Yellow and red metal oxides are commonly found on deteriorated water pipes and it has been reported that their main component is iron. X-ray absorption fine structure (XAFS) spectroscopy is an effective method which can be used to measure the chemical form of metals. However, when the iron concentration in water samples is extremely low, XAFS measurement and analysis can be difficult due to so much noise in the spectra. In this study, we examined various conditions related to XAFS measurements to analyze the chemical form of iron in water taken from water pipes. To optimize the XAFS measurement conditions, the energy range in pattern fitting was set at 7,100–7,140 eV. Piling up multiple pieces of filter paper was effective for improving the accuracy in the analysis of the XAFS spectra. The XAFS measurement of actual water samples was conducted and a low R factor was obtained from pattern fitting analysis. Differences were observed in the XAFS spectra depending on the chemical form of iron. These results showed that the chemical form of the iron of the actual water samples taken from the water pipes was able to be detected from the XAFS measurement.

INTRODUCTION

In Japan, many water pipes were installed during the 1970s and most of these will need to be upgraded at approximately the same time. Therefore, management of the operation will become important in the drinking water distribution system in the future. Determination of pharmaceuticals in water sources and drinking water has been conducted to investigate the probability of contamination in the water (Simazaki et al. 2014). In particular, the establishment of the means to prevent further deterioration and techniques for diagnosing deterioration are urgent problems for the near future. Fujita et al. (2014) suggested that analysis of the elemental composition of suspended particles can detect aging pipes in water distribution systems. On the other hand, yellow and red metal oxides are commonly found on deteriorated water pipes, and it has been reported that their main component is iron (Peng et al. 2010; Ishiwatari et al. 2013; Fujita et al. 2014). Lasheen et al. (2008) reported that differences in iron release were found among the materials of pipes and pipe ages. Many researchers have tried to measure the chemical form of iron in water pipes. According to Peng et al. (2010), who examined corrosion scales from unlined cast iron pipes and deposits mobilized during hydrant flushing events in drinking water distribution systems, siderite (FeCO3), magnetite (Fe3O4), and goethite (αFeOOH) were identified in drinking water. According to Sarin et al. (2001), who examined powdered samples from the entire tubercle from multiple water pipes, Fe3O4, αFeOOH, and lepidocrocite (γFeOOH) were identified as the main components. According to Sarin et al. (2004a), Fe3O4, αFeOOH, and maghemite (γFe2O3) were observed in a shell-like, dense layer near the top of the corrosion scales, while high concentrations of readily soluble Fe(II) content were observed inside the scales. Barkatt et al. (2009) reported that suspended solids included Fe3O4, αFeOOH, and γFeOOH in tap water from cold water taps in a 90-year-old building. Although there is much information on the chemical form of iron present in pipes, X-ray diffraction was the primary method of analysis, in which samples had to be dried in the pretreatment. Moreover, the measurement target was limited to the corroded pipes and there were no examples of direct measurement of the form of iron in wet conditions in water samples from these pipes.

X-ray absorption fine structure (XAFS) spectroscopy is another method available to measure the chemical form of metals. Samples are irradiated with X-rays to obtain absorption spectra near an absorption edge in the XAFS measurement. By analyzing these spectra, information such as the oxidation state of the atom can be obtained. One of the advantages of this method is that liquid and non-crystalline materials can be analyzed. In addition, sensitivity is especially high in the fluorescence method, which is beneficial when the concentration of the target material is low. Therefore, XAFS measurement of a water sample is possible despite the fact that the sample is a liquid and that the concentration of the target materials is low. XAFS can be divided into X-ray absorption near edge structure (XANES), which targets the spectrum at the absorption edge, around ± 50 eV, and extended X-ray absorption fine structure, which targets the spectrum beyond XANES (on the high-energy side). By comparing the XANES spectrum of the sample with those of standard reference materials containing iron with different oxidation numbers, such as FeCO3 and αFe2O3, proportions of different forms of iron in the sample can be determined. Szabo et al. (2009) analyzed the chemical form of cobalt by XAFS measurement in a model drinking water distribution system. However, there have been no reports in which XAFS measurement is applied to analysis of the chemical form of iron in the water taken from an actual water pipe.

When the iron concentration in a water sample is extremely low, XAFS measurement and analysis can be difficult due to so much noise in the spectra. Each iron state has its own specific XAFS spectra around the energy range of the absorption edge. However, the optimum analysis range for pattern fitting with XAFS spectra for samples has not yet been clarified. Thus, it is essential to develop a method that ensures the reliability of the obtained results with XAFS measurement. The aim of this study is to clarify the conditions under which XAFS measurement will be applied to analyze the chemical form of iron. In this study, standard reference materials and synthetic mixed samples were analyzed using XAFS measurements to establish the practical utility of this method in determining the form of iron in water samples. From these results, the target energy range for optimum pattern fitting, as well as the lower limit of analysis for XAFS were investigated. Moreover, XAFS measurement against actual water samples was performed and the potential for measuring the chemical form of iron in water pipes was considered.

MATERIALS AND METHODS

Conditions of XAFS measurements

XAFS measurements were performed at the Synchrotron Radiation Center of Ritsumeikan University, Japan to characterize the chemical form of iron in standard reference materials and collected water samples. It is possible to analyze the peculiar chemical state of iron using this XAFS measurement without inhibition by other elements. The Fe K-edge XAFS spectra were measured using BL-3 at the above-mentioned facility, which was equipped with a Si (220) double-crystal monochromator (Ishiwatari et al. 2013). FeCO3 (BOC Sciences), Fe3O4 (Wako Pure Chemical Industry), αFeOOH (Alfa Aesar), γFeOOH (Alfa Aesar), and αFe2O3 (Wako Pure Chemical Industry) were measured as standard reference materials for measurement in transmission mode using an ionization chamber filled with mixed gases: Ar 15% and N2 85% for the incident chamber, and Ar 50% and N2 50% for the transmission chamber (Takeuchi et al. 2012). Fluorescence yield mode was used for water sample analysis (Asaoka et al. 2012). A three-element Ge solid-state detector (Canberra) was used for the fluorescence detector.

For all samples, the obtained spectra were normalized to 7,300 eV. Pre-edge background contribution was subtracted from the raw XAFS spectra obtained in fluorescence yield mode by removing a constant extrapolated from the pre-edge region. The Victoreen extrapolation with a constant was used for background subtraction for XAFS spectra obtained in transmission mode. The commercially provided software (REX2000 ver. 2.5.93; Rigaku Co., Ltd), which has often been used for the analysis of XAFS spectra (Kodama et al. 2006; Takahashi et al. 2013; Moroki et al. 2014), was also used in this study. The R factors (%) were calculated based on a least squares fitting method using the following equation, where Iobs is the observed intensity and Ical is the calculated intensity with the standard reference materials in the XAFS spectra: 
formula

Examination of the fitting condition

The energy range used for analysis was varied in the pattern fitting between obtained spectra from the standard reference materials (FeCO3, Fe3O4, αFeOOH, γFeOOH, and αFe2O3), so that the optimum range could be determined. Next, Fe3O4, αFeOOH, and αFe2O3 were selected as typical iron compounds found in water pipes and were mixed at eight known ratios to prepare the synthetic mixed samples for the XAFS measurements, and spectra of the synthetic mixed samples were fitted using the known ratios of the standard samples. Then, R factors were calculated to verify the accuracy of pattern fitting. Additionally, αFe2O3 was added to the pure water and mixed completely. A certain amount was filtered in order to collect αFe2O3 on the filter paper. Through these procedures, the mixed standard samples containing known amounts of iron were prepared on the filter paper and the XAFS measurement of these samples were conducted by fluorescence yield mode. The obtained spectra of samples were fitted with the αFe2O3 spectrum as a standard reference sample. R factors were also calculated to provide information about the minimum limit of detection with this procedure.

XAFS measurement of the water samples

Water samples were taken directly from a water treatment plant, water distribution plants, and water pipes in Hitachi city, Ibaraki Prefecture. Materials and service ages of sampled water pipes are shown in Table 1. WTP1 and WTP2 were the same water treatment plant, but the date of water sampling was different. Approximately 20 L of the water sample was filtered using a membrane filter with a pore size of 0.45 μm and suspended materials were collected on the filter. Iron collected on this filter was measured with XAFS. Since iron can be oxidized easily, to maintain the chemical form of iron, the filtering process was completed as quickly as possible at the time of water sample collection. The filter papers on which iron was collected via the filtration were immediately water-sealed with sample water to avoid changes in chemical oxidation state. The component ratio of iron in the sample was estimated by pattern fitting the spectrum of the sample with the spectra of the standard reference materials: Fe3O4, αFeOOH, and αFe2O3. For the pattern fitting, the conditions which were optimized above were used. It was assumed that the R factor was minimized and the calculated composition ratio of the standard substance did not become negative in the pattern fitting. The suspended iron concentrations of the collected water samples were measured with inductively coupled plasma atomic emission spectroscopy (ICP-AES) after filtration using acid according to Ishiwatari et al. (2013).

Table 1

Conditions of the sampled water pipes

  Material
 
  
Sampling site Pipe Lining Service age (years) 
WTP1 
st.1 Ductile iron Mortar 41 
st.2 Ductile iron Mortar 39 
st.3 Cast iron Resin 28 
st.4 Cast iron Resin 28 
WTP2 
st.5 Cast iron Resin 29 
st.6 Cast iron Resin 29 
st.7 Cast iron Resin 0.5 
st.8 Steel – 53 
  Material
 
  
Sampling site Pipe Lining Service age (years) 
WTP1 
st.1 Ductile iron Mortar 41 
st.2 Ductile iron Mortar 39 
st.3 Cast iron Resin 28 
st.4 Cast iron Resin 28 
WTP2 
st.5 Cast iron Resin 29 
st.6 Cast iron Resin 29 
st.7 Cast iron Resin 0.5 
st.8 Steel – 53 

RESULTS AND DISCUSSION

Optimization of the analysis range

The Fe K-edge XAFS spectra of the standard reference materials are shown in Figure 1. The absorption edge was approximately 7,110 to 7,120 eV for all the standard reference materials and a rapid increase in intensity was noted around the absorption edge. Divalent FeCO3 had an absorption edge at a lower energy compared to other standard reference materials, and a second peak was found between 7,130 and 7,140 eV. FeCO3, Fe3O4, and αFe2O3 showed distinct characteristic spectra; however, the spectra of αFeOOH and γFeOOH were similar to each other. If the spectra of standard reference materials are similar, the pattern fitting of the spectrum of an unknown sample would be difficult, therefore it is desirable to utilize the energy range where the differences in the spectra of the standard reference materials are most significant.
Figure 1

Fe K-edge XAFS spectra of standard reference materials.

Figure 1

Fe K-edge XAFS spectra of standard reference materials.

Spectra of αFeOOH (Figure 2(a)) and αFe2O3 (Figure 2(b)), which were measured as the standard reference materials, were fitted with spectra of other standard reference materials. The analysis range was fixed with the minimum photon energy at 7,100 eV and the maximum photon energy was staggered by 10 eV intervals between 7,100 and 7,200 eV to calculate the R factor for each range, so that the differences in the spectra of the standard reference materials could be evaluated. The relationships between the maximum photon energy in the analysis range and the R factors are shown in Figure 2. When pattern fitting was done with FeCO3, the R factors were at least 6.8% and much higher than those of the other cases, with the result that FeCO3 is excluded from Figure 2. In Figure 2(a), αFeOOH is targeted. When Fe3O4 and αFe2O3 were used as explanatory reference materials, the R factor slowly increased, then reached a maximum at 7,130 eV, and finally, slowly decreased. This indicates that in the 7,100 to 7,130 eV range, the difference between spectra of standard reference materials is most significant. However, FeCO3 demonstrated a special characteristic in the second peak between 7,130 and 7,140 eV. Therefore, when the analysis range was set up to 7,130 eV, definite identification of FeCO3 would be difficult. In addition, if γFeOOH was used as an explanatory reference material, the R factor remained low. It was clear that αFeOOH and γFeOOH spectra were similar, therefore, the identification of αFeOOH and γFeOOH was more difficult compared to identification of other standard reference materials. In Figure 2(b), αFe2O3 is targeted. The R factor was relatively high at 7,130 eV and showed a decreasing trend above 7,140 eV. These findings show that iron ratios of these standard reference materials could be evaluated clearly by pattern fitting in the energy range of 7,100 to 7,140 eV. In a previous report, the analysis range was set from 150 below to 1,500 eV above the adsorption K edge of iron of 7,112 eV (Liu et al. 2013). The analysis range of 7,100 to 7,140 eV for pattern fitting in this study was narrower than this range but it was similar to the range of 7,110 to 7,150 eV which was reported by Takahashi et al. (2013). This analysis range was thought to improve the accuracy of the pattern fitting analysis.
Figure 2

Calculated R factors between standard reference materials ((a) αFeOOH was targeted; (b) αFe2O3 was targeted).

Figure 2

Calculated R factors between standard reference materials ((a) αFeOOH was targeted; (b) αFe2O3 was targeted).

Verification of pattern fitting and minimum limit of detection

Ratios of iron in the mixed standard samples are shown in Table 2. Spectra of mixed standard samples obtained from the XAFS measurement were fitted with the known ratio of each standard reference material and R factors were calculated. Results are shown in Table 2. The minimum R factor was 0.004%, the maximum value was 0.069%, and the average was 0.031%. These R factor values can be used to validate the pattern fitting. In this study, when a target environmental sample was fitted with a standard reference material, an R factor of less than 0.069% was thought to be required.

Table 2

Mixed ratio of mixed standard material samples and R factor values

Mixed ratio (%)
 
  
Fe3O4 αFeOOH αFe2O3 Calculated R factor (%) 
80.5 0.0 19.5 0.004 
51.4 8.9 39.7 0.069 
0.0 9.1 90.9 0.061 
0.0 47.3 52.7 0.010 
53.5 46.5 0.0 0.017 
50.8 0.0 49.2 0.043 
27.0 46.9 26.1 0.035 
35.3 30.6 34.1 0.006 
Mixed ratio (%)
 
  
Fe3O4 αFeOOH αFe2O3 Calculated R factor (%) 
80.5 0.0 19.5 0.004 
51.4 8.9 39.7 0.069 
0.0 9.1 90.9 0.061 
0.0 47.3 52.7 0.010 
53.5 46.5 0.0 0.017 
50.8 0.0 49.2 0.043 
27.0 46.9 26.1 0.035 
35.3 30.6 34.1 0.006 

XAFS measurement of samples in which αFe2O3 was collected on filter paper was conducted. The R factors calculated from pattern fitting with the spectrum of αFe2O3 are shown in Figure 3. Significant noise was present in the spectrum when the amount of collected iron in the filter was low. When the weight of added iron on the filter paper increased, the R factor decreased. The R factor was 0.007% in the case where the weight of iron was 2 mg. The decrease of the R factor suggested that the accuracy of the pattern fitting was improved due to the increase of collected iron on the filter paper. Figure 3 also shows the results of collecting 0.1 mg of iron on filter paper, cutting this paper into four pieces and piling them up. When the papers were not piled up, the R factor was 0.158%; however, with four pieces piled up, the R factor became 0.046% and lower than 0.069%, the value shown above. Therefore, at least 0.4 (0.1 × 4) to 0.5 mg of iron on the filter paper (0.003 to 0.004 mg Fe/mm2) would be required as the lower-limit level for accurate pattern fitting. From these results, it was helpful to pile up multiple pieces of filter paper for samples with extremely low iron concentrations. It is desirable to collect as much iron as possible for getting high intensity and to remove moisture content for preventing absorption of X-rays by water during XAFS measurement of wet samples in actual practice.
Figure 3

R factors of αFe2O3 collected on filter paper.

Figure 3

R factors of αFe2O3 collected on filter paper.

XAFS measurement of actual samples

Results of XAFS measurement of actual water samples and the iron concentrations are shown in Figure 4. In particular, WTP1 and WTP2 from the water treatment plant had extremely low iron concentrations and the spectrum noise was significant. The measurement of this sample was conducted three times and the obtained intensity was averaged to reduce the noise. The peak in the Fe K-edge XAFS spectra was obtained in all samples, however the XAFS spectra from all water samples had more noise than the spectra of the standard reference materials. There were slight variations in the position of the absorption edge and the shape of each spectrum. By pattern fitting of the spectrum of the sample with the spectra of standard reference materials, Fe3O4, αFeOOH, and αFe2O3, the component ratio of the chemical form of iron in the sample was estimated. Analysis in the case where FeCO3 was added as the standard reference material showed the ratio of FeCO3 was low in all samples. A maximum FeCO3 ratio of only 4% was obtained in st.8. Therefore, it was suggested that FeCO3 had an extremely low concentration in all water samples. This result agreed with previous reports in which the fraction of FeCO3 was small in iron corrosion (Sarin et al. 2004a, 2004b; Lee et al. 2008; Świetlik et al. 2012). Thus, in this analysis, FeCO3 was omitted from the explanatory reference materials in the pattern fitting.
Figure 4

Fe K-edge XAFS spectra and iron concentration of water samples.

Figure 4

Fe K-edge XAFS spectra and iron concentration of water samples.

Ratios of iron formed in water samples obtained by pattern fitting are shown in Figure 5. In WTP1, WTP2, and st.7, the concentration of iron was low and much noise was found. The weight of iron on the filter paper was much lower than 0.003 mg Fe/mm2 and R factors exceeded 0.069%. These results indicated that an accurate analysis could not be obtained. For other samples, R factors were lower than 0.069% and accurate pattern fitting results were obtained. For samples from st.1 and st.2 from mortar-lined iron pipes, the ratio of Fe3O4 was low, thus, these samples were thought to contain primarily trivalent iron. In addition, in st.4, most of the iron in the samples was thought to be αFeOOH, while the ratio of αFeOOH was thought to be low in st.8. The fact that αFeOOH was dominant in most samples was consistent with a previous report (Ishiwatari et al. 2013). In this way, differences in the chemical form of iron were observed among the different types of water pipes. According to Wang et al. (2012), who examined the effects of disinfectant and biofilms on the corrosion of cast iron pipes, αFeOOH and Fe2O3 played a major role in the corrosion of annular reactors. In addition, αFeOOH was the predominant species during the initial stages without disinfection, while CaCO3 and αFeOOH became predominant with increasing time. Świetlik et al. (2012) pointed out the importance of green rusts inside water pipes, and reported that Fe3O4, αFeOOH, γFeOOH, and low amounts of FeCO3 and quartz were found in corrosion scales and suspended deposits. The majority of previous results corresponded to the results obtained in this study. These results showed that the chemical form of iron in actual water samples taken from water pipes was able to be directly detected in wet condition from XAFS measurement.
Figure 5

Ratio of the calculated chemical form of iron by pattern fitting.

Figure 5

Ratio of the calculated chemical form of iron by pattern fitting.

CONCLUSION

In this study, we examined various conditions related to XAFS measurements to analyze the chemical form of iron in water taken from actual water pipes. To optimize measurement conditions, the energy range in the pattern fitting was set from 7,100 to 7,140 eV. Piling up multiple pieces of filter paper was effective in improving the accuracy in analysis of the XAFS spectra for samples with extremely low iron concentration. The XAFS measurement of actual water samples was conducted and a low R factor was obtained from pattern fitting. Since differences were observed in the XAFS spectra depending on the chemical form of iron, the measurement and analysis were thought to have been effectively conducted. These results showed that the chemical form of iron in the actual water samples taken from the water pipes was able to be detected from the XAFS measurement.

ACKNOWLEDGEMENTS

The authors thank members of Hitachi City for arranging the fieldwork and providing information on the drinking water supply system, and Hiromu Tsugane, Ryutaro Inoue, Tohru Kuroha, Shinobu Chinone, and Yoshio Hosaka for helping with sampling, analytical support and technical discussion. The authors also thank Dr Misaki Katayama at Ritsumeikan University for his cooperation in the XAFS measurement. This research was supported by JST A-STEP (12101523, 11100721, and 10101737).

REFERENCES

REFERENCES
Asaoka
S.
Hayakawa
S.
Kim
K. H.
Takeda
K.
Katayama
M.
Yamamoto
T.
2012
Combined adsorption and oxidation mechanisms of hydrogen sulfide on granulated coal ash
.
J. Colloid Interface Sci.
377
,
284
290
.
Barkatt
A.
Pulvirenti
A. L.
Adel-Hadadi
M. A.
Viragh
C.
Senftle
F. E.
Thorpe
A. N.
Grant
J. R.
2009
Composition and particle size of superparamagnetic corrosion products in tap water
.
Water Res.
43
(
13
),
3319
3325
.
Ishiwatari
Y.
Mishima
I.
Utsuno
N.
Fujita
M.
2013
Diagnosis of the ageing of water pipe systems by water quality and structure of iron corrosion in supplied water
.
Water Sci. Technol: Water Supply
13
(
1
),
178
183
.
Lasheen
M. R.
Sharaby
C. M.
El-Kholy
N. G.
Elsherif
I. Y.
El-Wakeel
S. T.
2008
Factors influencing lead and iron release from some Egyptian drinking water pipes
.
J. Hazard. Mater.
160
(
2–3
),
675
680
.
Lee
J. Y.
Pearson
C. R.
Hozalski
R. M.
Arnold
W. A.
2008
Degradation of trichloronitromethane by iron water main corrosion products
.
Water Res.
42
(
8–9
),
2043
2050
.
Liu
H.
Schonberger
K. D.
Peng
C. Y.
Ferguson
J. F.
Desormeaux
E.
Meyerhofer
P.
Luckenbach
H.
Korshin
G. V.
2013
Effects of blending of desalinated and conventionally treated surface water on iron corrosion and its release from corroding surfaces and pre-existing scales
.
Water Res.
47
(
11
),
3817
3826
.
Moroki
T.
Yasui
H.
Adachi
Y.
Yoshizawa
K.
Tsubura
A.
Ozutsumi
K.
Katayama
M.
Yoshikawa
Y.
2014
New insulin-mimetic and hypoglycemic hetero-binuclear zinc(II)/oxovanadium(IV) complex
.
Curr. Inorg. Chem.
4
(
1
),
54
58
.
Peng
C. Y.
Korshin
G. V.
Valentine
R. L.
Hill
A. S.
Friedman
M. J.
Reiber
S. H.
2010
Characterization of elemental and structural composition of corrosion scales and deposits formed in drinking water distribution systems
.
Water Res.
44
(
15
),
4570
4580
.
Sarin
P.
Snoeyink
V. L.
Bebee
J.
Kriven
W. M.
Clement
J. A.
2001
Physico-chemical characteristics of corrosion scales in old iron pipes
.
Water Res.
35
(
12
),
2961
2969
.
Sarin
P.
Snoeyink
V. L.
Bebee
J.
Jim
K. K.
Beckett
M. A.
Kriven
W. M.
Clement
J. A.
2004a
Iron release from corroded iron pipes in drinking water distribution systems: effect of dissolved oxygen
.
Water Res.
38
(
5
),
1259
1269
.
Sarin
P.
Snoeyink
V. L.
Lytle
D. A.
Kriven
W. M.
2004b
Iron corrosion scales: model for scale growth, iron release, and colored water formation
.
J. Environ. Eng.
130
(
4
),
364
373
.
Simazaki
D.
Hiramatsu
S.
Fujiwara
J.
Akiba
M.
Kunikane
S.
2014
Monitoring priority of residual pharmaceuticals in water sources and drinking water in Japan
.
J. Water Environ. Technol.
12
(
3
),
275
283
.
Świetlik
J.
Raczyk-Stanislawiak
U.
Piszora
P.
Nawrocki
J.
2012
Corrosion in drinking water pipes: the importance of green rusts
.
Water Res.
46
(
1
),
1
10
.
Szabo
J. G.
Impellitteri
C. A.
Govindaswamy
S.
Hall
J. S.
2009
Persistence and decontamination of surrogate radioisotopes in a model drinking water distribution system
.
Water Res.
43
(
20
),
5005
5014
.
Takeuchi
T.
Kageyama
H.
Nakanishi
K.
Inaba
Y.
Katayama
M.
Ohta
T.
Senoh
H.
Sakaebe
H.
Sakai
T.
Tatsumi
K.
Kobayashi
H.
2012
Improvement of cycle capability of FeS2 positive electrode by forming composites with Li2S for ambient temperature lithium batteries
.
J. Electrochem. Soc.
159
(
2
),
A75
A84
.