Fuel 187 (2017) 229–241

Contents lists available at ScienceDirect

Fuel
journal homepage: www.elsevier.com/locate/fuel

Full Length Article

Anode coke from coal – A low cost approach

John D. Craddock ⇑, Terry D. Rantell, James C. Hower, David T. Whitlow, John Wiseman,
Matthew C. Weisenberger
University of Kentucky, Center for Applied Energy Research, 2540 Research Park Drive, Lexington, KY 40511, United States

a r t i c l e i n f o a b s t r a c t

Article history: Declining quality and availability of petroleum cokes, which have traditionally been used for the produc-
Received 20 May 2016 tion of carbon anodes, is forcing aluminum producers to use cokes previously regarded as unsuitable.
Received in revised form 24 August 2016 Domestic reserves of coal represent a potential alternative carbon resource for anode production, pro-
Accepted 12 September 2016
vided coke specifications can be economically achieved. Previous studies have shown that the solvent
refining of coal could produce carbons suitable for aluminum production. To conform to coke specifica-
tions (e.g. metals content), it was necessary to remove mineral matter in the coal by a costly filtration
Keywords:
step. This adds significantly to the cost of the process, making it uncompetitive with coke from other
Anode coke
Low-ash coal
sources. However, cokes with properties very similar to the required specifications have been produced
Kentucky from very low ash coals by solvent extraction, but without the uneconomic solids removal step.
Petroleum coke Extraction in a suitable solvent was used to produce a coal digest, which when injected at high temper-
Low-cost ature into a coking vessel generated the anisotropic needle structure characteristic of high quality cokes.
Reported herein are the results of our investigation into using low-ash coals (1–2 wt.%) from the south-
eastern Kentucky coalfield to produce anode-grade cokes using similar methods. The high cost filtration
step was omitted, but the resulting coke did not meet the anode grade specifications.
Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction cracking processes; fluid catalytic cracking (FCC), hydrocracking

Carbon anodes used in the primary smelting of aluminum have
traditionally been manufactured from calcined petroleum coke [1].
The electrodes are made by blending the coke with a small fraction
of coal tar pitch as binder. The green anodes are then baked to pro-
duce carbon anodes with sufficient density to meet the required
performance standards. Consumption rates of the anodes during
primary aluminum smelting, in the molten fluoride salt electrolysis
cell are high, approaching 0.5 ton of carbon for each ton of alu-
minum produced [2]. Hence, large quantities of coke are required
for carbon anode manufacture to satisfy the demands of the alu-
minum industry. Worldwide, the majority (>85%) of calcined pet-
roleum coke is used to produce anodes for the primary smelting
of aluminum from alumina in the Hall-Héroult process [3].
Petroleum coke is a by-product from the oil refining process.
Modern refinery practice, controlled by the economics of the petro-
leum market, is aimed at maximizing the yield of lighter high value
products. Therefore, emphasis is placed on the conversion of the
heavy liquids from the distillation columns such as middle distil-
late, oil and residuum to higher value products by one of several
⇑ Corresponding author.
E-mail address: john.craddock@uky.edu (J.D. Craddock).
or thermal cracking (coking). The heaviest material (bottoms) is
fed to a delayed coker where thermal cracking of the large molec-
ular species occurs, which yields high value lighter products. Con-
sequent polymerization of the residuum results in the formation of
coke as a by-product. This ‘green coke’ is mechanically removed
from the coking drum and calcined by heating to 1300 °C to
remove volatiles and produce a material suitable for anode produc-
tion. Impurities in the crude oil become concentrated in the coke
and can have undesirable effects on aluminum production. For
example, metal impurities such as vanadium and nickel catalyze
carbon oxidation reactions, causing higher carbon consumption
in the electrolysis cell [4]. Stability of the coke in air and carbon
dioxide is a vital consideration in the economics of its use in
anodes. Other impurities collect in the refined aluminum and can
lower its value. Table 1 shows generic acceptable levels of impuri-
ties in the coke.
Impurity levels in petroleum cokes have been progressively
increasing in recent years as the refineries have been obliged to
accept a higher proportion of heavy sour crudes [6,7]. This trend
and changes in refinery practice have resulted in an increase in
the sulfur content of the petroleum coke, and/or an increasing coke
cost [1]. An alternative supply of anode coke is desired to
supplement calcined petroleum coke supplies, provided defined

http://dx.doi.org/10.1016/j.fuel.2016.09.045
0016-2361/Ó 2016 Elsevier Ltd. All rights reserved.
230 J.D. Craddock et al. / Fuel 187 (2017) 229–241
Table 1
Generic coke specifications for anode production [5].
Component Value
Ash 60.1 wt.%
S 61.0 wt.%
Fe 6300 ppm
Si, V & Ni 6200 ppm
Na 6100 ppm
Ca 650 ppm
P 65 ppm
specifications can be met and at a cost that is economically sus-
tainable. The significant domestic reserves of coal could represent
a viable carbon source for anode production. Specifically, the Blue
Gem coal in southeastern Kentucky is of initial interest due to its
inherently low ash content [8]. Cokes with properties very similar
to the required specifications have been produced from very low
ash coals by solvent extraction, in the absence of the uneconomic
solids removal step [9,10].
The principal objective of this research effort was to determine
whether Kentucky coal resources could be converted into coke
suitable for anode fabrication through a mild solvent extraction
process. Previous work has shown that the solvent refining of coal
yields carbons suitable for aluminum production [5]. Low ash coals
were extracted in a process solvent at relatively low pressure and
without the use of hydrogen gas. In addition, the uneconomic solids
removal step commonly used to separate any undissolved solids pre-
sent in the coal digest from the coal solution was omitted. Coals for
these tests were selected based on their inherently low mineral
matter content and their known coking behavior on the premise
that the structure and properties of the cokes produced from them
would likely be suitable for anode fabrication. High volatile A bitu-
minous coals of Eastern Kentucky produce anisotropic cokes, and
many have been used as blend constituents for production of met-
allurgical coke and were thus selected as prime candidates for this
application. Current anode coke specifications require very low
metals and sulfur contents, with V, Ni, Na, Ca, Si, and Fe being of
particular importance (see Table 1). To achieve anything close to
the current pet coke specs in terms of metals content, it was essen-
tial to use feed coals with minimal mineral matter content. To fur-
ther reduce the mineral matter content of the low-ash feed coals, a
coal cleaning procedure, selective oil agglomeration, was used on
the finely milled coals. This further cleaning of the coal would
remove mineral matter, which can poison the formation of the ani-
sotropic needle-like carbon structure [11] formed during coking
and required in a high quality anode coke. Removal of particulates
from the feed that inhibit growth and coalescence of the develop-
ing mesophase in the coking process, would lay the foundation for
achieving these goals. The as-received and cleaned coals were ana-
lyzed for composition and compared. A bench scale mild solvent
extraction process developed by University of Kentucky Center
for Applied Energy Research (CAER) to convert bituminous and
subbituminous coals to high value carbon products was used as
the basis for the work [5].
The work presented here addressed the issue of whether this
process could be used to produce cokes that exhibit the properties
and structure required for fabrication of carbon anodes and at an
economically viable cost, achieved by omission of the costly filtra-
tion step. The coal digests derived by this process were converted
into green coke by heating, under a N2 sparge of 230–260 SCCM to
promote solvent stripping and to simultaneously shear the poly-
merizing residue, at 380 °C for 1.5, 2.5 and 5.0 h, and then at
500 °C for 2.5 h. Subsequently, the green cokes were heat treated,
under an inert atmosphere to 1300 °C at 50 °C/min with a 1-h soak
time, to produce calcined cokes. The final cokes were characterized
by X-ray diffraction, optical microscopy and compositional
analysis.
2. Experimental
2.1. Identification, selection and procurement of candidate coals
Low ash coal samples from sites in the southeastern Kentucky
coalfield were identified and sourced. Three coals were selected
from the suite of candidate samples for conversion into coke by
solvent extraction and coking. Candidate coals were limited to
low-ash, low-S, high volatile A bituminous coals from southeastern
Kentucky: the Blue Gem coal from the Logan Hollow and Harps
Maple Creek mines, and the Jellico coal (sample no. 3037). These
coals were known to contain very little mineral matter (ash); in
the 1–2 wt.% range. The mineral content in the Blue Gem is marked
by low amounts of aluminosilicates and is dominated by siderite
[5,12–14]. Conversely, efforts have been made to arrive at low-
ash coal by chemically and thermally treating lower cost, high-
ash coals [15,16]. Other solvent extraction of coal research has
focused on producing one- and two-ring compounds for the jet fuel
industry [17].
Ash is known to hinder the formation of long-range anisotropy
in the final cokes [11]. Further, a central hypothesis was that the
ash content could be reduced even further, perhaps to <0.1 wt.%
through a coal cleaning (oil or pentane agglomeration) procedure,
to separate the ash fines from the coal. The cleaned ultra-low ash
coal would then be used to produce anode grade coke without
the need for a costly filtration step to remove ash from the coal
digests. The results of compositional analyses of the three starting
coals are shown in Table 2a. The native ash contents of the Logan
Hollow Blue Gem, Harps Maple Creek Blue Gem, and Jellico coals
were 1.02, 1.42, and 2.31 wt.%, respectively. Similarly, the metals
contents of these coals, stemming from the ash, were higher than
the generic coke specifications (Table 1). However, the sulfur con-
tent of the coals was quite low at 0.77, 1.08, and 0.99 wt.%, respec-
tively. The total vitrinite content (mineral-free basis) of the three
coals ranges from 71% for the Jellico coal to 75.4–77.6% for the
two Blue Gem samples (Table 2b). The Blue Gem samples have
lower inertinite content than the Jellico sample.
2.2. Coal cleaning: Pentane agglomeration procedure
Pentane agglomeration, commonly referred to as oil agglomer-
ation, was investigated as a method to reduce the ash content of
the selected coals to produce a desirable coking feedstock. Ash
levels of less than 0.1 wt.% were targeted, starting from coals con-
taining 1–2 wt.% ash. Oil agglomeration is a procedure commonly
used to separate finely disseminated hydrophobic and hydrophilic
minerals [18]. The procedure consists of adding hydrophobic oil to
a finely ground coal and water slurry. Upon agitation hydrophobic
agglomerates are formed. The mineral matter liberated from the
coal particles by the fine-grinding procedure accumulates in the
hydrophilic fraction of the product and can then be separated from
the organic hydrophobic fraction by decantation. The process has
been used successfully in the past at the industrial scale; however
no commercial installations are currently operating.
Pentane was used as the agglomerating oil, due to its low boil-
ing point and the fact that it leaves no residual components to
interfere with other upgrading processes. The parent coal was
slurred with water and ground (33% solids w/w) in a ball mill to
a size sufficient for liberation of the finely disseminated mineral
matter. Approximately 0.5 L of the ground slurry was transferred
to a blender, 25 ml of pentane added and the mixture was
subjected to intense agitation at 1000 rpm for 30 s to form
 J.D. Craddock et al. / Fuel 187 (2017) 229–241 231

Table 2a
Coal analyses.

hydrophobic agglomerates – coal particles bound together by agglomerated slurry was then allowed to equilibrate for 2 min,
pentane bridges. This was followed by an additional 30 s of low during which time agglomerates formed a stable mass above the
speed agitation (200 rpm) to produce larger agglomerates. The water phase, which contained non-agglomerated hydrophilic
232 J.D. Craddock et al. / Fuel 187 (2017) 229–241

Table 2b
Petrology of coals in study on mineral-included and mineral-free basis for each coal (left and right columns, respectively); with subtotals for vitrinite subgroups, totals for the
maceral groups and for the minerals, vitrinite maximum (Rmax) and random (Rrandom) reflectances and their respective standard deviations, and the V-types (v8, v9, and v10: the %
of Rmax readings from 0.80% to 0.89%, 0.90% to 0.99%, and 1.00% to 1.09%, respectively).

71,806 71,807 71,808

Jellico Blue Gem Blue Gem
(Logan Hollow) (Harps Crk./Maple Crk.)
Telinite 6.4 6.4 10.0 10.0 8.0 8.1
Collotelinite 51.2 51.5 52.0 52.0 49.8 50.3
Total telovitrinite 57.6 57.9 62.0 62.0 57.8 58.4
Vitrodetrinite 11.2 11.3 11.6 11.6 15.0 15.2
Collodetrinite 0.0 0.0 0.0 0.0 0.0 0.0
Total detrovitrinite 11.2 11.3 11.6 11.6 15.0 15.2
Corpogelinite 1.6 1.6 1.8 1.8 4.0 4.0
Gelinite 0.0 0.0 0.0 0.0 0.0 0.0
Total gelovitrinite 1.6 1.6 1.8 1.8 4.0 4.0
Total vitrinite 70.4 70.8 75.4 75.4 76.8 77.6
Fusinite 5.6 5.6 3.4 3.4 2.0 2.0
Semifusinite 7.6 7.6 2.4 2.4 2.6 2.6
Micrinite 1.8 1.8 3.2 3.2 2.2 2.2
Macrinite 1.6 1.6 0.0 0.0 0.0 0.0
Secretinite 0.6 0.6 0.2 0.2 0.0 0.0
Funginite 0.0 0.0 t t 0.0 0.0
Inertodetrinite 0.2 0.2 0.4 0.4 0.2 0.2
Total inertinite 17.4 17.5 9.6 9.6 7.0 7.1
Sporinite 10.8 10.9 13.0 13.0 13.2 13.3
Cutinite 0.6 0.6 1.2 1.2 1.0 1.0
Resinite 0.2 0.2 0.6 0.6 1.0 1.0
Alginite 0.0 0.0 0.0 0.0 0.0 0.0
Liptodetrinite 0.0 0.0 0.2 0.2 0.0 0.0
Suberinite 0.0 0.0 0.0 0.0 0.0 0.0
Exsudatinite 0.0 0.0 0.0 0.0 0.0 0.0
Total liptinite 11.6 11.7 15.0 15.0 15.2 15.4
Silicate 0.2 0.0 0.2
Sulfide 0.2 t 0.6
Carbonate 0.2 0.0 0.0
Other 0.0 0.0 0.2
Total mineral 0.6 0.0 1.0
Rmax 1.01 0.92 0.90
St dev 0.03 0.04 0.04
Rrandom 0.92 0.88 0.85
St dev 0.04 0.04 0.05
v8 0 18 44
v9 32 80 56
v10 68 2 0

particles (ash particles). The two phases were separated by
decanting the hydrophobic (agglomerated, ‘‘cleaned” coal) phase
from the water containing hydrophilic particles, through a 200
mesh (75 lm) screen. After decanting, the hydrophobic and
hydrophilic products were separately filtered, dried and weighed.
Initial trials, with coal ground to an average particle diameter of
11.0 lm, by a batch rotary ball mill, suggested that little, if any,
reduction in ash content occurred as a result of the pentane
agglomeration procedure. The results of compositional analyses
of the three starting coals pre- and post-pentane agglomeration
are shown in Table 3. Minor reductions in ash content of the Logan
Hollow Blue Gem and Jellico coals were observed post pentane
agglomeration with a batch rotary ball mill. The Logan Hollow Blue
Gem was reduced from 1.02 to 0.68 wt.% ash with a 2-h grind.
However, the ash content increased to 0.87 wt.% after a 4-h grind.
The Jellico coal ash content was reduced from 2.31 to 1.98 wt.%
after a 2-h grind, which further reduced to 1.81 wt.% with a 4-h
grind. The Harps Maple Creek Blue Gem actually was observed to
increase in ash after cleaning, from 1.42 to 1.40 to 1.65 wt.% ash,
as-received, 2- and 4-h grinds, respectively. None of the pentane
agglomerated, or ‘cleaned’ versions of the native low-ash coals
achieved an ash content of less than the targeted 0.1 wt.%. Only
slight reductions in ash, if any, were observed.
These results suggested that the ash particles were too small to
be liberated from the coal matrix during coal crushing in the rotary
ball mill, and that an attritor or stirred ball mill grinding method
was needed to reduce the coal size even further. With this method,
the average particle diameter was reduced to 9.2 lm after 1 h of
grinding. Attritor grinding longer than 1 h was observed to offer
diminishing returns in terms of particle diameter reduction. How-
ever, the results of pentane agglomeration from attritor-milled
Blue Gem coal showed, in all cases, that the process actually
increased the level of ash in the coal quite significantly as shown
in Table 3. Native ash concentrations increased by approximately
1.5–2.0 wt.% when the attritor mill grinding technique was uti-
lized. This was most likely caused by contamination from the small
steel balls used to mill the coal.
The results of the coal cleaning experiments suggested that the
use of pentane agglomeration to reduce the ash content of coal
offered diminishing returns for coals where the native ash content
is significantly less than 5 wt.% and, in particular, for low-ash coals
with ash contents in the 1–2% range. Further, significant energy
would need to be expended to mill the coal to particle sizes
<10 lm, where ash reduction would have the highest likelihood
of success. Thus, pentane agglomeration as a coal cleaning method
for ash reduction was abandoned. Work continued with the native
 J.D. Craddock et al. / Fuel 187 (2017) 229–241 233

Table 3
Analyses of the cleaned coals.

Coal analysis Proximate & ultimate analysis

Coal ID Logan hollow blue gem
Preparation As received Cleaned
Coal crushing Ball mill (2 h) Ball mill (4 h) Attritor mill
Sample no 72,257 72,258 72,454
Proximate analysis
Ash % 1.02 0.68 0.87 3.14
Moisture % 1.70 0.77 2.09 0.13
Volatile matter % 38.7 39.3 41.7 42.92
Fixed carbon % 58.6 59.2 55.4 53.81
Total % 100.0 100.0 100.0 100.0
Ultimate analysis
Carbon % 81.0 80.5 80.9 81.0
Hydrogen % 5.80 5.85 5.85 5.77
Nitrogen % 1.86 1.71 1.77 1.92
Sulfur % 0.77 0.70 0.70 0.70
Oxygen % 9.52 10.56 9.91 7.45
Total % 99.0 99.3 99.1 96.9

Coal analysis Proximate & ultimate analysis

Coal ID Harps maple Creek blue gem
Preparation As received Cleaned
Coal crushing Ball mill (2 h) Ball mill (4 h) Attritor mill Attritor mill
Sample no 72,302 72,304 72,455 72,456
Proximate analysis
Ash % 1.42 1.40 1.65 3.11 2.96
Moisture % 2.93 2.65 2.19 0.04 0.15
Volatile matter % 37.1 37.4 37.0 38.56 42.74
Fixed carbon % 58.6 58.6 58.8 58.29 54.15
Total % 100.0 100.0 99.7 100.0 100.0
Ultimate analysis
Carbon % 80.5 81.0 79.8 80.4 80.7
Hydrogen % 5.69 5.78 5.70 5.59 5.66
Nitrogen % 1.89 2.05 2.02 1.97 1.97
Sulfur % 1.08 0.75 0.75 0.71 0.66
Oxygen % 9.13 9.03 10.13 8.26 8.09
Total % 98.3 98.6 98.4 96.9 97.0

Coal analysis Proximate & ultimate analysis

Coal ID Jellico
Preparation As received Cleaned
Coal crushing Ball mill (2 h) Ball mill (4 h)
Sample no 72,305/06 72,307
Proximate analysis
Ash % 2.31 1.98 1.81
Moisture % 1.51 2.16 1.78
Volatile matter % 35.8 37.3 36.7
Fixed carbon % 60.4 58.6 59.8
Total % 100.0 100.0 100.0
Ultimate analysis
Carbon % 82.0 81.4 82.2
Hydrogen % 5.60 5.73 5.72
Nitrogen % 1.85 2.03 2.02
Sulfur % 0.99 0.74 0.71
Oxygen % 7.33 8.12 7.51
Total % 97.7 98.0 98.2

low-ash coals to determine whether coals with ash contents in the
1–2 wt.% range could be used to produce anode grade coke, but
without reducing the mineral matter and/or insoluble fraction
from the coal by either coal cleaning or by a digest filtration step.
Overall, the objective was to see if extremely low-ash coal could
produce a good anode coke, not to drive up the cost by subjecting
an already expensive coal to costly micron-scale cleaning.
2.3. Coal conversion optimization
The three southeastern Kentucky low-ash coal samples were
assessed for solvent extraction and coking. Solvent extraction tests
were conducted in micro-reactors to determine optimum process
conditions (solvent type, temperature, and time) aimed at maxi-
mizing conversion of the coal to soluble product. The presence of
undissolved material, (mineral matter and insoluble organic mat-
ter, IOM) in the digest can inhibit the growth and coalescence of
mesophase during coking that is crucial to the formation of long-
range anisotropic structure required to produce high quality anode
cokes. In the omission of an uneconomic filtration step, it was
therefore imperative to minimize the amount of mineral matter
in the feed coal and maximize the conversion of the coal to soluble
material prior to the commencement of the coking phase. Addi-
tionally, a low mineral matter content of the coke is also necessary
234 J.D. Craddock et al. / Fuel 187 (2017) 229–241
Fig. 1. Micro-reactor used to assess coal extraction parameters.
to minimize anode oxidation rate and contamination of the
aluminum during production as a result of metals present in the
carbon anodes. A simple and effective test was used to determine
the process conditions required for optimized conversion of the
coals to soluble products. A set of micro-reactors were constructed
consisting of 200-mm lengths of 12.7-mm diameter stainless steel
tubing (1.25-mm wall thickness) closed at one end with a cap and
at the other with a valve via a reducing union (Fig. 1).
In the test procedure, the reactor was loaded with a known
amount of coal and solvent, pressure tested, sealed, and immersed
in a heated fluidized sand bath controlled at the required temper-
ature. The reactor was mounted in a cradle driven in an oscillating
mode at 120 cycles/min to provide agitation for the coal/solvent
slurry. At the completion of the test, the reactor was quenched in
water and allowed to dry and equilibrate at ambient temperature.
The gas evolved during the reaction was determined by weighing
the reactor before and after venting the gas through the valve.
The reaction product (coal digest) was recovered and transferred
to a beaker. The amount of undissolved solids in the coal digest
was determined from the filter cake yield following filtration of
the reaction products through a GFA glass fiber filter paper (parti-
cle size retention of 1 lm). To aide in filtration, the viscosity of the
coal digest was reduced by dilution with quinoline, which is an
excellent solvent for the dissolution products from coal extraction
and does not normally result in precipitation of any of the dis-
solved coal. However, quinoline has a high boiling point (238 °C)
and cannot be easily removed from the cake by drying. The filter
cake was initially washed with fresh quinoline, which was then
replaced with a lighter miscible solvent, tetrahydrofuran (THF, bp
65 °C). The cake was then vacuum dried to remove the THF. The
dry filter cake contained mineral matter and IOM. Calculation of
the IOM from the weight of dry cake and the known ash content
of the coal allows determination of the amount of coal converted
to soluble products on a dry ash-free (DAF) basis. Tests were per-
formed in duplicate, and sometimes with multiple replicates to
gain confidence in the tests and for assessment of repeatability.
All tests were conducted at a solvent/coal ratio of 2:1, confined
gas-tight in the micro-reactor for residence times of 10, 30, and
60 min at reaction temperatures of 400 and 420 °C. Two process
solvents were used; anthracene oil (Reilly Industries) and decant
oil (Marathon Oil Co. from their Ashland refinery in Catlettsburg,
KY). Decant oil, the heaviest fraction from a fluid catalytic cracker,
was considered to be the most likely candidate to make an accept-
able solvent for the extraction of coal. It is a relatively low-value
refinery bottoms stream that is a common feed for the production
of pet coke and carbon black. Anthracene oil is produced from
overheads (condensed volatiles) of metallurgical coke production.
As seen in Table 4, both solvents contain low sulfur (<1 wt.%),
advantageous for the production of anode cokes.
2.3.1. Optimization of the coal digestion processes
Determination of the optimum coal digestion conditions, which
resulted in the highest coal conversions to a soluble form under the
mildest and shortest times, was critical. A matrix of tests was con-
ducted on the three low-ash coals using the two solvents in the
micro-reactors (2 parts solvent to 1 part coal powder, <200 mesh
(US and Tyler standard mesh) or <74 lm opening). Two process
temperatures and three process times were systematically
investigated. Replicate tests were performed at each condition to
gain statistically significant results, shown in Table 5. The test con-
ditions in bold for each coal/solvent type were those that yielded
the highest DAF conversion. 60-min digestion times were unam-
biguously determined to offer no significant increase in coal con-
version relative to shorter digestion times.
2.4. Production of coke from selected coal samples
Upon determination of the optimum digestion conditions, lar-
ger scale tests to prepare coal digests for assessment as feedstocks
for anode grade coke production were conducted using a 2-l stirred
Parr autoclave (300 g coal: 600 g solvent). After charging and purg-
ing with nitrogen, the reactor was rapidly brought to the operating
temperature to commence the digestion stage for the specified res-
idence time. The pressure was maintained at 1.4 MPa (200 psi) by
venting gas and light distillate, evolved through a manual control
valve, into condensate traps and gas collection bags. At completion,
the contents of the reactor were drained, sampled for analysis and
the viscous liquid coal digest set aside for use as a feedstock for
coking trials.
Reproducing the conditions found in a working, industrial-scale
delayed coker on a small scale is not trivial without a significant
investment in hardware. In an industrial system, the feed material
is commonly pumped through a pre-heater coil, raising the tem-
perature to above 475 °C, and injected at pressure obliquely into
the heated coking vessel held at a temperature of around 550 °C
and at a known pressure. While most of the low boiling fraction/
solvent flashes off from the feed and thermal cracking yields more
light material, growth and coalescence of mesophase proceeds by
polymerization reactions in the viscous turbulent melt. In the lab-
oratory, smaller batches of the coal digest were transferred to a
steel coking vessel (Fig. 2) for conversion to coke.
Turbulence and shear were promoted by sparging with nitrogen
to induce coalescence of the mesophase. The reactor was held
under these conditions at 380 °C for 1.5, 2.5, and 5.0 h before the
temperature was raised to 500 °C to complete the coking process.
The slow heating regime used, whereby the coal solution was held
at 380 °C for an extended period (up to 5.0 h) was introduced to
encourage mesophase growth and coalescence. Although far
removed from the conditions found in an authentic delayed coker,
it was considered that the coke produced by this method could be
characterized and its structure compared to a typical coker feed
treated in the same way. Low boiling material, generated by the
thermal cracking of the feed carried from the reactor, was collected
in cold traps downstream. The solvent fraction recovered from the
coker would normally be recycled to the front end of the process in
an industrial application. An additional test was conducted under
the same conditions but using neat decant oil as the feedstock in
order to make comparisons. Materials balances are discussed in
Table 6.
The green cokes recovered from the reactor were calcined to
remove residual volatile material by heating at 50 °C/min to
1300 °C with a 1 h soak time in a helium purge in a high-
temperature furnace. There was no provision to collect volatiles
during calcining. The structure and properties of the calcined cokes
were characterized by measurement of the proximate and ultimate
analyses and by determination of the metals content by ICP.
 J.D. Craddock et al. / Fuel 187 (2017) 229–241 235

Table 4
Solvent analysis.

Solvent Source Ultimate analysis (%)

C H O (by diff) N S
Anthracene oil Reilly industries 90.7 6.31 2.5 0.23 0.25
FCC Decant oil Marathon 91.0 8.22 0.5 0.08 0.16

Table 5
Micro-reactor coal – solvent digestion results.

Coke precursor Digest temp (°C) Digest time (min) DAF conversion (%) Std. dev. (%) N
Logan hollow blue gem + decant oil 400 30 87.1 2.2 8
400 10 80.7 7.4 8
420 30 74.9 7.7 8
420 10 84.5 5.5 8
Logan hollow blue gem + anthracene oil 400 30 94.9 0.9 8
400 10 79.1 13.1 8
420 30 87.7 2.9 8
420 10 82.4 9.8 8
Harps maple creek blue gem + decant oil 400 30 75.7 6.5 8
400 10 75.8 9.1 8
420 30 80.5 6.0 8
420 10 71.4 7.9 8
Harps maple creek blue gem + anthracene oil 400 30 95.4 4.1 8
400 10 93.4 4.7 8
420 30 79.3 10.8 8
420 10 82.8 10.3 8
Jelico + decant oil 400a 30 88.3 2.9 8
400 10 – – –
420 60 82.7 2.9 2
420 10 – – –
Jellico + anthracene oil 400 30 86.8 2.6 10
400 10 80.7 11.4 8
420 30 78.2 8.2 2
420 10 83.4 2.9 2

The optimized conditions for digestion of the coals into a soluble form are highlighted in bold.
a
Optimized per results in anthracene oil micro-reactor runs. 60-min digestion times were unambiguously determined to offer no significant increase in coal conversion
relative to shorter digestion times. All conversions are reported on a dry, ash-free basis (DAF).

3. Results and discussion

3.1. Coal cleaning

Coal cleaning by pentane agglomeration with both rotary and

attritor ball milling processes was found to be ineffective. There-
fore, the native low-ash coals (Logan Hollow Blue Gem, Harps
Maple Creek Blue Gem, and Jellico) containing 1.02, 1.42, and
2.31 wt.% ash, respectively, were used as received. It may be useful
to utilize oil agglomeration on a lower cost (higher ash content)
starting coal.

3.2. Micro-reactor tests

The results of the micro-reactor tests to determine the optimum

Fig. 2. Laboratory coking vessel.
Further, samples were ground to a fine powder for measurement of
crystalline properties by X-ray diffraction (an internal, Si-powder
standard was added to the ground cokes for XRD peak position
analysis). Samples were also set in epoxy resin, sectioned, and pre-
pared to a final 0.05-lm polish to observe petrographically the
coke structure by polarized reflected-light oil-immersion optical
microscopy.
digestion conditions – rendering the highest possible amount of
the coal into a digest form – are shown in Table 5. Logan Hollow
Blue Gem coal was digested in decant oil to 87.1 ± 2.2% (on a
DAF basis) after digestion for 30 min at 400 °C. In anthracene oil,
94.9 ± 0.9% DAF was digested, also after 30 min at 400 °C. Harps
Maple Creek Blue Gem coal was digested in decant oil to
80.5 ± 6.0% DAF after 30 min at 420 °C. In anthracene oil,
95.4 ± 4.1% DAF was solubilized after 30 min at 400 °C. Lastly, the
Jellico sample was digested in decant oil to 88.3 ± 2.9% DAF after
30 min at 400 °C. In anthracene oil, 86.8 ± 2.6% DAF was digested
after 30 min at 400 °C. In all cases, longer and shorter digestion
times, 60 and 10 min respectively, were found to offer no signifi-
236 J.D. Craddock et al. / Fuel 187 (2017) 229–241

Table 6
Larger scale digestion, coking and calcining results.

Coke precursor Digest number Coking Time at Calcining Digest Coke Calcin ing Overall coke
(daf coal conv.) number 380 °C (h) number yield (%) yield (%) yield (%) yield (%)
Logan hollow blue gem + decant oil DEDI 00 DEDI 01 1.5 GF 342 89.33 33.83 88.89 26.86
(85.1%) DEDI 02 2.5 GF 340 43.41 90.86 35.24
DEDI 04 5.0 GF 343 35.00 89.90 28.11
Harps maple creek blue gem DEDI 05 DEDI 06 1.5 GF 344 87.75 42.00 89.22 32.88
+ decant oil (63.6%) DEDI 07 2.5 GF 345 41.87 91.01 33.44
DEDI 08 5.0 GF 346 45.27 90.85 36.09
Jelico + decant oil DEDI 09 DEDI 10 1.5 GF 347 88.09 42.00 90.91 33.63
(78.2%) DEDI 11 2.5 GF 348 43.28 90.04 34.33
DEDI 12 5.0 GF 349 38.79 87.70 29.97
Neat decant oil DEDI 13 DEDI 14 1.5 GF 353 94.73 13.51 92.09 11.78
DEDI 15 2.5 GF 354 14.01 91.93 12.20
DEDI 16 5.0 GF 355 14.22 83.88 11.30
Neat decant oil (no digestion) N/A DEDI 03 2.5 GF 341 100 37.00 91.76 33.95a
Logan hollow blue gem DEDI 17 (86.8%) DEDI 18 2.5 GF 356 91.93 31.90 89.08 26.12
+ anthracene oil
a
No digest was done, so the yield assumes 100% digest yield.

cant benefit. The anthracene oil was a more effective solvent for
the Blue Gem coals. However, insufficient anthracene oil supplies
hampered the possibility to support a matrix of larger scale con-
versions from which anode cokes could be systematically analyzed.
Therefore, the optimum conditions, listed above in decant oil, were
used to make larger scale digests from which green and calcined
cokes were derived.
3.3. Larger scale digest, coking, and calcining
A 2-l reactor was used to prepare larger samples of coal digest
for the coking tests. The conditions selected, discussed in Sec-
tion 3.2, were based upon the results from the screening tests with
similar results. Harvested coal-solvent digests were then subjected
to green coking conditions, using the equipment shown in Fig. 2,
under the conditions outlined in Table 6. The digest yields, across
the board, were in the 90–95% yield range. Neat decant oil yielded
approximately 95%, and all of the coal digests yielded approxi-
mately 90%. Similarly, the calcining yield, across the board, was
approximately 90%. However, the coke yields after the green cok-
ing procedures varied more significantly; only 14% of the neat
decant oil digest was converted to green coke, with the majority
of the material escaping with the nitrogen sparge as volatiles.
However, undigested neat decant oil yielded 37% coke; therefore,
it is possible that the digestion process, even in the absence of coal
for 30 min at 400 °C, rendered the decant oil less efficient for cok-
ing. For the decant oil – coal digests, coking for 2.5 h at 380 °C
offered increased yield compared to 1.5 and 5.0 h at 380 °C. Here
coke yields for the Logan Hollow Blue Gem, Harps Maple Creek
Blue Gem, and Jellico coal digests in decant oil were as high as
43.41%, 41.87%, and 43.28% respectively, while the coke yield for
similarly processed neat decant oil was only 14.01%. Therefore,
overall calcined coke yields for the coal digests were approxi-
mately 35%, and that of similarly processed neat decant oil only
12%. A ca. 2 increase in coke yield was observed for the coal
digests in decant oil when compared to the control, neat decant
oil. Typical calcined coke yield in an industrial delayed coker is
20–25% (values acquired from internal communication with petro-
leum coke manufacturer). However, the structures of these cal-
cined cokes were significantly different (see Section 3.4.3). With
the remaining anthracene oil, one larger scale coke was produced
from the preferred coal digest conditions (Logan Hollow Blue
Gem, 30 min at 400 °C; see Table 5). Relative to decant oil, a lower
coke yield was observed, 31.90% as opposed to 43.41%, which
resulted in a lower overall coke yield of 26.12%. However, this cal-
cined coke showed a strikingly different microstructure than those
derived from digests in decant oil.
3.4. Characterization and analysis
The coal and coke samples were characterized by chemical, pet-
rographic, and XRD techniques, and their properties compared to
typical petroleum-derived anode cokes currently used by the alu-
minum industry.
3.4.1. Calcined coke analyses
Similar to the starting coals, the calcined cokes were composi-
tionally analyzed for proximate (ash, moisture, volatiles, and fixed
carbon by difference), ultimate (C, H, N, S, and O by difference), and
elemental analysis by ICP, and compared to a commercial pet coke
used for anodes (Table 7). For the Logan Hollow Blue Gem – decant
oil-derived samples, the concentration of ash in the calcined coke
was similar to that of the starting coal. However, surprisingly,
the ash concentration of the neat decant oil-derived calcined coke
was also approximately 1.6%, but the ash content of the neat
decant oil (without any processing prior to analysis) was found
to be 0.32%. Perhaps some concentrating of the ash in the calcined
coke occurred as a result of the digest and coking processes. Similar
levels of ash were observed in the Harps Maple Creek Blue Gem –
Decant oil derived calcined cokes, as in the Blue Gem calcined
cokes. Higher levels of ash were observed in the 3037 Jellico –
Decant oil derived calcined cokes, which is consistent with the ini-
tial coal analysis (Table 2). Relative to the commercial petcoke, all
calcined cokes (including coke derived from the undigested neat
decant oil) had unacceptably high metals contents. Coal digest-
derived cokes showed particularly high Fe (>1300 ppm), Ca
(>700 ppm), and Si (>1000 ppm), but they had low sulfur, approx-
imately 0.6 wt.%, relative to the commercial pet coke at 2.76 wt.%.
Further, the cokes all had low V and Ni content.
3.4.2. XRD analysis
All of the 1300 °C calcined cokes were subjected to X-ray
diffraction scanning of the (0 0 2) reflection. The peak position of
each (0 0 2) reflection was corrected with an internal Si-powder
standard. The d002 spacings are shown in Fig. 3. All of the spacings
were quite similar to the commercial petcoke, and no clear trend in
spacing was observed as a function of the dwell time at 380 °C dur-
ing green coking. The largest difference was observed between the
 J.D. Craddock et al. / Fuel 187 (2017) 229–241 237

Table 7
Calcined coke analyses.

Parent coal Logan hollow blue gem Harps maple creek blue gem 3037 Jellico Decant oil Commercial petcoke Anode grade
Sample no GF340 GF344 GF347 GF341 Specs
MA# DEDI 02 DEDI 06 DEDI 10 DEDI 03
Preparation As-received As-received As-received As-received As-received
Prox. analysis
Ash % 1.53 1.66 2.63 1.62 0.25 60.1
Moisture % 0.04 0.02 0.03 0.07 0.09
Volatile matter % 16.59 0.63 0.62 16.69 16.08
Fixed carbon % 81.84 97.69 96.72 81.62 83.58
Total % 99.7 100.0 100.0 99.6 100.3
Ultimate analysis
Carbon % 98.0 97.6 96.7 98.14 96.49
Hydrogen % 0.29 0.51 0.45 0.29 0.36
Nitrogen % 0.82 0.88 0.86 0.01 0.76
Sulfur % 0.63 0.63 0.58 0.49 2.76 61.0
Oxygen % 0.10 0.01 0.14 0.1 0.1
Total % 99.8 99.6 98.8 99.0 100.5
Coke elemental composition
Si ppm 1233 847 831 1369 14 200
Fe ppm 1350 2271 2119 339 32 300
Ca ppm 730 1560 759 166 22 50
Na ppm 258 332 339 190 31 100
P ppm 9 23 98 9 1 5
V ppm 14 9 6 5 14 200
Ni ppm 19 33 85 12 17 200

commercial petcoke and the Logan Hollow Blue Gem – Anthracene
oil-derived coke, with the latter showing a significantly larger d002,
indicating that it was less graphitic.
Similarly, Fig. 4 shows the crystallite thicknesses of the calcined
cokes as determined by the Scherrer relation, from the (0 0 2)
reflections. No clear trends were noted between the varyingly
coked materials. In all cases, the commercial petcoke had the lar-
gest crystallite thickness (Lc). The smallest Lc was observed for
the Logan Hollow Blue Gem – Anthracene oil-derived coke, which
is consistent with the (0 0 2) spacing results. All of the decant oil –
coal digest-derived cokes showed Lc values intermediate between
the commercial petcoke and the Logan Hollow Blue Gem – Anthra-
cene oil-derived coke.
3.4.3. Reflected polarized light optical microscopy
All calcined cokes were evaluated by reflected, polarized light
microscopy to verify and investigate anisotropic texture. Shown
in Fig. 5 are examples of the structures, including those from a

Calcined Coke XRD - (002) Spacing

0.356 Anthracene Oil - LGBG

0.354

0.352
Neat Decant Oil Decant Oil - LHBG Decant Oil - HMCBG Decant Oil - Jellico
0.350
d(002) (nm)

0.348

0.346

0.344

0.342

0.340

0.338
ke

9
3

6
34

35

34

34

34

34
35

35

34

34

34

34

34

35
Co

GF

GF

GF

GF

GF

GF
GF

GF

GF

GF

GF

GF

GF

GF
et

3-

6-

7-

0-

1-

2-
4-

5-

1-

2-

4-

6-

7-

8-
lP

-0

-1

-0

-1

-1

-1
-1

-1

-0

-0

-0

-0

-0

-1
cia

DI

DI

DI

DI

DI

DI
DI

DI

DI

DI

DI

DI

DI

DI
er

DE

DE

DE

DE

DE

DE
DE

DE

DE

DE

DE

DE

DE

DE
m
m
co

Fig. 3. d002 spacings of the (0 0 2) reflections of the calcined cokes. (Refer to Table 6 for process conditions of the DEDI numbers.)
238 J.D. Craddock et al. / Fuel 187 (2017) 229–241

Calcined Coke XRD - (002) Crystallite Thickness Lc

4.00

3.50 Anthracene Oil - LGBG

Neat Decant Oil Decant Oil - LHBG Decant Oil - HMCBG Decant Oil - Jellico
3.00

2.50
Lc (nm)

2.00

1.50

1.00

0.50

0.00
ke

9
3

6
34

35

34

34

34

34
35

35

34

34

34

34

34

35
Co

GF

GF

GF

GF

GF

GF
GF

GF

GF

GF

GF

GF

GF

GF
et

3-

6-

7-

0-

1-

2-
4-

5-

1-

2-

4-

6-

7-

8-
lP

-0

-1

-0

-1

-1

-1
-1

-1

-0

-0

-0

-0

-0

-1
cia

DI

DI

DI

DI

DI

DI
DI

DI

DI

DI

DI

DI

DI

DI
er

DE

DE

DE

DE

DE

DE
DE

DE

DE

DE

DE

DE

DE

DE
m
m
Co

Fig. 4. Crystallite thickness, Lc of the (0 0 2) reflection of the calcined cokes. (Refer to Table 6 for process conditions of the DEDI numbers.)

commercially relevant petcoke. The color changes from magenta to
blue-white are indicative of anisotropic orientation in the micro-
scopic crystallite domains of graphitic carbon.
Anisotropy is vitally important for thermo-mechanical and elec-
trical properties of the coke, which are essential for anode applica-
tions. Also visible are the long-range domains of the anisotropic
layered structures.
Fig. 6 shows optical images of the coal digest-derived calcined
cokes. Left to right, shows the effect of increasing the soak time
at 380 °C during the green coking process, 1.5, 2.5, and 5.0 h
respectively. Top to bottom, each line represents a different coke
precursor. The top shows the Logan Hollow Blue Gem coal – Decant
oil digest-derived calcined cokes; the Harps Maple Creek Blue Gem
coal – Decant oil-derived calcined cokes are in the second row; and
the Jellico – Decant oil-derived calcined cokes are shown on the
third row. The fourth row shows the digested, neat decant oil
derived calcined cokes. The single image at the bottom is the cal-
cined coke derived from Logan Hollow Blue Gem – Anthracene
oil digest (soaked at 380 °C for 2.5 h during the green coking pro-
cess). While the Jellico coal does have a slightly higher rank than
the two Blue Gem samples, the vitrinite in all three coals is ther-
moplastic. Fig. 7 further illustrates the behavior of the coals versus
the decant oil and the anthracene. As seen in Fig. 7A and B, the coal
melt is immiscible in the decant oil, as indicated by the texture of
the inertinite-bearing coke versus the texture of the surrounding
coke. The two coke phases are presumably derived from the coal
and the decant oil, respectively. In contrast, the coal + anthracene
melt was miscible, as shown by the uniform-texture, inertinite-
bearing coke textures (Fig. 7C and D).
The decant oil – coal digest-derived calcined cokes show evi-
dence of two phases; one composed of very fine, short-range abun-
dant anisotropic domains, and the other of long-range isotropic
domains. This is clear when compared to the neat decant oil
derived calcined cokes (GF 353–355), which contain only a single
phase of long-range, layered anisotropic domains. This long-
range structure is preferred. Increasing green coking time at
380 °C may slightly increase the divide between the observed
two phases. The micron-scale immiscibility of the coal digests in
decant oil may be leading to the formation of two phases, one
which is decant oil rich, and another which is coal digest rich. This

Commercial DED1-03
Pet Coke GF 341 GF325
Fig. 5. Polarized reflected light microscopy of non coal derived cokes - The images show the desirable, long-range anisotropic texture of (left to right) a commercial calcined
petcoke, calcined coke derived from neat decant oil (GF 341), and a calcined coke derived from a synthetic mesophase pitch (GF 325), which is known to produce a highly
anisotropic graphitic carbon.
 J.D. Craddock et al. / Fuel 187 (2017) 229–241 239

DEDl-01 DEDl-02 DEDl-04

GF 342 GF 340 GF 343

DEDl-06 DEDl-07 DEDl-08

GF 344 GF 345 GF 346

DEDl-10 DEDl-11 DEDl-12

GF 347 GF 348 GF 349

DEDl-14 DEDl-15 DEDl-16

GF 353 GF 354 GF 355

DEDl-17
GF 356
Fig. 6. Polarized reflected light microscopy of the coal digest-derived cokes.

coal digest-rich phase would be more likely to contain small ash
inclusions, which could account for the short-range anisotropic
order observed in one of the phases. Whereas in the other phase,
likely the decant oil-rich phase, long-range anisotropic order is still
observed. Lastly, only a single phase of very short-range anisotro-
pic domains was observed in the Anthracene oil – Logan Hollow
Blue Gem coal digest-derived calcined coke (GF 356). Here,
complete miscibility of the coal digest with the anthracene oil
occurred. Anthracene oil, being more aromatic than the decant
oil, would seemingly be a better solvent for the coals. This is
consistent with the observation of a single phase. However, in all
cases, the introduction of the coal to the digest resulted in
short-range anisotropic order. We believe this to stem from the
ash content native in the coals (1–2 wt.%), and that omission of
the hot filtration step to remove the insoluble matter and ash led
to the formation of undesirable short- range anisotropic order in
the calcined cokes.
3.5. Cost comparison
The prospect of utilizing a low-cost coal as a feedstock for a
value-added anode coke, was a driver in this effort – particularly
if the need for a hot filtration step of the coal digest could be omit-
ted. Using approximate current (1 July 2016) prices, the cost of a
Central Appalachian steam coal would be on the order of $41/t.
The cost of Blue Gem coal, a very low-ash, high-grade coal used
for metallurgical applications, would be on the order of $150/t.
High quality calcined petroleum needle coke, for use in graphite
electrodes, may be on the order of up to $2000/t, accounting for
a >10-fold increase in value.
In reality, needle coke is typically produced from high quality
decant oil, which itself can cost on the order of $500/t. Further,
high-grade needle cokes are typically produced on-site at a petro-
leum refinery, where the refiners make their own decant oil from
crude oil. Anode cokes are not as valuable as high-grade needle
240 J.D. Craddock et al. / Fuel 187 (2017) 229–241

Fig. 7. Distinct differences are noted in the behavior of coal in decant oil versus anthracene. In A and B (GF 342 and GF 343, respectively), the finer-domain Logan Hollow Blue
Gem coal-derived coke with included inertinite fragments (i) is surrounded by a coarser coke derived from the decant oil. In this case, the melted coal and decant oil were
immiscible, as seen in the distinct zones in the coke. In contrast, C and D, both from the coking of anthracene with the Logan Hollow Blue Gem coal (calcining number 356),
show that this melt was miscible, as indicated by the inertinite (i) fragments in the otherwise uniform fine-domain coke.

cokes, and cost in the neighborhood of $400 –$600/t. However,
anode cokes are typically produced from vacuum residuals (not
decant oil), which are somewhat less costly. But again, typically
crude refineries are their own suppliers of vacuum resids from
which anode coke is produced. In general practice, anode (and nee-
dle) cokes are produced from within refineries, which in turn sup-
ply themselves with their own precursor feedstock – at the lowest
cost and quality as needed. This precludes a simple estimation of
the potential cost impact of using a coal digest to make anode coke.
From a coal digest point of view, omitting the hot filtration step is a
cost-saving endeavor. However, the results of this work indicate
that omission of this step does not offer a feasible path towards
quality anode coke. It should be noted that utilization of 2:1
solvent to coal resulted in an approximate 2 increase in coke
yields. In other words, utilization of coal digests may be a possible
path to cost- effective ‘cutting’ of valuable precursor feedstocks
including decant and/or anthracene oil. Further, utilization of coal
digests may offer a way of reducing sulfur in pet coke – assuming
that low-S coal digests can offset high-S precursors. The cokes pro-
duced in this study were fairly low in S compared to the commer-
cial petcoke. In this regard, and likely with development of an
efficient hot filtration step to remove ash, allowing for long-
range anisotropy to develop, coal digests could find clear value in
the coke industry.
4. Conclusions
4.1. Summary
In summary, this work showed specific methodology for opti-
mized digestion of low-ash (1–2 wt.%) Eastern Kentucky coals in
decant and anthracene oils. Attempts were made to utilize oil
agglomeration as a method to further reduce the ash in the precur-
sor coals to less than 0.1 wt.%, which did not work well. Work pro-
ceeded utilizing the as-received low-ash coal. Using these
methodologies, larger scale digests were made and coked under
conditions of increasing time at 380 °C with N2 sparge, followed
by a final green coking at 500 °C. All green cokes were calcined
under a He purge to 1300 °C for 1 h. A set of control cokes from
neat decant oil was also prepared for purposes of comparison.
One calcined coke stemming from a digest in anthracene oil was
also made for comparison. All the coals, solvents and product cokes
 J.D. Craddock et al. / Fuel 187 (2017) 229–241
were compositionally analyzed. The desired long-range anisotropic
order of the layered crystallite domains was clearly observed in
cokes without coal digest. Cokes with coal digest in decant oil con-
tained two phases; one with short-range anisotropic domains and
the other with long-range anisotropic domains. Consistent with
Kimber’s findings [19], it is hypothesized that the short-range ani-
sotropy stemmed from ash particles concentrated in the final coke
and that complete miscibility between the decant oil and coal
digest did not occur, which led to the two phases. Stadelhofer
et al. reported that coal-tar pitch, low in quinoline insolubles,
was a useful starting material for needle cokes [20]. Much of
Marsh’s work reported that the presence of inert/insoluble/non-gra
phitizable components prevents the formation of desirable long-
range anisotropy in the final coke [21–23]. The decant oil-rich
phase produced the desired long-range anisotropic domains.
Utilization of the anthracene oil as a digest solvent did produce a
single phase, but it was wholly composed of short-range anisotro-
pic domains. The omission of a hot filtration of the coal digests to
remove the ash content is, therefore, not recommended for the
maximization of long-range anisotropic order. X-ray diffraction
analysis showed little difference in the d002 spacings and Lc crystal-
lite thicknesses between the neat decant oil and coal digest-
derived calcined cokes. The Lc of the commercial pet coke was
clearly larger than that of the anthracene oil - coal digest-derived
coke. Little difference in the final cokes was observed between
increasing time at 380 °C, under nitrogen sparge, from 1.5 to 2.5–
5.0 h. However, utilization of a coal digest allowed for an overall
coke yield as high as 35%, compared to the 12% of the neat decant
oil. The low-sulfur content of the low-ash coals and digest solvent
did produce a low-S calcined coke, also low in Ni and V, but high in
mineral ash metals (Na, Ca, Fe). Coal digests may find value in the
coke industry as a low cost filler material, which may also serve to
lower the S content in the final coke.
The central findings were that (1) the low-ash coal could be
converted into a digest in either decant oil or anthracene oil, with
very high yields (low concentrations of insoluble matter); (2) these
coal digests could be subjected to coking conditions, and converted
into a green coke, which could be subsequently calcined; but (3)
the inclusion of the coal digest, independent of the parent coal
ash concentration, resulted in the formation of very short-range,
abundant anisotropic domains while simultaneously mitigating
the formation of the desired long-range needle anisotropic
domains. These observations suggested that the ash content served
to prevent long-range anisotropic domain formation. More specif-
ically, the ash particles themselves, within the coal matrix, were
very fine (<10 lm) and abundant, which likely produced the
observed short-range microstructures, rather than long-range
microstructures. In this respect, it was found that even ‘low-ash’
native coals (1–2 wt.%) were simply too high in ash to feasibly form
anode grade coke without the costly filtration step. Although it
may be possible to reduce the ash content of the coal to less than
0.1 wt.%, thus rendering the coal more likely to produce quality
241
coke without a digest filtration step, the simple and energy-
efficient oil agglomeration method tested in this work was found
ineffective.
References
[1] Baker JJ, Rolle JG, Lee JM, Llerena R. Characterization of green and calcined coke
properties used for aluminum anode-grade carbon. Abstr Pap Am Chem Soc
1998;215:U609–10.
[2] Jones SS. Anode-carbon usage in the aluminum-industry. Acs Sym Ser
1986;303:234–50.
[3] AFPM. Petroleum coke overview; August 2016. <http://education.afpm.org/
refining/petroleum-coke/>.
[4] Radenovic A. Properties of carbon anode components for aluminum
production. Nafta 2012;63:111–4.
[5] Andrews RJ, Rantell T, Jacques D, Hower JC, Gardner JS, Amick M. Mild coal
extraction for the production of anode coke from Blue Gem coal. Fuel
2010;89:2640–7.
[6] Edwards L, Backhouse N, Darmstadt H, Dion M-J. Evolution of anode grade coke
quality. TMS Light Metals 2012:1207–12.
[7] Edwards L. The history and future challenges of calcined petroleum coke
production and use in aluminum smelting. Jom 2015;67:308–21.
[8] Rimmer SM, Hower JC, Moore TA, Esterle JS, Walton RL, Helfrich CT.
Petrography and palynology of the Blue Gem coal bed, southeastern
Kentucky USA. Int J Coal Geol 2000;42:159–84.
[9] Kimber GM, Whitehead JC, Cain M, Mukerjee A, Hilderbrandt RD. Production of
reduction grade coke from low ash buller coal. In: Carbon conference,
Bordeaux, France, 1984.
[10] Clarke JW, Rantell TD. Solids separation from the products of the solvent-
extraction of coal. Powder Technol 1984;40:223–34.
[11] Shi KS, Wang CF, Lin JF. Influence of microstructure on thermal-expansion and
graphitization of oil cokes. Fuel 1985;64:174–7.
[12] Hower JC, Rimmer SM, Bland AE. Geochemistry of the blue gem coal bed, Knox
County, Kentucky. Int J Coal Geol 1991;18:211–31.
[13] Hower JC, Taulbee DN, Rimmer SM, Morrell LG. Petrographic and geochemical
anatomy of lithotypes from the blue gem coal bed, Southeastern Kentucky.
Energy Fuel 1994;8:719–28.
[14] Rimmer SM, Hower JC, Moore TA, Esterle JS, Walton RL, Helfrich CT.
Petrography and palynology of the blue gem coal bed (Middle
Pennsylvanian), Southeastern Kentucky, USA. Int J Coal Geol 2000;42:159–84.
[15] Yoshida T, Takanohashi T, Sakanishi K, Saito I. Relationship between thermal
extraction yield and softening temperature for coals. Energy Fuels
2002;16:1006–7.
[16] Masaki K, Yoshida T, Li C, Takanohashi T, Saito I. The effects of pretreatment
and the addition of polar compounds on the production of ‘‘HyperCoal” from
subbituminous coals. Energy Fuels 2004;18:995–1000.
[17] Griffith JM, Clifford CEB, Rudnick LR, Schobert HH. Solvent extraction of
bituminous coals using light cycle oil: characterization of diaromatic products
in liquids. Energy Fuels 2009;23:4553–61.
[18] Mehrotra VP, Sastry KVS, Morey BW. Review of oil agglomeration techniques
for processing of fine coals. Int J Miner Process 1983;11:175–201.
[19] Kimber GM. Graphitising carbons from coal via solvent extraction: effects of
processing conditions on product quality. In: Proceedings of the international
conference on carbon, 1977. p. 316–7.
[20] Stadelhofer JW, Marrett R, Gemmeke W. The manufacture of high-value
carbon from coal-tar pitch. Fuel 1981;9:877–82.
[21] Marsh H. Carbonization and liquid-crystal (mesophase) development: Part 1.
The significance of the mesophase during carbonization of coking coals. Fuel
1973;52(3):205–12.
[22] Marsh H, Foster JM, Hermon G, Iley M. Carbonization and liquid-crystal
(mesophase) development. Part 2. Co-carbonization of aromatic and organic
dye compounds, and influence of inerts. Fuel 1973;52(4):234–42.
[23] Marsh H, Latham CS. The chemistry of mesophase formation. In: American
chemical society symposium series, n 1, 3, Washington, DC, 1986.