Attribution Non-Commercial (BY-NC)

Просмотров: 23

Attribution Non-Commercial (BY-NC)

- Ft Dow Corning 200 Fluid
- Chapter_ 2 Hydraulic Fluid
- Norma ASTM D1200
- Calculating Viscosities of Reservoir Fluids From Their Composition
- Exercise 8 Full Report
- Gao Bower Msmse04
- FM & F Machines 2013.pdf
- HY30-3400-UK-GP1-GPA
- Extraction
- K2 B 231116123141
- Wash Coat
- physicsinvestigatoryprojectforcolleges-140119070411-phpapp01
- CondensateStab2013AIChE91aPaper292728
- Squishy Materials
- Poster
- 211430934-Experiment-4
- Tangler t Pai Bul 1987
- My judge
- Lecture 2
- Low-Toxicity Oil Muds a Knowledge of Downhole Rheological Bhavior Assists Successful Field Application - Bailey 1986

Вы находитесь на странице: 1из 12

Chemical Physics

journal homepage: www.elsevier.com/locate/chemphys

Inﬂuence of the chain length on the dynamic viscosity at high pressure of some

2-alkylamines: Measurements and comparative study of some models

Masatoshi Yoshimura a,b, Christian Boned a,*, Guillaume Galliéro a, Jean-Patrick Bazile a,

Antoine Baylaucq a, Hideharu Ushiki b

a

Laboratoire des Fluides Complexes, Faculté des Sciences et Techniques, UMR CNRS 5150, Université de Pau et des Pays de l’Adour, BP 1155, 64013 Pau Cedex, France

b

Laboratory of Molecular Dynamics and Complex Chemical Physics, Department of Environmental and Natural Resource Science, Faculty of Agriculture,

Tokyo University of Agriculture and Technology, 3-5-8 Saiwai-cho, Fuchu-shi, Tokyo 183-8509, Japan

a r t i c l e i n f o a b s t r a c t

Article history: This work reports the dynamic viscosity data (a total of 93 points) of 2-alkylamines, which exhibit small

Received 24 December 2009 association, consisting of 2-aminobutane, 2-aminopentane, 2-aminoheptane and 2-aminooctane at four

In ﬁnal form 12 March 2010 temperatures between 293.15 K and 353.15 K (every 20 K), and pressures up to 100 MPa (every

Available online 18 March 2010

20 MPa) which allows to study the inﬂuence of the chain length. A falling-body viscometer with an uncer-

tainty of ±2% was used to perform these measurements.

Keywords: The variations of dynamic viscosity are discussed with respect to their behaviour due to chain length.

2-Alkylamine

Seven different models, most of them with a physical and theoretical background, are studied in order to

Chain length

Modeling

investigate how they take the chain length effect into account through their required model parameters.

Pressure The evaluated models are based on the empirical Vogel–Fulcher–Tamman (VFT) representation (com-

Viscosity bined with a Tait-like equation), the rough hard-sphere scheme, the concept of the free-volume, the fric-

tion theory, a correlation derived from molecular dynamics, a model based on Eyring’s absolute rate

theory combined with cubic EoS. A scaling viscosity representation has also been considered. These mod-

els need some adjustable parameters except the molecular dynamics correlation that is entirely predic-

tive. Overall a satisfactory representation of the viscosity of these 2-alkylamines is found for the different

models within the considered T, p range taking into account their simplicity. Moreover, it has been ver-

iﬁed that the viscosity is a unique function of TVc where the exponent c is generally related to the steep-

ness of the intermolecular repulsive potential (T: temperature and V: speciﬁc volume).

Ó 2010 Elsevier B.V. All rights reserved.

2-aminoheptane (CH3(CH2)4CH(NH2)CH3), and 2-aminooctane

In a previous work by our group on some linear primary amines (CH3(CH2)5CH(NH2)CH3), has been measured up to 100 MPa be-

[1], it has been underlined that there is a lack of information on the tween 293.15 K and 353.15 K. The structure of these molecules is

thermophysical properties of aliphatic amines. Aliphatic amines shown in Fig. 1. To the best of our knowledge, concerning these

are used as a solvent, as a raw material in the manufacture of a 2-alkylamines, which have branching structure (i.e. amino group

variety of other compounds, including emulsiﬁers and pharmaceu- is located at the second carbon of alkyl chain), there are no exper-

tical products. An accurate database of thermophysical properties imental data of viscosity versus pressure. Mention here that the

of these ﬂuids is of practical and fundamental value for the chem- density measurements for two 2-alkylamines (2-aminobutane

ical engineering application. and 2-aminooctane) have previously been performed with a

With the aim of completing a database relative to this chemical vibrating-tube densimeter as a function of temperature and pres-

family, we have published some experimental viscosity data con- sure [2]. In addition, the density of 2-aminopentane and 2-amino-

cerning linear amines (pentyl-, hexyl- and heptylamine) versus heptane measured as in Ref. [2] up to 100 MPa between 293.15 K

pressure up to 100 MPa in the temperature interval (293.15 K and 353.15 K are reported in this work.

and 353.15 K) [1]. From a theoretical point of view, an accurate experimental

In this work, the dynamic viscosity g (93 experimental data in database of dynamic viscosity is usable to investigate the appropri-

total) for four 2-alkylamines: 2-aminobutane (sec-butylamine, ateness of some viscosity models taking the chain length and

branching effect into account. In particular from the structural dif-

* Corresponding author. Tel.: +33 559 407 688; fax: +33 559 407 695. ference between two kinds of amines, linear 1-alkylamines and

E-mail address: christian.boned@univ-pau.fr (C. Boned). branched 2-alkylamines, it is possible to study how the difference

0301-0104/$ - see front matter Ó 2010 Elsevier B.V. All rights reserved.

doi:10.1016/j.chemphys.2010.03.016

M. Yoshimura et al. / Chemical Physics 369 (2010) 126–137 127

of the position of amino-group, which affects intermolecular inter- is designed with a ratio between its diameter and the tube diame-

actions, inﬂuences the macroscopic transport property i.e. viscos- ter greater than 0.98, which is substantially above the recom-

ity. In this work, seven different viscosity models are studied. mended value of 0.93 in order to ensure a concentric fall and to

The considered models are the empirical Vogel–Fulcher–Tamman minimize eccentricity effects [21–23]. Since the density of the sin-

(VFT) [3–5] representation combined with a Tait-like equation, ker is about 9 times higher than the density of the ﬂuids considered

the rough hard-sphere scheme [6,7], the model based on free-vol- in this work an error in the ﬂuid density of 0.1% results in an error

ume concept [8–10], the model based on the friction theory of about 1/7000 for qS qL (see relation (1)) and consequently a

[11,12], a model derived from molecular dynamics simulation very small error on viscosity. In this work, Ds corresponds to the

[13], the Eyring theory and cubic EoS model [14,15] and a scaling average value of six measurements of the falling time at thermal

viscosity representation [16,17]. and mechanical equilibrium with a reproducibility of 0.5%.

The calibration of the viscometer has been performed using

2. Experimental procedure accurate viscosity and density data for toluene and decane. The re-

quired reference viscosity and density data for toluene have been

The dynamic viscosity g under pressure was measured using a estimated by the correlation given for the viscosity and density

falling-body viscometer. This viscometer is of the type designed by Assael et al. [24]. The reported uncertainties for the calculated

by Daugé et al. [18]. The viscometer consists of two high-pressure density and viscosity values are 0.03% and 2%, respectively. For

cells, a measuring cell and a piston cell, which are connected by a n-decane, the viscosity data has been obtained by the correlation

capillary tube and a valve, see Fig. 1 in Ref. [18]. The piston cell is given by Huber et al. [25] using density values calculated by the

connected to a pneumatic oil pump, which is used to pressurize the expression given by Cibulka and Hnědkovský [26]. The reported

viscometer. The pressure of the sample within the viscometer is uncertainty for the calculated n-decane density and viscosity val-

measured by an HBM-P3M manometer connected directly to the ues are 0.1% and 2%, respectively. The apparatus parameters Ka

tube between the two cells, ensuring a measure of the real pressure and Kb are determined at each considered (T, p) condition by plot-

of the sample. The pressure is measured with an uncertainty of ting the reference viscosities of the two calibrating ﬂuids as a func-

0.2 MPa. The temperature is measured inside the measuring cell tion of (qS qL)Ds.

by a Pt100 probe connected to a classical AOIP thermometer with In this work, the densities of two 2-alkylamines (2-aminobu-

an uncertainty of 0.5 K. A circulating ﬂuid supplied by an external tane and 2-aminooctane) were taken from Ref. [2], where they

thermostatic bath controls the temperature of the sample in the have been measured up to 140 MPa between 293.15 K and

measuring cell and the piston cell. The viscometer is placed in an 353.15 K with a vibrating-tube densimeter. The uncertainty re-

automated air-pulsed thermal regulator oven in order to ensure a ported for these density measurements is ±5 104 g cm3

homogeneous temperature surrounding the system. Since the ﬂu- (around 0.05%). Taking into account the uncertainty due to the cal-

ids considered in this work are liquids at atmospheric pressure, the ibration, the temperature, the pressure and the density, the overall

ﬁlling procedure of the viscometer was done as described in Ref. uncertainty for the reported dynamic viscosities is of the order of

[19]. 2%. The accuracy of the device has been checked several times.

The basic principle of the falling-body viscometer is that a sin- For instance, recently in our group [19], the same instrument has

ker falls through a ﬂuid of unknown viscosity under a given tem- been used to measure the dynamic viscosity of methanol between

perature and pressure (T, p) condition. It has been emphasized by 293.15 K and 353.15 K up to 100 MPa. These data have been fur-

Daugé et al. [18] that, for this type of viscometer and for ﬂuids with ther used in a study about a new reference correlation [27] for

a low viscosity, a working equation of the functional form the viscosity of methanol, taking into account of many methanol

g(T, p) = f[(qS qL)Ds] should be used. This working equation re- literature data. Our data agree with the reference correlation with

lates the dynamic viscosity to the difference between the density an average absolute deviation of 1.0%, which is very satisfactory.

of the sinker qS and of the ﬂuid qL, and the falling time, Ds, be- At atmospheric pressure (0.1 MPa) the dynamic viscosity was

tween two detection sensors when the velocity of the sinker is con- obtained by measuring the kinematic viscosity, m = g/q, with a clas-

stant. For ﬂuids with a very low dynamic viscosity, such as sical capillary viscometer (Ubbelohde). For this purpose several

methane, Daugé et al. [18] used a second-order polynomial in tubes connected to an automatic AVS350 Schott Geräte Analyzer

(qS qL)Ds which implies the requirement of three reference ﬂu- were used. The temperature of the ﬂuid is controlled within 0.1 K

ids in order to perform the calibration of the viscometer. In this using a thermostatic bath. When multiplying the kinematic viscos-

work the lowest viscosity is 0.285 mPa s for 2-aminobutane at ity with the density, the dynamic viscosity is obtained with an

T = 353.15 K and p = 20 MPa, which is not too low, and conse- uncertainty less than 1%. Each capillary tube is provided with a cal-

quently it was found appropriate to use a linear relation for the ibration certiﬁcate but we checked the calibration of the capillary

working equation as follows: viscometer at several temperature using ‘‘Cannon Certiﬁed Viscos-

ity Standard” reference ﬂuids.

gðT; pÞ ¼ K a ðT; pÞ þ K b ðT; pÞðqS qL ÞDs; ð1Þ 2-aminobutane (sec-butylamine, C4H11N, molar mass M = 73.14

which relates the dynamic viscosity to two parameters Ka and Kb. A g mol1, boiling point at atmospheric pressure Tb = 336.15 K, CAS

similar working equation has recently been used by Pensado et al. number 13952-84-6), 2-aminopentane (C5H13N, M = 87.16

[20]. g mol1, Tb = 363.65 364.65 K, CAS number 63493-28-7), 2-ami-

The sinker used in this work is a solid stainless steel cylinder noheptane (C7H17N, M = mass 115.22 g mol1, Tb = 415.15

with hemispherical ends and a density of 7.72 g cm3. The sinker 417.15 K, CAS number 123-82-0) and 2-aminooctane (C8H19N,

Fig. 1. Molecular structure of (a) 2-aminobutane, (b) 2-aminopentane, (c) 2-aminoheptane, and (d) 2-aminooctane.

128 M. Yoshimura et al. / Chemical Physics 369 (2010) 126–137

tained from Sigma–Aldrich with purity of, respectively, 99%, 97%,

99% and 99% (with certiﬁcate of analysis by gas chromatography

of, respectively, 99.9%, 98.7%, 99.0% and 99.41%). These chemicals

were not subject to further puriﬁcation and were directly injected

into the high-pressure cell as soon as the bottle was opened.

3. Experimental results

ken from Ref. [2] and the values of viscosity and density measured

in this work for four 2-alkylamines are reported in Table 1. The

measurements have been performed at various temperatures

(293.15, 313.15, 333.15 and 353.15) K for pressures at (0.1, 20,

40, 60, 80 and 100) MPa. Note, however, that it was not possible

Fig. 2. Comparison with literature data of viscosity at atmospheric pressure for 2-

to measure viscosity or density at some conditions for 2-aminobu- aminobutane. }: Bai 2005 [29], s: Rao 2006 [28], h: Saleh 2001 [31], 4: Subha

tane (p = 0.1 MPa, T = 333.15 and 353.15 K) and 2-aminopentane 2004 [30], d: this work. Solid line is ﬁtting curve for all the values from literature

(p = 0.1 MPa, T = 353.15 K) as temperature is near or over their Tb. and this work with quadratic polynomial function.

As we did not ﬁnd in the literature viscosity values at high pres-

sure, the comparison was possible only at atmospheric pressure.

We found data only for 2-aminobutane (sec-butylamine) at amines from Ref. [2]. From Fig. 3, it can be seen that the viscosity

T = 303.15 K by Rao et al. [28], at T = 308.15 K by Bai et al. [29] decreases monotonically with increasing temperature. Fig. 4 shows

and Subha et al. [30], and at T = 298.15 up to 323.15 K (DT = 5 K) that the viscosity increases with increasing pressure. This behavior

by Saleh et al. [31]. Among these data, it is possible to do the quan- is the one generally observed in liquids. In fact, this is not surpris-

titative comparison only at T = 313.15 K with [31], and we obtained ing as primary amines exhibit a weak hydrogen bonding. Fig. 5

a deviation of 3.9%. Fig. 2 shows for 2-aminobutane our data and shows the variation of the viscosity versus density at p = 0.1 and

the various literature viscosity data versus temperature at atmo- 100 MPa. Fig. 6 shows the variation of the viscosity versus density

spheric pressure. The solid line corresponds to a polynomial qua- at T = (293.15 and 353.15) K. It is interesting to notice that, in the

dratic ﬁtting obtained using all the data. For the other three 2- case of constant pressure (Fig. 5), the viscosity data points versus

alkylamines, viscosity data was not found even at atmospheric density are roughly on the same curve independently of the alkyl

pressure. chain length of amines, however, it is not the case at constant tem-

The variation of the viscosity versus temperature at p = 60 MPa perature (Fig. 6). This indicates that it is the density which mainly

for four 2-alkylamines of this work and two 1-alkylamines (pentyl- determines the viscosity amplitude, more than the temperature or

amine and heptylamine) from Ref. [2] is shown in Fig. 3. Fig. 4 the chain length.

shows the variation of the viscosity versus pressure at The difference between four 2-alkylamines (2-aminobutane, 2-

T = 313.15 K for four 2-alkylamines of this work and two 1-alkyl- aminopentane, 2-aminoheptane and 2-aminooctane) is the length

Table 1

Dynamic viscosity g (mPa s) and density q (g cm3) versus temperature and pressure for four 2-alkylamines (2-aminobutane, 2-aminopentane, 2-aminoheptane and 2-

aminooctane).

g q g q g q g q

293.15 0.1 0.460 0.7253a 0.622 0.7478 0.903 0.7642 1.159 0.7720a

293.15 20 0.557 0.7412a 0.731 0.7625 1.083 0.7777 1.389 0.7848a

293.15 40 0.648 0.7545a 0.864 0.7748 1.294 0.7893 1.699 0.7961a

293.15 60 0.738 0.7661a 1.001 0.7855 1.540 0.7994 2.068 0.8060a

293.15 80 0.830 0.7763a 1.141 0.7951 1.819 0.8085 2.493 0.8150a

293.15 100 0.921 0.7855a 1.285 0.8038 2.148 0.8167 3.005 0.8232a

313.15 0.1 0.360 0.7050a 0.469 0.7293 0.661 0.7477 0.829 0.7558a

313.15 20 0.440 0.7234a 0.551 0.7457 0.793 0.7627 0.981 0.7704a

313.15 40 0.512 0.7382a 0.644 0.7594 0.950 0.7753 1.187 0.7825a

313.15 60 0.584 0.7508a 0.742 0.7712 1.118 0.7863 1.417 0.7932a

313.15 80 0.657 0.7619a 0.845 0.7816 1.300 0.7961 1.673 0.8027a

313.15 100 0.730 0.7718a 0.953 0.7909 1.494 0.8049 1.955 0.8114a

333.15 0.1 –b –b 0.364 0.7099 0.511 0.7302 0.630 0.7396a

333.15 20 0.350 0.7052a 0.432 0.7290 0.610 0.7473 0.740 0.7555a

333.15 40 0.409 0.7220a 0.504 0.7443 0.727 0.7612 0.887 0.7690a

333.15 60 0.469 0.7358a 0.578 0.7571 0.852 0.7732 1.048 0.7806a

333.15 80 0.529 0.7480a 0.653 0.7685 0.986 0.7839 1.228 0.7909a

333.15 100 0.593 0.7586a 0.730 0.7784 1.129 0.7934 1.426 0.8000a

353.15 0.1 –b –b –b –b 0.411 0.7132 0.496 0.7232a

353.15 20 0.285 0.6871a 0.348 0.7120 0.485 0.7322 0.596 0.7416a

353.15 40 0.336 0.7058a 0.410 0.7290 0.576 0.7477 0.708 0.7562a

353.15 60 0.386 0.7212a 0.471 0.7430 0.673 0.7607 0.825 0.7687a

353.15 80 0.436 0.7342a 0.531 0.7552 0.778 0.7721 0.949 0.7797a

353.15 100 0.488 0.7457a 0.590 0.7659 0.891 0.7822 1.080 0.7895a

a

The values for 2-aminopentane and 2-aminoheptane are taken from Ref. [2].

b

Viscosity and density cannot be measured on the condition in empty columns as temperature is near or over Tb (p = 0.1 MPa).

M. Yoshimura et al. / Chemical Physics 369 (2010) 126–137 129

Fig. 3. Dynamic viscosity versus temperature at p = 60 MPa for four 2-alkylamines Fig. 6. Dynamic viscosity versus density at T = 293.15 K (j: 2-aminobutane, : 2-

(j: 2-aminobutane, : 2-aminopentane, N: 2-aminoheptane and d: 2-aminooc- aminopentane, N: 2-aminoheptane and d: 2-aminooctane) and T = 353.15 K (h: 2-

tane) and two 1-alkylamines (}: pentylamine and 4: heptylamine) [2]. aminobutane, }: 2-aminopentane, 4: 2-aminoheptane and s: 2-aminooctane).

mine which has longer alkyl chain tends to form denser, less com-

pressible liquid than other 2-alkylamines which have shorter alkyl

chain at a given pressure.

From the viewpoint of the difference in the position of amino

group, Figs. 3 and 4 indicate that 1-alkylamines, which have linear

structure, show higher viscosity than 2-alkylamines, which have

branching structure, in the temperature and pressure range con-

sidered here. The same tendency was observed in the measure-

ment of density [2].

Generally, when a ﬂuid is brought under pressure (compressed),

the ﬂexibility and mobility of the molecules are reduced, since the

distance and space between the molecules become shorter, result-

ing in a reduction of the ﬂuid mobility and an increase in the vis-

cosity. Kioupis and Maginn [32] explained the fact that the

viscosity increases with pressure as a result of a reduction in the

Fig. 4. Dynamic viscosity versus pressure at T = 313.15 K for four 2-alkylamines (j: liquid void volume coupled with the molecular structure, resulting

2-aminobutane, : 2-aminopentane, N: 2-aminoheptane and d: 2-aminooctane) in a lower motion of the molecules, because the motion is related

and two 1-alkylamines (}: pentylamine and 4: heptylamine) [2]. either to molecules jumping or forcing adjacent molecules into

these voids. Therefore, when the number of voids decreases with

increasing pressure, complex molecules with a low ﬂexibility will

have difﬁculties of making these jumps or forcing other molecules

into these voids, resulting in the trapping of the molecules and a

higher viscosity.

For the four amines considered in this work, a slightly sharper

increase of dynamic viscosity against pressure has been found for

2-aminooctane than for 2-aminoheptane, -pentane and -butane

(Fig. 4) due to the increase of the alkyl chain length, which results

in an important reduction of the ﬂuid mobility (higher viscosity)

when brought under pressure.

4. Discussion

esting to investigate how the effect of the chain length is taken into

account by some viscosity models. First we consider an empirical

Fig. 5. Dynamic viscosity versus density at p = 0.1 MPa (j: 2-aminobutane, : 2-

aminopentane, N: 2-aminoheptane and d: 2-aminooctane) and p = 100 MPa (h: 2-

model based on Vogel–Fulcher–Tamman’s (VFT) viscosity repre-

aminobutane, }: 2-aminopentane, 4: 2-aminoheptane and s: 2-aminooctane). sentation [3–5] combined with Tait-like equation. Second, six more

or less recently derived models with a physical and theoretical

background are considered: rough hard-sphere scheme [6,7],

of alkyl chain in their molecular structure (Fig. 1) which affects the free-volume scheme [8–10], friction theory [11,12], correlation

dynamic viscosity for a given temperature and pressure. As ex- based on molecular dynamics simulation of Lennard–Jones ﬂuid

pected the viscosity increases with the chain length for a given [13], a scaling viscosity model [16,17], and model based on the Eyr-

set of T and p (Figs. 3 and 4), but notice that it is less sensitive to ing theory combined with a cubic EoS [14,15]. For the models using

the chain length for a given set of p and q or T and q (Figs. 5 and experimental density values, they are taken from Ref. [2] and from

6). Further, as observed in the density measurement [2], 2-alkyla- this work (Table 1).

130 M. Yoshimura et al. / Chemical Physics 369 (2010) 126–137

In order to make a comparative study of the performance of dif- that the worse overall MD, found for 2-aminooctane at T = 353.15 K

ferent viscosity models to represent the viscosity, the following and p = 0.1 MPa, is of the same order of magnitude as the experi-

deﬁnitions are used. mental error.

4.2. The rough hard-sphere scheme

1 X NP

AAD ¼ jDeviationi j;

NP i¼1 The second viscosity model employed is based on the fact that a

ð2Þ corresponding states relationship exists [6,7] between the experi-

MD ¼ MaximumjDeviationi j;

mental transport properties of rough non-spherical molecules

1 X NP

and the smooth hard-sphere values (subscript shs):

Bias ¼ Deviationi ;

NP i¼1 gexp ¼ Rg gshs ; ð4Þ

where NP is the number of data points, gexp the experimental vis- where the proportionality factor Rg, described as the roughness fac-

cosity and gcalc the calculated viscosity. The AAD (average absolute tor, accounts for the roughness and non-spherical shape of the mol-

deviation) indicates how close the calculated values are to the ecule. The roughness factor is related to the non-spherical shape of

experimental values, and the quantity Bias indicates how well the the molecule (see e.g. [35]) and Rg = 1 corresponds to a spherical

experimental points are distributed around the calculated points. molecule. Dymond and Awan [6] derived the following expression

If Bias is equal to AAD, then all of the calculated values are below relating the reduced smooth hard-sphere viscosity gshs to the exper-

the experimental values. Further, the quantity MD refers to the imental value gexp:

absolute maximum deviation. rﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ

gexp 1 gexp v 2=3

gshs ¼ ¼ 6:035 108 ð5Þ

4.1. Vogel–Fulcher–Tamman–Tait representation Rg MRT Rg

with v the molar volume, M the molar mass and R the gas constant.

The ﬁrst viscosity model considered in this work is one already

In order to determine Rg and v0, the close packing molar volume, for

proposed and described more precisely in a previous work of our

a given temperature a plot of log10 ðgexp Þ versus log10(m) from the

team [33]:

experiment is superimposed on a universal plot of log10 ðgshs Þ versus

EðTÞ þ p log10(m/m0) from the hard-sphere theory by vertical and horizontal

gðp; TÞ ¼ A exp½B=ðT CÞ exp D ln ; ð3Þ

EðTÞ þ 0:1MPa adjustments. The empirical expression for this curve is:

X i

where D was assumed to be temperature independent and E(T) is a gexp 7

1

second-order polynomial as E(T) = E0 + E1T + E2T2. This equation is log10 ¼ ag;i : ð6Þ

Rg i¼0

vr

derived from a Tait-like equation combined with the empirical Vo-

gel–Fulcher–Tamman’s model [3–5] which can be considered as The ag,i coefﬁcients [7] are universal, independent of the chemical

Andrade’s modiﬁed equation [34]. nature of the compound. This has been veriﬁed by Baylaucq et al.

Experimental viscosity data have been ﬁtted with Eq. (3) by [36,37]. Further, it is generally assumed that v0 is temperature

Levenberg–Marquardt algorithm and the obtained coefﬁcients dependent, whereas Rg is temperature independent [35].

are shown in Table 2 for four 2-alkylamines. In order to use them, The estimated hard-core volume v0 for each temperature

p is in MPa, T in K, M in kg m3 and the viscosity is in mPa s. Notice and the roughness factor Rg are given in Table 3 for the four

Table 2

Values of coefﬁcients and results obtained with the Vogel–Fulcher–Tamman’s representation (Eq. (3)).

A 0.00017525 0.017364 0.011448 0.047302

B 4941.2 835.53 1213.6 561.81

C 334.21 59.182 14.627 116.71

D 0.93693 1.5163 3.5456 7.1394

E0 17.154 165.64 1491 2517

E1 0.58743 1.8448 10.84 17.388

E2 0.0011376 0.002501 0.015464 0.021993

AAD (%) 0.250 0.410 0.541 0.736 0.484

Bias (%) 0.000514 0.00411 0.0338 0.0539 0.0228

MD (%) 0.76 0.97 1.16 2.64 2.64

Table 3

Adjusted values of the roughness factor Rg and the hard-core volume v0 (m3 mol1) in the rough hard-sphere viscosity scheme (Eqs. (5) and (6)).

2-Aminobutane 2-Aminopentane 2-Aminoheptane 2-Aminooctane Overall

Rg 1.261 1.287 1.241 1.290

v0 (293.15 K) 6.445E05 7.730E05 1.049E04 1.190E04

v0 (313.15 K) 6.378E05 7.621E05 1.036E04 1.174E04

v0 (333.15 K) 6.301E05 7.506E05 1.025E04 1.160E04

v0 (353.15 K) 6.198E05 7.394E05 1.014E04 1.147E04

AAD (%) 1.033 0.631 0.628 1.092 0.846

Bias (%) 0.086 0.036 0.053 0.050 0.005

MD (%) 2.07 1.50 1.34 2.77 2.77

M. Yoshimura et al. / Chemical Physics 369 (2010) 126–137 131

2-alkylamines. The calculation has been done in order to minimize dense supercritical states. Recently [10], the model has been gener-

MD for all the data. The modeling for the four compounds resulted alized to the simultaneous modeling of the self-diffusion coefﬁcient

in an AAD of 0.85% with a MD of 2.77% close to the magnitude of and dynamic viscosity at high pressure (up to 500 MPa).

the experimental error. The obtained v0 is increasing with molar The dilute gas viscosity term g0 can be obtained by any appro-

mass and is decreasing with temperature. The obtained Rg is not priate model, for instance the model by Chung et al. [40] which is

regular against molar mass but as the values are comprised be- able to predict the dilute gas viscosity of several polar and non-po-

tween 1.24 and 1.29 this fact is perhaps not signiﬁcant. Compared lar ﬂuids within an uncertainty of 1.5%. The model is an empirical

with the case of 1-alkylamines [1], it seems that Rg is less impor- correlation derived from the Chapman–Enskog theory [41] and the

tant for 2-amines than for 1-amines (1.394 for 1-heptylamine reduced collision integral expression for the Lennard–Jones 12-6

and 1.241 for 2-aminoheptane). This result is probably due to the potential of Neufeld et al. [42]. This model is related to the critical

branching effect. temperature Tc, the critical molar volume vc and the acentric factor

In order to make an analysis of the inﬂuence of the molecular x.

structure, the next modeling step is to represent the variation of Unfortunately, experimental critical property data of organic

the dynamic viscosity as a function of molecular mass. As the var- compounds are limited due to the fact that many compounds be-

iation of Rg is not simple, the parameter v0 has only been corre- come unstable during measurements near or even far from the

lated, by expressing it with a simple linear function, on the vicinity of the critical point. We found some experimental and rec-

overall data set (93 points). The correlation used for (m0(T, M) is ommended values in Marsh et al. [43], which is a part nine of re-

the following: view series for vapor–liquid critical properties of elements and

compounds. For 2-aminobutane (i.e. sec-butylamine)

m0 ðT; MÞ ¼ ða1 M þ a0 ÞT þ ðb1 M þ b0 Þ: ð7Þ

Tc = 514.3 ± 0.3 K (recommended value is 514 ± 1 K) and

This expression allows a reduction of the total number of parame- pc = 5.0 MPa (recommended value is 5.0 ± 0.5 MPa), and for 2-ami-

ters from 20 to 8. Table 4 shows parameters and result of re-estima- noheptane Tc = 598.0 ± 0.3 K (recommended value is 598.0 ± 0.6).

tion using the value of Rg obtained in Table 3. This parameter In such conditions, mathematical models can be used to provide

reduction resulted in an AAD of 4.37% with a MD of 8.80%. Both a reasonable estimate of these properties. A variety of estimation

AAD and MD have increased and are higher than the magnitude methods for critical property data are available in the open litera-

of experimental uncertainty. There seems to be branching effect, ture. A broad overview of these methods together with a detailed

which cannot be described very well with the simple linearization discussion of their reliability was given by Poling et al. [44]. Re-

of parameters against temperature and molar mass. This model, cently, Nannoolal et al. [45] developed a new group contribution

with the development of parameters Rg and v0, has already been method for the prediction of critical properties and their method

successfully applied in the case of 1-alkylamines [1] and in some has been compared with 10 well-known estimation methods from

other previous works [38,39]. literature. More recently, a position group contribution method,

which requires only the knowledge of their chemical structure,

4.3. Free-volume viscosity model for the prediction of critical temperature Tc [46], critical pressure

pc [47] and critical volume vc [48] have been proposed. Their meth-

Based on the free-volume concept, an approach has recently od, owing to the utilization of the position compensation factor,

been proposed in order to model the viscosity of Newtonian ﬂuids demonstrates signiﬁcant improvements compared to the previ-

in the gaseous and dense states [8,9]. In this approach, the total vis- ously used ﬁrst- or second- order method, especially in the capabil-

cosity g can be separated into a dilute gas viscosity term g0 and an ity of distinguishing between isomers. In order to evaluate the

additional term Dg, in the following way: dilute gas viscosity term g0, in this work, the critical properties

were estimated by the position group contribution method [46–

g ¼ g0 þ Dg: ð8Þ

48]. In addition, the acentric factor was estimated by the method

The term Dg characterizes the passage in the dense state and is con- recommended in Ref. [44], in which the estimated critical proper-

nected to the molecular structure via a representation of the free ties Tc and pc and the experimental Tb were used with the recom-

volume fraction. The general expression of the free-volume viscos- mended equation:

ity model is [8]:

" lnðPc =1:01325Þ þ f ð0Þ ðT br Þ

# x¼ ; ð10Þ

aq þ pM=q aq þ pM=q 3=2 f ð1Þ ðT br Þ

g ¼ g0 þ q‘ pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ exp B ; ð9Þ

3RTM RT

where T br ¼ T b =T c ; f ð0Þ and f ð1Þ : analytical expressions developed by

where ‘ corresponds to a characteristic molecular length. a is linked Ambrose and Walton [49]. Table 5 shows the values of Tb described

to the barrier energy E0 = aq, which the molecules have to exceed in as material speciﬁcation by Sigma–Aldrich and estimated values of

order to diffuse. B is characteristic of the free-volume overlap. This critical properties and acentric factor. As Tb of 2-aminopentane and

equation involves three physical parameters ‘, a and B, which are 2-aminoheptane are described as ranged value, we used the middle

characteristic of the molecule. This model has been shown to accu- value of the range. By the comparison with experimental value, we

rately represent the viscosity behavior of various hydrocarbons over found deviations for 2-aminobutane of Tc and pc are 1.4 and 9.1%,

wide ranges of temperature and pressure in the gaseous, liquid and and a deviation for 2-aminoheptane of Tc is 0.70%.

Table 4

Parameters and results of the rough hard-sphere viscosity scheme with the linearization of parameter v0 by Eq. (7).

v0 (m3 mol1) a0 a1 b0 b1

6.463E09 4.970E07 4.225E06 1.108E03

AAD (%) 7.01 6.53 1.72 2.22 4.37

Bias (%) 7.01 6.53 1.72 2.09 1.07

MD (%) 8.70 8.80 3.88 8.80 8.80

132 M. Yoshimura et al. / Chemical Physics 369 (2010) 126–137

Experimental boiling point (at atmospheric pressure) and evaluated values of critical is smaller for 2-aminooctane (0.04085 nm) than for 2-aminobu-

properties by the position group contribution method [46–48] and acentric factor by

the method in Ref. [44].

tane (0.06391 nm). This is related to the molecular structure that

is longer for 2-aminobutane molecule (Fig. 1). According to [8],

Tb (K) Tc (K) pc (MPa) vc (cm3 mol1) x in Eq. (9), ‘ ¼ L2 =bf , where L2 is an average characteristic molecular

2-Aminobutane 336.15 521.4 4.547 274.0 0.2757 quadratic length, and bf is the dissipation length of the energy

2-Aminopentane 364.15 547.5 4.190 333.1 0.3683 E = E0 + pM/q where the term pM/q = pv is related to the energy

2-Aminoheptane 416.15 593.8 3.671 455.6 0.5654

2-Aminooctane 438.15 615.3 3.466 518.0 0.6307

necessary to form the vacant vacuums available for the diffusion

of the molecules. Certainly as L2 seems to increase with the length

of alkyl chain, the decrease of ‘ means that bf increases with alkyl

chain length more than L2. Finally, it has to be mentioned that it is

It is important to notice that, as the compounds studied here are the value for the molecular energy which has a more important

in the dense state, the value of g0 is low compared to the total vis- contribution to the total viscosity in excess of the dilute gas viscos-

cosity. For the four compounds the maximum value for dilute gas ity. This is conﬁrmed by the fact that 2-aminobutane, which has

viscosity is g0 = 0.008 mPa s for 2-aminobutane at T = 353.15 K (to- lowest experimental viscosity in the four compounds, shows lower

tal viscosity of 2-aminobutane at p = 20 MPa and T = 353.15 K is value for the energy parameter a than the other compounds,

g = 0.285 mPa s). We noticed also that g0 is not very sensitive to though it shows higher value of ‘ and B than the other compounds.

small variations of the critical parameters and acentric factor. Compared with the case of 1-alkylamines [1], no particular effect

The three characteristic parameters in Eq. (9) have been esti- appears that could clearly be related to the branching effect.

mated by minimizing MD for each of the four amines. The esti- The procedure presented for the rough hard-sphere scheme can

mated parameters are presented in Table 6 with the deviation also be used to reduce the number of the parameters. Each main

results. The dynamic viscosity is obtained in (Pa s), when all vari- parameter ‘; a and B has then been correlated against the number

ables and properties are inserted in SI units. It should be noticed of carbon atoms n of the alkyl chain with linear equations as

that the AAD for each of the compounds are lower than the exper- A = A1M + A0 (with A ¼ ‘; a or B) so that only six parameters are

imental uncertainty with MD lower than 3.30% (obtained for 2- used, for the four amines. The results and parameters obtained

aminooctane at p = 20 MPa and T = 313.15 K). The overall results, by this correlation are shown in Table 7. Compared with the previ-

with an AAD of 1.01%, can be considered very satisfying because ous result (Table 6), AAD shows a little bit worse result (1.62% for

the calculations involve only three parameters for each one of all the data instead of 1.01%) but it is still lower than the experi-

the four amines. A comparison of the speciﬁc energy parameter a mental uncertainty. For the maximum deviation the result is

shows that it increases with the number of CH2 groups. By multi- 3.68% for the 2-aminobutane (T = 293.15 K, p = 0.1 MPa) and 2-

plying the speciﬁc energy parameter with the density, an inﬂuence aminooctane (T = 293.15, p = 100 MPa). It should be noticed that

on the energy barrier, E0 = aq, due to the chain length effects can a increases with M, and that ‘ and B decrease when M increases,

be seen, since the energy barrier E0, which the molecules has to ex- as already observed in Table 6.

ceed in order to diffuse, is approximately 73–78% higher for 2-ami- In Eq. (9) the density appears explicitly and for the calculation

nooctane than for 2-aminobutane. The larger energy will result in a we used the experimental density data. It is important to under-

lower mobility of the molecules in 2-aminooctane and conse- line, however, that the model could be applied even if the density

quently in a higher viscosity. In case of the B parameter, which is is not known by using an efﬁcient equation of state along with Eq.

related to the free-volume overlap of molecules, a pronounced ef- (9). Of course the parameters obtained in such way are partially

fect due to the chain length effect is also found: it decreases with linked to the chosen equation of state. Notice that this remark is

the chain length. The obtained B value of 2-aminooctane is signif- valid for all the models where density appears explicitly. In this

icantly lower than for 2-aminobutane. The effect of chain length aim, recently the free-volume model has been successfully coupled

Table 6

Results obtained on the four compounds with the free-volume viscosity model.

‘ (nm) 0.0639102 0.0442546 0.0469946 0.0408502

a (J m3 kg1 mol1) 93.766 125.364 140.117 160.181

B 0.00827095 0.00675202 0.00657969 0.00613258

AAD (%) 0.90 0.56 0.77 1.82 1.01

Bias (%) 0.051 0.217 0.071 0.555 0.090

MD (%) 1.81 1.46 1.53 3.30 3.30

Table 7

Results obtained on the four compounds with the free-volume viscosity model and the correlated parameters (A = A1 M + A0 with M in kg mol1).

A1 A0

‘ (nm) 0.0428623 0.0526912

a (J m3 kg1 mol1) 613.256 67.3793

B 0.00112600 0.00688086

2-Aminobutane 2-Aminopentane 2-Aminoheptane 2-Aminooctane Overall

AAD (%) 2.08 1.13 1.32 1.94 1.62

Bias (%) 1.506 0.913 1.191 0.177 0.263

MD (%) 3.68 3.55 3.15 3.68 3.68

M. Yoshimura et al. / Chemical Physics 369 (2010) 126–137 133

with the statistical associating ﬂuid theory (SAFT model) [50]. By suggesting that the critical viscosity is a linear function of

Moreover in this latter work the authors successfully demonstrate the molar mass and assuming that the third order friction constant

that it is possible to correlate and predict simultaneously the vis- is a constant independent of the 2-aminohydrocarbon, the three

cosity of n-alkanes using the free-volume model with SAFT theory, adjustable parameters were obtained and reported in the table

with parameters that are universal for the whole series of caption of Table 9. The deviation results obtained are reported in

n-alkanes with only six parameters. In the data-pool ﬁtting, they Table 9. The overall AAD = 3.08% with Bias = 0.35% and

assumed, like in the present work, that the parameters of the MD = 10.4%, which is a satisfactory result.

free-volume model scale linearly with molecular weight, and this It is important to note that when accurate and reliable viscosity,

simple assumption has been effective. density and phase behaviour predictions can be achieved, then

accurate models for these properties can be linked and connected.

In this sense, as already underlined above, the free-volume model

4.4. f-Theory model

has been successfully coupled with the statistical associating ﬂuid

theory for pure compounds [50]. The friction theory has also been

Starting from basic principles of mechanics and thermodynam-

coupled to different types of SAFT models [50,54]. In the f-theory

ics, the friction theory (f-theory) for viscosity modeling has been

the density does not appear explicitly, but it is necessary to know

introduced [11]. In the f-theory the total viscosity can be written

the attractive pressure and the repulsive pressure. In a recent pa-

as:

per it is shown how the f-theory can be linked to practically any

g ¼ g0 þ gf ð11Þ type of EoS [55] ranging from highly theoretical EoS to highly accu-

rate empirical reference EoS. However, the adjustable parameters

where g0 is the dilute gas viscosity and gf the residual friction con- in this model are linked to the equation of state chosen for the

tribution. The friction contribution is related to the van der Waals calculation.

attractive and repulsive pressure terms, pa and pr, of an equation

of states (EoS), such as the Peng and Robinson (PR) [51] or the 4.5. Molecular dynamics viscosity model

Soave–Redlich–Kwong (SRK) [52] ones. Based on this concept, a

general f-theory model [12] valid for hydrocarbons has been intro- Recently, a predictive viscosity approach has been introduced

duced with 16 constants identical for all hydrocarbons. In this work, for simple pure ﬂuids and mixtures over a wide range of tempera-

the f-theory approach in conjunction with the PR EoS, as described ture and pressure [13]. This approach is derived from molecular

in our previous work [53] has been used. In this approach there are dynamics simulations using a corresponding state scheme, where

two adjustable parameters, a ‘‘characteristic” critical viscosity gc the Lennard–Jones (LJ) ﬂuid is taken as the reference compound

and a ‘‘third order constant” d2. The required dilute gas viscosity and a one-ﬂuid approximation is applied to mixtures. A simple cor-

of the pure compounds has been obtained by the Chung et al. model relation has been developed in order to accurately reproduce re-

[40], as for free-volume theory. cent molecular dynamics results on the LJ ﬂuid over a large

The two adjustable parameters, the estimated critical viscosity range of thermodynamic states [13].

gc and the third order friction constant as well as the deviation re- In this model, the reduced viscosity g* has been expressed ver-

sults are shown in Table 8. The obtained results are satisfactory sus reduced temperature T* and reduced density q* (see [13] for

taking into account that they are obtained in conjunction with a details) as a sum of a classical Chapman–Enskog dilute density

simple cubic EoS with only two adjustable parameters. In compar- contribution g0 and of a residual viscosity contribution Dg [13] as

ison with the two previous other models with a physical back- in a free-volume and f-theory approaches:

ground (rough hard-sphere scheme and free-volume theory) g ðT ; q Þ ¼ g0 ðT Þ þ Dg ðT ; q Þ: ð12Þ

Table 8, shows that the overall values of AAD and MD found for this

model are slightly higher. In this model the dilute density contribution is deﬁned as

Plotting the critical viscosity reported in Table 8 as a function of pﬃﬃﬃﬃﬃ

T

the molar mass reveals that the critical viscosity is nearly a linear g0 ðT Þ ¼ 0:17629Ac ; ð13Þ

Xv

function of the molar mass (of the form gc = gc1M + gc0) The third

order friction constant reported in Table 8 cannot be represented where Ac = 0.95, and Xv is the collision integral. Neufeld et al. [42]

by a linear relationship with the molar mass. have derived expressions for different collision integrals. In Ref.

Table 8

Results obtained with the f-theory viscosity model in conjunction with the PR EOS by adjusting the critical viscosity gc and the third order friction constant d2.

gc (lP) 346.14 331.71 301.41 297.309

d2 (lP/bar3) 4.69969 109 5.21001 109 3.58445 109 3.81666 109

AAD (%) 1.61 1.09 3.43 3.97 2.56

Bias (%) 0.04 0.02 0.15 0.21 0.11

MD (%) 4.75 3.50 7.45 8.22 8.22

Table 9

Results obtained with the f-theory viscosity model in conjunction with the PR EOS by correlating the critical viscosity gc = gc1 M + gc0 with M in kg mol1 [gc1 = 1173.36

(lP mol kg1) and gc0 = 437.693 (lP)] and the third order friction constant d2 = 4.26789 109 (lP/bar3).

AAD (%) 1.68 2.06 4.44 3.99 3.08

Bias (%) 0.70 1.27 3.00 1.07 0.35

MD (%) 6.26 4.89 10.4 8.74 10.4

134 M. Yoshimura et al. / Chemical Physics 369 (2010) 126–137

Table 10

Results obtained with the molecular dynamics viscosity model.

AAD (%) 42.3 33.9 21.7 17.7 28.9

Bias (%) 42.3 33.9 21.7 17.7 28.9

MD (%) 46.4 39.5 29.9 30.8 46.4

vious methods, but this model is the only one purely predictive as

there is no parameter to adjust. Nevertheless, it is important to no-

tice that the model predicts correctly an increase of the viscosity

when the pressure increases and a decrease of the viscosity when

the temperature increases. Moreover the viscosity increases from

2-aminobutane to 2-amoinooctane (see Fig. 7 which correspond

to p = 60 MPa). From the viewpoint of the difference of molecular

structure, the overall results for branching amines of this work

are approximately twice worse than the ones for linear amines

studied in previous work [1]. The increase of the difference from

the theory value is thought to be due to the branching effect,

namely, this suggests the insufﬁciency of this method in order to

predict the structural isomers.

It is possible to improve the results with adjustment of the

molecular parameter r, but in this case the model is no more pre-

dictive. In fact, it should be mentioned that Eq. (16) in conjunction

Fig. 7. Molecular dynamics based correlation: comparison between the experi-

mental (closed symbols) and the estimated (open symbols) dynamic viscosity at with the critical molar volume vc is efﬁcient only for simple com-

p = 60 MPa for 2-aminobutane (j, h), 2-aminopentane (h, }), 2-aminoheptane (N, pounds [13]. For more complex molecules, vc (i.e. r, see Eq. (16))

4) and 2-aminooctane (d, s). should be adjusted on viscosity data. The values of vc adjusted by

minimizing MD and the results are shown in Table 11. By perform-

ing this adjustment in order that MD would be minimum, we

[13] the expression for the 12-6 collision integral is used. The resid- found, for 2-aminobutane, -pentane, -heptane and -octane, respec-

ual viscosity contribution is expressed as: tively, the adjusted values vc = 295.50, 350.75, 468.47 and

529.29 cm3 mol1 instead of the estimated values vc = 274.0,

Dg ðT ; q Þ ¼ b1 ½expðb2 q Þ 1 þ b3 ½expðb4 q Þ 1 333.1, 455.6 and 518.0 cm3 mol1 (Table 5). The AAD, Bias and

þ b5 ½expðb6 q Þ 1ðT Þ2 ; ð14Þ MD have been signiﬁcantly improved in comparison with the ones

in Table 10. This direct estimation of vc for the four compounds

where the bi coefﬁcients have been regressed against molecular using experimental viscosities clearly indicates that the model is

dynamics simulations results on the LJ pure ﬂuid [13]. very sensitive to the LJ parameter rii as previously shown [13].

In order to apply this approach to real ﬂuids, the two LJ molec- Finally we have supposed that the molecular length r has the

ular parameters (rii: molecular length and eii: energy parameter) relationship with molar mass as r = r1M + r0. The correlated

that are supposed to represent the real compound, are required. parameters for r, the values of vc calculated by Eq. (16) and results

These molecular parameters have been related to the critical tem- are presented in Table 12. The AAD, Bias and MD are worse than

perature Tc in K and the critical molar volume vc in m3 mol1, the ones obtained by the estimation of vc for each compounds (Ta-

through ble 11), however, they are still much better than the ones obtained

by the purely predictive method (Table 10).

kB T c As already underlined, in a predictive way this scheme provides

eii ¼ ; ð15Þ

1:2593 a reasonable estimation of the viscosity of these compounds. The

v c 1=3 overall results can be considered satisfactory compared to the sim-

rii ¼ 0:302 ; ð16Þ

plicity of this scheme. An improvement is obtained with only 1 ad-

NA

justed molecular parameter per compound (or 2 for the four

Eq. (15) has been proposed by Chung et al. [40] and Eq. (16) has compounds taken simultaneously), the critical molar volume of

been proposed by Galliéro et al. [13]. The unit for eii is (J), and for the pure ﬂuids (or the molecular size).

rii is (m).

It is worth to underline that this model is entirely predictive, 4.6. Eyring theory and cubic EoS model

contrary to the models described in previous sections, as no

parameter adjustment has to be done to the experimental viscosity On the basis of the Eyring’s Absolute Rate Theory [56], Lei et al.

database. Using the critical properties given in Table 5 we found [57] derived the following two-parameters model for correlating

the results indicated in Table 10. The model underestimates the the viscosity of pure liquids under saturated conditions:

viscosity. The results are much worse than the ones from the pre-

Table 11

Results obtained with the molecular dynamics viscosity model with the adjusted critical molar volumes.

vc (cm3 mol1) 295.50 350.75 468.47 529.29

AAD (%) 4.24 2.94 3.90 5.05 4.03

Bias (%) 2.27 0.83 3.06 3.59 1.30

MD (%) 8.44 6.07 5.95 8.88 8.88

M. Yoshimura et al. / Chemical Physics 369 (2010) 126–137 135

Table 12

Results obtained with the molecular dynamics viscosity model and the correlated parameters (r = r1M + r0 with M in kgmol1).

r (nm) r1 r0

2.023440 0.381871

vc (cm3 mol1) 296.65 346.89 463.87 531.06

AAD (%) 4.08 8.60 6.79 8.02 6.87

Bias (%) 0.90 8.60 6.79 7.50 1.75

MD (%) 13.0 12.7 13.0 13.0 13.0

RT 1 DU v ap with overall AAD values of 0.94% using Eyring-PR model, whose va-

g¼ exp a exp ðZ L Þ; ð17Þ

VL c RT lue is within experimental uncertainty. For the maximum devia-

tion, the result is 5.45% for 2-aminooctane (T = 293.15 K,

where c (s1) represents the frequency of activated molecule ﬂow

p = 100 MPa). In addition, the inﬂuence of the molar mass on

to vacancy site, the reciprocal of which can be considered the mean

parameters was modeled supposing that the parameters are linear

free residence time as sR = 1/c, a is the proportionality factor be-

function of the molar mass as A = A1M + A0 (with A = a or b or c). As

tween the activation energy DG– and the internal energy of vapor-

shown in Table 14, AAD shows a little bit worse result compared

ization DUvap, VL is the volume of the liquid, and ZL is the

with the previous result (overall 1.29% instead of 0.94%) but it is

compressibility factor of the liquid, respectively. On the other hand,

still within experimental uncertainty. For the maximum deviation,

in order to extend the applicability of Eq. (17) to a wider tempera-

the result is 6.00% for 2-aminooctane (T = 293.15 K, p = 100 MPa).

ture range, Macías-Salinas et al. [14,15] adopted successfully

The overall results of this model shown in Tables 13 and 14 are

power-law dependence between DG– and DUvap and their three-

very well, compared to other viscosity models used in this work.

parameters viscosity model is written as

From the viewpoint of the estimated values of characteristic

" #

RT 1 DU v ap b parameters, taking into account of the results of Lei et al. [57]

g¼ exp a expðZ L Þ: ð18Þ and Macías-Salinas et al. [14,15], there is no clear correlation be-

VL c RT

tween molecular shape and the value of the parameters. In addi-

In this work we use a slightly different equation than Eq. (18). tion, even if it is the same ﬂuid, the values of the parameters are

The viscosity model of the Eyring type for pure liquids is written different depending on the model expression and the type of EoS

as: used to calculate the equilibrium properties. However, as a satis-

factory result was obtained in this work by linearizing the param-

RT 1 DU v ap

g¼ exp a expðbZ L Þ: ð19Þ eters against molar mass, it is possible to consider that the value of

VL c RT parameters reﬂects the difference of the molecular motion that

This equation has still 3 adjustable parameters, but keep the linear originates in the difference of the molecular structure (length of

variation of ln(g) versus DUvap. It is worthwhile to say here that this the carbon chain). Focusing on the parameter c for the four 2-alkyl-

relation has been privately suggested to us by Prof. Macias-Salinas amines used here, the molecule which has the longer carbon chain

et al. [14,15]. The equilibrium properties VL, DUvap = Uvap UL and shows lower value of c, i.e., lower frequency of the activated mol-

ZL in Eq. (19) at given temperature and pressure are computed from ecule displacement to the vacancy site. Notice that when applied to

a cubic PR EoS [51]. linear amines, pentylamine, hexylamine and heptylamine [1] this

The three parameters of Eq. (19) were estimated for the four 2- model gives also a very good representation. The worst case is

alkylamines considered in this work. The correlating results are for heptylamine : AAD = 1.26%, Bias = 0.64% and MD = 4.0%. No par-

summarized in Table 13. The performance of Eq. (19) in correlating ticular effect appears that could clearly be related to the branching

saturated liquid viscosity of associated ﬂuids was remarkably good effect.

Table 13

Results obtained with the model based on Eyring theory and PR EoS.

a 0.26408 0.26399 0.22006 0.22040

b 0.11446 0.10304 0.10160 0.09770

c 1011 (s1) 9.1115 9.2957 6.3074 6.2322

AAD (%) 1.01 0.51 0.56 1.69 0.94

Bias (%) 0.020 0.001 0.010 0.043 0.019

MD (%) 2.69 1.56 2.40 5.45 5.45

Table 14

Results obtained with the model based on Eyring theory and cubic EoS and the correlated parameters (A = A1M + A0 with M in kg mol1).

A1 A0

a 1.5028 0.39482

b 0.21405 0.12609

11 1

c 10 (s ) 115.67 19.5938

2-Aminobutane 2-Aminopentane 2-Aminoheptane 2-Aminooctane Overall

AAD (%) 1.44 1.10 0.740 1.87 1.29

Bias (%) 0.162 0.278 0.324 1.46 0.393

MD (%) 5.16 3.24 1.85 6.00 6.00

136 M. Yoshimura et al. / Chemical Physics 369 (2010) 126–137

4.7. Roland model for viscosity scaling ﬂexibility also increases, and that softens the intermolecular po-

tential, namely, the value of c decreases.

Recently [16], it has been put into evidence the phenomenon of In our previous work [1], where we have estimated the values of

superposition of the relaxation times, s, for various glass-forming c for three 1-alkylamines in the same way, the same tendency of c

liquids and polymers when expressed as a function of TVc (V: spe- to decrease with the increase of carbon atoms has been observed.

ciﬁc volume V = 1/q, and c a constant characteristic of the mate- From the viewpoint of the difference of molecular structure, it is

rial). Roland et al. [17] have extended this thermodynamic worse to compare the value of c among the structural isomers of

scaling to the viscosity of several real ﬂuids. For a given compound 1-alkylamines and 2-alklamines. 2-alkylamines that have branch-

the viscosity g is only a functional of the quantity TVc ing molecular structure show lower scaling factor than 1-alkyl-

amines that have linear molecular structure, namely,

gðT; VÞ ¼ gðT; qÞ ¼ f ðTV c Þ ¼ f ðT qc Þ: ð20Þ pentylamine (c = 8.17) versus 2-aminopentane (c = 7.77) and hep-

tylamine (c = 7.53) versus 2-aminoheptane (c = 7.24). These differ-

According to Refs. [16,17,58] the parameter c reﬂects the mag- ences of c suggest that, in the comparison of molecules that have

nitude of the intermolecular forces. It links the thermodynamic to the same number of carbon atoms, molecules with the branching

the transport property (here viscosity). In Ref. [17], Eq. (20) has structure have a lower intermolecular potential than the ones with

been veriﬁed for several liquids. Eq. (20) has been later veriﬁed the linear structure.

by other authors. In particular in Refs. [59,60] the superposition

is very clearly observed for several pentaerythritol ester lubricants,

for linear and branching alkanes, polar liquids, ionic liquids, and 5. Conclusion

alcohols.

We have estimated the exponent c for the four 2-alkylamines A total of 93 experimental dynamic viscosity measurements are

considered in this work. In order to model the inﬂuence of the mo- reported for 2-aminobutane, 2-aminopentane, 2-aminoheptane

lar mass in this narrow molar mass interval, we made ﬁtting and 2-aminooctane, for temperatures between 293.15 K and

imposing c = c1M + c0. The function f(TVc) in Eq. (20) is not speci- 353.15 K and up to 100 MPa. At atmospheric pressure

ﬁed, however, using polynomial function of (TVc) we obtained a (p = 0.1 MPa) the dynamic viscosity was measured by a classical

very satisfactory scaling result as c = 0.018852 M + 9.411908 capillary viscometer (Ubbelohde) with an experimental uncer-

(M in g mol1). For each 2-alkylamine the viscosity data collapse tainty of 1%, whereas the viscosity under pressure was measured

onto a single master curve. In fact as shown in Fig. 8, there is a sin- with a falling-body viscometer with an experimental uncertainty

gle master curve identical for the four amines (c = 8.03, 7.77, 7.24 of 2%.

and 6.98 for 2-aminobutane, 2-aminopentane, 2-aminoheptane The experimental data for these systems have been used in or-

and 2-aminooctane, respectively). Fig. 8 shows really a very good der to evaluate the performance of one empirical correlation (Vo-

superposition of these four 2-alkylamines, which have branching gel–Fulcher–Tamman representation combined with Tait-like

structure. It is then clear that Roland et al. [16,17] scaling relation equation; seven adjustable parameters for each compound) as well

can be used to determine the volume and temperature dependence as six models with a more or less developed physical and theoret-

of viscosity, over broad temperature and pressure range. Neverthe- ical background: the rough hard-sphere scheme (ﬁve adjustable

less, for some strongly hydrogen-bonded materials (water, low parameters), a viscosity model based on the free-volume concept

molecular weight polypropylene glycol), the superpositioning fails (three adjustable parameters), the f-theory model based on friction

[17]. Despite the small association of these branching amines, the consideration (two adjustable parameters), a correlation derived

representation makes sense as Eq. (20) reproduces the experimen- from molecular dynamics (one adjustable parameter), a model

tal viscosity very well. According to Eq. (20), scaling factor c corre- based on Eyring theory combined with cubic EoS (3 adjustable

sponds to the weight on V = 1/q, the magnitude of the c value parameters) and a viscosity representation based on thermody-

means the magnitude of a relative inﬂuence of the density on the namic scaling (one adjustable parameter). This evaluation shows

viscosity in comparison with the temperature. For linear alkanes, that some simple models can represent the viscosity of these sys-

c decreases (13 for n-hexane and 6.3 for octadecane) with the mo- tems within an acceptable and satisfactory uncertainty. Further, by

lar mass [60]. Our result for 2-alkylamines suggests that, as for al- performing the linearization of characteristic parameters on molar

kanes, when the number of carbon atoms increases the molecular mass, and by comparing with the results of 1-alkylamines obtained

in previous work, the way different models take into account the

effect of molecular structure (carbon chain length and position of

amino group) has been revealed.

The free-volume model, the f-theory model, the molecular

dynamics viscosity model and the Eyring-PR EoS model are all

applicable to gases, liquids, and dense ﬂuids. Because of this, these

four models are suitable for industrial processes involving different

phases or phase changes. Moreover, from a fundamental point of

view, the rough hard-sphere scheme, the free-volume model, the

molecular dynamics model, and Eyring-PR EoS model provide

some insight on the microstructure of these complex systems. Fi-

nally, it should be underlined that the four 2-alkylamines consid-

ered here and also the three 1-alkylamines reported in previous

work [1] show a pressure and temperature viscosity behaviour

compatible with the superposition and scaling viscosity scheme,

providing one single master curve.

References

Fig. 8. Viscosity versus qc/T (g = f(TVc)) (h: 2-aminobutane, }: 2-aminopentane, 4: [1] M. Yoshimura, C. Boned, A. Baylaucq, G. Galliéro, H. Ushiki, J. Chem.

2-aminoheptane and s: 2-aminooctane). Thermodyn. 41 (2009) 291.

M. Yoshimura et al. / Chemical Physics 369 (2010) 126–137 137

[2] M. Yoshimura, A. Baylaucq, J.-P. Bazile, H. Ushiki, C. Boned, J. Chem. Eng. Data [32] L.I. Kioupis, E.J. Maginn, J. Phys. Chem. B 104 (2000) 7774.

54 (2009) 1702. [33] M.J.P. Comuñas, A. Baylaucq, C. Boned, J. Fernández, Int. J. Thermophys. 22

[3] H. Vogel, Phys. Z. Leipzig 22 (1921) 645. (2001) 749.

[4] G.S. Fulcher, J. Am. Ceram. Soc. 8 (1925) 339. [34] E.N. Andrade, Philos. Mag. 17 (1934) 497.

[5] G. Tamman, W. Hesse, Z. Anorg. Allg. Chem. 156 (1926) 245. [35] J. Millat, J.H. Dymond, C.A. Nieto de Castro, Transport Properties of Fluids,

[6] J.H. Dymond, M.A. Awan, Int. J. Thermophys. 10 (1989) 941. Cambridge University Press, UK, 1996.

[7] M.J. Assael, J.H. Dymond, M. Papadaki, P.M. Patterson, Fluid Phase Equilibr. 75 [36] A. Baylaucq, C. Boned, P. Dauge, P. Xans, Phys. Chem. Liq. 37 (1999) 579.

(1992) 245. [37] A. Baylaucq, M. Moha-ouchane, C. Boned, Phys. Chem. Liq. 38 (2000) 353.

[8] A. Allal, C. Boned, A. Baylaucq, Phys. Rev. E 64 (2001) 011203. [38] M.J.P. Comuñas, A. Baylaucq, F. Plantier, C. Boned, J. Fernández, Fluid Phase

[9] A. Allal, M. Moha-Ouchane, C. Boned, Phys. Chem. Liq. 39 (2001) 1. Equilibr. 222–223 (2004) 331.

[10] C. Boned, A. Allal, A. Baylaucq, C.K. Zéberg-Mikkelsen, D. Bessieres, S.E. [39] P. Reghem, A. Baylaucq, M.J.P. Comuñas, J. Fernandez, C. Boned, Fluid Phase

Quiñones-Cisneros, Phys. Rev. E 69 (2004) 031203. Equilibr. 236 (2005) 229.

[11] S.E. Quiñones-Cisneros, C.K. Zeberg-Mikkelsen, E.H. Stenby, Fluid Phase [40] T.H. Chung, M. Ajlan, L.L. Lee, K.E. Starling, Ind. Eng. Chem. Res. 27 (1988) 671.

Equilibr. 169 (2000) 249. [41] S. Chapman, T.G. Cowling, D. Burnett, The Mathematical Theory of Non-

[12] S.E. Quiñones-Cisneros, C.K. Zeberg-Mikkelsen, E.H. Stenby, Fluid Phase Uniform Gases, Cambridge University Press, UK, 1970.

Equilibr. 178 (2001) 1. [42] P.D. Neufeld, A.R. Janzen, R.A. Aziz, J. Chem. Phys. 57 (1972) 1100.

[13] G. Galliéro, C. Boned, A. Baylaucq, Ind. Eng. Chem. Res. 44 (2005) 6963. [43] K.N. Marsh, C.L. Young, D.W. Morton, D. Ambrose, C. Tsonopoulos, J. Chem. Eng.

[14] R. Macías-Salinas, F. García-Sánchez, O. Hernández-Garduza, AIChE J. 49 Data 51 (2006) 305.

(2003) 799. [44] B.E. Poling, J.M. Prausnitz, J.P. O’Connell, The Properties of Gases and Liquids.

[15] G. Cruz-Reyes, G. Luna-Bárcenas, J.F.J. Alvarado, I.C. Sánchez, R. Macías-Salinas, The Properties of Gases and Liquids, McGraw-Hill, New York, USA, 2001.

Ind. Eng. Chem. Res. 44 (2005) 1960. [45] Y. Nannoolal, J. Rarey, D. Ramjugernath, Fluid Phase Equilibr. 252 (2007) 1.

[16] C.M. Roland, S. Hensel-Bielowka, M. Paluch, R. Casalini, Rep. Progr. Phys. 68 [46] Q. Wang, P. Ma, Q. Jia, S. Xia, J. Chem. Eng. Data 53 (2008) 1103.

(2005) 1405. [47] Q. Wang, Q. Jia, P. Ma, J. Chem. Eng. Data 53 (2008) 1877.

[17] C.M. Roland, S. Bair, R. Casalini, J. Chem. Phys. 125 (2006). 124508. [48] Q. Jia, Q. Wang, P. Ma, J. Chem. Eng. Data 53 (2008) 2606.

[18] P. Daugé, A. Baylaucq, L. Marlin, C. Boned, J. Chem. Eng. Data 46 (2001) 823. [49] D. Ambrose, J. Walton, Pure Appl. Chem. 61 (1989) 1395.

[19] C. Zéberg-Mikkelsen, A. Baylaucq, G. Watson, C. Boned, Int. J. Thermophys. 26 [50] S.P. Tan, H. Adidharma, B.F. Towler, M. Radosz, Ind. Eng. Chem. Res. 44 (2005)

(2005) 1289. 8409.

[20] A.S. Pensado, M.J.P. Comuñas, L. Lugo, J. Fernandez, J. Chem. Eng. Data 50 [51] D.-Y. Peng, D.B. Robinson, Ind. Eng. Chem. Fund. 15 (1976) 59.

(2005) 849. [52] G. Soave, Chem. Eng. Sci. 27 (1972) 1197.

[21] M.C.S. Chen, J.A. Lescarboura, G.W. Swift, AIChE J. 14 (1968) 123. [53] C.K. Zéberg-Mikkelsen, A. Baylaucq, M. Barrouhou, S.E. Quiñones-Cisneros, C.

[22] Y.L. Sen, E. Kiran, J. Supercrit. Fluids 3 (1990) 91. Boned, Fluid Phase Equilibr. 222–223 (2004) 135.

[23] E. Kiran, Y.L. Sen, Int. J. Thermophys. 13 (1992) 411. [54] S.P. Tan, H. Adidharma, B.F. Towler, M. Radosz, Ind. Eng. Chem. Res. 45 (2006)

[24] M.J. Assael, H.M.T. Avelino, N.K. Dalaouti, J.M.N.A. Fareleira, K.R. Harris, Int. J. 2116.

Thermophys. 22 (2001) 789. [55] S.E. Quiñones-Cisneros, U.K. Deiters, J. Phys. Chem. B 110 (2006) 12820.

[25] M.L. Huber, A. Laesecke, H.W. Xiang, Fluid Phase Equilibr. 224 (2004) 263. [56] S. Glasstone, K.J. Laidler, H. Eyring, The Theory of Rate Processes: The Kinetics

[26] I. Cibulka, L. Hnědkovský, J. Chem. Eng. Data 41 (1996) 657. of Chemical Reactions Viscosity Diffusion and Electrochemical Phenomena,

[27] H.W. Xiang, A. Laesecke, M.L. Huber, J. Phys. Chem. Ref. Data 35 (2006) 1597. McGraw-Hill, New York, 1941.

[28] P.S. Rao, K.C. Rao, G.N. Swamy, Int. J. Chem. Sci. 4 (2006) 553. [57] Q.-F. Lei, Y.-C. Hou, R.-S. Lin, Fluid Phase Equilibr. 140 (1997) 221.

[29] M.E. Bai, K.G. Neerajakshi, K.S.V.K. Rao, G.N. Swamy, M.C.S. Subha, J. Indian [58] D. Coslovich, C.M. Roland, J. Phys. Chem. B 112 (2008) 1329.

Chem. Soc. 82 (2005) 25. [59] O. Fandino, M.J.P. Comuñas, L. Lugo, E.R. Lopez, J. Fernandez, J. Chem. Eng. Data

[30] M.C.S. Subha, G.N. Swamy, M.E. Bai, K.S.V.K. Rao, Phys. Theor. Anal. Chem. 43 52 (2007) 1429.

(2004) 1876. [60] A.S. Pensado, A.A.H. Pádua, M.J.P. Comuñas, J. Fernández, J. Phys. Chem. B 112

[31] M.A. Saleh, S. Akhtar, A.R. Khan, Phys. Chem. Liq.: Int. J. 39 (2001) 85. (2008) 5563.

- Ft Dow Corning 200 FluidЗагружено:Isabel Martínez Constan
- Chapter_ 2 Hydraulic FluidЗагружено:Pham Thị Thuy Ha
- Norma ASTM D1200Загружено:Sara Quiroz
- Calculating Viscosities of Reservoir Fluids From Their CompositionЗагружено:L Guillermo Rueda
- Exercise 8 Full ReportЗагружено:Princess Requiso
- Gao Bower Msmse04Загружено:aerosubra
- FM & F Machines 2013.pdfЗагружено:Chandan Kumar
- HY30-3400-UK-GP1-GPAЗагружено:Alejandro Cornejo
- ExtractionЗагружено:Ali M. Chehadeh
- K2 B 231116123141Загружено:hilda wiranda
- Wash CoatЗагружено:Bzes The-Sky
- physicsinvestigatoryprojectforcolleges-140119070411-phpapp01Загружено:michiemar
- CondensateStab2013AIChE91aPaper292728Загружено:Saras Unggul Utami
- Squishy MaterialsЗагружено:maria_d_k
- PosterЗагружено:juanmanuel_4615958
- 211430934-Experiment-4Загружено:Pavithran Krishnan
- Tangler t Pai Bul 1987Загружено:Djany Souza
- My judgeЗагружено:Loi Karlo Cruzat
- Lecture 2Загружено:chiang95
- Low-Toxicity Oil Muds a Knowledge of Downhole Rheological Bhavior Assists Successful Field Application - Bailey 1986Загружено:Géssica Palaoro
- Adv Fluid Handout(1)Загружено:Ahmad Idrees
- Cone PlateЗагружено:Lionel Candra
- ASTM D5125 Viscosity CupЗагружено:Muhammad Umer Draz
- Internal Viscosity 2018Загружено:Don MacInnes
- Stokes' law.docxЗагружено:Mayuresh More
- ARTÍCULOЗагружено:LuisJorgeMaldonadoRincón
- wing analysisЗагружено:karthikspeed
- 5-4OS-eqnЗагружено:rickyspaceguy
- Flumec-315.pptxЗагружено:Dainee Mae Cortez
- 10Загружено:Hariom Khungar

- 2opt ModelЗагружено:Miguel Hernandez
- Azeotropic distillationЗагружено:processpipingdesign
- Maranas Et Al, Computers and Chem. Eng., 1996Загружено:Miguel Hernandez
- EES TutorialЗагружено:Kiran Raj
- McCabeThiele - Graphical Design of Fractionating ColumnsЗагружено:asadashfaqlodhi
- J Phys Chem B 2005 RebeloЗагружено:Miguel Hernandez
- brits_vapour_2015.pdfЗагружено:Miguel Hernandez
- wsxtractЗагружено:Miguel Hernandez
- multcompЗагружено:Miguel Hernandez
- Steam TablesЗагружено:ajitsamal456
- programminginexcelvba_anintroductionЗагружено:Ramji Simhadri
- VBAPrimerЗагружено:Miguel Hernandez
- Book ListЗагружено:Miguel Hernandez
- Cuadro Open Tenis Sony Ericson MiamiЗагружено:Miguel Hernandez
- A 0Загружено:Miguel Hernandez

- Application Guide AdsorbentsЗагружено:Renu Sarath Babu
- Thermal MCЗагружено:Dora Aye
- Why Biomass?Загружено:Najib Altawell
- 08_HydroprocessingЗагружено:ThaiyalNayaki
- smc002Загружено:leizar_death64
- Ppisr Newsletter Jan-june 2012Загружено:drraghu123
- Glide User ManualЗагружено:jam head
- Tray Dryer - ExperimentЗагружено:Jade Dhalle Encarnacion
- The Factors Affecting the Performance of Activated CarbonЗагружено:Enrique Franco Ariza
- Curtis Hays Whitson NTH PhD ThesisЗагружено:Luis Alberto Izarra
- InTech-Factors Affecting the Stability of Crude Oil EmulsionsЗагружено:Kun Con
- Cross Section QMЗагружено:bigdrew89
- Alcohols IIT PACEЗагружено:Dushyant Gupta
- DNA–Protein Interactions- Methods for Detection and AnalysisЗагружено:brian8576
- is.15839.2009Загружено:Sohail Shaikh
- docslide.us_experiment-2-rdr.pdfЗагружено:alyssa
- Oxy Miser SeriesЗагружено:Karla Chumbiauca
- ch45Загружено:priya
- Ammonia.pptЗагружено:tamabiet
- Application of the EMEW Cell in Nickel IndustryЗагружено:Rahul_Jeur_957
- 10.1016@j.microc.2016.06.022Загружено:Iwan
- Fired Heater and CombustionЗагружено:Alex Gigena
- MSDS-HF10TQЗагружено:Yogig Tri
- Ch4+The+Periodic+TableЗагружено:v-imsellin
- 21q Chlorination ProcedureЗагружено:agritechregional
- introduction to inviromental control in refining industry.pptЗагружено:Amer Shaier
- Citric Acid and Other Organic CompoundsЗагружено:Keziah Cyra Papas
- Effective Spray Painting Booth Room.pdfЗагружено:Anurag Srivastava
- gummybearlabЗагружено:api-257187977
- ElectriCorr_VpCI-239_Aerosol_MSDS.pdfЗагружено:Jose Alberto Uribe Minier