Article Cite This: Environ. Sci. Technol. XXXX, XXX, XXX−XXX
pubs.acs.org/est
Stokes Shift and Specific Fluorescence as Potential Indicators of Organic Matter Hydrophobicity and Molecular Weight in Membrane Bioreactors Kang Xiao,†,‡,§ Bingjun Han,† Jianyu Sun,∥ Jihua Tan,† Jinlan Yu,† Shuai Liang,⊥ Yuexiao Shen,*,# and Xia Huang*,‡,§ †
College of Resources and Environment, University of Chinese Academy of Sciences, Beijing 100049, People’s Republic of China State Key Joint Laboratory of Environment Simulation and Pollution Control, School of Environment and §Research and Application Center for Membrane Technology, School of Environment, Tsinghua University, Beijing 100084, People’s Republic of China ∥ National Institute of Clean and Low-Carbon Energy, Beijing 102211, People’s Republic of China ⊥ College of Environmental Science and Engineering, Beijing Forestry University, Beijing 100083, People’s Republic of China # Department of Chemistry, University of California, Berkeley, Berkeley, California 94720, United States
Downloaded via UNIV OF SOUTHERN INDIANA on July 26, 2019 at 05:13:56 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
‡
S Supporting Information *
ABSTRACT: Hydrophobicity and molecular weight (MW) are two fundamental properties of dissolved organic matter (DOM) in wastewater treatment systems. This study proposes fluorescence Stokes shift and specific fluorescence intensity (SFI) as novel indicators of hydrophobicity and MW. These indicators originate from the energy gap and photon efficiency of the fluorescence process and can be readily extracted from a fluorescence excitation−emission matrix (EEM). The statistical linkages between these indicators and hydrophobicity/ MW were explored through investigation of DOM across 10 full-scale membrane bioreactors treating municipal wastewater. Stokes shift was found to exhibit a general rule among the hydrophobicity components in the order of hydrophilic substances (HIS) < hydrophobic acids (HOA) < hydrophobic bases (HOB). The Stokes shift of 1.2 μm−1 is a critical border, above which the relative fluorescence correlated significantly with the HOA-related content (Pearson’s r = 0.8). With regard to MW distribution (100 kDa), SFI was found to be the most sensitive to the change of MW of 0.9). Hydrophobicity-related π conjugation and MW-dependent light exposure might be responsible for the correlations. These fluorescence indicators may be useful for convenient monitoring of DOM in wastewater treatment systems.
1. INTRODUCTION Dissolved organic matter (DOM) plays substantial roles in water and wastewater treatment systems. DOM is inextricably linked to the physical processes (e.g., influencing sedimentation/coagulation/filtration), chemical reactions (e.g., acting as precursors of disinfection byproducts or complexants of heavy metals), and biological performances (e.g., affecting/reflecting biological activity and stability) of the treatment systems.1 Hydrophobicity and molecular weight (MW or molecular size) are two basic facets of DOM properties.1 By affecting the strength and breadth of interfacial interactions respectively,2,3 hydrophobicity and MW can have profound impacts on physicochemical interactions between DOM and co-existing materials (e.g., sludge biomass, adsorbents, air bubbles, membranes, and chemical pollutants) and, hence, the distribution and migration behaviors of DOM.4−7 Hydrophobicity and MW (size) are also two basic dimensions of © XXXX American Chemical Society
descriptors for biological activity of organic molecules, according to the philosophy of the famous Hansch’s model for biochemical structure−activity relationships.8 In relation to the fate of DOM, hydrophobic/hydrophilic and MW compositions of DOM may reflect the source of DOM (relating to, e.g., raw water quality), production of DOM (e.g., state of microbial metabolism), stability of DOM (e.g., solution environment), and elimination of DOM (e.g., biodegradability and chemical reactivity) in wastewater treatment processes.1 Therefore, rapid monitoring of hydrophobic/hydrophilic and MW compositions of DOM, at least qualitatively if possible, Received: Revised: Accepted: Published: A
April 8, 2019 May 21, 2019 June 12, 2019 June 12, 2019 DOI: 10.1021/acs.est.9b02114 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
Figure 1. Scheme of the fluorescence process and the meanings of (a) Stokes shift and (b) specific fluorescence.
specific molecular area for light exposure. With sufficient physical significance, Stokes shift and SFI merit further development as potential indicators of hydrophobicity and MW and need to be examined in a wider range of case studies. This study aims to explore the linkages between Stokes shift, SFI, and hydrophobicity and MW distributions of DOM in wastewater treatment systems. The correlations were conducted across 10 full-scale MBRs treating a variety of municipal wastewaters, to find out if there are any universal rules regarding these relationships. Statistical analyses were employed to evaluate the performances of Stokes shift and SFI in indicating hydrophobicity and MW at qualitative and quantitative levels. The independent as well as cross contributions of different hydrophobic/hydrophilic and MW components to the overall values of these indicators were also assessed for further insight into the fluorescence−hydrophobicity/MW linkages in wastewater treatment systems.
would be of great significance for timely diagnosis and control of the wastewater treatment state during process operation. Conventional measurement methods for hydrophobic/ hydrophilic composition include adsorptive resin column chromatography and reversed-phase chromatography,9,10 and those for MW distribution include sequential ultrafiltration and size-exclusion chromatography.9 These methods are usually laborious and difficult to be implemented for real-time onsite monitoring. In comparison, fluorescence spectroscopic measurements have the advantages of being rapid, sensitive, and easy to automate and provide abundant information to fingerprint organic matter properties.11−13 As an important representative of fluorescence spectroscopy, the excitation− emission matrix (EEM) has been widely used for characterization of DOM in wastewater treatment systems. 11 Fluorescence wavelength regions are qualitatively related to DOM chemical composition and migration/transformation fate in wastewater treatment systems.11,14,15 Semi-quantitative relationships have been established between fluorescence peak intensities and concentration measures (total organic carbon, chemical oxygen demand, protein content, etc.) for wastewater treatment systems.16−18 It is worthy of exploration if fluorescence spectroscopy could further be used to indicate hydrophobicity and MW distribution as well. To this end, it is critical to establish the relationship between fluorescence properties and hydrophobicity/MW and further to refine qualified proxies. A relatively limited number of literature reports have provided useful clues suggesting that the DOM components with different hydrophobicity or MW exhibit different excitation/emission (Ex/Em) wavelengths of fluorescence peaks, and the concentrations of the components could be empirically correlated with the intensity of the fluorescence peaks for surface water or wastewater DOM.19−24 From the essence of the fluorescence process, the Ex/Em wavelengths and fluorescence intensity can be related to fluorophore energy and fluorescence efficiency (fluorescence photon number per unit mass), respectively. The energy loss from fluorophore relaxation is expressed as Stokes shift (i.e., difference of Ex and Em wavenumbers), 25,26 and the fluorescence efficiency is described as specific fluorescence intensity (SFI),23 as illustrated in Figure 1. In our preliminary case study on two membrane bioreactors (MBRs) and an oxidation ditch, it was found that the hydrophobic fraction of DOM had a larger Stokes shift than the hydrophilic fraction27 and that the lower MW fraction had greater SFI than the higher MW fraction.23 These phenomena were tentatively explained from the scale of the π-conjugated system and the
2. MATERIALS AND METHODS 2.1. DOM Sampling. The DOM samples in this study were obtained from 10 full-scale MBR plants treating municipal wastewater in China. The 10 MBRs are located in different regions (North, East, Central, and Southwest China) and different river basins (Hai River, Yangtze River, Tai Lake, and Dian Lake). The process types are anaerobic/anoxic/ aerobic MBR and its variants. The treatment capacity is 20 000 m3/day or more for each plant. The treated wastewater types cover domestic sewage and integrated wastewater containing varied proportions of domestic and industrial wastewaters. Details of the plants (including geographical distribution, wastewater quality, process configuration, and operation parameters) have been provided previously.24 Sludge mixed liquor samples were taken from three separate corridors of membrane tanks as three replicas for each plant. At the time of sampling, all of the plants (except the third plant undergoing sludge bulking) had been stably operated for at least 1 year. The mixed liquor samples were filtered on the spot using a qualitative filter paper and a GF/F glass-fiber filter (0.7 μm, Whatman, U.K.) sequentially to obtain the DOM samples that were encapsulated in a 4 °C coolbox and immediately conveyed to the laboratory for analysis. For the three replicate DOM samples from each MBR plant, the hydrophobicity/MW fractionations, fluorescence EEM measurements, and concentration determinations were carried out separately, with the results averaged. 2.2. Hydrophobic/Hydrophilic and MW Fractionation. Hydrophobic/hydrophilic composition of DOM was characB
DOI: 10.1021/acs.est.9b02114 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
Figure 2. EEM spectra of organic matter from the 10 full-scale MBRs.
fractions were measured using a fluorescence spectrophotometer (F-7000, Hitachi, Japan) with a cuvette light path of 1 cm. Fluorescence intensity was scanned over the excitation (Ex) and emission (Em) wavelengths of 200−450 and 250−550 nm, respectively, at a speed of 2400 nm/min. The slit width was 5 nm, and the photomultiplier voltage was 700 V. Pure water background was subtracted from the EEM spectra, and an graphical interpolation technique was used to smooth the spectra for elimination of Rayleigh and Raman scatterings.32 The fluorescence intensity was corrected for inner-filter effect using ultraviolet−visible (UV−vis) absorbance25 and was devided by the Raman peak area of pure water at λEx = 350 nm to generate a standardized fluorescence intensity in Raman units (RU).33 Stokes shift (SS) is defined as (also illustrated in Figure 1)
terized using adsorptive column chromatography filled with DAX-8 resin (Supelite, Sigma-Aldrich, St. Louis, MO, U.S.A.). At a critical retention factor of 25, DOM was fractionated into hydrophilic substances (HIS) and hydrophobic acids/bases/ neutrals (HOA/HOB/HON). HOA was adsorbed by the resin at low pH (pH 2) and released at high pH (0.1 M NaOH); HOB was adsorbed at neutral pH and released at low pH (0.1 M HCl); HON was adsorbed at all pH values and eluted with organic solvent (methanol); and HIS remained unadsorbed. A detailed procedure of the fractionation has been given previously.24 The fractions, as obtained, were then diluted to have the same volume as the original DOM sample and adjusted to neutral pH. The ionic strength was not adjusted, given its minimal impact on fluorescence.12,28 Considering that methanol contained in HON may interfere with fluorescence,29 fluorescence properties of HON were no longer examined in the present study. HON accounted for no more than 11 ± 9% of total organic carbon (TOC) in the DOM over the full-scale MBRs (see Figure S1 of the Supporting Information) and is considered to be less important for the analysis in this study. MW distribution of DOM was characterized by sequential ultrafiltration using membranes with apparent MW cutoffs of 100, 10, and 1 kDa (PLHK, PLGC, and PLAC, Millipore, Burlington, MA, U.S.A.), thus dividing MW into four fractions: >100, 10−100, 1−10, and HOA > HOB in general. With great variation, the fractions of 100 kDa accounted for 20−54, 18− 44, 6−17, and 3−55% of TOC, respectively. Pearson’s partial correlation analysis showed that HIS correlated closely with the polysaccharide content (r = 0.670 and p = 0.069, excluding the effect of proteins and humics) and HOA correlated closely with the humic content (r = 0.800 and p = 0.017, excluding the effect of polysaccharides and proteins). 3.3. Stokes Shift as an Indicator of Hydrophobicity. Alongside the EEM spectra of the total DOM (Figure 2), the EEM spectra of their hydrophobic/hydrophilic fractions were also measured separately with the fluorescence intensity at each Stokes shift calculated and shown in panels a−c of Figure 3. The distribution of Stokes shift differed among the fractions. HIS had two distinct peaks on both sides of Stokes shift of ∼1.2 μm−1, while HOA and HOB mainly peaked at Stokes shift above 1.2 μm−1. This feature was manifested to be common across all 10 plants. The intensity-weighted average Stokes shifts of the hydrophobic/hydrophilic fractions were calculated and plotted in Figure 3d, revealing the overall trend of HIS < HOA < HOB (Jonckheere−Terpstra test; p = 0.000) with significant difference between any two of them (Wilcoxon signed rank test; p < 0.05). A similar trend was reported in a
Non-parametric tests were performed for qualitative analyses, particularly when the data under investigation disobeyed normal distribution (the normality checked via the Shapiro− Wilk test).
3. RESULTS AND DISCUSSION 3.1. EEM Spectra. EEM fluorescence spectra of DOM from the 10 MBRs are shown in Figure 2, expressed in fluorescence intensity per unit TOC concentration. Among the 10 MBRs, the EEM spectra differed in fluorescence contours (such as peak position and shape) or intensity. This might be related to the difference in DOM properties (such as chemical composition, hydrophobicity, and MW distribution) and originate from the difference in wastewater quality and operating conditions of the treatment plants (related to the source and formation of DOM). The disparities in fluorescence, on the other hand, are conducive to exploration of relationships between fluorescence and hydrophobicity/ MW across a wide variation range of fluorescence properties. 3.2. Hydrophobicity and MW Distributions. TOC concentrations of DOM from the MBR plants ranged from 6.6 to 22.2 mg/L. Polysaccharide, protein, and humic contents were 0.45−0.84, 0.02−0.45, and 0.31−0.52 g/g of TOC, respectively (Figure S1 of the Supporting Information). Polysaccharides and humics took considerable portions of DOM, while the protein content varied greatly among the plants. Proteins and humic substances have been widely reported to have fluorescence properties.12,17,25,35 Polysaccharidic compounds in DOM from the real wastewater treatment systems may also bear fluorophores that pertain to unsaturated segments of the complicated DOM macromolecule. It is also likely that polysaccharides may be tightly bound with proteins36 or exist as a part of a humic acid macromolecule37 to express overall fluorescence. A single D
DOI: 10.1021/acs.est.9b02114 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
Figure 4. (a) Scanning the optimal Stokes shift ranges for correlation of relative fluorescence with HOA proportion and (b) optimal ranges summarized.
Figure 5. (a) Significant and (b) insignificant correlations between SFI and different MW proportions.
also have masked the impacts of HIS and HOB on the total DOM to some extent (as will be discussed in section 3.5). As a result, only the HOA proportion correlated strongly with the average Stokes shift of the total DOM. The HOA proportion was further correlated with fluorescence intensity in specific ranges of Stokes shift. The ratio of fluorescence in a specific range of Stokes shift (SSi ± ΔSS) to that in the whole range of Stokes shift is calculated as follows:
narrower scope of the case study on three wastewater treatment plants.27 Hydrophobic fractions tend to have higher Stokes shifts, possibly as a result of the larger scale of the πconjugated system or higher π-electron density. In contrast, hydrophilic contents (such as polysaccharides) usually have lower aromaticity and, hence, smaller π-conjugated systems.38,39 HOA had a smaller average Stokes shift than HOB, possibly because HOA had a considerable amount of carboxyls (as a major type of acidic group40,41) that may act as electron-withdrawing substituents to increase molecular hardness and decrease Stokes shift.26 Because the different fractions had different Stokes shifts, it is plausible that Stokes shift distribution of the total DOM would vary as a function of the hydrophobic/hydrophilic proportion. Correlation analysis was performed between the hydrophobic/hydrophilic proportion and the average Stokes shift of the total DOM (Table S1 of the Supporting Information). The average Stokes shift correlated positively with the HOA proportion (Pearson’s and Spearman’s p < 0.01) but insignificantly with the HIS and HOB proportions. The insignificance of HIS and HOB might be due to the fact that the fluorescence density was low for HIS [0.035 ± 0.011 RU/ (mg of TOC/L)] and unstable for HOB [0.135 ± 0.089 RU/ (mg of TOC/L), with the error enlarged by its low TOC concentration; cf. Figure S1 of the Supporting Information] compared to that for HOA [0.083 ± 0.023 RU/(mg of TOC/ L)]. Any possible interfraction fluorescence quenching might
fi =
∑ SSi ±ΔSS
I/
∑ any SS
I (3)
where ΔSS is the span of the specific range. Pearson’s correlation was performed between f i and HOA proportion over the 10 MBR plants. For each given ΔSS, the whole Stokes shift distribution profile was scanned to obtain the best correlation with the corresponding SSi ± ΔSS range located (Figure 4a). At the scanning span from 0.8 to 1.1 μm−1, the Stokes shift ranges clearly divide into the negative correlation region (SS = 0−1.2 μm−1; r = −0.8) and positive correlation region (SS = 1.2−3 μm−1; r = 0.8). The Stokes shift of 1.2 μm−1 is a clear border between the positive and negative correlation regions (Figure 4b). The above shows that DOM from the 10 MBR plants demonstrated general rules regarding the relationship between Stokes shift and the hydrophobic/hydrophilic fractions. Stokes E
DOI: 10.1021/acs.est.9b02114 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
Figure 6. Scanning the optimal EEM wavelength regions for correlation of regional SFI with the MW proportions of (a)
