Received: 7 April 2019 Revised: 10 August 2019 Accepted: 12 August 2019

DOI: 10.1002/cjce.23632

MINI-REVIEW

Experimental methods in chemical engineering: specific surface

area and pore size distribution measurements—BET,
BJH, and DFT

Raoof Bardestani1 | Gregory S. Patience2 | Serge Kaliaguine1

1
Department of Chemical Engineering,
Laval University, Québec, Québec, Canada Abstract
2
Department of Chemical Engineering, Gas physisorption is an experimental technique based on equilibrium Van der Waals
Polytechnique Montréal, Montréal, Québec, interactions between gas molecules and solid particles, that quantifies the specific sur-
Canada
face area (SSA), pore size distribution (PSD), and pore volume of solids and powders.
Correspondence The performance of catalysts, absorbents, chromatography column materials, and poly-
Serge Kaliaguine, Department of Chemical mer resins depends on these morphological properties. Here we introduce the basic
Engineering, Laval University, 1065 Avenue
principles and procedures of physical adsorption, especially nitrogen physisorption, as
de la Medecine, Québec, Québec, G1V 0A6,
Canada. a guide to students and researchers unfamiliar with the field. The Brunauer-Emmett-
Email: [email protected] Teller theory (BET) is a common approach to estimate SSA that extends the Langmuir
monolayer molecular adsorption model to multilayer layers. It relies on an equilibrium
adsorption isotherm, measured at the normal boiling point of the adsorbate, eg, 77 K
or 87 K for N2 and Ar, respectively. Web of Science indexed 45 400 articles in 2016
and 2017 that mentioned N2 adsorption porosimetry—BET and BJH (Barrett-Joyner-
Halenda) keywords. The VOSViewer bibliometric tool grouped these articles into
four research clusters: adsorption, activated carbon in aqueous solutions for removal
of heavy metal ions; synthesis of nanoparticles and composites; catalysts perfor-
mance in oxidation and reduction processes; and photocatalytic degradation with
TiO2. According to the literature, the accuracy of the density function theory (DFT)
method is higher than with the BJH theory and it is more reliable.

KEYWORDS
BET surface area, BJH, DFT, pore-size distribution, specific surface area

1 | INTRODUCTION The Canadian Journal of Chemical Engineering published

these articles that are distinct from typical research papers
This paper is part of a series entitled “Experimental Methods and correspond to a tutorial review. Indeed this article, like
in Chemical Engineering”. As discussed in the introductory any other in the series, is not intended for established
paper,[1] the objective is to provide chemical engineering researchers in the field of BET measurements. We first
students and researchers unfamiliar with the method an over- describe the physisorption instrument and the laboratory pro-
view of the principles and experimental approaches. Often cedures to measure SSA and PSD. We then present the basis
these individuals need to master several disciplines in the of the theories and briefly highlight recent applications.
course of their research and cannot afford a systematic in- Finally, we mention the uncertainties and caution researchers
depth review of the literature dealing with all of the analyti- to report significant figures based on multiple measurements
cal methods involved in their work. It is in this spirit that rather than instrument precision.

Can J Chem Eng. 2019;97:2781–2791. wileyonlinelibrary.com/journal/cjce © 2019 Canadian Society for Chemical Engineering 2781
2782 BARDESTANI ET AL.

Specific surface area, SSA, is undoubtedly the most step in the process is to calibrate a quartz cell used for gas
meaningful morphological characteristic of solid substances adsorption: this cell is evacuated under high vacuum (<Pa) at
in applications related to porous structures, such as industrial 300 C for 2 hours (degassing), followed by a complete adsorp-
adsorbents, catalysts, pigments, cement, and polymers.[2,3] tion cycle in the empty cell at 77 K. The cell is then loaded with
Pores for which dpore<2 nm are classified as micropores, 50-150 mg of sample, and connected to the degassing station.
while mesopores lie between 2 to 50 nm, and macropores Degassing the sample is to remove physically adsorbed water
exceed 50 nm in diameter.[4] Micropores are further sub- and volatiles that accumulated during storage. For thermally
divided into ultra micropores (dpore<0.7 nm), medium-sized sensitive solids, mild degassing temperatures are used, and for
micropores (0.7 nm<dpore<0.9 nm), and super-micropores mechanically fragile porous materials, a stream of inert gas is
(dpore>0.9 nm).[5,6] introduced instead of vacuum, using the adsorbate path (ADS)
Gas physisorption equilibrium isotherms of nitrogen, connected to the degassing station (Figure 1). After degassing,
argon, or carbon dioxide adsorbates are the basis of the most the cell is transferred to the adsorption station and evacuated.
reliable methods to measure porous material SSA and pore The adsorption cycle is performed by incremental introduction
size distribution, as well as average pore diameter and vol- of adsorbate gas using the calibration volume (CAL) connected
ume.[7] At temperatures below 100 K, Van der Waals forces to the adsorption station. The difference between the measured
govern the interaction between gas molecules and surfaces pressure and that of the empty cell correlates with the number
with heat of adsorption lower than 4 kJ
mol−1. These forces of moles of adsorbed nitrogen. Nitrogen is the preferred gas
lead to the reversible physisorption of gas molecules on mate- adsorbate because it is inert, available in high purity, inexpen-
rial surface in multiple layers. Chemisorption, on the other sive, and interacts with most solids. During measurements, the
hand, is a different process, occurring with higher heat of cell is maintained at liquid nitrogen boiling point using a Dewar
sorption in a monolayer, irreversibly.[8,9] Brunauer et al[10] flask filled with liquid nitrogen.
extended the Langmuir theory of monolayer adsorption to Nitrogen physisorption isotherms are categorized into six
include multiple layers for which only adjacent layers of gas types (Figure 2).[15] Materials such as zeolites and some
molecules interact to calculate surface area, SBET. types of activated carbon with mainly narrow micropores,
Barrett et al[11] introduced the Kelvin equation to calcu- display type I or Langmuir isotherms, where the pores fill at
late pore size distribution, the Barrett-Joyner-Halenda (BJH) very low relative pressure with a steep uptake, because of
method. Seaton et al[12] were the first to propose the density considerable adsorbent-adsorbate interactions.[16] The lim-
functional theory (DFT) coupled with Monte Carlo molecu- ited amount of adsorbed gas is governed by the accessible
lar simulations to more precisely calculate pore size distribu- micropores volume instead of the actual internal surfaces.
tion from adsorption isotherms. For materials with highly Non-porous materials and those with mainly macropores
disordered slit-like micropores and spheres, non-local den- yield isotherms of type II (reversible isotherms), where N2
sity functional theory (NLDFT) apply best.[13] To determine molecules are absorbed in mono/multi-layers without restric-
pore size distribution, the BJH and DFT methods are com- tion. The transition point from monolayer coverage to multi-
plemented by porosimetry methods on the large size range. layer begins at point B in Figure 2. Type IV isotherms arise
Low pressure mercury porosimetry applies to macropores from solids with micro- and mesopores, where the interac-
(14 μm-200 μm) while high pressure Hg porosimetry mea- tions between gas molecules and adsorbent mesopore sur-
sures pore diameters as low as 3 nm (mesopores).[14] face lead to capillary condensation. Type VI corresponds to
The two major components of the N2 physisorption analyzer stepwise multilayer adsorption on a uniform non-porous sur-
are the degassing and adsorption stations (Figure 1). The first face. Adsorption of Ar or Kr on graphitized carbon black at

To vacuum To adsorption

Cell holder VAC CAL

ADS
V3
BET cell P

ADS V1 V2 V4 V6 V7

V5
Clamp VENT
F I G U R E 1 Gas physisorption
Heating Sample apparatus to obtain adsorption-desorption
Dewar
mantle N2(liq) isotherm (vacuum (VAC), calibration
(CAL), adsorbate (ADS), pressure gauge
Degassing Adsorption (P), V1-V7 valves)
BARDESTANI ET AL. 2783

FIGURE 2 Nitrogen adsorption-

desorption isotherms[15] I II

III IV
Amount adsorbed (cm3 g–1 )

V VI

Relative pressure (P/P0 )

liquid nitrogen temperature mostly leads to this isotherm condensation and adsorption rates in narrow necks versus the
type. The initial zero slope of types III and V at low relative wider bodies—ink bottle pores.[18] Loops of type H3 are
pressure reflects weak adsorbate-adsorbent interactions, found in solids consisting of aggregated non-rigid plate-like
comparable to the adsorbate-adsorbate interactions. Nitrogen particles, eg, some clays.[19] Type H4, on the other hand, is
adsorption on polyethylene is an example of this case. This characteristic of microporous substances with type I iso-
type of isotherms does not allow the estimation of the solids therms, with narrow slit-like pores. For some microporous
monolayer capacity and thus SSA.[15,16] materials, the desorption curve lies above the adsorption curve
In the case of capillary condensation occurring in meso- at low relative pressure (dashed lines in Figure 3). This phe-
pores, the desorption path will be different from that of nomena may be due to swelling of non-rigid micropores, irre-
adsorption, resulting in the formation of a hysteresis loop versible adsorption of adsorbate molecules (as long as the
(Figure 3).[15] The type of loop is mostly associated with the pore aperture is unaffected), or chemical interactions.
pore shape.[17] Materials with uniform distribution of spheri-
cal and cylindrical narrow mesopores exhibit H1 type of hys- 2 | THEORY
teresis loop. Type H2 hysteresis is common for inorganic
oxides with a complex network of interconnected narrow The BET theory assumes that the adsorption energy is inde-
pores. An oversimplified theory attributes it to difference in pendent of the adsorption sites and gas molecules interact
2784 BARDESTANI ET AL.

FIGURE 3 Types of hysteresis

H1 H2 loops obtained by nitrogen
physisorption[15]
Amount adsorbed (cm3 g–1 )

H3 H4

Relative pressure (P/P0 )

only in the vertical direction, lateral interactions between

adjacent adsorbed molecules being negligible[15,20,21]: Phy-
sisorption of the first adsorbate layer is as follows:

x 1 C −1
= + x ð1Þ
W ð1 −xÞ C × W ml C × W ml

where W is the mass adsorbed at relative vapour pressure;

x = P/P0 (P and P0 are the actual and saturated vapour pres-
sures of adsorbate); and Wml is the required mass of adsor-
bate forming a complete monolayer adsorbed on a given
FIGURE 4 Example of BET plot to derive β0 and β1
sample. C is a constant, expressing the differences in the
(Equations (1) and (4))[24]
heat of adsorption of the first and second or higher layers
and depends on temperature and on the heat of adsorption of
the first layer E1. For higher layers EL equals the latent heat
of condensation[22]: physisorption. Some researchers, however, suggest the upper

 limit P/P0 ratio at 0.1 for materials like graphitized carbon.[23]
E 1 −E L According to Equation (1), a plot of x/W(1−x)—y versus
C = exp ð2Þ
RT x gives a straight line with an intercept at β0 = C ×1W ml and a
slope of β1 = (C − 1)/(C × Wml).[3] For a mildly air oxidized
For type III and V isotherms, C = 1. Sing[23] rec-
biochar, for example, the slope equals 53 g−1 and the intercept
ommended the derivation of BET monolayer capacity, Wml,
is 0.48 g−1 (Figure 4).[24] The mass of gas that forms a mono-
from the best linear fit of the adsorption branch including
layer, Wml, is obtained from the sum of the intercept and slope:
point B. The BET theory can be applied over a relative pres-
sure (P/P0) from 0.05-0.35 in the adsorption branch of iso- 1
therm, where monolayer adsorption takes place. The linear = β0 + β1 ð3Þ
W ml
region of the BET plot (Figure 4), depends on the type of
adsorbate and adsorbent, as well as the temperature of The specific surface area, SSA, is as follows:
BARDESTANI ET AL. 2785

W ml sizes in the kernel, respectively. The non-local density func-

SSA = ×N ×A ð4Þ
M ×m tional theory (NLDFT) reliably predicts pore sizes of
ordered mesoporous molecular sieves (silicas) given suitable
where M is nitrogen molar mass; m it the sample mass; A the parameters for adsorbate-adsorbent interactions.[29–31] Phy-
nitrogen molecular cross sectional area (0.162 nm2); and sisorption analysis softwares, such as Autosorb-1, solve the
N is Avogadro’s number.[3] set of equations by non-negative least squares.[32] For carbo-
For the oxidized biochar with SSA = 65 m2
g−1, the naceous materials, NLDFT assumes slit-shaped pores with-
mass of the monolayer,Wml, equals 1.68 × 10−8 mg for a out network connections yet with ideal graphitic walls.[33]
46 mg sample. Quenched solid density functional theory, QSDFT, repre-
The Barrett, Joyner, and Halenda method, BJH, and den- sents roughness and the pore wall surface heterogeneity with
sity functional theory, DFT, allow estimating pore size dis- a single parameter.[29,30] DFT is considered a more realistic
tribution, PSD, based on the physisorption equilibrium and adaptable calculation approach of pore size distribution.
isotherms. The BJH theory implements two fundamental This is, for instance, evidenced by the results in Table 1,
assumptions; first the shape of pore is cylindrical, and sec- from the differences between average pore size diameter
ond the adsorbed amount results from both physical adsorp- determined by the BJH and DFT methods for mesoporous
tion on the pore walls and capillary condensation in silica after template removal, either by air calcination or
mesopores.[11] BJH, thus, considers the radius of pore as the ozone treatment.[31] For samples OTMS, BJH results are
sum of the multilayer thickness (t) and the meniscus radius close to DFT ones, while for samples CTAC and SBA-15
obtained from the Kelvin equation (Equation (5)): pore diameters estimated by DFT are almost 1 nm and 2 nm
higher, respectively.
P 2γV M
ln = ð5Þ
P0 rRT
3 | APPLICATIONS
where P/P0 is the relative pressure in equilibrium with a
In 2016 and 2017, 70 articles in Can. J. Chem. Eng. men-
meniscus; γ is the surface tension of the adsorbate in the liq-
uid form; VM is the molar volume of the liquid; R is the uni- tioned N2 adsorption, BET, and BJH-DFT, which ranks this
versal gas constant; r is the radius of the meniscus formed in topic 3rd among all experimental techniques. It was 4th
the mesopore; and T is temperature. BJH calculates the (45 400 occurrences) over the same period in the WoS
change in the thickness of adsorbed film from the decrease behind x-ray diffraction, scanning electron microscopy, and
of relative pressure in the desorption branch. Each decre- transmission electron microscopy.[1] Of the top 10 000 cited
ment is considered to result from evacuation of the largest articles, RSC Advances published 481 of them, followed by
pores from the capillary condensate, as well as a reduction Desalination and Water Treatment (219), Chemical Engi-
in thickness of the physically adsorbed layer. This theory neering Journal (185), Applied Surface Science (167), Inter-
yields a particular equation correlating pore volume and national Journal of Hydrogen Energy (151), Applied
radius, which is solved by numerical iteration.[11] Owing to Catalysis B-Environmental (125), Ceramics International
the restrictive hypotheses of BJH, this method fails to (123), and Journal of Environmental Chemical Engineering
describe both the micropore diameter and the narrow meso-
pores.[25] Groen et al,[26] for instance, associated this failure
TABLE 1 Examples of mesostructured silica, freed of template
with different physical effects during practical adsorption, either by ozone treatment (-oz) or air calcination (-cal) (adapted from
eg, tensile strength and pore network effects, which lead to Joshi et al[31])
incorrect both micro and mesopores size estimates from the
BJH pore DFT pore Total pore
desorption branch.
BET area size size volume
The International Standard Organization (ISO)[27] recom-
Sample (m2
g−1) (nm) (nm) (cm3
g−1)
mends DFT to calculate pore size distribution. The theory
ms-OTMSa-oz 945 2.8 3.0 0.73
relies on a theoretical description of the experimental
adsorption isotherm Nexp(P/P0) Equation (6)[28]: ms-OTMS-cal 926 3.0 3.2 0.77
ms-CTACb-oz 955 2.9 3.9 0.91
ð DM ms-CTAC-cal 1136 2.9 3.7 1.08
N exp ðP=P0 Þ = N DFT ðP=P0 , DÞ × f ðDÞdD ð6Þ
Dm
SBA-15-oz 711 10.0 12.0 1.16
SBA-15-cal 749 8.6 11.0 1.15
where f(D) is an unknown pore diameter distribution func- a
Mesoporous silica synthesized with octadecyltrimethoxysilane.
tion; and Dm and DM are the minimum and maximum pore b
Mesoporous silica synthesized with cetyltrimethylammonium chloride.
2786 BARDESTANI ET AL.

(122). The most prolific science category was physical Specific surface area plays a primary role in the adsorp-
chemistry with 2015 occurrences. Chemical engineering has tion of chemicals and ions from either aqueous, or gaseous
the second largest number of articles with 1744, followed by solutions, as well as in nanoparticle applications.[34] The
multidisciplinary chemistry (1493), then multidisciplinary blue cluster (Figure 5) groups keywords related to water
materials science (1493), environmental engineering (715), remediation, mostly including adsorption, activated carbon,
and energy and fuels (713). removal, wastewater, aqueous solutions, equilibrium, kinet-
VOSviewer generated a bibliometric map of keywords from ics, RSM (response surface methodology), heavy metal ions,
the 10 000 most cited articles in 2016 and 2017 (Figure 5). and methylene blue.[35]
The size of the circles is associated with the frequency of key- The nanoparticle cluster (green) is strongly connected to
words in the articles corresponding to SSA and pore size distri- adsorption, which implies that researchers are seeking sim-
bution, while the colours classify the subjects with similar ple methods to synthesize particles like carbon nanofibers
concepts. VOSviewer identified four main domains associated (CNFs). Wang et al[36] compared different methods to pro-
with SSA including water treatment and adsorption in blue, duce CNFs and mentioned template etching which yielded
nanoparticles and composites in green, photocatalysis reactions SSA = 970 m2
g−1. This was reported as a more efficient
in chartreuse, and catalysts in red. The most frequent keyword approach in CNFs synthesis than the conventional ones
is adsorption with 1795 occurrences (blue cluster), followed by such as catalytic growth, electrospinning activation, and
nanoparticles (1248, green cluster), and aqueous solution electrospinning PAN/pitch.[37–39] When surface area is the
(1077, blue cluster). The smallest circles (selective oxidation only targeted characterization of an adsorbent, researchers
and porosity) correspond to 107 occurrences. prefer carbonaceous materials such as biochar rather than

FIGURE 5 Bibliometric map of keywords created with VOSviewer software based on the 10 000 most cited articles in 2016 and 2017[54,55]
BARDESTANI ET AL. 2787

CNFs, because they are easier to synthesize and modify.[40] to a considerable accumulation of these acid molecules on the
Biochars are applied for water remediation to remove heavy external surface of SAC-13. Each type of nanoporous catalyst,
metal cations and chemicals. however, has special advantages for a targeted application.
Given its production conditions, biochar must be modi- Zeolites are thermally stable though relatively low in activity
fied using chemical or physical activation, in order to because of their microporous structure; carbons have high
increase its BET surface area. Chu et al[41] investigated thermal stability and surface area, yet low wettability. New
chemical activation using H3PO4. They reported that mixing catalysts could, however, be synthesized with the aim of com-
biochar with H3PO4, followed by carbonization at 350 C, bining required features. For instance, as discussed above, the
increased SSA from 9 m2
g−1 to 795 m2
g−1. They introduction of mesopores into zeolites removes some of their
obtained a higher surface area by first mixing raw pine saw- diffusional limitations in reactions with bulky molecules. Oxi-
dust with H3PO4, followed by pyrolysis at 350 C, resulting dation of carbonaceous materials increases oxygenated surface
in SSA of 1150 m2
g−1. The purpose of modifying the functional groups, thereby improving the surface wettability.
surface was to adsorb carbamazepine and bisphenol from Song et al[49] investigated the effect of thermal carboniza-
industrial wastewater. Bardestani and Kaliaguine[24] instead tion and chemical activation on walnut shell in catalytic
applied steam activation to increase biochar surface area, hydrolysis of CS2 and COS from tail gas to sulphur. The
which could be easily implemented in line with the pyrolysis samples with the highest SSA also had the highest sulfur
process (post-pyrolysis treatment). They used a pyrolysis retention capacity SSA>1000 m2
g−1. In a similar vein, a
biochar obtained from a mixture of spruce, fir, and pine Fe-Cu-KOH catalyst supported on walnut shell activated car-
shredded wood pyrolyzed at 475 C, and increased SSA of bon prepared by a sol-gel method was more active when the
the sample from 50 m2
g−1 to 1025 m2
g−1, by steam acti- surface area was higher, regardless of acidic site density or
vation at 900 C for 1 hour. The same biochar materials were thermal stability.[50] Acetaminophen was oxidized by ozone
also submitted to mild air oxidation at 200 C, yielding much over a high surface area modified MgO at a higher rate com-
higher surface density of oxygenated functional groups. This pared to a non-modified MgO.[51] Xu et al[52] reported that
made the biochar interesting for cation adsorption in spite of plasma-engraved Co3O4 had two times higher SSA than the
a modest increase in SSA.[42] Qian et al[43] reported that sur- pristine form, resulting in 10 times higher activity in oxygen
face area and micropores of biochar contribute significantly evolution reaction. Afshar-Taromi and Kaliaguine[18] investi-
to the sorption of organic contaminants from water. However, gated the effect of calcination time and temperature, along
the size of these pores is also important for this removal, since with that of the mass ratio of Pluronic123/Al(O-i-Pr)3, to
targeted chemicals are found in different molecular sizes.[44] identify optimal conditions to synthesize mesostructured
More keywords are associated with the red cluster than any γ-alumina as a catalyst support. They reported that a calcina-
other and it is centered around catalysis. Patience and tion temperature of 700 C for 3 hours produced a meso-
Bockrath[45] showed that in the catalytic partial oxidation of n- porous sample with SSA of 259 m2
g−1. Li et al[53] tried to
butane to maleic anhydride, yield decrease with time was build a three-dimensional porous nano-network to improve
directly proportional to the loss in surface area (whereas initial oxygen reduction (OR) electrocatalyst using iron and
yield was independent of initial surface area). As reviewed by nitrogen-doped carbon (Fe/N/C) by implementing a special
Sun et al,[46] narrow micropores of zeolites may reduce cata- sample preparation procedure using sodium chloride as a
lyst activity in the conversion of heavy oil-based bulky mole- template. NaCl promoted the formation of a 3D-Fe/N/C with
cules in different reactions, due to the decrease in reactants a much higher SSA and, as a result, higher activity in the
diffusivity. The suggested solution introduces mesopores into electrocatalytic process, owing to the facilitated transfer of
a solely microporous matrix. For example, mesoporous ZSM- O2 molecules. Joshi et al[31] used two approaches for tem-
5 shows much higher activity in alkylation of benzene with plate removal including air calcination and ozone treatment,
ethylene than conventional ZSM-5, owing to its much wider and reported that except for octadecyltrimethoxysilane meso-
diffusion paths in the mesoporous structure.[47] Wang and porous silica (OTMS), the former provides pore structures
Xiao[48] reviewed recent applications of nanoporous catalysts, with higher BET surface area (Table 1).
eg, resins, metal oxides, carbons, mesoporous silicas, poly-
divinylbenzene, and zeolites, that exhibit interesting catalytic
activity, recyclability, and product selectivity in a variety of
4 | UNCERTAINTIES
biomass conversion reactions. For example, Nafion/silica
4.1 | Sources of error and limitations
nanocomposites, SAC-13, are highly active and recyclable in
the esterification of short-chain carboxylic acids, eg, acetic Errors in gas physisorption isotherms determination include
acid, owing to their proper pore diameter, while they show instrumental errors and those related to sample preparation.
low activity for the long-chain ones like caprylic acid, owing To minimize operational errors, the vacuum pump operation
2788 BARDESTANI ET AL.

must first be checked before degassing. Degasser filters elimi- assumptions: it neglects lateral interactions between adsorbed
nate particles that are entrained during degassing. The filter molecules (circumvented by working at low monolayer cover-
maintenance thus improves pores evacuation. O-rings to hold age); and it assumes all adsorption sites are energetically
the cell and insulate the cell connection port must be crack- equivalent.[58] Kruk et al[59] reported that BET is inaccurate to
free, and therefore they should be changed frequently. We measure SSA of MCM-41 silica, since they have strongly het-
recommend wearing gloves when manipulating the quartz erogeneous surfaces that do not satisfy the BET assumptions.
and other cells to avoid adding natural oils from the fingers to As a result, they reported that SSA obtained by BET analysis
the sample cell. Calibrating the cell before each gas phy- was 10% to 15% higher than what is obtained from BJH anal-
sisorption isotherm improves measurement accuracy and ysis or an equation considering geometrical properties (some
reduces operational error, though at the expense of productiv- special considerations such as pore shape were expressed as
ity. Exceedingly high degassing temperatures and degassing an equation by these authors). The difference was attributed
times can change the morphology of a sample, with a to BET overestimation due to overlapping the monolayer for-
corresponding effect on the sample surface area. Too mild mation with multilayer adsorption. Other criticisms are also
degassing conditions may, however, not effectively evacuate related to the lack of mobility of the adsorbed molecules,
the pores from physisorbed water and volatiles, with pores which may result in the preferential location of the adsorbate
remaining inaccessible to nitrogen molecules. This would molecules in the outermost surface, or a lack of consistency
result in underestimation of pore volume and SSA. Thus, with the heat of adsorption values determined using other
samples, especially those with micropores, need to be methods.[9]
degassed under more severe conditions, while avoiding alter-
ing the material properties. As long as degassing conditions
4.2 | Detection limits
change the sample mass, as in the case of some carbonaceous
materials, BET measurements are unreliable and porosity esti- The SSA detection limits depend on sample preparation
mates change with degassing time and temperature. Sigmund (degassing conditions) and especially the type of adsorbate.
et al[56] studied the role of degassing temperature in the BET N2 could be used for most of the porous materials. Some
analysis of four types of biochar, carbon nanotubes, and researchers, however, question whether N2 physisorption
Al2O3. They reported that this parameter strongly influenced can allow the SSA of undeveloped porous structure to be
SSA, changing the elemental composition of biochar samples, determined, which mainly consists of ultra micropores. They
even at a low temperature of 105 C. They proposed that suggest performing adsorption using gases with smaller
researchers in the biochar community prepare standardized molecular size like argon and carbon dioxide. Wang et al[60]
degassing protocols, which, however, might be difficult to reported a very low SSA of 0.7 m2
g−1 for a biochar
achieve due to the different biochar sources and pyrolysis obtained by pyrolysis of Lauan wood (Philippine mahogany)
conditions. For instance, Bachmann et al[57] proposed a at 500 C. Chemical activation using KOH then developed
degassing temperature range of 100 C to 200 C, while Bar- pores, which led to a decrease in average pore diameter from
destani and Kaliaguine[24] found that degassing under 300 C 9.5 to 2.2 nm with an SSA of 1100 m2
g−1. Sedghkerdar
does not allow nitrogen physisorption, owing to the presence et al[61] also reported a very low SSA of 0.5 m2
g−1 for
of a heavy fraction of pyrolysis bio-oil filling biochar pores. Cadomin limestone with an average pore diameter of
In order to obtain more precise SSA, the sample mass should 3.5 nm, used as the base material for synthesizing new sor-
be measured after degassing. For samples that readily adsorb bents for CO2 capture. These reported values indicate that a
humidity, they should be weighed shortly after degassing. very low SSA is possible to measure using N2. On the other
Another common experimental mistake is associated with hand, Suliman et al[62] reported that nitrogen physisorption
insufficient time allowed to reach a stable sample temperature did not allow the SSA of a biochar prepared from wood
after each incremental addition of the adsorbate. The adsorp- hybrid poplar at a pyrolysis temperature below 500 C to be
tion process is exothermic and the temperature increases dur- determined; however, it could for samples prepared at higher
ing the process and sufficient time must be allowed for the pyrolysis temperatures. They reported that the values of SSA
temperature to return to the isotherm temperature. for a pyrolysis biochar produced at 600 C were estimated to
Often articles report SSA with too many significant fig- be 200 m2
g−1 and 400 m2
g−1, using N2 and CO2 as
ures. The absolute precision of SSA measurement by BET is adsorbates, respectively. These authors ascribe these differ-
not higher than 5% and reproducibility is in the range of 1%. ences to the higher accessibility of CO2 to micropores.
Thus, reporting data with more than three significant figures Garrido et al[63] attributed the difference in the value of SSA
is unwarranted. between N2 and CO2 to the temperature at which adsorption
Despite the relative accuracy and simplicity of the BET takes place. They mentioned that for solely microporous
theory, it has been criticized mainly because of its two main solids, narrow pores kinetically restrict the entry of N2
BARDESTANI ET AL. 2789

molecules at 77 K. The kinetic diameters of CO2 (0.33 nm) adsorption sites and lateral interactions between adjacent mol-
and N2 (0.364 nm) are actually similar. An increase in tem- ecules is negligible introduces error. We identified surprising
perature would lead to an increase in the diffusion rate of the contradictions in data reporting associated with the types of
molecules, only resulting in a faster adsorption of gas. adsorbate for the SBET and highlighted work that successfully
Therefore, adsorption at 273 K or even 298 K using CO2 applies N2. The technique is straightforward for ceramics and
would only accelerate the entry of the adsorbate into ultra metals, however for fragile substrates, we recommend that
micropores, not the adsorption equilibrium. researchers complete a systematic literature review to identify
There are also some concerns about using CO2 adsorption operating conditions that minimize mechanical and thermal
for the SSA measurement of pyrolysis biochars, which are stresses. These are some of the difficulties that confront
often loaded with heavy bio-oil. CO2 may dissolve in these researchers, particularly students and those unfamiliar with
heavy liquids, thus yielding overestimated SSA.[64] Contrary the technique who are the target audience for this article.
to the results of Suliman et al,[62] Bae et al[5] systematically Advances in the instrumentation include multiple stations to
investigated the SBET of ultra-micropores zeolites and MOFs accelerate the process and minimize downtime, controlling
and reported that N2 was appropriate for both these materials
the jacket temperature more precisely, and improving the sig-
to determine the ultra-micropore fraction (D<0.7 nm). Com-
nal to noise ratio of the pressure transducers. Older systems
paring with other studies, they concluded that BET calcula-
took up to 18 hours per sample whereas recent technology
tions underestimate the SBET of MOFs and zeolites, with the
has reduced this to 30 minutes.
standard BET relative pressure range (0.05<P/P0<0.3). They
proposed a lower P/P0 range of 0.0001-0.01 for a better esti-
mation of SBET since ultra-micropores are filled with the A C KN O W L E D G E M E N T
adsorbate at much lower relative pressures than the standard
conditions. They mentioned that Ar at 87 K could also be a Brendan A. Patience collected the data from WoS and cre-
reasonable choice for substances with ultra-micropores. The ated the VoSViewer bibliographic map.
results with Ar adsorption agreed well with those measured
by N2 over a range of low relative pressure.[65] Kim et al,[6] NOMENCLATURE
however, criticized the conclusions of Bae and Walton’s find-
ings, arguing that it is impossible to measure nitrogen A molecular cross-sectional area of gas (m2)
(at 77 K) and argon (at 87 K) isotherms for ultra-microporous C BET constant (Equation (2))
zeolites and MOFs with pore diameters close to the kinetic d diameter (m)
diameters of the adsorbates. They suggested that these mole- dpore pore diameter (m)
cules cannot overcome the activation energy corresponding to Dm minimum pore size in the kernel (m)
pores entrance at cryogenic temperatures. They then proposed DM maximum pore size in the kernel (m)
that the SBET over a relative pressure range of up to 0.034 E1 heat of condensation of the first monolayer
determined from CO2 isotherm may be the most meaningful (kJ
mol−1)
EL latent (L) heat of condensation of multilayer
value for the materials with only ultra-micropores. Thus this
(kJ
mol−1)
issue of measuring SSA of ultra-microporous materials is still
g gravitational constant (m
s−2)
debated.
m sample mass charged to cell (g)
M molecular weight (g
mol−1)
5 | CONCLUSI ON S N Avogadro constant (mol−1)
P pressure (Pa)
Specific surface area along with pore size distribution (PSD) Ps saturated vapour pressure (Pa)
are fundamental physico-chemical properties that chemical r radius of a meniscus (m)
engineers measure and control while synthesizing solids and R gas constant (8.314 J
mol−1
K−1)
monitoring changes on stream. These physico-chemical prop- SSA specific surface area (m2
g−1)
erties are determining factors for catalysts and adsorbents per- SBET BET surface area (m2)
formances, and other applications requiring high porosity. t time (s)
Although the technique was developed nearly a century ago, T temperature (K)
challenges remain with respect to instrument operation and V volume (m3)
calibration; researchers often misrepresent the uncertainty in VM molar volume of gas covering a monolayer of
the measurement and report more significant figures than the solids (m3)
method warrants. The theory is well documented, however, vpore specific pore volume (mL
g−1)
assuming that the N2 adsorption energy is independent of the W adsorbate mass adsorbed on solid (g)
2790 BARDESTANI ET AL.

[22] G. S. Patience, Experimental Methods and Instrumentation

Wml adsorbate mass adsorbed in monolayer on solid (g) for Chemical Engineers, 2nd ed., Elsevier B.V, Amsterdam,
x P/P0 The Netherlands 2017.
x
y W ð1− xÞ [23] K. Sing, Colloid. Surface. A 2001, 187-188, 3.
[24] R. Bardestani, S. Kaliaguine, Biomass Bioenerg. 2018, 108, 101.
[25] M. Luisa Ojeda, J. Marcos Esparza, A. Campero, S. Cordero,
Greek letters I. Kornhauser, F. Rojas, Phys. Chem. Chem. Phys. 2003, 5, 1859.
[26] J. C. Groen, L. A. Peffer, J. Pérez-Ramírez, Micropor. Mesopor.
β0 constant in Equation (3) Mat. 2003, 60, 1.
β1 constant in Equation (3) [27] ISO-15901-3, Pore size distribution and porosity of solid mate-
γ surface tension rials by mercury porosimetry and gas adsorption: Part 3: Analysis
of micropores by gas adsorption 2019, https://www.iso.org/
O R C ID standard/40364.html (accessed: October 2018).
[28] J. Landers, G. Y. Gor, A. V. Neimark, Colloid. Surface. A 2013,
Gregory S. Patience https://orcid.org/0000-0001-6593- 437, 3.
7986 [29] P. I. Ravikovitch, A. Vishnyakov, A. V. Neimark, Phys. Rev. E
2001, 64, 011602.
[30] M. Thommes, B. Smarsly, M. Groenewolt, P. I. Ravikovitch,
R E F E REN CE S A. V. Neimark, Langmuir 2006, 22, 756.
[31] H. Joshi, D. Jalalpoor, C. Ochoa-Hernández, W. Schmidt,
[1] G. S. Patience, Can. J. Chem. Eng. 2018, 96, 2312. F. Schüth, Chem. Mater. 2018, 30, 8905.
[2] J.-B. Donnet, E. Custodero, J. E. Mark, B. Erman, C. M. Roland [32] C. L. Lawson, R. J. Hanson, Solving least squares problems,
(Eds.), The Science and Technology of Rubber, 4th ed., Academic SIAM, Philadelphia 1995.
Press, Boston 2013, p. 383. [33] A. V. Neimark, Y. Lin, P. I. Ravikovitch, M. Thommes, Carbon
[3] K. S. Sing, F. Rouquerol, J. Rouquerol, K. Sing, P. Llewellyn, 2009, 47, 1617.
G. Maurin (Eds.), Adsorption by Powders and Porous Solids, [34] Y. Feng, Y. Li, L. Shi, M. Wu, J. Mi, Can. J. Chem. Eng. 2017,
2nd ed., Academic Press, Oxford 2014, p. 237. 95, 2087.
[4] J. Rouquerol, D. Avnir, D. Everett, C. Fairbridge, M. Haynes, [35] S. Hemidouche, L. Boudriche, A. Boudjemaa, S. Hamoudi, Can.
N. Pernicone, J. Ramsay, K. Sing, K. Unger, J. Rouquerol, J. Chem. Eng. 2017, 95, 508.
F. Rodrguez-Reinoso, K. Sing, K. Unger (Eds.), Characterization of [36] J. Wang, Z. Zhang, Q. Zhang, J. Liu, J. Ma, J. Mater. Sci. 2018,
Porous Solids III, Elsevier, Amsterdam, The Netherlands 1994, p. 1. 53, 16466.
[5] Y.-S. Bae, A. Ö. Yazaydin, R. Q. Snurr, Langmuir 2010, 26, [37] C. Ma, R. Wang, Z. Xie, H. Zhang, Z. Li, J. Shi, J. Porous Mat.
5475. 2017, 24, 1437.
[6] K. C. Kim, T.-U. Yoon, Y.-S. Bae, Micropor. Mesopor. Mat. [38] O. Duman, S. Tunç, T. G. Polat, B. K. Bozoglan, Carbohydr.
2016, 224, 294. Polym. 2016, 147, 79.
[7] A. Peigney, C. Laurent, E. Flahaut, R. Bacsa, A. Rousset, Carbon [39] N.-N. Bui, B.-H. Kim, K. S. Yang, M. E. D. Cruz, J. P. Ferraris,
2001, 39, 507. Carbon 2009, 47, 2538.
[8] A. H. Berger, A. S. Bhown, Enrgy. Procedia 2011, 4, 562. [40] V. Krstic, T. Uroševic, B. Pešovski, Chem. Eng. Sci. 2018,
[9] D. Dollimore, P. Spooner, A. Turner, Surf. Technol. 1976, 4, 121. 192, 273.
[10] S. Brunauer, P. H. Emmett, E. Teller, J. Am. Chem. Soc. 1938, [41] G. Chu, J. Zhao, Y. Huang, D. Zhou, Y. Liu, M. Wu, H. Peng,
60, 309. Q. Zhao, B. Pan, C. E. Steinberg, Environ. Pollut. 2018, 240, 1.
[11] E. P. Barrett, L. G. Joyner, P. P. Halenda, J. Am. Chem. Soc. [42] R. Bardestani, C. Roy, S. Kaliaguine, J. Environ. Manage. 2019,
1951, 73, 373. 240, 404.
[12] N. Seaton, J. Walton, N. quirke, Carbon 1989, 27, 853. [43] K. Qian, A. Kumar, H. Zhang, D. Bellmer, R. Huhnke, Renew.
[13] G. Kupgan, T. P. Liyana-Arachchi, C. M. Colina, Langmuir Sust. Energ. Rev. 2015, 42, 1055.
2017, 33, 11138. [44] H. Li, X. Dong, E. B. da Silva, L. M. de Oliveira, Y. Chen,
[14] C. H. Bartholomew, R. J. Farrauto, Fundamentals of Industrial Cat- L. Q. Ma, Chemosphere 2017a, 178, 466.
alytic Processes, 2nd ed., John Wiley & Sons, Hoboken, NJ 2005. [45] G. S. Patience, R. E. Bockrath, Appl. Catal. A-Gen. 2010, 376, 4.
[15] K. S. W. Sing, Pure Appl. Chem. 1985, 57, 603. [46] Q. Sun, Z. Dai, X. Meng, F.-S. Xiao, Chem. Soc. Rev. 2015, 44,
[16] P. Schneider, Appl. Catal. A-Gen. 1995, 129, 157. 6018.
[17] K. Kaneko, J. Membrane Sci. 1994, 96, 59. [47] B.-L. Su, C. Sanchez, X.-Y. Yang, Hierarchically Structured
[18] A. Afshar-Taromi, S. Kaliaguine, Micropor. Mesopor. Mat. 2017, Porous Materials, Wiley-VCH, Berlin 2011.
248, 179. [48] L. Wang, F.-S. Xiao, Green Chem. 2015, 17, 24.
[19] M. M. Labani, R. Rezaee, A. Saeedi, A. A. Hinai, J. Petrol. Sci. [49] X. Song, K. Li, P. Ning, C. Wang, X. Sun, L. Tang, H. Ruan,
Eng. 2013, 112, 7. S. Han, Appl. Surf. Sci. 2017a, 425, 130.
[20] G. Fagerlund, Mater. Struct. 1973, 6, 239. [50] X. Song, K. Li, C. Wang, X. Sun, P. Ning, L. Tang, Chem.
[21] S. Lowell, J. E. Shields, M. A. Thomas, M. Thommes, Surface Eng. J. 2017b, 330, 727.
Area Analysis from the Langmuir and BET Theories, Springer, [51] A. Mashayekh-Salehi, G. Moussavi, K. Yaghmaeian, Chem.
Dordrecht, The Netherlands 2004, p. 58. Eng. J. 2017, 310, 157.
BARDESTANI ET AL. 2791

[52] L. Xu, Q. Jiang, Z. Xiao, X. Li, J. Huo, S. Wang, L. Dai, Angew. [61] M. H. Sedghkerdar, N. Mahinpey, Z. Sun, S. Kaliaguine, Fuel
Chem. Int. Edit. 2016, 55, 5277. 2014, 127, 101.
[53] Y. Li, W. Liao, Z. Li, T. Feng, L. Sun, C. Guo, J. Zhang, J. Li, [62] W. Suliman, J. B. Harsh, N. I. Abu-Lail, A.-M. Fortuna,
Carbon 2017b, 125, 640. I. Dallmeyer, M. Garcia-Perez, Biomass Bioenerg. 2016, 84, 37.
[54] N. J. van Eck, L. Waltman, Science 2010, 84, 523. [63] J. Garrido, A. Linares-Solano, J. M. Martin-Martinez, M. Molina-
[55] Web of Science™ Core Collection, http://apps.webofknowledge. Sabio, F. Rodriguez-Reinoso, R. Torregrosa, Langmuir 1987,
com (accessed on July 12, 2018). 3, 76.
[56] G. Sigmund, T. Hüffer, T. Hofmann, M. Kah, Sci. Total Environ. [64] A. Samanta, A. Zhao, G. K. H. Shimizu, P. Sarkar, R. Gupta,
2017, 580, 770. Ind. Eng. Chem. Res. 2012, 51, 1438.
[57] H. J. Bachmann, T. D. Bucheli, A. Dieguez-Alonso, D. Fabbri, [65] K. S. Walton, R. Q. Snurr, J. Am. Chem. Soc. 2007, 129, 8552.
H. Knicker, H.-P. Schmidt, A. Ulbricht, R. Becker, A. Buscaroli,
D. Buerge, A. Cross, D. Dickinson, A. Enders, V. I. Esteves,
M. W. H. Evangelou, G. Fellet, K. Friedrich, G. G. Guerrero,
B. Glaser, U. M. Hanke, K. Hanley, I. Hilber, D. Kalderis,
J. Leifeld, O. Masek, J. Mumme, M. P. Carmona, R. C. Pereira, How to cite this article: Bardestani R, Patience GS,
F. Rees, A. G. Rombolá, J. M. de la Rosa, R. Sakrabani, S. Sohi, Kaliaguine S. Experimental methods in chemical
G. Soja, M. Valagussa, F. Verheijen, F. Zehetner, J. Agric. Food
engineering: specific surface area and pore size
Chem. 2016, 64, 513.
[58] A. Seri-Levy, D. Avnir, Langmuir 1993, 9, 2523.
distribution measurements—BET, BJH, and DFT.
[59] M. Kruk, M. Jaroniec, A. Sayari, Langmuir 1997, 13, 6267. Can J Chem Eng. 2019;97:2781–2791. https://doi.
[60] S. Wang, H. Wang, Q. Yin, L. Zhu, S. Yin, New J. Chem. 2014, org/10.1002/cjce.23632
38, 4471.