Fractionation and spectroscopic properties of fulvic acid and its extract

Chemosphere 72 (2008) 1425–1434

Contents lists available at ScienceDirect

Chemosphere journal homepage: www.elsevier.com/locate/chemosphere

Fractionation and spectroscopic properties of fulvic acid and its extract Xiaodong Ma a,*, Sarah A. Green b a b

Graduate Institute of Technology, University of Arkansas at Little Rock, 2801 South University Avenue, Little Rock, AR 72204, USA Department of Chemistry, Michigan Technological University, 1400 Townsend Drive, Houghton, MI 49931, USA

a r t i c l e

i n f o

Article history: Received 19 December 2007 Received in revised form 11 May 2008 Accepted 15 May 2008 Available online 7 July 2008 Keywords: Fulvic acid Liquid–liquid extraction Thin-layer chromatography (TLC) UV–vis and fluorescence spectra 3D Matrix fluorescence Model compounds

a b s t r a c t Novel results were obtained when a fulvic acid was isolated from Acros humic acid and fractionated by traditional preparative thin-layer chromatography. Eight colorful bands were directly viewed and analyzed showing very different fluorescence and absorption properties. The fluorescence quantum yield of the bands ranged from 2% to 9.4%, significantly higher than that of natural humic substances. An aqueous fulvic acid solution was also extracted with methylene chloride (CH2Cl2) by continuous liquid–liquid extraction. The CH2Cl2 extract was further fractionated by thin-layer chromatography. Eleven highly fluorescent colorful bands and six weakly fluorescent bands were observed and examined. UV–vis absorption and fluorescence (including 3D matrix) spectra and fluorescence quantum yields revealed that each band still represented a mixture of compounds. Moreover, substantial differences in optical properties were observed among bands. A single band possessed the highest fluorescence quantum yield (6%) and highest specific fluorescence (fluorescence/mass), and accounted for 21% of the total fluorescence of the extract. The mass of individual bands varied from 1.6% to 14.1% of the total materials recovered. Components of all fractions were grouped into 11 fluorophore families according to their maxima on 3D matrix fluorescence spectra. No component is dominant in the whole fulvic acid or extracted portion in terms of optical properties. Over 40 natural products are proposed for model chromophores. Ó 2008 Elsevier Ltd. All rights reserved.

1. Introduction The importance of humic substances to the natural world is well documented (Zepp and Schlotzhauer, 1981; Blough and Green, 1995; Vodacek et al., 1997). Many experimental attempts (Khairy, 1980; Leenheer, 1981; Thurman and Malcolm, 1981; Santos et al., 1994; Busseler et al., 1996; Wander and Traina, 1996; Trubetskoj et al., 1997; Wu and Tanoue, 2001) including PARAFAC analysis (Stedmon et al., 2007) have been made to fractionate these extremely complicated mixtures. XAD resin series have been widely used to extract humic acid (HA) or fulvic acid (FA) as a whole from ocean, river or lake water (Leenheer, 1981; Thurman and Malcolm, 1981; Malcolm et al., 1994; Santos et al., 1994) or soil (Wander and Traina, 1996). Cross-flow filtration has been employed to concentrate organic carbon and to fractionate them by molecular weight (Busseler et al., 1996). Reverse osmosis was able to separate DOM from water by removing solvent and thus leaving the DOM unfractionated (Serkiz and Perdue, 1990). These three methods were not able to further fractionate HA or FA into subfractions of specific characteristics. Preparative and non-preparative thin-layer chromatography (TLC) methods have been applied

* Corresponding author. Tel.: +1 501 569 8041; fax: +1 501 569 8039. E-mail address: [email protected] (X. Ma). 0045-6535/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.chemosphere.2008.05.029

to fractionate FA or HA into 2, 3 or 4 portions (Khairy, 1980; Andres and Romero, 1988; Khairy, 1990). Reverse phase HPLC (Saleh and Ong, 1989; Lombardi et al., 1994; Woelki et al., 1997), size-exclusion chromatography (SEC) (Lin et al., 1995; Trubetskoj et al., 1997; Franke et al., 2004), ultrafiltration (Trubetskoj et al., 1997), flow field-flow fractionation (Zanardi-Lamardo et al., 2002), capillary electrophoresis (Pompe et al., 1996), CE/MS (de Lourdes Pacheco and Havel, 2004) and LC/MS (Zwiener and Frimmel, 2004) have been employed to fractionate HA or FA successfully into 4–7 peaks as determined by UV–vis absorption or fluorescence. Nevertheless, because of lacking narrow-band separation of individual fractions, the numerous components of HA or FA with broad affinity to the chromatographic column produced a continuous chromatogram in the spectrophotometric detector and overwhelmed the spectrum details of different fractions. Moreover, the limited number of absorption and fluorescence spectra collected on HPLC is insufficient for detailed optical analysis, especially fluorescence matrix scans which are essential for multi-fluorophore molecules. Therefore, the chemical and photochemical properties of individual fractions are hard to evaluate. In this study, liquid–liquid extraction was employed to extract the most hydrophobic (partitioned in CH2Cl2) organic materials from a FA isolated from an Acros HA. Preparative TLC was used to fractionate this FA organic extract as well as the methanol-soluble portion of FA. As the modern instruments, like GC/MS and

1426

X. Ma, S.A. Green / Chemosphere 72 (2008) 1425–1434

LC/MS, are developed, traditional separation method like preparative TLC are used less and less due to the labor skill requirement and lacking automation and computerization. However, preparative TLC method has advantages over the modern instruments in terms of large sampling and fractions amount, versatile mobile phase, unmatched direct 2-dimensional image, simple equipment, easy to perform and low cost. In this study, a non-fluorescent TLC plate was adopted to prevent interference from the fluorescent indicator. The developing reagent was selected from tens of tests on small TLC strips by varying the kind, polarity and ratio of many organic solvents (e.g., methanol, acetone, methylene chloride, ethanol, chloroform, propane, hexane and water) to optimize the separation. The mass, UV–vis absorption and fluorescence (including 3D matrix) spectra of the separated fractions and extracted materials were examined, from which fluorescence quantum yield (QY) were

computed for each fraction. The goal of this work was to separate specific less polar portions of FA from the continuous distribution and to study the relationship between the spectroscopic properties of the extract and that of the identifiable sub-fractions of FA. 2. Experimental FA (110 mg) were isolated from 20 g of a commercial humic acid (Acros Organics) at pH 2.0. The dried FA was dissolved into 0.6 ml methanol to form a brown solution which was applied onto a preparative TLC plate (Fisherbrand, 250 lm, 20  20 cm2, Redi/ Plate, Silica Gel G, without fluorescence indicator) using a capillary tube (10 ll). The material formed a dark brown straight line of 16 cm  0.5 cm and was 2 cm from the bottom of the plate. After the methanol evaporated (over night), the TLC plate was developed in a mixed solvent of methanol and ethyl acetate (2:1) for 50 min

a FA

1

A2

Absorbance (normalized)

A3

0.8

A5 A7 A8

0.6

0.4

0.2

0 200

250

300

350

400

450

500

Wavelength (nm)

b CH2Cl2 extract

1.0

B3

Absorbance (normalized)

B6 0.8

B9 B10 B13

0.6

B15

0.4

0.2

0.0 200

250

300

350

400

450

500

Wavelength (nm) Fig. 1. Selected UV–vis spectra of: (a) fractionated FA bands; and (b) fractionated B bands. All spectra were normalized to 1.000 at their peak wavelength. FA was measured in aqueous phase, CH2Cl2 extracted material was measured in CH2Cl2. All others were measured in methanol.

1427

X. Ma, S.A. Green / Chemosphere 72 (2008) 1425–1434

In the second fractionation, 750 ml of filtered fulvic acid solution (containing 3.8 g FA) was extracted with 700 ml of methylene chloride at 40 °C in a continuous liquid–liquid extractor for 42 h to obtain a yellow material (9.2 mg), which was dissolved in 1 ml methanol and was dropped onto a preparative TLC plate. Then the TLC plate was developed in methanol and ethyl acetate (1:1) for 52 min. Seventeen bands of different fluorescence colors and intensity were observed on the plate under the mercury UV lamp. Bands were processed as above to obtain materials of interest and are labeled B1–B17.

in a rectangular glass tank with a sheet of solvent-saturated filter paper. The developed TLC plate was then air-dried over night. Eight bands of different fluorescence colors and intensity were observed under a mercury UV lamp. Each band was individually removed from the plate, and then soaked in methanol to desorb the materials of interest from the silica gel substrate for optical property measurements. These are numbered as band A2–A9. A band of blank silica gel was also removed from the plate and soaked in methanol to obtain a reference for optical measurements. All band spectra were measured in methanol.

2.0E+06 1.8E+06

: s lope 29 2/3 34

29 2 /3 34 33 : s lo 7/4 pe = 1.4X 30 105 2 :s R = lop 0.98 e= 7 6. 8x 10 4 R2 = 0. 98 5

Fluorescence Intensity (cps)

1.6E+06 1.4E+06 1.2E+06 1.0E+06 8.0E+05 6.0E+05 4.0E+05

4 2 0.434 0 R = = 1 .0x 1

0: /4 3 33 7

9 .99 =0

2

R x 10 2.6 : e p s lo 4

2

R

4

sl 30: 0/4 6 3

2.0E+05

ope

6 .99 =0

10 .7x =2

0.0E+00 0

10

20

30

40

Extraction Time (hr) Fig. 2. Extraction kinetics of fluorophores extracted by CH2Cl2 from FA water solution. Wavelengths of excitation and emission are shown on the regression lines respectively. N: Ex/Em = 292 nm/334 nm; r: Ex/Em = 337 nm/430 nm; +: Ex/Em = 360 nm/430 nm; (Ex/Em = 263 nm/334 nm, not shown).

Table 1 Summary of developed bands of FA and CH2Cl2 extracted materials Band

Rf

Mass (mg)

Mass percentage (%)

As (mg1)

Fs (cps * mg1)

QY (%)

Fluor. percentage (%)

Ex/Em (nm/nm)

A2 A3 A4 A5 A6 A7 A8 A9 B1 B2 B3 B4 B5 B6 B7 B8 B9 B10 B11 B12 B13 B14 B15 B16 B17

0.00 0.03 0.06 0.10 0.35 0.69 0.84 0.93 0.00 0.09 0.18 0.26 0.34 0.48 0.60 0.65 0.68 0.73 0.77 0.79 0.81 0.84 0.92 0.95 0.98

11.7 0.6 0.4 0.1 2.8 18.9 1.0 1.3 0.5 1.0 0.4 0.4 0.6 1.1 0.7 0.4 0.4 0.1 0.3 0.2 0.3 0.2 0.5 0.4 0.1

31.8 1.6 1.1 0.3 7.6 51.4 2.7 3.5 7.0 12.5 5.5 5.5 7.8 14.1 8.6 4.7 5.5 1.6 3.9 3.1 3.9 3.1 7.0 4.7 1.6

0.17 0.22 2.01 3.55 0.42 0.09 0.77 1.77 4.51 1.11 1.04 5.21 1.57 1.89 1.41 1.55 4.37 7.55 1.35 6.24 1.73 11.51 6.39 4.97 5.68

5.84  107 2.18  108 2.65  108 5.07  109 3.12  108 7.35  107 2.79  108 5.97  108 1.33  109 2.19  108 8.56  108 9.91  108 4.49  108 8.01  108 3.93  108 7.07  108 4.24  109 3.87  109 1.06  109 6.87  108 7.88  108 1.06  109 1.09  109 2.48  109 2.91  109

2.19 6.62 0.86 9.37 4.83 5.16 2.37 2.22 1.94 1.30 5.41 1.25 1.88 2.78 1.83 2.98 6.36 3.36 5.14 0.72 2.99 0.60 1.12 3.27 3.36

14.4 2.8 2.2 10.7 18.4 29.3 5.9 16.4 8.6 2.5 4.3 5.0 3.2 10.3 3.1 3.0 21.2 5.5 3.8 2.0 2.8 3.0 7.0 10.6 4.2

435/519, 465/533 255/427, 310/425 343/422 237/315, 255/431, 281/314, 313/428 258/435, 316/428–333/436 257/435, 314/427–353/441 255/319, 274/316, 309/421 257/420, 306/420 262/452, 280/317, 314/429, 343/435–368/454 270/319, 308/423, 344/432–450/511 257/443, 315/433, 340/437 264/454, 280/317, 317/428, 345/429–364/451 260/437, 280/317, 336/427 271/468, 370/465 261/425, 306/415, 334/428 281/314, 306/418, 334/427 300/412 277/312, 301/413 234/349, 246/431, 287/346, 302/416 235/351, 277/313, 287/346, 306/416, 338/423 236/352, 249/435, 290/352, 307/425, 348/432 236/353, 252/438, 289/351, 308/429, 352/437 266/377, 280/325, 313/390, 363/449 262/370, 280/330, 313/369 237/345, 268/406, 276/323

1428

X. Ma, S.A. Green / Chemosphere 72 (2008) 1425–1434

An HP 8452A UV–vis spectrophotometer was employed to measure the UV–vis absorption spectrum of all samples. The fluorescence spectra of the samples were collected on a SPEX Fluorolog F112 fluorometer with the reference spectrum subtracted for Raman peak and background. Fluorescence excitation wavelengths were chosen to give the maximum emission intensity for each sample. 3D matrix fluorescence spectra were also measured. All samples were diluted to avoid inner filter effects. All spectra were not corrected for instrument-specific effect. To express the absorptivity per unit mass material, specific absorbance (As) is defined as

A m

As ¼

ð1Þ

where A stands for total absorbance (corrected for dilution) of the material at a given wavelength, and m is the mass of the material. Fluorescence QY was calculated using quinine sulfate monohydrate (in 0.1 N H2SO4, U = 0.51) as the standard (Vodacek et al., 1994). Specific fluorescence (Fs) (fluorescence per unit mass) is defined in this work as

IF m

Fs ¼

3. Results and discussion The absorption of the aqueous FA water solution decreases approximately exponentially from <200 nm to around 500 nm with a shoulder around 210 nm (Fig. 1a). The fluorescence spectrum with an excitation of 280 nm has a broad emission from 300 to 700 nm showing a peak at 435 nm. The position of the emission peak shifts as a function of excitation wavelength. These observations indicate that the FA is a mixture of many components which have different optical properties. 3.1. Methylene chloride extraction of aqueous FA solution To discover highly fluorescent chromophoric fractions within FA, an extraction experiment was conducted to recover the most non-polar organic matter from an aqueous FA solution. Subsequently, methylene chloride extracted 10 mg yellow material from the aqueous solution. Although the extracted material was a small fraction of the total mass, it has a considerable higher QY (4.9%) than that of the parent FA (1.5%). Considering the solvent

ð2Þ

where IF stands for total integral fluorescence (corrected for dilution) of a sample excited at its peak excitation wavelength (from 255 to 318 nm).

a

6.0E+09

4.0 Specific Absorbance

Specific Fluorescence

8.0 7.0

40

6.0 5.0

30

4.0 20

3.0

2.0 1.5

2.0E+09 1.0 1.0E+09

0.0

A2

A3

A4

A5

A6

A7

A8

b

A9

12.0

12

6.0 5.0

10 4.0 8 3.0 6 2.0

4

Quantum Yield (%)

Percent of Mass QY

Specific Absorbance (1/mg)

7.0

14

A4

A5

A6

A7

A8

A9

14.0

Band Number 16

A3

Band Number

0.0

0

Mass Percentage (%)

0.0E+00 A2

1.0

b

3.0E+09

0.5

2.0

10

4.0E+09 2.5

Specific Fluorescence (cps/mg)

9.0

Specific Absorbance Specific Fluorescence

5.0E+09

4.0E+09

10.0

8.0

3.0E+09

6.0 2.0E+09 4.0 1.0E+09 2.0

2

1.0

0

0.0 B1 B2 B3 B4 B5 B6 B7 B8 B9 B10 B11 B12 B13 B14 B15 B16 B17

0.0

B1 B2 B3 B4 B5 B6 B7 B8 B9 B10 B11 B12 B13 B14 B15 B16 B17

Mass Percentage (%)

50

5.0E+09 3.0

Specific Fluorescence (cps/mg)

Percent of Mass QY

Quantum Yield (%)

a

Specific Absorbance (1/mg)

3.5 10.0

60

0.0E+00

Band Number

Band Number Fig. 3. Mass percentage and fluorescence QY of: (a) fractionated FA bands; and (b) fractionated B bands.

Fig. 4. Specific absorbance and specific fluorescence of: (a) fractionated FA bands; and (b) fractionated B bands. UV absorbance and fluorescence emission intensity were measured at the peak wavelengths.

1429

X. Ma, S.A. Green / Chemosphere 72 (2008) 1425–1434

from the parent FA. The fluorescence spectrum of the extracted material showed multiple peaks, suggesting that multiple fluorophores were extracted from FA aqueous solution. A large well-defined fluorescence emission peak appeared consistently at 333–335 nm with excitations at 252, 261, 280 and 292 nm. However, a second peak at >380 nm shifted to the red with increasing excitation wavelength from 333 nm to 400 nm.

Fluorescence Intensity (cps)

difference between the extracted material (in CH2Cl2) and the parent FA (in aqueous phase), a separate study was done showing that only half of the net increase could be due to solvent effect, i.e., the QY of the extracted material should be 3.3% in aqueous phase. The UV–vis spectrum of the extracted material had an absorption peak at 232 nm, with two shoulders at 265 and 300 nm (Fig. 1b). These characteristics differentiated the extracted material

a

m)

th(n

ng ele

v

Emission Wavelength (nm)

Fluorescence Intensity (cps)

b

ion itat

Wa

Exc

300000

200000

100000 450 400 300 250

0 300

400

500

600

m)

h(n

gt len

350

ve

on

ti cita

Wa

Ex

Emission Wavelength (nm)

c Fluorescence Intensity (cps)

19E+05

14E+05

09E+05

500000

450 400

0

350 300 250 400

450

500

550

Emission Wavelength (nm)

600

ion itat

)

nm

th (

ng ele

v

Wa

Exc

Fig. 5. Matrix scanned fluorescence spectrum of band A5 (a), band B11 (b), and band B9 (c) dissolved in methanol; partial contour map of CH2Cl2 extracted material (nonfractionated) with the positions of the fluorescence maxima of the fractionated bands indicated by the band numbers (d).

1430

X. Ma, S.A. Green / Chemosphere 72 (2008) 1425–1434

Fig. 5 (continued)

Table 2 Model compounds with similar fluorophores Fluorophore family

Maxima center (nm/nm)

Stokes shift (nm)

Model compounds

Bands involved

Ref.

1 2 3

235/350 248/317 260/435

115 69 175

Biphenyl Adenosine monophosphate (AMP); CMP; TMP 5a-Androstan-3-one

Berlman (1971) Wolfbeis (1985) Wolfbeis (1985)

4

268/406

138

5a-Pregnan-3-ol-2/-one, TMPO

B11, B12, B13, B14, B17 A3, A5, A8 A3, A5, A6, A7, A8, A9, B1, B2, B3, B4, B5, B6, B7, B8, B10, B11, B12, B13, B14 B17

5

275/305

30

Tyrosine; tannins

B3, B15

6 7

275/350 308/425

75 117

Tryptophan 2-Aminobenzoic acid; salicylic acid

8

313/380

67

9

350/434

84

10

>350/>460

110

11

450/526

76

7-Hydroxy-6-methoxy-3-methylchromones; 3-hydroxy coumarin; 3-aminobezoic acid; 4-hydroxycinnamic acid; gallic acid; pyridoxine; cinchroninic acid; nistatin 6-Hydroxy coumarin; 7-hydroxy-3-phenyl coumarin; 7hydroxy-4-phenyl coumarin; 6-hydroxy-7-methoxy coumarin; isodrimatol coumarin; ferulic acid; xanthurenic acid; NADH; colchicines; quinidine;chlortetracycline; quinones Zearalenone; 1,6-diphenylhexatriene; 1-amino anthracene; fluoranthene; 3,30 -bifluoranthenyl; quinine bisulfate; xerocomic acid; amphotericin; quinones Riboflavin tetrabutyrate; pulvinic acid; bilirubin; quinizarin; xerocomic acid; FMN; FAD

B11, B12, B13, B14, B15, B16 B1, B2, B3, B4, B7, B8, B9, B10, B11, B12, B13, B14 B15, B16

For fluorophores with different excitation/emission characteristics, the extraction kinetics varied. Plot of fluorescence vs. extraction time at four different excitation and/or emission wavelengths showed different extraction rates. So, there must be at least four fluorophore families in the CH2Cl2 extract (Fig. 2). During first 8 h of the extraction, fluorophores with short wavelength fluorescence had much higher extraction rates (1.5  105 and 1.4  105 cps/h for fluorophores with Ex/Em = 263 nm/ 334 nm and Ex/Em = 292 nm/334 nm respectively) than that with longer wavelength fluorescence (6.8  104 cps/h for fluorophores with Ex/Em = 337 nm/430 nm) (Fig. 2). Generally speaking, larger fluorophores emit fluorescence at longer wavelengths (Ma and

Berlman (1971); Wolfbeis (1985) Maie et al. (2006); Wolfbeis (1985) Wolfbeis (1985) Wolfbeis (1985) Wolfbeis (1985)

A4, B1, B2, B3, B4, B5, B7, B8, B12, B13, B14, B15

Cory and McKnight (2005); Wolfbeis (1985)

A3, A4, A5, A6, A7, A8, A9, B1, B3, B4, B5, B6, B7, B8, B10, B12, B14, B15 A2

Berlman (1971); Cory and McKnight (2005); Gaylord and Brady (1971); Wolfbeis (1985) Gaylord and Brady (1971); Wolfbeis (1985)

Green, 2004). Therefore, these rates suggest that smaller fluorophores entered the organic phase more easily in short-term extraction. After 8 h, the rates of all four fluorophores decreased, with that of the short wavelength fluorophores decreasing more than that of the longer wavelength ones. After 18 h, the rates for all fluorophore families were very close (2.5  104 cps/h). At the end of the 42 h extraction, the rates of small fluorophores were even lower than that of large ones, perhaps due to saturation. From the mass obtained, it clearly shows that only a small part of FA can be extracted into CH2Cl2 phase. While short-term extraction is in favor of small chromophores and

X. Ma, S.A. Green / Chemosphere 72 (2008) 1425–1434

fluorophores, long-term extraction is more appropriate for larger compounds. 3.2. FA fractionation on preparative TLC plate Eight bands (A2–A9) appeared on the preparative TLC plate for the methanol-soluble FA. The retention factors (Rf) of those bands are shown in Table 1. Five bands (A2, A4, A6, A7 and A9) were detectably fluorescent by human eyes, showing very different colors and intensities under UV light, with one (A7) having strong fluorescence. After dissolution in MeOH, study of the individual band spectra revealed that band A8 had very short wavelength fluorescence emission at 318 nm when excited at 255 nm. The other seven bands emitted fluorescence at around 430 nm with excitation at either 255 nm or 312–318 nm. Therefore, there exist at least two different yet separable chromophores in this FA sample, exemplified by A9, which had a Stokes shift of only 63 nm, and A7, with a Stokes shift as large as 115 nm. A separate study showed that fluorescence peaks of FA in methanol shifted up to 10 nm to red for both excitation and emission wavelengths while decreasing Stokes shift for up to 6 nm compared to that in water. Further evidence for multiple chromophores was found in the UV–vis absorption spectra of developed bands. Several bands have absorption peaks or shoulders close to their peak excitation wavelengths. For example, bands A7 and A8 have a shoulder at 254 nm, A3 has a peak at 300 nm, and A5 has a peak at 280 nm (Fig. 1a). The distribution of mass among the developed bands varied greatly, from 0.1 mg (A5) to about 19 mg (A7), corresponding to 0.3% to 51% of the total mass (Fig. 3a). The fluorescence distribution was likewise uneven, ranging from 1.1  108 (A4) to 1.4  109 cps (A7), or 2.2% to 29% of the total fluorescence on the plate. The percentage of mass did not correspond to that of fluorescence, i.e., Fs was not constant from band to band. Because Fs = IF/m and QY / IF/A, Fs / As  QY. Thus, the variation in Fs value indicated the variation in fluorescence specific absorbance and QY. Calculations showed that most FA bands had a higher QY (from 2.2% to 9.4%) than humic substances in natural water (1%), with only one band (A4) having a smaller value (0.86%) (Fig. 3a). Although the band with the strongest fluorescence (A7) possessed only a moderate QY (5.2%), its high mass percentage (>51%) made it the brightest on the plate under UV 254 nm light irradiation. The QY of A5, a yellow colored band, was remarkably high (9.4%), almost 10 times that of the natural water sample. Because of that, band A5 contributed 11% of the total fluorescence although it comprised only 0.3% of the total mass of FA on the plate (Fig. 3a). As it developed in the middle lower part of the TLC plate, the polarities of the organic compounds in the A5 fraction were intermediate. As discussed before, part of the increase was due to the solvent effect. Through calculation, the As values at 260 nm of eight individual bands ranged from 0.22 (A7) to >12 mg1 (A5) respectively (Fig. 4a). The As value of A5 band was over 55 times higher than that of the lowest one (A7). Varying more widely, the Fs value of the developed bands was found to be from 5.8  107 to 5.07  109 cps/mg (Fig. 4a). Again, band A5 displayed the highest Fs value of all bands, 87 times higher than that of the lowest band (A2). Its highest specific absorbance and specific fluorescence, together with its highest QY, distinguish A5 from other bands for retaining the strongest capability of not only absorbing light per unit mass, but also emitting light both per unit mass and per photon absorbed.

1431

acetate = 1:1) was tested to give the best separation. Seventeen bands (B1–B17) were clearly resolved on the TLC plate under UV 254 nm light, with one band (B9) showing very bright blue fluorescence. The Rf values are shown in Table 1. More bands were developed from the FA extract than from FA itself. This apparent paradox is an artifact of the complexity of FA. In the parent FA, the numerous components formed a continuum of colored background, in which the color and fluorescence from the components of small concentration were overwhelmed. In addition, there could exist a chain of energy transfer and charge-transfer processes among the components, quenching the fluorescence of some fluorophores and causing fewer observable bands (Del Vecchio and Blough, 2004). After being extracted into CH2Cl2, the much smaller subset of compounds was likely to spread on the TLC plate from each other enough to show clearer separation. Moreover, the interaction force chain among components probably broke due to loss of most components, making the fluorescence of each individual band observable. The fluorescence emission spectra of the bands differed dramatically when exposed to the same excitation wavelength. When excited at 280 nm, B16 and B17, emitted very short wavelength fluorescence at 320 nm, but other bands (like B3 and B13) produced much longer wavelength emission, from 430 to over 450 nm. These results confirmed that the extracted material possessed multiple fluorophores. The mass of the extracted and fractionated materials ranged from 0.1 to 1.1 mg. One strongly fluorescent band, B17, and a weakly fluorescent band, B10, had the lowest mass, only 1.6% (0.1 mg) of each of the total mass on the TLC plate. Band B6 had the highest mass, 14% (1.1 mg) of the total mass on plate. Extending from 2% to 21% of the total fluorescence, the contribution of each band to the total fluorescence was different from that to the total mass of the whole extracted material (Fig. 3b). For example, bands B17 generated 4.2% of the total fluorescence, ranking 9th among all 17 bands. B6 was ranked only 3rd highest in fluorescence. The QY of individual bands varied from 0.7% to 6.4% (Table 1). Excluding the contribution from solvent effect (1/3), there were 12 bands holding higher QY than natural humic substances, up to over four times higher (B9). These observations suggest that those partition coefficients of highly fluorescent compounds in CH2Cl2 are higher than that of less fluorescent material, consistent with the fact that most fluorescent materials are conjugated ring systems, which are likely hydrophobic, and hence have a greater chemical affinity to organic solvents than to water. Being not correlated well, As and Fs values changed in a wide range. The value of As varied from 1.04/mg (B3) to 11.5/mg (B14) (Fig. 4b), while that of Fs ranged from 2.2  108 cps/mg (B2) to 4.2  109 cps/mg (B9). Although it had only 5.5% of the total mass on the TLC plate, band B9 gave off more than 21% of the total fluorescence. The factors affecting the apparent fluorescence of each band on the plate include QY, Fs value, and mass percentage. Some bands, like B2, B5 and B7 that contributed a high proportion to the total mass (12.5%, 7.8% and 8.5%, respectively), contributed very little fluorescence (2.5%, 3.2% and 3.1%, respectively) due to their small Fs values and low QYs. High QY is not the only determining reason for apparent fluorescence. For instance, band B16 generated 10.6% of the total fluorescence with only 4.7% of the mass, due to its high Fs value (2.5  109 cps/mg) and moderate QY (3.3%) (Fig. 4b). 3.4. Matrix fluorescence spectra

3.3. TLC study of extracted material The CH2Cl2 extracted material was less polar than the FA solution. Therefore a less polar developing solvent (methanol:ethyl

Three-dimensional excitation/emission matrix fluorescence spectra (Coble et al., 1990) provide a powerful spectroscopic method for determining fluorophore diversity. Fluorescence matrix

1432

X. Ma, S.A. Green / Chemosphere 72 (2008) 1425–1434

scans revealed that almost all developed bands from both sets of TLC plates were still composed of multiple fluorophores. All FA bands (A series) exhibited multiple maxima on the matrix spectra. For instance, band A8 has three maxima at Ex/Em = 274 nm/ 316 nm, 309 nm/421 nm and 255 nm/421 nm, while A5 has four at 237 nm/315 nm, 281 nm/314 nm, 255 nm/431 nm, and 314 nm/ 428 nm (Fig. 5a). Some of the maxima were also found by other researchers (Boehme and Coble, 2000). Clearly, the composition of the fractionated bands is still complex. Likewise, most bands of the extracted materials (B) showed multiple maxima: Band B3 had three maxima and band B11 had four maxima (Fig. 5b). However, band B9 had only one strong maximum at 300 nm/412 nm with very faint longer wavelength fluorescence (Fig. 5c), suggesting that B9 was an almost pure fluorophore. Some maxima from different bands occurred at the same Ex/Em position, implying that the chromophores within these bands are either the same or within the same fluorophore family. Based on these observations, we propose that developed bands contain no fewer than 11 fluorophore families (Table 2). Among all bands, there was not one whose maxima were identical to those of the extracted material before TLC fractionation. Instead, the maxima of individual bands were clustered around the maxima of the latter (Fig. 5d). These results reinforce the evidence that FA is a very complex material incorporating many components possessing diverse structural and optical properties. The center bands in each category, A5 and B9 each possessed the highest QY on their plates, implying that band B9 might be a sub-fraction of band A5 extracted into methylene chloride. However, the Rf value showed B9 (Rf = 0.68) was less polar than that of A5 (Rf = 0.10). Counting polarity difference in the development solvents, B9 is much less polar than A5. Comparison of both A and B bands demonstrated that no two bands possessed same number of maxima at the same wavelengths. The UV–vis absorption spectra of B1 and B9 were very similar (peak at 216 nm with a shoulder at 294 nm), but B1 showed four fluorescence matrix maxima, while B9 only had one maximum at 300/412 nm. On the other hand, although B1, B3, and B4 each had three maxima (around 340/437, 314/430 and 262/450 nm) with same relative intensity in each band, their UV–vis spectra were significantly different: B1 had a peak at 218 nm, B3 had two peaks at 236 and 296 nm, and B4 had a peak at 228 nm. Interestingly, some fluorophore families are located in close wavelengths as the fluorophores approached by the PARAFAC statistical method (Stedmon et al., 2007), which implies that the chemical separation in this work and the statistical analysis method corroborate each other somehow. 3.5. Comparison of FA and extract fractions There are some similarities between the developed bands of FA (A set) and FA extract (B set). In both sets, bands with the largest Rf values (A8, A9, B16 and B17) have fluorescence peaks at short wavelengths around 318–320 nm with excitation wavelength from 255 to 280 nm, implying similar fluorophore structures in the least polar fractions. In contrast, the most polar bands (low Rf values) have fluorescence peaks at longer wavelengths, 430–450 nm, which may reflect the red shift in fluorescence expected with the addition of heteroatoms (O, N, S) which also increase polarity. Reckoning solvent effect, the QY of 19 bands was higher (>1.5%) than that of natural humic substances (1%), with bands holding the highest fluorescence QYs (A5 and B9) falling in the middle of each set. Compared to FA, the contributions from each fraction of the extracted materials are more diverse. Among FA fractions, two bands

(A2 and A7) comprised 31.8% and 51.4% of the total mass respectively, while in the FA extract, the heaviest band (B6) comprised only 14.1% of the total mass. The highest As value of the FA set (A5, 3.55/mg) is 38 times that of the lowest value (A7, 0.093/mg), whereas the highest As value of the extract set (B14, 11.5/mg) is only 11 times that of the lowest value (B3, 1.04/mg). The Fs value of A5 (highest) is almost two orders of magnitude higher than that of A2 (lowest), while the Fs value of B9 (highest) is only 19.3 times that of B2 (lowest). In contrast, the range in fluorescence was similar for A and B sets: A bands ranged from 2.2% (A4) to 29.3% (A7) of the total fluorescence, and B bands ranged from 2.0% (B12) to 21.2% (B9). Again, the continuum of components and the interaction force in FA could account for the dense contribution. After extracted into CH2Cl2, the much fewer components and much less interaction force made the division of bands clearer and more divisions. The highest QY (6.4%) in the B bands was not as high as that (9.4%) from the original FA. Therefore, some fluorophores with high QYs in the FA aqueous solution were not extractable by methylene chloride. 3.6. Model fluorophores To probe the chemical species in the developed TLC bands, some organic compounds were proposed for model compounds within FA. From Table 2, it can be seen that the Stokes shift of individual groups ranged from 30 nm to over 175 nm, indicating the structures of these fluorophores and their interactions with the solvent, methanol, were quite different. Published spectra of a variety of compounds, including 1,2,4-benzenetricarboxylic acid, maleic acid, 2,4,6-trihydroxybenzoic acid, resorcinol, vanillic acid and crotonaldehyde, natural coenzymes, vitamins, alkaloid, heterocyclic compounds (chromones, coumarins, aflatoxins, pulvinic acid, etc.), pyrrole pigments, quinines, quinones, polyenes, amino acid metabolites, urinary metabolites, phenols and acids were examined to compare the fluorescence properties with the developed bands of FA and its extract. A tryptophan-like fluorescence (Ex/Em = 288 nm/342 nm) was observed in developed FA extract bands B11–B14 and B16, and a tyrosine-like fluorescence (Ex/Em = 275 nm/305 nm) (Wolfbeis, 1985; Coble et al., 1990) was observed in five A bands and six B bands. These peaks may indicate that amino acids or dissolved proteins are present in FA. However, tannins also have been found to fluoresce in this region (Maie et al., 2006). The fluorescence maximum of band B9 was close to that of 2-aminobenzoic acid and salicylic acid (Wolfbeis, 1985), both of which are likely constituents of FA. Nevertheless, metabolites of contaminants, like succinic acids of toluene and methylnaphthalene, can also contribute strong fluorescence (Kumke et al., 1999; Zwiener et al., 1999; Ohlenbusch et al., 2002). An adenosine monophosphate (AMP)-like fluorescence was found in bands A3, A5 and A8 (Wolfbeis, 1985). Many bands had coumarin-type fluorescence. And gallic acid, due to its existence in many plants, was at least as likely to be present as tryptophan. These model chromophores, all of which exist in plants and other living organisms, could be incorporated into fulvic or humic acid through microbial action and/or low-temperature chemical reactions (Rullktter, 1993). Two fluorescence families, #2 and #11, were only associated with A bands and might contain relatively hydrophilic groups. In contrast, many more maxima were associated only with the B bands, e.g., fluorescence families #1, #4, #6, #7, and #8. These hydrophobic fluorophores were probably too dilute to appear in less fractionated FA or were overwhelmed by more intense or more concentrated components. The rest fluorescence families, #3, #5, #9, and #10, exhibited in both A and B bands.

X. Ma, S.A. Green / Chemosphere 72 (2008) 1425–1434

The structures of these model compounds range from short chain fatty acids to substituted aromatic and heterocyclic compounds. This list is intended to demonstrate the diverse mixtures that could contribute to FA optical properties. Certainly other chromophores may have absorption/fluorescence spectra in these spectra regions. The combination of compounds actually present in FA must contribute to both the observed exponentially decreasing absorbance from UV to visible spectral regions, and to the broad multi-component fluorescence, that is, evident from the matrix scans. It is worth noting that some of the model compounds were not analyzed in water (or methanol), which can cause solvent effect shift in spectrum. Disentangling FA fluorescence is complex because, unlike absorbance, the sum of individual components in a mixture is not necessarily equivalent to the fluorescence of the whole. Specifically, overlapping emission/excitation peaks of different components can result in energy transfer, charge migration, structural reorganization, or some combination of these which can quench the fluorescence of the donor and produce fluorescence of the acceptor. These donor–acceptor interactions may exist within all natural hydroxy- or polyhydroxy-aromatic polymers that form appropriate acceptors upon partial oxidation (Del Vecchio and Blough, 2004), and are especially probable for non-polar compounds that associate into larger structures in aqueous solution (Sutton and Sposito, 2005). These types of energy or charge-transfer absorbance and/or fluorescence would not be observed when the species are separated. 4. Conclusions We employed traditional preparative TLC method to fractionate a fulvic acid and its organic extract, and visually observed many clearly separated colorful bands successfully. The desorbed compounds existed in the bands vary extensively in mass, absorption and 3D fluorescence spectroscopies. Some fractions exhibited significantly higher QY than the parent fulvic acid despite the solvent effect. Fluorescence to absorbance ratio, specific absorbance and specific fluorescence revealed the complexity of fulvic acid from different aspects. More than 40 model organic compounds are proposed to represent the chromophores and fluorophores in this natural material. Fulvic acid is a complex polydisperse mixture of compounds that can be partially fractionated by various methods. Preparative TLC method is by no means inferior to modern chromatography instruments in separation and visualization of complex systems. Acknowledgements The authors are grateful to the partial financial support from Michigan Space Grant Consortium (National Aeronautics and Space Administration) and the National Science Foundation through the Keweenaw Interdisciplinary Transport Experiment in Superior Project (OCE-9712872) for this work. References Andres, J.M., Romero, C., 1988. Fractionation of raw and methylated fulvic acids from lignite by thin-layer chromatography. Fuel 67 (March), 441–443. Berlman, I.B., 1971. Handbook of Fluorescence Spectra of Aromatic Molecules. Academic Press, New York. Blough, N.V., Green, S.A., 1995. Spectroscopic characterization and remote sensing of nonliving organic matter. In: Zepp, R.G., Sonntag, C. (Eds.), The Role of Nonliving Organic Matter in the Earth’s Carbon Cycle. John Wiley & Sons, New York, pp. 23–45. Boehme, J.R., Coble, P.G., 2000. Characterization of colored dissolved organic matter using high-energy laser fragmentation. Environ. Sci. Technol. 34, 3283– 3290.

1433

Busseler, K.O. et al., 1996. An intercomparison of cross-flow filtration techniques used for sampling marine colloids: overview and organic carbon results. Mar. Chem. 55, 1–31. Coble, P.G., Green, S.A., Blough, N.V., Gagosian, R.B., 1990. Characterization of dissolved organic matter in the Black Sea by fluorescence spectroscopy. Nature 348 (6300), 432–435. Cory, R.M., McKnight, D.M., 2005. Fluorescence spectroscopy reveals ubiquitous presence of oxidized and reduced quinones in dissolved organic matter. Environ. Sci. Technol. 39 (21), 8142–8149. de Lourdes Pacheco, M.H., Havel, J., 2004. Mass spectroscopic and capillary electrophoresis studies of humic acids interactions with various compounds and xenobiotics. In: Ghabbour, E.A., Davies, G. (Eds.), Humic Substances: Molecular Details and Applications in Land and Water Conservation, Proceedings of the Seventh Humic Substances Seminar, Boston, MA, March 17–19, Taylor and Francis, Inc., New York, pp. 79–98. Del Vecchio, R., Blough, N.V., 2004. On the origin of the optical properties of humic substances. Environ. Sci. Technol. 38 (14), 3885–3891. Franke, K., Roessler, D., Kupsch, H., 2004. Use of radioactive tracers for the characterization of humic and fulvic acids in high performance size exclusion chromatography. In: Ghabbour, E.A., Davies, G. (Eds.), Humic Substances: Nature’s Most Versatile Materials, Proceedings of the 11th Biennial Conference of the International Humic Substances Society and the 6th Humic Substances Seminar, Taylor and Francis, Inc., New York, pp. 3–8 (Boston, MA, July 21–27, 2002). Gaylord, M.C., Brady, L.R., 1971. Comparison of pigments in carpophores and saprophytic cultures of paxillus panuoides and paxillus atrotomentosus. J. Pharm. Sci. 60, 1503. Khairy, A.H., 1980. New thin-layer chromatographic systems for separation and identification of humic substances. Z. Pflanzenernaehr. Bodenkd. 143 (6), 730– 742. Khairy, A.H., 1990. Contribution to the quantification of humic acids by thin layer chromatography. Z. Pflanzenernahr. Bodenk. 153 (3), 133–140. Kumke, M.U., Zwiener, C., Abbt-Braun, G., Frimmel, F.H., 1999. Spectroscopic characterization of fulvic acid fractions of a contaminated groundwater. Acta Hydrochim. Hydrobiol. 27 (6), 409–415. Leenheer, J.A., 1981. Comprehensive approach to preparative isolation and fractionation of dissolved organic carbon from natural waters and wastewaters. Environ. Sci. Technol. 15 (5), 578–587. Lin, C.F., Lee, D.Y., Chen, W.T., Lo, K.S., 1995. Fractionation of fulvic acids: characteristics and complexation with copper. Environ. Pollut. 87, 181–187. Lombardi , A.T., Seritti, A., Morelli, E., 1994. Marine dissolved organic matter. Fractionation by RP-HPLC. In: Humic Subst. Global Environ. Implic. Hum. Health, Proc. Int. Meet. Int. Humic Subst. Soc., sixth, pp. 851–856. Ma, X., Green, S.A., 2004. Photochemical transformation of dissolved organic carbon in lake Superior—an in situ experiment. J. Great Lakes Res. 30 (Suppl.), 97–112. Maie, N., Pisani, O., Jaffé, R., 2006. Fate of Mangrove Tannins in aquatic environments, ASLO 2006 Summer Meeting, Victoria, British Columbia, Canada. Malcolm, R.L., Aiken, G.R., Bowles, E.C., Malcolm, J.D., 1994. Isolation of Fulvic and Humic acids from the Suwannee River, Humic Substances in the Suwannee River, Georgia: Interactions, Properties, and Proposed Structures. US Geological Survey Water-Supply Paper 2373, pp. 13–19. Ohlenbusch, G., Zwiener, C., Meckenstock Rainer, U., Frimmel Fritz, H., 2002. Identification and quantification of polar naphthalene derivatives in contaminated groundwater of a former gas plant site by liquid chromatography–electrospray ionization tandem mass spectrometry. J. Chromatogr., A 967 (2), 201–207. Pompe, S., Heise, K., Nitsche, H., 1996. Capillary electrophoresis for a finger-print characterization of fulvic acids. J. Chromatogr., A 723, 215–218. Rullktter, J., 1993. Organic Geochemistry. In: Engel, M.H., Macko, S.A. (Eds.). Plenum Press, New York, pp. 377–393. Saleh, F.Y., Ong, W.A., 1989. Structural features of aquatic fulvic acids. Analytical and preparative reserved-phase high-performance liquid chromatography separation with photodiode array detection. Anal. Chem. 61, 2792–2800. Santos, M.E., Esteves, V.I., Amado, F.L., Duarte, A.C., 1994. Fractionation and characterization of aquatic organic matter. In: Senesi, N., Miano, T.M. (Eds.), Humic Substances in the Global Environment and Implication on Human Health. Elsevier Science B.V., pp. 877–882. Serkiz, S.M., Perdue, E.M., 1990. Isolation of dissolved organic matter from the Suwannee river using reverse osmosis. Water Res. 24 (7), 911–916. Stedmon, C.A. et al., 2007. Characteristics of dissolved organic matter in Baltic Coastal Sea ice: allochthonous or autochthonous origins? Environ. Sci. Technol. 41 (21), 7273–7279. Sutton, R., Sposito, G., 2005. Molecular structure in soil humic substances: the new view. Environ. Sci. Technol. 39 (23), 9009–9015. Thurman, E.M., Malcolm, R.L., 1981. Preparative isolation of aquatic humic substances. Environ. Sci. Technol. 15 (4), 463–466. Trubetskoj, O.A., Trubetskaya, O.E., Afanas’eva, G.V., Reznikova, O.I., Saiz-Jimenez, C., 1997. Polyacrylamide gel electrophoresis of soil humic acid fractionated by size-exclusion chromatography and ultrafiltration. J. Chromatogr., A 767, 285– 292. Vodacek, A., Blough, N.V., DeGrandpre, M.D., Peltzer, E.T., Nelson, R.K., 1997. Seasonal variation of CDOM and DOC in the middle Atlantic bight: terrestrial inputs and photooxidation. Limnol. Oceanogr. 42 (4), 674–686. Vodacek, A., Green, S.A., Blough, N.V., 1994. An experimental model of the solarstimulated fluorescence of chromophoric dissolved organic matter. Limnol. Oceanogr. 39 (1), 1–11.

1434

X. Ma, S.A. Green / Chemosphere 72 (2008) 1425–1434

Wander, M.M., Traina, S.J., 1996. Organic matter fractions from organically and conventionally managed soils: II. Characterization of composition. Soil Sci. Soc. Am. J. 60, 1087–1094. Woelki, G., Friedrich, S., Hanschmann, G., Salzer, R., 1997. HPLC fractionation and structural dynamics of humic acids. Fresenius J. Anal. Chem. 357, 548–552. Wolfbeis, O.S., 1985. The Fluorescence of Organic Natural Products. In: Schulman, S.G. (Ed.), Molecular luminescence spectroscopy, Part 1. Methods and Applications. Chemical Analysis. John Wiley & Sons, New York, pp. 167–370. Wu, F., Tanoue, E., 2001. Isolation and partial characterization of dissolved coppercomplexing ligands in streamwaters. Environ. Sci. Technol. 35 (18), 3646– 3652. Zanardi-Lamardo, E., Clark, C.D., Moore, C.A., Zika, R.G., 2002. Comparison of the molecular mass and optical properties of colored dissolved organic material in

two rivers and coastal waters by flow field-flow fractionation. Environ. Sci. Technol. 36 (13), 2806–2814. Zepp, R.G., Schlotzhauer, P.F., 1981. Comparison of photochemical behavior of various humic substances in water: III. Spectroscopic properties of humic substances. Chemosphere 10 (5), 479–486. Zwiener, C., Frimmel, F.H., 2004. LC–MS analysis in the aquatic environment and in water treatment technology – a critical review. Part II: Applications for emerging contaminants and related pollutants, microorganisms and humic acids. Anal. Bioanal. Chem. 378 (4), 862–874. Zwiener, C., Kumke, M.U., Abbt-Braun, G., Frimmel, F.H., 1999. Adsorbed and bound residues in fulvic acid fractions of a contaminated groundwater. Isolation, chromatographic, and spectroscopic characterization. Acta Hydrochim. Hydrobiol. 27 (4), 208–213.