Erin Partlan, Kathleen Davis, Yiran Ren, Onur Guven Apul, O. Thompson Mefford, Tanju Karanfil, David A. Ladner
Published in the Journal Water Research
Citation: Erin Partlan, Kathleen Davis, Yiran Ren, Onur Guven Apul, O. Thompson Mefford, Tanju Karanfil, David A. Ladner (2016). “Effect of bead milling on chemical and physical characteristics of activated carbons pulverized to superfine sizes.” Water Research 89, 169-170. https://doi.org/10.1016/j.watres.2015.11.041
Abstract
Superfine powdered activated carbon (S-PAC) is an adsorbent material with particle size between roughly 0.1–1 μm. This is about an order of magnitude smaller than conventional powdered activated carbon (PAC), typically 10–50 μm. S-PAC has been shown to outperform PAC for adsorption of various drinking water contaminants. However, variation in S-PAC production methods and limited material characterization in prior studies lead to questions of how S-PAC characteristics deviate from that of its parent PAC. In this study, a wet mill filled with 0.3–0.5 mm yttrium-stabilized zirconium oxide grinding beads was used to produce S-PAC from seven commercially available activated carbons of various source materials, including two coal types, coconut shell, and wood. Particle sizes were varied by changing the milling time, keeping mill power, batch volume, and recirculation rate constant. As expected, mean particle size decreased with longer milling. A lignite coal-based carbon had the smallest mean particle diameter at 169 nm, while the wood-based carbon had the largest at 440 nm. The wood and coconut-shell based carbons had the highest resistance to milling. Specific surface area and pore volume distributions were generally unchanged with increased milling time. Changes in the point of zero charge (pHPZC) and oxygen content of the milled carbons were found to correlate with an increasing specific external surface area. However, the isoelectric point (pHIEP), which measures only external surfaces, was unchanged with milling and also much lower in value than pHPZC. It is likely that the outer surface is easily oxidized while internal surfaces remain largely unchanged, which results in a lower average pH as measured by pHPZC.
1. Introduction
Activated carbon adsorption is a best available technology for the removal of small molecular weight organic contaminants from water (Clark and Lykins, 1990, Crittenden et al., 2005a, Osantowski and Wullschleger, 1986). Powdered activated carbon (PAC) is a common form of activated carbon, produced by grinding larger granular activated carbon particles (GAC) (Neely and Isacoff, 1982, Snoeyink and Chen, 1985). PAC is defined as particles passing through an 80-mesh sieve (177 μm) (ASTM Standard D5158-98, 2013) but typical particle sizes are between 10 and 50 μm. A recent development in activated carbon technology is to further reduce the size of PAC. Superfine powdered activated carbon (S-PAC) is activated carbon that has been reduced to sizes near or less than one micrometer.
S-PAC can improve adsorption kinetics for small molecules and circumvents competition with larger molecules, such as natural organic matter (NOM) (Ellerie et al., 2013, Heijman et al., 2009a, Matsui et al., 2013, Matsui et al., 2009). The mechanism for faster adsorption lies in shortened diffusion-limited adsorption pathways, which allows faster transport to terminal adsorption sites for small molecules (Ando et al., 2010, Ellerie et al., 2013, Matsui et al., 2008, Pelekani and Snoeyink, 2000). Terminal sites for small molecules lie in micropores where sorption occurs due to increased sorption strength for pore diameters less than twice an adsorbate diameter (Li et al., 2002). Conversely, adsorption of large molecules is only governed by external surface area and not by the characteristics of inner pores. In the shortened pathways of S-PAC, the time of transport for small molecules is decreased, but the rate of large molecule adsorption to outer surfaces and pores remains the same. The increase in the rate of large molecule adsorption due to increased specific external surface area is much less than the increase in small molecule adsorption due to the shortened diffusion pathway. As a result, particle size is the main driving force behind improved kinetics. However, changes in other physical and chemical properties can also impact adsorption performance. For example, it is possible for pulverization to shift the distribution of pore sizes by exposing smaller diameter pores, effectively becoming larger diameter pores (Dunn and Knappe, 2013, Ellerie et al., 2013, Matsui et al., 2014). An in-depth examination of carbon characteristics on adsorption performance revealed that carbon properties, namely hydrophilicity in the form of oxygen content, had strong impacts on the penetration depth of adsorbates (Matsui et al., 2015).
Since S-PAC is not commercially available, there is no standard production method. Individual groups have produced their own materials or outsourced their production on a case-by-case basis, so material production parameters may vary widely. These parameters have been rarely reported as there is no available guidance regarding parameters to be controlled and recorded. S-PAC researchers describe production through wet milling using phrases such as wet grinding in a bead mill, pulverization in a ball mill, wet mill micro-grinding and micro-grinding in a wet bead mill (Ando et al., 2010, Ellerie et al., 2013, Heijman et al., 2009a, Matsui et al., 2007). Wet milling generally encompasses high velocity contact between grinding media and the product material—activated carbon in this case—while suspended in a carrier fluid. Grinding media can be made from a variety of materials and in a variety of sizes; none of the previous studies specifically describe the grinding media used. Additionally, the carrier fluid and additives such as surfactants are not described, though many are available for enhancing grinding performance. The specific outcome of grinding depends on these factors since it is the result of energy transfer from the rotation of the mill through the grinding media and the carrier fluid.
Among carbons, differences in abrasion resistance, also termed hardness, affect the rate of particle breakdown. Abrasion is a surrogate measure of hardness since harder carbons have a reduced tendency to degrade upon impact than softer carbons (Toles et al., 2000, Yalcin and Arol, 2002). Activated carbons are relatively soft; commercial coal and wood based carbons were found to have a Mohs hardness between 2 and 3 on the 1–10 Mohs hardness scale (Patni et al., 2008). Carbon hardness depends on the carbon precursor. Coconut shell and other nut shells are very hard precursors (Heschel and Klose, 1995, Mohd Din et al., 2009). Of coal carbons, bituminous coal-based carbons are harder than lignite-based carbons (Greenbank and Spotts, 1993). Wood-based carbons have a much wider range of hardness, from softer than lignite coal-based carbons to as hard as coconut-shell based carbons, depending on the type of wood (Hernández et al., 2014).
Wet milling is designed to physically alter the carbon but may also cause chemical changes, which is important for activated carbon applications since chemical characteristics play large roles in adsorption processes (Karanfil and Kilduff, 1999, Karanfil et al., 1999, Li et al., 2002, Quinlivan et al., 2005). Surface functional groups affect the type of molecules that are preferentially adsorbed (Apul et al., 2015, Karanfil and Kilduff, 1999, Pendleton et al., 1997). Carbon modification has been explored for enhanced or targeted contaminant removal (Akmil Başar et al., 2003, Dastgheib et al., 2004, Liu et al., 2009). For example, oxidation of carbon results in increased adsorption of metal cations but decreased adsorption of low molecular weight organic compounds (Biniak et al., 1999, Jia and Thomas, 2000, Karanfil and Kilduff, 1999, Tessmer et al., 1997). Activated carbon typically has oxygen content around 3–4% by mass; those modified to increase oxygen-containing functional groups can have contents around 10% (Apul et al., 2013, Li et al., 2002).
Due to the small sample size of S-PACs in previous studies combined with limited characterization data, trends in material changes as a result of PAC size reduction cannot be determined. Most recently, Matsui et al. (2015) measured a large number of parameters regarding milled carbons; however, the study measured properties of only one particle size and did not examine changes as a result of milling. Hence, the main objective of this study was to systematically examine a number of carbon types under varying degrees of milling in order to elucidate specific effects of wet milling on the chemical and physical characteristics of activated carbons.
2. Materials and methods
2.1. Activated carbon
Activated carbon materials were obtained from several commercial sources and chosen to cover a range of characteristics. Seven commercial carbons were used with variation in specific surface area, pore volume, and pHPZC: Watercarb-800 (bituminous coal, Standard Purification), Filtrasorb 400 (bituminous coal, Calgon), PAC 20 B (bituminous coal, Norit), Hydrodarco 3000 (lignite coal, Norit), Hydrodarco B (lignite coal, Norit), Aqua Nuchar (wood, Mead Westvaco), and Aquacarb 1230C (coconut shell, Siemens). The carbons are labeled as BC1, BC2, BC3, LC1, LC2, WD, and CS, respectively (Table 1). BC1, BC3, LC2, and WD were received as PAC, and they were sufficiently small to pass the 200 μm mesh screen within the bead mill. LC1, BC2, and CS were received as GAC, and they were processed using a coffee grinder and sieved through a #100 screen (150 μm openings) to isolate the smaller particles for milling.
2.2. Bead mill
PAC was pulverized to S-PAC using a bead mill (MiniCer, Netzsch Premier Technologies, Exton, PA, USA) containing 0.3–0.5 mm yttrium-stabilized zirconium oxide ceramic beads. The mill was operated at 85% loading capacity; 120 mL of beads were measured with a graduated cylinder and added to the 140 mL mill. Thus, the design mill void was 20 mL, but the total mill void was 50 mL after accounting for void space when measuring the beads. The 120 mL loading volume corresponds to approximately 2 million beads. PAC was added to the mill as a slurry comprised of 24 g of dry carbon in 300 mL of distilled and deionized water (DDI) with 18 MΩ resistivity, yielding a percent solids concentration of approximately 6–7%. The slurry was kept chilled at 10 °C during milling.
It is common in milling applications to use a dispersant to aid the milling process. Without dispersants the breakup of particles results in an increase in solution viscosity; dispersants help maintain fluidity and decrease pumping energy. One dispersant type (Reax, Mead Westvaco) was tested to evaluate its effects on carbon milling; however, all of the samples used for further characterization were milled with no dispersant addition.
All S-PAC samples were produced using the same flow rate, mill loading, and mill rotational speed; thus, the rate of energy transfer to the carbon particles was kept constant through all tests. Total energy applied was varied by changing the total milling time, which varied from one pass through the mill to six hours. In the one-pass scenario carbon was fed into the mill and then collected as product in the output. This was repeated once more for the two-pass scenario. Each pass took approximately 45 s. In all other millings, the carbon slurry was recirculated through the system by connecting the output line back to the feed tank.
2.3. Characterization methods
Milled and unmilled carbons were analyzed for physical and chemical parameters. Elemental analysis, measuring weight percent compositions of carbon, oxygen, hydrogen, and nitrogen, was performed using a Flash Elemental Analyzer 1112 series (Thermo Electron Corporation). Sizes for particles less than 6 μm in diameter were measured using dynamic light scattering (DLS) with a Zetasizer NanoZS (Malvern, Worcestershire, UK). Readings were taken in distilled water after bath sonication and Z-avg hydrodynamic diameters are reported. Particles larger than 6 μm were measured by optical microscopy imaging using a Zeiss Axioskop 2 Plus optical microscope with a Zeiss AxioCam MRc5 camera attachment running AxioVision AC version 4.2 software. Particles were sonicated in DDI before imaging and Zeiss Immersionsol 518C immersion oil was used to view the particles at 40× magnification. The images were processed using ImageJ software to determine the average Feret diameter of the particles and the particle size distribution. Particles were also visually observed with scanning electron microscopy (SEM, Hitachi SU6600) after sputter coating with platinum for 3-min.
Nitrogen gas adsorption was performed at 77 K with an ASAP 2020 analyzer (Micromeritics Instrument Corp. U.S.) and pore size distributions were determine using density functional theory (DFT) and calculated surface area from the Brunauer–Emmett–Teller (BET) equation. The DFT model allows for categorization of pore volumes into micropore (<2 nm), mesopore (2–50 nm), and macropore (>50 nm) fractions.
pHPZC in the bulk material was measured by a pH drift method where the point of zero charge is defined as the pH were no drift occurs after 48 h (Dastgheib et al., 2004, Lopez-Ramon et al., 1999). For each pH point and carbon, 100 mg of dry carbon were added to 20 mL of pH adjusted 0.1 M NaCl in a CO2-free atmosphere in a glove box. After a minimum of 48 h on a shaker table, pH was measured in each vial and compared to a no carbon blank. The isoelectric point (pHIEP) of the carbon was determined by measuring the zeta potential of S-PACs using the Zetasizer NanoZS (Malvern, Worcestershire, UK) and observing the pH that produced a zeta potential reading of zero. Samples were prepared by probe sonication in DDI and manually titrated for pH adjustment.
3. Results and discussion
3.1. Milling
Each pass through the mill corresponded with approximately 5 s of contact time through the 50 mL mill void. Thus, a milling time of 30 min corresponded to approximately 40 passes through the mill and a total milling time of 200 s, based on average residence time. However, due to short circuiting and dead zones within the mill feed tank, the true number of passes was likely higher for some particles and lower for others. Additionally, viscosity increased at different rates for each carbon, so the number of passes made by one carbon may differ slightly from the number made by a different carbon due to a changing flow rate. DDI was added as needed during milling to decrease viscosity and allow for complete mixing.
To examine the effect of dispersant on milling, WD was milled with varying dispersant concentrations. Dispersant mitigated viscosity increase, reducing the need to add DDI, and did not affect the particle size reduction. Notably, the coconut shell-based carbon, CS, did not noticeably increase in viscosity though no dispersant was added. As mentioned previously, none of the samples milled with dispersants were used for further characterization in this study.
3.2. Carbon particle size
PAC particle sizes were very similar among carbons, with median sizes ranging from 11 to 14 μm (Table 1). Milling times as brief as one pass through the mill resulted in particle sizes near or below one micrometer and subsequent milling further reduced median particle sizes but with diminishing returns (Fig. 1). Due to the different analytical measurements for determining particle size above and below 6 μm—DLS for smaller particles and visible-light microscope image analysis for larger ones—comparison of S-PAC size with parent PAC is indirect. The hydrodynamic diameter measurements from DLS are found using models that assume spherical shape. The spherical assumption is supported by SEM images, especially for particles milled for longer times, but particles milled for shorter times have more angularity. SEM also qualitatively supports Z-avg particle size measurements with observable declines in particle size as milling time increases. Images of PAC and all S-PACs from BC1 taken by SEM at 10,000 times magnification reveal the range of particle sizes from large PAC and 1 pass particles to small particles after 6 h that have a tendency to aggregate (Fig. 2). The smallest particles—after 1 h, 2 h, and 6 hrs—were also imaged at 20,000 times magnification where individual particles are more clearly identifiable (Fig. S1).
Fig. 1. Particle size of milled carbons with milling times varied from one pass through the mill to 6.3 h. Data are split into two panes to avoid excessive overlap.
Fig. 2. Scanning electron microscopy images (10K magnification) of all forms of BC1, including (A) PAC and all S-PACs. Particles visibly decrease in size as milling increases from (B) 1 pass to (C) 15 min, (D) 30 min, (E) 1 h, (F) 2 h, and (G) 6 h.
The longest milling times resulted in the smallest particle sizes for each carbon. Most carbons reduced to median particle sizes of 200–400 nm; the largest size was 440 nm for WD after 6 h of milling and the smallest was LC2 at 170 nm after 4 h of milling (Fig. 1). In addition to producing the smallest S-PAC, LC2 also had the largest reduction in size in the shortest time; particles had a median size of 342 nm after 5 min and 270 nm after 20 min. The second lignite-based carbon, LC1, also exhibited relatively fast size reduction resulting in a median size of 390 nm after 15 min; however, it did not continue decreasing in size with further milling, with a size of 324 nm after 6 h. WD, BC2, and CS had little particle breakdown under short milling times, with median sizes above 700 nm and up to 1000 nm in the case of WD after one pass through the mill.
Overall, CS and WD were the slowest to decrease in size during milling, as seen by a low slope in its concentration over milling time (Fig. 3). Concentration was calculated from number-based particle size distributions measured in triplicate and conservation of mass; particle volumes calculated from the distribution bins were divided into the total initial mass to arrive at a total particle number. LC2 had the fastest breakdown (i.e. the fastest increase in particle concentration), followed by BC3, while BC1, BC2, and LC1 had similar rates. Rates of breakdown are interpreted as ease of milling—faster breakdown rates correspond to softer carbons that are more easily milled.
Fig. 3. Concentration of particles increases as particle size decreases. Concentrations were calculated from measured average particle sizes and conservation of mass.
The particle size results indicate that the same milling energy has different effects for different carbons. The three bituminous coals exhibited similar milling responses. The lignite coals acted similarly at first, but one continued to decrease in particle size while the other leveled off. Both WD and CS remained as large particles after initial milling, though CS responded to continued milling while WD did not. Thus, the rates of particle breakdown as well as the minimum achievable particle size were highly specific to each carbon and generalizations based on material are difficult to draw.
3.3. Carbon chemical properties
The majority of PACs had a basic pHPZC, though WD was neutral and LC1 had an acidic pHPZC. As milling time increased, the pHPZC of all carbons decreased and the oxygen content of all carbons increased (Table 2). (Carbon, hydrogen, and nitrogen were also measured, though no clear trends with milling time were noticed [Table S1]). LC1 and LC2 had the least observed pHPZC change, shifting down about half a pH point, while the other carbons dropped approximately two pH points over the longest milling times. BC1, BC2, and BC3 had similar PAC oxygen content around 2%, and increased to 4.87%, 6.45%, and 7.86%, respectively. WD had the highest initial oxygen content, 7.14%, as well as the highest final oxygen content, 10.50%. CS experienced the smallest increase in oxygen content from 3.38% to 5.38%. pHPZC values were inversely correlated with oxygen content (Fig. S2). CS, BC2, and BC3 correlated with an R2 above 0.95 while WD and BC1 had R2 values of 0.90 and 0.83, respectively. A decrease in pHPZC and an increase in oxygen content as a result of milling have been reported previously (Dunn and Knappe, 2013). The data shown here for several milling times provide convincing evidence that these changes are directly caused by milling.
Table 2. Surface charge properties measured by point of zero charge, isoelectric point, and percent oxygen content. The difference between pHPZC and pHIEP is shown as ΔpH. Carbons are distinguished by material and milling time. Dashed lines indicate that measurements were not taken.
pHIEP values could be expected to decrease as did the pHPZC values. Interestingly, pHIEP did not change appreciably with milling time; some variability was observed as milling progressed, but overall trends were nearly flat. As such, pHIEP did not correlate with oxygen increases. Additionally, the values were notably lower than pHPZC values, ranging from approximately 2.4 to 4.2 for the S-PAC of four carbons. The pHIEP of PAC could not be measured using the electrophoretic mobility technique since the PAC particles settled too quickly; it is unknown whether the pHIEP changed during the initial pass of PAC through the mill.
The difference in pHPZC and pHIEP is a result of the qualities measured by each technique. The pH drift method for pHPZC utilized in this study used a 48 h equilibration time, thus allowing for complete diffusion of solutes into and out of carbon pores. The pHPZC is thus a measure of acidity throughout the carbon material. pHIEP measurements are based on electrophoretic movement which is affected only by charges on the outer surface of the carbon; charged functional groups within the pores should have little or no effect on pHIEP. There are two possible mechanisms for a decreasing pHPZC and unchanging pHIEP: either chemical changes are only occurring at internal sites, which are measured by pHPZC but not pHIEP, or chemical changes are happening rapidly at external sites, wherein all S-PAC have the same external condition and the average pH value decreases as particle size decreases and total external surface area increases. The second scenario is consistent with previous work in activated carbon where it has been reported that incomplete oxidation results in preferential oxidation on the external surface, and can be indicated by measuring both pHIEP and pHPZC (Menendez et al., 1995). The low and unchanging pHIEP values imply that there are similar external surface charge densities of an acidic nature on all S-PAC, regardless of milling time. Correlation between pH shifts and specific external surface area, calculated from particle size measurements, reveal that the second mechanism is more likely; the difference between pHPZC and pHIEP is used as the correlating pH factor here and is termed ΔpH (Fig. 4). WD, CS, and BC2 had R2 values above 0.90 while BC1 had a poor correlation due to an outlier point. Oxygen increases also correlated strongly with specific external surface area, which is consistent with the idea that oxidation is happening primarily on the external surfaces (Fig. 5). The concept of an easily oxidized surface, once exposed to the abrasive milling environment, is sketched in Fig. 6 and shows (a) the initial surface oxygen groups (b) the breakup of particles to create new external surface area, and (c) the subsequent oxidation of those new surfaces.
Fig. 4. The difference between pH measured using the pHPZC method and the electrokinetic measurement of the isoelectric point is correlated with the S-PAC external surface area, which increases as particle size decreases.
Fig. 5. Oxygen increased with newly revealed surface area as a result of milling. External surface area was calculated from Z-avg particle size measurements.
Fig. 6. Illustration of PAC pulverization into S-PAC with proposed rapid surface oxidation.
Manifestation of oxygen on the outer surfaces occurs in multiple ways, such as through the formation of organic surface functional groups, or through inorganic oxidation. The variety of oxygen-containing surface functional groups that can occur at graphite edges in activated carbon have been documented (Biniak et al., 2007, Boehm, 1994). Based on the approximate mass of oxygen attached to S-PAC as a result of milling—2–5% by weight, or 0.13–0.31 mmol of oxygen per gram of carbon—the external surface concentration of oxygen-containing functional groups is on the order of 0.02 mmol/m2. In order to evaluate the possibility that some of the oxidation was inorganic, we compared ash content to oxidation. Ash content did not change appreciably as PAC was milled to S-PAC (Fig. S3). Percent oxygen increase between PAC and S-PAC correlated well with total ash content, and even better when only considering potentially reactive elements: Ca, Mg, Zn, Cu, Mn, Fe, and Al (Fig. S4). The high correlation is likely the result of thorough surface oxidation on elements evenly distributed throughout the carbon matrix, however, its contribution is limited: for a concentration of divalent elements of approximately 10 mmol per gram of carbon, it is likely that only 1% of those elements would be accessible on external surfaces for oxidation, or 0.01 mmol/m2.
3.4. Carbon physical properties
Total pore volume and distributions of volume to micropores, mesopores, and macropores shifted slightly as a result of milling (Table 3). BC1, BC2, BC3 and CS increased in the mesoporous fraction. BC2 increased in its macroporous fraction as well. WD had a decreased mesoporous fraction and an increased macroporous fraction. Primary and sub-microporous fractions—pores with diameters less than 1 nm (10 Å) and from 1 nm to 2 nm, respectively—were also examined due to their importance in adsorption of small molecules; however, there was little difference observed with respect to increased milling time (Stoeckli et al., 1993). Changes to pore volumes were not a distinct function of milling time, though it is reasonable that shifts in pore volume distributions following milling will trend towards increases in meso- and macroporous fractions as internal pores are opened. Most notable is the development of a mesoporous fraction in the longest milled S-PAC of CS where previously there was none. Typically—and CS PAC was no exception—coconut-shell based activated carbons are microporous upon activation (Crittenden et al., 2005b). Thus, despite a lack of direct correlation between milling and pore distribution changes to the carbons, changes were observed that may still have an impact on adsorption performance.
Surface areas were also generally unchanged, with the exception of WD, which decreased from over 1500 m2/g to 1000 m2/g. It is likely that WD had thinner pore walls than other carbons, as indicated by a high specific surface area, and thus had channels that were easily crushed during milling. In practice it may be possible to avoid crushing pore channels by optimizing milling conditions, including reduction of milling time, operation at lower mill power, or processing at a lower percent solids content.
3.5. Comparison with S-PAC in literature
A handful of publications over the past decade have explored milled carbon for use as an adsorbent; however, since most of these studies focus on adsorption performance, few have reported detailed physical and chemical carbon characteristics differences (Ando et al., 2010, Dunn and Knappe, 2013, Ellerie et al., 2013, Heijman et al., 2009b, Matsui et al., 2005, Matsui et al., 2004, Matsui et al., 2014, Matsui et al., 2013, Matsui et al., 2009, Matsui et al., 2008, Matsui et al., 2007). Available characterization data were extracted from those publications and is presented in Table 4. Here we consider all reported carbons with sizes under one micrometer. All studies reported particle size, which have been measured by various techniques, including dynamic light scattering (DLS), laser light scattering (LLS), laser diffraction, LED measurement, and optical and scanning electron microscopy. After particle size, surface area, as measured by nitrogen adsorption, was the next most common parameter reported. Since particle size and surface area are two characteristics often used to describe performance of PAC, reporting of these parameters is logical. Some papers have reported carbon characteristics not examined in this study, including uniformity coefficient and geometric standard deviation as related to particle size (Matsui et al., 2013), as well as acidic and basic functional groups as determined by Boehm titration (Matsui et al., 2015). The next most commonly reported parameters regard micropore fractions and mesopore fractions, though reporting is inconsistent; some studies report the volume of the pore fraction while others report the surface area of the pore fraction. The total pore volume is rarely reported. Lastly, pHPZC has only been reported by Ellerie et al. (2013) and Dunn and Knappe (2013). Notably, only Dunn and Knappe (2013) have reported oxygen content (not included in Table 4). It is clear that data gaps exist for an analysis of S-PAC beyond particle size and select physical characteristics.
Another data gap lies in the reporting of milling parameters. No paper has discussed the details of milling beyond a description of the grinding media and, in one case, the milling duration. It is clear from the number of variables present in wet milling, and the variety of carbons produced in this study, that processing details are important to reproducible data. For this reason, grinding is often viewed as an art, rather than a science. The most critical information to be reported relate to the transfer of energy: mill speed, milling duration, grinding media characteristics (size, composition, and loading rate), grinding carrier fluid plus dispersants used, and carbon loading rate (percent solids content). However, it is likely that all process variables play some role or another towards the final outcome, thus it is recommended that as many details be reported as possible, including such parameters as mill configuration, rate of flow, and temperature of milling.
The studies in Table 4 have found S-PAC to perform well, despite having few characterization parameters. It is understood that the improved rate of small molecular adsorption is governed by the path length, which is determined by particle size, a parameter that is reported in each study. However, the changes to chemical and physical characteristics observed in this study, and which likely have also occurred in carbons produced for previous studies, are known to have effects on adsorption. The increasingly negative surface charges observed could result in shifts in the adsorption affinity for hydrophobic compounds. Increases in the mesoporous and macroporous fractions would increase the adsorption capacity for lower molecular weight NOM. Lastly, optimization of S-PAC production is an important consideration. Between highly variable grinding procedures and the heterogeneity of activated carbon, an S-PAC with particular characteristics is produced. Additionally, extended milling yielded diminishing returns, thus there is a tradeoff between energy input and production of a useful S-PAC. Material choice is another major design consideration; for example, the lignite coals in this study reached small sizes very quickly, but the bituminous coals were more predictable in breakdown. An understanding of the grinding parameters and parent PAC that produce certain S-PAC types will result in the ability to fine tune the process for creating an S-PAC with certain desired characteristics.
4. Conclusions
The primary conclusions from this study are as follows:
- Pulverization of activated carbon by bead milling predictably reduces the mean particle size. The ability to reduce in size was generally related to the carbon precursor material; we observed faster size decline in the coal-derived carbons; however, we recognize that other wood and coconut shell carbons may behave differently than the samples used here. Successful pulverization in all carbon types supports the potential of S-PAC technology for application.
- Chemical property measurements led to discovery of an oxidative process occurring during bead milling. Oxidation appears to be most prevalent on external surfaces and the extent of surface oxidation did not increase with milling time. Since surface charge is known to affect adsorption processes, oxidation due to bead milling may affect S-PAC performance in relation to its parent PAC.
- Pore volume and surface area changes were not a predictable function of milling time; clear trends across all carbon types were not apparent. However, some changes in individual samples, particularly shifts towards meso- and macroporous fraction, are likely significant enough to affect adsorption performance in systems with competitive adsorption.
- Material reporting is lacking among S-PAC studies. Improved reporting of material processing and characteristics would improve the ability to predict which carbon materials will be easiest to mill and what surface chemistry changes to expect. This will enable more accurate design of future S-PAC processes.
5. Acknowledgments
This research is sponsored by a grant from the National Science Foundation, CBET-1236070. SEM images were created at the Clemson University Electron Microscopy Laboratory. Ian DeMass is thanked for assistance with particle characterization.
6. References
Akmil Bas¸ ar, C., Karagunduz, A., Keskinler, B., Cakici, A., 2003. Effect of presence of Ions on surface characteristics of surfactant modified powdered activated carbon (PAC). Appl. Surf. Sci. 218, 170e175. http://dx.doi.org/10.1016/S0169-4332(03)00576-2.
Ando, N., Matsui, Y., Kurotobi, R., Nakano, Y., Matsushita, T., Ohno, K., 2010. Comparison of natural organic matter adsorption capacities of super-powdered activated carbon and powdered activated carbon. Water Res. 44, 4127e4136.
Apul, O.G., Wang, Q., Zhou, Y., Karanfil, T., 2013. Adsorption of aromatic organic contaminants by graphene nanosheets: comparison with carbon nanotubes and activated carbon. Water Res. 47, 1648e1654. http://dx.doi.org/10.1016/j.watres.2012.12.031.
Apul, O.G., Zhou, Y., Karanfil, T., 2015. Mechanisms and modeling of halogenated aliphatic contaminant adsorption by carbon nanotubes. J. Hazard. Mater. 295,138e144. http://dx.doi.org/10.1016/j.jhazmat.2015.04.030.
ASTM Standard D5158-98, 2013. Standard Test Method for Determination of Particle Size of Powdered Activated Carbon by Air Jet Sieving. ASTM International, West Conshohocken, PA.
Biniak, S., Pakula, M., Szymanski, G., Swiatkowski, A., 1999. Effect of activated carbon surface oxygen-and/or nitrogen-containing groups on adsorption of copper (II) ions from aqueous solution. Langmuir 15, 6117e6122.
Biniak, S., Swiatkowski, A., Pakula, Mc, P. of C., 2007. Electrochemical studies of phenomena at active carbon-electrolyte solution interfaces. In: Radovic, L. (Ed.), Chemistry & Physics of Carbon. CRC Press, pp. 125e216.
Boehm, H., 1994. Some aspects of the surface chemistry of carbon blacks and other carbons. Carbon N. Y. 32, 759e769.
Clark, R., Lykins, B., 1990. Granular Activated Carbon: Design, Operation and Cost. Lewis Publishers, Chelsea, MI.
Crittenden, J.C., Trussell, R.R., Hand, D.W., Howe, K.J., Tchobanoglous, G., 2005a. Water quality management strategies. In: Water Treatment: Principles and Design, pp. 221e286.
Crittenden, J.C., Trussell, R.R., Hand, D.W., Howe, K.J., Tchobanoglous, G., 2005b. Adsorption. In: Water Treatment: Principles and Design. John Wiley & Sons, Inc., Hoboken, NJ, pp. 1245e1358.
Dastgheib, S.A., Karanfil, T., Cheng, W., 2004. Tailoring activated carbons for enhanced removal of natural organic matter from natural waters. Carbon N. Y. 42, 547e557. http://dx.doi.org/10.1016/j.carbon.2003.12.062.
Dunn, S.E., Knappe, D.R.U., 2013. DBP Precursor and Micropollutant Removal by Powdered Activated Carbon (Web Report #4294). Water Research Foundation, Denver, Colorado.
Ellerie, J.R., Apul, O.G., Karanfil, T., Ladner, D.A., 2013. Comparing graphene, carbon nanotubes, and superfine powdered activated carbon as adsorptive coating materials for microfiltration membranes. J. Hazard. Mater. 261C, 91e98. http://dx.doi.org/10.1016/j.jhazmat.2013.07.009.
Greenbank, M., Spotts, S., 1993. Effects of starting material on activated carbon characteristics and performance. In: Proc. WATERTECH Expo.
Heijman, S.G.J., Hamad, J.Z., Kennedy, M.D., Schippers, J., Amy, G., 2009a. Submicron powdered activated carbon used as a pre-coat in ceramic micro-filtration.
Desalin. Water Treat. 9, 86e91.
Heijman, S.G.J., Rabinovitch, E., Bos, F., Olthof, N., van Dijk, J.C., 2009b. Sustainable seawater desalination: stand-alone small scale windmill and reverse osmosis system. Desalination 248, 114e117.
Hernandez, A.M., Labady, M., Laine, J., 2014. Granular activated carbon from wood originated from tropical Virgin forest. Open J. For. 4, 208e211.
Heschel, W., Klose, E., 1995. On the suitability of agricultural by-products for the manufacture of granular activated carbon. Fuel 74, 1786e1791. http://dx.doi.org/10.1016/0016-2361(95)80009-7.
Jia, Y.F., Thomas, K.M., 2000. Adsorption of cadmium ions on oxygen surface sites in activated carbon. Langmuir 16, 1114e1122. http://dx.doi.org/10.1021/la990436w.
Karanfil, T., Kilduff, J.E., 1999. Role of granular activated carbon surface chemistry on the adsorption of organic compounds. 1. Priority pollutants. Environ. Sci. Technol. 33, 3217e3224.
Karanfil, T., Kitis, M., Kilduff, J.E., Wigton, A., 1999. Role of granular activated carbon surface chemistry on the adsorption of organic compounds. 2. Natural organic matter. Environ. Sci. Technol. 33, 3225e3233.
Li, L., Quinlivan, P.A., Knappe, D.R.U., 2002. Effects of activated carbon surface chemistry and pore structure on the adsorption of organic contaminants from aqueous solution. Carbon N. Y. 40, 2085e2100.
Liu, G., Ma, J., Li, X., Qin, Q., 2009. Adsorption of bisphenol A from aqueous solution onto activated carbons with different modification treatments. J. Hazard. Mater. 164, 1275e1280. http://dx.doi.org/10.1016/j.jhazmat.2008.09.038.
Lopez-Ramon, M., Stoeckli, F., Moreno-Castilla, C., Carrasco-Marın, F., 1999. On the characterization of acidic and basic surface sites on carbons by various techniques. Carbon N. Y. 37, 1215e1221.
Matsui, Y., Aizawa, T., Kanda, F., Nigorikawa, N., Mima, S., Kawase, Y., 2007.
Adsorptive removal of geosmin by ceramic membrane filtration with superpowdered activated carbon. J. Water Supply Res. Technol. 56, 411. http://dx.doi.org/10.2166/aqua.2007.017.
Matsui, Y., Ando, N., Sasaki, H., Matsushita, T., Ohno, K., 2009. Branched pore kinetic model analysis of geosmin adsorption on super-powdered activated carbon.
Water Res. 43, 3095e3103. http://dx.doi.org/10.1016/j.watres.2009.04.014.
Matsui, Y., Murai, K., Sasaki, H., Ohno, K., Matsushita, T., 2008. Submicron-sized activated carbon particles for the rapid removal of chlorinous and earthy-musty compounds. J. Water Supply Res. Technol. 57, 577. http://dx.doi.org/10.2166/aqua.2008.070.
Matsui, Y., Murase, R., Sanogawa, T., Aoki, N., Mima, S., Inoue, T., Matsushita, T., 2005 Rapid adsorption pretreatment with submicrometre powdered activated
carbon particles before microfiltration. Water Sci. Technol. 51, 249e256.
Matsui, Y., Nakao, S., Sakamoto, A., Taniguchi, T., Pan, L., Matsushita, T., Shirasaki, N.,2005 Adsorption capacities of activated carbons for geosmin and 2-methylisoborneol vary with activated carbon particle size: effects of adsorbent and adsorbate characteristics. Water Res. 85, 95e102. http://dx.doi.org/10.1016/j.watres.2015.08.017.
Matsui, Y., Nakao, S., Taniguchi, T., Matsushita, T., 2013. Geosmin and 2-methylisoborneol removal using superfine powdered activated carbon: shell adsorption and branched-pore kinetic model analysis and optimal particle size.Water Res. 47, 2873e2880. http://dx.doi.org/10.1016/j.watres.2013.02.046.
Matsui, Y., Sakamoto, A., Nakao, S., Taniguchi, T., Matsushita, T., Shirasaki, N., Sakamoto, N., Yurimoto, H., 2014. Isotope microscopy visualization of the adsorption profile of 2-methylisoborneol and geosmin in powdered activated carbon. Environ. Sci. Technol. 48, 10897e10903.
Matsui, Y., Murase, R., Sanogawa, T., Aoki, N., 2004. Micro-ground powdered activated carbon for effective removal of natural organic matter during water treatment. Water Sci. Technol. Water Supply 4 (4), 155e163. Retrieved from http://www.iwaponline.com/ws/00404/ws004040155.htm.
Menendez, J., Illan-G omez, M., Radovic, L., 1995. On the difference between the isoelectric point and the point of zero charge of carbons. Carbon N. Y. 33, 1655e1657.
Mohd Din, A.T., Hameed, B.H., Ahmad, A.L., 2009. Batch adsorption of phenol onto physiochemical-activated coconut shell. J. Hazard. Mater. 161, 1522e1529. http://dx.doi.org/10.1016/j.jhazmat.2008.05.009.
Neely, J., Isacoff, E., 1982. Carbonaceous Adsorbents for the Treatment of Ground and Surface Waters, vol. 21. Pollution Engineering and Technology. Marcel Dekker Inc, New York, United States.
Osantowski, R., Wullschleger, R., 1986. Evaluation of Activated Carbon for Enhanced COD Removal from Pharmaceutical Wastewater. U.S. EPA, Cincinnati, OH.
Patni, A., Ludlow, D., Adams, C., 2008. Characteristics of ground granular activated carbon for rapid small-scale column tests. J. Environ. Eng. 216e221.
Pelekani, C., Snoeyink, V.L., 2000. Competitive adsorption between atrazine and methylene blue on activated carbon: the importance of pore size distribution.
Carbon N. Y. 38, 1423e1436. http://dx.doi.org/10.1016/S0008-6223(99)00261-4.
Pendleton, P., Wong, S.H., Schumann, R., Levay, G., Denoyel, R., Rouquero, J., 1997.
Properties of activated carbon controlling 2-methylisoborneol adsorption.
Carbon N. Y. 35, 1141e1149. http://dx.doi.org/10.1016/S0008-6223(97)00086-9.
Quinlivan, P. a, Li, L., Knappe, D.R.U., 2005. Effects of activated carbon characteristics on the simultaneous adsorption of aqueous organic micropollutants and natural organic matter. Water Res. 39, 1663e1673. http://dx.doi.org/10.1016/j.watres.2005.01.029.
Snoeyink, V.L., Chen, A.S.C., 1985. Removal of organic micropollutants by coagulation and adsorption. Sci. Total Environ. 47, 155e167.
Stoeckli, F., Huguenin, D., Greppi, A., 1993. Primary and secondary filling of micropores in active carbons. J. Chem. Soc. Faraday Trans. 89, 2055. http://dx.doi.org/10.1039/ft9938902055.
Tessmer, C.H., Vidic, R.D., Uranowski, L.J., 1997. Impact of oxygen-containing surface functional groups on activated carbon adsorption of phenols. Environ. Sci.Technol. 31, 1872e1878. http://dx.doi.org/10.1021/es960474r.
Toles, C.A., Marshall, W.E., Johns, M.M., Wartelle, L.H., McAloon, A., 2000. Acidactivated carbons from almond shells: physical, chemical and adsorptive properties and estimated cost of production. Bioresour. Technol. 71, 87e92. http://dx.doi.org/10.1016/S0960-8524(99)00029-2.
Yalcin, M., Arol, A.I., 2002. Gold cyanide adsorption characteristics of activated carbon of non-coconut shell origin. Hydrometallurgy 63, 201e206. http://dx.doi.org/10.1016/S0304-386X(01)00203-1.