Relations between Environmental Black Carbon Sorption and

Relations between Environmental Black Carbon Sorption and ...https://pubs.acs.org/doi/pdfplus/10.1021/es0498742?src=recsys...
7 downloads 0 Views 141KB Size
Environ. Sci. Technol. 2004, 38, 3632-3640

Relations between Environmental Black Carbon Sorption and Geochemical Sorbent Characteristics GERARD CORNELISSEN,† ZOFIA KUKULSKA,† STAVROS KALAITZIDIS,‡ KIMON CHRISTANIS,‡ AND O ¨ R J A N G U S T A F S S O N * ,† Institute for Applied Environmental Research (ITM), Stockholm University, 10691 Stockholm, Sweden, and Department of Geology, University of Patras, Greece

Pyrogenic carbon particles in sediments (soot and charcoal, collectively termed “black carbon” or BC) appear to be efficient sorbents of many hydrophobic organic compounds, so they may play an important role in the fate and toxicity of these substances. To properly model toxicant sorption behavior, it is important to (i) quantify the magnitude of the role of BC in sorption and (ii) elucidate which geochemical BC characteristics determine the strength of environmental BC sorption. Sorption isotherms of d10phenanthrene (d10-PHE) were determined over a wide concentration range (0.0003-20 µg/L), for five sediments with widely varying characteristics. From the sorption isotherms, we determined Freundlich coefficients of environmental BC sorption, KF,BCenv. These varied from 104.7 to 105.5. From the data, it could be deduced that BC was responsible for 49-85% of the total d10-PHE sorption at a concentration of 1 ng/L. At higher concentrations, the importance of BC for the sorption process diminished to 38 µm means the 38-63-µm-size fraction. Sorption Isotherms. In Figure 2, the sorption isotherms of d10-PHE in the four Finnish sediments are presented. The KET sorption isotherm is in a recent paper (5). The isotherms were fit to the Freundlich equation

CS ) fTOCKF,TOCCnWF

(2)

where fTOC is the fraction of TOC and KF,TOC is the TOCnormalized Freundlich sorption coefficient. The Freundlich parameters are presented in Table 2. For all five sediments, KF,TOC is rather high compared to literature values for KOC of PHE sorption to sediments that are frequently around 4.04.5 (32). Two possible reasons can be given for this: (i) many of the literature KOC values have been measured in batch experiments where PHE sorption to dissolved organic carbon has led to underestimation of KOCsthis “particle concentration effect” often has not been adequately corrected for but is here avoided by the POM-SPE methodsand (ii) we deduced our KF,OC from measurements in mainly the nanogram per liter range whereas literature KOC’s have often been measured at far higher concentrations; isotherm nonlinearity results in smaller KOC’s at higher concentrations. Freundlich n values were just below unity, with the most nonlinear sorption for the HOY sediment (nF ) 0.85). The

most linear sorption was shown by MEK sediment (nF ) 0.98), the sediment with the smallest BC:TOC ratio (BC:TOC ) 0.004) and, thus, the smallest relative amount of nonlinearly sorbing BC. Freundlich coefficients below unity are an indication of adsorption to a limited number of sites, as opposed to a dissolution-like partitioning process for which nF ) 1 would be expected. Organic Petrography. Maceral analyses were performed on the light fraction of MEK, KUO, and KET samples. The results (in volume percent of the OM of the light fraction) were recalculated in weight percent using densities for huminite (1.35 g/cm3), liptinite (1.2 g/cm3), and inertinite (1.5 g/cm3) given in the literature (33; Table 1). For MEK, the huminite content was relatively high (82 wt %). Because huminites originate from tissues of the humic parts of the plants, this is in accordance with the terrestrial δ13C signal of this sediment. The main huminite component of the MEK and KUO sediments was attrinite, a maceral that is ubiquitous (>90%) in soft brown coals (28). The maceral group that is probably most relevant for strong sorption to BC and unburned coal is inertinites, the most condensed, aromatic, and inert macerals (29). Earlier, Kleineidam et al. (34) reported high KOC values for inertinites. KET had a far higher inertinite content than the other VOL. 38, NO. 13, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3635

FIGURE 2. PHE sorption isotherms of the original sediments: log CTOC (µg/kg dw) versus log CW (µg/L). The sorption isotherm of the KET sediment is in ref 5.

TABLE 2. Freundlich Parameters of TOC Sorption, as Well as log KTOC at CW of 1 mg/L and 1 ng/L Calculated Directly from the Freundlich Parametersa

VAR MEK KUO HOY KET a

TOC sorption log KF,TOC [(µg/kgTOC)/(µg/L)nF] 5.03 ( 0.04 5.40 ( 0.14 5.12 ( 0.10 4.76 ( 0.07 5.05 ( 0.06

nF

log KTOC at 1 mg/L (L/kg TOC)

log KTOC at 1 ng/L (L/kg TOC)

0.85 ( 0.02 0.98 ( 0.03 0.92 ( 0.05 0.83 ( 0.03 0.93 ( 0.03

4.6 5.3 4.9 4.3 4.8

5.5 5.4 5.3 5.2 5.2

Note that the log KTOC at 1 µg/L is equal to log KF,TOC.

sediments, in line with the high BC content of this sample. Pyrofusinite is probably the maceral that best resembles BC because it consists of primarily fossil remains of charcoal (29). Calculated as a percentage of whole sediment mass (instead of a percentage of the light fraction as in Table 1), KET sediment contained 0.62% pyrofusinite, a value that is very close to the 0.72% BC of this sediment. However, also inertodetrinite may include significant portions of pyrogenic particles because this maceral consists of small ( KUO > MEK. Size Fractionation of BC and TOC. The distributions of total mass, TOC mass, and BC mass over the six size fractions are shown in Figure 1. Most of the BC resides in the small size fractions for all the Finnish sediments except VAR. The opposite is true for the KET sediment. The reason for this difference could be that the KET BC consists of near-source fall-out material from the industrialized and heavily populated Rhine basin, while the BC in the pristine Finnish sediments probably stems from small particles that are more easily transported over long distances and that, therefore, have settled further from their sources. The observation is in line with a Lake Michigan study by Griffin and Goldberg (35), where the onset of industrialization and the accompanying fossil fuel BC emissions (around 1930) increased both the absolute amounts of sediment BC and the size of the BC particles. Both large absolute amounts and largesized BC particles are observed for the KET sediment, which is from an industrialized area. Also, Rockne et al. (23) observed a relative depletion in BC for the smallest size fraction for sediments that were strongly anthropogenically contaminated. BET SSA. Comparison of the presently measured BET SSA values of the noncombusted bulk sediments with literature values for sediments leads to the following observations: (i) the small values for the low-TOC KUO and VAR sediments (1.9 and 0.4 m2/g, respectively; Table 1) are in line with the low values for sand and aquifer materials (36-38); (ii) the value of 3.9 m2/g for the highly organic MEK sediment (Table 1) is in accordance with literature values for pure OM that are around 4 m2/g (39, 40); (iii) the value for KET is slightly higher (8.2 m2/g; Table 1) and in line with the usual range of sediments with 2-8% TOC (36, 41-43); and (iv) the high BET surface area for HOY sediment (30.9 m2/g; Table 1) is probably due to a high clay content; SSA of clays can be high, up to 40-200 m2/g (44-46). The clayey character of the HOY sediment is confirmed by its small average mesopore diameter (89 Å) compared to the other sediments (161-207 Å). However, ICP-MS analysis revealed that its Si and Al contents were not significantly higher than those of the other four sediments (t test, 95%). OM Resistance to Oxidation. The carbon fraction remaining after the various combustion temperatures is plotted as a function of temperature in Figure 3. The first OC starts to disappear at about 150 °C. Previous research on the combustion of positive (pure) soot standards and negative (nonsoot) standards at various temperatures has shown that 375 °C is the optimal temperature where most of the BC is left whereas all other OC has been combusted (6, 15). About 70-95% of the BC had disappeared after combustion at 504 °C. At 600 °C, 40% modern carbon) than both the PAHs and BC (>90% fossil carbon). PAHs and BC had similar ages. If part of the OC had been charred to BC, the BC age would have been between the value for the PAHs and the value for the TOC. (iii) In high-resolution transmission electron microscopy pictures

before and after CTO-375 treatment, no charcoal was observed (15). (iv) In lake sediment cores, PAH contents were shown to correlate with BC (determined with the CTO-375 method) but not with TOC (6); if charring had occurred, this correlation would have been lost. (v) The environmental distribution of PAHs (47) and polychlorodibenzo-p-dioxin/ furans (10) was significantly better correlated with BC than with TOC. (vi) In a ring test with various other methods of BC determination (22), the CTO-375 method produced the smallest BC contents, indicating that it is more likely that some BC is combusted during CTO-375 treatment than that new BC is formed through charring of other OC. (vii) In the present study, an extremely low BC:TOC ratio of 0.004 was observed for the organic-rich MEK sediment. We conclude that the risk of charring is low for our presently used sediments, but that the CTO-375 method might be too harsh for some BC materials, especially charcoal formed at relatively low temperatures. Therefore, the measured BC contents might be lower limits of the total BC contents and the total BC in the sediments might be more biomass-derived than the 14C signatures of the combusted sediments suggest. On the other hand, high fractions of modern carbon (85 and 74%) were recently observed for two sediments from the Stockholm region treated by exactly the same CTO-375 procedure (Mandalakis and Gustafsson, unpublished results; Cornelissen and Gustafsson, unpublished results). Environmental BC Sorption. In this section, we will deduce KF,BCenv, the “environmental” BC sorption coefficients. KF,BCenv is smaller than the “intrinsic” KF,BC for BC after combustion (KF,BCint) probably because the native PAHs and OM attenuate the BC sorption strength (5). The calculation of KF,BCenv was made possible by the large concentration interval of our sorption isotherms (0.0003-20 µg/L). For this calculation, we used the iterative procedure outlined in detail in ref 5. In this three-step procedure, we first calculated the “BC-free” KF,OC from the highest 3-5 points of the sorption isotherm (in the micrograms per liter range), using eq 1. Here, KF,BCint was used as a starting value. Because we had measured KF,BCint only for the KET sediment [105.53 (5)], we used this as a starting value for all the sediments. nF,BC was 0.54 for the KET sediment (5), and also this value was used for all sediments. In the second and third steps, we calculated KF,BCenv from the BC-dominated low-concentration isotherm points (the nanogram per liter range). Using the KF,BCenv of the first round as a new starting value, the procedure was repeated until a constant KF,BCenv was obtained. The calculated values for KF,BCenv are presented in Table 3. Because we measured the KF,BCint for only one sediment (KET), we evaluated the influence of the generic value of KF,BCint that was used as a starting value for the first iteration. The iteration proved to be stable over a log KF,BCint interval of 4.0-6.0. VOL. 38, NO. 13, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3637

FIGURE 4. Importance of BC for the d10-PHE sorption process (plotted as the percentage of total sorbed d10-PHE that is in the BC) as a function of aqueous concentration CW. The asterisks are added for clarity and do not represent actually measured data points. We also repeated the calculations with various values of nF,BC. The calculated KF,BCenv values proved to be slightly dependent on the nF,BC that was used in the calculation. For example, KF,BCenv was about 0.3 log units higher when using nF,BC ) 0.64 instead of 0.54. Importantly, however, the calculated KF,BCenv at 1 ng/L (also presented in Table 3) proved to be insensitive to the nF,BC used in the calculation, KF,BCenv at 1 ng/L remaining constant within 0.01 log units upon changing nF,BC from 0.54 to 0.64 or 0.74. For the MEK sediment, it proved not to be possible to calculate KF,BCenv from the sorption isotherm. This was caused by the extremely low BC:TOC ratio (0.004; Table 1) that caused OC sorption to overwhelm BC sorption, even at the lowest concentrations. That made it impossible to carry out step ii of the iteration. For the other sediments, the KF,BCenv values were around 105. As a result of the high nonlinearity of BC sorption, KBCenv was much higher at 1 ng/L, between 106 and 107 (Table 3). The “BC-free” log KOC values are also presented in Table 3. They appear to be on the order of 4.6-4.9 for all sediments. These values are rather high compared to most published values for sorption of PHE to sediments that are on the order of 4.0-4.5 (18-20, 34). As discussed above, many literature values have been measured at high concentrations and possibly suffer from a particle concentration effect. On the other hand, our present log KOC values are below most published values for PHE sorption to coal, kerogen, and inertinite that can be as high as 6 (18-20). A log KOC of 5 could, therefore, also point to the presence of a mixture of relatively weakly sorbing amorphous humic/fulvic material and strongly sorbing material in more advanced stages of coalification. Our KF,BCenv values are in the same order of magnitude as literature KBC values for pure soot and charcoal in the nanogram per liter range (1, 2, 17). For environmental BC, Accardi-Dey and Gschwend (14) calculated log KF,BC values that were slightly higher than the currently observed ones [5.6-6.4 (for one sediment, 4.5)] for 11 Environmental Protection Agency soils and sediments. They based their calculations on literature sorption isotherms from Huang et al. (18) and BC contents measured with the CTO-375 method and assumed “BC-free” log KOC values of 4.0. Thus, the reason that the KF,BC values calculated by Accardi-Dey and Gschwend are higher than our current ones may be that their assumed log KOC value was 0.7-1.0 log units smaller than the currently observed ones. Using the KF,TOC and KF,BCenv values and the BC:TOC ratios, it was possible to calculate the percentage of total sorbed d10-PHE that was associated with the BC as a function of CW (Figure 4). It turned out that BC is relatively unimportant for PHE sorption at 1 µg/L, whereas it was more important than 3638

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 13, 2004

the other OC at 1 ng/L (49-85% of PHE in BC). Note that these percentages are independent of the values of KF,BCint and nF,BC that were assumed for the calculation of KBCenv. Relation between BC Sorption and Geochemistry. Two factors probably control BC sorption: site sorption affinity and site capacity. Site affinity is related to aromaticity (BC oxidation resistance), while site capacity is related to BC surface area (BC particle size) and the amounts of competing sorbates (TOC and native PAHs). KF,BCenv values were linearly regressed against the aforementioned geochemical sorbent properties. However, none of the single-parameter relationships between KF,BCenv and geochemical BC characteristics were statistically significant (p < 0.1; r2 < 0.6). Three reasons can be given for the lack of significance: (i) both KF,TOC and KF,BCenv were reasonably constant among the sediments, despite the wide variation of basic sediment characteristics such as TOC, BC, BC:TOC, and particle size distribution (Table 1, Figure 1); (ii) because we could not determine KF,BCenv for MEK sediment, our interpretation of KF,BCenv was restricted to four sediments; and (iii) the potential for spurious correlations. Nevertheless, some trends between KF,BCenv and BC properties are observed. As discussed previously, these trends are not statistically significant. When KF,BCenv increases, there is a general increase of fC, BC fraction