Thermodynamic assessment of carbazole-based organic polycyclic compounds for hydrogen storage applications via a computational approach

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 8 ) 1 e1 0

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.elsevier.com/locate/he

Thermodynamic assessment of carbazole-based organic polycyclic compounds for hydrogen storage applications via a computational approach Byeong Soo Shin a, Chang Won Yoon b, Sang Kyu Kwak c,**, Jeong Won Kang a,* a

Department of Chemical and Biological Engineering, Korea University, 145, Anam-ro, Seongbuk-gu, Seoul, 136-701, Republic of Korea b Fuel Cell Research Center, Korea Institute of Science and Technology, 5, Hwarang-ro 14-gil, Seongbuk-gu, Seoul, 02792, Republic of Korea c School of Energy and Chemical Engineering, Ulsan National Institute of Science and Technology, Ulsan, 44919, Republic of Korea

article info

abstract

Article history:

Liquid organic hydrogen carriers (LOHCs) are promising candidates for storage and

Received 15 December 2017

transport of renewable energy due to their reversible reaction characteristics. For the

Received in revised form

proper assessment of candidate molecules, various thermochemical properties are

17 April 2018

required, and significant experimental efforts are necessary. In this work, we suggest a

Accepted 23 April 2018

systematic method for the estimation of thermochemical properties for LOHC candidate

Available online xxx

molecules combining Density Functional Theory (DFT) calculations, Conductor-like Screening Model (COSMO) and Molecular Dynamics (MD) simulations. We applied the

Keywords:

suggested method for the assessment of previously reported LOHC materials. Based on the

Hydrogen storage

analysis, new candidates of carbazole-derivative compounds (N-acetylcarbazole, N-phe-

Liquid organic hydrogen carriers

nylcarbazole, N-benzoylcarbazole, and 4-methyl-4H-benzocarbazole) are suggested, and

Reaction enthalpy

their properties are estimated and reviewed. Calculation results show that these candi-

Thermodynamic assessment

dates can provide high theoretical hydrogen uptake capacities above 6 wt% and optimal heats of dehydrogenation in the liquid phase. Analysis on the stereoisomerism showed that the structure-selectivity toward less stable stereoisomers of the hydrogen-rich form is preferable for the dehydrogenation process. © 2018 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.

Introduction Hydrogen storage is one of the essential elements of technologies for utilizing renewable energy systems. Recently, liquid organic hydrogen carriers (LOHCs) have received

increasing attention because they can store a significant amount of hydrogen via a catalytic chemical reaction under ambient conditions compared with high-pressure gas compression or low-temperature cryogenic hydrogen storage [1]. Moreover, the use of LOHCs is economically beneficial because LOHCs have similar thermo-physical properties (e.g.,

* Corresponding author. ** Corresponding author. E-mail addresses: [email protected] (S.K. Kwak), [email protected] (J.W. Kang). https://doi.org/10.1016/j.ijhydene.2018.04.182 0360-3199/© 2018 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved. Please cite this article in press as: Shin BS, et al., Thermodynamic assessment of carbazole-based organic polycyclic compounds for hydrogen storage applications via a computational approach, International Journal of Hydrogen Energy (2018), https://doi.org/10.1016/ j.ijhydene.2018.04.182

2

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 8 ) 1 e1 0

density, viscosity, heat capacity) to diesel oil, and thus current infrastructures for transportation can be used [2]. Thermochemical properties of LOHCs such as reaction enthalpy and equilibrium constants are essential for the initial design of hydrogen storage systems because they determine the maximum possible efficiency of the system. The thermochemical properties targets suggested by Mu¨ller et al. [3] are as follows: (1) the optimal range of enthalpy of dehydrogenation is from 39.78 to 58.60 kJ/mol H2 and (2) negative Gibbs free energies for hydrogenation and dehydrogenation are preferable at the reaction conditions. The enthalpies of dehydrogenation for liquid organic hydrides (i.e., hydrocarbons) such as methylcyclohexane are above 67 kJ/mol H2, typically requiring a high dehydrogenation reaction temperature around 300
C [4]. One strategy to overcome the problem of the high reaction temperature is to incorporate heteroatoms (i.e., N, S, O, B) into the molecular structure. N-ethylcarbazole is one of the well-known compounds for a LOHC, and a reversible reaction is possible at a low temperature ranging from 120 to 180
C [5,6]. The presence of an N-heteroatom in the 5-membered ring decreases the enthalpies of dehydrogenation compared with hydrocarbon materials with similar structures like dodeca-hydrofluorene [7]. Also, ammonia borane (AB), a compound containing boron (B) and nitrogen (N), has been considered as a potential hydrogen storage candidate due to its high hydrogen storage capacity up to 19.6 wt % [8]. Campbell et al. [9] proposed BN heterocyclic compounds (N-R-1,2-dihydro-1,2-azaborane, R ¼ H, Me, t-Bu) and demonstrated that the spent fuel of BN heterocyclics could be converted to fully hydrogenated materials under ambient conditions. Recently, Li et al. [10] proposed 2-methylindole as a potential substance for chemical hydrogen storage because of its low melting point and fast kinetic reaction rate compared with N-ethylcarbazole. Forberg et al. [11] developed a lignin-based energy storage system. They synthesized octahydrophenazine from a lignin hydrogenolysis product and ammonia and demonstrated that phenazine is a promising candidate suitable for reversible hydro/dehydrogenation reactions with high storage capacity. Assessment of LOHCs requires comprehensive reviews of thermodynamic properties. Arlt and coworkers [3,12] reported surveys on thermodynamic properties LOHCs including Nethylcarbazole and amine borane-based compounds. They also published studies on carbazole and N-alkylcarbazoles, which are helpful for understanding the influence of alkylchain length on the thermodynamic characteristics [13,14]. Such review of thermodynamic properties requires a significant amount of experimental efforts. Development of estimation technology using thermodynamic correlations and molecular-modeling enables us to reduce efforts in the screening of candidate LOHCs among millions of components. In this paper, we propose a systematic computation-based estimation for thermodynamic properties of LOHCs and assessment scheme by comparing the hydrogen storage capacity, melting point, enthalpies of dehydrogenation, and Gibbs free energy of dehydrogenation in the gas phase. In the proposed method, molecular configurations were constructed by Density Functional Theory (DFT) methods, and enthalpies of dehydrogenation in the gas phase were computed. Heat effects associated with phase change

were estimated by the COSMO-RS method or thermodynamic correlations. Finally, molecular dynamics (MD) simulations predicted densities of components, which are closely related to the volumetric storage capacities. Based on the analysis of thermochemical properties of seven components previously reported (Table 1), we further suggested additional carbazolebased candidate molecules (acetylcarbazole, phenylcarbazole, benzoylcarbazole, and 4-methyl-4H-benzocarbazole) and properties were evaluated using the same manner. The proposed method can be used for assessment of any candidate molecule at the early stage of development.

Computational details Density functional theory calculation Atomic arrangement and electronic structures of candidate LOHC molecules were calculated using DFT method implemented in the DMol3 program [15]. Three types of exchangecorrelation functionals (GGA-PBE [16], mGGA-TPSS [17], Hybrid GGA-B3LYP [18]) were used to compare the feasibility of calculating the enthalpies of dehydrogenation in the gas phase. The atomic orbitals were composed of double and triple numerical basis sets plus polarization functions (DNP, TNP). All core electrons were explicitly included for relativistic effects. The Grimme dispersion correction model [19] was used in all cases to describe long-range electron correlations. The SCF convergence criterion was within 1.0  106 Hartree per atom. In many cases, a component can be locally stable for several conformers. We used the following procedure to obtain the structure of conformers. Initial structures of conformers were generated by the Boltzmann jump method where molecular torsion angles were randomly changed, and then the generated structures were optimized with the COMPASS force field [20]. Five conformers which have lower energies were selected, and they were further optimized by density functional theory (DFT) calculations. For carbazole-derivatives, the hydrogen-rich form of the hydrogen carriers may have different configurations of hydrogen in the 5-membered ring depending on the surface condition of the catalyst. Morawa Eblagon et al. [21] reported three stereoisomers when N-ethylcarbazole was fully hydrogenated, as shown in Fig. 1 (Stereoisomers type A, type B, and type C). For simple comparison of properties, as shown in Table 1, stereoisomer A was only used for calculations. For reaction equilibrium calculations of newly suggested carbazole derivatives (perhydro-N-ethylcarbazole, -acetylcarbazole, -phenylcarbazole, -benzoylcarbazole, and 4-methyl-4Hbenzocarbazole), all three possible configurations were used in the calculations. Vibrational frequencies were calculated to determine the thermochemical properties using the harmonic oscillation approximation. Most quantum chemical calculation methods (including DFT and Hartree-Fock) overestimate the zero-point vibrational energy (ZPVE) because the anharmonicity effect is neglected [22]. One of the strategies to adjust vibrational frequencies is to introduce a scaling factor (0.89e1.06) into the vibration energy [23]. Many ZPVE scale factors have been reported and evaluated for various functionals such as BLPY,

Please cite this article in press as: Shin BS, et al., Thermodynamic assessment of carbazole-based organic polycyclic compounds for hydrogen storage applications via a computational approach, International Journal of Hydrogen Energy (2018), https://doi.org/10.1016/ j.ijhydene.2018.04.182

3

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 8 ) 1 e1 0

Table 1 e Comparison of thermophysical and thermochemical properties of the reported LOHCs. No.

Substance

1

Ammonia Borane (AB)

2

1,2-dihydro-1,2-azaboraine

3

4

3-methyl-1, 2-BN-cyclopentane

N-t-Bu-1,2-dihydro-1, 2-azaboraine

Dr HD ðgÞ at 298.15 K (kJ/mol H2)

Dr GD ðgÞ at 298.15 K (kJ/mol)

47

13.90c 21.35 [9] 55.25c

49.90c 56.93 [9] 438.23c

4.7 [46]

42 [46]

50.52c

416.00c

96e98a [9]

4.3

25

41.57c 35.16 [9]

7.70c 1.67 [9]

Capacity

Hydrogen rich form

Hydrogen lean form

Melting point (
C)

H3 N  ВH3

H2 N  BH2

112

6.5

e

62e63a [37]

4.7

18a [46]

NH2 BH2

NH2 BH2

g H2/L

N B B N N B

N B B N N B

t-Bu

H

wt%

t-Bu

N BH2

N BH

5

H-carbazole

H N

H N

246.3b [47]

6.7 [44]

e

48.12c

58.48c

6

N-methylcarbazole

CH3 N

CH3 N

89b [48]

6.2 [44]

e

49.10c

59.73c

7

N-ethylcarbazole

70b [49]

5.7 [44]

54 [44]

47.80c 50.70 [13]

53.96c

H3C N

a b c

H3C N

The melting point of Hydrogenated form. The melting point of Dehydrogenated form. DFT calculation results using TPSS/DNP level of theory with dispersion correction, in this work.

B3LYP, and MP2 coupled with Pople-style basis sets [24,25]. Since the determination of appropriate scaling factors for the methods is beyond the scope of this paper, and the anharmonicity effect is assumed to be marginal due to relatively high symmetries of targeted molecules, the calculated ZPVE with scaling factor 1 was not further corrected.

Prediction of enthalpy of dehydrogenation in the gas phase

Fig. 1 e Three types of stereoisomer for carbazolederivative compound. R represents functional group. For stereoisomer type A, all four H atoms in the 5-membered ring face down (symmetric structure). The stereoisomer type B and type C have a different configuration of C-H bond (red-dotted ellipse) in the 5-membered ring (asymmetric structure). Yellow lines indicate the additional bonds in the 4-methyl-4H-benzocarbazole. (For interpretation of the references to color/colour in this figure legend, the reader is referred to the Web version of this article.)

The difference in the heat of formation between reactant and product is a simple and reasonable method to estimate the enthalpies of reaction. However, due to the variation of results depending on the choice of isodesmic reactions, this method is not reliable for our purposes. Verevkin et al. [26] suggested a modified atomization procedure using high-level first principles (e.g., G3MP2, Gn series) as an alternative method to calculate the heat of formation. This method also tends to underestimate the heat of formation applied to heterocyclic compounds compared with experimental data. There is a simple way to calculate the enthalpies of dehydrogenation in the gas phase, Dr HD ðgÞ, using the DFT method. Instead of using the calculated heat of formation, the

Please cite this article in press as: Shin BS, et al., Thermodynamic assessment of carbazole-based organic polycyclic compounds for hydrogen storage applications via a computational approach, International Journal of Hydrogen Energy (2018), https://doi.org/10.1016/ j.ijhydene.2018.04.182

4

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 8 ) 1 e1 0

enthalpies of the reaction can be obtained by combining the change in total electronic energies E0 at T ¼ 0 K and thermal corrected enthalpies Hcorr as shown in Equation (1),

Enthalpy of dehydrogenation, equilibrium constant and conversion in the liquid phase

Dr HD ðgÞ ¼ DE0 þ DZPVE þ Hvib ðSÞ þ n½Hvib ðH2 Þ þ Hrot ðH2 Þ þ Htrans ðH2 Þ þ RT  Hvib ðS þ nH2 Þ

enthalpies of vaporization were calculated using the ClausiusClapeyron equation.

(1)

where, the subscripts of vib, rot, and trans stand for vibrational, rotational, and translational contributions, respectively, R is the ideal gas constant, S is the hydrogen-lean form, and SþnH2 is the hydrogen-rich form of LOHC. The detailed formula can be found in detail elsewhere [27].

Fig. 2 describes the calculation procedure for the enthalpy of dehydrogenation in the liquid phase. The enthalpies of dehydrogenation in the liquid phase were calculated using the following thermodynamic equation,  3 g Dr HD ðliqÞ ¼ Dr HD g; DMol þ Dl HH2 rich ðCOSMO  RSÞ g

 Dl HH2 lean ðCOSMO  RSÞ

Prediction of vapor pressure and enthalpy of vaporization for LOHCs Enthalpies of dehydrogenation calculated in Section Prediction of enthalpy of dehydrogenation in the gas phase should be corrected when the reaction occurs in a liquid phase. When compared with heats of reaction, the effect of latent heats is normally small. When the boiling point and/or critical constants are available, the vapor pressure and the enthalpies of vaporization can be estimated by theoretical or empirical correlations. NIST TDE (Thermo Data Engine) provides a systematic estimation method [28] depending on the level of experimental information available. Without such information, one should rely on a group-contribution method or a method based on computational molecular modeling method. In many cases, group parameters for the fragments in LOHCs are not available, and we have to rely on a method based on the molecular modeling techniques. The Conductor-like Screening Model for Real Solvents (COSMO-RS) method [29,30] allow us to predict vapor pressures of LOHCs by combining DFT calculations in solvation effects (DFT/COSMO) with statistical thermodynamic theory. In this work, optimized structures of LOHCs using the method described in Section Density functional theory calculation were used as input files in the DFT/COSMO calculation. A polarization charge density (s) immersed in a perfect conductor of infinite permittivity (ε ¼ ∞) was calculated by COSMO algorithm implemented in DMol3. To obtain reliable thermodynamic properties (enthalpies of vaporization in this work) from COSMO-RS, a consistent functional/basis set combination should be used in DFT/COSMO and COSMO-RS calculations. For example, when the sigma profile is generated at the PBE/DNP level in the DFT/COSMO calculation, COSMOtherm parameterizations that have been derived from the PBE/DNP level should be employed in the COSMO-RS calculation. In COSMO-RS, the chemical potential was calculated from differences in the interaction energy between new contacts of the molecular surfaces by adding a molecular surface with polarity s into the real solvent (condensed phase) [31,32]. The vapor pressure was obtained from Equation (2). " X # mX  mXig

PXvap ðTÞ ¼ exp

RT

(3)

3

(2)

where mXig and mXX are the chemical potentials of the compound X in the ideal gas phase and pure bulk conditions, respectively. The detailed expressions are given in Ref. [33]. The

where, Dr HD (g, DMol ) is the enthalpies of dehydrogenation in the gas phase calculated from DMol3 program, g g Dl HH2 rich (COSMO-RS) and Dl HH2 lean (COSMO-RS) are the enthalpies of vaporization for hydrogen-rich and hydrogen-lean compounds, respectively. The equilibrium constant, K, is calculated as follows:


d lnK DH ¼ dT RT2

(4)




where DH is the standard enthalpy change of a reaction. Because of the temperature dependency of enthalpy of reaction, the equilibrium constant was calculated according to the following equations, 2 1 KðTÞ ¼ K373K $exp4 $ R

ZT

3 DHðTÞ 5 dT T2

(5)

373

The standard pressure and temperature were set to 1 bar and 373 K, respectively. Because the difference in the changes of the standard Gibbs free energy between liquid and gas phase is small, the following assumption can be used:


K373K ¼ exp

DGliquid at¼373K R,373

! zexp




DGgas at¼373K R,373

! (6)

Thus, the equilibrium conversion (Xe) was obtained using Equation (7), Xe ¼

KðTÞ 1 þ KðTÞ

(7)

Fig. 2 e The calculation procedure for enthalpies of dehydrogenation in the liquid phase, Dr HD ðliqÞ. Dr HD ðg; DFTÞ denote the enthalpies of dehydrogenation in the gas phase calculated by DFT method. Dvap H(COSMO-RS) is the enthalpies of vaporization predicted by COSMO-RS method.

Please cite this article in press as: Shin BS, et al., Thermodynamic assessment of carbazole-based organic polycyclic compounds for hydrogen storage applications via a computational approach, International Journal of Hydrogen Energy (2018), https://doi.org/10.1016/ j.ijhydene.2018.04.182

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 8 ) 1 e1 0

Prediction of density for hydrogen-rich form of LOHCs by molecular dynamics Densities of candidate LOHCs are essential for the estimation of volumetric storage capacities. Densities were estimated using a combined constant temperature (NVT) - constant pressure (NPT) molecular dynamics (MD) simulation method. Initial molecular configuration and electron static partition charges (ESP) are from the result of DFT calculation described in Section Density functional theory calculation. MD simulations were performed using the COMPASS force field [20] with the fixed charges calculated from DFT. Periodic bulk systems were constructed with 200 molecules and the (NVT) MD simulation was performed for 100 ps followed by the (NPT) MD for 300 ps at 1 bar. Berendsen barostat and Andersen thermostat were used for stabilizing pressure and temperature, respectively.

Results and discussion Enthalpy of dehydrogenation in the gas phase Prediction results of enthalpy of dehydrogenation may vary depending on the choice of the functional (PBE, TPSS, and B3LYP) and the basis set (DNP, TNP). By comparing the estimation with available data, we try to choose the best functional and basis set which are appropriate for LOHCs assessment, as shown in Table S1. We found that PBE and B3LYP functionals overestimated the enthalpies of dehydrogenation. The TPSS functional shows fair agreements with experimental data. Normally, choice of functional has more influence than that of the basis sets. The exchange-correlation functional is strongly related to the enthalpy of reaction because it describes the kinetic energy and Colombic interactions of electrons. TPSS combined with DNP or TNP was used for estimation of enthalpies of dehydrogenation in the gas phase for components without experimental data.

Vapor pressure and enthalpy of vaporization ThermoData Engine (TDE) software developed by NIST/Thermodynamics Research Center [28] provides a systematic method to estimate thermodynamic properties based on the property relationships using available experimental information. The prediction methods implemented in TDE start from the estimation of the normal boiling temperature (NBP) and enthalpy of vaporization at NBP using a groupcontribution method. The predicted boiling temperature is validated with critically evaluated experimental data and used in prediction of the critical properties (i.e., critical pressure and critical temperature) and acentric factor. Using the information, a corresponding states theory (e.g., AmroseWalton method [34]) predicts the vapor pressure, followed by verification procedure with experimental data. The vapor pressure and enthalpies of vaporization for indoline and N-ethylcarbazole were predicted using the TDE algorithm and compared with experimental data and COSMORS results, as shown in Figure S13eS16. TDE shows the best performance in predicting thermodynamic properties

5

because the predicted properties were updated by verification using experimental data. However, when the pure properties were predicted without the help of experimental data, the predicted values do not agree due to missing group parameters for LOHC components. Moreover, group contribution methods cannot identify property differences among stereoisomers described in Fig. 1. According to Lin et al. [35], the COSMO-solvation model shows AAD of 76% for vapor pressure estimation and RMSD of 4.81 kJ/mol for enthalpies of vaporization estimation. Considering the logarithmic nature of vapor pressure equation, relatively small effects on the reaction enthalpies, and equilibrium conversion, COSMO-based method would be enough for our purposes for heterocyclic molecules when group contribution parameters are not available. COSMO-RS calculations results with several different combinations of functionals were compared with experimental data for heterocyclic materials with experimental data (pyridine, piperidine, quinoline, indane, tetra hydroquinoline, 2,3-dihydro benzofuran, indoline, and N-ethylcarbazole) to select the best set. The calculation results of vapor pressure and enthalpies of vaporization were compared with experimental data as shown in Fig. 3 (detailed comparisons are given in Supplementary Materials; Figures S1eS16). Combination of PBE/DNP shows reasonable results for vapor pressures and enthalpies of vaporization within 56% and 10% of AAD (RMSD: ~8 kJ/mol), respectively. We already showed that PBE/DNP overestimates the reaction enthalpies in the gas phase as explained in Section Enthalpy of dehydrogenation in the gas phase. For the consistency in the calculation, TPSS functional with DNP basis set was employed in the DFT/COSMO calculation, and then COSMO-RS calculations were performed without adjusting the COSMOtherm parameterizations (DMol3_PBE_C30_1401.ctd file) derived from PBE/DNP level of theory. Although the combination of different models may lead to inconsistencies, the TPSS/DNP model shows comparable results to PBE/DNP; because TPSS is a series of GGA formulas similar to PBE, and the same types of numerical basis sets (i.e., DNP) are used. In the case of pyridine, piperidine, and quinoline, TPSS/TNP is much more reliable than the other methods. However, for the heterocyclic compounds containing 5-membered rings, the TPSS/TNP model shows relatively larger deviations in both vapor pressure and enthalpy of vaporization. Based on the comparison of calculation with experimental data, the structures of hydrogen-rich and hydrogen-lean forms of ethylcarbazole, acetylcarbazole, benzoylcarbazole, phenylcarbazole, and 4-methyl-4H-benzocarbazole were estimated using the TPSS/DNP method for both the gas phase (DFT) and the COSMO calculation.

Assessment of previously reported LOHCs LOHC components listed in Table 1 were compared in terms of thermodynamic properties such as hydrogen storage capacity, melting point, enthalpies of dehydrogenation, and Gibbs free energy of dehydrogenation in the gas phase. The dehydrogenation Gibbs free energy Dr GD ðgÞ of ammonia borane (component 1 in Table 1), and 1,2-dihydro-1,2-azaboraine (component 2 in Table 1) were small and the dehydrogenation reaction is thermodynamically favorable at room temperature. However, a

Please cite this article in press as: Shin BS, et al., Thermodynamic assessment of carbazole-based organic polycyclic compounds for hydrogen storage applications via a computational approach, International Journal of Hydrogen Energy (2018), https://doi.org/10.1016/ j.ijhydene.2018.04.182

6

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 8 ) 1 e1 0

a)

300

AAD(%) in Pvap

250

GGA-PBE / DNP mGGA-TPSS / DNP / D-corr (Grimme) mGGA-TPSS / TNP / D-corr (Grimme)

200 150 100 50 0

b)

25

AAD(%) in ΔvapH

20

e e e e dro zol ro line idin dan trahy ne dihyd an idin rba ino P yr r n a u e I e c i e l n Q T inol 2,3 ofur P ip ol i thy nz Ind n-e qu -be

GGA-PBE / DNP mGGA-TPSS / DNP / D-corr (Grimme) mGGA-TPSS / TNP / D-corr (Grimme)

15

10

5

0

le e ro azo e e idin dro arb ineIndan trahyd e din r y c l i l e e r o h y i p n n in Py Pi Te inolin 2,3-d fura ndoli n-eth Qu o I qu enz

Fig. 3 e Comparisons of experimental data with COSMO-RS calculation results. a) AAD (%) in vapor pressure, b) AAD (%) in enthalpies of vaporization. The different density functionals and basis sets were used in DFT/COSMO calculation.

solvent is required to prevent solidification of the hydrogen-rich form (the melting points of component 1 and 2 are 112
C and 63
C, respectively). To overcome the problems of solidification, Luo et al. [36] reported 3-methyl-1,2-BN-cyclopentane (component 3), which is in liquid form at room temperature with a 4.7 wt% hydrogen storage capacity. Due to the large negative value of the Gibbs free energy of dehydrogenation reaction (Dr GD ðgÞ ¼ e 416.00 kJ/mol), it is difficult to achieve reversible reactions and a strong reducing agent (LiAlH4) was required to regenerate the spent fuel [37]. The N-t-Bu-1,2-dihydro-1,2azaboraine (component 4) satisfies the requirements of a reversible reaction, Dr GD ðgÞ ~0 kJ/mol and the range of heat of dehydrogenation (39.78e58.60 kJ/mol H2). This component also suffers from the problem of the high melting point of the hydrogenated form. Moreover, decompositions of the hydrogenrich form (component 1e4) occur near 150
C [37]. At this temperature, volatile impurities such as borazine, ammonia, and

diborane can be generated as vapors so that it is possible to damage, for instance, the proton exchange membrane (PEM) fuel cell. Also, the storage capacity of boron- and nitrogencontaining chemical hydrides is less than 5.5 wt%, which is the US Department of Energy (DOE) target except for AB. Among the carbazole-based compounds, H-carbazole (component 5) shows largest hydrogen storage capacity (6.7 wt%) but it is unfavorable due to large Gibbs energy of dehydrogenation (58.48 kJ/mol) and high melting point (246
C). With additions of alkyl chains, melting points are lowered (Nmethylcarbazole, 89
C, N-ethylcarbazole 70
C) even though molecules becomes larger. In case of carbazole-derivatives, asymmetry in molecular arrangement plays a significant role in the solid-liquid phase transition. N-ethylcarbazole shows the most favorable Gibbs free energy of dehydrogenation among carbazole derivatives (53.96 kJ/mol) whereas the storage capacity compared with H-carbazole was slightly reduced by 15%

Please cite this article in press as: Shin BS, et al., Thermodynamic assessment of carbazole-based organic polycyclic compounds for hydrogen storage applications via a computational approach, International Journal of Hydrogen Energy (2018), https://doi.org/10.1016/ j.ijhydene.2018.04.182

7

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 8 ) 1 e1 0

(5.7 wt %). Addition of other component and forming eutectic mixture may lower the melting point of the dehydrogenated form of N-ethylcarbazole (70
C) [38]. Mehranfar and coworkers [39,40] reported DFT studies on step-wise dehydrogenation of perhydro-N-ethylcarbazole in the gas phase (1 bar, 298 K). They reported a negative value of Gibbs free energy (4.9 kcal/mol) of dehydrogenation between perhydro-N-ethylcarbazole and Nethylcarbazole using B3LYP/6-311þþG(d,p) method. This result does not agree with our estimation result of 53.96 kJ/mol (12.89 kcal/mol). Mehranfar et al. [39,40] focused on the ratedetermining step by finding the optimum pathway of dehydrogenation reaction. They optimized a reactant or intermediate with hydrogen molecule surrounding unsaturated carbonrings to search transition state. There are many degrees of freedom in these configurations, which may affect the calculation of Gibbs free energy of the reaction. Our result may be justified by comparison of enthalpy value reported by Stark et al. [13] (see Table 1, compound 7).

Assessment of newly suggested LOHCs Based on the assessment of the previous section, N-alkylcarbazole is a promising hydrogen storage candidate due to reversible reactions and high storage capacities. Three carbazole-derivatives (methyl-, ethyl-, and propyl-) were experimentally investigated [41e43] for thermodynamic and catalytic characteristics. As already mentioned in the previous section, the addition of alkyl chain is preferable for a reversible reaction, but it has a disadvantage in the storage capacity. For further investigation, we suggested carbazole-derivative compounds functionalized with different types of functional

groups (rather than alkyl chains) and analyzed their thermochemical properties using a molecular modeling technique. Nacetylcarbazole (S/C2), N-benzoylcarbazole (S/C3), N-phenylcarbazole(S/C4), and 4-methyl-4 H-benzocarbazole (S/C5) were considered and estimated properties are shown in Table 2. N-ethylcarbazole (S/C1) was also included for comparison. Note that S and C denote the spent and charged states of the carrier, respectively. The ideal gravimetric capacities of carbazole-derivatives are all above 6 wt%. Volumetric capacities calculated by MD simulations are listed in Table S2 of Supplementary Material. MD simulation results for perhydro-N-ethylcarbazole (54.34 g H2/L) was in good agreement with experimental data (54 g H2/L) [44]. The DOE gravimetric and volumetric capacity targets (by 2020) are 5.5 wt% and 40 g H2/L, respectively. These capacities are based on the system levels, and theoretical values should be slightly higher considering incomplete conversions. Calculated capacities of four suggested materials were 6e7 wt% and 60e70 g H2/L (material basis) and they will meet the DOE suggestions. There is a trade-off between the storage capacity and the melting point due to the structural characteristics. The flat structure of the aromatic rings in the hydrogen-lean form is responsible for the high melting point above 76
C, while the hydrogen-rich forms (C2 - 5) are tin a liquid state at ambient conditions. By mixing several LOHCs, the formation of a eutectic mixture may overcome the problem of the high melting point. Fig. 4 (Table S3eS7) shows a comparison of calculated enthalpies of dehydrogenation in the liquid state for an equilibrium constant K ¼ 1 using the procedure described in Section Enthalpy of dehydrogenation, equilibrium constant

Table 2 e Comparison of thermos-physical properties for the screened hydrogen storage candidates.

Entry

Spent carrier

Melting point (
C)

Capacity a

b

wt% /g H2/L S/C1

N-ethylcarbazole

S/C2 S/C3 S/C4 S/C5

N-acetylcarbazole N-benzoylcarbazole N-phenylcarbazole 4-methyl-4H-benzocarbazole

a b

5.70 [43]/54.0 [43] 5.83/54.34 6.32/63.83 6.92/68.92 6.94/67.11 6.87/67.39

70 [49] 76 [50] 98 [51] 91e93 [52] e

Gravimetric hydrogen storage capacity (carrier only). Calculated by molecular dynamics.

Please cite this article in press as: Shin BS, et al., Thermodynamic assessment of carbazole-based organic polycyclic compounds for hydrogen storage applications via a computational approach, International Journal of Hydrogen Energy (2018), https://doi.org/10.1016/ j.ijhydene.2018.04.182

8

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 8 ) 1 e1 0

and conversion in the liquid phase. Three different stereoisomers shown in Table 1 were considered in the calculations. The enthalpies of dehydrogenation of perhydro-N-acetylcarbazole (C2) are slightly higher than those of perhydro-Nethylcarbazole (C1). Due to the presence of oxygen, higher temperature is required to release hydrogen from the hydroxyl-group in hydrogen-rich form. For perhydro-N-benzoylcarbazole (C3) and ephenylcarbazol (C4), the highest enthalpy of dehydrogenation was estimated to be 58.60 kJ/mol H2, indicating that cyclohexyl-groups require high thermal energies to release hydrogen compared with nitrogencontaining heterocyclic rings. Although the different types of functional groups and structures increased the enthalpy of dehydrogenation compared to ethyl functional groups, all proposed candidates have reasonable enthalpies of reaction within the optimal range of dehydrogenation enthalpies (39.78e58.60 kJ/mol H2). It is interesting to note that stereoisomers type B or type C for each compound require lower heats to release hydrogen than stereoisomer type A. Heats for dehydrogenation have a close relation with steric energy levels of optimized stereoisomers. For perhydro-N-ethylcarbazole, the stereoisomer type B shows the lowest enthalpy of dehydrogenation (see Fig. 4) and the highest total steric energy in stereoisomers as listed in Table S8. Optimized steric energy levels varies depending on the dispersion correction. Before the correction, stereoisomer type C showed the lowest steric energy. When the dispersion correction was applied, the stereoisomer type A is more stable than type C and type B (i.e. type A < type C < type B). Sotoodeh et al. [45] reported that the stereoisomer type A and type C for perhydro-N-ethylcarbazole was more stable than other possible stereoisomers and the relative energy between them was within 3 kJ/mol. However, Morawa Eblagon et al. [21] reported that the asymmetric structure (Stereoisomer type B and type C) is more thermodynamically stable than symmetric form (Stereoisomer type A). It should be noted that calculated relative-energies between stereoisomers in our results are more close to the trend reported by Sotoodeh et al. [45]. When the fully hydrogenated product is in lower energy level (unstable stereoisomer form), the dehydrogenation reactions may readily occur at a lower temperature. These

results are in line with the experimental observation from Air Products [44]. They investigated the effect of stereoisomer distribution on perhydro-N-ethylcarbazole (hydrogen-rich form) catalytic dehydrogenation and demonstrated that the reactant consisting of less stable stereoisomers was dehydrogenated at a much lower temperature than the reactant consisting of stable stereoisomers.

Isomer A Isomer B Isomer C

S/C 5 S/C 4 S/C 3 S/C 2 S/C 1

Optimal range of dehydrogenation enthalpy

40

45

50

ΔrHD (kJ/mol H2)

55

60

Fig. 4 e Enthalpies of dehydrogenation for hydrogen storage candidates with different stereoisomers. Each enthalpy value is calculated at equilibrium constant, K ¼ 1.

Fig. 5 e Equilibrium conversion of dehydrogenation for hydrogen storage candidates with different stereoisomers, which are shown in Fig. 1. a) type A, b) type B, c) type C. The table inserted in figure indicates the temperature where the 99% of reaction conversion is achieved.

Please cite this article in press as: Shin BS, et al., Thermodynamic assessment of carbazole-based organic polycyclic compounds for hydrogen storage applications via a computational approach, International Journal of Hydrogen Energy (2018), https://doi.org/10.1016/ j.ijhydene.2018.04.182

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 8 ) 1 e1 0

9

financial resource from the Ministry of Trade, Industry & Energy, Republic of Korea (No.20153030041030). This research was also supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT & Future Planning (No. 2014R1A5A1009799). This work was also supported by the Technology Innovation Program (10045068, development of flow assurance and organic acid/calcium removal process for the production of offshore opportunity crude) funded by the Ministry of Trade, Industry and Energy (MI, Korea). Computational resources are from UNIST-HPC and KISTI-PLSI.

These tendencies were further investigated by calculating the dehydrogenation equilibrium conversion (Xe) as shown in Fig. 5. When comparing the temperature where the conversion is achieved up to 99%, stereoisomer type B for S/C1, S/C3, S/C4 and type C for S/C2, S/C5 were shown to be effective structures for the dehydrogenation in terms of thermodynamics. In the case of stereoisomer type C (see Fig. 5, c), the full dehydrogenation temperature of S/C2 is lower than that of S/C1. For S/C5 LOHC, a large variation was observed in equilibrium conversion depending on the type of stereoisomer structures. Therefore, the structure-selectivity toward less stable stereoisomers of the hydrogenated carriers (hydrogenrich form) is important for the dehydrogenation process when considering the thermal efficiency.

Appendix A. Supplementary data

Conclusion

Supplementary data related to this article can be found at https://doi.org/10.1016/j.ijhydene.2018.04.182.

Assessment of LOHC candidate molecules was performed in thermodynamic aspect using a combined computational approach employing DFT, MD, and COSMO-RS methods. To evaluate the DFT simulation results, different combinations of density functionals and basis sets were tested using experimental observations. The DFT method that gave the most consistent results was the TPSS functional with a DNP basis set in this study. Based on the calculation results, ammoniaborane-based compounds showed high thermodynamic driving forces for dehydrogenation reactions, and therefore they may face difficulty in regenerating fuel. Carbazolederivatives were promising LOHCs candidates considering their thermophysical properties and enthalpy of reaction in the gas phase. We suggested the modification of functional groups of carbazole to improve the thermophysical and thermochemical properties. N-acetylcarbazole, N-benzoylcarbazole, N-phenylcarbazole, and 4-methyl-4H-benzocarbazole were suggested, and high storage capacities above 6 wt% were estimated. Although the enthalpies of dehydrogenation of screened LOHCs were slightly higher than that of N-ethylcarbazole due to their specific functional type (i.e., hydroxyl and cyclohexyl groups), the estimated values were within the optimal range of reaction enthalpies. By mixing several LOHC components, the problem of the high melting point may be solved. We found that unstable stereoisomers were favorable for the dehydrogenation reaction at low temperature. When considering energy efficiency, the structure-selectivity toward stereoisomers type B or type C, which are in an unstable state, is an important factor in consideration of the dehydrogenation process. The framework used in this study can be used for assessment of other candidate molecules for LOHC purposes. It should be noted that calculations performed in this work only considered thermodynamic aspects and we did not take into account the catalytic reaction.

Acknowledgements This work was supported by the New & Renewable Energy Core Technology Program of the Korea Institute of Energy Technology Evaluation and Planning (KETEP), granted

references

[1] Yang J, Sudik A, Wolverton C, Siegel DJ. High capacity hydrogen storage materials: attributes for automotive applications and techniques for materials discovery. Chem Soc Rev 2010;39:656e75. [2] Teichmann D, Arlt W, Wasserscheid P, Freymann R. A future energy supply based on liquid organic hydrogen carriers (LOHC). Energy Environ Sci 2011;4:2767e73. € lkl J, Arlt W. Thermodynamic evaluation of [3] Mu¨ller K, Vo potential organic hydrogen carriers. Energy Technol 2013;1:20e4. [4] Zhang C, Liang X, Liu S. Hydrogen production by catalytic dehydrogenation of methylcyclohexane over Pt catalysts supported on pyrolytic waste tire char. Int J Hydrogen Energy 2011;36:8902e7. [5] Pez GP, Scott AR, Cooper AC, Cheng H, Bagzis LD, Appleby JB. International patent WO/2005/000457 in hydrogen storage reversible hydrogenated of Pi-Conjugated substrates. 2008. [6] Yang M, Han C, Ni G, Wu J, Cheng H. Temperature controlled three-stage catalytic dehydrogenation and cycle performance of perhydro-9-ethylcarbazole. Int J Hydrogen Energy 2012;37:12839e45. [7] Sotoodeh F, Huber BJ, Smith KJ. The effect of the N atom on the dehydrogenation of heterocycles used for hydrogen storage. Appl Catal A Gen 2012;419:67e72. [8] Marder TB. Will we soon be fueling our automobiles with ammoniaeborane? Angew Chem Int Ed 2007;46:8116e8. [9] Campbell PG, Zakharov LN, Grant DJ, Dixon DA, Liu S-Y. Hydrogen storage by Boron nitrogen heterocycles: a simple route for spent fuel regeneration. J Am Chem Soc 2010;132:3289e91. [10] Li L, Yang M, Dong Y, Mei P, Cheng H. Hydrogen storage and release from a new promising liquid organic hydrogen storage Carrier (LOHC): 2-methylindole. Int J Hydrogen Energy 2016;41:16129e34. [11] Forberg D, Schwob T, Zaheer M, Friedrich M, Miyajima N, Kempe R. Single-catalyst high-weight% hydrogen storage in an N-heterocycle synthesized from lignin hydrogenolysis products and ammonia. Nat Commun 2016;7:13201. [12] Mu¨ller K, Stark K, Mu¨ller B, Arlt W. Amine borane based hydrogen carriers: an evaluation. Energy Fuel 2012;26:3691e6. [13] Stark K, Emel’yanenko VN, Zhabina AA, Varfolomeev MA, Verevkin SP, Mu¨ller K, et al. Liquid organic hydrogen carriers:

Please cite this article in press as: Shin BS, et al., Thermodynamic assessment of carbazole-based organic polycyclic compounds for hydrogen storage applications via a computational approach, International Journal of Hydrogen Energy (2018), https://doi.org/10.1016/ j.ijhydene.2018.04.182

10

[14]

[15]

[16] [17]

[18]

[19]

[20]

[21]

[22] [23]

[24]

[25]

[26]

[27] [28]

[29]

[30]

[31] [32] [33]

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 8 ) 1 e1 0

thermophysical and thermochemical studies of carbazole partly and fully hydrogenated derivatives. Ind Eng Chem Res 2015;54:7953e66. Emel’yanenko VN, Varfolomeev MA, Verevkin SP, Stark K, Mu¨ller K, Mu¨ller M, et al. Hydrogen storage: thermochemical studies of N-alkylcarbazoles and their derivatives as a potential liquid organic hydrogen carriers. J Phys Chem C 2015;119:26381e9. Delley B. An all-electron numerical method for solving the local density functional for polyatomic molecules. J Chem Phys 1990;92:508e17. Perdew JP, Burke K, Ernzerhof M. Generalized gradient approximation made simple. Phys Rev Lett 1996;77:3865. Tao J, Perdew JP, Staroverov VN, Scuseria GE. Climbing the density functional ladder: nonempirical metaegeneralized gradient approximation designed for molecules and solids. Phys Rev Lett 2003;91:146401. Stephens P, Devlin F, Chabalowski C, Frisch MJ. Ab initio calculation of vibrational absorption and circular dichroism spectra using density functional force fields. J Phys Chem 1994;98:11623e7. Grimme S. Semiempirical GGA-type density functional constructed with a long-range dispersion correction. J Comput Chem 2006;27:1787e99. Sun H. COMPASS: an ab initio force-field optimized for condensed-phase applications overview with details on alkane and benzene compounds. J Phys Chem B 1998;102:7338e64. Morawa Eblagon K, Tam K, Yu KK, Zhao S-L, Gong X-Q, He H, et al. Study of catalytic sites on ruthenium for hydrogenation of N-ethylcarbazole: implications of hydrogen storage via reversible catalytic hydrogenation. J Phys Chem C 2010;114:9720e30. Hehre WJ. Ab initio molecular orbital theory. Accounts Chem Res 1976;9:399e406. Scott AP, Radom L. Harmonic vibrational frequencies: an evaluation of Hartree Fock, Møller Plesset, quadratic configuration interaction, density functional theory, and semiempirical scale factors. J Phys Chem 1996;100:16502e13. Pople JA, Scott AP, Wong MW, Radom L. Scaling factors for obtaining fundamental vibrational frequencies and zeropoint energies from HF/6e31G* and MP2/6e31G* harmonic frequencies. Isr J Chem 1993;33:345e50. Merrick JP, Moran D, Radom L. An evaluation of harmonic vibrational frequency scale factors. J Phys Chem A 2007;111:11683e700. Verevkin SP, Emel’yanenko VN, Pimerzin AA, Vishnevskaya EE. Thermodynamic analysis of strain in the five-membered oxygen and nitrogen heterocyclic compounds. J Phys Chem A 2011;115:1992e2004. Stewart JJ. MOPAC manual. Tokyo: Fujitsu Ltd; 1993. Frenkel M, Chirico RD, Diky V, Yan X, Dong Q, Muzny C. ThermoData Engine (TDE): software implementation of the dynamic data evaluation concept. J Chem Inf Model 2005;45:816e38. Klamt A. Conductor-like screening model for real solvents: a new approach to the quantitative calculation of solvation phenomena. J Phys Chem 1995;99:2224e35. Klamt A, Jonas V, Bu¨rger T, Lohrenz JC. Refinement and parametrization of COSMO-RS. J Phys Chem A 1998;102:5074e85. Eckert F, Klamt A. COSMOtherm, Version C3. 0, release 13.01. Leverkusen: COSMOlogic GmbH & Co KG; 2013. Eckert F, Klamt A. Fast solvent screening via quantum chemistry: COSMO-RS approach. AIChE J 2002;48:369e85. Klamt A. COSMO-RS: from quantum chemistry to fluid phase thermodynamics and drug design. Amsterdam: Elsevier; 2005.

[34] Ambrose D, Walton J. Vapour pressures up to their critical temperatures of normal alkanes and 1-alkanols. Pure Appl Chem 1989;61:1395e403. [35] Lin S-T, Chang J, Wang S, Goddard WA, Sandler SI. Prediction of vapor pressures and enthalpies of vaporization using a COSMO solvation model. J Phys Chem A 2004;108:7429e39. [36] Luo W, Campbell PG, Zakharov LN, Liu S-Y. A singlecomponent liquid-phase hydrogen storage material. J Am Chem Soc 2011;133:19326e9. [37] Chen G, Zakharov LN, Bowden ME, Karkamkar AJ, Whittemore SM, Garner III EB, et al. Bis-BN cyclohexane: a remarkably kinetically stable chemical hydrogen storage material. J Am Chem Soc 2014;137:134e7. [38] Stark K, Keil P, Schug S, Muller K, Wasserscheid P, Arlt W. Melting points of potential liquid organic hydrogen Carrier systems consisting of N-alkylcarbazoles. J Chem Eng Data 2016;61:1441e8. [39] Mehranfar A, Izadyar M, Esmaeili AA. Hydrogen storage by N-ethylcarbazol as a new liquid organic hydrogen Carrier: a DFT study on the mechanism. Int J Hydrogen Energy 2015;40:5797e806. [40] Mehranfar A, Izadyar M. Theoretical evaluation of Nalkylcarbazoles potential in hydrogen release. Int J Hydrogen Energy 2017;42:9966e77. [41] Yang M, Dong Y, Fei S, Pan Q, Ni G, Han C, et al. Hydrogenation of N-propylcarbazole over supported ruthenium as a new prototype of liquid organic hydrogen carriers (LOHC). RSC Adv 2013;3:24877e81. [42] Sotoodeh F, Huber BJ, Smith KJ. Dehydrogenation kinetics and catalysis of organic heteroaromatics for hydrogen storage. Int J Hydrogen Energy 2012;37:2715e22.  n BS, Staudt T, [43] Sobota M, Nikiforidis I, Amende M, Zano € fert O, et al. Dehydrogenation of dodecahydro-NHo ethylcarbazole on Pd/Al2O3 model catalysts. Chem Eur J 2011;17:11542e52. [44] Cooper AC. Design and development of new carbon-based sorbent Systems for an effective containment of hydrogen. Allentown: PA: Air Products and Chemicals Inc; 2012. [45] Sotoodeh F, Smith KJ. Analysis of H2 release from organic polycyclics over Pd catalysts using DFT. J Phys Chem C 2013;117:194e204. [46] Luo W, Campbell PG, Zakharov LN, Liu SY. A singlecomponent liquid-phase hydrogen storage material. J Am Chem Soc 2011;133:19326e9. [47] Lide DR. CRC handbook of chemistry and physics. 80th ed. 2000. Boca Raton: FL. [48] Steele WV, Chirico R, Collier W, Hossenlopp I, Nguyen A, Strube M. Thermochemical and thermophysical properties of organic nitrogen compounds found in fossil materials. 1986. Bartlesville: OK. [49] Storrie FR, Tucker SH. Syntheses in the carbazole series. The alleged synthesis of 3-nitro-N-ethylcarbazole. J Chem Soc 1931;0:2255e63.  z ME, Dobrowolski JC, Polaczek J, Szafran  ski AM, [50] Jamro Kazimirski JK, Lisicki Z. (Solidþ liquid) equilibria, i.r., and semi-empirical studies of carbazole with selected tertiary amides. J Chem Thermodyn 2001;33:565e79. [51] Tucker SH. Iodination in the carbazole series. J Chem Soc 1926;129:546e53. [52] Dunlop HG, Tucker SH. Attempts to prepare optically active tervalent nitrogen compounds. Part I. Syntheses of 1, 9phenylenecarbazole and derivatives. J Chem Soc 1939:1945e56.

Please cite this article in press as: Shin BS, et al., Thermodynamic assessment of carbazole-based organic polycyclic compounds for hydrogen storage applications via a computational approach, International Journal of Hydrogen Energy (2018), https://doi.org/10.1016/ j.ijhydene.2018.04.182