Preparation and process optimization of porous carbons using different carbon sources and activating agents are frequently and commonly reported in open literature. However, only scanty references are made on utilization of petroleum coke for conversion to high surface area porous carbon using KOH as the activating agent. Hence, the present work attempts a process optimization exercise to prepare high surface area porous carbon from Petroleum coke using chemical activation (KOH) utilizing design of experiments. The effect of activation temperature, petroleum coke to KOH ratio (KPR) and activation duration were assessed on the surface area and yield of the porous carbon. The process optimization was performed covering experimental parameters in the range of 500?C - 800?C, 2 - 5 and 30 - 120 min. The optimal process conditions for maximizing the yield and BET surface area was identified to be an activation temperature of 639?C, KPR of 4.5 and activation duration of 43 min, having BET surface area 1765 m 2/g and yield of 89.8%. However, an attempt to maximize only the BET surface area, ignoring yield has resulted with a porous carbon with maximum surface area of 2061 m 2/g, with the optimal process conditions being an activation temperature of 688?C, KPR of 3.8 and activation duration of 74 min, with the corresponding yield of only 77%. The characterization of porous carbon was performed using nitrogen adsorption isotherm, FT-IR and SEM analysis.
Porous carbons are well known and widely used adsorbent in industries for variety of separation applications. Even as compared with the new generation adsorbents, porous carbons still are the most preferred industrial adsorbent due to cost, availability and its suitability for different applications. A mere comparison of the volumes of different adsorbents being commercially manufactured would highlight its importance. Generally, the porous carbon manufacturing methods are either based on physical activation which are basically gasification reactions of carbon with steam/CO2/combination of both or based on the chemical activation methods which include dehydrating agents such as phosphoric acid, sodium hydroxide, potassium hydroxide, zinc chloride etc., the precursors for preparation of porous carbons could be either coal, lingocellulosic material or Biomass based materials. However, characteristics of porous carbon highly depend on its precursor characteristics and method of activation [
Petroleum coke is the most abundant byproduct of oil refining with over 100 million tons global production reported in the year 2010. Efforts have been directed towards petroleum coke utilization in the fields of combustion and gasification to generate electric power or produce syngas [
Petroleum coke with surface area of 11.7 m2/g converted into porous carbon utilizing KOH as the activated agent having a surface area and pore volume of 692 m2/g, 0.264 cc/g respectively [
However, as per author’s knowledge process optimization of porous carbon from petroleum coke was not reported in open literature. Considering the fact there are a number of reactions that occur simultaneously during the process of activation, results based limited random experimental runs could vary vastly as compared to product generated at optimal experimental conditions. The experimental attempts reported in literature predominantly report single point experiments or assess effects of parameter varying one parameter at a time. Hence the present work attempts to assess the effect of the process variables such as activation temperature, KOH/Petroleum coke ratio (KPR), activation duration and to optimize the process conditions for maximizing the porous carbon yield and the BET surface area.
In this paper, petroleum coke received from Reliance Industries, India was used as the precursor. The petroleum coke was washed with deionized hot water to remove the foreign materials and dried in oven at 105˚C for 24 h. The dry materials were stored in a in a desiccator for further experiment. The KOH was purchased from sigma, USA. Double distilled water was used for all the washing purposes in the process. 35% Aqueous HCl was purchased from Sigma, USA for neutralization of unreacted KOH in washing process. Litmus paper from Sigma, USA was used to test pH of filtrate.
The BET surface area, along with the pore size distribution was estimated using the standard nitrogen adsorption isotherm obtained by an Autosorb 1-C adsorption apparatus (Quanta Chrome Instruments, USA). Prior to analysis, the samples were first dried in an oven at 110˚C overnight and were quickly placed in the sample tube. The tube was then heated to 170˚C and was evacuated for 4 h until the pressure was less than 10−4 Torr. The BET surface area was calculated from the isotherms using multipoint Brunauer-Emmett-Teller (BET) method [
FT-IR spectra of various samples were recorded on a Nicolet 740 FT-IR spectrometer at ambient conditions using KBr as the diluent to identify the functional groups in the activated carbon. The samples were loaded into the sample holder and scanned in the mid IR region viz., 100 to 4000 cm−1 to generate the spectrum. A pellet made of nearly the same amount of KBr was used as the background. FEG-250 SEM instrument (FEI, Holland) was employed at an accumulation voltage of 30 KV with 2.5 K magnification to estimate the surface pore structure of the porous carbon. The carbon, hydrogen, nitrogen, and sulfur (CHNS) content of the adsorbent samples have been evaluated employing ELEMETAR-VARIO-EL analyzer.
1 g of petroleum coke was taken as the precursor and mixed with KOH powder at an impregnation ratio (KPR) of 2 - 5. The mixture was activated in a horizontal furnace at a temperature of 500˚C - 800˚C for an activation time of 30 - 120 min, in an inert atmosphere (Nitrogen flow of 150 mL/min). Upon completion of the each experiment, the carbonized samples were cooled to the room temperature in an inert atmosphere and washed repetitively with aqueous hydrochloric acid (HCl) to remove ash present in the materials, followed by washing with distilled water several times until the pH of the filtrate was reduced to neutral. The samples were dried in oven at 105˚C for 12 h to ensure complete dryness. The yield of porous carbon was estimated based on grams of bone-dry porous carbon prepared to the grams of bone-dry petroleum coke utilized for activation.
Response Surface Methodology (RSM) is a statistical and mathematical technique that relates experimental data to regression models, which is utilized to assess the effect of parameters and interaction among them in addition to optimizing the process conditions [
Factor | Low Level (−1) | Center Point (0) | High Level (+1) |
---|---|---|---|
Temperature (X1) | 500˚C | 650˚C | 800˚C |
KPR (X2) | 2 | 3.5 | 5 |
Activation time (X3) | 30 | 75 | 120 |
The BBM utilizes an empirical second-order polynomial model relating the dependent and independent variables as expressed below,
Y = β 0 + β ∑ i = 1 n β i X i + ∑ i = 1 n β i j X i X j + ∑ i = 1 n β i i X i 2 (1)
where Y is the predicted response variable, β 0 is the constant, β i , β i j , and β i i are the linear, interaction and quadratic parameters respectively. The experimental data are analyzed using statistical software MINITAB-15 for the regression analyses and analysis of variance (ANOVA). The process optimization was performed using optimizer tool to maximize the individual objective functions (Y1, and Y2).
The complete design matrixes along with the responses from the experimental results are presented in
According to the sequential model sum of squares, were selected based on the highest order polynomials where the additional terms were significant and the models were not aliased. The final empirical equations after eliminating the insignificant parameters, relating to the dependent variables in the form of uncoded units to the % Yield (Y1) and BET surface area (Y2) are given by Equations (2) and (3):
Y 1 = − 171.6 + 0.56 X 1 + 35.38 X 2 + 0.66 X 3 − 0.001 X 3 2 − 0.032 X 1 X 2 − 0.18 X 2 X 3 (2)
Y 2 = − 14921.9 + 39.1 X 1 + 1419.1 X 2 + 22.7 X 3 − 187.7 X 2 2 − 0.2 X 3 2 (3)
Positive sign in front of the terms indicates synergistic effect, whereas negative sign indicates antagonistic effect. The quality of the model developed was evaluated based on the correlation coefficient value (R2). The R2 for % Yield is 0.99, while that for BET surface area is 0.97 validating the appropriateness of the model. The closer R2 value to unity, the better is the models ability in predicting the response.
The coefficients of the models are listed in
Run | X1 (˚C) | X2 | X3 (min) | Y1 (% Yield) | Y2 (BET Surface area) |
---|---|---|---|---|---|
1 | 800 | 3.5 | 30 | 67.17 | 1312 |
2 | 500 | 2.0 | 75 | 68.37 | 317 |
3 | 650 | 3.5 | 75 | 77.95 | 2004 |
4 | 800 | 3.5 | 120 | 51.74 | 1285 |
5 | 800 | 5.0 | 75 | 58.18 | 1423 |
6 | 500 | 5.0 | 75 | 86.34 | 733 |
7 | 500 | 3.5 | 30 | 82.73 | 826 |
8 | 650 | 3.5 | 75 | 78.16 | 2010 |
9 | 650 | 5.0 | 30 | 97.01 | 1445 |
10 | 500 | 3.5 | 120 | 67.16 | 796 |
11 | 650 | 3.5 | 75 | 78.52 | 1995 |
12 | 650 | 2.0 | 30 | 68.69 | 1120 |
13 | 650 | 5.0 | 120 | 60.08 | 1458 |
14 | 650 | 2.0 | 120 | 80.75 | 1060 |
15 | 800 | 2.0 | 75 | 68.60 | 1297 |
Term | Coefficient | T | P |
---|---|---|---|
Constant | −166.22 | −12.37 | 0.000 |
X1 | 0.56 | 16.12 | 0.000 |
X2 | 32.90 | 13.31 | 0.000 |
X3 | 0.65 | 8.82 | 0.000 |
X12 | 0 | −15.11 | 0.000 |
X22 | 0.35 | 1.39 | 0.222 |
X32 | −0.001 | −4.16 | 0.009 |
X1X2 | −0.032 | −12.93 | 0.000 |
X1X3 | 0 | 0.064 | 0.952 |
X2X3 | −0.181 | −22.31 | 0.000 |
R-Sq = 99.5%, R-Sq (adj) = 99.2%.
Term | Coefficient | T | P |
---|---|---|---|
Constant | −15,578.6 | −11.20 | 0.000 |
X1 | 40.2 | 11.18 | 0.000 |
X2 | 1608.3 | 6.28 | 0.002 |
X3 | 21.7 | 2.83 | 0.037 |
X12 | 0 | −10.79 | 0.000 |
X22 | −187.7 | −7.14 | 0.001 |
X32 | -0.2 | −5.24 | 0.003 |
X1X2 | −0.3 | −1.28 | 0.258 |
X1X3 | 0 | 0.013 | 0.990 |
X2X3 | 0.3 | 0.321 | 0.761 |
R-Sq = 97.4%, R-Sq (adj) = 95.5%.
and BET surface respectively. Statistically lower the p value or higher the F value more significant is the model parameters.
The validity of model in addition to R2 is based on the ANOVA, which is a statistical technique that subdivides the total variation in a set of data into component parts associated with specific sources of variation for the purpose of testing hypotheses on the parameters of the model.
Source | Degree of freedom (DF) | Sum of squares (SS) | Mean squares (MS) | F | P |
---|---|---|---|---|---|
Model | 9 | 1949.9 | 216.7 | 179.73 | 0.000 |
Linear | 3 | 852.9 | 156.9 | 130.14 | 0.000 |
Square | 3 | 295.6 | 98.5 | 81.72 | 0.005 |
Interaction | 3 | 801.5 | 267.2 | 221.63 | 0.007 |
Error | 5 | 6.0 | 1.2 | - | - |
Lack of fit | 3 | 5.9 | 1.9 | 23.51 | 0.041 |
Pure error | 2 | 0.2 | 0.08 | - | - |
Total | 14 | 1955.9 |
Source | Degree of freedom (DF) | Sum of squares (SS) | Mean squares (MS) | F | P |
---|---|---|---|---|---|
Model | 9 | 3,324,403 | 369,378 | 28.59 | 0.001 |
Linear | 3 | 1,075,883 | 609,328 | 47.16 | 0.000 |
Square | 3 | 2,226,160 | 742,053 | 57.43 | 0.000 |
Interaction | 3 | 22,359 | 7453 | 0.58 | 0.655 |
Error | 5 | 64,600 | 12,920 | - | - |
Lack of fit | 3 | 64,486 | 21,495 | 377.11 | 0.003 |
Pure error | 2 | 114 | 57 | - | - |
Total | 14 | 3,389,003 |
indicate the appropriateness and validity of the model. Among the model parameters, all the interaction parameters were found to be insignificant. From the ANOVA results it can be concluded that the model predictions using Equations (2) and (3) are satisfactory and that the model can be utilized to identify the optimum process conditions.
In the activation process a significant amount of K2CO3 and H2 were produced in comparison to the small amount of CO2 during the activation, which implied that most of the carbon consumption was due to the reaction of K2O with CO2 to form K2CO3 [
2KOH → K2O + H2O (Dehydration) (4)
C + H2O → H2+CO (Water-gas Reaction) (5)
CO + H2O → H2 + CO2 (water-gas shift reaction) (6)
K2O + CO2 → K2CO3 (Carbonate Formation) (7)
At higher activation temperatures, the reduction of K2O by Carbon and Hydrogen takes place to produce Metal Potassium:
K2O + C → CO + 2K (Reduction by Carbon) (8)
K2O + H2 → 2K + H2O (reduction by hydrogen) (9)
An overall reaction of the above reactions, which showed to be thermodynamically stable according to Gibbs energy calculations [
6KOH + C → 2K + 3H2 + 2K2CO3 (10)
The metallic K becomes mobile at activation temperatures and can intercalate into the carbon matrix, which causes a widening of the carbon layers, which results in pore formation [
To summarize how activation by KOH takes place, three main methods take place [
1) Chemical activation which involves redox reactions between Potassium compounds and the carbon material which results in the formation of a porous structure.
2) Physical activation by the formation of CO2 and H2O which further contributed to the pore structure development by carbon gasification.
3) Carbon lattice expansion by metallic Potassium, which has the capability of intercalating into the carbon lattice resulting in expansion of the lattice. Once the Potassium and its compounds were washed out, they left behind pores, which increased the porosity.
The effect of activation temperature, KPR, and activation time on the % Yield of porous carbon (Y1) is presented in this section.
KPR while
Nevertheless, from
The highest yield corresponds to the minimum to medium of all three variables covered in the present work. Similar results were reported in preparation of porous carbon from coconut husk [
BET surface area is an important parameter which determines the adsorption capacity of porous carbon. In general higher the BET surface area, higher will be the adsorption capacity provided the pore size are suitable to the adsorbing molecule. The effect of activation temperature, activation time and KPR ratio on BET Surface area (Y2) is presented in this section. All parameters show significant effect towards BET surface area as shown in
An increase in the activation temperature as well as the KPR ratio contributes to
an increase in the BET surface area while it contributes to the decrease in yield of porous carbon. However, in the interest of the commercial manufacturing process, it is desirable identify the optimum condition that maximizes the yield as well as the BET surface area. Therefore, in order to compromise between these two responses, the function of desirability was applied using Design of Experiment Software (Minitab-16, USA). The optimum process conditions are estimated to be an activation temperature of 639˚C, KPR of 4.5 and activation time 43 min, with the resultant porous carbon BET surface area 1765 m2/g and yield of 89.8%. In order to ensure the acceptability of the identified optimized conditions, results of the experimental repeat runs conducted at the optimized conditions are reported in
X1 | X2 | X3 | % Yield | % Error | BET Surface Area | % Error | ||
---|---|---|---|---|---|---|---|---|
Predicted | Experimental | Predicted | Experimental | |||||
639.4 | 4.5 | 42.7 | 90.5 | 89.8 | 0.8 | 1739 | 1765 | 1.5 |
exercise. It also confirms that the RSM approach is most suitable for optimization of process conditions for preparation of porous carbon.
The present work clearly confirms that all the process parameters are effective on BET surface area and % Yield of porous carbon. In general optimum process conditions vary depending on the type of precursor [
The structural heterogeneity of porous carbon is very important for all adsorption processes.
The BET surface area of the porous carbon is estimated as 2010 m2/g, with average pore volume of 0.85 cc/g, and average pore diameter of 1.42 nm, with the proportion of micropore volume being 0.76 cc/g with 89.4% in total pore size distribution as shown in
In order to further investigate their surface chemistries, FT-IR spectra of porous carbon is presented in
Sample | BET surface area (m2/g) | VT (cc/g) | Micropore volume % | Mesopore volume % | Average pore diameter (nm) |
---|---|---|---|---|---|
Petroleum coke carbon | 2010 | 0.85 | 89.4 | 8.2 | 1.42 |
olefin functional groups and the one at 1150 cm−1 to the stretching vibration of aromatic ring (methoxy-O-CH3) [
The SEM image of porous carbon is shown in
Petroleum coke having low surface area (11 m2/g) and pore volume (0.012 cc/g), was utilized to convert it into porous carbon with KOH activation using process optimization. An increase in the activation temperature, KPR and activation duration contributed to an increase in the BET surface area while the yield of porous carbon decreased. An increase in the activation duration was found to influence the BET surface area while its effect on the yield was found to be marginal. The optimum conditions have been identified to be an activation temperature of 639˚C, KPR ratio of 4.5 and duration of 43 min, with the corresponding
BET surface area and yield being 1765 m2/g and 89.8%. On the other hand, if the objective is restricted to maximize the BET surface area, an activation temperature of 688˚C, KPR of 3.8 and activation duration of 74 min, results in a porous carbon with 2061 m2/g having an yield of 77%. CHNS analysis of petroleum coke to porous carbon shows a good improvement in oxygen content from 6.2 to 28.2 and a drastic decrease in sulfur content from 5.2 to 0.2, and improved physico-chemical properties.
The Authors wish to acknowledge the financial support from the petroleum Institute for giving an opportunity to work on Porous carbon development from petroleum coke (RIFP-15318/2015).
Srinivasakannan, C., El-Mootassem, M., Reddy, K.S.K. and Al Shoaibi, A. (2018) Synthesis and Characterization of Porous Carbon from Petroleum Pitch Using KOH. Journal of Materials Science and Chemical Engineering, 6, 53-69. https://doi.org/10.4236/msce.2018.67007