136 Amlyst, February, 1979, Vol, 104, @p. 136-142 Apparatus for the Automatic Preparation of Soil Extracts for Mineral-nitrogen Determination J. A. P. Marsh, R. Kibble-White and C. J. Stent Agricitltural Research Council Weed Research Organizatzon, Begbroke Hill, Yarnton, Oxford, OX5 1PF An apparatus is described that automatically prepares samples and feeds an AutoAnalyzer system. It consists of a reagent adder, which adds the correct volume of extractant for an approximately weighed amount of soil, and a sample preparation unit, which mixes, filters, dilutes and loads samples on to an AutoAnalyzer sampler. The results obtained using the apparatus were in good agreement with those obtained by manual sample preparation. Keywovds : Mineral-nitrogen determination ; soil analysis ; automatic extraction The Microbiology Group at the Weed Research Organization analyses about 3 500 samples for mineral-nitrogen each year.The manual preparation of such large numbers of soil samples for the determination of ammonium- and nitrate- plus nitrite-nitrogen on an Auto- Analyzer is tedious, time consuming and prone to considerable operator error. As accurate weighing of soil samples and addition of a fixed volume of extractant are two of the slowest jobs involved, it was decided that a considerable saving of skilled labour could be made by building a modified version of the reagent adder manufactured by the British Sugar Corporation Ltd. for use in their Tarehouse Laboratories. The remainder of the system was improved by designing and building a sample preparation unit (SPU) that would automatically stir and filter the mixture, dilute the extract and load it on to an AutoAnalyzer sample tray.Apparatus Reagent Adder The reagent adder (Fig. 1) consists of a basic beam balance (Denward Instruments Ltd.) modified to weigh between 19 and 21 g and tared to use standard-mass plastic beakers. One balance pan has been removed and replaced with a palladium probe connected to the equip- ment electronics. This probe is suspended in the neck of a specially manufactured pipette with a precision-bore neck (T. W. Wingent Ltd., Cambridge) designed such that a 1-g variation from 20 g on the balance gives a difference of 2 ml in the of the equipment cs dependent on the use of an electrolytic solution were chosen for the liquid in use (2 M potassium chloride solution).KCI reservoir pipette. The working and component values I Control panel I Inlei. Outlet f- valve Microswitch sv 2 (MS 1) Fill Empty Normal Inhibit Inhibit Fill Fig. 1. Layout of the reagent adder.MARSH, KIBBLE-WHITE AND STENT 137 A schematic drawing of the electronic circuit is shown in Fig. 2. When the mains switch S1 is closed, power is applied to the mains transformer whose nominal output of 20 V is rectified to produce a supply, allowing for losses in the transformer and rectifiers, of approxi- mately 24 V d.c.; S2, the Inhibit empty switch, and S3, the Inhibit fill switch, are shown in the inhibit position, and S4, the Normal - Flush switch, is shown in the flush position. During normal operation these switches are closed. When power is applied to the circuit, the two resistors (R1 and R2) apply sufficient current to the two transistors (in a Darlington pair) to turn on the solenoid valve SV1 and its associated indicator lamp LP1.The potassium chloride solution rises in the pipette until it reaches the mid probe (Fig. l), when the low-impedance path through the liquid removes the drive from the transistors and valve SV1 turns off. Between 19 and 21 g of moist soil are placed in a beaker on the balance pan, then push-button PB1 is operated. This applies power to the relay RA/2 and the mid probe is disconnected by contact RA2, thus re-establishing drive to the transistors and turning SV1 on again. The relay is latched on by contact RA1. The pipette then continues to fill until the liquid reaches the palladium probe, when the potassium chloride solution again removes drive from the transistors, thus turning off SV1 and RA2, which is unlatched. Indicator lamp LP1 also goes out, showing that the liquid has reached the required level.The beaker is removed from the balance pan and placed on a platform, which operates microswitch MS1. This applies power to valve SV2 and removes the transistor drive, via S4, from the Fill valve circuit, allowing the contents of the pipette to drain into the beaker. YT), N Liquid Mid Circuit diagram of the reagent adder. Fig. 2. As soon as MS1 has been operated the next sample is weighed out, while the pipette is draining. If the whole of the system is to be drained or washed out then switch S4 is opened and the microswitch MS1 is closed, thus applying power to SV1 and SV2 at the same time.It is important that the system should be thoroughly washed out after use because if potassium chloride solution is left in the system it crystallises and blocks the valves. If part of the pipework is to be drained, S2 or S3 is operated as appropriate. Sample Preparation Unit (SPU) After addition of the reagent, a magnetic follower (length 40 mm) is placed in each beaker, and the beakers are fitted into cups on the conveyor on the SPU (B in Fig. 3). The beakers are carried over four rows of revolving magnets to give a total of 48-min stirring. Subse- quently, the beakers stand, unstirred, for 12 min and then tip into filter-funnels (C in Fig. 3) around the periphery of a 10-sided Perspex table.The soil extract, filtered through a Whatman No. 1 filter-paper, is collected in a beaker (D in Fig. 3) fitted to a similar table lying below the filter-table. The tables rotate on a common spindle until the sample reaches a sample pick-up arm (E in Fig. 3) and is drawn up, via a peristaltic pump (F in Fig. 3), and is automatically diluted when necessary and dispensed via another arm (G in Fig. 3) into the cups on an AutoAnalyzer tray (H in Fig. 3).138 MARSH et al.: APPARATUS FOR THE AUTOMATIC PREPARATION Analyst, VoZ. 104 T I 470'rnrn 360imm K I Sick view L Fig. 3. Layout of the sample preparation unit. The frame of the conveyor is constructed from angle-iron. The conveyor consists of two lengths of Renolds chain with a 30-mm pitch, joined by 30 aluminium slats (350 x 25 mm) at right-angles to the chain.Each slat carries four plastic cups into which 100-ml plastic beakers fit tightly. The conveyor, in a triangular configuration with gear wheels at the angles (J in Fig. 3), is driven by a Parvalux 9.0 N m torque motor (K in Fig. 3) running at 5 rev min-l. The stirring mechanism consists of 16 horseshoe magnets (Gallenkamp, Cat. No. XJP- 780-T) set in four rows of four (L in Fig. 3) and attached to ball races by spindles, each of which carries a toothed pulley. The magnets are rotated by a toothed drive belt, and their polarities are arranged such that the magnetic field helps the turning and, therefore, only a small motor (Parvalux SD8S, 50 W) is required to drive them.l The two 10-sided Perspex tables are fixed 150 mm apart on a vertical spindle driven by a motor (95 W).The top disc is 0.965 m across the flats, 9 mm thick and each of the 10 sides has four holes drilled to hold 63 mm diameter polypropylene funnels (M in Fig. 3). The funnels are fitted with rubber grommets (9.5 x 6.3 mm), which sit flush on the upper table to prevent unfiltered sample from running down the outside of the funnel and dripping into the beaker below. The lower disc has depressions drilled in it to locate the collection beakers, under the funnels (D in Fig. 3). The collection beakers are 100-ml plastic beakers that have been cut off at approximately 35 mm to facilitate entry of the pick-up arm. At the edge of the upper surface of the lower disc are cams (N in Fig.3), which are positioned to contact a microswitch that controls the position of the tables at each movement. The movements of the Perspex tables and the conveyor are co-ordinated so that as each beaker moves over the end of the conveyor it tips its contents into a funnel aligned with it. A sheet of polyethylene draped from a bar (P in Fig. 3) prevents cross-contamination of the samples as they move round. The sample pick-up unit is situated approximately one third of the way round the table (E in Fig. 3) and is designed to pick up the sample from the table, flush with air, pick up a wash solution and then flush with air again until it picks up the next sample. It consists of an arm connected to a 12-V car windscreen-wiper motor withFebruary, 1979 OF SOIL EXTRACTS FOR MINERAL-NITROGEN DETERMINATION 139 limit switches to stop movement at the end of the track.For sampling or washing, the arm is lowered into the sample or wash solution by means of an electromagnet, and is returned to its upper position after sampling, by means of a spring. The sample is picked up through a small-bore polyethylene tube, the suction being provided by tubing on the peristaltic pump of the AutoAnalyzer. If required, the sample can be diluted at this stage using the pump and AutoAnalyzer coils, and it is then pumped to the AutoAnalyzer sample tray via a dispenser unit (G in Fig. 3). The dispenser unit is similar to the pick-up unit, but has a simple horizontal movement from the AutoAnalyzer sample tray to waste. The timing and switching sequence of the SPU operates on a 3-min cycle, which is started and maintained by the presence of a 110-V a.c.supply, that is switched by the timing cam on a Technicon AutoAnalyzer. Individual timings, which are all given as time after initiation, are controlled by standard plug-in (octal) timers and the auxiliary switching by standard plug-in (octal) relays. Schematic drawings of the circuit are shown in Fig. 4. The two halves of the circuit are drawn separately for clarity only. Fig. 4 (a) is the timing circuit and Fig. 4 ( b ) is the low-voltage switching circuit. The presence of mains voltage at the inlet powers timer TF, which closes the 110-V circuit. When the 110-V supply from the Auto- Analyzer switches on, relay RA operates and connects the mains supply to the timer circuit and the 12-V d.c.supply to the low-voltage circuit. Power is thus supplied to the pick-up solenoid via pin 4 and the pick-up arm drops into the sample on the lower table. After 60 s TA stops and the +12-V supply is transferred from the pick-up solenoid to the pick-up motor via pin 3. This allows the pick-up arm to spring back to its upper position. The motor swings the pick-up arm to the wash position where a double-limit switch removes power from the motor and connects the +12-V supply to relay RD. This relay latches on and removes the +12 V from pin 3 for the remainder of the cycle. At the same time mains voltage is applied to timer TG and via contact TG1 to the table motor. The table rotates and allows microswitch C (MSC) to slide off the table cam (N in Fig.3) and close. This powers RB, which by-passes TG1. TG cuts out while the table is in motion so that when the next cam position is reached MSC removes the +12 V from RB, which in turn removes power from the table motor, thus stopping the table at the appropriate point. At every fourth movement of the table, one corner of the table momentarily closes micro- switch MSB, which powers RC. This applies mains voltage to the conveyor motor and a cam on the conveyor allows microswitch MSA to close before MSB opens again, thus main- taining power to RC. This allows the conveyor to continue to move to its next position when the next cam opens MSA and RC removes power from the conveyor motor. After 75 s TD makes contact TD1, enabling contact TE1 to apply +12 V to the dispenser motor, thereby driving the dispenser arm to the sample position, where it is stopped by a limit switch.Timer TE is initiated by mains power from the other contact on TD. After 90 s TB completes its cycle and contact TB1 is made, enabling TC1 to apply +12 V to the pick-up solenoid via pin 2, thus lowering the pick-up arm into the wash solution. Contact TB2 applies mains voltage to TC. After 115 s TE switches, TE1 applies +12 V to the dispenser motor via pin 3 and the dispenser arm drives to the waste position, where it is stopped by a limit switch. After 120 s TC completes its cycle, TC1 removes +12 V from the pick-up solenoid, allowing the pick-up arm to return to its upper position, and applies +12 V to the pick-up motor, which drives the pick-up arm to its original position over the disc, where a limit switch stops it.The equipment remains in this position for the remainder of the 3 min. The end of the sequence is signified by a short interruption in the 110-V supply from the AutoAnalyzer, which causes a similar break in the 12 V and mains supplies, thus re-setting the timers. When the 110-V supply is re-applied, by the timing cam of the AutoAnalyzer, the sequence re-st arts. Results Tests were carried out to compare the efficiency of extraction results obtained using a manual weighing and sample preparation method, and the reagent adder and sample preparation unit. Extracts of four replicates of 10 soils were prepared by each method and140 MARSH et al. : APPARATUS FOR THE AUTOMATIC PREPARATION Analyst, VoZ.104 analysed for nitrate- plus nitrite-nitrogen by a diazotisation and coupling reaction with sulphanilic acid and N-( 1-naphthy1)ethylenediamine and ammonium-nitrogen by an indo- phenol method. These methods q e described fully by Greaves et aL2 (4 2 s I I I 11ov N 8 TD2 6 L 5 I s 4 6 RB 1 3 1 T l f N r - - - y - - - - I Pickup -7 - 6 I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I t I I I I I -1 I I I I I I I I I I r -.- - - - - - - TE 1 I 1 3 I I ' 2 (- Fig. 4. Circuit diagrams of the sample preparation unit.February, 1979 OF SOIL EXTRACTS FOR MINERAL-NITROGEN DETERMINATION 141 Results for the nitrate plus nitrite determinations (Table I) show that there was very little difference in the amounts extracted with either technique.TABLE I DETERMINATION OF NITRATE- PLUS NITRITE-NITROGEN IN SOIL AFTER EXTRACTION BY MANUAL AND AUTOMATIC TECHNIQUES Sample 4 5 6 7 8 9 10 Manual preparation determinations/ deviation/ Coefficient of pg g-l dry soil pg g-' variation, % 23.15 0.671 2.9 14.00 0.303 2.2 8.38 0.120 1.4 38.00 1.281 3.4 19.08 0.426 2.2 4.80 0.293 6.1 10.08 0.086 0.9 47.45 0.426 0.9 14.80 0.172 1.2 57.94 1.536 2.7 7 A \ Mean of 4 Standard Automatic preparation determinations/ deviation/ Coefficient of pg g-l dry soil pg g1 variation, % 23.26 0.587 2.5 13.54 0.220 1.6 8.18 0.160 2.0 36.82 0.913 2.5 19.04 0.323 1.7 4.82 0.223 4.6 10.54 0.250 2.4 47.66 0.798 1.7 15.28 0.215 1.4 58.49 0.538 0.9 A f \ t Mean of 4 Standard Automatic - manual A \ Difference/ Standard w g - l error/trggl t 0.11 0.446 0.26 - 0.46 0.187 2.46 - 0.20 0.100 2.03 - 1.17 0.786 1.49 - 0.04 0.268 0.15 0.01 0.184 0.07 0.46 0.132 3.49 0.20 0.462 0.45 0.47 0.138 3.44 0.54 0.814 0.67 Ammonium-nitrogen (Table 11) extracted by the automatic method was slightly higher than that extracted manually, and was significantly higher when less than 2 pg of ammonium- nitrogen per gram of dry soil was present.This may be due to the timing for the filtering of the samples and loading the AutoAnalyzer being shorter than with the manual method. The regression coefficient for results of automatic against manual methods is significantly less than 1 at P = 0.01, the relationship between the two being TABLE I1 DETERMINATION OF AMMONIUM-NITROGEN IN SOIL AFTER EXTRACTION BY MANUAL AND AUTOMATIC TECHNIQUES Sample 1 2 3 4 5 6 7 8 9 10 Manual preparation determinations1 deviation/ Coefficient of pg g-l dry soil pg gl variation, % 6.93 0.264 3.8 4.41 0.261 5.9 I * If Mean of 4 Standard 3.10 0.082 2.6 4.28 0.101 2.4 2.49 0.113 4.5 0.48 0.024 5.1 3.18 0.030 0.9 1.46 0.017 1.2 0.55 0.024 4.3 1.48 0.024 1.7 Automatic preparation Mean of 4 Standard determinations/ deviation/ WLg g-l dry soil !J.g g-' 6.53 0.155 4.33 0.088 3.20 0.051 4.37 0.278 2.42 0.118 0.66 0.025 3.37 0.136 1.63 0.062 0.74 0.025 1.61 0.015 Coefficient of variation, yo 2.4 2.0 1.6 6.4 4.9 4.4 4.0 3.8 3.4 0.9 Automatic - manual L I \ Diff erencel w g-l - 0.40 - 0.08 0.09 0.09 - 0.07 0.08 0.18 0.17 0.19 0.13 Standard errorlpg g-1 t 0.153 2.64 0.138 0.69 0.048 1.93 0.148 0.68 0.082 0.82 0.018 4.69 0.070 2.66 0.032 5.20 0.017 11.28 0.014 8.89 Discussion The equipment described in this paper has now been used for the preparation of over 4000 samples for determining ammonium- and nitrate- plus nitrite-nitrogen in soil and has proved reliable.A comparison of the labour-intensive stages in the manual and automatic sample prepara- tions is shown in Table 111. The equipment has reduced the labour requirement by approxi- mately 60%. The SPU has also been used for the extraction of available phosphate from soil, although analytical problems have, at present, made it impossible to automate the loading of the A4utoAnalyzer for this analysis. We thank Mr. D. F. A. Horsley and Mr. P. R. Leaton of the British Sugar Corporation Ltd. for their guidance and advice in building the reagent adder, Mr. R. W. Foddy for constructing the equipment and Mr. R. C. Simmons who designed the prototype circuits and did the original wiring.142 MARSH, KIBBLE-WHITE AND STENT TABLE I11 COMPARISON OF STAGES REQUIRING LABOWR IN THE MANUAL AND AUTOMATIC NITRATE- PLUS NITRITE-NITROGEN ON AN AUTOANALYZER PREPARATION OF SOIL SAMPLES FOR DETERMINATION OF AMMONIUM- AND Stage 1 2 3 4 5 6 7 8 9 10 11 Manual preparation Accurate weighing into bottles Accurate addition of extractant Stoppering bottles Placing on shaker Removing from shaker Transport to filter area Unstoppering bottles Setting up filters Filtering Diluting Loading AutoAnalyzer tray Stage Automatic preparation I Rough weighing into pre-tared beakers 2 Placing on conveyor belt 3 Setting up filters References 1. 2. Baker, K. F., Analyst, 1970, 95, 885. Greaves, M. P., Cooper, S. L., Davies, H. A., Marsh, J. A. P., and Wingfield, G. I., “Technical Report, Received September 13th, 1978 Accepted September 20th, 1978 Agricultural Research Council Weed Research Organization,” 1978, No. 45, p.55.