over determination of uncertainty67, 77 due

to its clear and objective protocol. Ideally, a model chemistry could be found with zero MSD and $%&' . Pragmatically, the model chemistry tested herein that primarily minimizes MSD and

secondarily $%&' for CMRs 1 – 44 will be used to determine BDEs for methyl linoleate and

methyl oleate.

Table 1. Mean Signed Deviations (MSDs) and Population Standard Deviations ($%&' ) in Calculated Homolytic Bond Dissociation Energies (BDEs) for Chemical Moiety Representative (CMR) Reactions 1 – 44 (MSD ± $%&' in kcal/mol).a

B3LYP M06-2X B97D MP2 6-31G(d) 3.7 ± 5.6 -1.1 ± 9.3 3.8 ± 5.9 0.9 ± 14.2b 6-31+G(d,p) 5.7 ± 11.4 0.5 ± 9.4 5.5 ± 11.4 -1 ± 12.3 6-311G++(2d,p) 6.5 ± 10.9 0.7 ± 9.5 6.4 ± 10.4 -1.7 ± 12.2 cc-pVDZ 5.9 ± 9.2 0.7 ± 9.5 5.6 ± 9.1 2.4 ± 40.9b aug-cc-pVDZ 6.7 ± 10.1 1.5 ± 9.5 6.4 ± 9.7 -1.7 ± 13 a b CBS-QB3 = 0.2 ± 9.1 kcal/mol. Reaction 14 required inclusion of an ultrafine integration grid to obtain converged geometries; given that a change to the integration grid changes the accuracy

ACS Paragon Plus Environment

14

Page 15 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

of the computed value, all structures required to describe Reaction 14 were recomputed using this method. The results in Table 1 are striking: the B3LYP and B97D density functionals struggle to approach MSD less than 5 kcal/mol. The wavefunction based MP2 method has MSDs less than 2 kcal/mol, however $%&' is in the 12 – 14 kcal/mol range and MP2 is known to be much more

computationally costly than the M06 family of DFT functionals.78 The M06-2X method shows