CALORIMETRIC MEASUREMENTS OF THE ADSORPTION OF COLLAGEN
AND OTHER ORGANIC ONTO OXIDE SURFACES
PAUL J. BUSCEMI
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMNETS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
TABLE OF CONTENTS
ACKNOWLEDGEMENTS .. . . . . . . .ii
ABSTRACT . . . . . ... . . v
I INTRODUCTION . .. . . . . 1
Use of Protein Models . .. . . 2
Calorimetry as an Analytic Tool .. .. 4
Protein Adsorption . . .. . . 6
Surfaces. ............ . 10
Forces of Adsorption .. . . . . 15
Thermodynamics of Adsorption. . . ... 17
Heterogeneous Adsorption . . . ... 24
Experimental. . . . . . . ... 33
II ADSORPTION OF CARBOXYLIC ACIDS
ON ALUMINA AND HYDROXYAPATITE 41
Introduction. . . . . . . ... 41
Experimental . . . . . . ... 43
Results . . . . . . . . 44
Discussion. . . . . ... . . 51
Conclusions . . . . . .. .... 59
III ADSORPTION OF CHONDROITIN SULFATE
AND OTHER CARBOHYDRATES . . . .. 61
Introduction. . . . . . . ... 61
Experimental . . . . . . ... 62
Results . . . . . . . . 62
Discussion. . . . . . ... 70
Conclusions . . . . . .. . . 77
IV ADSORPTION OF POLYPEPTIDES. . . .. 79
Introduction. . . . . . . ... 79
Experimental. . . . . . . ... 81
Results . . . . . . . .. 81
Discussion. . . . . . ... 86
Conclusions . ... . . . . ... 92
TABLE OF CONTENTS Continued
V REACTION OF PEPTIDES AND CARBOHYDRATES
IN SOLUTION . . . . . .
Introduction. . . .
Experimental. . . .
Results . . . .
Discussion . ..
Conclusions .. ...
VI ADSORPTION OF COLLAGEN.
Results . .
VII BIOGLASS. . .
Results . .
VIII SUMMARY .. ..
BIBLIOGRAPHY . . .
BIOGRAPHICAL SKETCH .
. . . . . 93
. . . . 95
. . . . . 97
. . . . 108
. . . . . 115
. . . . 117
. . . . 117
. . . .. . 117
. . . . 117
. . . . 132
. . . . 138
. . . . . . 141
. . . . . . . 141
S . . . . . . 142
. . . . . . . 142
. . . . . . 148
. . . . . . . 150
. . 152
Abstract of Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of the
Requirements for'the Degree of Doctor of Philosophy
CALORIMETRIC MEASUREMENTS OF THE ADSORPTION OF COLLAGEN
AND OTHER ORGANIC ONTO OXIDE SURFACES
Paul J. Buscemi
Chairman; R. E. Loehman
Major Department: Materials Science and Engineering
The present work is the result of the application of
solution microcalorimetry to the problem of determining the
energies of adsorption of organic molecules onto ceramic
surfaces. The systems studied were chosen to model the
attachment of collagen to ceramics and to provide some expla-
nation for the observed bonding of ceramic implants to bone.
An aqueous solution of an organic molecule such as a
polyamino acid, polysaccharide, or smaller molecules with
similar functional groups was automatically mixed in a micro-
calorimeter with a slurry of a powdered oxide such as A1203,
SiO2, or a special glass composition and the heat evolved or
adsorbed was determined. Calorimetric measurements were
performed on increasing concentrations of reacting organic
molecules for a fixed weight of powder with known surface
area. Plots of the reaction heat, Q, versus the initial con-
centration of the organic, Co, yield thermometric titration
curves which were analyzed to give the enthalpy of the
reaction AH, the free energy change AG, and by difference
the entropy change for the reaction AS.
The systems were studied in order of increasing
structural complexity from simple carboxyls, amines, and
organic sulfates to amino acids, polyamino acids, polysaccha-
rides, and collagen. Changes in the aqueous solutions by
additions of salts or changes in pH and combinations of
organic molecules were also studied.
Results indicate that there are at least two forces
which contribute to the bonding of the collagen and other
organic to oxide surfaces, hydrogen bonding and ionic bond-
ing, the former releasing from 8 to 12 kcal/mole of functional
groups while the latter releases 4 to 6 kcal/mole of
functional groups depending on the relative polarities of
the adsorbing molecules and the surface. There are strong
indications of the denaturation of collagen at some surfaces
at which hydrogen bonding and ionic bonding act cooperatively.
The study of adsorption and interaction of proteins and
other biological molecules onto non-biological surfaces is
important because of the increasing use of prosthetic materials
. It is essential to know how each of these two distinctly
different components will interact in biological media. In
this study, a further understanding of the reaction between
the connective tissue protein, collagen, and various oxide
surfaces is sought.
Two properties of collagen adsorption are of major
concern: how well does the protein adhere to the oxide surface
and does the surface change the structure of the protein?
Protein adhesion relates to the binding of tissue to a prosthetic
material and can be approached by the determination of the
enthalpy AH of the adsorption reaction . Changes induced
by the surface on the protein lead to denaturation. This
increases its vulnerability to enzyme attack and eventual
rejection from the host . This question can be approached
by comparison of the enthalpy of reactions of model systems
with that of reactions which have been found not to be
disruptive to the structure of protein.
Three materials serve as substrates: silica, alumina,
and hydroxyapatite. Each of these materials is well character-
ized and has potential for use as a prosthetic material .
The calorimetric measurements made, therefore, are not for
the purpose of further characterizing these materials but to
study their influence on the adsorption of organic molecules.
The values of the thermodynamic parameters (AG, AH, and
AS), determined from the calorimetric measurements, do depend,
however, on the structural features of the surface as well as
those of the adsorbing molecule and on their mutual environment.
A'practical approach for studying complicated systems encountered
in actual application is to study simpler model systems ,
which provide singular features for observation.
Within a series of model compounds, the structure of
the molecules, the solvent, and the surface can be systematically
varied and correlations can be made between the thermodynamic
data and the variations in experimental conditions. In this
study, extensive use is made of several types of molecular
model compounds including those representing collagen as well
as those representing carbohydrate and other physiologically
relevant organic structures.
Use of Protein Models
Past workers have used molecular model compounds designed
to study collagen. Specifically, the polyamino acids have
been well studied in this way. Poly-l-proline has been used
in conformational studies in CaC12 solution  showing that
disordering of the molecule is associated with its increasing
rotational ability. X-ray diffraction studies of poly-l-proline
have demonstrated that the backbone conformations of the
molecule are similar to native collagen . Poly-l-lysine
and poly-l-arginine have been used as models for collagen in
relation to the structure of amorphous ground substance [8,9].
Poly-l-lysine and poly-l-glutamic acid have been used in
conformational studies using differential capacitance techniques
. In other studies, workers have used combinations of
synthetically prepared amino acids to model collagen [11,12].
Smaller molecules representing isolated residues of
the collagen molecule have also been investigated but not as
frequently as the polyamino acids. Dyes containing amino
groups have been shown to selectively adsorb chondroitin sulfate
, an important carbohydrate in structural tissue .
Surface viscosity measurements using amines, amides, and
carboxylic acids as model proteins have been studied in relation
to bilayer film formation  in membrane studies.
In biological studies of proteins other than collagen,
molecular models have been widely studied. Enthalpies of
aqueous solution have been determined calorimetrically for
amines and carboxylic acids  as part of a quantitative
description of biological systems. Heat capacity measurements
 on several amino acid solutions have been made to help
explain protein structure. The binding of short amino acid
chains  in subunit studies of immunoglobulin has also been
investigated by calorimetry. Differential scanning calorimetry
has been used to study conformation changes of many polypeptides
used as models for collagen [19,20].
Molecular models in non-biological studies have also
been used. The adsorption of molecules containing the same
functional groups which proteins possess, amines , sulfates
, and organic acids , have been examined by various
methods. Calorimetry has not been extensively used for this
purpose. In general, the use of molecular models in biological
and non-biological systems appears widespread for the determina-
tion of various properties.
Calorimetry As An Analytic Tool
Calorimetry has long been used to measure the enthalpy
of adsorption (heat of wetting) of various liquids onto dry
oxide surfaces. Such measurements are made by immersing a
clean, dry solid powder into a liquid. The heat change is
measured in a calorimeter. For our purposes, the most
relevant liquid used in previous studies was water. Heats of
wetting of silica [24-26], of alumina [27-29], and of hydroxy-
apatite  have each been measured. The results of these
works show several consistent features. First, there are
differences in the heats of wetting with variation in the out-
gassing pressure and with temperature of evacuation, indicating
surface heterogeneities. Also there are differences in heats
of wetting with variation in particle size. Finally, the heats
of adsorption of water range from -10 to -20 kcal/mole of water
adsorbed and are attributed to hydrogen bonding of the water
to the surface .
Calorimetry has also been used in a similar manner
to measure the heat of adsorption of a second component from
an aqueous solvent . The heat of adsorption of sodium
dodecyl sulfonate (SDS) was found by subtracting the heat of
wetting of alumina in pure solvent from that found with SDS
present. In this case the initial adsorption attraction was
attributed to coulombic forces between the surface and ion
and was calculated to be -12 kcal/mole.
Results for similar experiments have been confirmed by
other methods [32,33] using water and other polar and ionic
liquids as adsorbents. For example, the differential heat of
adsorption for the adsorption of octadecyl alcohol on alumina
from a benzene solution, calculated from the temperature
dependence of the adsorption coefficient was found to be -8.6
kcal/mole while that found directly from calorimetric measure-
ments was -8.68 kcal/mole .
The heat of adsorption of water on quartz determined
by adsorption measurements at several temperatures was found
to be between -11 and -14 kcal/mole of water adsorbed .
There were differences found when the quartz was ground and
exposed to water vapor prior to drying and evacuation. This
was attributed to the formation of an amorphous layer of silica
on the surface. These differences disappeared when the rough-
ened surfaces were annealed at 7000C. The heats of wetting
agree well with those found from calorimetric measurements .
Calorimetry has been extensively used in biochemical
applications . However, there are few calorimetric data
on the adsorption of proteins onto oxide surfaces and there
is apparently none for the adsorption of collagen. The
calorimetric data most relevant to adsorption primarily
involve such globular serum proteins as serum albumin .
The remainder of this section will therefore be devoted to
previous studies of protein adsorption use with particular
emphasis on oxide substrates.
The demonstration of molecular attachment of cell
proteins on foreign substances has been accomplished by various
methods. Multiple internal reflection spectroscopy has
been used to measure protein interaction using germanium 
as a substrate. A KRS-5 prism pressed against protein on a
hydroxyapatite substrate allowed protein-hydroxyapatite inter-
action to be studied . Although energy calculations were
not carried out in these studies, the changes in adsorption
frequencies indicate chemical interaction with the surface of
Film compression studies  using collagen, gelatin,
and poly-l-alanine with silica gel showed adsorption hysterisis
indicating an irreversible process. The maximum interaction
of the silica gel and collagen occurred at pH 5.2. The iso-
electric point,where there is charge neutralization of the
protein, is 5.5. There was also interaction between alanine,
which has no ionic side groups, and the gel. The interaction
appeared to be of the same type as that of collagen. The
primary binding force was assumed to be hydrogen bonding .
Adsorption of bovine serum albumin (BSA) on hydro-
philic silica  exhibited a maximum surface density at
pH 5.5. The isoelectric point (IP) of this protein is 4.9.
The surface of the silica is negatively charged at this pH.
Desorption occurred readily at pH's away from the IP indicating,
as suggested by the author, that binding was due to hydrophobic
interaction. Other studies [42,43] showed that even after
extensive washing with water and EDTA that not all of the BSA
adsorbed onto pyrex glass could be recovered. Maximum
adsorption was near the IP of the protein and the free energy
change was estimated to be -2.5 kcal/mole of protein. The
enthalpy was not calculated.
Serum globulins have been shown to be preferentially
adsorbed by silicic acid  and silica  and by other
minerals . Maximum adsorption took place at the isoelectric
point on these surfaces as well as on calcium phosphate gel
. In none of these studies were determinations of the
enthalpy of the various adsorption reactions made.
The adsorption of albumin, fibrinogin, and globulin on
polyethylene has been determined by internal reflection spectro-
scopy . The adsorption isotherms followed a Langmuir isotherm,
a common finding in which the quantity of solute adsorbed, X,
at the equilibrium concentration, C, is given by X = aC/(l + bC)
where a and.b are constants. The adsorption was assumed to be
due to hydrogen bonding because of the shift in the amide I
(C=O stretching) band at 1640 cm-1
Calorimetric measurements of the adsorption of human
serum albumin on negatively charged polystyrene (PS) 
were shown to be pH dependent. Maximum adsorption of the
protein occurred near the IP (4.9) where the enthalpy of the
reaction was positive. At pH values removed from the IP the
reaction was exothermic. It was suggested that at pH values
away from the IP the conformation of the adsorbed protein
changes for energetic reasons. Denaturation of the protein
is not surprising since it is known that the internal bonding
of serum albumin is weak . The' enthalpy of the adsorption
reaction varied between zero and 8 kcal/mole as the surface
charge varied between -1.0 mpCcm-2 and -7.5 mpCcm-2. The
most negative enthalpy values were recorded for pH 3.8 and 9
and were near -8.4 kcal/mole.
Enthalpy values for the adsorption of albumin, gamma
globulin, and fibrinogin, were calculated from adsorption
measurements at several temperatures . The results indicated
that the adsorption took place in two distinct ways. Both
types were apparently Langmuir and took place on separate
membrane sites. One type of adsorption was easily reversible
with a heat of adsorption in the neighborhood of -10 kcal/mole.
The other type of adsorption reaction was hydrophobic, endothermic,
and with heats of adsorption in the range of 5 to 20 kcal-
In another study , the forces involved in the
adsorption reactions between several globular proteins and
glass surfaces were determined to be primarily ionic amine-
silanol bonding and hydrogen bonding. Two rates of adsorption
were noted. The first appeared to be related to the number
of amines present on the surface of the protein. The second
was slower and seemed to be dependent on the molecular weight
of the protein. Hydrogen bonding was suspected since the
proteins could not be completely washed from the surface with
Ionic bonding of ribonuclease to glass was indicated
to be strong  since very little of the protein could be
removed by rinsing in several solvents. No enthalpy determina-
tions were made.' There was a decrease in adsorption with
an increase in ionic strength.
Further review of the literature reveals that the
various adsorption studies cannot be readily compared due to
the large number of experimental variables and to the random
manner in which they are controlled in each experiment. A
few common features in the study of protein adsorption do
There is usually more than one type of interaction
present for any particular system and one of these is usually
hydrogen bonding. The observed enthalpy values are in the
range of -10 to +10 kcal/mole of protein. Finally, maximum
adsorption density appears to take place near the IP of the
protein. There are many exceptions to these general results,
The effect of changes in ionic strength on adsorption
is also not well understood . Dissolved salts disrupt
hydrogen bonds which proteins depend on for conformational
stability . Changes in the ionic strength of the solvent
will also have effects on the adsorbing surface. For example,
phosphate, a common component in buffers, will change the surface
charge of alumina . Generalizations are difficult to
make about the action of specific ions on adsorption unless the
specific system understudy is clearly defined.
Silica, alumina, and hydroxyapatite, the three materials
used in this study, are oxides. Hydroxyapatite is sparingly
soluble at neutral pH whereas silica and alumina are virtually
insoluble . All are hydrophilic and each displays a
surface charge which varies with pH.
The surface charge results from exposed surface atoms
attempting to complete their coordination of nearest neighbors
. Exposed cations do this by pulling an OH" ion or H20 from
solution and anions by attracting a proton from the aqueous
phase. The result is adsorbed H+ or OH- ions which assume their
respective charges on the surface.
Any other ion which can pass between the solid and
liquid phases may also help to establish the surface charge.
Such ions are called potential determining ions. Thus,
OH and H are potential determining ions for eachof the
OH and H are potential determining ions for each of the
three surfaces used here. In addition Ca and P04- are
potential determining ions for hydroxyapatite. Certain
ions added as impurities may alter the surface charge such as
aluminum ions  or cobalt  on silica or phosphate on
alumina . The wide variety of buffering systems used in
biological studies involving adsorption can thus lead to
differing results for proteins even if the same material is
used as a substrate.
The surface potential can be altered by a change in
pH. For each of the three oxide substrates in this study, there
exists a pH at which the surface charge is zero. This pH is
called the point of zero charge (pzc) and is listed in Table
1 [62,63] for the substances used as adsorbents.
As the pH varies from one side of the pzc to the
other, the sign of the surface charge will change as will the
adsorption properties of the protein.
Ions in solution which do not pass through both phases
but are attracted near the surface by electrostatic forces
are called counter ions. They will form a diffuse layer of
ions in solution near the surface and will tend to neutralize
the surface charge. The concentration of the ions generally
decreases exponentially with distance from the surface .
The.higher the concentration, the more compact the diffuse
layer will be. The thickness of the diffuse layer ranges from
about 10 A at .1 M solutions to about a few hundred A in .001 M
salt solutions .
PZC of Substances Used as Adsorbents
Due to the shape of the diffuse layer, the influence
of dissolved salts on proteins will be greater near a surface
than in bulk solution. Some proteins, because of their
large size, may extend entirely through a double layer. The
effect of dissolved salts on adsorbed proteins would then be
difficult to explain in detail.
The adsorption properties of the substrates are due
as much to adsorbed water as to their intrinsic stucture.
The -surface of silica in aqueous solution has been shown by
infrared spectroscopy to possess three types of surface ions
 as represented below:
Unless the adsorbed water and associated ions are driven off
by heating, silicon ions cannot chemically react with organic
molecules arriving from solution. It has been shown that
ammonia will not react with hydroylated silica, although
chemisorption will occur if the silica has been subjected to
a prior vacuum degassing at temperatures in excess of 400 C
[66,67]. Silica treated with ammonium fluoride solution
showed evidence for Si-F bonding instead of silanol  but
this reaction occurred after heating the substrate to 4000C
in vacumm.* Trimethylsiloxane can be covalently bonded to
silica by refluxing them in acetone for 24 hours at 500C .
This gives some idea of the difficulty of penetrating the
adsorbed layers on the silica surface. It can be seen that
the reactions of aqueods solutions of proteins with silica may
occur with either surface oxygen or hydrogen, depending on
the compositional purity of the surface.
The same effect can be seen with alumina. Steric
acid was adsorbed from CC14 solvent after the alumina had been
evacuated at 8000C for one hour. Without the pretreatment,
steric acid would not covalently bind to the alumina surface
. Methanol has been shown to adsorb on alumina  after
successive evacuation and heating at 4000C, heating in oxygen
to rid any hydrocarbons present and then heating again at
10-5 torr at not less than 3500C for 1/2 hour. A methoxide
surface is formed when the clean dry surface is exposed to
methanol vapor. From studies of adsorbed acetylene on alumina
it was concluded that the surface contains electron poor
and electron rich  sites (oxide ions, hydroxyl groups, and
aluminum ions) after the sample had been heated to 8000C.
Even at these temperatures not all the hydroxyl groups were
removed from the surface . Hydrogen and hydroxyl ions on
alumina are exposed to the solution interface, yielding a
surface structure similar to that of silica.
Hydroxyapatite is assigned the formula Cal0(PO4)6(0H)2.
In solution the surface undergoes hydrolysis, yielding a surface
having the formula Ca2(HP04) (OH)2 . It is different from
silica and alumina in that, in addition to surface OH- and H
ions, there are also Ca+2 ions which are capable of binding
adsorbing anions . The multiple internal infrared spectra
of several synthetic and naturally occurring calcium phosphates
exposed to organic acids show shifts in the P-0 stretching
frequency  which were attributed to hydrogen bonding. Other
workers  have shown that H30+ ions are hydrogen bonded to
the calcium. There have been suggestions that there is some
covalent bonding between organic constituents and hydroxyapatite
in bone . Binding of calcium ions by collagen has been
demonstrated by solution analysis . It has not been shown
that collagen can attach to the surface of the hydroxyapatite
without an intervening water molecule or hydroxyl ion.
Forces of Adsorption
The relationship between the enthalpy of a reaction and
the total energy is H = E + PV. Most biological processes occur
in liquids rather than in the gas phase . In this case
the changes in pressure and volume are small. To a good
approximation then, dE ~ dH.
The total energy of adsorption is affected by the type of
interaction between the surface and the molecule. This energy
is comprised of several components. They may be classified as
non-polar, ionic, hydrogen, and covalent bonding [77-79].
Non-polar dispersivee) forces are always present between
molecules. They arise because the time-averaged electron cloud
.interaction between uncharged atoms is attractive . They
are moderately strong, producing energies in the range of 1 to
10 kcal/mole. Hydrophobic bonding is a result of dispersive
forces. This is a consequence of a decrease in displace-
ment of a polar medium when less polar components coalesce,
thus creating a lower energy state.
Ionic or electrostatic attraction can occur between
oxides and proteins both of which are normally charged in
aqueous solutions. Ionic bonding strength is decreased by
an increase in the ionic strength of the solvent because of
shielding, whereas dispersion forces are not affected. Ion
interaction is essentially independent of temperature [80,813.
Hydrogen bonding is partially ionic and partially
covalent [82,83]. It arises from the electrostatic force
acting between hydrogen and a lone-pair of electrons of
nitrogen, oxygen, or fluorine. The small size and the close
approach of the hydrogen atom accounts for the partial (20%)
covalent character . Typical bond strengths are of the
order 1 to 10 kcal/mole. Hydrogen bonds are also weakened
by an increase in ionic strength of the solvent. Completely
covalent bonding rarely occurs in the adsorption phenomena
in which we are interested .
The classification of physical or chemical adsorption
is somewhat arbitrary [79,83]. If the adsorption is found
to be readily reversible and has an energy of the same order
of magnitude as the liquefaction of gas, it usually is classified
as physical [77 ]. Osipow states that Van der Waals forces
are responsible for physical adsorption whereas Fuerstenau
also includes coulombic attraction. Chemical adsorption is
irreversible and the magnitude of the energy change is of the
order of chemical reactions .
Thermodynamics of Adsorption
Complete reviews on the thermodynamics of adsorption
are given elsewhere [76-78]. In this section, only those
points needed to explain the following data will be presented.
Limited explanation of the ideas of earlier workers would
be in order, however.
Gibbs gave the first rigorous thermodynamic explanation
for why a given material should either adsorb or desorb at a
surface. He was able to predict the functional relationship
between surface tension and surface concentration and the bulk
concentrations of the surface-active solutes. His derivations
assume that substances tend to minimize the free energy of
the surface region by either becoming concentrated or depleted
there. As a result, it has been the free energy which has been
traditionally determined, and surface concentration measurements
are the most common method of doing this. Other methods, such
as the thermometric titration method , may be useful for
obtaining thermodynamic data.
The thermometric titration technique is an analytical
method in which the heat effect of a titration reaction is
used to measure the titer of a sample. It is applicable to
reactions of the type
RI + R2 P (1.1)
which entails a heat of reaction Q. 'In equation (1.1) Ri
refers to reactants i and P refers to the product. For single
step reactions an equilibrium constant, K, may be written
K = [P]
where the brackets denote concentrations. Through knowledge
of equation (1.1) and the reaction heat, Q, the enthalpy change
of the reaction and the equilibrium constant may be determined.
The method by which this is done will be detailed for adsorption
For an adsorption reaction equation (1.1) can be written
Su + R So
where R1 in equation 1.1 has been replaced by Su which is an
unoccupied site on a solid surface capable of adsorbing from
solution a reacting component R to produce an occupied site So.
The equilibrium constant is then written
K = [S ,
The concentration of occupied sites [Sol is equal to
the number of adsorbed molecules per unit area, Na/A, while
the concentration of unoccupied sites is equal to the total
number of sites N minus the number of occupied sites per unit
area, (N0 N )/A. The equilibrium constant is then
K = N
(No N ) ER]
Dividing;- through by Ns gives
K = N /N
(1-N /Ns) [R]
where 0 = Na/N0 is the fraction of occupied sites. Equation
(1.4) is the Langmuir adsorption isotherm [ ] and will be
used later. Its use requires that each site is occupied by no
more than a single molecule and that no two sites interact.
These conditions are satisfied when the concentration of R is
The concentration of reacting molecules is equal to the
number of moles of R in solution divided by the total volume
Nr/V. If No is the total number of moles of R on the surface
and in solution then N No Na. The equilibrium constant
r r a
can now be written as
K= Na V
(N Na) (N Na)
j r a
The enthalpy change for the adsorption reaction, AH, is
related to the reaction heat Q by
AH = -Q/Na (1.6)
where the minus sign denotes, by convention, that an exothermic
reaction (positive Q) will yield a negative enthalpy change.
Substituting equation (1.6) into (1.5), one obtains
-K = -QAHV (1.7)
(N AH + Q)(AHNo + Q)
To determine K, at least two adsorption experiments are completed
in which Q is measured but a set of values is usually completed
to determine the endpoint of the titration. The original
number of moles of reacting molecules, No, is varied, and No
is held constant by keeping the surface area of the adsorbing
substrate constant. Values for Ql, Q2, N1, N, and sare
recorded where N1 and N are values for Nr. The value of K is
1 2 r
assumed to be nearly constant if N1 is not too different from
N2 Equation (1.7) can then be solved for AH by using the
quadratic equation (1.8)
AH2 Ns (Q1o 2 Q2N) + AH Q1Q2 (N2-N1) +Q1Q2(Q2-Ql)=O
for the two sets of values. The total number of surface sites
can be estimated by dividing the total surface area by the
known cross-sectional area of the adsorbing molecule. This
method is valid as long as the surface concentration of adsorbed
molecules is low and lateral interaction does not occur. Alterna-
tively three sets of values can be used to eliminate N from
equation 1.7. Both methods give results = 10% of each other,.
The volume V is held constant. The value of AH is substituted
into equation (1.7) to find K, and therefore AG by using the
AGo = -RT In K
In order to complete the thermodynamic data (AGO =
AHo TASo) Ho must first be determined or approximated.
The number AH used in equations (1.6) to (1.8) is the total
enthalpy change for the reaction under experimental conditions
while AHo is the standard enthalpy change. Under suitable
conditions, AH can be shown to be = AH so that ASo can be
determined. Those conditions will now be explained.
The chemical potential /i or partial molar free energy
Gi of component i in a chemical reaction is defined by
'i = Gi = ('G/aNi)tp (1.10)
where G is the free energy change for all components in the
reaction. The chemical potential can be expressed as a function
of the activity ai of component i and of the chemical potential
in some reference state
^i = /iref + RTlnai (1.11)
The term RTlnai takes into account the energies of interaction
of component i with other components at a given concentration in
the mixture. The choice of reference state is quite arbitrary
and varies for experimental convenience. Generally, no matter
what reference state is chosen, the activity is expressed as a
function of the mole fraction component i, Xi, and a parameter
known as the activity coefficient fi
ai = xi fi
The activity coefficient approaches 1 in pure solutions for
the solvent while in dilute solutions of component i, fi
approaches a constant which may be greater or lesser than one.
The partial molar enthalpy is found to be
Hi = -RT2 ( lni/ T)pn (1.12)
or in view of equation (1.11)
H-i = -RT2(ln i ref/DT) RT2 ( n ai/DT)pn (1.13)
or = -RT2(OIn iref/T)n RT2 (~in fi/DT)p
Under the constant composition Xi does not vary with temperature.
The first term of equation (1.13) is the enthalpy
change for component i which would occur if the reaction was
held under reference conditions. The standard enthalpies, HO,
of pure compounds are the enthalpies of reaction of building
up those compounds from their elements under standard conditions
(P = 1 atm T = 298K). The chemical elements themselves have
zero standard enthalpies of formation. If the reaction is
carried out under standard conditions, the enthalpy change for
the reference state is equal to H?. Then
H. = H. RT (~ln f./ T) (1.14)
Knowledge of fi for the systems under study is lacking.
We therefore make the approximation that AHo is significantly
larger than the natural logarithm of the temperature variation
of fi. The validity of this approximation relies upon the non-
interaction of solute molecules. This condition is assumed to
hold for dilute solutions. Therefore
Hi = Hi (1.15)
To relate HO to AH it is noted that AH can be written
as the difference of the sums of the partial molar enthalpies
of products and reactants
AH = ZiHiX + kHkZk (116)
where each sum is taken over each of the different components
for products and reactants and X. and Zk are the respective
mole fractions. In view of equation (1.15) the enthalpy
change for the adsorption reaction is
AH = H X + EkHk Zk (1.17)
AH = AH (1.18)
where AHo is the standard enthalpy change for the complete
reaction. Thus, under the rather ideal conditions in which
there is no interaction between solute molecules in solution or
on the surface at 1 atm and 2980K AH may substitute for AH.
The value of AH can be used with AGo to find at least an approximate
value for AS0.
The experimental conditions in this work meet the contraints
of pressure and temperature. The constraint of non-interaction
of solute molecules holds only for dilute solutions. In as
much as enthalpy values tended towards.constant rather than
steadily decreasing values,lateral interaction on the surface
between adsorbed molecules does not appear to have occurred.
Interaction of solute molecules in solution can only be assumed
to be small in the concentration ranges used, typically 10- to
As mentioned earlier, from the standpoint of adsorption
studies, the surface of a substrate is often not uniform.
Heats of adsorption may vary at different positions on the
surface. If the condition is maintained that the different
sites are non-interacting, the adsorption onto a heterogenous
surface may be regarded as simultaneous independent reactions
of the type expressed by equation (1.3)
Sl + R Sol
S + R + So1
S2 + R So2
S + + R Soj
where the subscript j enumerates the different types of sites.
The concentration of a single type of solute molecules, R,
is common to all surface sites. Each adsorption reaction,
according to equation (1.19), would evolve a reaction heat
Qj. The total reaction heat, Qt, would be the sum of Qj for
each reaction on the different types of sites.
Qt Ql + Q2 + .. .Qj (1.20)
If AHi is the enthalpy change per mole of adsorbed molecules
on sites of type.j and Nj is the number of moles adsorbed then
equation (1.20) can be expressed according to equation (1.6) as
-Qt = AH1 N1 + AH2N2 + ....AHjNj (1.21)
In a calorimetric measurement it is the value of Qt
which is measured. If the solution is analyzed to determine
the total number of moles adsorbed, Nt a total apparent
enthalpy is found
AH = -Qt/Nt (1.22)
From equation (1.20) then
AHt = -Qt/Nt = 1/Nt (AH1 N1 + AH2 N2 +...AHj N )
AHt = AH1 Y + AH2 Y2 +....AH. yj (1.23)
where Yj is the fraction of the total number of moles of
molecules bound to sites of type j. Applying equation (1.18)
to each reaction
AHt = EjAH9 yj
where AHO is the weighted sum of the standard enthalpy changes
for all adsorption processes. The standard free energy
change for each reaction expressed by equation (1.19) is given
AG = AH? TAS? (1.25)
3 1 1
Rewriting equation (1.25) for A and substituting into (1.24)
AH = E.(AG9 + TAS9) y (1.26)
t i I I
= AC yj + TASjyj
AH = AGo + TASo
t t t
AGo = G. AG y.
L I 3 I
AS = ES? yj (1.28)
express the weighted sums of the standard free energy and
standard entropy changes for the complete adsorption reactions.
For each reaction expressed in equation (1.19) there
is an equilibrium constant Kj which can be written according to
equation (1.4) as
K. = 6. (1.29)
or related to AGQ according to equation (1.9) as
AG? = -RT In K. (1.30)
Equations (1.29) and (1.30) express the fact that each site
carries on an equilibrium reaction independently of all others.
The fraction 6. = N./N? is the ratio of occupied sires of type
j to the total number of sites of type j.
For any equilibrium concentration [R] of solute molecules
the total fraction of occupied sites of all types, 8t, can be
6 = N /No
t t s
S = X1 61 + X2 0 + ...X. (1.31)
where X. = N9/No is the fraction of sites of type j and is a
] 3 s
If equation (1.30) is substituted into equation (1.27)
AGo = -E. (RT In Kj) Yi
= -RT'ln K' K^ ..... K (1.33)
Thus, even though each AG? for the individual.reactions is
constant, the overall free energy change will vary as each
fraction y. varies.
The net effect is that if any single type site adsorbs
a large percentage of all molecules adsorbed then
AGt approaches AG? as v 1
Generally, however, there is no simple number AGo which can
be expressed in terms of a single equilibrium constant Kt,
having the form
K = KJ K2 ........KY (1.34)
t 1 2 3
The value of Kt would vary as the fraction of occupied sites
The physical interpretation is that each site contributes
a specific amount of energy to the total energy change. At
very low concentrations only a very few sites react, presumably
those of higher energy. At higher concentrations a greater
number of sites react, but of overall lower energy. The result
is a lowering in the average energy change as the concentration
Under the conditions of independently acting sites
evaluation of K would give AGo exactly. However, to do this,
knowledge of each Ki and yi has to be available. In the absence
of such knowledge, the evaluation of K can be approximated by
evaluation of equation (1.29) by replacing 6j by Ot. The number
found from this method, K', could be used for determination of
AGO under suitable conditions.
Those conditions may be determined by estimated
values of Kj for calculation of K' and Kt and using each for
evaluation of AG The value of K' is found from
K' = Ot
where 0t is defined by equation (1.31). The percent errors in
AGt is found from
% error AGt = n K In K' x 100
The percent error is calculated by setting values for K. and
Xj and directly calculating K' from equation (1.34) and Kt
from equation (1.40). If the percent error is acceptable, AGt
may be calculated.
An example for the case of two different types of
adsorption sites is given. The equations necessary to carry
out this calculation are given below for convenience
61 = K1 [R]/(l + K1 ER])
2 = K2 [R]/(l + K2 [R])
yl = 01No /(O1NO + 0 NO)
11 IN 1 2 2
y = 2N /(62No + 2NO)
Kt = K1 K2
X NO / (NO + NO)
X= N / (No + N
2 2 1 2
ot = X161 + X202
K' = t/(1 + t) [R]
In the example the values of [R] were carried over several
orders of magnitude while the total number of sites NO =
Nl + N2 was kept constant at 10-6 moles. The results are
shown in Figure 1
Two cases are presented. In (a) the fraction Ni/No
0.1 and K = 5000 are held constant, while K1 is given
values of 1000 and 5000. In (b) K1 and K2 are held constant
at 1000 and 2000 respectively while X varies over three
values; 0.002, 0.02, and 0.2.
While K2 is less than five times the value of K1 and
X1 is a good approximation of Kt with the error remaining
within 1%. As X1 approaches lthe error goes to zero. Also,
at very low concentrations, it is assumed that most molecules
would adsorb only onto the highest energy sites so all values
of y (equation 1.34) go to zero except Y and the error again
goes to zero. For oxides the overall fraction of highly
reactive sites is small [85,71]. Moreover, the hydrated
oxide surface will be of lower energy than a perfectly dry
surface, aiding in meeting the condition that K1 not be too
much larger than K2 [60,28]. Under these constraints and using
equations (1.8) and (1.7) with Q = Qt, AHt can be calculated.
For the calculations in the later chapters, the
difference between the values of NO for use in equation (1.8)
is kept small so that variation of AH in that concentration
interval is small. The values of AGo calculated tend to remain
within 20% of the highest to lowest values. This corresponds
S K2 = 5000
Sx = 0.1
10 K',Kt ( K1=50000) ,\ 4
9 (K = 50000)
( ) 8 -
(K1 = 10000)
5 K', Kt ( K1 = 10000) o
SK2 =2000 4
0 -x --"--" l =0 2 K1 =10000 0
.- ."' 2
(h) 6 0
4 Kt (.02)
I I I I I
7 6 5 4 3 2 1
Figure 1. Examples for the calculation of K' and Kt
for constant K2 and xl (a), and for constant K2 and K1 (b).
to a range of K. varying by about a factor of seven. Under
these conditions the maximum error in AGt is less than 2%.
The value of AS calculated from
ASo = AGo AH
can be given only simple interpretations. It is known that
values of AS will be between -20 and +20 cal/mole-deg [49,51].
A decrease in entropy is typically explained as a loss of
freedom of solute molecules as they adsorb. Increases in
entropy, generally found in experiments using macromolecules,
are explained as solvent molecules gaining additional freedom
as the largestructuring molecules are removed from solution.
Exceptions to this general rule are present.
The various plots of AG, AH, and AS presented in the
following chapters, in accordance with the previous discussion,
are to be understood as the composite values AGt, AHt, and
ASt. The values for these parameters are closer to single
values of AH?, AG?, and AS? in those regions of the curves
where they tend towards constant values. In these regions the
percent error is generally less than .5%.
The determination of the thermodynamic properties for
the adsorption of collagen on hydroxyapatite is presented as
an example of the calculations made in the following chapters.
The first step in the analytical procedure is to plot Q vs. C
(Figure 2). From this graph two values of Q and Co are chosen
for the sample calculation. In this case values of
I I I i
0,5 1,0 1,5 2.0
Figure 2. The first step in the calculation of the thermo-
dynamic functions is plotting the reaction heat, Q, vs the
original concentration Co. Here, two points are taken from
the measurements of the adsorption of collagen on alumina.
Q1 = 5.0 x 10"3 cal
Q2 = 5.9 x 103 cal
C = 1 x 10-6 M
C = 1.33 x 10-6 M
are arbitrarily chosen. The values of CI and C2 correspond
N = 2 x 10-9 moles
N = 2.66 x 10-9 moles
The number of moles of surface sites, No, in this example
is taken as 2.5 x 10-9. This number was determined from
consideration of the surface area occupied by a collagen mole-
cule, the number of sites as calculated by the program, and
study of the reaction heat curve. Substitution of these
values into equation 1.8 yields -3.3 x 106 cal/mole for AH.
Substitution of AH, Q1, No and V (4 x 10 -3 l)into equation 1.7
gives 4.3 x 105 for K or -7.6 kcal/mole for AGo. This in turn
yields -1.1 x 104 e.u. for AS0.
Calorimetric measurements were made using an LKB model
10700 batch microcalorimeter . The basic calorimetric unit
consists of two identical gold cells situated in an aluminum
heat sink (see Figure 3). Each cell has two compartments
capable of holding 2 and 4 ml of fluid. Mixing of the fluids
in each compartment is accomplished by rotation of the entire
calorimeter. There is no stirring, and after rotation, the full
Figure-3. Schematic of the operation of an LKB model 10700
microcalorimiter. A-calorimeter, B-reaction cell, C-thermo-
pyle, D-heat sink, E-amplifier, F-chart recorder, G- reaction
heat curve. The reaction heat, Q, is proportional to the
area under the curve.
4 ml of fluid are contained in the forward compartment. In
all experiments, 2 ml of fluid were used in each compartment.
Measurement of the heat loss or gain incurred by the
mixing procedure is made through multiple thermopiles located
between the heat sink and cells on two sides of each cell.
The thermopiles are connected in opposition so that the signals
from reactions producing equal amounts of heat are cancelled
electronically. One cell is then arbitrarily chosen as a
reaction cell and the other as a reference cell in which
unwanted heats can be cancelled.
Determination of the reaction heat is made by manual
integration of the voltage vs. time curve produced during the
course of a reaction. The energy is calibrated against a
known heat produced in the reaction cell using a precision
resistor and a known current generated for a specific time
Each reaction in this work is of the type
A BI A + B
where A is a solution of organic molecules and B is a slurry
of powdered oxide used as a substrate. In the mixing operation
several reactions are possible, each contributing to the over-
all heat produced. They are due to 1) wetting of the cell
wall; 2) dilution of the organic molecule; 3) friction of
mixing; 4) chemical reaction. Only the last heat is desired.
The others must be eliminated.
The first three heats are accounted for in various
ways. The cells are first wet with the solvent being used
and emptied prior to filling with reacting components. This
eliminates the heat of wetting of the cell wall. The heat of
dilution of the organic molecules is accounted for in the
reference cell, while the heat of dilution of the oxide powders
and frictional heat are measured separately and subtracted
from the reaction heat.
The heat measured in this way gives a measure of the
reaction A + B C where C is a complex of solid particulates
and adsorbed organic molecules.
Calibration of the calorimeter was carried out as suggested
by the manufacturer and consisted of two procedures. The first
procedure determined the sensitivity of the thermopiles. The
manufacturer listed the sensitivity of the thermopiles as 28.0
and 30.0 microvolts for a constant current of 30 miliamps
through the calibration heaters. The measured values were
consistently within 28.0 L .05 microvolts and 30.0 1 .05 micro-
volts. The second procedure judged the accuracy of the unit's
calibration mechanism. The heat of dilution of a six percent
sucrose solution was measured periodically to be 6.36 .05
kcal/mole. The literature value is 6.36 t .03 kcal/mole.
No literature data could be found for the heats of
adsorption of surfactants from aqueous solution on prewetted
surfaces. However the heat of adsorption of sodium dodecyl
sulfonate (SDS) measured from the heat of immersion of dry
alumina (Linde A) was estimated to be 12 kcal/mole 1 kcal/
mole . Using the method described in this work the
heat of adsorption of SDS on Linde A alumina was found to
be 10.2 kcal/mole of SDS. The discrepancy, -1 kcal/mole,
is thought to be due to the surface being wet prior to contact
with SDS. This would prevent any possibility of SDS coming
into contact with a drier, and presumably, higher energy
A Perkin-Elmer-Hitachi model 139 U.V. Vis spectro-
photometer was used for concentration determinations. Separa-
tion of particulates from supernatent solutions was carried
out by centrifuging or by filtering through micropore glass
filters. In those cases where filtering was used, the filter
was first saturated with a solution of the organic molecule to
be analyzed, then rinsed thoroughly. In all cases a standard
was used which had been put through identical procedures as the
Protein and polypeptide determination was made through
the use of Biuret reagent [87 ] and measuring at 550 millimicrons.
Carbohydrates were determined using a phenol solution and
measuring at 490 millimicrons . Amino acid concentrations
were determined by use of ninhydrin . Carboxylic acids
were titrated with phenothalein as the indicator.
Distilled water, having an initial pH near 7, was used
for preparing solutions. The pH was varied by adding HC1 or
NaOH. Three solutions were used as solvents: a low ionic strength
solution (LISS) in which the only ions present were those added
by pH adjustment, a 0.165 M salt solution, and a buffered solution.
The concentration of NaCI in the low ionic strength solution
did not exceed 0.001 M., The 0.165 M salt (Ringers) solution
consisted of 9 gm/l of NaC1, 0.25 gm/1 of CaC12, and 0.42 gm/1
of KC1 and is known. The buffered solution was a 0.2 M solution
of mono- and di-basic phosphate. The phosphate buffer was
used only in Ringers solution bringing the total molarity to
Except in those instances where solids were not used
at all, 0.1 gm solid particulates were added to the solution.
The solids were placed into suspension by mixing 1.0 gm of solid
powder with about 18 ml of distilled water, adjusting the pH
and then bringing final volume to 20 ml. One hour was allowed
for the pH to equilibrate. With stirring, 2 ml aloquots were
distributed into 10 tubes by pipetting. Using this method 0.1
gm 0.01 gm of solid was delivered to each tube.
The gold calorimeter reaction cells were washed daily
with detergent. The washing procedure included injecting a
solution of the surfactant into each compartment, rotating the
calorimeter and withdrawing the solution by aspiration. The cells
were then continually rinsed with distilled water while being
evacuated. By moving the tip of the aspirator tube from the
top of the cell to the bottom in one compartment, while filling
the other compartment, a good turbulent rinsing reaction
developed. Experience showed that five minutes of such procedure
cleaned the cell. Approximately once a week, the detergent
was left in the cells overnight to permit it to react more
Between daily experiments, the same procedure was used
to clean the cells. The oxide powder slurry was removed for
analysis after mixing, however, and the detergent was not used.
Cleaning with 1 M HC1 or NaOH was found to be necessary only
Organic constituents were weighed out to 1 -'.01 mg
and mixed volumetrically. Adjustment of pH was the same as
with solids. Incremental concentrations were made from standard
batches and measured volumetrically by pipet. Distribution of
organic, and solid, solutions into the reaction cell was made
by syringe. The lowest concentration was always used first.
The materials used as substrates include silica, alumina,
and tricalcium phosphate. The silica  was described by the
manufacturer as amorphous. It has. a specific surface area of
0.7 m2/gm and a pzc of 3. The alumina used  consisted of
two types: Linde A, a a-alumina of specific surface area 15 m2/gm
and a pzc of near 9 and Linde B, a mixture of y and a alumina
of specific surface area 82 m2/gm and a pzc near 9. The tri-
calcium phosphate, referred to as hydroxyapatite in the text,
consisted of 85 volume per cent hydroxyapatite and had a specific
surface area of 57 m2/gm. Surface measurements were made in
this laboratory using a multi-point B.E.T. nitrogen adsorption
isotherm. Measurements of pzc were also conducted in this
laboratory using a Zeta meterR .
All experiments were run at 250 C.
All solids were used without modification. Prior to
weighing, large (approximately 10 gm) batches of solids were
rinsed in distilled water, decanted and evacuated for 24
hours at 10'3 atm. Prepared powders were stored under vacuum
at room temperature.
All organic substances used were stored under refrigeration
prior to use. None was repurified or modified in any way.
Batch solutions were used within one week and were stored at
4C. Particular information on each substance used is given
in the appropriate chapter.
Reaction heat data were plotted against Co and fed into
a statistical analysis program available through the Northeast
Florida Regional Computing Center. The reaction heat, Q, was
held as the independent variable. The program generated an
approximating function which was used to calculate the thermo-
dynamic data. In all chapters, referral to "calculated thermo-
dynamic data" refers to this procedure. The graphs of Q versus
Co presented are original data.
ADSORPTION OF CARBOXYLIC ACIDS
ON ALUMINA AND HYDROXYAPATITE
Several of the amino acids which make up collagen
possess charged side groups, principally amines and carboxyl.
The polysaccharide, chondroitin sulfate, often associated with
collagen, also possesses the carboxyl groups as well as the
sulfate group. Because each of these charged species has the
potential to interact with an aqueous oxide surface, knowledge
of the type and strength of the surface reaction is sought.
The thermodynamic and adsorption behavior of molecules
containing the sulfate group have been previously described.
The adsorption of sodium dodecyl sulfate on alumina has been
studied by calorimetric techniques  and by solution depletion
techniques . Heats of adsorption for this molecule have
values near -6 kcal/mole at low concentrations when only ionic
forces drive the adsorption reaction. The free energy change
lies near -11 kcal/mole of ions. The adsorption of sulfate
ions onto solid barium sulfate from aqueous solution has been
measured by the thermometric titration technique. The heat of
precipitation was found to be -4.5 kcal/mole ..
The adsorption of alkyl ammonium acetate on quartz
has been followed as a-function of temperature and concen-
tration [92-93].It was found that at 25C at neutral pH the
isoteric heat of adsorption was between zero and 2 kcal/mole
of ions. The positive enthalpy was expected because of charge
repulsion of the quartz surface and the negatively charged
ions in solution. The free energy change at 250 remained near
-3.5 kcal/mole. In most cases the hydrocarbon chain of the
molecules caused abrupt changes in the thermodynamic properties
due to lateral interaction of the adsorbing molecules as the
equilibrium concentration increased.
Binding studies of sulfates, citrates, and amino acids
on calcium oxalate and calcium phosphate have been carried
out in relation to bone formation and kidney stone growth
[94,95]. Thermodynamic data are not readily available for these
reactions. Equilibrium constants of magnesium oxalate [94 ],
however, have been shown to be near 4000 which would correspond
to a free energy change near -5 kcal/mole for this precipitation
The purpose of this chapter is to discuss the variation
of the heat of adsorption of simple and polymeric carboxlic
acids on alumina and hydroxyapatite in relation to surface charge,
molecular conformation and ionization and solution pH. Qualita-
tive results indicate that while electrostatic interactions
are required to initiate adsorption, other interactions such
as hydrogen bonding, take place.
The sodium salts of acetic acid, oxalic acid, citric
acid, and polyacrylic acid (PAA), containing one, two, three
and multiple carboxyl groups were selected for study. The
different number of charged groups, different degrees of
ionization and molecular structure of each of these molecules
should provide a sufficient variety of detectable changes in
the calorimetric measurements. Analysis of the various changes
should furnish a clear understanding of the adsorption
The structural: formulas of each of the molecules used
in the work described in this chapter are given below:
Acetic Acid Oxalic Acid
H OH H H H
I \ I I I
HOOC C C C COOH -4 C C-
H I H H COOH
Citric Acid Repeat Unit of
Sodium salts of the carboxlic acids were purchased
from Sigma Scientific, Inc. . Poly(acrylic acid) in a
65% aqueous solution was purchased from Aldrich Chemical
Company  and was reported by them to have a molecular weight
of 2000. Linde A alumina, and hydroxyapatite were used as
Titrations of the carboxlic acids were carried out
with HC1 or NaOH. Determination of the amount of acid
adsorbed was based on calibration against known concentrations.
Other methods and procedures were described earlier.
Reaction heats for the adsorption of the three simple
carboxylic acids on alumina are presented in Figure 4. For
each acid, the maximum reaction heat was found to occur when
the solution was near pH 5. The reaction heats at pH 3 were
second and the lowest curves were recorded for pH 7. The
highest heat was recorded for sodium oxalate which also has
the lowest dissociation constant (see Table 2).
The enthalpy change upon adsorption was first found
by determining the amount of each acid adsorbed by titration.
The method was suitable only for pH 5. At pH 3 and 7 poor
precision resulted because of the small amount of each acid
adsorbed. Results are given below:
acetic oxalic citric
-AH (kcal/mole molec.) 4.2 4.8 4.6
(Co = .001M)
Thermodynamic data were calculated for pH 5 using a
smaller concentration range (C = 10'4 to 10-3 M) (see Figure
5). These results show that there are differences in the
enthalpy change for the three acids used. Each curve displays
a tendency towards more negative exothermicc) values at the
lower end of the concentration range. The heat of adsorption
S/ /' 3
20- /5 / 5
o/ ./ 3
I I I
4 3 2 1 0
Figure 4. Reaction heats for the adsorption of three
carboxylid acids on alumina.
of Carboxylic Acids
Acetic Oxalic Citric
4.75 1.23 3.14
A ACETIC ACID
0 OXALIC ACID
30 C CITRIC ACID 60
o 20 c C 0
-4- -. .- .
lies between -6 and -24 kcal/mole. The free energy changes
are nearly equal at about -5.5 kcal/mole, while the entropy
changes are each negative and lie between -10 and -20 cal/mole-
deg (1 cal/mole'deg = 1 entropy unit or e.u.).
Calculated values for the thermodynamic functions for
the adsorption of the three carboxlyic acids on hydroxyapatite
are presented in Table 3. The values show less variation for
hydroxyapatite than they did for alumina. The enthalpy
change tends to be less negative than for alumina, while the
free energy and entropy changes are about the same.
Polyacrylic acid (PAA) had such a high affinity for
alumina and hydroxyapatite that a cotton-like gel formed at
pH 5 and 7, causing great difficulty in cleaning the gold
reaction cells. Three concentrations were run with alumina,
however, and were repeated several times to improve precision.
Titration determinations of the amount of PAA adsorbed leads
to an enthalpy of 82 cal/gm of PAA for the adsorption reaction.
Taking 72 gm/mole as the molecular weight of the monomer, the
determined enthalpy change is -5.7 kcal/mole of acrylic acid
monomer assuming all acid groups participate in the adsorption
The calculated thermodynamic data for the adsorption
of PAA onto aluminaare presented in Figure 6. The enthalpy
change per residue lies between -7 and -12 kcal/mole, while
the .free energy change is concentration invarient at -5 kcal/
mole. The entropy change is negative as it is with the other
carboxylic acids used in this section.
Themodynamic Variables for the Adsorption of
Carboxylic Acids on Hydroxyapatite at pH 5 and 7
-AG -AH AS
(kcal/mole) (kcal/mole) (cal/mole-deg)
5 -- --"- .. ----- --.- i0
0.02 0.04 0.06 0.08
Figure 6. Thermodynamic data for the adsorption of poly-
acrylic acid on alumina at pH 7 in low ionic strength solu-
tion; AG----, AH-- AS.-.-
As molecules are adsorbed from solution, other
molecules will ionize to try to maintain the original concen-
tration. At the same time the alumina or the hydroxyapatite
will act as a buffer to maintain the pH. So long as the pH
is constant, the degree ionization of the solute will remain
the same. This leads to a qualitative explanation for the
reaction heat curves of Figure 4.
The total ionic charge in solution is directly propor-
tional to the degree ionization, a, which is related to the
pH by :
pH = pKo log [ (1 a)/a ] (2.1)
where pK is defined as
pK = -In [H+] [A-] (2.2)
This relation holds for single as well as polyelectrolytes
In the absence of any specific interaction between a
solid and electrolytes in solution, the surface potential is
given by [5b]:
S= RT In a/a (2.3)
where z is the valence (including sign) of the potential
ion, F is the Faraday constant, a is the activity of the
potential determining ion in solution and a is the activity
of the potential determining ion at the pzc. As mentioned
in the introduction, the potential determining ion for the
systems under consideration is H+. The surface potential
may then be written from equation (2.3) as
= 2.3 RT (log [H ] log [Ho]) (2.4)
where concentration are substituted for activities and
where [Ho] is the hydrogen ion concentration at the pzc.
The interaction due to the surface potential and the
charge, q, in solution gives rise to an interaction energy,
E = o dq (2.5)
where integration is necessary since the charge concentration
is a differential process. The surface potential affecting
ions in solution will decrease as saturation of the surface by
charged ions is approached. The total interaction energy is
nearly equal to Poq in this model if the concentration of charged
ions in solution is low so that there is little interaction of
the adsorbed ions on the surface. In this case the surface
potential which each ion encounters will be the same. The
energy lost by the ions is transferred to other ions and solvent
in the form of kinetic energy and flows as heat out of the
A plot of o from equation (2.4) and a from equation (2.1)
versus pH for a hypothetical monovalent acid with a pKo near 5
and an oxide with a pcz near 9 is given in Figure 7. At low
pH, a, and thus the charge in solution decreases. At high pH
the surface potential decreases. The highest interaction should
be expected to occur where o and a are not near zero.
2 4 6 8
Plot of 4o and a for a hypothetical acid.
Figure 8. Values of the reaction heat calculated by use
of equation 2.6 and found by calorimetric experiments.
For example, at pH 5 equation (2.4) gives for a
monovalent electrolyte '(H+)
to = .059 (9 5) = .24 volts
The total charge in solution, using acetic acid as the mono-
valent acid at .001 M is
q = a Ane (2.6)
= .12 coul.
where A is Avogadro's number, n is the total number of mole-
cules in solution, and e is the electric charge. The total
electronic interaction energy is
E = qo q
= (,24) (.12) = .028 j = 6.7 meal
The value of Q found from microcalorimetry is 4.7 meal. The
agreement is fair. Plots for other values of Q and E for other
acids at different pH values are shown in Figure 8 for C =
.001 M. Although the experimental and theoretical results do
not fit well for all cases, the simple model qualitatively
explains the experimental findings, which is what was sought.
Such factors, as treating the ions in solution as other than
point charges, and the potential at the Stern layer are not
taken into account. The most important result is that the
reaction heat and electronic interaction energy decrease as
either o or a decrease. The tentative conclusion develops that
electronic interaction will probably be the essential factor
in enthalpy changes measured for charged adsorbing molecules.
The enthalpy and free energy change values are within the
range expected from the investigations of other workers.
Some idea of the number of active groups of each
molecule actually on the surface can be obtained from the
following analysis. Using the data from Table 4and Table 5
the average number of ionized groups on acetic acid, oxalic
acid, or citric acid can be determined from their dissociation
constants. The numbers are given in Table 4. If we divide
the enthalpy change per molecule by the average number of
ionized groups. AH" is obtained. This assumes that the enthalpy
change for the adsorption of carboxyl groups is about the
same regardless of the molecule being considered. Results are
given in Table 5 for pH 5 and 7 at .001 M.
The values for AH* are fairly consistent in each case,
Considering the enthalpy change of the adsorption for acetate
as an arbitrary baseline, we can argue that if AH* for oxalate
or citrate had been much larger than that for acetate it
would have implied that more groups per molecule were participat-
ing in the surface reaction than expected from the degree of
ionization. Asis,AH* for oxalate is slightly lower and citrate is
slightly higher than the AH* for acetate at both pH values.
Oxalate is the most strongly dissociated of the three acids
implying, perhaps, a slightly stronger reaction with the surface.
It should be realized that a particular group cannot be partially
ionized at a particular instant in time. The invarience in
Ionized Groups per Molecule
for Three Carboxylic Acids
pH Acetic Oxalic Citric
3 0.02 1.0 0.42
5 0.63 1.85 1.9
7 1.0 2.0 2.8
Enthalpy Change per Mole, AH, and Enthalpy
Change per Mole of Ionized Groups, AH"
AH* is evidence that only ionized groups participate in
adsorption. This supports the supposition that electro-
static attraction is primarily responsible for adsorption.
The same procedure applied to the data for the adsorp-
tion of the three acids on alumina does not give such consis-
tent results. At pH 5 AH* lies between -6 to -11 kcal/mole
for acetate, -2 to -7 kcal/mole for oxalate and -7 to -3 kcal/
mole for citrate.
Near Co = .4mM (see Figure 5), a medium concentration,
we find that AH* lies near -9.8 kcal/mole for acetate, -4.1
kcal/mole for oxalate and -8.4 kcal/mole for citrate. This
indicates that while acetate has one, and citrate two ionized
surface groups, oxalate has on the average only one of its
two ionized groups on the surface.
It was argued earlier that only ionized molecules are
attracted to the surface and that the number of such molecules
drops as the surface charge drops. It should be recalled
that the surface charge on alumina and hydroxyapatite is
produced by the adsorbed H+ ions. This suggests the possibility
that an approaching ionized carboxyl group and surface hydrogen
ion participate in hydrogen bonding.
Zeta potential measurements  show that citrate
changes the surface charge of hydroxyapatite at low concentra-
tions. This indicates that electrostatic attraction by itself
is not the only factor in adsorption. If it were, when the
surface charge had been neutralized by adsorption of a sufficient
amount of citrate, adsorption would have ceased and the surface
charge would not reverse.
The single charged acetate ion does not reverse the
surface charge of hydroxyapatite or alumina, this would
indicate the absence of other than electrostatic interaction.
It will be recalled from the introduction that .close approach
of hydrogen to an anion is required for hydrogen bonding to
occur. Since, at pH 7, acetate has one and citrate has three
ionized groups, it is thought that multiple groups are required
to pull the molecule close enough to the surface for hydrogen
bonding to occur. The interplay between surface charge,
molecular size and charge density, ionic and hydrogen bonding
becomes apparent in these situations.
Polyacrylic acid has such multiply-charged groups. It
is known to be a linear molecule which is fully ionized at pH 7
[ ]; therefore, there are no hydrogen bonds to be broken due
to an unfolding of the molecule upon adsorption. The AH between
-5.7 and -6.9 kcal/mole of residues of PAA is close to that
found for the adsorption of the carboxyl group of the other
molecules on alumina and hydroxyapatite. The similarity in all
these instances implies that the same type interaction occurs,
and is not a strict function of surface composition or of
There appears to be no particular differences in the
enthalpy of adsorption of a carboxyl group onto alumina or
hydroxyapatite due to the number of groups on a molecule. In
each case the enthalpy change is near -6 kcal/mole and an
attracting force is required to initiate adsorption. Other-
wise a plot of E or Q versus pH would be similar in shape to
the a versus pH and not bell shaped.
Once adsorption occurs, the influence by multiply-
charged groups on the molecule was evidenced by a change in
surface charge. For this condition to arise, specific adsorp-
tion has to occur which requires forces other than electro-
static. The proximity of oxygen in the carboxyl groups and
H on the surface suggest hydrogen bonding.
ADSORPTION OF CHONDROITIN SULFATE
AND OTHER CARBOHYDRATES
Polysaccharides, notably chondroitin sulfate (CS),
which contain carboxyl and sulfate groups, are present in
dentin and enamel  and in bone [102,103]. Under proper
conditions of pH and ionic strength, these polysaccharides
will complex with collagen . In the presence of a
foreign surface, these polysaccharides, like other charged
molecules, will adsorb and react not only with collagen but
also with the surface. It is the purpose of the studies of
this chapter to explain the interaction of aqueous solutions
of chondroitin sulfate with alumina, hydroxyapatite, and
silica. A later chapter will discuss the interaction of CS
Chondroitin sulfate is a polysaccharide made up of
basic dimer units of glucoronic acid and galactosamine. Several
simpler carbohydrates were chosen to model CS: (a) galactose,
glucose, D-acetyl galactosamine; (b) glucose-6-sulfate (G6S),
D-galacturonic acid; (c) dextran, and (d) polygalacturonic
acid (PGA). Each carbohydrate was chosen because it possessed
a single feature of the CS molecule: (a) the carbohydrate
residue; (b) a charged carbohydrate; (c) the polymer back-
bone structure; and (d) the charged polymer. The structural
formulaefor these molecules are shown in Figure 9a and 9b.
All carbohydrates were purchased from Sigma Chemical
Company  and used without further purification. The
chondroitin sulfate (sodium), dextran, and polygalacturonic
acid were reported to have molecular weights of 45,000, 60-
90,000, and 25,000 respectively. The chondroitin sulfate was
determined to be chondroitin-6-sulfate by infrared spectroscopy
using the KBr pellet technique, and had a molecular weight of
45-60,000. The materials used as substrates, Linde B alumina,
hydroxyapatite, and silica, and the low ionic strength solution
were described previously.
The first set of experiments was conducted with
galactose, glucose, and dextran (see Figure 9b). None of these
molecules possesses charged groups. By measuring the heat of
adsorption of these molecules on the oxides, the contribution
to the total enthalpy change on adsorption of uncharged
carbohydrate monomer and polymer could be estimated. The initial
concentration, C was varied from .01 to .1 moles/liter. By
comparison with the reaction heats produced by the carboxylic
acids in this concentration range, it was estimated that 20 to
40.meal would be considered a significant reaction. The results
OH / 3
.------ 28.7 A .
Figure 9. Molecular structure of chondroitin sulfate (a)
( D-glucoronic acid N-acety galactosamine-6-sulfate)
(b) the three-fold helix of chondroitin sulfate.
HO- C H
HO- C H
Figure 9b. Molecular structures of the carbohydrates
used in the experiments of this section.
are presented in Figure 10. The maximum reaction heat, Q,
produced was about .3 mcal at a concentration of .1 M on
alumina. The enthalpy change, as determined by solution
depletion, was approximately -100 cal/mole adsorbed for galactose
and glucose on alumina. For dextran, AH was about -20 cal/
mole of residues (.11 cal/gm) and -15 cal/mole of residues
(.8 cal/gm) on hydroxyapatite.
The enthalpy change for the adsorption of D-acetyl
galactosamine on alumina was found to be -430 cal/mole of
molecules. Using the thermometric titration method described
earlier, the free energy change was found to be -5.2 kcal/mole.
The enthalpy change for the adsorption of this molecule on
hydroxyapatite was found to be -400 cal/mole of molecules.
Calorimetric measurements for the adsorption of D-
galacturonic acid on alumina indicated a stronger reaction than
with the uncharged carbohydrates. The reaction heat, Q, reached
a maximum of 8.9 meal. The calculated enthalpy change and
that determined by solution analysis are presented in Figure 11
and lie near -10.5 kcal/mole. There is good agreement for the
enthalpy change using both methods except in the lower concentra-
tion range where the calculated values are more negative. The
free energy change varies between -4.6 and -5.5 kcal/mole
adsorbed. The entropy change is negative and lies between -8
and -30 cal/mole*deg per molecule adsorbed.
Measurements of the adsorption of D-galacturonic acid on
hydroxyapatite showed a similar enthalpy change to that on
o I I I
.02 .04 .06 .08
Figure 10. Reaction heats for the adsorption of dextran o ,
galactose ,, and glucose o, on alumina (closed symbols) and
hydroxyapatite ( open symbols) in low ionic strength solution
at pH 7.
- :- -- --Z --
Figure 11. Thermodynamic data for the adsorption of D-galac-
turonic acid on alumina at pH7 in low ionic strength solution;
AG--- AH- AS---. Enthalpy change determined by solution
alumina, as determined by solution depletion (see Figure 12).
Values for AH lie between -6 and -8 kcal/mole.
The adsorption of D-galacturonic acid on silica is
endothermic. At low concentration, the enthalpy is +700
cal/mole, becoming more positive at higher concentrations.
The amount adsorbed was determined by solution depletion. The
free energy change and entropy change were not calculated.
The change in enthalpy for the adsorption of glucose-6-
sulfate on alumina, hydroxyapatite, and silica was determined
to be -7.6, -5.4, and 0.3 kcal/mole of adsorbed molecules,
Calorimetric measurements for the adsorption of poly-D-
galacturonic acid on alumina and hydroxyapatite were hampered
by agglomeration of the particles by the polymer. The thermo-
dynamic functions for this reaction were calculated and are
shown below.. In the concentration range used, .001 M to .01 M
of residues, these values were constant (1 0.1 kcal/mole).
AG AH AS
(kcal/mole) (kcal/mole) (kcal/mole/
alumina -3.4 -2.54 3.5
hydroxyapatite -4.2 -.31 15.0
In contrast to the decrease in entropy found for the
adsorption of PAA on alumina or hydroxyapatite, the entropy
change is positive for adsorption of poly-D-galacturonic acid
on both alumina and hydroxyapatite indicating an over-all
decrease of ordering.
o -2 -
Figure 12. Heat of adsorption for D-galacturonic acid on
silica-e-- and hydroxyapatite -- in low ionic strength
solution at pH 7.
The results of the calorimetric measurements of the
adsorption of chondroitin sulfate (CS) on alumina are
presented in Figure 13. Comparison of the enthalpy of
adsorption for CS on silica, alumina, and hydroxyapatite are
The enthalpy changes for the adsorption of CS on the
uncharged hydroxyapatite is found to be 15-30% more negative
than that for the positively charged alumina. The enthalpy
change for adsorption of CS on silica is found to be positive
endothermicc) and was determined by solution depletion, the
free energy change AG was not calculated.
Glucose, galactose, and dextran are uncharged molecules.
Because of the availability of hydroxyl groups it is possible
that these molecules can undergo hydrogen bonding with an
oxide surface. The low value of the enthalpy change for the
adsorption of these molecules, however, does not indicate very
strong reaction with the oxide surface in comparison with the
S 4 .. 20
0 I I
0.2 0.4 0.6 0.8
Figure 13. Thermodynamic data for the adsorption of chon-
droitin sulfate on alumina in low ionic strength solution
at pH 7; AG----, AH- AS--.
The only difference in the experimental conditions
in using the uncharged versus charged molecules is
attributable to the functional groups. Therefore the large
differences in AH observed are seen as due to the presence of
the charged groups.
If it is assumed that, in the case of dextran, only
two or'three points of contact are made per molecule, then
the enthalpy change could be on the order of 3-4 kcal/mole
molecules. Comparison of dextran with poly-d-galacturonic
acid (PGA), however, still shows that dextran is much less
strongly bound than PGA.
In the concentration ranges used, 10-6 M dextran
molecules (not residues), there is little hydrogen bonding
of dextran chains to one another [76,98]. Therefore, the
breaking of interchain hydrogen bonds should not contribute
substantially to the low enthalpy change actually measured.
If the hydrogen bonding does take place between the oxide sur-
face and many dextran residues, it is not manifested in the
measured enthalpy change. In the absence of other interactions,
it is concluded that the charged groups of the carbohydrate
molecules are necessary to provide sufficient attraction of
the entire molecule to the surface.
The increase in the heat of adsorption of D-acetyl
galactosamine over that of galactose is attributed to the
presence of the NH2COCH3 side chain. The exact cause can only
be speculated. Perhaps the nonpolar methyl group is forced
from solution by more polar solvent ions, drawing the molecule
to the surface. Whatever the mechanism, if these uncharged
molecules are strongly bound to the surface, it is not
reflected in the enthalpy determination. The type of bond
which would occur would almost certainly be hydrogen.
In any event, the adsorption of D-galacturonic acid
on alumina and hydroxyapatite is much more energetic than that
of galactose or galactosamine. The similarity in the molecules
and in the adsorption experiment strongly suggests that the
charged carboxyl group is responsible for the higher enthalpy
change and that binding between solute molecules or desolvation
effects do not account for the noted change. The negative
entropy change indicates an overall increase in ordering. This
increase in ordering may be due to confinement of the carbo-
hydrate molecules to the surface and subsequent loss of freedom
Adsorption of D-galacturonic acid on silica produces a
positive enthalpy change. This can be accounted for by the
charge repulsion which exists between the surface and the molecule.
There was a finite amount of acid adsorbed, however. This
would indicate perhaps a second stronger force necessary to
overcome the charge repulsion or that the negatively charged
molecules are occupying the fewer positive siteS on the silica
surface, or reaction with high energy sites. Since the reaction
heat and adsorption measurements leveled off quickly, the last
two possibilities appear more likely; especially in view of the
finding that charge attraction appears necessary for strong
Likewise the adsorption and calorimetric measurements
of glucose-6-sulfate suggest that the presence of a charged
group on the carbohydrate, opposite to that of the surface,
is required for stronger (more exothermic) reactions. The
negative sulfate group is attracted to the positive alumina
surface. As with'D-acetyl galacturonic acid on silica, the
adsorption of glucose-6-sulfate on silica is endothermic.
Poly-D galacturonic acid is obviously strongly attached
to the alumina and hydroxyapatite surfaces. The enthalpy
change is more negative for alumina, than for hydroxyapatite,
demonstrating the greater attraction for this surface. If
we consider that the enthalpy change is produced by the
charged groups bonding to the surface, then, using a figure
of -9kcal/mole as the enthalpy change found for the adsorption
of D-galacturonic acid on alumina, we can estimate that one
in three residues bonds to the surface. For hydroxyapatite,
this figure is perhaps one in twenty or thirty.
The chondroitin sulfate molecule is known to exist in
a threefold or eightfold helix [105,106] which is rigid in
solution  (see Figure 9). It possesses the ability to
change the conformation of positively charged polypeptides
from an extended coil to a helical structure (see Chapter 5).
The sulfate group extends further away from the carbohydrate
backbone than does the carboxyl group which is located on the
other side of the same dimer. The acetyl amine group extends
slightly further away from the backbone than does the
carboxyl and is located on the same side of the backbone.
From these considerations--helical structure, position of
the charged groups, and possible steric hinderance--it is
reasonable to assume that not all the charged groups participate
in bonding with the surface at the same time,
To help analyze the binding of CS to a surface, consider
that the interaction of a single dimer with the surface
permits interaction of both the carboxyl group and sulfate
group with the surface. Both groups would then contribute to
the enthalpy of the reaction. From the data in this chapter
and the previous one, it is seen that the change in enthalpy
is fairly constant for each type molecule, as it is between
carboxyl groups and that only charged groups contribute
significantly to the reaction heat, Q. The reaction heat due
to the adsorption of a carboxyl and sulfate group would be
between -12 and -16 kcal/mole. The measured enthalpy value is
about -2 kcal/mole of dimers for alumina and between -2.2 and
-2.8 kcal/mole for hydroxyapatite. Dividing the total enthalpy
possible by the measured value would indicate that between one
in three to one in seven dimers interact with the surface.
We may assume that only one of the charged groups inter-
acts per dimer. Using an enthalpy change between -6 and -12
kcal/mole of charged groups then one in three to one in five
groups would be indicated as interacting with the surface.
From the physical picture and the calorimetric data it seems
plausible to conclude that the.chondroitin sulfate molecule is
positioned horizontally on the surface with approximately
one out of every four dimers on the average coming into
contact with the oxide surface. In this instance, the charged
group interacting with the surface would be the sulfate group
since it extends further away from the CS backbone.
The entropy change due to adsorption of CS on alumina
is positive, indicating an increase in entropy or a decrease
in the order of the system. Since the molecule is rigid in
solution and is not likely to greatly change conformation on
the surface, no entropy contribution is attributable to a
change in shape. An increase in entropy can be attributed to
a release of solvent molecules from around the molecule or
from the surface into solution [106,107].
The adsorption of CS on silica can also be explained by
an increase in entropy. The surface charge on the silica is
the same as that on the carbohydrate. Overall, there is an
electrostatic repulsion between the surface and the charged
molecule. There must be some other energy supplied to over-
come this repulsion. Since it is the free energy change which
drives the reaction, and AH is positive, there must be at
least an equivalent positive entropy change so that TAS is
greater than AH. This entropy change, as suggested above,
can be supplied by the solvent ions.
It is difficult to speculate on the conformation of
CS on the silica surface as was done above with alumina. This
is so because the carbohydrate monomers are not attracted to
the surface of the silica as they are to alumina because of
In this section we have determined some of the thermo-
dynamic features of the adsorption of chondroitin sulfate on
alumina, hydroxyapatite, and silica by the use of model
carbohydrates. The results show that for positively charged
alumina the enthalpy change for the adsorption of charged
carbohydrates is about the same as that for the carboxylic
acids and lies between -7 and -9 kcal/mole of adsorbed species.
In these experiments the entropy change is negative and the
enthalpy change forms the major portion of the free energy
change. The enthalpy change for the adsorption of the charged
monomeric carboxylated carbohydrates on hydroxyapatite is close
to -7.6 kcal/mole. The adsorption of carbohydrates containing
the sulfate group on alumina or hydroxyapatite is about -5.4
kcal/mole. The similarity in enthalpy for the reaction suggests
a similar type reaction. The adsorption of charged carbohydrate
monomers on silica is weak and endothermic. The uncharged
monomers produce an enthalpy change only a fraction of that of
the charged monomers.
It was found that a model of the uncharged polymer backbone
of CS does not produce a large reaction heat or an enthalpy
change, suggesting that the presence of charged groups is
required to enhance the adsorption reaction. The data on
adsorption of polygalacturonic acid supports this conclusion.
Polygalacturonic acid adsorbed strongly, producing
agglomeration of the solid particles at high concentration (.1 M).
The enthalpy change per residue is lower than that found
for its monomeric counterpart. The conclusion drawn here,
as with polyacrylic acid in the previous chapter, is that
fewer points of contact are made, but that each point of
contact contributes essentially the same heat change as the
monomer. The entropy change was positive further increasing
the driving force.
Chondroitin sulfate was shown to adsorb to each of the
oxide powder substrates. The negative enthalpy change for
alumina and hydroxyapatite indicates that adsorption is strongly
enhanced by the opposite charges. Since the chondroitin sulfate
molecule is comparatively bulky relative to the models used,
fewer points of contact would be expected. The thermodynamic
calculations show, however, that perhaps as much as one-third
to one-fourth of possible bonding sites touch the surface.
The results for CS on silica are more speculative. The
enthalpy change is positive and it.is assumed that the major
contribution to the free energy change of adsorption is from
a positive entropy change related to the molecule size. Smaller
molecules were not found to produce a positive entropy change.
Relying on the results of the previous chapter, it
is assumed that once the molecules are attached to the surface,
that hydrogen bonding will take place.
In a subsequent chapter, we will investigate the type
interaction which CS and other molecules undergo with collagen
and collagen models. In the next chapter a study of the adsorp-
tion of molecules possessing amine side groups is discussed.
ADSORPTION OF POLYPEPTIDES
The calorimetric measurements discussed in the previous
chapters were related to the adsorption of molecules which
possessed a carboxyl group. The results showed that the mole-
cules on which the carboxyl group was ionized displayed greater
reaction heat and adsorption density than those molecules
which were not ionized. In this chapter, molecules containing
charged amine side groups are studied for the possible informa-
tion they can give on the adsorption of collagen onto silica,
alumina, and hydroxyapatite.
Several of the primary amino acids and their respective
polymers were investigated as collagen models. The molecules
used were alanine, poly-l-alanine (PA), proline, poly-l-
proline (PLP), poly-1-hydroxyproline (PLHP), lysine, poly-1-
lysine (PLP), and poly-l-arginine (PLA). Lysine and arginine
and their polymers possess basic side chains. The structural
formulaeof the monomers are given below. The polymers are linked
at the carboxyl and amino groups.
CH3CHCOO- ( COO-
+NH3 H H
H H NH2
The amino acids are dipolar ions. For the dibasic
amino acids, arginine and lysine, adsorption onto negative
surfaces will be enhanced. In polymeric form only poly-l-
lysine and poly-l-arginine retain any charge in neutral
The solubility in water of the other amino acids will
decrease as a result of their polymerization. Acidic amino
acids, aspartic acid and glutamic acid were not studied because
the carboxyl group has been discussed in the previous chapters.
Furthermore, interpretation of calorimetric data would be
difficult because of the presence of three charged groups.
The amino acids used in this section were purchased
from Sigma Chemical Company. Both amino acid monomers and
polypeptides were chloride salts, except for poly-l-lysine,
which was a bromide salt. The molecular weights of the poly-
peptides were reported to be: 1,000-5,000, poly-l-alanine;
15,000-50,000, poly-1-arginine; 70,000, poly-l-lysine;
10,000-30,000, poly-l-proline; and 10,000-30,000, poly-l-
The substrates and the solvent are the same as used
in the previous chapter. A few experiments were performed in
the .165 M salt solution (and will be indicated as such in
Mixing and calculation procedures were described
Calorimetric and adsorption results for alanine and
PA on the three oxides are presented in Figure 14. The
amino acid monomer adsorbs strongly on all three surfaces at
this pH. It acts much as the carboxylic acids do. There is
no sharp endpoint in the Q vs. C curve for the monomer, but
the slope of the surve is greater than that for the polymer.
The enthalpy change for each surface tends towards -5 to -8
kcal/mole of monomers.
3 alanine A1203
PA Al 203
I p I
0.0 0.1 0.2 0.3 0.4 0.5
Figure 14. Heats of adsorption for alanine and poly-l-alanine
(PA) onto alumina, silica, and hydroxyapatite (TCP) in low
ionic strength solution at pH 7. run in .165M salt solution.
When the dissolved salt (R-C1) concentration in the
solvent is increased to .16 M, the enthalpy change of the
adsorption of alanine on alumina decreases in the higher
concentrations range of alanine, but tends toward the -6
kcal/mole in the lower end.
The polymer, PA, exhibits much different behavior
showing no specific tendencies to adsorb. Reaction heats are
less than .5 kcal/mole and enthalpy changes are of magnitude
less than -1 kcal/mole.
Measurements made with proline and PLP and hydroxy-
proline and PLHP (see Figure 15) show results similar to
results for alanine and PA. The monomer again produces a
higher enthalpy change than the polymer (-4 to -8 kcal/mole).
However, PLP and PLHP are apparently more strongly attracted
than the poly-l-alanine with heats of adsorption lying between
-0.5 and -2.5 kcal/mole of residues. PLHP is somewhat more
strongly attracted on all three surfaces than is PLP. An
increase in ionic strength is noted by a decrease in the
enthalpy change for PLHP on alumina.
Lysine and Arginine
The addition of charged side groups causes a marked
change in the enthalpy curves, as shown in Figure 16. The
heat of adsorption of lysine on silica is, as expected,
1 4. HP-HA
2 9. PLHP-A*
.02 .04 .06 .08
Figure 15., Heats of adsorption for proline (P), hydroxyproline
(HP), poly-1-hydroxyproline (PLHP), and poly-l-proline (PLP),
on alumina (A), silica (S), and on hydroxyapatite (HA) in low
ionic strength solution and Ringers solution (*).
0.2 00.4 0.6 0.8
Figure 16. Heats of adsorption for lysine (L), poly-1-lysine
(PLL), and poly-1-arginine (PLA), on silica (S), alumina (A),
and hydroxyapatite (HA) in low ionic strength solution at
pH 7. Values determined by solution analysis -e- others
greater in magnitude than on alumina and is more pronounced
in the lower concentration end. The enthalpy change lies
in the range of -2 to -8 kcal/mole of monomers. The greatest
enthalpy change, however, for the polypeptides was recorded
for PLA on alumina. The second largest was on hydroxyapatite
and third on silica. This order also happens to be the order
of decreasing specific surface area. The enthalpy of adsorp-
tion of PLA was greater than that for PLL on hydroxyapatite.
The free energy change for PLA and PLL lie in the
range between -4 and -7 kcal/mole of residues (see Figure 17).
There is a sharp decrease in the free energy change of PLA
on hydroxyapatite at low concentrations. The entropy changes
for these molecules are small because of the similarity of
AH and AG.
The higher enthalpy change for the adsorption of
alanine, compared to poly-l-alanine, is due to the electro-
static attraction of the amine group or the carboxyl group
to the surface. PA, having no charge except for its terminal
groups is not strongly adsorbed despite its greater molecular
weight. PA is in a helical form , not coiled.
Since the molecule is uncharged,.the reaction heat of
the entire molecule due to adsorption only would be small in
S/ PLL-SiO 2
4 PLL-Al /
I I I I
0.0 0.2 0 .4 0.6 0.8 1.0
Figure 17. Free energy change for the adsorption of poly-1-
lysine (PLL) and poly-l-arginine (PLA) on silica, alumina and
hydroxyapatite at pH 7.
comparison with ionized molecules . A large conformational
change would then cause the overall reaction to be endo-
thermic. The molecular concentrations are low. Therefore,
breaking of intermolecular hydrogen bonds should contribute
little to the enthalpy change.
The charged monomers of proline and hydroxyproline
are also more strongly attracted to the oxide surfaces than
their polymers (Figure 15). This is taken as a result of
electrostatic attraction. The heat of adsorption, measured
by solution depletion and found to be between -4 and -6 kcal/
mole for both monomers, results from reaction of the charged
groups with the oxide surface.
Calorimetric measurements for the adsorption of PLP
and PLHP did not show a specific pattern for any of the
surfaces. Both of these molecules have a helical conformation
in solution . There was no attempt to determine whether
or not this structure was grossly disturbed upon adsorption,
or if it was, what contribution to the enthalpy change such
a disruption would make. Neither was there an attempt to
determine how many points of contact were made. The definite
conclusions which can be drawn from these data are relatively
few. There are some reasonable assumptions, however, that can
be made which, if accepted, will further explain the situation.
Since these molecules are in an extended conformation,
not coiled, there are no intramolecular hydrogen bonds. At
low concentrations (2 x 10-5 M of residues) there should be
little intermolecular hydrogen bonding . The disruptions
which would primarily occur, then, would correspond to
rotational movement of the molecule , There is no
reason to suspect that these molecules should undergo grotesque
distortion on the surface since there are no strong attractive
forces. Therefore, the contribution to energetic changes due
to conformation alterations should be small.
Poly-l-proline and poly-l-hydroxyproline possess a
ring structure which is relatively nonpolar compared to the
polar solvent. Because of this, it is plausible to assume
that these less polar structures are in a lower energy state
on the surface, rather than in the solution. This is termed
hydrophobic bonding and could possibly account for the energetic
changes measured if most of the residues were near the surface
and not surrounded by the mobile polar ions in solution.
By comparison with the carbohydrates studied earlier
and in absence of detailed information on the geometric
smoothness of the surface, it would also be possible to suggest
that only a few points of contact are made on the surface .
Each of these contacts would assume a higher energy than
indicated by the average of 1-2 kcal/mole of residues measured.
If these contacts are hydrogen bonds made up of hydrogen
atoms on the ring structure and oxygen atoms on the surface,
then each bond would entail an energy change of about
-7kcal/mole. On the average then one out of 7 residues
would be in contact with the surface.
Comparison of the adsorption of these uncharged
molecules with those that possess charged functional groups
indicates that the role of PLP or PLHP would be minor in
comparison. Although the mechanism of adsorption has not
been fully explained in this case, there should be little
doubt that when positioned next to a charged molecule in a
peptide chain, the latter will play the dominant role in
adsorption to an oxide surface.
Lysine and Arginine
The adsorption heats of lysine on silica and alumina
are similar (Figure 16). Since this molecule possesses two
basic and one acidic group at neutral pH, it is reasonable
to assume that the charged carboxyl group is attracted to
the positive alumina surface, and that the positive amine
groups are attracted to the negative silica surface. Because
of the more basic properties of this molecule, it might be
suspected that the reaction with the silica surface would be
somewhat stronger than with the alumina surface. This appears
to be the case.
Poly-l-lysine and poly-l-arginine are known to exist
in an extended charged coil conformation in solution at
neutral pH [8,9]. If the coils, which are stabilized by
hydrogen bonding, were to break down, the enthalpy change
would be due to both the adsorption and the unfolding processes.
Enthalpy changes measured in this and the two previous chap-
ters indicate that enthalpy changes between -6 and -8
kcal/mole of single, charged groups are to be expected. The
magnitude of the unfolding process, the breaking of hydrogen
bonds, would lie between 5 and 7 kcal/mole [19,82], an
endothermic process. The resulting enthalpy changes for both
processes would be comparatively small with a value near 0
kcal/mole. Such a situation is encountered in experiments
presented in the next chapter where the coil-to-helix
transition is known to take place. Instead,the enthalpy
change is much more negative, decreasing in magnitude from
-14 kcal/mole to -6'kcal/mole. The variation is thought to
be due to adsorption on the fewer negatively charged sites
on the alumina and hydroxyapatite surface. These sites possess
a distribution of high to low energy within their own group.
Adsorption of PLL or PLA to neutral high energy sites
can be eliminated since PA, PLHP, and PLP each have heats of
adsorption which are smaller in magnitude. If the adsorption
had not depended on surface charge the neutral molecules
should have been just as strongly attracted to the surface.
The order of decreasing enthalpy change (alumina >
hydroxyapatite > silica) corresponds to the specific surface
area of the solids. The higher specific area of the alumina
(87m2/gm) provides more edges and peaks which are assumed to
form high energy sites. For equal amounts adsorbed, a
greater fraction of adsorbed molecules would be on these
sites for alumina, than for either hydroxyapatite or silica.
Negatively charged polysaccharides also displayed an increase
in enthalpy change at lower concentrations attributable to
high energy sites. More than a single species of these
specially adsorbing areas appears likely . The free
energy changes are of the same order of magnitude as those
found for the carboxylic acids and polysaccharides.
As found in the previous chapters, molecules with
charged ionic side groups react more energetically with oxide
surfaces than those without. The enthalpy change per
charged group lies between -4 and -10 kcal/mole. For those
polyamino acids which possess no charged side groups, the
magnitude of the enthalpy change is found to be less, near
-1 kcal/mole of residues. Although the mechanism for uncharged
molecules is not clearly defined, it is believed that conforma-
tional deformation does not contribute significantly to the
enthalpy changes measured. It is possible than an uncharged
polyamino acid could be bound to an oxide surface by a few
relatively high energy contacts. Polyamino acids with charged
side groups, however, will play the dominant role in adsorption.
REACTION OF PEPTIDES AND
CARBOHYDRATES IN SOLUTION
In the earlier chapters, an investigation of the
adsorption of molecules onto oxide surfaces has been discussed.
These calorimetric studies of compounds which are models for
collagen provided some information on the state of the adsorbed
molecules, the energy changes on adsorption and the type of
interaction they undergo.
In vivo, it is unlikely that a collagen molecule will
come into contact with a clean surface. In general, there
will be other substances present in the body fluids which
will adsorb first because of factors such as greater concentra-
tion. Also, collagen may not be present at all when the
hydrated oxide surface is first exposed to body fluids .
We should have some indication, therefore, of how these adsorbed
molecules will affect the adsorption of collagen. In order
to understand their interaction at a liquid-solid interface,
it would be helpful to first investigate their .interactions
It is the purpose of this chapter to provide further
insight to the interaction of collagen, chondroitin sulfate,
and serum albumin in solution.
The same compounds which have been used previously
to model collagen and chondroitin sulfate have been used
here. For collagen they are poly-l-arginine (PLA), poly-l-
lysine (PLL), poly-lalanine (PA), and poly-l-proline (PLP).
For chondroitin sulfate they are dextran, galactose, galacturonic
acid, and polygalacturonic acid.
The reaction of chondroitin sulfate with collagen has
been studied by model systems, as mentioned in earlier
chapters. The reaction of collagen with dyes containing acidic
groups and with CS have been studied in regard to the under-
standing of the role of CS and collagen in connective tissue
. It was found that the cationic groups of collagen bond
with the anions of the dyes and CS in a pH range of 1.5 to
7 with a sharp drop in the number of anions fixed below pH 2
and a more gradual decrease from maximum adsorption at pH 3
to zero at pH 7.
In experiments to determine the role of CS in the
calcifying mechanisms of bone , it was found that calcifi-
cation would not occur or would occur more slowly in an aqueous
collagen mixture when CS was not present. These experiments
were performed near pH 7. The authors suggested that binding
of CS to collagen at neutral pH would aid the natural calcifica-
In another experiment with chondroitin sulfate 
and cationic dyes, it was concluded that aggregation of
dyes on the surface of the CS rather than ionic interaction
was mainly responsible for bonding. The thermodynamic functions
indicated an enthalpy change of -7 to -12 kcal/mole of dye
The binding of cobalt hexammine (Co(NH3)6+3) to connective
tissue, micropolysaccharides, heparine, and sulfated chitosans
has been studied by a spectrophotometric procedure . The
cobalt hexammine was used to represent amino functions of fibrin.
Ion pair formation was found to be the primary binding mechanism,
but was influenced by local binding factors, electrostatic attrac-
tion of neighboring charged groups, and competition with other
cations for binding sites.
Such previous work generally indicates that CS-collagen
or CS-polypeptide binding will be primarily ionic, pH, and
structure dependent. These previous results and the results
discussed in this chapter will serve as an aid in understanding
later calorimetric measurements.
In these calorimetric measurements, a solid substrate
was not used. In the first set of experiments, the concentra-
tion of saccharides was held constant at a concentration Co
of 10-3 M of saccharide monomers or residues. In the case of
CS, this refers to dimer residues. Aquisition of the organic