A FIELD-BASED SPECTROPHOTOMETRIC FLOW-INJECTION SYSTEM FOR AUTOMATIC DETERMINATION OF CHLORIDE IN SOIL WATER N I C H O L A S M. HOLDENI, * JOHN E D O W D 2, A N D R E W G. W I L L I A M S 1 and DAVID S C H O L E F I E L D 3
IDepartment of Geographical Sciences, University of Plymouth, Plymouth, PL4 8AA, U.K.* 2Department of Geology, University of Georgia, Athens, Georgia, 30602, USA. 3Institute of Grassland and Environmental Research, North Wyke, Okehampton, Devon, EX20 2SB, U.K. (Received: October 1994; revised: February 1995)
Abstract. The use of chloride as a tracer for soil water investigations is discussed. Limitations with
laboratory based soil core and field based sampling are considered with respect to the poor suitability of the data for rigorous assessment of mechanistic models. Investigation of water movement in soil has been restricted by limitations in spatial and temporal sampling. Fine resolution sampling generates large numbers of samples which cause problems with post sampling laboratory analysis. This paper describes a field-based system for the analysis of chloride in soil water. There are three component parts to the system, (i) a custom sampling sub-system comprising of ceramic cup suction samplers and sample traps, (ii) a sample routing sub-system to channel sample to (iii) a sample analysis sub-system utilizing a flow injection method for sample analysis using a custom built photo-diode detector. The three sub-systems were controlled by a suitably equipped personal computer. A calibration procedure is described with a third order polynomial equation derived to convert millivolt response from the detector into chloride concentration. Field and laboratory data from a tracer experiment are presented and discussed, and it is concluded that the system is well suited to field-based applications. Finally it is noted that the photo-detector is suitable for colourimetric analysis of any tracer with suitable chemical determination.
1. Introduction Environmental pollution due to agricultural practises is of great interest to environmental scientists. Water movement through soils, and mass transport of chemicals is of particular importance, with much research being devoted to investigation of nitrate (Addiscott et al, 1991) and pesticide behaviour (Ftihr & Hance, 1992). To monitor water movement, tracers are used which provide indication (by means of concentrations) of rates of flow and interactions of new water (that applied in an 'event') with old water (that residing in the soil prior to an 'event'). A conservative tracer which has little chemical reaction in the soil is best used for this type of work so that tracer concentrations are not inadvertently influenced by interaction with colloids or chemical species in the soil. Chloride is typically utilized for this purpose and has been extensively used in the past (Addiscott et al, 1978; Tyler & Thomas, 1981; Jarvis et al, 1991; Roth et al, 1991). Chloride is favoured because * Address for correspondence: Institute of Grassland and Environmental Research, North Wyke, Okehampton, Devon, EX20 2SB, UK
Environmental Monitoring and Assessment 36: 217-228, 1995. (~) 1995 Kluwer Academic Publishers. Printed in the Netherlands.
218
NICHOLAS M. HOLDEN ET AL.
it is cheap, can be determined reliably by wet chemistry, and can be applied to soil in the field without constituting a serious pollution effect. Laboratory based investigations using small soil cores may make possible intensive sampling regimes (Marshall, 1994) but are restricted in the spatial information they can provide, while field based experiments tend to have sampling rates in the order of once per day which provide insufficient data to characterize rapid flow events (Roth eta/, 1991). This has meant that the scope of field investigations of soil water movement have tended to be limited due to spatial and temporal resolutions of sampling being too coarse to provide data suitable for testing of mechanistic models in a rigorous manner (Hosang, 1993). Soil water is usually extracted from the soil using ceramic cups with a negative pressure applied internally, and a means of extracting water from the cup available at the soil surface (Webster et al, 1993). Traditionally such water samples are stored and analysed in the laboratory with the associated risk of mislabelling, cross-contamination and chemical and biological degradation. When monitoring water movement through the soil, ceramic cups are placed at a number of depths down the soil profile, a tracer is applied to the soil surface, and its migration through the profile monitored by means of concentration/time relationships obtained from each suction sampler. When a large number of ceramic cups are used to characterize a field site, fine temporal resolution sampling generates large numbers of samples. This situation can be accommodated in various of ways: (i) by having a number of technicians collecting and storing samples during the experiment with later analysis; (ii) by having an automatic sample collection and storage system that permits later analysis, and (iii) by having automatic sampling and field based analysis. This paper describes a field based method of sample chloride analysis using a low-cost spectrophotometric detector, and a routing system that permits multiple samplers to be connected to a single detector. The methodology may be adapted for other tracers for which suitable analytical methods are available.
2. Instrumentation
The equipment developed for in situ analysis of chloride in soil water samples consists of three component parts, a sampling sub-system, a routing sub-system and a flow injection sub-system. The three components are integrated and controlled using a personal computer. Each sub-system will be considered separately. 2.1.
SUCTION SAMPLING
The size of ceramic cups, and the magnitude of the tension applied are dictated by the requirements of the experiments. Examples and details can be found in Litaor (1988) and Grossmann & Udluft (1991). The system described here was developed to investigate rapid preferential flow mechanisms with soil close to, but not totally
FIELD-BASEDETERMINATIONOFCHLORIDEIN SOILWATER
2]9
Ceramic suction sampler 0.2tim Sample trap
T
"--J.t!l
W
~ u
i O.04m
~
A~phere
Vacuum
m
trap
A
!
1
t
i
So,e/o,, ' r~n~
I
val~
, " C-"
I . . . .
coiour
r~gent:
,
Deionised ;: ~!~i" ,
:
~
~
~
T i
[J. _ [.... Jr|~e~ion v a ' !
2
!t Waste
:i
~
i !
B
Control lines i
Data aquisition link
PC with ADC and DIO cards
Fig. 1. Schematicrepresentation of the field based spectrophotometric flow-injection system. A: the suction sampling sub-system. B: the sample routing sub-system. C: the flow-injection sub-system.
220
NICHOLASM. HOLDENET AL.
3 Inlet/1 outlet routing valves
cleaning agent
?
standard
2 Inlet/1 outlet routing valves
to Injection system
Fig. 2. Detail of the sample routing sub-system LOOP
CARRIER
/ ~ Filling position
REAGENT ......
REACTOR
SAMPLE
WASTE
~ Emptying position
II
~
CARRIER
REORENT
i
Fig. 3. Diagram of the method of operation of the sample injection valve.
FIELD-BASE DETERMINATION OF CHLORIDE IN SOIL WATER
221
saturated. The routing and flow-injection subsystems can be connected to any type of ceramic cup. The ceramics used here were tubes 25 cm long by 4 cm diameter. Figure la illustrates the construction of a sampler and surface trap. The ceramic tube was designed to be installed horizontally into the soil thus eliminating the possibility of creating preferential flow pathways along access holes. To ensure a sample could always be drawn to the sample trap, the sampler was inclined, and the capillary tube weighted with a glass tube placed around it which resulted in water accumulation around the end of the capillary tube. The vacuum used (0. 1 kPa) to draw sample into the system was applied through a trap (to protect the pump), and a sample trap. A pinch valve was located between the sample and vacuum traps to allow access to atmosphere. This was necessary so that the peristaltic pump could drain the sample trap whilst permitting the other samplers to remain evacuated and operating. 2.2.
SAMPLE ROUTING
A routing system was required for maximum utilization of the flow-injection system. The system took about 3 minutes to process a sample, while the samplers took much longer to yield sufficient sample for analysis. By staggering the timings of the system, best use of the resources was possible. The routing network developed to connect nine samplers, standards and cleaner to the flow-injection system is illustrated in Figure 2. Four 3 inlet/1 outlet solenoid control valves created a nine into one structure. A 2 inlet/1 outlet valve was placed in-line to allow a standard to be routed to the flow-injection system for periodic reliability checking. After calibration tests, an additional valve was added to allow nitric acid to be routed through the system as a cleaning agent, which improved the quality of the output. These two valves could have been replaced with a fifth 3 inlet/1 outlet valve. Figure lb indicates a small part of the total routing structure and how it linked the sample and flow-injection systems. 2.3.
SAMPLE ANALYSIS
The flow-injection system used was based on a method presented by Clinch et al (1987), and Williams et al (1992). There are three major components to the system (Figure lc): a peristaltic pump ("Mini-S', Ismatec UK, Carshalton, Surrey, U.K.), an injection valve ("Sample Injection Valve", Bukard Scientific, Uxbridge, Middlesex, U.K.), and a photo-diode detector (made in house). The pump was used to propel sample, de-ionized water (the carrier) and chloride colour reagent through the system. The colour reagent comprises equal volumes of mercuric thiocyanate and ferric nitrate. Thiocyanate ions, released through a replacement reaction with chloride (to produce mercuric chloride) combine with ferric ions to form ferric thiocyanate. Ferric thiocyanate is highly coloured and is formed in concentrations proportional to the original concentration of chloride. This makes
222
NICHOLAS M. HOLDENET AL.
TABLE I Calibration data for a photo-diode detector system. The mean offset from two calibration performed on different days was used to produce the final calibration curve. standard conc (mg 1 - 1 )
mean (1) offset(mV)
S.D. (1) (mV)
mean (2) S.D. (2) offset (mV) (mV)
mean offset
5 10 20 50 100 150 200 250 300
0.006699 0.012300 0.021498 0.041879 0.062742 0.077988 0.091619 0.104952 0.112196
0.000859 0.000453 0.000706 0.001024 0.001142 0.001250 0.002733 0.001199 0.001480
0.007667 0.013054 0.023512 0,045623 0.068393 0.082492 0.093082 0.101832 0.110178
0.007183 0.012677 0.022505 0.043751 0.065557 0.080240 0.092350 0.103392 0.111187
~-~
0.000696 0.000941 0.000946 0.001178 0.001303 0.002985 0,001576 0,001734 0.001057
photodetector colour reagent
I--7--1 LEDs colour reagent + sample
~ photodel~ctor Fig. 4. The optical arrangement within the photodiode detector.
it highly suited to the colourimetric determination of chloride (Zall et al., 1956; Standing Committe of Analysts, 1984; Skalar Analytical, 1992). The injection valve (Figure 3) was used to inject samples into the flow of carrier which was routed through the specially fabricated photo-diode detector. Figure 4 illustrates the optical arrangement within the detector. There are two detectors, which run in parallel. One measures absorbence of light (535 nm) through colour reagent, and the other, absorbence of the same light through colour reagent mixed with carrier or sample. The output from the detector is a millivolt response proportional to the difference between the absorbences. The concentration of chloride is determined by comparing the mV response for the carrier, to the mV response for sample. The difference in response is related to the chloride concentration. It is necessary to calibrate the detectors to produce a standard curve.
FIELD-BASE DETERMINATION OF CHLORIDE IN SOIL WATER
2.4.
223
COMPUTER CONTROL
The complete field-based system schematically represented in Figure 1 was controlled using a IBM PC compatible personnal computer. The computer was equipped with an analogue/digital converter (ADC, Avantech PCL818) and a digital input/output controller (DIO, Avantech PCL722). The DIO controller was used to switch solid state relays which actuated the sample trap pinch valves, routing valves, pump and injection valve. The ADC monitored mV response with time over the period when sample was injected. The system was controlled by specially written software using the C language which integrated system control and data acquisition.
3. Calibration
The detector needed to be calibrated before field-installation so that millivolt response data could be converted into chloride concentrations. Calibration was performed using nine potassium chloride (KC1) standards of increasing concentration (5, 10, 20, 50, 100, 150, 200, 250 and 300 mg 1-1). The calibration procedure was automated using the routing valves and software written to control the system. Each standard was injected into the carrier thirty times, and the response recorded. The whole procedure was replicated once. Table 1 details the replicate calibration data obtained for a system. It can be seen that the variation in offset values for the 5 mg 1-1 standard is greater than those for higher concentrations. For field application where background chloride levels are low, this could prove to be a problem, but at the site in Devon, U.K. which this system was designed for, background chloride levels are about 10 mg 1-1, thus eliminating precision problems at low concentrations. To generate a calibration curve, the data from each replicate were used to establish a mean offset value, these were plotted against concentrations and a third order polynomial fitted to the data. Figure 5 illustrates the curve for the data in Table 1. The 300 mg 1-1 standard was superfluous because the intensity of colour for the ferric thiocyanate reaction was near its operational limit and was very close to that for 250 mg 1-1. A good fit was achieved (r2 = 0.99980) using the equation: y = 0.769830 + 587.25288x + 7563.64023x 2 + 102077.74487x 3. The equation describing the calibration curve was used in the field control software to permit calculation of concentration values for samples during analysis. Thus an observer could watch the progress of a tracer in close to real-time, and observe any problems and rectify them as they occurred. The concentration of chloride to be applied in the field was determined by an optimum range of 10-250 mgC11-1 for the detector. A concentration of 225 mgC1 1-1 was chosen because, when summed with background the maximum chloride concentration would not exceed 250 mg 1-1 .
224
NICHOLAS M. HOLDEN ET AL.
400
300
200
I
=o o
100
0
~l~l'l
o
o
L l~
o
o
¢:;
f'
o
offset
[~
o
I~I
o
' 1'1
o
¢~
¢:;
(mY)
Fig. 5. The calibration curve for the photo-diode detector.
4. Field Experimentation The system described in this paper was developed for use with samplers installed in a large undisturbed soil block located in Devon, U.K. (Holden et al., 1994, 1995a, 1995b). The block was designed for conducting fine resolution tracer experiments using automated sampling procedures to permit many samples to be analysed without recourse to a laboratory. The first experiment conducted was using a
225
FIELD-BASE DETERMINATION OF CHLORIDE IN SOIL WATER
°'°I g
0(1)
14.29
1(3)
°
0.85
.~ 0 . 7 0
0.80]
~ o.6s
0.75
t
I
~
..~
0
50
100
150
0.60
'
0
t
r
i
50
100
150
Time 2(1)
7.645
Time
14.29
B(1)
0.95]
8.696
0.70-]
i
0
50
100
150
0
50
Time 4(1)
.o
-~ o . s 6
IyA.
100
150
Time
8.241
5(I)
.
,
~0.
15.99
° I
0.52
I 0
~ 50
i 100
n 150
0.60
I 0
I 50
i 100
150
Fig. 6. Sample traces from the system for background chloride. These traces were obtained from a large soil block installation in Devon, UK (Holdenet al., 1994).
constant surface flux of 10 mmhr-1 simulated rainfall with a chloride tracer applied with the simulated rainfall for a 2 hour period. Samples were analysed by 6 field based flow-injection systems, each with 9 sample traps. Sample in each trap was analysed every 25-30 minutes. Sample was collected manually every hour for laboratory analysis to confirm the correct functioning of the systems before further experimentation occurred.
5. Results and Discussion Traces from samples analyses in the field are presented in Figure 6. Laboratory determined chloride concentrations were 10.214-1.02 mgl- ~ while field determined were 11.44-0.6 mgl - I , as calculated from samples extracted from the soil a few minutes apart, Chloride breakthrough curves from 0.4 m in the soil block determined in the field and laboratory are presented in Figure 7. The laboratory curve had less noise and a lower peak concentration value due to the poorer sampling resolution (at best 1 hour). Figure 8 presents chloride breakthrough at 0.5 m in the soil block at 4 locations illustrating the type of data obtainable using the flow-injection system.
226 90 80
'l
N I C H O L A S M. H O L D E N ET AL.
T +
70
i
6O T
g tO
.m
!'
C
40
C 0
30 +
l
r
20 4i
I 04 -0.5
0
0.5
1
1.5
2
time (days)
Fig. 7. Chloride breakthrough curves from the same sampler analysed in the field (~) and laboratory (B). 160 -
14o
12o
E
1ooi
C
.9
o
60
u
40!
8°
4
.2
1
-0.5
0
0.5
1
1.5
2
time (days) Fig. 8. Chloride breakthrough curves from 4 locations at 0.5 m in the soil block during a tracer experiment with a 10 mmhr-1 constant surface flux.
The s y s t e m proved to require little maintenance once installed and tested. Operator input w a s restricted to ensuring a supply o f colour reagent and deionized water, and retrieval o f data from the computer controlling the system. The small sample
FIELD-BASEDETERMINATIONOF CHLORIDEIN SOIL WATER
227
volumes required by the system to perform a replicated analysis (c.5ml) meant that temporal sampling resolution could be fine, thus the system was well suited to experiments aimed at obtaining reliable data for testing mechanistic models. The low rates of flow used in the system kept the production of spent colour reagent to a minimum but disposal of waste was a problem that needed to be well managed to prevent pollution in the field. It can be concluded that the flow injection system presented here is a useful technique for chloride determination, and when combined with the necessary sampiing and routing equipment, provides a much needed method of data acquisition in relation to rapid flow mechanisms in soils. The photo-diode detector has been used for determination of nitrate and phosphorus and can potentially be used for any tracer which can be colourimetrically quantified.
References Addiscott, T. M., Rose, D. A., and Bolton, J.: 1978, Chloride leaching in the Rothamsted drain guages: Influence of rainfall pattern and soil structure. Journal of Soil Science 29, 305-314. Addiscott, T. M., Whitmore, A.P. and Powlson, D. S.: 1991, Farming, fertilizers and the nitrate problem. CAB International, Wallingford, U.K. Clinch, J. R., Worsfold, P. J. and Casey, H.: 1987, An automated spectrophotometric field monitor for water quality parameters. Determination of nitrate. Analytica Chimica Acta 200, 523--531. Fiihr, F and Hance, R. J.: 1992, Lysimeterstudies of thefate ofpesticides in the soil. BCPC Monograph 53. British Crop Protection Council. Grossmann, J. and Udluft, P.: 1991, The extraction of soil water by the suction-cup method: A review. Journal of Soil Science 42, 83-93. Holden, N. M., Scholefield, D., Williams, A. G. and Dowd, J. E: 1994, Design of a large soil block for fine temporal and spatial resolution investigations of preferential flow. In: Etchevers, B. (ed.), Transactions of the 15th World Congress of Soil Science, Acapulco, International Society of Soil Science/Mexican Society of Soil Science, Mexico, p 169-170. Holden, N. M., Scholefield, D., Williams, A. G. and Dowd, J. E: 1995a, A large soil block for in situ real-time fine spatial and temporal resolution measm'ement of preferential flew. In: Wiezik, B. (ed.), Hydrological Processes in the Catchment, Cracow Urfiversi~y of Tectmology, Cracow, Poland, p 239-247. Holden, N. M., Dowd, J. F., Williams, A. G. and Scholefield, D.: 1995b, Computer control for investigating water and chemical transport in a large isolated soil Nock. Computers and Electronics in Agriculture 12, 225-236. Hosang, J.: 1993, Modelling preferential flow of water in soils - a two-phase appr~ach for field conditions. Geoderma 58, 149-163. Jarvis, N. J., P-E. Jansson and Dik: RE.: 1991, Modelling water and solute transport in macroporous soil. II. Chloride breakthrough under non-steady flow. Journal of Soil 'Science 42, 59-70~ Litaor, M.I. 1988. Review of soil solution samplers. Water Resources Research 24, 727-733. Marshall, J. E. 1991.: Comparison of tracer movement through disturbed vs. intact cores taken from a structured piedmont soil. M.S. thesis, Graduate Faculty, University of Georgia. Roth, K., Jury, W.A., Fliihler, H. and Attinger, W.: 1991, Transport of chloride through an unsaturated field soil. Water Resources Research 27, 2533-2832. Skalar Analytical: 1992, The SANplu~ Segmented flow analyser. Soil and Plant analysis.Skalar Analytical B.V. Breda, The Netherlands. Standing Committee of Analysts: 1984, Methods for the examination of water and associated materials. Methods for assessing the treatability of chemicals and industrial wastewaters. HMSO London.
228
NICHOLAS M. HOLDEN ET AL.
Tyler, D. D. and Thomas, G. W.: 1981, Chloride movement in undisturbed soil columns. Soil Science Society of America Journal 45, 459-461. Webster, C.P., Shepherd, M. A., Goulding, K. W. T. and Lord, E. I.: 1993, Comparrison of methods for measuring the leaching of mineral nitrogen from arable land. Journal of Soil Science 44, 49-62. Williams, A. G., Dowd, J. E and Marshall, J.: 1992, A flow injection system for temporal analysis of soil waters. AGU 1992 Fall meeting, San Fransisco, USA. (Abstract EOS, October 1992). Zall, D. M., Fisher, D. and Garner, M. Q.: 1956, Photometric determination of chloride in water. Analytical Chemistry 28, 1665-1668.
IDepartment of Geographical Sciences, University of Plymouth, Plymouth, PL4 8AA, U.K.* 2Department of Geology, University of Georgia, Athens, Georgia, 30602, USA. 3Institute of Grassland and Environmental Research, North Wyke, Okehampton, Devon, EX20 2SB, U.K. (Received: October 1994; revised: February 1995)
Abstract. The use of chloride as a tracer for soil water investigations is discussed. Limitations with
laboratory based soil core and field based sampling are considered with respect to the poor suitability of the data for rigorous assessment of mechanistic models. Investigation of water movement in soil has been restricted by limitations in spatial and temporal sampling. Fine resolution sampling generates large numbers of samples which cause problems with post sampling laboratory analysis. This paper describes a field-based system for the analysis of chloride in soil water. There are three component parts to the system, (i) a custom sampling sub-system comprising of ceramic cup suction samplers and sample traps, (ii) a sample routing sub-system to channel sample to (iii) a sample analysis sub-system utilizing a flow injection method for sample analysis using a custom built photo-diode detector. The three sub-systems were controlled by a suitably equipped personal computer. A calibration procedure is described with a third order polynomial equation derived to convert millivolt response from the detector into chloride concentration. Field and laboratory data from a tracer experiment are presented and discussed, and it is concluded that the system is well suited to field-based applications. Finally it is noted that the photo-detector is suitable for colourimetric analysis of any tracer with suitable chemical determination.
1. Introduction Environmental pollution due to agricultural practises is of great interest to environmental scientists. Water movement through soils, and mass transport of chemicals is of particular importance, with much research being devoted to investigation of nitrate (Addiscott et al, 1991) and pesticide behaviour (Ftihr & Hance, 1992). To monitor water movement, tracers are used which provide indication (by means of concentrations) of rates of flow and interactions of new water (that applied in an 'event') with old water (that residing in the soil prior to an 'event'). A conservative tracer which has little chemical reaction in the soil is best used for this type of work so that tracer concentrations are not inadvertently influenced by interaction with colloids or chemical species in the soil. Chloride is typically utilized for this purpose and has been extensively used in the past (Addiscott et al, 1978; Tyler & Thomas, 1981; Jarvis et al, 1991; Roth et al, 1991). Chloride is favoured because * Address for correspondence: Institute of Grassland and Environmental Research, North Wyke, Okehampton, Devon, EX20 2SB, UK
Environmental Monitoring and Assessment 36: 217-228, 1995. (~) 1995 Kluwer Academic Publishers. Printed in the Netherlands.
218
NICHOLAS M. HOLDEN ET AL.
it is cheap, can be determined reliably by wet chemistry, and can be applied to soil in the field without constituting a serious pollution effect. Laboratory based investigations using small soil cores may make possible intensive sampling regimes (Marshall, 1994) but are restricted in the spatial information they can provide, while field based experiments tend to have sampling rates in the order of once per day which provide insufficient data to characterize rapid flow events (Roth eta/, 1991). This has meant that the scope of field investigations of soil water movement have tended to be limited due to spatial and temporal resolutions of sampling being too coarse to provide data suitable for testing of mechanistic models in a rigorous manner (Hosang, 1993). Soil water is usually extracted from the soil using ceramic cups with a negative pressure applied internally, and a means of extracting water from the cup available at the soil surface (Webster et al, 1993). Traditionally such water samples are stored and analysed in the laboratory with the associated risk of mislabelling, cross-contamination and chemical and biological degradation. When monitoring water movement through the soil, ceramic cups are placed at a number of depths down the soil profile, a tracer is applied to the soil surface, and its migration through the profile monitored by means of concentration/time relationships obtained from each suction sampler. When a large number of ceramic cups are used to characterize a field site, fine temporal resolution sampling generates large numbers of samples. This situation can be accommodated in various of ways: (i) by having a number of technicians collecting and storing samples during the experiment with later analysis; (ii) by having an automatic sample collection and storage system that permits later analysis, and (iii) by having automatic sampling and field based analysis. This paper describes a field based method of sample chloride analysis using a low-cost spectrophotometric detector, and a routing system that permits multiple samplers to be connected to a single detector. The methodology may be adapted for other tracers for which suitable analytical methods are available.
2. Instrumentation
The equipment developed for in situ analysis of chloride in soil water samples consists of three component parts, a sampling sub-system, a routing sub-system and a flow injection sub-system. The three components are integrated and controlled using a personal computer. Each sub-system will be considered separately. 2.1.
SUCTION SAMPLING
The size of ceramic cups, and the magnitude of the tension applied are dictated by the requirements of the experiments. Examples and details can be found in Litaor (1988) and Grossmann & Udluft (1991). The system described here was developed to investigate rapid preferential flow mechanisms with soil close to, but not totally
FIELD-BASEDETERMINATIONOFCHLORIDEIN SOILWATER
2]9
Ceramic suction sampler 0.2tim Sample trap
T
"--J.t!l
W
~ u
i O.04m
~
A~phere
Vacuum
m
trap
A
!
1
t
i
So,e/o,, ' r~n~
I
val~
, " C-"
I . . . .
coiour
r~gent:
,
Deionised ;: ~!~i" ,
:
~
~
~
T i
[J. _ [.... Jr|~e~ion v a ' !
2
!t Waste
:i
~
i !
B
Control lines i
Data aquisition link
PC with ADC and DIO cards
Fig. 1. Schematicrepresentation of the field based spectrophotometric flow-injection system. A: the suction sampling sub-system. B: the sample routing sub-system. C: the flow-injection sub-system.
220
NICHOLASM. HOLDENET AL.
3 Inlet/1 outlet routing valves
cleaning agent
?
standard
2 Inlet/1 outlet routing valves
to Injection system
Fig. 2. Detail of the sample routing sub-system LOOP
CARRIER
/ ~ Filling position
REAGENT ......
REACTOR
SAMPLE
WASTE
~ Emptying position
II
~
CARRIER
REORENT
i
Fig. 3. Diagram of the method of operation of the sample injection valve.
FIELD-BASE DETERMINATION OF CHLORIDE IN SOIL WATER
221
saturated. The routing and flow-injection subsystems can be connected to any type of ceramic cup. The ceramics used here were tubes 25 cm long by 4 cm diameter. Figure la illustrates the construction of a sampler and surface trap. The ceramic tube was designed to be installed horizontally into the soil thus eliminating the possibility of creating preferential flow pathways along access holes. To ensure a sample could always be drawn to the sample trap, the sampler was inclined, and the capillary tube weighted with a glass tube placed around it which resulted in water accumulation around the end of the capillary tube. The vacuum used (0. 1 kPa) to draw sample into the system was applied through a trap (to protect the pump), and a sample trap. A pinch valve was located between the sample and vacuum traps to allow access to atmosphere. This was necessary so that the peristaltic pump could drain the sample trap whilst permitting the other samplers to remain evacuated and operating. 2.2.
SAMPLE ROUTING
A routing system was required for maximum utilization of the flow-injection system. The system took about 3 minutes to process a sample, while the samplers took much longer to yield sufficient sample for analysis. By staggering the timings of the system, best use of the resources was possible. The routing network developed to connect nine samplers, standards and cleaner to the flow-injection system is illustrated in Figure 2. Four 3 inlet/1 outlet solenoid control valves created a nine into one structure. A 2 inlet/1 outlet valve was placed in-line to allow a standard to be routed to the flow-injection system for periodic reliability checking. After calibration tests, an additional valve was added to allow nitric acid to be routed through the system as a cleaning agent, which improved the quality of the output. These two valves could have been replaced with a fifth 3 inlet/1 outlet valve. Figure lb indicates a small part of the total routing structure and how it linked the sample and flow-injection systems. 2.3.
SAMPLE ANALYSIS
The flow-injection system used was based on a method presented by Clinch et al (1987), and Williams et al (1992). There are three major components to the system (Figure lc): a peristaltic pump ("Mini-S', Ismatec UK, Carshalton, Surrey, U.K.), an injection valve ("Sample Injection Valve", Bukard Scientific, Uxbridge, Middlesex, U.K.), and a photo-diode detector (made in house). The pump was used to propel sample, de-ionized water (the carrier) and chloride colour reagent through the system. The colour reagent comprises equal volumes of mercuric thiocyanate and ferric nitrate. Thiocyanate ions, released through a replacement reaction with chloride (to produce mercuric chloride) combine with ferric ions to form ferric thiocyanate. Ferric thiocyanate is highly coloured and is formed in concentrations proportional to the original concentration of chloride. This makes
222
NICHOLAS M. HOLDENET AL.
TABLE I Calibration data for a photo-diode detector system. The mean offset from two calibration performed on different days was used to produce the final calibration curve. standard conc (mg 1 - 1 )
mean (1) offset(mV)
S.D. (1) (mV)
mean (2) S.D. (2) offset (mV) (mV)
mean offset
5 10 20 50 100 150 200 250 300
0.006699 0.012300 0.021498 0.041879 0.062742 0.077988 0.091619 0.104952 0.112196
0.000859 0.000453 0.000706 0.001024 0.001142 0.001250 0.002733 0.001199 0.001480
0.007667 0.013054 0.023512 0,045623 0.068393 0.082492 0.093082 0.101832 0.110178
0.007183 0.012677 0.022505 0.043751 0.065557 0.080240 0.092350 0.103392 0.111187
~-~
0.000696 0.000941 0.000946 0.001178 0.001303 0.002985 0,001576 0,001734 0.001057
photodetector colour reagent
I--7--1 LEDs colour reagent + sample
~ photodel~ctor Fig. 4. The optical arrangement within the photodiode detector.
it highly suited to the colourimetric determination of chloride (Zall et al., 1956; Standing Committe of Analysts, 1984; Skalar Analytical, 1992). The injection valve (Figure 3) was used to inject samples into the flow of carrier which was routed through the specially fabricated photo-diode detector. Figure 4 illustrates the optical arrangement within the detector. There are two detectors, which run in parallel. One measures absorbence of light (535 nm) through colour reagent, and the other, absorbence of the same light through colour reagent mixed with carrier or sample. The output from the detector is a millivolt response proportional to the difference between the absorbences. The concentration of chloride is determined by comparing the mV response for the carrier, to the mV response for sample. The difference in response is related to the chloride concentration. It is necessary to calibrate the detectors to produce a standard curve.
FIELD-BASE DETERMINATION OF CHLORIDE IN SOIL WATER
2.4.
223
COMPUTER CONTROL
The complete field-based system schematically represented in Figure 1 was controlled using a IBM PC compatible personnal computer. The computer was equipped with an analogue/digital converter (ADC, Avantech PCL818) and a digital input/output controller (DIO, Avantech PCL722). The DIO controller was used to switch solid state relays which actuated the sample trap pinch valves, routing valves, pump and injection valve. The ADC monitored mV response with time over the period when sample was injected. The system was controlled by specially written software using the C language which integrated system control and data acquisition.
3. Calibration
The detector needed to be calibrated before field-installation so that millivolt response data could be converted into chloride concentrations. Calibration was performed using nine potassium chloride (KC1) standards of increasing concentration (5, 10, 20, 50, 100, 150, 200, 250 and 300 mg 1-1). The calibration procedure was automated using the routing valves and software written to control the system. Each standard was injected into the carrier thirty times, and the response recorded. The whole procedure was replicated once. Table 1 details the replicate calibration data obtained for a system. It can be seen that the variation in offset values for the 5 mg 1-1 standard is greater than those for higher concentrations. For field application where background chloride levels are low, this could prove to be a problem, but at the site in Devon, U.K. which this system was designed for, background chloride levels are about 10 mg 1-1, thus eliminating precision problems at low concentrations. To generate a calibration curve, the data from each replicate were used to establish a mean offset value, these were plotted against concentrations and a third order polynomial fitted to the data. Figure 5 illustrates the curve for the data in Table 1. The 300 mg 1-1 standard was superfluous because the intensity of colour for the ferric thiocyanate reaction was near its operational limit and was very close to that for 250 mg 1-1. A good fit was achieved (r2 = 0.99980) using the equation: y = 0.769830 + 587.25288x + 7563.64023x 2 + 102077.74487x 3. The equation describing the calibration curve was used in the field control software to permit calculation of concentration values for samples during analysis. Thus an observer could watch the progress of a tracer in close to real-time, and observe any problems and rectify them as they occurred. The concentration of chloride to be applied in the field was determined by an optimum range of 10-250 mgC11-1 for the detector. A concentration of 225 mgC1 1-1 was chosen because, when summed with background the maximum chloride concentration would not exceed 250 mg 1-1 .
224
NICHOLAS M. HOLDEN ET AL.
400
300
200
I
=o o
100
0
~l~l'l
o
o
L l~
o
o
¢:;
f'
o
offset
[~
o
I~I
o
' 1'1
o
¢~
¢:;
(mY)
Fig. 5. The calibration curve for the photo-diode detector.
4. Field Experimentation The system described in this paper was developed for use with samplers installed in a large undisturbed soil block located in Devon, U.K. (Holden et al., 1994, 1995a, 1995b). The block was designed for conducting fine resolution tracer experiments using automated sampling procedures to permit many samples to be analysed without recourse to a laboratory. The first experiment conducted was using a
225
FIELD-BASE DETERMINATION OF CHLORIDE IN SOIL WATER
°'°I g
0(1)
14.29
1(3)
°
0.85
.~ 0 . 7 0
0.80]
~ o.6s
0.75
t
I
~
..~
0
50
100
150
0.60
'
0
t
r
i
50
100
150
Time 2(1)
7.645
Time
14.29
B(1)
0.95]
8.696
0.70-]
i
0
50
100
150
0
50
Time 4(1)
.o
-~ o . s 6
IyA.
100
150
Time
8.241
5(I)
.
,
~0.
15.99
° I
0.52
I 0
~ 50
i 100
n 150
0.60
I 0
I 50
i 100
150
Fig. 6. Sample traces from the system for background chloride. These traces were obtained from a large soil block installation in Devon, UK (Holdenet al., 1994).
constant surface flux of 10 mmhr-1 simulated rainfall with a chloride tracer applied with the simulated rainfall for a 2 hour period. Samples were analysed by 6 field based flow-injection systems, each with 9 sample traps. Sample in each trap was analysed every 25-30 minutes. Sample was collected manually every hour for laboratory analysis to confirm the correct functioning of the systems before further experimentation occurred.
5. Results and Discussion Traces from samples analyses in the field are presented in Figure 6. Laboratory determined chloride concentrations were 10.214-1.02 mgl- ~ while field determined were 11.44-0.6 mgl - I , as calculated from samples extracted from the soil a few minutes apart, Chloride breakthrough curves from 0.4 m in the soil block determined in the field and laboratory are presented in Figure 7. The laboratory curve had less noise and a lower peak concentration value due to the poorer sampling resolution (at best 1 hour). Figure 8 presents chloride breakthrough at 0.5 m in the soil block at 4 locations illustrating the type of data obtainable using the flow-injection system.
226 90 80
'l
N I C H O L A S M. H O L D E N ET AL.
T +
70
i
6O T
g tO
.m
!'
C
40
C 0
30 +
l
r
20 4i
I 04 -0.5
0
0.5
1
1.5
2
time (days)
Fig. 7. Chloride breakthrough curves from the same sampler analysed in the field (~) and laboratory (B). 160 -
14o
12o
E
1ooi
C
.9
o
60
u
40!
8°
4
.2
1
-0.5
0
0.5
1
1.5
2
time (days) Fig. 8. Chloride breakthrough curves from 4 locations at 0.5 m in the soil block during a tracer experiment with a 10 mmhr-1 constant surface flux.
The s y s t e m proved to require little maintenance once installed and tested. Operator input w a s restricted to ensuring a supply o f colour reagent and deionized water, and retrieval o f data from the computer controlling the system. The small sample
FIELD-BASEDETERMINATIONOF CHLORIDEIN SOIL WATER
227
volumes required by the system to perform a replicated analysis (c.5ml) meant that temporal sampling resolution could be fine, thus the system was well suited to experiments aimed at obtaining reliable data for testing mechanistic models. The low rates of flow used in the system kept the production of spent colour reagent to a minimum but disposal of waste was a problem that needed to be well managed to prevent pollution in the field. It can be concluded that the flow injection system presented here is a useful technique for chloride determination, and when combined with the necessary sampiing and routing equipment, provides a much needed method of data acquisition in relation to rapid flow mechanisms in soils. The photo-diode detector has been used for determination of nitrate and phosphorus and can potentially be used for any tracer which can be colourimetrically quantified.
References Addiscott, T. M., Rose, D. A., and Bolton, J.: 1978, Chloride leaching in the Rothamsted drain guages: Influence of rainfall pattern and soil structure. Journal of Soil Science 29, 305-314. Addiscott, T. M., Whitmore, A.P. and Powlson, D. S.: 1991, Farming, fertilizers and the nitrate problem. CAB International, Wallingford, U.K. Clinch, J. R., Worsfold, P. J. and Casey, H.: 1987, An automated spectrophotometric field monitor for water quality parameters. Determination of nitrate. Analytica Chimica Acta 200, 523--531. Fiihr, F and Hance, R. J.: 1992, Lysimeterstudies of thefate ofpesticides in the soil. BCPC Monograph 53. British Crop Protection Council. Grossmann, J. and Udluft, P.: 1991, The extraction of soil water by the suction-cup method: A review. Journal of Soil Science 42, 83-93. Holden, N. M., Scholefield, D., Williams, A. G. and Dowd, J. E: 1994, Design of a large soil block for fine temporal and spatial resolution investigations of preferential flow. In: Etchevers, B. (ed.), Transactions of the 15th World Congress of Soil Science, Acapulco, International Society of Soil Science/Mexican Society of Soil Science, Mexico, p 169-170. Holden, N. M., Scholefield, D., Williams, A. G. and Dowd, J. E: 1995a, A large soil block for in situ real-time fine spatial and temporal resolution measm'ement of preferential flew. In: Wiezik, B. (ed.), Hydrological Processes in the Catchment, Cracow Urfiversi~y of Tectmology, Cracow, Poland, p 239-247. Holden, N. M., Dowd, J. F., Williams, A. G. and Scholefield, D.: 1995b, Computer control for investigating water and chemical transport in a large isolated soil Nock. Computers and Electronics in Agriculture 12, 225-236. Hosang, J.: 1993, Modelling preferential flow of water in soils - a two-phase appr~ach for field conditions. Geoderma 58, 149-163. Jarvis, N. J., P-E. Jansson and Dik: RE.: 1991, Modelling water and solute transport in macroporous soil. II. Chloride breakthrough under non-steady flow. Journal of Soil 'Science 42, 59-70~ Litaor, M.I. 1988. Review of soil solution samplers. Water Resources Research 24, 727-733. Marshall, J. E. 1991.: Comparison of tracer movement through disturbed vs. intact cores taken from a structured piedmont soil. M.S. thesis, Graduate Faculty, University of Georgia. Roth, K., Jury, W.A., Fliihler, H. and Attinger, W.: 1991, Transport of chloride through an unsaturated field soil. Water Resources Research 27, 2533-2832. Skalar Analytical: 1992, The SANplu~ Segmented flow analyser. Soil and Plant analysis.Skalar Analytical B.V. Breda, The Netherlands. Standing Committee of Analysts: 1984, Methods for the examination of water and associated materials. Methods for assessing the treatability of chemicals and industrial wastewaters. HMSO London.
228
NICHOLAS M. HOLDEN ET AL.
Tyler, D. D. and Thomas, G. W.: 1981, Chloride movement in undisturbed soil columns. Soil Science Society of America Journal 45, 459-461. Webster, C.P., Shepherd, M. A., Goulding, K. W. T. and Lord, E. I.: 1993, Comparrison of methods for measuring the leaching of mineral nitrogen from arable land. Journal of Soil Science 44, 49-62. Williams, A. G., Dowd, J. E and Marshall, J.: 1992, A flow injection system for temporal analysis of soil waters. AGU 1992 Fall meeting, San Fransisco, USA. (Abstract EOS, October 1992). Zall, D. M., Fisher, D. and Garner, M. Q.: 1956, Photometric determination of chloride in water. Analytical Chemistry 28, 1665-1668.
