1
A hybrid cylindrical model for characterization of MCM-41 by density functional theory

Type: Object | Advantage: None | Novelty: New | ConceptID: Obj1

2
A hybrid cylindrical model for characterization of MCM-41 by density functional theory (DFT) is proposed in this work, where the surface heterogeneity of MCM-41 is taken into account by using a hybrid potential model to represent the interactions between a pore wall and molecules inside the pore.

Type: Object | Advantage: None | Novelty: New | ConceptID: Obj2

3
This model consists of two parts: (1) the potential energies from the oxygen atoms inside the wall, represented by the potential model proposed by our group; (2) the potential energies from the silanol coverage and/or other unknown factors in the surface of the channel of MCM-41, represented by the cylindrical surface potential function of Tjatjopoulos et al. (G. J. Tjatjopoulos, D. L. Feke and J. A. Mann, J. Phys. Chem., 1988, 92, 4006–4007).

Type: Model | Advantage: None | Novelty: None | ConceptID: Mod1

4
To test the new model, the DFT method was used to calculate the adsorption isotherm of nitrogen in MCM-41 at 77 K. The isotherm calculated is compared with the experimental data as well as the calculated results of Maddox et al., who divided the surface of MCM-41 into eight sectors and adopted different parameters for each sector to consider the heterogeneity of the surface.

Type: Object | Advantage: None | Novelty: New | ConceptID: Obj3

5
Compared with the work of Maddox et al.

Type: Conclusion | Advantage: None | Novelty: None | ConceptID: Con1

6
(M. W. Maddox, J. P. Olivier and K. E. Gubbins, Langmuir, 1997, 13, 1737–1745), our model gives a much better fit to the experimental isotherm of nitrogen at 77 K in the pressure range of P/P0 = 0.2–0.5 with much less parameter and computation effort, where phase transition and capillary condensation occur.

Type: Conclusion | Advantage: None | Novelty: None | ConceptID: Con1

7
Furthermore, the relationship between the reduced pressure, at which capillary condensation takes place, and the pore diameter by the hybrid model is in good agreement with that obtained by Maddox et al.

Type: Conclusion | Advantage: None | Novelty: None | ConceptID: Con2

8
In addition, adsorption and phase behavior of methane and ethane are studied by the model, and the calculated results also coincide well with the experimental isotherms of methane and ethane at 264 K–373 K. Therefore, the hybrid potential model incorporating into the DFT method provides a useful tool for characterization of MCM-41.

Type: Conclusion | Advantage: None | Novelty: None | ConceptID: Con3

Introduction

9
Porous materials have been applied in many fields, such as gas separation, purification, and reaction processes, etc.

Type: Background | Advantage: None | Novelty: None | ConceptID: Bac1

10
In the meantime, when a fluid is confined to a region of a molecular scale, its phase behavior can be strongly affected, and a rich variety of new types of phase transitions can occur.

Type: Background | Advantage: None | Novelty: None | ConceptID: Bac2

11
Therefore, investigations on characterization and the adsorption behavior of porous materials are of great importance from both scientific and practical points of view.

Type: Motivation | Advantage: None | Novelty: None | ConceptID: Mot1

12
In 1992, a new family of mesoporous molecular sieves designated as M41S was discovered by Mobil scientists.1,2

Type: Background | Advantage: None | Novelty: None | ConceptID: Bac3

13
As is pointed out by Ravikovitch et al.,3 the main advantages of MCM-41, a representative of M41S, are (1) of an array of uniform hexagonal channels with very narrow pore size distributions, (2) the pore length is significantly larger than the pore diameter, (3) pore networking effects are negligible.

Type: Background | Advantage: None | Novelty: None | ConceptID: Bac4

14
Therefore, because of its ideal pore structure, there exists a strong incentive for researchers to study and use it.

Type: Motivation | Advantage: None | Novelty: None | ConceptID: Mot2

15
Not only can it be used in adsorption, separation and catalysis, it is also regarded as the most suitable model adsorbent currently available for verification of theoretical predictions for cylindrical pores.

Type: Motivation | Advantage: None | Novelty: None | ConceptID: Mot3

16
Many experimental investigations have been performed on the adsorption properties of MCM-41,1,4–8 which provides deep insights into the adsorption behavior of simple fluids in MCM-41, and also serves as an experimental basis for the test and development of theoretical models.

Type: Background | Advantage: None | Novelty: None | ConceptID: Bac5

17
On the other hand, both the molecular simulation and density functional theory (DFT) methods have been used to study the adsorption behavior of simple fluids in MCM-41 and its pore size distribution (PSD).

Type: Background | Advantage: None | Novelty: None | ConceptID: Bac6

18
For example, Ravikovitch et al3,9,10. carried out many investigations to study the adsorption isotherms of nitrogen in MCM-41 and PSD of MCM-41 by the DFT method, Maddox et al11–14. used the grand canonical Monte Carlo (GCMC) model to study the adsorption behavior of a pure fluid and binary mixtures in MCM-41.

Type: Background | Advantage: None | Novelty: None | ConceptID: Bac7

19
In both the DFT and MC methods, a potential model for the wall–fluid interactions is required, which plays a key role in the accuracy of the calculated properties.

Type: Background | Advantage: None | Novelty: None | ConceptID: Bac8

20
As a result, many potential models have been proposed for Lennard-Jones (LJ) cylindrical pores, such as the cylindrical surface potential model of Tjatjopoulos et al.,15 the potential model as the analog of 10–4–3 for planar graphite surface,11–13 the pseudoatomic model of Nicholson and Gubbins,16 the model of Peterson et al.,17 and a simplified model proposed by Zhang and Wang.18

Type: Background | Advantage: None | Novelty: None | ConceptID: Bac9

21
Recently, a complete analytical potential model for a cylindrical pore of finite thickness was further proposed by our group.19

Type: Background | Advantage: None | Novelty: None | ConceptID: Bac10

22
It gives identical results to the model of Peterson et al.,17 while inconvenient numerical integration in the latter is avoided.

Type: Background | Advantage: None | Novelty: None | ConceptID: Bac11

23
In addition, it reduces to the cylindrical surface model of Tjatjopoulos et al.,15 when the wall thickness approaches zero.

Type: Background | Advantage: None | Novelty: None | ConceptID: Bac12

24
Previous works9,14,18 show that the experimental isotherms can not be reproduced very well by using the existing potential models in the whole pressure range if the surface heterogeneity of MCM-41 is not taken into account.

Type: Background | Advantage: None | Novelty: None | ConceptID: Bac13

25
Therefore, Maddox et al14. divided the surface of MCM-41 into eight sectors and adopted different parameters for each sectors to consider the heterogeneity of its surface.

Type: Background | Advantage: None | Novelty: None | ConceptID: Bac14

26
Their work gives a good representation of the adsorption isotherms of nitrogen in MCM-41 by the GCMC method.

Type: Background | Advantage: None | Novelty: None | ConceptID: Bac14

27
However, considerably greater parameter and computation effort is needed.

Type: Background | Advantage: None | Novelty: None | ConceptID: Bac14

28
In this work, we focus on the heterogeneity of MCM-41 by using the DFT method, because the DFT method is remarkably efficient in the computation of the meso-porous materials, which are of simple, low-dimensional geometry and their external potentials, i.e. the interactions between the solid surface and fluid molecules, can be described by explicit analytical functions.

Type: Object | Advantage: None | Novelty: New | ConceptID: Obj4

29
A hybrid potential model is proposed here to consider the surface heterogeneity of MCM-41, which consists of two parts: (1) the potential energies from the oxygen atoms inside the wall, represented by the complete analytical model developed by our group recently.19

Type: Object | Advantage: None | Novelty: New | ConceptID: Obj5

30
(2) the potential energies from the silanol coverage and/or other unknown factors in the surface of the channels of MCM-41, represented by the cylindrical surface potential function of Tjatjopoulos et al.15

Type: Model | Advantage: None | Novelty: None | ConceptID: Mod2

31
To test this new model, first, the adsorption isotherm of nitrogen in MCM-41 at 77 K is calculated.

Type: Object | Advantage: None | Novelty: New | ConceptID: Obj6

32
Then, calculated and experimental adsorption isotherms of methane and ethane confined in MCM-41 pores at different temperatures are compared.

Type: Object | Advantage: None | Novelty: New | ConceptID: Obj7

33
Based on our results, the hybrid potential model incorporated into the DFT method is recommended as a useful tool for characterization of MCM-41.

Type: Conclusion | Advantage: None | Novelty: None | ConceptID: Con4

34
In addition, some discussion is also addressed.

Type: Background | Advantage: None | Novelty: None | ConceptID: Bac15

Density functional theory and potential models

Density functional theory

35
In the DFT approach, the grand potential (GP) of the fluid in a pore at a given temperature T, and the chemical potential μ is a functional of the local fluid density ρ(r):where F[ρ(r)] is the intrinsic Helmholtz free energy functional and v(r) is the potential of the pore walls.

Type: Model | Advantage: None | Novelty: None | ConceptID: Mod3

36
The criterion for equilibrium is that the grand potential reaches a minimum, namely, the density profile ρ(r) of the adsorbate within the pore satisfies the condition:In this paper the expression of free energy in eqn. (1) is split into attractive and repulsive contributions.

Type: Model | Advantage: None | Novelty: None | ConceptID: Mod4

37
The attractive part of the fluid–fluid potential is given by the Weeks–Chandler–Anderson perturbation scheme20 for a cut and shifted LJ potential.

Type: Model | Advantage: None | Novelty: None | ConceptID: Mod5

38
The repulsive part of hard spheres, Fhs, is then split into an ideal part, Fhsid, and an excess part, Fhsex:The ideal part is trivial, and the excess part is given by Tarazona’s recipe.21,22

Type: Model | Advantage: None | Novelty: None | ConceptID: Mod6

39
Detailed information can be referred to in our previous work23.

Type: Background | Advantage: None | Novelty: None | ConceptID: Bac16

The potential model

The potential model between a pair of fluid molecules

40
In this work, we use the LJ potential to represent the interaction between a pair of fluid moleculeswhere r is the interpartical distance, ε is the well depth , and σ is the molecular size distance between a pair of molecules when the potential is zero.

Type: Model | Advantage: None | Novelty: None | ConceptID: Mod7

41
The full description can be referred to the ref. .24

Type: Model | Advantage: None | Novelty: None | ConceptID: Mod8

42
Parameters of the LJ potential are listed in Tables 1 and 2.

Type: Model | Advantage: None | Novelty: None | ConceptID: Mod9

The potential model between the fluid molecules and MCM-41 pore wall

43
A hybrid potential model is proposed to represent the interactions between the pore wall and the molecules inside the pore.

Type: Model | Advantage: None | Novelty: None | ConceptID: Mod10

44
This model consists of two parts: (1) the potential energies from the oxygen atoms inside the wall, represented by the complete analytical potential model proposed by our group19 (see Appendix A); (2) the potential energies from the silanol coverage and/or other unknown factors on the surface of the channels of MCM-41, represented by the cylindrical surface potential model of Tjatjopoulos et al.15

Type: Model | Advantage: None | Novelty: None | ConceptID: Mod11

45
A schematic diagram of this model is shown in Fig. 1(a).

Type: Model | Advantage: None | Novelty: None | ConceptID: Mod12

The interactions between the fluid molecules and the oxygen atoms in MCM-41

46
Traditionally, we ignore the silicon atoms in the adsorbent structure, and just take into account the interactions between the fluid molecules and the oxygen atoms in MCM-41.

Type: Model | Advantage: None | Novelty: None | ConceptID: Mod13

47
In this work, a complete analytical potential model proposed19 (see Appendix A) is used to represent this interaction, in which the interaction energy, U, of the testing fluid molecule within a cylindrical pore of infinite thickness is given bywhere F is the hypergeometric series, R is the radius of the cylindrical pore, l is the radial distance from the testing particle to the center of the pore, and ρsolid is the number density of interaction sites inside the pore wall.

Type: Model | Advantage: None | Novelty: None | ConceptID: Mod14

48
A skeletal density of 2.0 g cm−3 is used in this work, as adopted in the theoretical model of Feuston and Higgins.25

Type: Model | Advantage: None | Novelty: None | ConceptID: Mod15

49
Parameters εsf, σsf for solid–fluid interactions are obtained by the Lorentz–Berthelot combining rules:If the wall thickness is finite, the interaction energy of the testing fluid molecule experienced can be calculated by U(R1,l) − U(R2,l), where R1 and R2 are the inside radius and outside radius of the cylindrical pore, respectively.

Type: Model | Advantage: None | Novelty: None | ConceptID: Mod16

The interactions between the fluid molecules and the silanol coverage and/or other unknown factors on the MCM-41 surface

50
As an approximation, we made the surface heterogeneity of MCM-41 caused by the silanol coverage and/or other unknown factors on the surface of the pores of MCM-41 “smooth”, that is, it is assumed that all the heavy cations and/or functional groups attached to the surface of MCM-41 are uniformly distributed.

Type: Model | Advantage: None | Novelty: None | ConceptID: Mod17

51
Therefore, the interactions between the attractive sites and a fluid molecule inside MCM-41, U(r,R), can be represented by the cylindrical surface potential function proposed by Tjatjopoulos et al.:where R is the radius of the cylindrical pore, r is the radial distance from the test particle to the surface of the wall and F is the hypergeometric series.

Type: Model | Advantage: None | Novelty: None | ConceptID: Mod18

52
Because we know little about the composition, density and array mode of these attractive sites present on the surface of MCM-41, we take εsurf,fρsurf as an adjustable parameter.

Type: Model | Advantage: None | Novelty: None | ConceptID: Mod19

53
It consists of two parts: εsurf,f is the surface heavy cations and/or functional groups–fluid interaction parameter, ρsurf is the density of these attractive sites on the wall.

Type: Model | Advantage: None | Novelty: None | ConceptID: Mod20

54
The parameter εsurf,fρsurf for nitrogen was obtained by using the method proposed in the literature,10,26 and the parameters were determined for methane and ethane as Yun et al. recommended.27

Type: Model | Advantage: None | Novelty: None | ConceptID: Mod21

55
The values of the parameters obtained are listed in Tables 1 and 2.

Type: Model | Advantage: None | Novelty: None | ConceptID: Mod22

56
The parameter for the surface molecule, σsurf is set to 0.276 nm,24 and parameter σsurf,f is obtained by the Lorentz–Berthelot combining rules.

Type: Model | Advantage: None | Novelty: None | ConceptID: Mod23

Results and discussion

Adsorption isotherm of N2 in MCM-41 at 77 K

57
The DFT method combined with the hybrid cylindrical potential model was used to calculate the adsorption isotherm of nitrogen in MCM-41 initially at 77 K. The calculated results and the experimental data14 are shown in Fig. 2.

Type: Object | Advantage: None | Novelty: New | ConceptID: Obj8

58
Parallel calculations were also carried out for the models of the complete analytical potential model and the model of Tjatjopoulos et al.

Type: Object | Advantage: None | Novelty: New | ConceptID: Obj9

59
(see Table 1 for the parameters) to see the improvement of the hybrid model.

Type: Object | Advantage: None | Novelty: New | ConceptID: Obj9

60
From Fig. 2 it is found that the hybrid potential model shows improved results over the two models.

Type: Observation | Advantage: None | Novelty: None | ConceptID: Obs1

61
For a detailed comparison between the three models, results are shown in Fig. 3.

Type: Observation | Advantage: None | Novelty: None | ConceptID: Obs2

62
Obviously, when the reduced pressure P/P0 varying from 0 to 0.1, the model of Tjatjopoulos et al. works better than the complete analytical model, and the hybrid model gives nearly identical results to that of the model of Tjatjopoulos et al., as is shown in Fig. 3a.

Type: Observation | Advantage: None | Novelty: None | ConceptID: Obs3

63
The results from these three models are comparable, while the hybrid model being slightly better in the range of P/P0 = 0.1–0.3.

Type: Observation | Advantage: None | Novelty: None | ConceptID: Obs4

64
When P/P0 is in the range of 0.3–0.6, the complete analytical model is better than the model of Tjatjopoulos et al., and the hybrid model works slightly better than the complete analytical model.

Type: Observation | Advantage: None | Novelty: None | ConceptID: Obs5

65
Consequently, it is concluded that the hybrid model can incorporate the advantages of the two constituent models, leading to a better description of the experimental adsorption isotherm in the pressure ranges.

Type: Result | Advantage: None | Novelty: None | ConceptID: Res1

66
Our calculated results are further compared with the work of Maddox et al.,14 who divided the surface of MCM-41 into eight sectors and adopted different parameters for each sector to consider the heterogeneity of its surface (see Fig. 1b).

Type: Object | Advantage: None | Novelty: New | ConceptID: Obj10

67
Their results from GCMC method are shown in Fig. 4, along with our calculated results.

Type: Observation | Advantage: None | Novelty: None | ConceptID: Obs6

68
Obviously, in the very low pressure region, the method of Maddox et al. is better than our model.

Type: Observation | Advantage: None | Novelty: None | ConceptID: Obs7

69
This observation is not surprising, because adsorption at very low pressures is very sensitive to the potential model used, and the cylindrical wall is divided into eight sectors with different parameters in the method of Maddox et al.

Type: Result | Advantage: None | Novelty: None | ConceptID: Res2

70
In contrast, in the range of P/P0 = 0.2–0.4, our model gives better results.

Type: Observation | Advantage: None | Novelty: None | ConceptID: Obs8

71
More impressively, our hybrid model describes phase transition and capillary condensation very well, while the model of Maddox et al. shows large discrepancies in the region.

Type: Observation | Advantage: None | Novelty: None | ConceptID: Obs9

72
Besides, it is noticed that much more complicated potential functions and more parameters are used in their method.

Type: Observation | Advantage: None | Novelty: None | ConceptID: Obs10

Relationship between pore size and capillary condensation pressure for N2 at 77 K

73
Furthermore, the relationship between the reduced pressure at which capillary condensation takes place and the pore diameter is determined with our model.

Type: Result | Advantage: None | Novelty: None | ConceptID: Res3

74
Fig. 5 presents eight isotherms by DFT method for pore size ranging from 2.5 nm to 4.2 nm.

Type: Observation | Advantage: None | Novelty: None | ConceptID: Obs11

75
The reduced pressure at which capillary condensation takes place can then be accurately determined here.

Type: Result | Advantage: None | Novelty: None | ConceptID: Res4

76
Fig. 6 shows the relationship between the capillary condensation pressure and the pore diameter.

Type: Object | Advantage: None | Novelty: New | ConceptID: Obj11

77
The results are in good agreement with those obtained by the Maddox et al. using the Monte Carlo method14.

Type: Observation | Advantage: None | Novelty: None | ConceptID: Obs12

Adsorption isotherms for methane and ethane

78
To further test the hybrid model, the experimental adsorption isotherms of methane and ethane in MCM-41 pores with a pore diameter of 4.1 nm and a wall thickness of 1.0 nm measured by Yun et al27. were collected from the literature.

Type: Object | Advantage: None | Novelty: New | ConceptID: Obj12

79
They measured two isotherms, T = 303.15 and 373.15 K, for methane and four isotherms, T = 264.75, 273.55, 303.15 and 373.15 K, for ethane.

Type: Observation | Advantage: None | Novelty: None | ConceptID: Obs13

80
In addition, they used the GCMC approach to calculate the isotherms and obtained good agreement between simulated and experimental values.

Type: Background | Advantage: None | Novelty: None | ConceptID: Bac17

81
In this work, these experimental data are adopted to further test the new model.

Type: Object | Advantage: None | Novelty: New | ConceptID: Obj13

82
The parameters for methane and ethane molecules are listed in Table 2.

Type: Observation | Advantage: None | Novelty: None | ConceptID: Obs14

83
The calculated results by using the DFT method with the hybrid model are shown in Figs. 7 and 8.

Type: Observation | Advantage: None | Novelty: None | ConceptID: Obs15

84
Obviously, the calculated isotherms are in good agreement with the experimental data for methane, while the DFT predictions of ethane adsorptions are somewhat less accurate, possibly caused by the simplified treatment of the adsorbate–adsorbate interactions.

Type: Result | Advantage: None | Novelty: None | ConceptID: Res5

85
Nevertheless, even under these conditions the DFT predictions show a consistent tendency with experimental observations.

Type: Observation | Advantage: None | Novelty: None | ConceptID: Obs16

86
This again illustrates that the hybrid potential model incorporating into the DFT method provides a useful tool for characterization of MCM-41.

Type: Conclusion | Advantage: None | Novelty: None | ConceptID: Con5

Conclusions

87
A hybrid cylindrical model for characterization of MCM-41 is proposed in this work, where the heterogeneity of the pore of MCM-41 is taken into account by dividing the interactions between the fluid molecules in the pore and the wall into two parts: (1) the potential energies from the oxygen atoms inside the wall, represented by the complete analytical potential model proposed by our group recently;19 (2) the potential energies from the silanol coverage and/or other unknown factors in the surface of the channel of MCM-41, represented by the cylindrical surface potential function of Tjatjopoulos et al.15

Type: Object | Advantage: None | Novelty: New | ConceptID: Obj14

88
(see Fig. 1a).

Type: Background | Advantage: None | Novelty: None | ConceptID: Bac18

89
The hybrid potential model proposed enables the DFT method to represent better the adsorption isotherm of nitrogen in MCM-41 at 77 K than the two constituent potential models, and gives a better fit in the range of P/P0 = 0.2–0.4, and accurate capillary condensation pressure, compared with the work of Maddox et al.,14 in which the pore of MCM 41 consists of eight sectors with different potential parameters (see Fig. 1b).

Type: Result | Advantage: None | Novelty: None | ConceptID: Res6

90
Impressively, the dependence of the capillary condensation pressure on pore diameter for adsorption of nitrogen at 77 K can be reproduced very well with the hybrid model.

Type: Result | Advantage: None | Novelty: None | ConceptID: Res7

91
To test the new model, the adsorption isotherms of methane and ethane ranging from 264–373 K in MCM-41 were predicted with good agreement.

Type: Result | Advantage: None | Novelty: None | ConceptID: Res8

92
As a result, the present work demonstrates that the hybrid potential model incorporating into the DFT method provides a useful tool for characterization of MCM-41.

Type: Conclusion | Advantage: None | Novelty: None | ConceptID: Con6

93
It should be noted that this work addresses the heterogeneity induced by a smooth potential energy on the surface of MCM-41.

Type: Conclusion | Advantage: None | Novelty: None | ConceptID: Con7

94
In fact, there exist some defects and domains of active groups on the surface of MCM-41, which can not be simply represented by a smooth potential function.

Type: Conclusion | Advantage: None | Novelty: None | ConceptID: Con8

95
We probably should introduce a more complicated method to take into account the effects of both the energy and geometry heterogeneity.

Type: Conclusion | Advantage: None | Novelty: None | ConceptID: Con9

96
This will be discussed in our future work.

Type: Background | Advantage: None | Novelty: None | ConceptID: Bac19