Green Chemistry – with a Special Emphasis on Tannin
Molecules for the Protection of Aluminum Boilers
Gaudreault*, R.; Dargahi**, M.; Weckman**, N.; Olsson**, A.L.J.; Omanovic**,
S.; Schwartz*, G.; Tufenkji**, N.
*TGWT Clean Technologies Inc, **McGill University, Department of Chemical
This paper reinforces the view that innovative green molecules can meet today’s
industrial water treatment needs while reducing overall water consumption in commercial
and industrial facilities. Tannin molecules, extracted from renewable resources, are
examples of green molecules that have continued to protect steam boilers well above
ASME guidelines, for more than 40 years. In fact, the use of tannins and lignin (there are
more than a thousand different tannin molecules identified) has been a primary boiler
water treatment technology since the advent of steamships on the high seas over 150
Today, high-efficiency condensing boilers employed in many hot-water closed loop
systems, often contain significant quantities of aluminum construction components, and
an ideal water treatment technology for aluminum protection is still challenging to many
water treatment companies. We have to comply with manufacturer’s water specifications
in terms of pH and chemistries, in order to minimize corrosion rates, on different alloys
in the systems, and also ensure the validity of the manufacturer’s warranty program.
This paper reviews some of the complexities and interactions that take place at
metal/water interfaces, and also discusses some different investigative methods
employed, including electrochemical corrosion studies and quartz-crystal-microbalance
with dissipation monitoring (QCM-D), to characterize the protective tannin-based layer.
We were able to demonstrate that TGWT tannins (TG 3000) form a protective layer that
prevents aluminum corrosion in a pH range from 7.0 up to 9.0, i.e. above manufacturers’
specifications (7.0 to 8.5). Moreover, corrosion inhibition efficiencies up to 80% were
obtained for aluminum (pH 7-9) and mild steel (pH 10-11). The optimum pH / corrosion
inhibition efficiency of a closed loop aluminum/mild-steel boiler system, using tannin-
based chemistry, would be about 9 or lower.
Finally, results showed that TG 3000 tannins quickly adsorb and reach an equilibrium
within about 15 minutes, compared to some conventional tannin molecules that continue
to adsorb even up +28 hours, giving rise to potential consequences of fouling, loss of
boiler efficiency and overheating. Hence, effective tannin molecules must be carefully
Most production facilities, hospitals, cities and factories use too much water. One-fifth of
the world’s population on earth lives under conditions of water scarcity (Science, June 8,
2012). Consequently, innovative and sustainable technologies are thus essential to reduce
water use in domestic and industrial processes. Hence, it is necessary to clean the water
and to reuse/recycle it as many times as possible. However, while reducing water use, the
concentration of electrolytes (inorganic salts dissolved in water) increases and the
corrosion rate of high pressure/temperature vessels (e.g., boilers and cooling towers),
piping, equipment, and tanks made of different types of metal alloys, also increases. Thus
corrosion inhibitors are added to the water to minimize corrosion [1-4]. Unfortunately,
traditional corrosion inhibitors such as phosphates, sulfites, and amines are not
environmentally friendly [1, 4-9], and their use requires the system to operate at low
inorganic salt content, resulting in the discharge of larger quantities of water into the
rivers, lakes, oceans, etc. Consequently, more efficient and environmentally friendly
corrosion inhibitors for boilers and hot water closed loop systems are needed, which
could also provide protection at extreme operating conditions (high pressure/temperature,
high concentration of dissolved inorganics). The best choice would be to use “greener”
(naturally occurring) molecules as corrosion inhibitors to significantly reduce the
environmental footprint. For example, tannin-based chemistries are used in industrial
steam boiler systems to prevent corrosion. These “greener” molecules enable operation
under much higher conductivity (8000-10000 µS/cm) than traditional chemistries (lower
than 3000 µS/cm). Hence, these renewable molecules reduce water usage, resulting in
reduced energy consumption. The common assumption is that these compounds adsorb
and form a protective film on metal surfaces, thus minimizing the corrosion rate. The
chemicals that are suspected to form the corrosion-protective film are tannins, and their
derivatives. Tannins are large polyphenolic compounds with hydroxyls and carboxylic
groups. They are natural biodegradable compounds and can form strong complexes with
metal ions, proteins and other macromolecules. Tannins can be hydrolysable and
condensable. Original scientific work was performed on the molecular structure of
tannins, e.g. tannic acid and corilagin , and its hydrolysed form, e.g. gallic acid [11, 12].
Moreover, experimental work was performed to study the stability of tannic acid,
corilagin and similar molecules in water [13-16]. However, we still have a poor
understanding of the characteristics of the protective organic layer; namely, we do not
know which specific molecules provide the corrosion protection, how strongly they are
attached to the metal surface, the stiffness or flexibility of the adsorbed layer, the number
of layers, the coordination chemistry, and the effects of varying conditions and operating
parameters such as the presence of the monovalent and multivalent metal ions, pH,
temperature, shear rate (flow rate), type of dissolved solids and conductivity.
Although it has empirically been shown that tannin-based molecules reduce mild steel
corrosion, and that a visual change in the color of the metal surface occurs, the layer
formation on the metallic surface still needs to be experimentally proven. Moreover,
during the last decade numerous aluminum hot water boilers were installed because they
have a much better thermal conductivity, are smaller and lighter. However, very little
research has addressed the corrosion inhibition of these aluminum boilers, and even
more, no scientific studies were performed with tannin-based chemistries. To optimize
the inhibitor formulation and achieve much higher corrosion protection efficiency at
lower inhibitor concentrations and over a longer period of time and under extreme
operating conditions, it is necessary to characterize the physico-chemical properties of the
adsorbed layers under various conditions resembling those used in actual industrial
applications of the inhibitor, including in aluminum boilers.
Although, steel is well established in the boiler industry, the prevalence of aluminum in
the market is increasing because of more efficient thermal properties, e.g. thermal
conductivity and diffusivity. The scientific challenge is to develop inhibitor formulations
that are capable of protecting aluminum boilers at higher pH values, at which condition
they are particularly prone to corrosion. Commercial specifications from the boilers
suppliers are to operate the boiler in the pH range of 7.0-8.5. However, other equipment
in the systems such as heaters and pipes, made of mild steel or copper-based metals,
require different optimum pH. For example, hot water closed-loop steel boilers operate in
the range of 8.5-11.0. The lower pH favors reduced aluminum corrosion, but at a cost of
increased carbon steel corrosion, and vice versa for higher pH . Moreover, Colin
Frayne recommends 7.5-8.5 for boilers having significant aluminum content .
In this work, a range of advanced scientific experimental techniques, have been employed
to investigate the corrosion inhibition efficiency and for surface characterization of
tannin-based layers developed on mild steel (MS) and aluminum (Al) surfaces used in
hot-water closed loop systems.
WHAT ARE TANNINS?
They are large polyphenolic natural product, biodegradable, and can form strong
complexes with metal ions, proteins and other macromolecules. More than 1000 tannins
were identified , having molecular weights from 500 to over 3,000 (gallic acid esters)
and up to 20,000 (proanthocyanidins) .
HYDROLYSABLE AND CONDENSABLE
Tannins can be hydrolysable (Figure 1) and condensable (Figure 2).
Hydrolysable tannin refers to acidic hydrolysis propensity, e.g. the hydrolysis of tannic
acid (TA) would give by-product such as gallic acid (Figure 1, top left and right
respectively). Moreover, the hydrolysis of corilagin would also give similar by-products,
including gallic acid. The common characteristics of these molecules is that they have
aromatic rings with hydroxyls and carboxylic groups, and are negatively charged. Tannic
acid (TA) has a molecular formula C76H52O46 and a molecular weight of 1701.18 g mol-1.
Figure 1. Molecular structures of hydrolysable tannins: top left) tannic acid and top right) one of the
hydrolized product, i.e. gallic acid. Bottom left and right show corilagin, a pyrogalloyl .
The most abundant polyphenols are the condensed tannins, found in virtually all families
of plants, and comprising up to 50% of the dry weight of leaves. Tannins of tropical
woods tend to be of a cathetic nature rather than of the gallic type present in temperate
woods , e.g. catechin shown in Figure 2.
Figure 2. Catechin codensable tannin: left) Molecular structure, and right) electrostatic surface potential.
Courtesy of Professor M.A. (Tony) Whitehead and Kevin Conley, McGill University.
MATERIALS AND METHOD
CLOSED LOOP PILOT UNIT
The pilot unit is used to investigate the corrosion inhibition efficiency on aluminum, mild
steel and copper used in boilers, hot-water closed loop systems (Figure 3). The flow rate
is maintained at 8 USGPM (30 LPM) at an average temperature of 145°F (63°C) and the
pressure in the loop is 80 +/- 10 psig. The corrosion rate is monitored by three coupons:
aluminium Al1100, mild steel C1010, copper CDA110, and by three linear polarisation
resistance (LPR) probes with two electrodes each of the same materials than the coupons
connected to controllers giving direct corrosion rate in mpy. Every two weeks the loop is
stopped and the coupons are changed for new ones and all the sensors are recalibrated.
The coupons are then weighed, using a 5-digit analytical balance, for the evaluation of
the corrosion rate.
Figure 3. Description of the TGWT corrosion inhibition closed loop pilot unit: 1) Hot Water Tank; 2)
Expansion Tank; 3) Coupon Holder (Al, mild steel, Cu); 4; Corrosion Probe (Al, mild steel, Cu); 5)
Rotameter; 6) Corrosion Probe Display, 4-20mA output; 7) Pressure Gauge; 8) Air Vent; 9) Thermometer;
10) Controller Recorder; 11) pH probe; 12) Conductivity Probe; 13) injection Point; 14) Chemical
Metering Pump; and 15) Recirculation pump (not shown).
This pilot unit complies with AWT recommendations and guidelines for corrosion
coupons, and ASTM D2688–05. Corrosion coupons are analyzed according to ASTM G1
Standard Practice for Preparing, Cleaning, and Evaluation of Corrosion Test Specimens
CHEMICALS AND SOLUTIONS
The corrosion inhibitor, TG 3000 solutions, were prepared by diluting concentrated TG
3000 1:1 using deionized water and then further diluting 1:1000 in tap water. Montreal
city tap water (2012 results: pH of 7.5 (min. /7.3; max. /8.0); conductivity of 304 µS/cm
(min/275; max. /323); and Ryznar Index of 8.6 (min./-0.5; max. /9.4), TH: 117 mg/l
CaCO3; Chlorides: 25.3 mg/l; Sulfates 25.3 mg/l; Silica: 0.83 mg/l; Total Phosphates:
0.007 mg/l), was used as a corrosive liquid in the corrosion cell, in the absence and
presence of the TG 3000 inhibitor. For pH adjustment, aqueous 0.1 M sodium hydroxide,
and 0.1 M citric acid or 0.1 M sulfuric acid, were used.
ELECTROCHEMICAL/CORROSION CELL AND EQUIPMENTS
A standard three-electrode electrochemical/corrosion cell was utilized in all
electrochemical experiments. The counter electrode (CE) was a graphite electrode. The
reference electrode (RE) was a saturated calomel electrode (SCE) separated from the cell
by a glass frit. All potentials in this manuscript are expressed with respect to the SCE.
The working electrode (WE) was prepared from an Al 1100 rod (Metal Samples), and
sealed with epoxy resin to give a two-dimensional surface exposed to the electrolyte.
(1) (2) (3) (4)
Figure 4. A schematic diagram of an electrochemical cell used for the corrosion measurement experiments.
(1) Reference Electrode, (2) Working electrode (Al 1100), (3) Counter Electrode, (4) Air Tube, (5)
Electrolyte, and (6) Thermostatic Water Bath.
Table 1: The chemical composition (%, w/w) of coupon alloys.
NA: Not available.
Electrochemical measurements were performed using an AUTOLAB
potentiostat/galvanostat/FRA PGSTAT 30 controlled by FRA2 and GPES v.4.9 software.
To ensure complete characterization of the interface and the surface processes,
electrochemical impedance spectroscopy (EIS) measurements were made over a
frequency range, from 100 kHz to 10 mHz with the alternating current (AC) voltage
amplitude of ±10 mV.
Prior to each experiment, the working electrode surface was polished with 600-gradation
abrasive sandpaper, and then thoroughly rinsed with ethanol. After this, the electrode was
kept in an ultrasonic bath for 5 min in ethanol to degrease it and remove any possible
polishing residues, and then rinsed with deionized water. The electrode was then
immersed in the test electrolyte and equilibrated for 1 hr at 158°F (70°C) or 32 hours at
22°C at open-circuit potential (OCP), followed by a specific type of experiment. All
measurements were carried out by continuous purging of air through the liquid. All the
values reported in this work represent mean values of at least three replicate experiments.
QUARTZ-CRYSTAL-MICROBALANCE WITH DISSPATION (QCM-D)
Montreal city tap water was filtered using a 0.22µm membrane to remove dust or
impurities before solution preparation. The corrosion inhibitor TG 3000 and conventional
tannins were used as received. TG 3000 solutions were prepared by diluting concentrated
TG 3000 1:1 using deionized (DI) or Montreal tap water and then further diluting 1:1000
in tap water. All other solutions were prepared by dissolving the compound in tap water
to the desired concentration of 100ppm. Solutions were stored at 46°F (8°C) for a
maximum of one week before measurements. For pH adjustments aqueous sodium
hydroxide and citric acid were used.
An E4 QCM-D unit from Q-Sense (Q-Sense AB, Göteborg, Sweden) was used for the
QCM-D experiments. Alumina (Al2O3) coated AT-cut quartz crystals (QSX301) with a
fundamental resonance frequency of 5MHz were purchased from Q-sense and used as the
model substrate. The thickness of the aluminum layer is about 100 nm. The crystals were
cleaned by soaking and sonicating for at least 10 minutes in a 2% Hellmanex solution
before being rinsed with DI water and dried with nitrogen gas. The crystals were exposed
to ultraviolet light for 20min immediately before the experiment.
Figure 5. Experimental setup for the corrosion inhibitor adsorption on Al2O3 surface.
A flow rate of 100µL/min for the short experiments and 10µL/min for the long
experiment was maintained through the QCM-D using a peristaltic pump (RegloDigital,
Ismatec). The temperature was controlled by the QCM-D and was maintained at 72°F
(22°C). In this work, only one experiment at higher temperature of 122°F (50°C) was
performed to show the effect of temperature. Before beginning the experiments, a
frequency and dissipation baseline in tap water was allowed to stabilize. Then the desired
solution was flowed across the crystal surface in order to measure any adsorption on the
crystal. In certain cases this was followed by a rinse in tap water to see if the corrosion
inhibitor was irreversibly adsorbed. After each experiment, the QCM-D setup was
cleaned by flowing a 2% Hellmanex solution then DI water through the QCM-D for at
least 10min each. The water was purged from the setup using air and nitrogen gas.
The Quartz Crystal Microbalance with Dissipation monitoring (QCM-D) consists of a
quartz crystal sensor which oscillates at its fundamental resonance frequency of 5MHz,
as well as at the frequencies of its odd overtones. As molecular deposition occurs on the
crystal surface, a negative shift in these frequencies can be measured in real time.
The Sauerbrey relation is used to calculate the mass of molecular adsorption to the sensor
surface, referred to as the Sauerbrey mass . The Sauerbrey mass (Δm) is calculated
from the measured shift in the quartz crystal resonance frequency (Δf) by:
where C is the sensitivity constant which for the AT-cut quartz crystals used in Q-Sense
device equals -17.7 ng cm-2 Hz-1, and n is the resonance frequency overtone number. The
Sauerbrey mass (Δm) is the adsorbed mass averaged over surface area and is in units of
ng/cm2. The detection limit of QCM-D is 0.1 Hz (1 ng/cm2).
When applying the Sauerbrey relation the following assumptions are made: 1) The
adsorbed mass is continuous and homogenously distributed on the surface; 2) we assume
a no-slip condition (i.e. that the mass does not slip on the sensor while oscillating); 3) The
mass is rigid. Assumption 3 can be tested by considering the dissipative energy losses at
each oscillation cycle in the QCM. The QCM-D measures these dissipative energy losses
during crystal oscillation in terms of oscillation decay time (i.e. the so-called dissipation
factor, D). The dissipated energy can also be determined from the width of the oscillation
peak (i.e., half bandwidth at half maximum Г ).
The dissipation energy and the bandwidth are fully equivalent and are related to one
another via Eq.(2):
where Γ and D, are the bandwidth and dissipation factor at a specific overtone, and fn is
the resonance frequency at overtone n. The bandwidth is the imaginary part of the
complex frequency shift and thus is in units of Hz. Since fn is in units of MHz, the
dissipation factor is in units of 10-6.
A "Sauerbrey like" rigid film dissipates less energy per oscillation cycle than a "soft like"
viscoelastic film does. For example, a decrease in the frequency shift (Δf) indicates that
tannic acid (TA) adsorbs on Al2O3 surface, and a small shift in dissipation (ΔD) indicates
that the TA layer is rigid (Figure 6). However, viscoelastic films do not fully couple to
the crystal oscillation which leads to a "missing mass" effect [23, 24].
Figure 6. Adsorption of tannic acid (TA) at pH 7.8 to Al2O3 surface measured by QCM-D. TA
concentration of 100ppm and room temperature.
In order to determine if the Sauerbrey equation will provide an adequate estimation of the
adsorbed mass, the ratio between the dissipation factor and the normalized frequency
shift (normalized with respect to overtone number) is considered. The threshold value for
the Sauerbrey to be valid is 10-7 according to Q-Sense and 4x10-7 according to Reviakine
. In this study, the ratio ΔD/Δf was always lower than 10-7, i.e. justifying the use of the
Sauerbrey equation to calculate the adsorbed mass. Moreover, the frequency shifts were
practically identical at all overtones and therefore only the data from the 3rd overtone is
RESULTS AND DISCUSSION
CORROSION INHIBITION RESULTS FOR AL 1100
Short term, high temperature experiment with TG 3000 inhibitor
Electrochemical impedance spectroscopy measurements
Electrochemical impedance spectroscopy (EIS) was applied to investigate the aluminum
electrode/electrolyte interface and processes that occur on the Al 1100 surface at OCP in
the presence and absence of TG 3000 inhibitor in the solution, most notably the general
corrosion resistance of Al 1100. Figure 7 shows the EIS spectra of the Al 1100 electrode
recorded in the absence and presence of inhibitor at pH 8. The spectra were recorded at
158°F (70°C) after the stabilization of the electrode at OCP for 1 hour. The EIS spectra
are characterized by two semicircles, each representing one time constant. Figure 7 shows
that the diameter of both semicircles increases in presence of inhibitor, indicating an
increase in corrosion resistance of the material.
0500 1000 1500 2000 2500 3000 3500
Without TG 3000
With TG 3000
Figure 7. Nyquist plot of an Al 1100 electrode recorded in the presence (□) and the absence (o) of TG 3000
inhibitor at 158°F (70°C) and pH 8 after 1 hour of incubation at OCP. Symbols are experimental data and
solid lines represent the simulated (modeled) spectra.
In order to extract qualitative information, a nonlinear least-squares (NLLS) fit analysis
was used to model the spectra, employing an electrical equivalent circuit (EEC) presented
in Figure 8.
Figure 8. EEC model used to fit EIS data recorded on Al 1100.
In this EEC, Rel represents the ohmic resistance between the working and the reference
electrode; R1 is the charge transfer resistance related to the corrosion reaction at
OCP; CPE1 is the capacitance of the electrochemical double-layer at the
electrode/electrolyte interface; R2 can be prescribed to the pseudo-resistance of the
surface-adsorbed inhibitor and/or porous passive aluminum oxide layer in presence and
absence of the inhibitor, respectively, while the element CPE2 is its pseudo-capacitance.
The EEC in figure 8 was used in fitting of the spectra recorded in the presence and
absence of the inhibitor. The corresponding R1 and R2 values, obtained by fitting the
experimental spectra, are presented in Table 2 (note that CPE values are not listed since
they are not used in the evaluation of corrosion resistance).
Table 2. Dependence of EEC parameters and corrosion inhibition efficiency (η) of TG 3000 for Al 1100 on
pH of the electrolyte. The data were obtained by modeling EIS spectra recorded at 158°F (70°C) after 1
hour of incubation. SD = standard deviation.
TGWT TG 3000
R1 / Ω
R2 / Ω
Ri / Ω
η / %
Taking into account the physical meaning of the EEC parameters of the circuits in figure
8, the corrosion resistance of the bare (control) Al 1100 surface is equivalent to the sum
of charge-transfer resistance, R1, and the corresponding resistance of the aluminium oxide
porous layer R2, R0 = (R1 + R2)control, while when the inhibitor layer is formed on the Al
1100 surface, the corresponding corrosion resistance is equivalent to the sum of the
charge-transfer and inhibitor+oxide layer resistance R2, Ri = (R1 + R2)inhibitor.
Consequently, the corrosion inhibition efficiency of the inhibitor, η, was calculated by
comparing the total resistance value, Ri, recorded at various pH in presence of inhibitor,
to the corresponding R0 value recorded in the absence of inhibitor (control sample) and
the results are shown in Table 2:
Tafel Polarization Measurements
Tafel polarization measurements were made to complement/verify the data obtained from
EIS measurements. The technique is based on the polarization of the working electrode
(test-sample) ca. ±200 mV around OCP at a slow scan rate (1 mV s−1).
Figure 9 clearly shows that in the presence of inhibitor in the bulk solution, both the
cathodic (oxygen reduction) and anodic (Al dissolution) current decreases, which is due
to the blockage of the surface with adsorbed inhibitor. Also, the corrosion potential
shifted to a more positive direction in the presence of inhibitor, which indicates corrosion
protection of the Al 1100 surface by the inhibitor layer.
E / VSCE
-1.4 -1.3 -1.2 -1.1 -1.0 -0.9 -0.8 -0.7
log I (A)
With TG 3000
Without TG 3000
Figure 9. Tafel plots of an Al 1100 electrode recorded at pH 8 in the presence (blue) and absence (red) of
the inhibitor (TG 3000) at 158°F (70°C). The curves were recorded after 1 hour of incubation at OCP. Scan
rate = 1 mV s−1.
The corresponding corrosion current (Icorr.) were determined by the extrapolation of the
linear (Tafel) part of the cathodic and anodic curves to the corresponding corrosion
potential, and presented in Table 3 (mean value) together with the corresponding standard
deviation. The inhibition efficiency was then calculated using Eq.(4):
Table 3. Corrosion current in the presence (Icorr,i) and absence (Icorr,0) of the inhibitor (TG 3000), calculated
form Tafel measurements recorded on Al 1100 at various pH after 1 hour of incubation at 158°F (70°C) at
OCP. The table also lists the corresponding corrosion inhibition efficiency values, η. SD = standard
Icorr,0 / µA
Icorr,i / µA
η TG 3000 / %
The comparison between the corrosion inhibition efficiencies obtained from EIS and
Tafel measurements show a very good agreement between both methods (Figure 10).
6 7 8 9 10 11 12
Corrosion Inhibition Efficiency / %
Figure 10. Dependence of the corrosion inhibition efficiency of TG 3000 as a function of pH obtained
from EIS (o) and Tafel (□) measurements after 1 hour of incubation at 158°F (70°C).
Long term, low temperature experiments for TG 3000 inhibitor
Low temperature experiments were performed for Al 1100 with TG 3000 at 72°F (22°C)
for 32 hours at pH 7 and 9 (Table 4). Figure 11 shows that it takes approximately 5 hours
for TG 3000 to reach to the highest corrosion inhibition. Interestingly, the online linear
polarization resistance (LPR) monitoring of the TGWT pilot unit also shows higher
corrosion rate (lower corrosion inhibition) in the first few hours (results not shown here).
Table 4: Dependence of Al 1100 corrosion inhibition efficiency (η) of TG 3000 on incubation time. The
data were obtained by modeling EIS spectra recorded at 72°F (22°C) at pH 7 and 9. SD = standard
time / hour
η / %
η / %
Time / hour
0 5 10 15 20 25 30 35
Corrosion Inhibition Efficiency / %
Figure 11. Corrosion inhibition efficiency of TG 3000 inhibitor as a function of time for Al 1100 at 72°F
(22°C) at pH 7 (red) and 9 (blue).
CORROSION INHIBITION RESULTS FOR C1010
Short term, high temperature experiments
Short term (one hour) experiments were performed for C1010 at 158°F (70°C) over a
range of pH with TG 3000 as the inhibitor. As expected the corrosion inhibition
efficiencies are low at lower pH, due to high corrosion rate of C1010 at low pH (Table 5,
Figure 12). However, careful interpretation is required when analyzing the corrosion rate
after one hour at 158°F, since the results from the pilot unit showed higher corrosion rate
(lower inhibition efficiency) in the first few hours to a few days.
Table 5. Corrosion inhibition efficiency values (η) of TG 3000 inhibitor for C1010 obtained by modeling
EIS spectra at various pH after one hour at 158°F (70°C).
η / %
6 7 8 9 10 11 12
Corrosion Inhibition Efficiency / %
Figure 12. Corrosion inhibition efficiency (η) of TG 3000 inhibitor as a function of pH for C1010 obtained
by modeling EIS spectra after one hour at 158°F (70°C).
Figure 13 also shows the corrosion inhibition efficiencies of TG 3000 inhibitor for both
Al 1100 and C1010, at various pH after one hour of incubation at 158°F (70°C). The
results indicate that the optimum pH / corrosion inhibition efficiency of a closed loop
Aluminum/Mild-Steel boiler system, using tannin-based chemistry, would be about 9 or
6 7 8 9 10 11 12
Corrosion Inhibition Efficiency / %
Figure 13. Corrosion inhibition efficiency (η) of TG 3000 inhibitor as a function of pH for C1010 (o) and
Al 1100 (□) obtained by modeling EIS spectra at various pH after one hour of incubation at 158°F (70°C).
INHIBITOR SURFACE FILM FORMATION
QCM-D Measurement of the Adsorption of Tannins to an Al2O3 Surface
Figure 14 clearly shows that TG 3000 adsorbed to the Al2O3 surface and that the
adsorption is highly pH dependent of the system. The adsorption is similar at the lower
pHs of 7.0 and 7.8. At these pHs the adsorption rapidly increases upon injection of the
TG 3000 and then reaches an equilibrium value within 15 minutes. This behavior is likely
due to molecules quickly adsorbing to the surface and then changing conformations. As
the pH is increased to 9.1 then 9.5, higher adsorption is observed. Interestingly, the
highest adsorption rates occur within 7-8 minutes and vary from 2.6 to 5.7 mg m2 min-1.
Figure 14. Adsorbed amount (left) and adsorption rate (right) of TG 3000 as a function of time and pH to
Al2O3 surface (number (n) of experiments: n=3 for pH 7 and 9.1, n=4 for pH 7.8 and 9.5).
Then, a long term experiment was run at a pH of 9.11 to determine how the adsorbed
layer of TG 3000 behaves over extended time periods (Figure 15).
Figure 15. Adsorption of TG 3000 at pH 9.1 over a long time-scale
As observed in the short term experiments at pH 9.1, there is rapid adsorption of TG 3000
within the first 10-15 min to an average of about 2.5-3 mg/m2. This is slightly more than
was observed in the short term experiment which is probably due to the fact that a slower
flow rate was used for the long term experiment allowing more adsorption to occur. Over
the course of the experiment, there is a slight decrease in adsorption. This trend could
indicate desorption of heavier molecules versus attachment of lighter ones,
conformational change, water molecules that are removed from the layer, or is simply
part of the experimental error (see error bars). The signal is otherwise suggesting the
adsorption of a stable layer.
Adsorption of Tannins to an Al2O3 Surface: effect of temperature (72° and 122°F)
The adsorption of TG 3000 to the alumina surface at pH 8 and a temperature of 72°F
(22°C) and 122°F (50°C) was performed (Figure 16). In general, the signal from the
higher temperature experiment is very noisy due to bubbles and other complications at
the high temperature in the QCM-D module. Nevertheless, it can be seen that at the
higher temperature, about twice the adsorption is occurring than at the lower temperature.
Moreover, it is speculated that the difference in adsorbed amount between 72 and 122°F
is not due to a higher amount of water molecules in the layer. Due to the noise, we should
not try to extract absolute values from the high temperature experiment. Otherwise, the
adsorption rate behavior is approximately the same in both cases.
Figure 16. Adsorption of TG 3000 to Al2O3 surface at 72°F (22°C) and 122°F (50°C)
Adsorption of conventional tannin onto Al203
The adsorption of conventional tannin to the Al2O3 QCM-D crystal exhibited different
behavior than the TG 3000 (Figure 17). In other words, over long time scales, the
adsorption at pH 10 levels off and reaches equilibrium while at other pHs there is
continued adsorption, albeit at a very slow rate. This difference between the lower pH
and pH of 10 could possibly be explained by many phenomena, e.g. going from a
positively charged surface to a negatively charged surface when the pH gets beyond 9.1
 or 8.7 . Another factor that could influence the adsorption of conventional tannin
is the fact that it dissolves differently at different pH, and the solutions have a distinct
color difference depending on the pH. Again, the highest rate of adsorption occurs at
about 7-8 minutes. Moreover, conventional tannins continue to adsorb up to +28 hrs (not
shown here), giving rise to potential fouling and loss of boiler efficiency. Obviously, the
critical amount adsorbed is unknown, hence further experiments are needed to answer
Figure 17. Top: Short- (left) and long-time (right) scales adsorption and bottom: short- and long-time
scales adsorption rate of conventional tannin to Al2O3 surface.
Tannic acid adsorption onto Al2O3
The adsorption of tannic acid to the Al2O3 QCM-D crystal shows the opposite trend to
the other tannin molecules studied in that as the pH is increased, the adsorption decreases
Figure 18. Adsorption (left) and adsorption rate (right) of Tannic Acid to Al2O3 surface.
The decrease in adsorption at the higher pH could be due to the fact that as the pH was
adjusted to 9.0 using NaOH, a dark yellow precipitate was formed and settled out of the
solution. Due to this precipitation, the amount of soluble tannic acid in the solution would
be decreased which could result in decreased adsorption. These experiments were
repeated and there was a visual change in color starting at pH 8. Here, we speculate that
calcium ions (dissolved in Montreal city water with concentration of 32.5 ppm) can
induce the precipitation. Hence additional experiments, using calcium hydroxide
(Ca(OH)2) for pH control, showed that precipitation was gradually increasing starting at
pH 8 until 10.2, where dark yellow precipitates were observed. The maximum adsorption
rates also occur between 7-8 minutes as evidenced by Figure 18 (right).
Adsorption of Tannins on Mild Steel Surface
Figure 19 shows that TG 3000 does adsorb to the mild steel surface at pH 8. Although not
shown in this manuscript, the TG 3000 is fairly tightly bound to the aluminum and mild
steel surface, i.e. does not desorb with rinsing. This is further supported by the fact that
even after cleaning the system and the crystals with Hellmanex, there was still an
observed color change on the mild steel crystal surface. This observation corroborates the
ones made inside the industrial boilers. However, if the crystal surfaces are to be re-used
for more QCM experiments, another cleaning protocol may need to be developed to
completely clean the TG 3000 from the surface.
Figure 19. Adsorption of TG 3000 to alumina (pH 7.8) and mild steel (pH 8.0) surfaces.
Calculated adsorbed mass of corrosion inhibitors on Al2O3 surface
The objective of using corrosion inhibitors is to protect the metal from the transport of
solvated corrosive ions and oxygen molecules to the metal surfaces. The amount of
adsorbed inhibitor can give insights on how the protective layer can help achieve this
goal. This section is an attempt to estimate the maximum adsorbed amount of corrosion
inhibitors on an aluminum surface as a function of pH, and to determine if the protective
layer is rigid or flexible.
The dissipation (ΔD) for all tannin samples is very low, 0.1-0.4E-06 , and the ratio
between the dissipation and the frequency (ΔD/Δf ratio) is less than 10-7 Hz-1, i.e. -1E-08 to
-7E-09 Hz-1, indicating that the adsorption is in the Sauerbrey regime and thus the
Sauerbrey equation can be used to calculate the adsorbed mass.
Table 6 summarizes the adsorbed mass of corrosion inhibitors, the maximum adsorption
rate (kmax) and the time to reach kmax, as calculated from the QCM-D frequency shifts (Δf).
It shows that adsorption occurs and is a function of chemistry and pH for all molecules
investigated. The Al2O3 surface has a positive surface charge up to a pH of 9.1  or 8.7
, and then a negative surface charge above this value, will likely influence the
attachment of the corrosion inhibitor and consequently the corrosion rate.
Table 6. Adsorption of corrosion inhibitors on Al2O3 at 72°F (22°C) (Δm includes water).
Adsorbed mass at
Time to kmax
(mg m-2 min-1)
Values within parenthesis indicate standard deviation of 3-4 measurements. Values without
parenthesis indicate the results of one single fold measurement.
In addition to calculating the adsorbed mass of corrosion inhibitor, this section is an
attempt to estimate the type of adsorbed layer (mono- or multi-layers), and elucidate the
orientation of the molecules on the surface, whether laying side by side or stacked on top
of each other. To do such an investigation, tannic acid was selected as a model molecule
because the molecular weight is known and the molecular dimensions have been
measured from molecular modeling .
Using Δm of TA adsorption at a pH of 7.8 (Table 6) and the Sauerbrey equation,
preliminary calculations (Δf of -31.77 Hz and ΔD of 0.91*10-6, C76H52O46, molecular
weight of 1701.18 g / mole, length = 3.5 nm and width = 2.0 nm) suggest that less than
two TA molecules per nm2 are adsorbed on the aluminum surface of the QCM crystal.
This is an estimate since Δm includes TA and water molecules, where the proportion of
water within the protective tannins layer is unknown and is most likely function of the pH
of the system.
Finally, the adsorbed amount generally increases with increasing pH. However, too much
adsorption might have negative consequences such as fouling, loss of boiler efficiency, or
overheating. Hence, effective molecules must be carefully selected among more than a
thousand tannin molecules.
Our results show that TGWT tannins protect aluminum surface from pH 7.0 up to 9.0, i.e.
above manufacturer’s specifications (pH = 7.0 to 8.5).
Moreover, corrosion inhibition efficiencies up 80% were obtained for aluminum (pH 7-9)
and mild steel (pH 10-11).
We were able to prove that TG 3000 tannins form a protective layer on aluminum and
mild-steel surfaces using electrochemical techniques and QCM-D method. Interestingly,
the protective layer is rigid and strongly bonded to the surface.
Although all molecules rigidly bind to the surface, the adsorption is highly pH dependent.
The optimum pH / corrosion inhibition efficiency of a closed loop aluminum/mild-steel
boiler system, using tannin-based chemistry, would be about 9 or lower.
TG 3000 tannins quickly adsorb and reach equilibrium within about 15 minutes.
However, between pH 7.0 and 9.0, conventional tannins continue to adsorb even up to,
and/or after 28 hours, the potential consequences being fouling, loss of boiler efficiency,
or overheating. Hence, the selection of effective tannin molecules is critical.
The authors gratefully acknowledge the financial support from the Natural Science and
Engineering Research Council of Canada (NSERC) and from TGWT Clean
Technologies. The authors also thanks Mr. Colin Frayne for valuable comments, and Mr
Benoit Gaboriau and Mrs Melanie Cardin for the design of the TGWT pilot unit.
1. M. Ash, I. Ash, Handbook of Corrosion Inhibitors, Synapse Information
2. A.U. Smart, Industrial Boiler System Corrosion Inhibitors: Studies on the High
Temperature Reactions and Properties of Aliphatic Amines in Water, University
of York, 1989
3. M. Schweinsberg, B. Mayer, W. Hater, Evaluation of Corrosion Inhibitors for
Cooling Water by Physico-chemical Methods, Henkel Surface Technologies,
4. B. Mayer, W. Hater, M. Schweinsberg, Environmentally Sound Corrosion
Inhibitors for Cooling Water, Henkel Surface Technologies, TL6-2002.
5. V.S. Sastri, Green Corrosion Inhibitors: Theory and Practice, John Wiley &
6. S. Gartiser, E. Urich, Environmentally compatible cooling water treatment
chemicals, Environmental Research of the Federal Ministry of the Environment,
Nature Conservation and Nuclear Safety - Water Economy - Research Report 200
24 233, 2002.
7. S. Ghareba, Inhibition of carbon steel corrosion by long alkyl-chain amino
acid corrosion inhibitors, PhD Thesis, McGill University, 2011.
8. M.A. Quraishi, D. Jamal, Technical Note, CAHMT—A New and Eco-Friendly
Acidizing Corrosion Inhibitor, Corrosion, 56 (2000) 983.
9. M.S. Morad, Inhibition of iron corrosion in acid solutions by Cefatrexyl:
Behaviour near and at the corrosion potential, Corros. Sci., 50 (2007) 436.
10. R. Gaudreault, M-A. Whitehead, T.G.M. van de Ven. Molecular modeling of
poly(ethylene oxide) model cofactors; 1,3,6-tri-0-galloyl-β-D-glucose and
corilagin. Published online March, 8th, 2002, © Springer-Verlag 2002.
11. R. Gaudreault, M-A. Whitehead, T.G.M. van de Ven. A theoretical study of
the interactions of water with gallic acid and a PEO/TGG complex. Molecular
Simulation, Vol.32, No.1, 15 January 2006, 17-27.
12. R. Gaudreault, M-A. Whitehead, T.G.M. van de Ven. Molecular Orbital
Studies of Gas-Phase Interactions between Complex Molecules. J.Phys.Chem. A
2006, 110, 3692-3702.
13. R. Gaudreault, “Mechanisms of Floculation with Poly(ethylene oxide) and
Novel Cofactors: Theory and Experiment”, Ph.D. Thesis, Dept. of Chem. MGill
University, Montreal, Canada, December 2003.
14. R. Gaudreault, T.G.M. van de Ven, M.A. (Tony) Whitehead, “Mechanisms of
Flocculation of Microcrystalline Cellulose by Poly(ethylene oxide) and Cofactor
Corilagin”, The 13th FRC Symposium, Robinson College, Cambridge, 1269-
1292, 11th-16th September 2005.
15. R. Gaudreault, M.A. (Tony) Whitehead, T.G.M. van de Ven, “Mechanisms of
Flocculation with Poly(Ethylene Oxide) and Novel Cofactors”, Colloids and Surf.
A: Physicochem. Eng. Aspects, 268:131-146 (2005).
16. R. Gaudreault, T.G.M. van de Ven, M.A. (Tony) Whitehead, “Salt Necessary
for PEO-Cofactor Association: The role of Molecular Modelling in PEO
Flocculation Mechanisms”, First Applied Pulp & Paper Molecular Modelling
Symposium, Montréal, Canada, August 24-26, 2005, 155-193. Eds. R.
Gaudreault, M.A. Whitehead and T.G.M. van de Ven, Desktop Publishing,
17. Labrosse, M.; Donovan E., Pilot Research to Determine Effective Aluminum
Corrosion Inhibition, AWT 2011 Annual Convention and Exposition, September
14-17, 2011, Atlanta, Georgia.
18. Frayne, C., Internal Note on Optimum pH for Closed-Loop Water Systems
Containing Corrosion and Other Inhibitor Packages, April 2012.
19. Karamali Khanbabaee, Teunis van Ree, « Tannins: classification and
definition », Nat. Prod. Rep., vol. 18, 2001, p. 641-649.
20. Bate-Smith and Swain (1962). "Flavonoid compounds". In Florkin M., Mason
H. S. Comparative biochemistry III. New York: Academic Press. pp. 75–809.
21. Tannin in Tropical Woods. Doat J, Bois. For Tmp., 1978, volume 182, pages 34-37.
22. Sauerbrey, G. Verwendung Von Schwingquarzen Zur Wiigung Dunner.
Schichten Und Zur Mikrowägung. Z. Phys 1959, 155, 206–222.
23. Voinova, M. V; Jonson, M.; Kasemo, B. “Missing Mass” Effect in
Biosensor’s QCM Applications. Biosens. Biolectron. 2002, 17, 835–841
24. Johannsmann, D.; Mathauer, K.; Wegner, G.; Knoll, W. Viscoelastic
Properties of Thin Films Probed with a Quartz-crystal Resonator. Phys. Rev. B
1992, 46, 7808–7815.
25. Reviakine, I.; Johannsmann, D.; Richter, R. P. Hearing What You Cannot See
and Visualizing What You Hear. Anal. Chem. 2011, 23, 8838-8848.
26. J. A. Yopps I and D. W. Fuerstenau, The Zero Point Charge of Alpha-Alumina,
Journal of Colloid Science, 19, 61-71 (1964).
27. Michael F. Cuddy, Aimee R. Poda, and Lauren N. Brantley, Determination of
Isoelectric Points and the Role of pH for Common Quartz Crystal Microbalance
Sensors, ACS Appl. Mater. Interfaces 2013, 5, 3514−3518.
28. Cohen, Jon J. Film Formation, stability and Corrosion Inhibition of Surface
deposited Film Inhibitors, CTI Journal, Vol. 34, No. 2, pp. 8-12.