1
SI Appendix for “Coal fly ash is a major carbon
flux in the Chang Jiang (Yangtze River)
basin
”
by Li et al.
Contents
SI Appendix Text S1-S7
SI Appendix Figures S1-S6
SI Appendix Tables S1-S2 (upl
oaded as Datasets files)
S1. Sample collection
To characterize organic carbon in the Chang
Jiang (Yangtze River) basin, we collected a
comprehensive set of samples from fly ash, shale,
river sediment, Holocene sediment, and standing
biomass. All sample information is compiled in A
ppendix Table S1. In summary, the fly ash samples
were collected from wasted coal ash piles from se
veral coal-fired power pl
ants, a coking plant, and
an aluminum plant in east China, and from a coal
ash pile aside a boiler-heat
er unit in the campus of
Nanjing University, China. Fresh shale samples we
re collected from sedimentary sequences in the
lower-middle CJ reach, nearby the
cities of Nanjing, Chaohu, and
Yichang, and the profiles were
excavated during sampling to remove the surface weat
hered layers. Plant leaf samples were collected
in the campus of Nanjing University. The Holocene core sediment samples were taken from the CM
97 core drilled at the Chongming Island at the Chang
Jiang estuary, which were identified as deltaic,
estuarine, fluvial and floodplain sedimentary facies
(1). The Chang Jiang river suspended sediment
samples were collected monthly from August
2007 to September 2008 in the lower reach near
Nanjing, covering a complete hydrol
ogical year. Sampling details and
results for the river suspended
sediments were presented in a prior study (2). All
sediment samples were dried at 40°C in the oven
and grounded using an agate pestle and mortar.
S2. Laboratory analyses
Decarbonation and bulk OC measurement
The Holocene core sediment, shale, and fly ash
samples were decarbonated using 5 M HCl at 75 °C
to remove detrital carbonates follo
wing a previous work studying OC
in river sediments in the Andes
(3). Bulk OC content was measured on decarbonated samples using an element analyzer. Samples
were weighed before and after decarbonation to
correct for the OC content in raw samples.
Chemical oxidation
We conducted oxidation experiments following the prot
ocol described in ref.(4
). In summary, a solid
sample of 0.5 g was put in 250 mL DI wate
r and the mixture was sonicated. Then, 20 g Na
2
S
2
O8 and
22 g NaHCO
3
(as buffering agent) were added to th
e solution. The oxidation experiment was
conducted at 80 °C for 48 hours on a heater equippe
d with a magnetic stirrer. After oxidation, the
sample was treated with 20 mL 0.01 M HCl and 20
mL DI water to remove any trace carbonate
produced during oxidation. The sample was then washed
with DI water to neutralize pH, and dried in
the oven.
2
Raman spectral analysis
To characterize the structure of OC in the samp
les, Raman spectral analysis was performed on raw
samples using a Renishaw RM2000 Raman spectrometer in the School of Earth Sciences and
Engineering at Nanjing University, China.
We adopted a synchrosan band mode 100 cm
-1
to 2000
cm
-1
to capture the bands for inorganic minerals (200-1100 cm
-1
) and carbonaceous matter
(1300-1600 cm
-1
) (Fig. S2), with the other
configuration parameters set
as suggested in an earlier
study of OC in the Himalayan river systems (5).
The position of a standard silicon wafer was
repeatedly measured for calibration.
We decomposed the carbonaceous band into G and D bands (Figs. S2-S3), which correspond to
graphite and defects, respectively (5, 6). Followi
ng a prior study (6), we
resolved the carbonaceous
material band as G band and D1, D2, D3, and D4 ba
nds by fitting Lorentzian profiles to five Raman
peaks (Fig. S3).
Ramped pyrolysis oxidation (RPO) analysis
Ramped pyrolysis oxidation analysis was done on
a shale sample and a fly ash sample using the
instrumentation at the National Ocean Sciences
Accelerator Mass Spectrome
try (NOSAMS) facility.
Technical details and workflow were described in pr
evious studies (7-9). In brief, a solid sample
containing around 100-300 μg carbon was loaded into
a pre-heated quartz r
eactor in an oven. The
oven was programmed to heat at a ramp rate of 20 °C min
-1
, from ambient temperature to 1000 °C.
The experiment was done under the oxidation mode us
ing carrier gas composed
of 98% He and 2%
O
2
and with a gas flow rate of 35 mL min
-1
. As the sample was heated and oxidized, the eluent gas
was sent into an infrared gas analyzer where CO
2
concentration was measured at 1 s temporal
resolution. As a result, a thermogram was obtai
ned showing the evolution of the monitored CO
2
concentration over time and temperature. To
characterize OC reactivity and bond strength, we
converted the obtained thermogram
s to the activation energy (
E
a
) spectra using the ‘rampedpyrox’
Python package developed by Hemingway et al. (8)
(code available from th
e GitHub repository at
https://github.com/FluvialSeds/rampedpyrox
), which implemented an inverse model to find the
optimal solution of a set of parallel
first-order kinetic decay reactions.
S3. Sources for data compilation and synthesis
Coal consumption
The coal consumption data in China from 1950-2010 wa
s adopted from Table 7-2 in China Statistical
Yearbook (1999, 2012) by the National Bureau
of Statistics of China (NBSC)
(
http://www.stats.gov.cn/tjsj/ndsj/
,
http://www.stats.gov.cn/yearbook/indexC.htm
) (10, 11)
Fly ash production and utilization
The fly ash production and utilization
data were adopted from Yao et
al. (12) and references therein.
Specifically, the fly ash data in China were obtained from the
Annual Report on Comprehensive
Utilization of Resources of China
released by the National Devel
opment and Reform Commission of
the People's Republic of China (
http://www.gov.cn/gzdt/2013-04/08/content_2372577.htm
) (13).
China fly ash data were only avai
lable from 2001 to 2015 (12). To va
lidate the coal and fly ash data,
we use the two datasets to calculate an average fl
y ash content in coal by
linearly regressing the two
datasets with a defined Y-intercept
of 0 (Fig. S1). Thus we estimate
an average fly ash content of 17%
3
for coal consumed in China, well within the normal
range of fly ash content in coals of 5-20% (12).
To quantify fly ash production in
years prior to 2001 when fly ash da
ta is not available but coal
consumption data is available, we multiply the co
al consumption in those years by our calculated
average fly ash content of 17% in 2001-2015 (Fig. S1
). This approach likely provides a lower
estimate as the ash content is expected to have
decreased over time with improvement in combustion
efficiency and coal cleaning (Fig. S1d).
To determine the total fly ash production in the
CJ basin, we summed coal consumption in all
provinces within the CJ basin and
calculated the ratio between coal
consumption in the CJ basin and
the national total consumption, from 1990 to 2011 (T
able 4-15 in China Energy Statistical Yearbook,
2012, by the National Bureau of Statistics of the
People’s Republic of China) (14). We find an
average value of 36.3±0.1% (1990-2011) as the fraction
of coal consumed in the CJ basin versus in
the whole country.
To estimate how much fly ash is wasted rather than
utilized (Fig. 3c), we
calculate the difference
between the totally produced fly as
h and utilized fly ash. The utili
zed part is determined as the
product of the total production and
utilization rate. The utilization
data is available from 2001 to
2015 with an increase in proportion of total coal
combustion from 62% to 70%, as compiled by Yao
et al. (12) from the reports from
the National Development and Re
form Commission of the People's
Republic of China (12). For years prior to 2001, we
adopt a utilization pr
oportion of 62% as of 2001,
and recognize that actual utilizat
ion would be lower (wasted ash
would be higher) considering a
general growing trend over time (12). Thus our ca
lculations of fly ash pr
oduction and release in the
CJ basin provide a lower and conservative estimate.
Chang Jiang suspended sediment flux
The Chang Jiang sediment flux data for year wh
en samples were collected (2007-2008) was adopted
from the Chang Jiang Sediment Bulletin (2009) (
http://www.cjw.gov.cn/zwzc/bmgb/2018gb/
) (15) at
the Datong station in the lower CJ reach, as de
termined from hydrological gauging. The annual CJ
sediment flux data during 1950-2000s
was compiled in ref. (16).
Coal fire power plant
The map of coal-fired power plants in east and central China (Fig. 1b) is based on multiple databases
including the
Global Coal Plant Tracker
database (Global Energy Monitor, 2019,
https://endcoal.org/glob
al-coal-plant-tracker/
,
https://www.carbonbrief.org/mappe
d-worlds-coal-power-plants
) (17), and the
Carbon Dioxide
Emissions From Power Plants Worldwide
(CARMA) database
(
https://www.cgdev.org/article/ca
rma-v30-reveals-new-data-co2-emi
ssions-corporate-ownership-and
-locations-60000-power
) (18). We use this map to show the ge
neral regional patter
n, with coal-fired
power plants clustered in the middle-lower Chang
Jiang basin. These provide a major source of fly
ash and FOC
ash
. However, we do not use the map to c
onduct any quantitative calculations because
the databases may not be complete in documenting fly ash sources.
4
Large reservoir data
The map of large reservoi
rs (with capacity > 0.1 km
3
) in the Chang Jiang River basin was made from
multiple sources including ref. (
19), the ‘Database of the Basic Characteristics of Chinese Large and
Medium-Scale Reservoirs’ by the Inland Water Bi
ological Division of Ch
inese Biodiversity
Information Center
(
https://web.archive.org/web/20120330041630/ht
tp://brim.ihb.ac.cn/indexen.aspx
) (20), and a
database,
Code for China Reservoir Name
, compiled by the Ministry of Water Resources, People’s
Republic of China (2011) (21), with referen
ce to the satellite images on Google Earth.
S4. End-member mixing model in the oxidation fraction (
f
ox
)-1/OC space
Here we derive the mixing model used to distinguis
h the contributions of di
fferent components to the
riverine OC, with results presented in the main text (additional mixing trends used to constrain the
model are shown in Fig. 2d and SI Appendix Fig. S2).
Riverine-carried particulate OC is a mixture
of OC sourced from biosphere (OC
bio
) and lithosphere (OC
fossil
) (22, 23), which can be expressed
based on the mass balance of concentrati
ons within a given sediment sample:
[OC] = [OC]
bio
+ [OC]
fossil
(Eq. S1)
During the oxidation experiment, the oxidized carbon, [OC]
ox
, is composed of oxidized OC
bio
and
oxidized OC
fossil
, so we can write:
[OC]
ox
= [OC]
bio-ox
+ [OC]
fossil-ox
(Eq. S2)
where the subscripts ox, bio-ox, and fossil-ox mean
total oxidized carbon, oxidized biospheric OC,
and oxidized fossil OC, respectively.
For bulk OC in river sediments, we use an oxidation fraction (
f
ox
) to quantify OC loss during
oxidation experiments.
[OC]
ox
=
f
ox
×[OC] (Eq. S3)
We then define an analogous oxidation fraction for OC
bio
,
f
ox-bio
, which gives:
[OC]
bio-ox
=
f
ox-bio
×[OC]
bio
(Eq. S4)
Similarly, we define an
oxidation fraction for OC
fossil
,
f
ox-fossil
:
[OC]
fossil-ox
=
f
ox-fossil
×[OC]
fossil
(Eq. S5)
5
Combining Equations S2-S5, we have:
f
ox
×[OC] =
f
ox-bio
×[OC]
bio
+
f
ox-fossil
×[OC]
fossil
(Eq. S6)
We rewrite Eq. S6, yielding:
f
ox
=
f
ox-bio
×[OC]
bio
/[OC] +
f
ox-fossil
×[OC]
fossil
/[OC] (Eq. S7)
A prior study (2) showed that in
the Chang Jiang river sediment,
f
ox-fossil
= 0, which is expected
because after substantial remineralization and ox
idation of fossil OC during erosion and fluvial
transfer (5, 24-26), only the most
refractory component can be pr
eserved in the river sediment.
Taking this observation
into Eq. S7 gives:
f
ox
=
f
ox-bio
×[OC]
bio
/[OC] (Eq. S8)
As [OC]
bio
= [OC] – [OC]
fossil
(Eq. S1), we have:
f
ox
=
f
ox-bio
×([OC] – [OC]
fossil
)/[OC] (Eq. S9)
Letting
f
ox
= Y, and 1/[OC] = X, then:
Y =
f
ox-bio
–
f
ox-bio
×[OC]
fossil
×X (Eq. S10)
Thus, Equation S10 demonstrates
a linear relationship between
f
ox
and 1/[OC], where the intercept at
the X-axis is [OC]
fossil
, the intercept at the Y-axis is
f
ox-bio
, and the slope is their product,
f
ox-bio
×[OC]
fossil
(Figs. 2d, 2e and S2). By fitting lines to the observed data, we can determine these
unknown parameters.
S5. Flux estimate and error propagation
We use two primary approaches to determine the fr
action of fly ash-sourced fossil OC in the Chang
Jiang-exported fossil OC (
f
FOC-ash
).
Mass balance approach
We conduct a mass balance calculation between FOC
ash
and FOC in ash-uncontaminated river
sediment (FOC
CJ0
), as:
[FOC]
ash
×(
f
sed-ash
) + [FOC]
CJ0
×(1 -
f
sed-ash
) = [FOC]
CJ
(Eq. S11)
where
f
sed-ash
is the mass fraction of fly ash in
the CJ sediment flux, and [FOC]
CJ
is the apparent
content of FOC in the CJ sediment (0.45±0.10%) de
termined from radiocarbon measurements (2).
[FOC]
CJ0
is 0.15±0.02% from the oxidation experiment.
6
The fraction of FOC
ash
in the CJ-exported FOC is calculated as:
f
FOC-ash
= [FOC]
ash
×(
f
sed-ash
)/[FOC]
CJ
(Eq. S12)
And the riverine FOC
ash
flux is calculated as:
Q
FOC-ash
= [FOC]
ash
×(
f
sed-ash
) ×Q
sed
(Eq. S13)
where Q
sed
is the total sediment flux (~130 Mt yr
-1
in the sampling year) (15, 27).
To propagate errors on
f
sed-ash
,
f
FOC-ash
, and Q
FOC-ash
, we conduct Monte Carlo random sampling
calculations. In each iteration,
we randomly sample [FOC]
ash
following the probability distribution of
the measured histogram of [FOC]
ash
, sample [FOC]
CJ0
and [FOC]
CJ
following two normal
distributions of 0.15±0.02% (this st
udy) and 0.45±0.10% (2) (mean ± 1
σ
), respectively, and solve
Equations S11-S13 simultaneously. We conduct 10,000
iterations and generate a population of results.
We report the median values and define the uncertainties from the 16
th
-84
th
percentiles of the resulted
population. We also use this approach to estimate the
errors on the fraction of
riverine-carried fly ash
in the totally produced fly ash a
nd in the wasted (produced – util
ized) fly ash in the CJ basin.
Constraint from magnetic su
sceptibility (MS) study (28)
We refer to an independent study th
at observed increases in MS of CJ
sediment and attributed these
changes to input of fly ash (28). They estimate a
f
sed-ash
of 7% on this basis. Taking this value into
Equations S11-S13, we can estimate
f
FOC-ash
and Q
FOC-ash
, respectively, and propaga
te the errors using
the same Monte Carlo random sampling approach.
S6. Expanded discussion on FOC
ash
: origins, forms, separation, and controlling factors
FOC
ash
is also termed ‘unburned carbon’. Major res
earch efforts have been devoted to imaging,
characterizing, separating, and
recovering or removing FOC
ash
. Referring to recent studies (29-34)
and review papers (35-39) on FOC
ash
, below we summarized (1) the
industrial standards for FOC
ash
content, (2) the identifica
tion and classification of di
fferent carbon species in FOC
ash
, (3) the
separation methods of FOC
ash
, and (4) the controlling factors of FOC
ash
, to complement the
discussion of the characteristics of FOC
ash
in the main text.
Industrial standards of FOC
ash
content
FOC
ash
provides a measure for the efficiency of co
mbustion (39, 40). When utilizing fly ash as
concrete, air entraining agents are added to fly as
h to improve air entrainment performance, but are
easily absorbed by FOC
ash
(39-42). Thus FOC
ash
amounts need to be controlled for ash quality
assurance (34, 35, 39). As a result,
industrial standards of FOC
ash
contents have been established.
Dong et al. (34) compiled the
industrial standards of FOC
ash
content in different regions and
countries, finding that these are consistently ar
ound 5-10%. We list here
the standards for FOC
ash
contents for major coal-consumption countries and
regions. The US ASTM-C618 standard states that
for class-F (from anthracite and bituminous coal
, siliceous with <10% CaO by mass) and class-C
(from subbituminous and lignite coal, containing
mainly lime with 10-40% CaO) fly ash, loss on
7
ignition (LOI, as an approximation for FOC
ash
) needs to be within 6% if using ash for construction
materials (34, 37, 43). The Chinese
national standard GB/T477-2008 for
fly ash used in cement and
concrete states that LOI needs to
be within 5% and 8% for grade
I and II fly ash, re
spectively (34,
35). In Europe, the EU standard EN 450 states th
at LOI should be within 5%, 2-7%, and 4-9% for
category A, B, and C fly ash, respectively, for us
e in concrete and cement (34, 44). The Australian
standard AS3852.1 recommends LOI within 4% for use
in concrete (37, 45). The Indian standard
3812 states a upper limit of LOI as 5% (34, 44)
. The Russian standard GOST 25818 has permitted
limits of 20%, 10%, and 3% for type I fly ash fo
r anthracite coal, hard coal, and brown coal,
respectively (34, 46).
Classification of carbon species in FOC
ash
FOC
ash
contains a spectrum of carbon species with va
rying compositions, sizes, origins, structures,
and relationships with minerals. Here we introduc
ed the major methods employed to classify FOC
ash
to facilitate intercomparison.
Considering elemental composition, FOC
ash
can be classified as elem
ental carbon (EC) and organic
carbon (OC) (29, 30, 35). EC is mainly graphite or
graphitic carbon, whereas OC is carbon bonded to
other atoms such as hydrogen and oxygen (29, 30).
Observational studies based on scanning el
ectron microscopy (SEM) and high-resolution
transmission electron microsc
opy (HRTEM) indicated that FOC
ash
is composed of two major carbon
forms: individual carbon part
icles and mineral-associated
carbon (31, 35, 37, 41, 47). The carbon
particles include (1) monomer char
particles from incomplete co
mbustion featured by irregular
shapes with sizes of 1s-10s μm (37, 47, 48), and (2
) aggregates of fine carbon particles (1s-10s nm)
that are likely sourced from soot (37, 42, 48, 49). Th
e mineral-associated ca
rbon contains (1) carbon
attached to mineral surfaces, and (2) carbon embe
dded in minerals as inclusions (35, 47).
Other studies combined the sizes and charac
teristics of different
carbon species in FOC
ash
to infer
their sources and origins. Specifical
ly, soot-sourced carbon tends to clus
ter as spherical
particles, in
the size range of ones to tens of nm, that can attach
to the surfaces of and aggregate in the structures
of minerals and char par
ticles (ones to tens of μm), and form aggregates themselves (37, 42, 48, 49).
Char particles, from the incomplete combustion
of coal, have irregular shapes and highly porous
structures (37, 47, 48). Mineral-as
sociated carbon can contain inhe
rited carbon that
has fine sizes
from incomplete combustion of coal (e.g. the in
ertinite maceral), therma
lly altered carbon during
combustion, and carbon from other origins (e.g. soot) th
at later form associati
ons with ash particles
(30, 37, 38, 50).
Overall, these are the different
carbon species that compose FOC
ash
, and thus we accounted for these
in our budget calculations. These
forms of carbon may still have
have variable reactivity (e.g.
graphitic carbon vs. non-graphitic carbon), but are e
xpected to be substant
ially recalcitrant, as
evidenced by our chemical oxidation experiments
and the dominance of gr
aphitic structures and
mineral associations which are effective
mechanisms for carbon preservation (7, 51).
8
Separation of FOC
ash
Significant research efforts have
been devoted to separating th
e unburned carbon species from fly
ash to improve the efficiency of ash utilizati
on. Several studies summarized the major approaches
used for separating FOC
ash
including size separation, gravity
separation, electrostatic separation,
froth flotation and oil agglomeration, and therma
l processing (35, 52). Here we described the major
separation methods in brief, with
reference to recent review papers
and relevant studies (35, 37, 52),
to complement our discussion of the characteristics of FOC
ash
. The separation methods listed below
were developed based on characteristics of FOC
ash
including its size, grav
ity, structure, morphology,
pore sizes, and charge property, a
nd different methods have different
ways of separating the carbon
species.
Size separation, including both dry
and wet sieving, is a simple a
nd economical method to separate
and recover unburned carbon from fly ash. Size separa
tion through sieving is mo
st effective for fly
ash within which FOC
ash
is dominated by coarse, unburned pa
rticles such as char, but cannot
separate carbon species of finer si
zes (e.g. soot carbon attached to as
h particles) (35). This method is
often used as a first step to screen fly ash be
fore more advanced proc
essing (e.g. electrostatic
separation) (53).
Gravity separation is based on the
density differences between FOC
ash
and fly ash minerals (35).
Laboratory experiments showed th
at gravity separation combined with other techniques such as
centrifuging and triboelectrostat
ic enrichment can work effectively to remove unburned carbon
(e.g. >80%) for the coarse fraction of fly ash (54).
Thermal treatment is another commercially used a
pproach to remove unburned carbon from fly ash
(37). One commonly used equipment is fluidized-b
ed reactors, which can combust fly ash in a
continuous stream (55). Fluidized-
bed reactors can recycle burned
fly ash with high carbon content
to maintain bed temperature and reduce consumpti
on of supplemental fuel (55). This technique has
no requirement on the freshness of fly ash and can be
used to process long-dis
posed ash as well (37).
More advanced separation methods include elec
trostatic separation, fr
oth flotation, and oil
agglomeration, which have been widely used in
oil and mining industry (35, 48, 52, 54). Electrostatic
separation takes advantage of
the charge property of FOC
ash
: ash particles and unburned carbon have
distinct electron affinities, such
that during charging, fly ash partic
les tend to gain electrons and
becoming negatively charged, whereas unburned carbon mo
re easily loses electr
ons to be positively
charged (35, 49, 56). This method charges fly
ash and carbon particles, and once charged, the
charged fly ash and unburned carbon
particles are delivered to a hi
gh-voltage electrostatic field
where the particles carrying distin
ct charges and polarities are de
flected and separated (35, 55, 57).
Combining different charging mechanisms (e.g. tr
iboelectric charging and
induction charging) and
equipment (e.g. drum separator, triboelectric be
lt separator, and para
llel and louvered plate
separators), electrostatic separa
tion can be applied to both coarse
- and fine-sized (e.g. < 1 μm) fly
ash (56). Drying is a common procedure when prep
aring ash for electrostatic separation, because
moisture decreases the efficiency
of electrostatic
separation (57).
9
Froth flotation is a widely used technique in coal
and mining industry to sepa
rate materials based on
their hydrophobicity (58-60), and has been used
to separate the hydr
ophobic unburned carbon and
the hydrophilic ash minerals as well (35, 48, 61-63).
In froth flotation, fly ash is mixed with a
surfactant (e.g. polar reagents) and water, and is deliv
ered to a tank (i.e. flot
ation cell) where air is
introduced and bubbles are generated (48, 59, 60).
The hydrophobic carbon partic
les attach to the
bubbles and float to the surface forming a froth, wher
eas the hydrophilic ash minerals remain in the
solution and can be separated. Recent studies ha
ve focused on improving and developing flotation
devices (e.g. Denver flotation ce
ll, cyclonic-static microbubble flot
ation column, and concurrent
flotation column) and testing diffe
rent reagents (e.g. diesel oil
and acetic acid) to increase the
efficiency of bubble generation and
particle separation (63-67). Howe
ver, very fine carbon particles
collide less frequently wi
th bubbles, and thus cannot be effectivel
y separated using fr
oth flotation (35,
60, 62). Oil agglomeration complements froth fl
otation by separating unburned carbon from
fine-sized fly ash (68, 69). In brief, fly as
h is wetted by oil such that the hydrophobic, unburned
carbon particles are coated with oil and form ag
glomerates, whereas the hydrophilic, ash mineral
particles do not agglomerate and remain in susp
ension, with a recovery rate of unburned carbon as
high as ~50-60% (53, 69).
We note that although the above methods can separate FOC
ash
, none of them can quantitatively
recover all carbon species from fly ash. Each
method has its limitations and preferable carbon
species in FOC
ash
that can be recovered. For example,
the froth flotation method works more
effectively for carbon attached to mineral surf
aces. Removing carbon aggregated in mineral
structures often requires gri
nding and other methods to dissoci
ate mineral and carbon (35, 52, 54).
Controlling factors of FOC
ash
A large number of studies have focused on unders
tanding the factors determining the amount and
forms of FOC
ash
. Several reviews, notably Hower et al
. (37), have summarized those factors
comprehensively, and so here we briefly introduce t
hose factors. In general,
the controlling factors
are related to (1) the nature of coal and (2) th
e design, configuration, and
condition/operation of the
combustion systems (35, 37, 41, 70, 71). Modeling studies also show that the FOC
ash
content can be
predicted using these tw
o variables (32, 40).
The factors describing the nature of coal include
coal rank, sizes, maceral composition, mineralogical
composition, volatile content, moisture, and calori
fic value (32, 37, 40). Nota
bly, it has been shown
that decreasing particle size of
the feed coal can increase combustion efficiency and reduce the
FOC
ash
content (72, 73). The abundances of
different types of macerals (e
.g. inertinite, vitrinite, and
liptinite) that make up coal influence FOC
ash
contents as well, whereby the inertinite macerals are
more resistant during combustion than vitrinite
macerals (74-77). Fly ash from coals with high
proportions of intertinite m
acerals tend to have high FOC
ash
amount (74-77). Hower et al. (37)
showed in detail that coal rank plays
a key role regulating the forms of FOC
ash
as well. In summary,
the fly ash produced from low-rank (lign
ite and subbituminous) coal contains FOC
ash
of intact
inertinite-derived carbon, thermally
-altered vitrinite-sourced carbon,
and isotropic char (37, 78, 79).
FOC
ash
produced from medium-rank (bituminous) coal ma
inly contains unaltered inertinite-sourced
carbon, mixtures of isotropic and anisotropic coke
originated from vitrinite, and amorphous and
crystalline carbon in association with minerals
(49, 80-84). The fly ash
produced from high-rank
10
(anthracite) coal inherits car
bon from the coal macerals expe
riencing limited alteration during
combustion (85, 86).
The factors related to the combustion system incl
ude combustion time, coal feeding rate, damper
position, burner tilt, coal-oxygen-air ratios, temperat
ure and pressure in combustion systems, heat
flow rates, and flame patterns (32, 37, 40).
S7. Compilation of FOC
ash
content data in global fly ash samples
We compiled 247 FOC
ash
content data from previous studies
to constrain the amount of unburned
carbon in fly ash worldwide (Table S2) (30, 33, 40-42, 47, 48, 50, 52, 71, 87-96). Those data were
measured for fly ash samples from Australia, Chin
a, Canada, Spain, South Korea, South Africa, and
the US (Table S2). Most samples were industrial co
al fly ash collected from
power stations, with a
small fraction from fly ash produced from si
mulated combustion in laboratories. The FOC
ash
contents
were estimated either by loss on ignition (n = 212) or
by ultimate analysis (i.e. elemental analysis) (n
= 35) (30, 33, 40-42, 47, 48, 50, 52, 71, 87-96). The compiled FOC
ash
contents likely overestimated
the actual FOC
ash
contents, because loss on ignition (LOI) measures mass loss once burned to high
temperature (e.g. 950 °C, ASTM-Standard D7348,
2013) (37, 97) without excluding non-carbon
elements and volatiles, and elemental analysis
does not separate inor
ganic carbon (e.g. carbonate)
(89, 90). The compiled FOC
ash
contents show a skewed dist
ribution with a median of 4.70%
(1.30-14.39%, 16
th
-84
th
percentiles) and a major
ity of data within 10%. Note that the compiled
FOC
ash
content is comparable to the worldwide i
ndustrial standards of unbur
ned carbon content in
coal ash (5-10%, Appendix S6)
(34) and our measured FOC
ash
content in the CJ ash samples
(2.25
+1.63
/
-1.18
%), lending confidence to
our estimates of FOC
ash
fluxes.
Figure S1.
R
R
elationsh
i
Sup
p
i
p between
c
ortin
g
Inf
o
c
oal consu
m
11
o
rmation
A
m
ption rate
A
ppendix F
i
and fly ash
ig
ures
production
n
rate in Ch
i
na.