Shedding light on the governing mechanisms for insufficient CO and H2 burnout in the presence of potassium, chlorine and sulfur

Abstract

Based on the experiences of insufficient burnout in industrial fluidized bed furnaces despite adequate mixing and availability of oxidizer, the influence of potassium on CO and H2 oxidation in combustion environments was investigated. The combustion environments were provided by a laminar flame burner in a range relevant to industrial furnaces, i.e. 845 °C to 1275 °C and excess air ratios ranging from 1.05 to 1.65. Potassium, in the form of KOH, was homogeneously introduced into the hot gas environments to investigate its effect on the radical pool. To quantitatively determine key species that are involved in the oxidation mechanism (CO, H2, KOH, OH radicals, K atoms), a combination of measurement systems was applied: micro-gas chromatography, broadband UV absorption spectroscopy and tunable diode laser absorption spectroscopy. The inhibition effect of potassium on CO and H2 oxidation in excess air was experimentally confirmed and attributed to the chain-terminating reaction between KOH, K atoms and OH radicals, which enhanced the OH radical consumption. The addition of chlorine or sulfur could reduce the concentrations of KOH and K atoms and consequently eliminated the inhibition on CO and H2 oxidation. Existing kinetic mechanisms underestimate the inhibiting effect of potassium and they fail to predict the effect of temperature on CO and H2 concentration when potassium and sulfur co-exist. This work advances the need to revise existing kinetic mechanisms to fully capture the interplay of K and S in the oxidation of CO and H2 in industrial fluidized bed furnaces.

Full text

Contents lists available at ScienceDirect
Fuel
journal homepage: www.elsevier.com/locate/fuel
Full Length Article
Shedding light on the governing mechanisms for insufficient CO and H
2
burnout in the presence of potassium, chlorine and sulfur
Teresa Berdugo Vilches
a,⁎
, Wubin Weng
b
, Peter Glarborg
c
, Zhongshan Li
b
, Henrik Thunman
a
,
Martin Seemann
a
a
Division of Energy Technology, Chalmers Tekniska Högskola, Sweden
b
Division of Combustion Physics, Lund University, P.O. Box 118, S-221 00 Lund, Sweden
c
Chemical Engineering, Technical University of Denmark, DK-2800 Kgs., Lyngby, Denmark
ARTICLE INFO
Keywords:
CO oxidation
Potassium
Inhibition
Combustion
UV absorption spectroscopy
TDLAS
ABSTRACT
Based on the experiences of insufficient burnout in industrial fluidized bed furnaces despite adequate mixing and
availability of oxidizer, the influence of potassium on CO and H
2
oxidation in combustion environments was
investigated. The combustion environments were provided by a laminar flame burner in a range relevant to
industrial furnaces, i.e. 845 °C to 1275 °C and excess air ratios ranging from 1.05 to 1.65. Potassium, in the form
of KOH, was homogeneously introduced into the hot gas environments to investigate its effect on the radical
pool. To quantitatively determine key species that are involved in the oxidation mechanism (CO, H
2
, KOH, OH
radicals, K atoms), a combination of measurement systems was applied: micro-gas chromatography, broadband
UV absorption spectroscopy and tunable diode laser absorption spectroscopy. The inhibition effect of potassium
on CO and H
2
oxidation in excess air was experimentally confirmed and attributed to the chain-terminating
reaction between KOH, K atoms and OH radicals, which enhanced the OH radical consumption. The addition of
chlorine or sulfur could reduce the concentrations of KOH and K atoms and consequently eliminated the in-
hibition on CO and H
2
oxidation. Existing kinetic mechanisms underestimate the inhibiting effect of potassium
and they fail to predict the effect of temperature on CO and H
2
concentration when potassium and sulfur co-
exist. This work advances the need to revise existing kinetic mechanisms to fully capture the interplay of K and S
in the oxidation of CO and H
2
in industrial fluidized bed furnaces.
1. Introduction
Combustion technologies are important sources of heat and elec-
tricity in today’s society, and they are likely to play a role in the future
energy system as a stabilizing complement to intermittent energy
sources. To ensure that combustion takes place in the most efficient and
clean way, combustion units are subject of strict regulations on emis-
sions, e.g. limits on CO, NO
x
,SO
x
, dioxins. Among the regulated
emissions, CO should be kept below 150–500 ppm at the stack for a
given excess air concentration of usually 6% for biomass and 11% for
waste. The exact values depend on the permissions for different plants
and are defined in national and local legislation, e.g. Regulation
(20013:253) for combustion of waste in Sweden [1].
A common reason for CO emissions is a local deficit of O
2
due to
mixing limitations between fuel and air. Therefore, commercial com-
bustors operate with excess air, resulting in a concentration of some
percent O
2
at the stack. The problem of CO emissions is exacerbated in
facilities that combust biomass and inhomogeneous solid fuels such as
waste streams. In those cases, moderate temperatures (typically
750–900 °C) are applied to reduce the risk of uncontrolled ash melt and
corrosion issues derived from the impurities in the fuel, as well as
prevent overheating of the furnace material; a measure that can also
challenge the complete oxidation of CO into CO
2
. Besides mixing and
temperature, CO oxidation can be influenced by the chemical interac-
tion between the inorganic impurities in the fuel and the fuel oxidation
reactions. In fact, even trace levels of active species like alkali com-
pounds can drastically influence the combustion chemistry in thermal
conversion processes [2].
The literature on the effect of alkali on the oxidation of CO is in-
conclusive and the underlying mechanisms are still under discussion.
Both, promoting and inhibiting effects of alkali species on CO oxidation
have been observed and are theoretically possible [2]. This reflects the
strong dependence of the reactions on operating conditions, as well as
the complex interrelation between inorganic and organic chemistry.
https://doi.org/10.1016/j.fuel.2020.117762
Received 30 January 2020; Accepted 31 March 2020
⁎
Corresponding author.
E-mail address: berdugo@chalmers.se (T. Berdugo Vilches).
Fuel 273 (2020) 117762
0016-2361/ © 2020 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/BY/4.0/).
T

Previous work in the 12-MW Chalmers circulating fluidized bed (CFB)
boiler [3,4] has evidenced the strong inhibiting effect of K on CO oxi-
dation during combustion of biomass in a bed of olivine. Potassium was
introduced in the combustor as part of the fuel ash and it accumulated
in the bed material, wherefrom it could be released in gas-phase to the
reaction environment [4].Fig. 1 shows the increasing levels of CO over
time, as the ash elements accumulate in the olivine bed. Despite the
temperature (850 °C) and oxygen concentrations (3.5% vol) were the-
oretically sufficient to ensure a complete oxidation of the fuel, CO
concentrations steadily increased reaching values above 1000 ppm at
the stack [3]. The causality between the CO emissions and the presence
of K was further confirmed by adding small batches of K salts to the
fluidized bed (not shown in the figure). Similar inhibitory effect of CO
by Na salts have been observed in smaller units, e.g. in a laboratory-size
fluidized bed reactor by Bulewiz et al. [5], and in a 300-kW down-fired
radiant furnace by Lissianski et al. [6]. Hindiyarti et al. [7] found that
the inhibiting effect of K is also relevant to the oxidation of CO by water
vapour in the absence of oxygen in the temperature range 500–1100 °C.
The inhibiting effect was proportional to the concentration of po-
tassium and it levelled offat high concentrations (above 500 ppm under
the investigated conditions). Contrarily, Ekvall et al. [8] did not observe
an influence of K on the oxidation of CO during air-combustion of
propane in a 100-kW experimental rig at temperatures above 1000 °C,
while a promoting effect was found under oxy-fuel conditions.
The inhibiting effect of alkali is mostly attributed to gas-phase ra-
dicals removal reactions [2,9,10]. According to Glarborg [2], the main
radical removing reactions are R1 and R2. The sequence corresponds to
the overall step H + OH →H
2
O, where two radicals are removed.
KOH+H→K+H
2
O (R1)
K+OH+M→KOH+M (R2)
H+O
2
→OH+O (R3)
CO+OH→CO
2
+H (R4)
Other trace species than alkali, e.g. sulfur and chlorine, are also
known to influence the CO oxidation [2]. The experience of the Chal-
mers fluidized bed boiler shown in Fig. 1 exemplifies the strong re-
sponse of the CO emissions to the addition of elemental sulfur. With
1 kg/h of sulfur the CO emissions went rapidly from 1200 ppm to zero,
in a system containing 3 tons of olivine bed and fed with 1.8 ton/h of
biomass. A similar drastic response was observed in a 26-kW bench-
scale diffusion flame during combustion of natural gas under slight sub-
stochiometric conditions and with the addition of 0–100 ppm of SO
2
[11]. In contrast, other work has reported that the addition of SO
2
or
higher levels of sulfur in the fuel inhibit CO oxidation, in a laboratory
flow reactor [12] and in the fluidized bed reactors [13,14], respec-
tively. Alzueta et al. [15] describe that sulfur promotes the oxidation of
CO under nearly stochiometric conditions, while it mildly inhibits it
under fuel lean and rich conditions. Yet, this explanation disagrees with
the observations at the Chalmers boiler, where the promoting effect of S
on CO oxidation occurred with excess air (3.5% vol in the flue gases or
excess air ratio 1.2).
Most previous research indicates that the sulfur chemistry has a
direct influence on CO oxidation [11,15], while the experience at
Chalmers with K-containing biomass and addition of S [3] also points at
interrelations between the S and K chemistry that affect the CO oxi-
dation. Hypothetically, this can occur via e.g. sulfation reactions [8,16]
that alter the speciation of K in the reaction environment. Similarly,
other species present in biomass such as chlorine could also interfere
with the K chemistry [12] and, thereby, influence the CO oxidation.
Chlorine by itself have been found to inhibit CO oxidation in a number
of investigations. Under moist conditions, Roesler et al. [17] found that
trace levels of HCl (0–200 ppm) at atmospheric pressure clearly in-
hibited CO oxidation in the temperature range of 500–900 °C. Wu et al.
[18] found that the inhibiting effect was more pronounced in oxygen-
rich conditions than at stochiometric or oxygen-lean conditions. The
inhibiting effect of chlorine has also been observed under fluidized bed
operation in the CANMET 0.8 MW CFB combustor using coal as fuel and
operating at 850 °C [19,20].
Biomass and waste fuels typically contain chlorine, sulfur and po-
tassium. An understanding of the CO oxidation in large scale combus-
tors calls for a better knowledge of the interplay of these species on the
oxidation chemistry. The aim of this work is to shed light on the gov-
erning mechanisms for insufficient CO burnout due to the presence of K,
despite adequate mixing and availability of oxidizer that has been ob-
served in large scale combustors. A laminar burner system was em-
ployed to map the influence of K in the oxidation of CO covering re-
action conditions relevant to combustion of biomass and solid waste in
fluidized bed combustors, i.e., mild temperature, excess air and co-ex-
istence of trace elements. CO oxidation was investigated in the presence
of K, S and Cl. The work involves a quantitative study in combustion
environments with well-controlled temperature and oxidation condi-
tions, with quantitative measurement of KOH, K-atom, OH radical, CO
and H
2
. The suitability of existing kinetic mechanisms to explain the
results was assessed and discussed.
2. Experimental
Experiments were carried out with a CH
4
/N
2
/O
2
flame generated in
a multi-jet burner and the flue gas composition was analysed under
various flame conditions, notably temperature and excess air (λ). The
conditions investigated are summarized in Table 1, where the excess air
ratio ranges from 1.04 to 1.67 and the temperature ranges from 1275 to
845 °C. The range was chosen to cover λ= 1.2 and temperature of
845 °C, as they correspond to the flue gas conditions typically found in
commercial fluidized bed combustors, as well as those applied in the
Chalmers system when the effect of K and S has been evidenced.
Seeding of K-, S- and Cl-containing species were conducted under
the flame conditions described in Table 1 to investigate their effect on
the oxidation of CO. Further details on the seeding system are described
in subsection 2.2. Two sets of experiments were defined to system-
atically address two questions: (1) the influence of potassium on the
oxidation of CO at different flame conditions; and (2) the influence of
sulfur and chlorine, respectively, on the oxidation of CO in co-existence
Fig. 1. Influence of accumulation of K in the bed material on the CO emissions
from the 12-MW Chalmers circulating fluid bed boiler, and effect of elemental S
in mitigating the CO emissions. Combustion of woody biomass at 850 °C with
excess air, i.e. 3.5%vol O
2
at the stack. Bed material: Olivine [3].
T. Berdugo Vilches, et al. Fuel 273 (2020) 117762
2

with potassium. The series of experiments carried out in each set are
listed in Table 2. In the first set, potassium was firstly added in different
concentrations at constant temperature and excess air ratio (series K1);
secondly, the excess air ratio was varied at constant temperature and
with a fixed seeding rate of potassium (series K2); and thirdly, the
temperature was varied at constant excess air ratio and seeding rate of
potassium (series K3). In the second experimental set, the influence of
sulfur in co-existence with a fixed seeding flow of potassium was in-
vestigated at different temperatures while keeping the excess air ratio
constant (series SK). Finally, the effect of chlorine was investigated with
a single test (labelled ClK in Table 2) that involved the simultaneous
addition of K and Cl, and that was compared to a similar case with
addition of K and in the absence of Cl.
The concentrations of CO and H
2
in the flue gas was measured with
a micro gas chromatograph (µ-GC); the concentration of K atoms by a
TDLAS system; and the concentrations of KOH, KCl and OH radicals by
UV absorption spectroscopy. The burner, seeding system and the
measurement techniques applied are described in detail below.
2.1. Burner
The multi-jet burner applied enables the generation of a hot gas
environments with an even temperature distribution. The detailed
structure of the burner was described by Weng et al. [21], and only a
brief description is given here. The burner was comprised of two
chambers, namely the jet chamber and the co-flow chamber. The pre-
mixed CH
4
/air/O
2
gas mixtures were introduced into the jet chamber
and evenly distributed to 181 steel tubes with an inner diameter of
1.6 mm. Premixed flames anchored on the jet tubes to provided hot flue
gas, which generated a hot gas environment above the outlet
(85 mm × 47 mm) of the burner for the present study. Each tube was
surrounded by even co-flow gases, air/N
2
, which was introduced
through the co-flow chamber. All the gas flows were controlled by mass
flow controllers (Bronkhorst). The temperature information was ob-
tained through a two-line atomic fluorescence thermometry system of
indium atoms as described by Borggren et al. [22].
2.2. Seeding of K, S and Cl
To introduce potassium into the hot gases, a water solution of po-
tassium carbonate with a concentration of 0.5 mol/l was used. As
shown in Fig. 2, solution fogs were generated through an ultrasonic
fogger and transported by air into the jet chamber. As the potassium
carbonate passed through the flames, gas-phase potassium hydroxide
can be formed in the hot gas. The potassium concentration in the hot
gas could be adjusted from 0 to 31 ppm by varying the flow rate of the
air for transporting solution fogs. The concentration of total potassium
seeded into the hot flue gas was estimated from the measurements of
KOH and K atoms. The seeding of trace amount of K
2
CO
3
was observed
to have negligible effect on the flue gas temperature.
In the experiments with sulfur, SO
2
was seeded from a gas cylinder
into the hot gas after the mixing with co-flow air/N
2
. To seed chlorine
to the flame, chloroform (CHCl
3
) was added into the hot gas system.
The CHCl
3
liquid was stored in a bubbler and kept at 10 °C using a
chilled bath (PolySicence). A N
2
flow was used to carry CHCl
3
vapor
into the jet chamber. Potassium chloride was generated by the co-
seeding of potassium carbonate. The seeding of sulfur was done at a
concentration of 150 ppm while the seeding of chlorine was approxi-
mately 400 ppm, which could fully convert KOH in to K
2
SO
4
or KCl at
the temperature of 845 °C.
2.3. GC system
The micro-gas chromatograph (µ-GC) was a Varian CP4900 with
two channels equipped with a PoraplotQ and a M S5Å column, and
using He and Ar as carrier gases, respectively. A sampling tube made by
stainless steel was used to extract the flue gas at a height of about 5 mm
above the burner outlet. A gas sample of approximately 300 mL was
collected in a gas bag with the help of a syringe. During sampling, the
gas was cooled down in the sampling tube, and it was filtered and dried
in a LC-NH2 adsorption column before entering the gas bag. Thereafter,
the gas bag was connected to the inlet of the µ-GC, so the instrument
could take samples from it.
For each test, one gas bag was filled with a sample gas, and its
content was analysed five times, i.e. five repeat chromatograms were
created per flame conditions. Each chromatogram is the result of a
point-injection (10–30 ms sampling time), which was taken in intervals
of 4 min to allow the necessary analysis time required by the µ-GC. The
concentrations of H
2
and CO presented in this work for each test are the
average of the five repeat chromatograms, and the error bars represent
the corresponding standard deviation.
2.4. TDLAS system
The TDLAS system applied in the measurement of K atoms has been
previously described in refs. [23,24]. In this system, a laser at wave-
length of 769.89 nm was used to detect the potassium atoms with its
transition of 4
2
S
1/2
⟶4
2
P
1/2
. The laser was provided by an external
cavity laser (Toptica, DL100), and it had a power of about 3 mW and a
beam size of about 1 mm
2
. The cavity laser was controlled by an
Table 1
Flame conditions investigated.
Flame Case Gas flow rate (sl/min) Excess
air ratio
λ
Gas product
Temperature T
(°C)
Jet-flow Co-flow
CH
4
Air O
2
N
2
Air
T1O1 2.69 13.29 2.80 30.30 0.00 1.04 1275
T2O1 2.48 12.91 2.45 35.35 0.00 1.04 1115
T3O1 2.27 11.84 2.25 40.40 0.00 1.04 985
T4O1 1.86 9.68 1.83 40.40 0.00 1.04 845
T4O2 1.86 9.68 1.83 35.35 5.04 1.32 845
T4O3 1.86 9.68 1.83 29.24 11.13 1.67 845
Table 2
Experimental series according to the question investigated.
Experimental sets Series Variable Range of variation Flame conditions* Others
(1) Effect of potassium on CO oxidation K1 Potassium seeding 0–32 ppm T4O1
K2 Excess air ratio 1.04–1.67 T4O1, T4O2, T4O3 K-seeding11 ppm
K3 Temperature 845–1275 °C T1O1, T2O1, T3O1, T4O1 K-seeding 11 ppm
(2) Effect of Cl and S on CO oxidation in co-existence with K SK Temperature 845–1275 °C T1O1, T2O1, T3O1, T4O1 K-seeding 11 ppm
S-seeding 150 ppm
ClK ––T4O1 K-seeding 11 ppm
Cl-seeding 400 ppm
*Flame conditions described in detail in Table 1.
T. Berdugo Vilches, et al. Fuel 273 (2020) 117762
3

analogy control package with a temperature control module (DTC 110),
a current control module (DCC 110), and a scan module (SC 110). The
scan module was adopted to make the laser have scanning of the wa-
velength with a range over 35 GHz at a repetition rate of 100 Hz. The
power of the laser was monitored by two photodiodes (PDA100A,
Thorlabs), and the scanning range was measured by a high-finesse
confocal Fabry-Perot etalon (Topoca, FPI 100), shown in Fig. 2(b). As
the laser passed through the hot flue gas containing potassium atoms,
the light was absorbed and the concentration of potassium atoms can be
derived through the Beer-Lambert law [24,25]. In the present study, the
laser beam located at about 5 mm above the burner outlet with a path
length of 8.5 cm. The potassium line was determined to have a Voigt
profile with a full width at half maximum (FWHM) of 5.34 GHz.
2.5. UV absorption spectroscopy system
In the experiments, the concentrations of K atoms, KOH, KCl and OH
radicals were measured by UV absorption spectroscopy in the hot gas
environments. As shown in Fig. 2(c), a deuterium lamp was used to
generate a collimated UV light with a beam size of about 5 mm. The UV
light was guided by five UV-enhanced aluminium mirrors to pass
through the hot gas with a path length of 522 mm [26]. After the
passage, the light was collected and analysed by a spectrometer.
The experimental absorbance data were obtained through the di-
vision between the intensity of UV light after the passage of the flame
with potassium seeding and the one without potassium seeding. It was
used to determine the concentration of KOH and KCl in the hot gas
through the Beer-Lambert law. In the calculation, the UV absorption
cross-sections of KOH and KCl measured by Weng et al. [24] was used.
Typical UV absorption spectrum obtained in KOH concentration mea-
surement are shown in Fig. 3 represented by the black dots. Two ab-
sorption peaks centred at around 250 nm and 330 nm can be observed
due to the absorption of gas-phase KOH, as described in our previous
studies. The absorbance of KOH was simulated using its UV absorption
cross-sections, and it could be well overlapped to the measured results
as the concentration was set to 11.5 ppm (see Fig. 3 with a red line).
The differences between the measured one and the simulated one are
the additional absorption at around 404 nm and two dips at around
283 nm and 310 nm. The former one was from the K-atom absorption,
while those two dips were formed due to the reduction of OH radicals in
the flame after the seeding of potassium. In the measurement of OH
radicals, the absorption at around 310 nm were obtained through the
division between the intensity of UV light after the passage of the flame
and the one without flame. A typical experimental result was shown in
the inset of Fig. 3 with black dots. This measured OH absorbance can be
fitted (as shown by the solid line) to retrieve the OH concentration
Fig. 2. Schematic of the experimental setup. (a) The structure of the burner system with GC sampling system. (b) TDLAS system for K-atom measurement. (c) UV
absorption spectroscopy system for quantitative measurements of KOH, KCl and OH radical.
Fig. 3. UV absorption spectrum obtained for KOH and OH measurement. The
measured absorbance values were presented with black dot and the corre-
sponding fitting curves based on UV-absorption cross-section of KOH [24] and
OH [27,28] were shown with red lines. (For interpretation of the references to
colour in this figure legend, the reader is referred to the web version of this
article.)
T. Berdugo Vilches, et al. Fuel 273 (2020) 117762
4

through LIFBASE [28]. As a result of the line-of-sight nature of the
absorption spectroscopy technique, the concentration obtained by this
method is the average concentration along the horizontal direction
from the centre to the edge of the hot flue gas.
3. Modelling
The simulations were conducted using Chemkin PRO [29] as de-
scribed by Weng et al. [30]. A one-dimensional stagnation reactor in
Chemkin PRO was adopted to simulate the reaction occurring in the hot
flue gas along the central axial direction. The gas inlet of the reactor
was assigned to be the mixture of the hot flue gas at 3 mm from the jet
flame fronts and the co-flow gas. The composition of the hot flue gas
produced from the premixed jet flames was obtained using a one-di-
mensional free propagation premixed flame model in Chemkin PRO
with the GRI-3.0 mechanism [31]. The temperature along the axial
direction in the reactor was set to be the one measured by a type B
thermocouple in the burner. The measured temperature was corrected
based on the heat transfer theory due to heat losses by thermal radia-
tion. The axial distance in the reactor was set to 64 mm, which was the
distance between the flame front and the flow stabilizer. The horizontal
distribution of the species concentration was obtained using a one-di-
mensional opposed-flow model in Chemkin PRO as reported by Weng
et al. [30]. Through this model, the reaction between the hot flue gas
and the ambient air on the edge of the hot flue gas was mimicked.
Average concentrations on the horizontal direction were derived from
the horizontal distribution of species obtained in the simulations to
obtain a result that could be compared to the measurements by ab-
sorption spectroscopy. Thus, both the theoretical and experimental
concentrations shown in the results section refer to the average on the
horizontal direction.
The modelling was conducted with a chemical kinetic model that
accounted for both the K–Cl–S interactions, including chlorination and
sulfation of potassium, and the impact of the trace species on the O/H
radical pool. With a few changes, the K–Cl–S mechanism reported by
Weng et al. [16] was chosen for the calculations. This mechanism draws
on previous work on alkali metals [32–34] as well as sulfur [35] and
chlorine [36] chemistry. In the present work, the alkali subset was re-
vised, with a particular emphasis on the interaction of potassium spe-
cies with the radical pool. The full mechanism is available as
Supplementary Material.
Reactions of potassium species with the radical pool were discussed
in previous work [2,7,32]. Few of these steps have been characterized
experimentally, and most rate constants have been estimated by ana-
logy with reactions of sodium and lithium. Most of the K/O/H subset is
the same as that proposed by Glarborg and Marshall [32]. However, the
thermochemistry for KO
2
and the rate constant for
K+O
2
+M=KO
2
+ M were drawn from the more recent work of
Sorvajärvi et al. [37]. Under oxidizing conditions, the KO
2
radical may
play an important role due to its relatively high thermal stability. Alkali
superoxides have been reported to enhance radical removal in pre-
mixed flames [38–41] and KO
2
is believed to participate in the reaction
sequences leading to sulfation of KOH [42]. While the K + O
2
+M
reaction is now well established, there are no measurements available
for the chain terminating step KO
2
+OH=KOH+O
2
. In the present
work, we have chosen a rate constant for KO
2
+OHof10
14
cm
3
mol
−1
s
−1
; this value is possibly an upper limit. Another reaction of interest is
the oxidation of CO by KO
2
to form CO
2
+ KO. Perry and Miller [43] set
this step to be very fast, with a rate constant of 10
14
cm
3
mol
−1
s
−1
, but
their modelling predictions were not sensitive to the value. The Perry
and Miller rate constant, which is sufficiently fast to eliminate inhibi-
tion of CO oxidation by potassium species under the present conditions,
is not supported by any experimental or theoretical work, and we be-
lieve that a more realistic value is 2 × 10
12
cm
3
mol
−1
s
−1
.
Fig. 4. The variation of the concentration of K atom and OH radical with dif-
ferent amount of potassium seeded into the hot gas at temperature of 845 °C
and excess air ratio of 1.04.
Fig. 5. The variation of the concentration of CO and H
2
with different amount
of total potassium seeded into the hot gas at temperature of 845 °C and excess
air ratio of 1.04.
Fig. 6. The variation of the concentration of K atoms and OH radical at dif-
ferent excess air ratio with a temperature of 845 °C and 11 ppm K seeding.
T. Berdugo Vilches, et al. Fuel 273 (2020) 117762
5

4. Results
The concentration of CO, H
2
, KOH, OH and K for the experimental
series listed in Table 2 are shown in Figs. 4–11. The measurements are
presented and discussed in two parts corresponding to the two sets of
experiments in Table 2: those related to the influence of potassium on
CO oxidation at different operating conditions; and those related to the
influence of sulfur and chlorine in co-existence with K.
4.1. The effect of operating conditions on inhibition by KOH
Fig. 4 shows results for K atoms and OH measured in a hot flue gas
temperature of 845 °C and an excess air ratio of 1.04, i.e. series labelled
as K1 in Table 2. The concentration of K atoms increased with the total
potassium seeded but levelled out at high potassium loadings. The
concentration of OH decreased strongly as the total potassium con-
centration was increased from 0 to 2 ppm. With potassium levels above
2 ppm, the OH radical concentration in the hot gas was below the de-
tection limit. This shows that the presence of K atoms strongly enhances
the consumption of OH radicals in the hot gas environment.
The corresponding concentrations of CO and H
2
, measured by the µ-
GC, are shown in Fig. 5. The levels of CO and H
2
increased from 0 ppm
to 1800 ppm and 700 ppm, respectively, as the potassium seeding in-
creased to 31 ppm. As the concentration of potassium in the flame in-
creases, the concentrations of CO and H
2
increase until they level off.
This trend is in agreement with that obtained by Hindiyarti et al. [7] in
Fig. 7. The variation of the concentration of CO and H
2
at different excess air
ratio with a temperature of 845 °C and 11 ppm K seeding.
Fig. 8. The variation of the concentration of KOH (a) and K atoms (b) at different temperature and SO
2
seeding at excess air ratio of 1.04 and about 11 ppm of K
seeding.
Fig. 9. The variation of the concentration of OH radical at different tempera-
ture with an excess air ratio of 1.04. The red line indicates the case without K
and SO
2
seeding; black line indicates the case with K seeding but without SO
2
seeding; blue line indicates the case with K and SO
2
seeding. (For interpretation
of the references to colour in this figure legend, the reader is referred to the web
version of this article.)
Fig. 10. The variation of the concentration of CO at different temperature with
an excess air ratio of 1.04. The red line indicates the case without K and SO
2
seeding; black line indicates the case with K seeding but without SO
2
seeding;
blue line indicates the case with K and SO
2
seeding. (For interpretation of the
references to colour in this figure legend, the reader is referred to the web
version of this article.)
T. Berdugo Vilches, et al. Fuel 273 (2020) 117762
6

moist conditions and in the absence of oxygen. In their case, the con-
centration of potassium required to saturate the inhibiting effect of
potassium was one order of magnitude larger than that obtained in this
work in the presence of oxygen. Note that the increase in the levels of
unburned CO and H
2
is consistent with the depletion of OH shown in
Fig. 4. The results confirm that potassium has a strong inhibiting effect
on the oxidation of CO and H
2
.
The modelling predictions are in qualitative agreement with the
measurements, but there are quantitative differences. While the K atom
concentration is slightly underpredicted, the impact of K on the OH
level is significantly smaller than observed. The simulations correctly
predict a significant inhibition of the oxidation of CO by the addition of
potassium, but the effect on H
2
is underpredicted.
The inhibiting effect of potassium on the oxidation of CO/H
2
in the
hot flue gas was also observed at higher excess air ratios, i.e.,
λ= 1.04–1.67, as shown in Fig. 6. In this experimental series (series K2
in Table 2), the excess air was adjusted by modifying the amount of air
and N
2
in the co-flow. The temperature was kept constant to 845 °C and
the potassium seeding to 11 ppm. The concentration of K atoms de-
creased significantly with the increase of excess air ratio. However, all
the cases showed a significant consumption of OH. The concentrations
of CO and H
2
are shown in Fig. 7, where the inhibiting effect becomes
weaker with the increase of excess air ratio. Under the conditions
tested, the inhibiting effect of K on the oxidation of H
2
can no longer be
detected at excess air ratios above 1.35.
The simulation results are in reasonable agreement with the ex-
perimental data. Similar to the observations at λ= 1.04, the OH
concentration in the presence of potassium is overpredicted, while at
zero KOH feeding, OH is predicted within 30%.
In the experimental series K3 (see Table 2), the temperature was
varied in the range 845–1275 °C, while the excess air ratio was kept at
1.04 and the seeding of K
2
CO
3
corresponded to 11 ppm of total po-
tassium. The resulting concentrations of KOH and K atoms are pre-
sented in Fig. 8, and the concentration of OH in the cases with and
without K is shown in Fig. 9 (see cases without sulfur seeding in the
figures). Potassium can enhance the consumption of OH under different
temperatures. The corresponding concentrations of CO and H
2
in the
hot gas with and without potassium seeding are shown in Fig. 10 and
Fig. 11, respectively. The inhibition effect was weakened with the in-
crease of temperature and there is almost no effect as the gas tem-
perature increased to 1275 °C. This is in line with the findings of Ekvall
et al. [8] that found no inhibiting effect by potassium at temperatures
above 1000 °C.
The model predicts the variation in the concentration of potassium
and OH radicals as a function of flue gas temperature satisfactorily. The
difference between the trends of the concentration of K atoms obtained
in the simulation and in experiment might be caused by the small in-
homogeneity of the oxygen concentration in the hot flue gas con-
sidering the strong influence of oxygen concentration on the K atom
concentration as shown in Fig. 6, but still the simulation and experi-
mental results were at the same concentration level and much smaller
than the concentration of KOH as shown in Fig. 8.
4.2. Influence of sulfur and chlorine on K-inhibition
The results derived from the tests with seeding of sulfur are shown
in Figs. 8–11, where they are labelled as with SO
2
. In the cases with co-
seeding of potassium (series SK in Table 2), the addition of sulfur show
larger influence in the low temperature range compared to the cases at
higher temperature. For instance, with the addition of sulfur at 845 °C
the concentrations became closer to the conditions without potassium
seeding (compare black and blue solid lines in Figs. 10–11). This ob-
servation is in line with the sulfation reaction [8]. At 845 °C, the pre-
sence of sulfur reduces the formation of K atoms and promotes con-
version of KOH to K
2
SO
4
particles. Consequently, more OH survived at
this temperature, as shown in Fig. 9. Due to the removal of gas phase
potassium by the sulfation process the inhibitory effect of potassium is
mitigated.
The results prove the interplay between potassium and sulfur in the
CO oxidation under the conditions tested, and confirm the hypothesized
mechanism involved in the large scale tests at the Chalmers unit [3].
Therefore, the direct effect of sulfur in the oxidation process, as de-
scribed by several authors for fuels that are free of impurities [11–15] is
not sufficient to capture the influence of sulfur in combustion units with
more complex fuels. The complexity of the reaction process is increased
by the dependence of the behaviour of sulfur at different temperatures
as seen in Figs. 10–11 and air excess ratio as described by Alzueta et al.
[15].
In the test with seeding of chlorine (ClK in Table 2), the presence of
approximately 400 ppm of Cl caused all the potassium in the hot gas to
be converted into KCl, which was measured by the UV absorption
system, as shown in Fig. 8. After seeding of Cl, a much smaller level of K
atoms was detected; about 30 times smaller compared to the KOH case.
At the same time, the OH concentration increased from 0 ppm to
10 ppm, as presented in Figs. 8 and 9, while the H
2
and CO con-
centrations reduced to 0 ppm (cf. Figs. 10 and 11). This is in contrast to
the inhibiting effect of Cl on the oxidation of CO commonly reported in
literature [17–20], which emphasizes the relevance of the combined
effect of active species in complex systems. The data show clearly that
the presence of sulphur or chlorine effectively eliminate the inhibition
effect of potassium on the CO/H
2
.
The model predicts correctly the removal of KOH and K atoms with
SO
2
or Cl seeding, especially at the temperature of 845 °C. Potassium
sulfate and potassium chlorine are more stable in the gas phase than
potassium hydroxide, resulting in lower predicted concentrations of K
atoms in the hot flue gas and higher levels of OH radicals (cf. Fig. 9). As
a result, the inhibition effect on CO oxidation is weakened (cf. Fig. 10).
However, at the temperature above 985 °C, a larger deviation between
the simulation and experimental results was observed.
5. Conclusion
This work sheds light on the true mechanism behind insufficient
combustion in industrial furnaces in the presence of potassium despite
adequate mixing and availability of oxidizer. It is demonstrated that
trace level of potassium, i.e. tenths of ppm, inhibits the gas-phase oxi-
dation of CO and H
2
in the temperature range of 845–1275 °C and
equivalence ratio of 1.04–1.65. Up to 2000 ppm CO and 700 ppm H
2
was observed to survive the oxidation in oxygen rich hot environments
Fig. 11. The variation of the concentration of H
2
at different temperature with
an excess air ratio of 1.04. The red line indicates the case without K and SO
2
seeding; black line indicates the case with K seeding but without SO
2
seeding;
blue line indicates the case with K and SO
2
seeding. (For interpretation of the
references to colour in this figure legend, the reader is referred to the web
version of this article.)
T. Berdugo Vilches, et al. Fuel 273 (2020) 117762
7

due to the co-existence of trace amount of potassium. The experiments
showed that higher temperature and higher oxygen concentration, re-
spectively, weakens the inhibiting effect of potassium. The reduction of
OH concentration due to the existence of K atoms is experimentally
measured and identified as the main cause for the inhibition of the
oxidation of CO and H
2
. It was further shown that sulfation of the po-
tassium species led the effect to be eliminated. These findings provide
an explanation to the casualty between increasing CO concentration
and the presence of potassium in flue gases, as well as to the positive
impact of sulfur additions in large scale combustion units on the CO
emissions. Existing kinetic models capture the qualitative interrelation
between potassium, OH radicals, CO and H
2
, respectively, at 845 °C and
excess air ratio 1.04, as well as the effect of varying excess air ratio.
However, the inhibiting effect of potassium is more intense in practice
than what the predictions indicate. The model fails to predict the ex-
perimental trends at varying temperature when sulfur and potassium
coexist in the combustion environment, and the predictions of H
2
concentration are generally less accurate than those for CO. Overall,
this work serves as a basis for the improvement of the theoretical de-
scription of the reaction mechanism involved in the inhibition of H
2
and
CO oxidation by potassium species and calls for the revision of existing
mechanisms.
Declaration of Competing Interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influ-
ence the work reported in this paper.
Acknowledgements
The research leading to these results received funding from the
Swedish Energy Agency (STEM) through the CECOST project and the
Swedish gasification centre SFC.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://
doi.org/10.1016/j.fuel.2020.117762.
References
[1] M.-o. energidepartementet, Förordning (2013:253) om förbränning av avfall, in: M.-
o. energidepartementet (Ed.), 2013.
[2] Glarborg P. Hidden interactions—trace species governing combustion and emis-
sions. Proc Combust Inst 2007;31:77–98.
[3] Thunman H, Seemann M, Berdugo Vilches T, Maric J, Pallares D, Ström H, et al.
Advanced biofuel production via gasification –lessons learned from 200 man-years
of research activity with Chalmers’research gasifier and the GoBiGas demonstration
plant. Energy Sci Eng 2018;6:6–34.
[4] Marinkovic J, Thunman H, Knutsson P, Seemann M. Characteristics of olivine as a
bed material in an indirect biomass gasifier. Chem Eng J 2015;279:555–66.
[5] Bulewicz EM, Janicka E, Kandefer S. Halogen inhibition of CO oxidation during the
combustion of coal in fluidized bed; 1989. p. 163–8.
[6] Lissianski VV, Zamansky VM, Maly PM. Effect of metal-containing additives on nox
reduction in combustion and reburning. Combust Flame 2001;125:1118–27.
[7] Hindiyarti L, Frandsen F, Livbjerg H, Glarborg P. Influence of potassium chloride on
moist CO oxidation under reducing conditions: experimental and kinetic modeling
study. Fuel 2006;85:978–88.
[8] Ekvall T, Andersson K. Experimentally observed influences of KCl and SO2 on CO
oxidation in an 80 kW oxy-propane flame. In: International Pittsburgh Coal
Conference, Pittsburgh, USA; 2017.
[9] Friedman R, Levy JB. Inhibition of opposed jet methane-air diffusion flames. The
effects of alkali metal vapours and organic halides. Combust Flame
1963;7:195–201.
[10] McHale ET. Flame inhibition by potassium compounds. Combust Flame
1975;24:277–9.
[11] Andersen J. Emission control through sulfur addition. M.Sc. Thesis, Department of
Chemical Engineering, Technical University of Denmark, 2006.
[12] Dam-Johansen K, Åmand L-E, Leckner B. Influence of SO2 on the NON2O chemistry
in fluidized bed combustion: 2. Interpretation of full-scale observations based on
laboratory experiments. Fuel 1993;72:565–71.
[13] Anthony EJ, Lu Y. Relationship between SO2 and other pollutant emissions from
fluidized-bed combustion. Symp (Int) Combust 1998;27:3093–101.
[14] Miccio F, Löffler G, Wargadalam VJ, Winter F. The influence of SO2 level and
operating conditions on NOx and N2O emissions during fluidised bed combustion of
coals. Fuel 2001;80:1555–66.
[15] Alzueta MU, Bilbao R, Glarborg P. Inhibition and sensitization of fuel oxidation by
SO2. Combust Flame 2001;127:2234–51.
[16] Weng W, Chen S, Wu H, Glarborg P, Li Z. Optical investigation of gas-phase KCl/
KOH sulfation in post flame conditions. Fuel 2018;224:461–8.
[17] Roesler JF, Yetter RA, Dryer FL. The inhibition of the CO/H20/02 reaction by trace
quantities of HCl. Combust Sci Technol 1992;82:87–100.
[18] Wu D, Wang Y, Wei X, Li S, Guo X. Experimental investigation and numerical si-
mulation of CO oxidation with HCl addition. Fuel 2016;179:221–8.
[19] Julien S, Brereton CMH, Lim CJ, Grace JR, Anthony EJ. The effect of halides on
emissions from circulating fluidized bed combustion of fossil fuels. Fuel
1996;75:1655–63.
[20] Wang J, Anthony EJ. CO oxidation and the inhibition effects of halogen species in
fluidised bed combustion. Combust Theor Model 2009;13:105–19.
[21] Weng WB, Borggren J, Li B, Alden M, Li ZS. A novel multi-jet burner for hot flue
gases of wide range of temperatures and compositions for optical diagnostics of
solid fuels gasification/combustion. Rev Sci Instrum 2017;88.
[22] Borggren J, Weng WB, Hosseinnia A, Bengtsson PE, Alden M, Li ZS. Diode laser-
based thermometry using two-line atomic fluorescence of indium and gallium. Appl
Phy B-Lasers Opt 2017;123.
[23] Gao Q, Weng WB, Li B, Li ZS. Quantitative and spatially resolved measurement of
atomic potassium in combustion using diode laser. Chin Phys Lett 2018;35.
[24] Weng W, Brackmann C, Leffler T, Aldén M, Li Z. Ultraviolet absorption cross sec-
tions of KOH and KCl for nonintrusive species-specific quantitative detection in hot
flue gases. Anal Chem 2019;91:4719–26.
[25] Qu ZC, Steinvall E, Ghorbani R, Schmidt FM. Tunable diode laser atomic absorption
spectroscopy for detection of potassium under optically thick conditions. Anal
Chem 2016;88:3754–60.
[26] Weng WB, Leffler T, Brackmann C, Alden M, Li ZS. Spectrally resolved ultraviolet
(UV) absorption cross-sections of alkali hydroxides and chlorides measured in hot
flue gases. Appl Spectrosc 2018;72:1388–95.
[27] Luque J, Crosley DR. LIFBASE: database and spectral simulation program (Version
1.5). SRI International Report MP 99-009; 1999.
[28] Luque J, Crosley DR. LIFBASE: database and spectral simulation program (Version
1.5); 1999.
[29] Design R. CHEMKIN-PRO 15131 San Diego; 2013.
[30] Weng W, Zhang Y, Wu H, Glarborg P, Li Z. Optical measurement of KOH, KCl and K
for quantitative K-Cl chemistry in thermochemical conversion processes, Submitted
to Fuel.
[31] Smith GP, Golden DM, Frenklach M, Moriarty NW, Eiteneer B, Goldenberg M,
Bowman CT, Hanson RK, Song S, Gardiner WC, VVL Jr, Qin Z. GRI-Mech 3.0.
http://www.me.berkeley.edu/gri_mech/.
[32] Glarborg P, Marshall P. Mechanism and modeling of the formation of gaseous alkali
sulfates. Combust Flame 2005;141:22–39.
[33] Hindiyarti L, Frandsen F, Livbjerg H, Glarborg P, Marshall P. An exploratory study
of alkali sulfate aerosol formation during biomass combustion. Fuel
2008;87:1591–600.
[34] Li B, Sun Z, Li Z, Aldén M, Jakobsen JG, Hansen S, et al. Post-flame gas-phase
sulfation of potassium chloride. Combust Flame 2013;160:959–69.
[35] Song Y, Hashemi H, Christensen JM, Zou C, Haynes BS, Marshall P, et al. An ex-
ploratory flow reactor study of H2S oxidation at 30–100 bar. Int J Chem Kinet
2017;49:37–52.
[36] Pelucchi M, Frassoldati A, Faravelli T, Ruscic B, Glarborg P. High-temperature
chemistry of HCl and Cl2. Combust Flame 2015;162:2693–704.
[37] Sorvajärvi T, Viljanen J, Toivonen J, Marshall P, Glarborg P. Rate constant and
thermochemistry for K + O2 + N2 = KO2 + N2. J Phys Chem A
2015;119:3329–36.
[38] Kaskan WE. The reaction of alkali atoms in lean flames. Symp (Int) Combust
1965;10:41–6.
[39] McEwan M, Phillips L. Dissociation energy of NaO2. Trans Faraday Soc 1966;62.
[40] Carabetta R, Kaskan WE. Oxidation of sodium, potassium, and cesium in flames. J
Phys Chem 1968;72:2483–9.
[41] Hynes AJ, Steinberg M, Schofield K. The chemical kinetics and thermodynamics of
sodium species in oxygen-rich hydrogen flames. J Chem Phys 1984;80:2585–97.
[42] Mortensen MR, Hashemi H, Wu H, Glarborg P. Modeling post-flame sulfation of KCl
and KOH in bio-dust combustion with full and simplified mechanisms. Fuel
2019;258:116147.
[43] Perry RA, Miller JA. An exploratory investigation of the use of alkali metals in
nitrous oxide control. Int J Chem Kinet 1996;28:217–34.
T. Berdugo Vilches, et al. Fuel 273 (2020) 117762
8