Adsorption, Surface Area,

and Porosity

Aaron Nackos
Purpose and outline
Why: discuss conceptual framework for
understanding isotherms so that details
on calculations can be easily understood
 Physical picture of porous
solid blind ink bottle
surface
roughness

closed Other
shapes =
slit or
funnel

cylindrical
blind cylindrical
open
through
Three pore size regimes
Micropores: dpore < 2 nm (N2 diameter
around 0.35 nm)
Mesopores: 2 nm < dpore < 50 nm
Macropores: dpore > 50 nm

Distinctions arise from differences in how

adsorbates physisorb, as we shall see
shortly
Why are pore size and
surface area important
Changes in these properties are
indicative of pore plugging, material
sintering
Comparison between samples
possible
Used in computations such as Thiele
modulus (pore diffusion)
Terms
Adsorptive: substance
that is capable of being
adsorbed as it exists in
the fluid state
Adsorbate: material in
adsorbed state
Adsorbent: surface of
the solid on which
adsorption occurs Solid surface
Physisorption vs. Chemisorption
Weak attraction Strong attraction of
produced by induced adsorbate to surface
dipole, Van der Waals due to sharing of
forces electrons
Easily reversible
Not easily reversible
(desorbed)
Adsorbent surface not Surface often altered
usually altered Example: poisoning of
Example: H2O H2S on Co or Fe sites
condensation on cold on FT catalyst
glass
 Ideal vs. actual
physisorption behavior
ideal
Ideal solid surface

ith layer
actual
3rd layer

2nd layer

1st layer

Actual solid surface

N2 is most commonly used
adsorbate because:
1. N2 has permanent quadrupole
moment (responsible for forming
well-defined monolayer on most
surfaces)
2. Level of specificity on most
surfaces not too strong for N2
3. Liquid and gaseous N2 are readily
available and cheap
Running experiment 1:
Loading sample tube
Small amount (~0.5
grams) of sample to be
analyzed is placed in
clean, dry sample tube
Tube must be pre-weighed
before and after sample
introduction
Avoid touching with fingers
so oils do not interfere
with weight measurements
Running experiment 2:
Outgassing sample
Sample tube is placed
in degasser to remove
moisture and other
contaminants
Degasser flows dry N2
over sample, which is
heated to >100 C (up
to 400 C)
Running experiment 3:
Sample loaded into Tristar
Up to 3 sample
tubes loaded into
analyzer,
evacuated to near-
vacuum, and
immersed in N2
bath @ 77K
N2 gas introduced to sample
tubes in small increments;
volume of gas added and
equilibrium pressure recorded
Results obtained plotted as
isotherm

P0 is saturation pressure of N2 (= ambient press)

Different isotherm shapes
possible

Non-porous; strong Non-porous; weak

adsorbate-adsorbent adsorbate-adsorbent
interactions interactions
Microporous

Porous;
analogous
to Type II
Porous;
analogous
to Type III Homogeneous
surface
Type I notes
Indicates sample
contains micropores Type I
only
Sometimes a narrow
hysteresis is observed

Amt adsorbed
Observed in
Charcoal
Silica
Molecular sieves p/p0 1.0
Stannic oxide gel
Ammonium
phosphomolybdate
Type I notes (cont.)
Theory predicts fit to Langmuir isotherm:
x Bp e E RT
=
N av
=
1

B=
(2MRT )
1
xm 1 + Bp z m v1 2

p is pressure
B is constant at given temp
zm is # molecules adsorbed per cm2 in
completed monolayer
v1 is oscillation frequency of adsorbed
molecules
E1 is energy of desorption
Type I notes (cont.)
In practice, Type I isotherms not useful for
reliable values of specific surface, but may
predict within 2-3 times in either direction
In general, plateau of Type I represents
volume filled with adsorbate in liquid-like form,
rather than monolayer predicted by theory
Inaccuracies occur because of:
Activated passage of molecules through fine
restrictions (=slow adsorption)
Pore filling even at low pressures because potential
fields from opposite walls overlap, increasing
attractive force on adsorbate molecules as compared
to larger pores
 Notes on Type II adsorption
On oxides and hydrated oxides, N2 adsorption
gives high values of c (see discussion on BET
method)
N2 is most often used adsorbate because on
majority of nonporous solids it yields type II
with sharp, well-defined knee
On system with well-defined knee and minimal
surface heterogeneity, total gas adsorption at
knee (xknee) is quite close to the monolayer
adsorption as calculated from the BET equation
(xm)
For gases other than N2, BET is less reliable
overall if knee is less well-defined
 Type III adsorption
The more molecules that are adsorbed, the
easier it is for further molecules to adsorb
Type III observed when adsorbate-
adsorbent interactions are weak (low values
of cSee following BET discussion)
In BET eq., xm is so small that accuracy of
equation is diminished
Bottom line: BET does not work for Type III
isotherms
Conclusions for non-porous
isotherms (Type II and III)
Optimal conditions for determining surface
area from monolayer adsorption:
Isotherm is Type II
Isotherm has sharp knee
Sharp knee results from high net Hads (i.e. c>50)
A sharp knee sometimes results from
localized adsorption (such as adsorption on
active sites) where surface is energetically
heterogeneous
Am (the area per molecule adsorbed)
depends on adsorbate packing mode, which
depends on nature of solid surface
 Type IV Isotherm
Porous Type IV Type IV
analogous to non-
porous Type II Desorption
branch
Type IV appears in

Amt adsorbed
solids that possess
mesopores (2-50 Adsorption
branch
nm)

p/p0 1.0
Type IV Isotherm
Capillary
condensation in
course pores
Mesopore
desorption branch;
behaves according
to Kelvin eq. Mesopores completely
filled (full hysteresis
Amt adsorbed

curve seen only if p/p0

exceeds this point)

Mesopore adsorption branch

(general opinion is
multilayer gradually built
up)
Monolayer adsorption
and micropore filling

p/p0 1.0
Conclusions for surface area
calculations of Type IV
Surface area for Type IV computation
same as for Type II, with same rules
If knee is not sharp, then values for
specific surface become more uncertain (if
c < 10, then uncertainty in value of surface
area can be as high as 100%)
Presence of micropores can throw off
calculation, making apparent surface area
higher than actual area b/c micropores
take up several layers of adsorbate before
monolayer is complete on larger pore walls
 Type V adsorption
Uncommon
Same problems as for Type III
isotherms (Type V porous, Type III
non-porous)
Bottom line:
Only plots of Type II and Type IV
(most common for N2 adsorption on
most oxides) are useful for
calculations now discussed
Determining surface area of
adsorbent
If total amount of adsorbate taken
up by adsorbent at monolayer
coverage is known, total surface
area of adsorbent can be calculated
Multilayer info would be misleading
Area of N2 molecule = 0.162 nm2
BET (Brunauer, Emmett,
Teller) surface area method
Details of derivation left out, but
following assumptions important:
Energetically uniform surface (all
adsorption sites equivalent)
Horizontal interactions between
adsorbent molecules are neglected
2nd and subsequent layers treated as
equivalent and as having properties of
bulk liquid
 x c ( p p0 )
BET equation: =
xm {1 ( p p0 )}{1 + [c 1]( p p0 )}

Used to predict adsorption of gaseous species

based on relative pressure (p/p0)
p0 is vapor pressure of gas at sample temperature
p is absolute pressure inside sample chamber
c is constant that depends on adsorbate,
adsorbent, and temperature
x is (usually) volume of gas adsorbed per gram of
material @ STP (can be grams adsorbate/gram
material or moles adsorbate/gram adsorbent)
xm is amount of gas corresponding to 1 monolayer
on solid surface
 Comments on constant, c,
in BET equation
x c ( p p0 ) E1 L
= c exp
xm {1 ( p p0 )}{1 + [c 1]( p p0 )} RT
E1 is heat of adsorption of first layer onto solid
surface
L is latent heat of condensation (bulk liquid)
The larger c is, the greater the disparity in average
lifetime of molecules in 1st layer compared to that
of molecules in higher adsorbed layers
(in Aarons words, higher c means molecules more
strongly attracted/bound to surface than to each
other)
 Value of c changes isotherm
shape
Isotherm is Type II (or IV) if c > 2, Type III (or
V) if c < 2 (c > 50 desired for good fit to data)
(stop presentation and double-click on Excel image below to observe effects on isotherm by varying c)

c= 50

Sample Isotherm
2
1.8
1.6
1.4
1.2
x/xm

1
0.8
0.6
0.4
0.2
0
0 0.1 0.2 0.3 0.4 0.5 0.6
p/p0
 Linearized BET equation
BET equation can be linearized to obtain:

p p0 1 c 1
= + p p0
x(1 p p0 ) xm c xm c
Linear behavior usually observed between p/p0 = 0.05-0.30

Plot left side versus p/p0 to calculate xm

and c, as shown below
1 slope
xm = c= +1
slope + intercept intercept
 Tristar instrument BET
output
# of points used in BET
calculation depends on # points
taken by instrument and on user
settings
nonlinearity reflected in plot
and correlation coefficient

slope
c= +1
intercept
1
xm =
slope + intercept
BET surface area single
point method
p p0 1 c 1
= + p p0
x(1 p p0 ) xm c xm c
If C is sufficiently large, then linearized
form of BET simplified to
xm = x(1 p p0 )
Micromeritics Tristar analyzer chooses
p/p0 closest to 0.30 and reports this on
summary report (last page in report)
Error is only few % if C = 100, but when
C < 80, error is substantial
Considerations in pore
volume and size computation
Customary to take total specific pore
volume @ p/p0 0.95, but this may not
be satisfactory because adsorption
capacity as p/p0 1 is dependent on
both upper limit of pore size distribution
and on external area
Amount adsorbed @ isotherm plateau at
high p/p0 ratios can be regarded as total
volume in that range of mesopores
 Laplace equation
Models the pressure gradient across a small, curved
meniscus

Pconcave (top) > Pconvex (bottom)

2
p ' ' p ' =
r
Assuming that molar volume of liquid is both constant
and negligible compared to that of vapor, and that
the vapor is ideal, Kelvin equation can be derived
Estimation of pore size from
Kelvin equation
Kelvin equation predicts equilibrium vapor
pressure of liquid contained in cylindrical
capillary
Molar volume Liquid surface
of liquid tension
~
p 2 V
ln = cos( )
p0 rK RT
Wall-liquid
Kelvin pore radius contact angle

usually assumed to be zero (may not

always be a good assumption)
 Comments on Kelvin
Equation (cont.)
Previous form of Kelvin Equation obtained for
cylindrical pores, where dv r
=
dS 2
~
p
General Kelvin equation: ln = V cos( )
p dv
0 RT
dS
v is pore volume; S is surface area of pore
walls
If pores not cylindrical, the Kelvin radius
may no longer be equal to the parameter(s)
that describe the actual pores
 Comments on Kelvin
Equation (cont.)
For pore systems that are not very complicated,
relationship between r and pore size parameter(s)
can be worked out from principles of geometry
(figures from Gregg and Sing, 1967)
Slit-shaped pores
w p = rK + 2t If hemicylindrical-shaped
gas-liquid interface is
assumed; wp is pore width
 Evacuated pore still retains
adsorbed multilayer film
rK (nm) is radius of Reproduced from
pores in Kelvin eq. Rouquerol et al. 1999
0.415
rK =
log10 ( p p0 )
rp is actual pore radius vapor
t is thickness of
adsorbed film that t 2rK t

remains on surface
after pore evacuation solid solid
(Tristar has equations 2rp
to estimate thickness)
rp = rK + t liquid
BJH (Barrett, Joyner,
Halenda) method for pore size
Estimates pore size distribution and
average pore size using:
Isotherm data (p/p0 and volume)
Kelvin equation
Model for thickness of adsorbed multilayer
Simple geometry relationships for cylinder
Pores of different size devolve gas at
different relative pressures
Measuring volume gas injected/devolved
at various p/p0 values allows estimation
of volume of pores within a certain size
range
Assumptions used in BJH pore
size distribution computation
Kelvin eq. applicable over entire range
Meniscus curvature controlled by pore
size/shape
= 0
Pores are rigid and of well-defined shape
Distribution is confined to mesopore range
Filling/emptying of pores do not depend on
location
Adsorption on pore wall proceeds by same
mechanism as on corresponding open
surface
Important comment:
In the following 12 slides, a simplified
picture of pore filling/emptying is
presented where all pores are cylindrical,
are open on one end, and have access to
the pellet surface. This illustrates the
basic principles of adsorption/desorption
and can explain part of the reason the
hysteresis is seen in a Type IV isotherm.
In reality, pore structure is very
complicated, with pores being
interconnected and having varying
diameters and shapes.
 Amt ads.
Adsorption 1
p/p0 1.0

Adsorbent surface (exposed to gaseous

adsorptive)

Pores
 Amt ads.
Adsorption 2
p/p0 1.0
 Amt ads.
Adsorption 3
p/p0 1.0
 Amt ads.
Adsorption 4
p/p0 1.0
 Amt ads.
Adsorption 5
p/p0 1.0
 Amt ads.
Adsorption 6
p/p0 1.0
Adsorption 7

Amt ads.
Pores filled
p/p0 1.0
 Amt ads.
Desorption 1
p/p0 1.0
 Amt ads.
Desorption 2
p/p0 1.0
 Amt ads.
Desorption 3
p/p0 1.0
 Amt ads.
Desorption 4
p/p0 1.0
Desorption 5

Amt ads.
adsorbate removed
p/p0 1.0
Summary: Key features of
Type IV isotherm
Capillary
condensation in
Mesopore
course pores
desorption branch;
behaves according
to Kelvin eq.
Mesopores completely
Linear range filled (full hysteresis
Amt adsorbed

for BET; p/p0 curve seen only if p/p0

0.05 to 0.30 exceeds this point)

Mesopore adsorption branch

(general opinion is
multilayer gradually built
up)
Monolayer adsorption
and micropore filling

p/p0 1.0
Comments on Hysteresis
Much work has gone into understanding
hysteresis. Pore filling mechanism not
same as emptying mechanism. General
conclusions:
1. Metastable layer develops on adsorption
branch (pores dont fill with capillary
condensation like they empty, but rather a
thick multilayer shell is gradually built up
inside pore).
2. On desorption, condensate is entrapped by
network percolation (pore mouths are not
directly open to the particle surface).
Comments on Hysteresis
(cont.)
Difference in contact angle
(assumed to be zero for simplicity)
during adsorption and desorption
can also be a factor (see next slide)
For pore size, generally desorption
branch used, although neither
branch truly behaves according to
theory
 Causes of Type IV
hysteresis (cont.)
>0
Amt adsorbed

p/p0 = 1

p/p0 1.0

If isotherm cuts saturation pressure ordinate =0

(where p/p0 = 1) at finite angle (top) rather
than asymptotically (bottom), > zero at high
pressure end and will remain so as desorption
occurs.
Even if = 0 at high pressure end, as
desorption branch is traversed it is likely that
increases
 Forms of Type IV Isotherm

Amt adsorbed
Amt adsorbed

Amt adsorbed

p/p0 1.0 p/p0 1.0 p/p0 1.0

Open-ended Capillary space Tubular or ink-

cylindrical pores between parallel bottle pores of
lying in narrow plates or open slit- varying radius (e.g.
range of radius shaped capillaries pores with
restrictions)
[See slide]
[See slide]
 Notes on parallel plates
model
During adsorption, meniscus
not formed between parallel
plates until vapor pressure
reaches saturation

Amt adsorbed
Upon desorption, cylindrical A
meniscus is present between
plates, which gradually come
together as liquid evaporates
(horizontal part of desorption B
curve, A)
Plates continue to merge until
they cannot come together p/p0 1.0
further due to mode of
packing or small particles in-
between (vertical part of
desorption curve, B)
 Notes on varying pore
radius model
Adsorption (should be used
for calculating pore size
distribution)
Pores of radius rn begin to
fill first, corresponding with
p/p0 ratio in Kelvin eq.
Pores continue to fill as p/p0 A
increases, the meniscus

Amt adsorbed
gradually increasing in
radius
Desorption B
Pores should fail to empty
(A) until pressure falls to
value corresponding to rn
At B, all pores empty at
once p/p0 1.0
 Example pore size
distribution plot from Tristar

These plots show peak at around 10 nm

Other plot types exist (see manual)
Tristar 3000 software has built-in functions
that smooth curves
Hydraulic radius
If pore geometry can be specified by
single parameter, hydraulic radius
(rh) can be used to approximate
average pore radius
Example: cylindrical pores

rh L rh
2 Vol
Vol
= = rh = 2
Area 2 rh L 2 Area
Vol
dh = 4
Area
Tristar includes this in summary
report (volume and area from BJH)
Hydraulic radius (cont.)
Hydraulic radius can be compared to
maximum of pore distribution curve.
rh is usually approximately equal to
rp. If rh > rp, there may be
significant micropore volume, which
will also show up as a wide
discrepancy between BET area and
cumulative BJH surface area.
Summary report

How surface area

calculated found in
manual
Summary, final remarks
SA, pore sizes as determined by gas
physisorption are useful for comparison
of samples and are good as an estimate
of true parameters, but may be very
inaccurate
Derived mesopore distribution may give
misleading picture of actual pore size
distribution
Pores are rarely cylindrical as models
say
Summary, final remarks
(cont.)
Other sources give more instruction on
calculations (see appendix in
Micromeritics manual, B&F, Sing)
No generally reliable method available
for micropore volume or area, although
some calculation methods are included
in instrument (t-plot estimates
micropore surface area contribution,
Langmuir, MP method also)
Mercury intrusion useful for macropore
size determination
Sources
Adsorption, Surface Area, and
Porosity (Gregg and Sing, 1967)
Fundamentals of Industrial Catalytic
Processes (Farrauto and
Bartholomew)
Adsorption by Powders and Porous
Surfaces (Rouquerol, Rouquerol, and
Sing, 1999)
Sources (cont.)
Appendix C of Tristar operators
manual
Reporting Physisorption Data for
Gas/Solid Systems (Sing et al.,
1985; Pure & Appl. Chem, Vol. 57,
No. 4)
Recommendations for the
Characterization of Porous Solids
(Rouquerol et al., 1994; Pure &
Appl. Chem, Vol. 66, No. 8)