3.2.9.4. Determination of Suspended Matter in Sea and Drinking Water

We shall compare the performance of flow nephelometry with stop flow turbidimetry by analyzing water samples collected from different sources. This is a critical test because the instrument is calibrated by a homogenous, stabilized suspension of a standard material of well known chemical composition, while Suspended Matter in sea water contains particles of sand, clay, and organic matter of different sizes and optical properties.

The turbidity of sea water changes considerably between the open ocean where it can be as low as 0.05 NTU and as high as 60 NTU in an estuary if sediment is stirred by tidal activity. In coastal waters, the turbidity increases during the rainy season: in Hawaii, it may be as high as 5 NTU due to erosion of soil from the mountainside. Freshwater in clear rivers and lakes is about 10 NTU, and groundwater is as low as 2 NTU. Drinking water should be below 1 NTU (EPA) and never more than 4 NTU (WHO). To accommodate such a wide range of concentrations, we prepared calibrations in 0 to 20 NTU (P) and 0 to 2 NTU ranges and analyzed samples collected in Hawaii Kai Marina (U) and water supply in Honolulu while using ‘Analysis’ software (Appendix 2).

Suspended Matter concentrations determined by nephelometry (V left) and turbidimetry (right) follow a similar pattern: higher turbidity in Marina, which is heavily silted, and lower turbidity outside Marina by the bridge, where sea water from the bay dilutes water from the Marina at flood. The turbidity at Dock is high because a culvert in its vicinity brings in muddy water from the mountainside.  (Response to drinking water sample (TAP) is included to show baseline – note absence of the schlieren effect).

The unexpected, striking difference between large fluctuations of nephelometric (V left), and well stabilized turbidimetric data (V right) is due to a large difference in volumes of interrogated suspensions and physical properties of monotored particles. In turbidimetry, the illuminated and monitored volume is 50× larger than that for nephelometry, while the light path of 10 cm is 100× longer (B right) than of a nephelometric flow cell with orthogonal illumination (B, left). Therefore, in turbidimetry, the beam, passing through the vertical column, interrogates a larger volume of particles and integrates their individual responses, which vary due to their size, shape, and optical density, while the sample is arrested in the flow cell. The result is a smooth response and increase in sensitivity with increase of flow cell length. In contrast, in nephelometry mode, a stream of heterogenous particles reflects light, which is monitored as large fluctuations in signal amplitude are scanned by a 1 μL flow cell and monitored as large fluctuations in the signal of amplitude scanned by the flow cell, which monitors 50x less particles as they fly by. Therefore, FI nephelometry fails to analyze Suspended Matter (V left) because monitored particles differ so much in size and shape that the reflected light is no longer statistically smoothed, while the response of the same Suspended Matter by FI turbidimetry is reproducible. (V right).

The inevitable conclusion is that particles of sediment passing through the flow cell reflect light in flow nephelometry differently than particles of Formazin (compare V left with I) while in flow turbidimetry, responses to Formazin and sediment are similar (compare V right with S). This observation argues against the use of flow nephelometry for monitoring the turbidity of sediments because there's no way how to increase the illuminated volume of the flow cell constructed by an orthogonal configuration of optical fibers.

Next must be shown that Formazin suspended particles scatter away the same flux of light as particles of Suspended Matter. This is confirmed in two ways: 

1) by comparing spectra of Formazin with spectra of Suspended Matter recorded at a wide range of concentrations.

2) by correlation of slopes of calibration lines of Formazin with calibration line obtained by dilution of particles of Suspended Mater.

Spectra of Formazin in DI (N) and in SSW (R) are the same as of unfiltered SM spectra in SSW (X). 

Calibrations of Formazin in DI (P) and in SSW (T) and in DI for drinking water (AB) correlate well in terms of linearity slope and reproducibility over ranges 0 to 2 and 0 to 20 NTU.

Calibration of Formazin (AA) correlates within 2% with Suspended Matter in Sea Water with excellent reproducibility and linearity.

Furthermore, autodilution of SM particles also results in a linear calibration line. We used a 0.5-μm filter to prepare from sea water the 1.5 NTU Suspended Matter filtered sample (see Y dock) and used it to generate a calibration graph (Q right). Surprisingly linear, slope of 0.02 NTU/ 10cm @550nm is the same as that of Formazin (Q left), and as all other calibrations in this work. The poor regression and reproducibility of individual dilutions of SM compared to Formazin of a similar concentration range is due to differences in light scattering by fever and far more heterogonous SM particles than those of Formazin.

TAKEOUT

There is a close relationship between slopes and LINEAR calibration lines of Formazin particles and Suspended Matter from HK Marina. This finding is supported by the close similarity of turbidity spectra of these suspensions, where uniform absorbance/concentration lines gradually decrease in the range of 500 nm to 900 nm due to scatter of light.
Considering the vast difference in composition of Formazin and of Suspended Matter in SW from HK, this finding is remarkable because it indicates that scatter of light in heterogeneous suspensions of vastly different composition causes the logarithmic decrease of light intensity (Lambert) and that Formazin can serve as a universal calibration standard for analysis of suspensions of different compositions. Obviously, the ranges and limits of Beer’s Law must be investigated to establish a broader scope of this claim.
 

Filtration and particle size

Prior to chemical analysis sea water is routinely filtered through 0.5 μm filter that allows soluble species and colloids to pass into a filtrate. To gain insight into effect of this sample treatment, we collected three SW samples (not the same as the samples analyzed above) selected to encompass a wide NTU range and analyzed them unfiltered and filtered (Y). Surprisingly while the unfiltered Dock samples contained 23.9, 9.8, and 4.9 NTU, the filtered Dock samples contained the same concentration of (colloidal ?) particles (1.6 NTU blue lines). Therefore, the mass of Suspended Matter in filtered samples of sea water from HK Marina is 1.6 ppm (Appendix 1).

These findings open Pandora’s Box of uncomfortable questions: what are the effects of sample filtering? Is it necessary to filter, and if so, is a 0.5 μm sufficient cutoff when it leaves as much as 1.6 NTU solids in filtrate? Considering vast number of samples of sea water, presumably filtered manually each day on a cruise or in the laboratory, screening of NTU level should be considered because it will identify large SM blank or other errors due to high NTU, while it might on the other hand, eliminate the need to filter samples with NTU below 0.5 equal to A= 0.01A (Q) with 10cm long LP flow cell. However, without question, when NTU 5 or higher is encountered, the value of LOD obtained by spectrophotometry should be revised. Sample screening for NTU level can easily be done since the same instrument (zz) furnished with a 20 cm long cell as used in this work has been used to automate analysis of nutrients and trace elements in sea water. Determination of phosphate, silicate, nitrate, and nitrite.

Analysis of drinking water

Since the quality of potable water is evaluated by measuring suspended matter in the 0 to 2 NTU range, we ran a calibration protocol (L) by using a 2.0 standard prepared in DI. The resulting outputs: 1/ absorbance vs. time (AB left) and 2/ calibration graph (AB right) are similar to data in the range 0 to 20 NTU, with strictly linear calibration response. However, the higher values of LOD and BLANK (AB) compared to those obtained in 0 to 20 NTU (P) are likely due to the varying quality of the daily output of our DI source (avoidable by additional filtration). The calibration (AB right) was used with the software protocol in the (Appendix 2) to analyze the following samples of drinking water: 1) Tap water UH, 2) Tap water from home, 3) PELLEGRINO^® bottled water, 4) Laboratory DI water, 5) Drinking water fountain UH, and 6) SMART^® bottled water. Results (AD) show that unfiltered tap water did not meet the strict EPA standard, but the filtered tap (UH Font.) and all bottled waters did. Interestingly, PELLEGRINO® is carbonated mineral water, while SMART^® is tap water, distilled with taste adjusted by the addition of inorganic salts and by carbonation.

TAKEOUT

Flow Injection turbidimetry covers a range of 0.1 to 20NTU (0.001 A to 0.4 A) in a strictly linear calibration (AB right), which can be extended 5x by increasing the flow cell length from 10 cm up to 50cm. This makes turbidimetry as sensitive as nephelometry and applicable to the analysis of potable and sea water. 
 

[HOME]

[PREVIOUS PAGE]      [NEXT PAGE]