You are on page 1of 5

Microporous and Mesoporous Materials 103 (2007) 108–112

www.elsevier.com/locate/micromeso

Temperature-dependent thermal conductivity of powdered zeolite NaX


a,b b,c,1 a,b,c,*
Michael B. Jakubinek , Bi-Zeng Zhan , Mary Anne White
a
Department of Physics, Dalhousie University, Halifax, Canada B3H 3J5
b
Institute for Research in Materials, Dalhousie University, Halifax, Canada B3H 3J5
c
Department of Chemistry, Dalhousie University, Halifax, Canada B3H 4J3

Received 2 November 2006; received in revised form 18 January 2007; accepted 22 January 2007
Available online 30 January 2007

Abstract

The thermal conductivity of evacuated packed powders of zeolite NaX have been measured as a function of particle size (2 lm and
800 nm) and temperature (ca. 5–390 K ). The temperature dependence of the thermal conductivity of polycrystalline zeolite NaX over the
particle size range studied here is glass-like and the results show a clear particle size effect. Room temperature values of the effective ther-
mal conductivity of NaX, adjusted to 100% packing fraction, are 0.16 and 0.08 W m1 K1, for the 2 lm and 800 nm particle sizes,
respectively.
Ó 2007 Elsevier Inc. All rights reserved.

Keywords: Zeolite; Thermal conductivity; Particle size effect; Thermal properties

1. Introduction and sodalite [8,9], have been measured and the effects of
temperature [4,5,7–9], filling gas [8,9], pressure [6–9],
Knowledge of the thermal conductivity (j) of zeolites hydration [6,8], and particle size [5,7–9] on j have been
can be important for applications such as catalysis, sorp- explored. In all cases the measured values of j are found
tion vacuum pumps, hydrogen storage, and heat pumps. to be low, less than 0.4 W m1 K1 and in many cases sub-
Zeolites can exhibit the uncommon property of contracting stantially lower. The experimental measurements are also
on heating (i.e., negative thermal expansion) [1], which, for low in comparison with molecular dynamics simulations,
framework materials, has been linked to the occurrence of which have predicted values between 0.6 W m1 K1
unusually low thermal conductivity [2,3]. However, the (NaA [11]) and 4 W m1 K1 (sodalite [12]) at room tem-
thermal conductivity of zeolitic materials can be difficult perature and up to 7 W m1 K1 at 100 K (sodalite [12]).
to quantify because of complications due to effects of Discrepancies between the values obtained in simulations
hydration, pressure and filling gas. Applications often rely and experiments result, at least partially, because the simu-
on powdered zeolites, where the measured thermal conduc- lations are for single crystals and calculate the thermal con-
tivity also depends on the particle size and fraction of void ductivity of the zeolite lattice, whereas experiments on
space. polycrystalline powders measure effective thermal conduc-
Thermal conductivities of several zeolite powders, viz. tivities, which include the lattice thermal conductivity but
NaX [4–7], CaA [5,8], NaA [5,8–10], KA [8,9] NaY [8] are also affected by packing voids, heat transported by
the filling gas, adsorbed water and thermal resistance asso-
ciated with boundaries.
*
Corresponding author. Address: Department of Chemistry, Dalhousie Thermal conductivity values for powdered zeolites are
University, Halifax, Canada B3H 4J3. Tel.: +1 902 494 3894; fax: +1 902 particularly relevant given the prominence of powders in
494 8016.
E-mail address: Mary.Anne.White@Dal.Ca (M.A. White).
applications. In this work, we determined the effective ther-
1
Present address: Chevron Energy and Technology Company, Rich- mal conductivities of powdered zeolite NaX in vacuum as a
mond, CA 94802, USA. function of temperature, including particle size effects.

1387-1811/$ - see front matter Ó 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.micromeso.2007.01.040
M.B. Jakubinek et al. / Microporous and Mesoporous Materials 103 (2007) 108–112 109

These results give definitive information on the lattice ther- Table 1


mal conductivity of the zeolite powders; contributions for Densities of the 4.80 mm diameter, cylindrical samples after pressing, in
comparison to the theoretical densities: 1468 kg m3 for bulk-NaX and
the effects of packing void space and adsorbed gases could 1464 kg m3 for micro- and nano-NaX, calculated from the chemical
be added separately to produce useful thermal models of formulae and unit cell dimensions
zeolites for real situations. Sample Density (kg m3) Percentage of theoretical density
bulk-NaX 1280 ± 50 87 ± 3
2. Experimental methods micro-NaX 1210 ± 60 83 ± 4
nano-NaX 1190 ± 50 81 ± 3
Zeolite NaX (also called zeolite X, Linde X, and molec- nano-NaX(2) 1230 ± 40 84 ± 3
ular sieve 13X), an analogue of the natural zeolite faujasite,
was purchased from Aldrich and calcined in flowing dry air
(4 h, T = 773 K) to remove any organic template residue. drical mold and pressing with an axial load (equivalent to
This sample, with particle sizes of approximately 2 lm as 0.5 GPa) for 5 min in a hydraulic press. The densities of
determined from high-resolution scanning electron micros- samples, as calculated from their masses and dimensions,
copy (SEM, uncoated samples, Hitachi S4700 cold-field are reported in Table 1. The resulting cylinders were
emission SEM) [13], is subsequently referred to as bulk- attached using a 5-min epoxy loaded with a fine, silver
NaX. We have recently developed techniques to produce powder to disk-shaped, gold-coated copper leads (Quan-
zeolites with controlled particle sizes of approximately tum Design, Part No. 4084-609) in a two-wire configura-
800 nm and 20–100 nm [13,14], referred to as micro- and tion and then loaded into the PPMS. Our calculations,
nano-NaX, respectively. Extensive characterizations of where the measured thermal resistance is modeled by a ser-
these samples, including X-ray diffraction, X-ray flores- ies addition of the thermal resistances of the sample and
cence, Fourier transform infrared spectroscopy, solid state epoxy, indicate that the thin layers of epoxy do not signif-
29 icantly affect the measured, effective thermal conductivity
Si NMR spectroscopy, nitrogen adsorption isotherms
and heat capacity measurements, have been published pre- for low thermal conductivity samples over the temperature
viously [13–15]. range reported here. These preparation procedures were
The bulk-, micro-, and nano-NaX samples described conducted while keeping the samples inside a helium atmo-
above had been stored under ambient conditions and were sphere; however the samples were exposed briefly to air
dehydrated in Pyrex tubes under dynamic vacuum at 473 K when loading into the PPMS. The results (see below) illus-
for 24 h and then cooled to room temperature and sealed trate the importance of rigorous inert/dry atmosphere
under argon before use in this study. The effectiveness of preparation when measuring physical properties of zeolites.
a similar dehydration procedure has been verified previ-
ously by thermogravimetric analysis [16]. As the samples 3. Results and discussion
are prone to take in water, they were handled in ultra-
dry helium gas and stored in a vacuum desiccator when The apparent thermal conductivity as a function of time
not in use. was observed to decrease before reaching a stable value,
Thermal conductivity measurements from ca. 2–300 K shown in Fig. 1 for micro-NaX. This finding is attributed
were carried out using a Physical Property Measurement water loss occurring under the high vacuum conditions of
System (PPMS, Quantum Design, San Diego, CA) equipped the experiment, which has been observed for other porous
with Thermal Transport Option (TTO, Quantum Design). materials [19]. This suggests that the samples were incom-
‘‘Single Measurement Mode’’ [17], which employs a pletely dried or had adsorbed moisture when exposed to
steady-state thermal conductivity technique [18], was used ambient air during the loading processes. These observa-
for all the measurements. In this mode a known heating tions were made at 390 K (for the micro-NaX sample) or
power, chosen to achieve a temperature rise of ca. 3%, is 350 K (for bulk- and nano-NaX). The highest temperature
applied and the sample’s thermal conductance, K, is deter- of the system was approximately 390 K, but was reduced
mined directly from this power and the steady-state temper- to 350 K after samples were found to have lost contact with
ature difference across the sample, DT1. Measurements the leads, possibly due to softening of the epoxy at higher
were conducted under vacuum (<105 kPa), rendering con- temperatures. Reported thermal conductivity measure-
vective heat transfer and gas conduction negligible. A radi- ments as a function of temperature were initiated only after
ation shield was employed to reduce radiative heat loss a stable value was achieved (Fig. 1); agreement between
and the PPMS software also corrects for radiative loss results at the same temperature before and after the tem-
[17], which, in this work, was no more than 14% (at 390 K perature-dependent measurements indicates that neither
for micro-NaX; less than 5% for micro-NaX below 200 K further water loss nor sample cracking or loss of contact
and for bulk-NaX below 300 K). with the leads occurred during the measurements reported
The TTO is designed to measure consolidated materials here. The apparent thermal conductivity attained a con-
so it was necessary to press the samples into short cylin- stant value in a reasonable timescale for the bulk- and
ders. Pressed samples 4.80 mm in diameter were produced micro-NaX samples but not for nano-NaX; it was still
by loading 30–40 mg of the powdered samples into a cylin- dropping at a rate of 4% in the 20 h following 200 h under
110 M.B. Jakubinek et al. / Microporous and Mesoporous Materials 103 (2007) 108–112

0.14 Evaluation of the measured data should consider that


the powder packing is less than 100% efficient. The packing
fraction, f, of each sample, which includes only packing
0.13
voids and not the intrinsic porosity of zeolite crystals,
was calculated from
κ / Wm-1 K-1

0.12 qtheory  qsample


f ¼1 ; ð1Þ
qtheory
0.11
Start where qsample is the density of the sample, determined from
Finish
its mass and dimensions, and qtheory is the theoretical den-
0.10 sity calculated from the chemical formula and cubic unit
cell dimension (24.7701 Å) [15]. The packing fractions of
0.09
all the samples were equal within the limits of the uncer-
0 10 20 30 40 50 60 70 80 tainties (f = 0.87 ± 0.04, 0.83 ± 0.05, 0.81 ± 0.04, and
time / hours 0.84 ± 0.04 for bulk-NaX, micro-NaX, and two nano-
NaX samples, respectively) and therefore, relative compar-
Fig. 1. Apparent thermal conductivity as a function of time under high isons of the measured thermal conductivities can be
vacuum for micro-NaX at 390 K. Temperature-dependent thermal
conductivity measurements were started only after a stable value was
made. There is a clear reduction in thermal conductivity
achieved as indicated by the ‘‘Start’’ and ‘‘Finish’’ arrows. for micro-NaX in comparison to bulk-NaX. While results
for nano-NaX are not reported here, because of the diffi-
culty in achieving stable values, its thermal conductivity
high vacuum (in comparison to ca. 0.5% over 40 h for the reached values equal to that of micro-NaX at 350 K and,
bulk- and micro-NaX samples). The long time required to given that it was still decreasing at an appreciable rate over
reach stability makes it impractical to accurately measure time, this can be considered to be an upper limit such that
the thermal conductivity of nano-NaX by this technique j(nano-NaX)  j(micro-NaX) at 350 K.
and more rigorous inert/dry atmosphere sample handling Assuming that the voids are randomly distributed and
likely is required. the void-fraction is not so large that it forms a continuous
The effective thermal conductivity results for bulk- and minority phase, theory developed by Klemens [20] can be
micro-NaX are given in Table 2. Although, in principle, applied to determine the thermal conductivity of an equiv-
the PPMS can be used to obtain thermal conductivity mea- alent NaX sample with no voids (i.e., including only the
surements down to ca. 2 K, even the minimum possible lattice thermal conductivity as present in a theoretical bulk
heater power resulted in too large a temperature difference sample and any particle size effects) from the measured
between the hot and cold thermometers for these low ther- result according to
mal conductivity samples, allowing reliable measurements
only for T > ca. 5 K. jmeasured 4
¼ 1  ð1  f Þ; ð2Þ
jNaX 3
Table 2 where Eq. (2) is valid for voids of both isotropic and aniso-
Measured, effective thermal conductivity for bulk-NaX and micro-NaX tropic shape, provided that they are randomly oriented and
bulk-NaX micro-NaX distributed. Thermal conductivities for 100% packing frac-
T (K) j (W m 1 1
K ) T (K) j (W m1 K1) tion, using Eq. (2), are displayed in Fig. 2. The uncertainty
6.02 0.0083 4.08 0.0025 in jNaX was calculated from the uncertainties in jmeasured
9.99 0.015 5.95 0.0047 and f by standard propagation of error methods.
20.17 0.026 7.99 0.0036 The thermal conductivity results from 50 K to 250 K are
30.34 0.034 9.96 0.0047 within the range of the results reported for zeolite 4 A pow-
40.74 0.040 19.91 0.0092
der under vacuum [10]. Although our results are high in
60.39 0.051 29.91 0.013
80.52 0.060 39.99 0.016 comparison with lower packing fraction results for NaX
100.27 0.068 60.00 0.022 at comparable pressures [7] and a selection of other zeolites
150.29 0.085 80.02 0.026 at 0.10 kPa [8,9], the main difference undoubtedly is due to
201.02 0.099 100.29 0.030 much lower packing fractions (f  0.3–0.5) in the earlier
251.38 0.11 150.44 0.038
studies. The results of Griesinger et al. [8], when used to
301.87 0.13 200.42 0.044
352.09 0.15 250.59 0.053 calculate the effective, fully dense thermal conductivity
391.41 0.17 302.72 0.062 via Eq. (2), are in good agreement with the present results.
352.48 0.073 Furthermore, the results of Mori et al. [7], which used lar-
391.42 0.095 ger NaX particles (1 mm to 10 lm), show the same trend of
Our measurements of other low thermal conductivity materials indicate decreasing j with decreasing particle size as observed here
uncertainties for this technique of ca. ±20% [2,3,19]. for NaX samples with particle sizes of 2 lm and 0.8 lm.
M.B. Jakubinek et al. / Microporous and Mesoporous Materials 103 (2007) 108–112 111

0.24 in j below room temperature (near one-tenth of the mate-


0.06
rials Debye temperature) and dj/dT < 0 at room tempera-
0.20 0.04 ture [21]. It is important to note that the calculated mean
free paths are effective values and not quantitatively useful
κNaX / Wm-1 K-1

0.02
0.16 because of assumptions involved in the Debye model,
0.00 which lead to large underestimates. However, the model
0 20 40
0.12 does provide useful information about the temperature
dependence and also about the relative magnitudes of the
0.08 mean free paths of the samples studied here.
The only other low-temperature investigation of zeolite
0.04 thermal conductivity did not show clearly whether dj/dT
was positive or negative [10]. However, the glass-like
0.00 dependence (i.e. dj/dT > 0) observed here is consistent
0 50 100 150 200 250 300 350 400 with other observations on zeolite powders at higher tem-
T/K peratures (210–570 K [8] and 303–440 K [4]). This is in con-
Fig. 2. Temperature-dependent effective thermal conductivity of bulk- trast to molecular dynamics simulations, which have
NaX (h) and micro-NaX () for 100% packing fraction. predicted higher thermal conductivities and crystal-like
temperature dependences for several zeolites [11,12] includ-
ing silaceous faujasite [12]. The temperature dependences
The thermal conductivity results allow an assessment of
of j and ‘mfp indicate that the mean free path is reduced
the effective phonon mean free paths (‘mfp) from the Debye
so much that heat transport now occurs by mechanisms
model:
similar to those in amorphous, or glass-like, materials.
1 Shortening of the phonon mean free path in a zeolite struc-
j ¼ Cv‘mfp ; ð3Þ
3 tures likely is due both to mechanisms present in a single
crystal, such as distortions of the SiO4 tetrahedra caused
where C is the heat capacity per unit volume and v is the by complex unit cells [8] or framework cations changing
phonon velocity. The results (Fig. 3) clearly exhibit glass- the vibrational modes by locally disturbing the oxygen sub-
like temperature dependences, with ‘mfp longer at very lattice [11], and factors present only for polycrystalline
low-temperatures but becoming short and temperature- powders. While the effect of particle size is not presently
independent on warming to room temperature [21]. This understood, the decrease in thermal conductivity from
short ‘mfp leads to low values of j and the temperature bulk- to micro-NaX indicates that particle size effects play
dependence of ‘mfp leads to dj/dT > 0, in contrast with a prominent role. It would be very interesting to measure
typical (simpler) crystalline materials which exhibit peaks a single crystal sample but due to the difficulty of produc-
ing zeolite single crystals of sufficient size, advances in the
0.7 understanding of the thermal conductivity of zeolites will
likely have to rely on an interplay between measurements
0.6 on powders and simulations.

0.5 4. Conclusions
0.4
lmfp / nm

The measurements reported here show that the thermal


0.3
conductivity of highly packed, powdered polycrystalline
zeolite NaX is glass-like, which contrasts to the results of
0.2 existing simulations that predict crystal-like thermal con-
ductivity. In addition, the results show that particle size
0.1 influences the thermal conductivity, despite its negligible
effect on the heat capacity of the same samples [15].
0.0
0 50 100 150 200 250 300 Acknowledgments
T/ K
Fig. 3. Effective phonon mean free path, ‘mfp, as a function of temper- The authors thank Vladimir Murashov for translating a
ature for bulk- (h) and micro-NaX () with 100% packing fraction. The table from Russian to English and Patricia Laws for help-
‘mfp values were calculated from the Debye model, Eq. (3), using literature ful discussions. Financial support from NSERC (through
heat capacity data for these same samples determined between ca. 40 and
grants to M.A.W. and a postgraduate scholarship to
300 K [15]. As there are relatively few data available on the elastic
properties of zeolites and none for NaX, phonon velocities were estimated M.B.J.), the Killam Trusts (through a research professor-
from the Debye temperatures for these materials, as reported by Qiu et al. ship to M.A.W., postdoctoral fellowship to B.Z.Z. and pre-
[15]. doctoral scholarship to M.B.J.), and the Canada
112 M.B. Jakubinek et al. / Microporous and Mesoporous Materials 103 (2007) 108–112

Foundation for Innovation, Atlantic Innovation Fund, [9] K. Spindler, A. Griesinger, E. Hahne, High Temp.–High Press. 29
and other partners which fund the Facilities for Materials (1997) 45.
[10] V.V. Murashov, M.A. White, Mater. Chem. Phys. 75 (2002) 178.
Characterization managed by the Institute for Research [11] V.V. Murashov, J. Phys.: Condens. Mat. 11 (1999) 1261.
in Materials, is gratefully acknowledged. [12] A.J.H. McGaughey, M. Kaviany, Int. J. Heat Mass Transf. 47 (2004)
1799.
[13] B.-Z. Zhan, M.A. White, K.N. Robertson, T.S. Cameron, M.
References Gharghouri, Chem. Commun. (2001) 1176.
[14] B.-Z. Zhan, M.A. White, M. Lumsden, J. Mueller-Neuhaus, K.N.
[1] P. Lightfoot, D.A. Woodcock, M.J. Maple, L.A. Villaescusa, P.A. Robertson, T.S. Cameron, M. Gharghouri, Chem. Mater. 14 (2002)
Wright, J. Mater. Chem. 11 (2001) 212. 3636.
[2] C.A. Kennedy, M.A. White, Solid State Commun. 134 (2005) 271. [15] L. Qiu, P.A. Laws, B.-Z. Zhan, M.A. White, Can. J. Chem. 84 (2006)
[3] C.A. Kennedy, PhD thesis, Department of Chemistry, Dalhousie 134.
University, Halifax, Canada, 2005. [16] P. Laws, MSc thesis, Department of Chemistry, Dalhousie Univer-
[4] R.K. Vyas, S. Kumar, Ind. Eng. Chem. Res. 34 (1995) 4058. sity, Halifax, Canada, 2002.
[5] S.Z. Vasil’ev, V.I. Letichevskii, V.L. Mal’ter, M.Ya. Solntsev, G.M. [17] Physical Properties Measurement System: Thermal Transport Option
Yusova, O.M. Kostenok, N.I. Sokolova, Khimicheskoe I Neftyanoe User’s Manual, Quantum Design, San Diego, CA, 2002.
Mashinostroenie (1979) 16. [18] R. Berman, Thermal Conduction in Solids, Oxford University Press,
[6] H. Sahnoune, PH. Grenier, Chem. Eng. J. 40 (1989) 45. Oxford, 1976, pp. 3–5.
[7] H. Mori, Y. Hamamoto, S. Yoshida, Nippon Reito Kucho Gakkai [19] M.B. Jakubinek, C. Samarasekera, M.A. White, J. Mater. Res. 21
Ronbunshu 17 (2000) 171. (2006) 287.
[8] A. Griesinger, K. Spindler, E. Hahne, Int. J. Heat Mass Transf. 42 [20] P.G. Klemens, High Temp.–High Press. 23 (1991) 241.
(1999) 4363. [21] D.G. Cahill, R.O. Pohl, Annu. Rev. Phys. Chem. 39 (1988) 93.

You might also like