Use of single superphosphate fertiliser produced using Eppawala rock phosphate as a source of phosphorous for rice cultivation

CP Udawatte, PVA Panagoda, WMADB Wickramasinghe, JDH Wijewardena, DN Sirisena, S Emitiyagoda and HRUD Bandara 1 Department of Physical Science and Technology, Faculty of Applied Sciences, Sabaragamuwa University of Sri Lanka, Belihuloya. 2 Department of Earth Resource Engineering, Faculty of Engineering, University of Moratuwa, Moratuwa. 3 Rice Research and Development Institute, Batalagoda, Ibbagamuwa. 4 Sri Lanka Council for Agricultural Research Policy, Wijerama Mawatha, Colombo 07. 5 Horticulture Crop Research and Development Institute, Gannoruwa, Peradeniya. 6 Extension and Training Centre, Department of Agriculture, Peradeniya. 7 Lanka Phosphate Limited, Eppawala.

Abstract: Sri Lanka is an agricultural country totally depending on imported fertilisers for paddy cultivation. Importation of fertilisers including triple superphosphate (TSP) is a heavy burden on the country's national economy. Hence, optimum utilisation of the locally available Eppawala phosphate deposit (EPD) will save a signifi cant amount of foreign exchange. The objective of this study was to assess the eff ectiveness of locally produced Eppawala single superphosphate (ESSP) as a source of phosphate fertiliser for rice in comparison to TSP. Small tonnage of ESSP was produced for the purpose of experiments using fi nely ground Eppawala rock phosphate (ERP: 90 % < 150 μm) mixed with 70 % sulphuric acid stoichiometrically, allowing the resulting slurry to solidify. Resulting product granulated after curing for 2-3 weeks was characterised by X-ray fl uorescence (XRF), X-ray powder diff raction (XRD) and the available phosphorus was determined. The available phosphorus was found to be in the range of 20 -22 %, which is far in excess of the stipulated requirement of 17 %. Agronomic eff ectiveness of ESSP over the imported TSP was tested for rice cultivation in diff erent soil conditions in the dry, wet and intermediate zones of Sri Lanka in fi ve consecutive seasons. The results showed that important parameters such as the grain yield and root dry weight in plots applied with ESSP as a source of phosphorus were comparable with that of TSP applied plots. Application of TSP and ESSP showed similar results in dry and intermediate zones of the country. Therefore, if ESSP is produced in suffi cient quantities INTRODUCTION Phosphorus (P) is one of the most important plant nutrients in rice cultivation. Continuous cultivation of high yielding rice varieties has resulted in a decline in the ability of soil to provide the adequate P requirement (Senanayake 1984, Sirisena et al., 2000, Wickramasinghe et al., 2000, Seneviratne et al., 2002. Therefore, application of P as fertiliser is important to maintain soil fertility (Rezania, 1992). At present, triple superphosphate (TSP) is the major source of P fertiliser recommended by the Department of Agriculture (DOA) by the 2000 recommendations for rice in Sri Lanka with nearly 80 % of the total imported TSP used in the paddy sector alone (Hemachandra et al., 1988, Wijewardena et al., 19901998, Wijewardena 1998, 1999. The country's annual requirement of TSP is imported at a cost of nearly Rs. 6.5 billion (NFS, 2018). Experiments have been conducted during the past four decades to investigate the suitability of Eppawala rock phosphate (ERP) in diff erent forms as a P fertiliser for June 2020 Journal of the National Science Foundation of Sri Lanka 48 (2) rice cultivation. Nagarajha et al. (1979;1980), Maraikar et al. (1983) and Joseph (1986) have tested concentrated superphosphate (CSP), Rhenania phosphate (RhP), fused magnesium phosphate (FMP), Eppawala rock phosphate (ERP) and imported rock phosphate (IRP) as sources of P for rice. Long-term experiments have shown that there is hardly any residual or cumulative eff ect of ERP while IRP showing slightly better results. The eff ect of RhP and CSP however, were very signifi cant. It was viewed that the better performance of RhP and FMP could be partly due to their silicate content. Wijesundara et al. (1993) and Wickramasinghe et al., (unpublished data; Annual Report, RRDI) reported that rice showed no response to FMP, but a signifi cant response to di-ammonium phosphate in intermediate and dry zones of Sri Lanka. Ratnayake et al., (1994) reported that TSP is the best source of P fertiliser than selectively mined Eppawala rock phosphate (SERP) or high grade Eppawala rock phosphate (HERP) and ERP. However, the increase in plant P uptake was higher in SERP treated soils than ERP. Although SERP/HERP has a total P 2 O 5 content in the range of 30 to 40 %, citric acid soluble P 2 O 5 is only 5.8 %.
ERP's low solubility prevented its use in short-term food crops. However, performance of SSP produced from ERP over that of TSP in the food crop sector has not been reported yet. Since the nature of the Eppawala apatite deposit favours the production of SSP as reported by NSF (1999), a research programme was launched as per the recommendations of the report.
The manufacturing process of ESSP consists of reacting fi nely-ground phosphate rock with conc. sulfuric acid, resulting ESSP containing ~18 % P 2 O 5 .
The net reaction is given below: The process takes place in two stages as follows: The fi rst stage represents the diff usion of sulfuric acid in the ground rock mass accompanied by a rapid chemical reaction on the particle surface, continuing until the acid is completely consumed with the crystallisation of calcium sulphate.
The second stage represents the diff usion of the formed phosphoric acid into the pores of the rock particles which did not decompose. This stage is accompanied by a second reaction. This reaction ends in the reaction mixer in 30-60 minutes, during the period of settling and hardening of the superphosphate slurry caused by the relatively rapid crystallisation of the low solubility calcium sulphate. The next stage is the ageing of the superphosphate, i.e. the formation and crystallisation of monocalcium phosphate. The end product still containing a certain amount of uncombined phosphoric acid makes the fertiliser more hygroscopic, although it is not a serious disadvantage. However, the hydroscopic nature of the product may be minimised by subsequent treatment of the product with ammonia or ammonium sulphate. This may be carried out prior to and after the curing period.
The objective of this research was to assess the eff ectiveness of locally produced Eppawala single super phosphate fertiliser for rice in comparison with triple super phosphate.

METHODOLOGY
A limited quantity of Eppawala single superphosphate (ESSP) produced from ERP by Lanka Phosphate Limited (LPL) was evaluated for its agronomic effi ciency for rice at the Rice Research and Development Institute, Batalagoda and Regional Agriculture Research and Development Centre, Bombuwala. After conducting the initial experiments, the study was extended to farmer fi elds in Polonnaruwa and Kalutara districts. The adaptability of ESSP was tested in 146 farmer fi elds distributed island-wide with varying soil conditions in collaboration with the Department of Agriculture (DOA), Provincial Departments of Agriculture (PDOA) and the Mahaweli Authority of Sri Lanka.

Preparation of Eppawala single superphosphate (ESSP)
The commercial product, ERP (28 % wt. P 2 O 5 ) of Lanka Phosphate Limited is manufactured by crushing and grinding the run-of-quarry (ROQ) phosphate to < 100 μm powder form using roller mills (SLS-645). The powder was well mixed with commercially available conc. H 2 SO 4 according to the ratio of 100 kg of ERP: 2000 cm 3 of 70 % H 2 SO 4 by manually pouring the acid straight from the plastic container and stirring portion-by-portion for 30 mins in a stainless steel vessel fi xed with a blender operated at 10 rpm throughout its preparation. Twenty tonnes (20 t) of ESSP were produced in batches with each batch producing one tonne of the mixture. The resultant slurry was allowed to cool and stand at room temperature for one month for curing followed by sun drying and grinding to obtain ESSP in powder form (100 μm).
Journal of the National Science Foundation of Sri Lanka 48(2) June 2020 The available P 2 O 5 content of the starting ERP and the resulting ESSP were measured by the Vanodomolybdate method (Jaff ery, 1971). The experimental procedure started taking 1 g of the sample into a 50 mL conical fl ask and adding 25 mL of 2 % citric acid followed by stirring for 30 min. The resultant suspension was fi ltered through No. 40 Watmann fi lter paper with 1 cm 3 of the fi ltrate diluted to 100 cm 3 . The phosphate concentration in the solution was determined colorimetrically (Model-Corning-252) at 460 nm wavelength.
Chemical analysis of ERP and ESSP was conducted with X-ray fl uorescence technique (Model Rigaku NEX-CG, USA), with a beam diameter of 1 mm and an x-ray energy level of 8-10 keV. Phases present in the ERP and the resulting ESSP were determined using a Siemens D5000 powder x-ray diff ractometer with Cu-K α radiation generated at 40 kV and 30 mA.

Field experiment at RRDI at Batalagoda
A long-term experiment was initiated in the Maha 2003/2004 season at the RRDI, Batalagoda to evaluate the performance of ESSP in comparison with TSP. Soil of the experimental site was a low humic gley soil (Batalagoda series) with sandy loam texture. Plot size was 6 m × 3 m. Treatments were arranged in accordance with randomised complete block design (RCBD) with three replications.
The four treatments of the experiment were: (45 kg P 2 O 5 /ha as TSP is the recommended level of P for dry zone and intermediate zone at that time -DOA 2003 fertiliser recommendation for rice) P0 -Control (no P fertiliser) P1 -45 kg P 2 O 5 /ha as TSP P2 -45 kg P 2 O 5 /ha as ESSP P3 -30 kg P 2 O 5 /ha as ESSP At the beginning of the experiment, composite soil samples from each replicate were tested for initial P content. All the plots received N, K and Zn (140 kg N, 40 kg K 2 O and 1 kg Zn/ha) at the rates recommended by the DOA in its 2003 fertiliser recommendations. P treated plots received respective P treatments and applied as basal. Eighteen days old seedlings of the variety Bg 358 were transplanted at the spacing of 15 cm × 15 cm at two plants per hill. Pest, diseases and weed management, and other cultural practices were done as recommended by the DOA. At the time of maturity, plant samples were collected from 10 hills and yield components and dry weights were measured. Air dried samples were used for the determination of plant P content. Plots were harvested at maturity after discarding the border rows. Grains were separated and air dried to calculate the plot yield at 14 % moisture content. Experiment was continued in the Land preparation and cultural practices were similar to the experiment conducted at RRDI, Batalagoda. Plots with the dimensions 6 m × 3 m were arranged in accordance with RCBD with 3 replicates. Rice variety Bg 300 was direct seeded. All the plots received N, K and Zn (140 kg N, 40 kg K 2 O and 1 kg Zn) at the rates recommended by the DOA. P treated plots received respective P treatments and was applied as basal. At the time of maturity, plant samples were collected from 900 m 2 area and yield components and dry weight of plants were measured. Air dried samples were used for the determination of plant P content. Plots were harvested in a similar manner followed at the RRDI experiment.
Field experiments in low country wet zone (LCWZ)

Experiment 1
Field experiments were conducted at 3 locations in the low country wet zone (LCWZ), namely, at the Regional Agriculture Research and Development Centre (RARDC), Bombuwala (location 1) and at two locations in farmers' fi elds in Palayangoda (location 2) and Keenagasmanana (location 3) in Maha 2004/2005 season.
Treatments of the experiment in all locations consisted of 30 Kg P 2 O 5 /ha as per the recommended P level for wet zone, in 2003 DOA.

June 2020
Journal of the National Science Foundation of Sri Lanka 48 (2) the experiments varied from location to location and were on the basis of farmer preference. Fertiliser rates were applied as recommended by the DOA for the respective agro-ecological zone. Testing was carried out in large plots (1000 m 2 ) in adjacent fi elds. Phosphorus was applied as basal at the rate of 45 kg P 2 O 5 /ha in the form of TSP and ESSP for respective plots. Plots were harvested at maturity and the grain yield was recorded at 14 % moisture content.

Characterisation of ERP and ESSP
It was noted that the powdered phase is not suitable for fi eld application since it is partly blown away with the wind. However, this loss can be overcome by granulation or palletisation in the manufacturing process.
The XRF analysis of average CaO and P 2 O 5 of run-of-Quarry apatite sample of ERP and ESSP shows 49.40 wt.%, 27.80 wt.% and 66.30 wt.%, 18.01 wt.% respectively. Analysis did not indicate any signifi cant presence of Cd and As in Eppawala Rock Phosphate. However, it was noted that the Fe and Al content in Sri Lankan apatite is considerably higher (as Fe 2 O 3 9.10 wt.%, Al 2 O 3 5.90 wt.%, respectively) than the reported values in other countries (Jayawardena, 1988). Further, analysis of ESSP showed that it contains 22.40 wt.% S, an added advantage for crop production.
The major element in ERP was Ca within a range of 21.5 -24.5 wt.%. ERP contains apatite [Ca 5 (PO 4 ) 3 .OH, F, Cl] as the main constituent mineral (Dissanayake et al., 2009). Commercial ERP consists of both apatite and the matrix, containing 34.9 -43.2 wt.% Ca (Hewawasam et al.,1999). Percentages of Ca and P obtained were below the levels reported previously using X-ray diff ractometry as the analytical method (Dahanayake et al.,1991, Hewawasam et al.,1999. This could be due to lower sensitivity of the XRF used in the study. Cu and Sr are naturally available in earth and can be assumed to be present in the apatite matrix. Early studies also indicate the presence of 3.3 -11.9 wt.% Fe in secondary apatite and the matrix in the form of hematite -Fe 2 O 3 (Dahanayake et al., 1991;1995). The chemical analysis of ESSP shows a considerable amount of S, since it is accumulated as CaSO 4 .2H 2 O after acid treatment. In addition, F and Cl were not detected in the ESSP since it evolves into gases such as HF and HCl, respectively during the acid treatment.
The treatments were arranged in a RCBD with three replicates, with the plot size of 6 m × 3 m. N and K were applied according to the DOA recommendation of 55 kg/ha and 60 kg/ha, respectively. Rice variety Ld-356 was grown as a test crop. Soil samples were collected from each replicate before the commencement of the experiment and from each plot after each season in all 3 locations for analysis. Available P was analysed by Olsen's method. Ten (10) plants per each plot were uprooted at 45 days after planting and the root dry weight and root length were measured.

Experiment 2
A long-term fi eld experiment was initiated at the RARDC, Bombuwala in 2006. The treatments consisted of : No P, T1 -ERP, T2 -HERP, T3 -ESSP, T4 -TSP and T5 -50 % TSP + 50 % HERP. The experiment was laid out in a RCBD and replicated three times. The level of P was kept constant at the rate of 30 kg P 2 O 5 /ha as recommended by the DOA for rice grown in the LCWZ. N and K were added to every treatment according to the rates and times recommended by the DOA. The plot size was 4.5 m × 4.5 m and rice variety Ld 356 was grown as a test crop. The respective treatment plots were maintained the same during the entire experimental period. Soil samples were collected from each replicate before commencement of the experiment and from each plot after each season for analysis. In addition, 10 plants from each plot were uprooted 45 days after planting and dry weight of foliage, dry weight of roots and root length were determined. After curing the product for a maximum (Gunawardena et al., 1994) of 4 weeks, the measured available P 5 O 5 content of the product was 17.5 wt% agreeing with the similar result of 18 % obtained in the study. It was also observed that the available phosphorus content of ESSP increases with increase in the extent of acidulation.

Pilot scale farmer fi eld demonstrations
in the rice yield in the control plots treated with no P fertiliser. A similar reduction was not observed in the plots treated with TSP or the ESSP.

Soil P content
The initial P level of the experiment site was 6 mg/kg of soil and it is clear that the rice yield is always higher in the plots treated with P fertiliser than no-P plots. A considerable variation was observed in the residual P content in soil after harvesting in the fi rst two seasons ( Table 2). Plots treated with ESSP at the rate of 45 kg P 2 O 5 /ha had signifi cantly higher soil P content at the time of harvesting (7 mg/kg) in the 1 st season, but in the 2 nd and 3 rd seasons TSP 45 kg P 2 O 5 /ha treated plots had higher soil P than all the other treatments followed by ESSP 30 kg P 2 O 5 /ha and ESSP 45 kg P 2 O 5 /ha. In the 3 rd season, the diff erence in soil P contents among four (4) treatments was not signifi cant, but soil P content in all 4 treatments were lower than that of 1 st and 2 nd seasons.

Phosphorous content of plant samples at the time of harvesting
The P content in plant samples showed larger variation among treatments (Table 2). Plant P content was higher in the plots treated with TSP while the lowest P content was recorded in the control plots without P fertiliser.
Between the two ESSP rates, higher P content was recorded in plant samples in the plots treated with ESSP at the rate of 45 P 2 O 5 /ha. P content in the plant samples had a very good relationship with grain yields. When there was lower grain yield, P content of the plant was also lower and vice versa (Tables 1 and 2). Table 3 shows that the application of P fertiliser increased the 1000 grains weight of rice as well as number of fi lled grains per panicle. However, the diff erences among P treatments are not signifi cant.

Performance of ESSP under farmer fi elds in Polonnaruwa
The results show that grain yield among 5 treatments were almost similar in all the 3 locations in Polonnaruwa (Table 4). Yield at the Jayanthipura location was higher compared with the other two locations, while the lowest yield was recorded at Hingurakgoda. The initial soil P contents of the sites were 4.05, 4.22 and 2.22 mg/kg soil for Jayanthipura, Thopawewa and Hingurakgoda, respectively. Although the soil P levels are low, there was no yield diff erence between no-P plots and P-treated plots.
XRD patterns of ERP and ESSP are given in Figure 1  (a and b, respectively). The XRD pattern of ERP can be attributed to hydroxyl apatite Ca 10 (OH) 2 (PO 4 ) 6 which is quite similar to that of the XRD pattern reported by Wimalasena et. al. (2009). XRD pattern for the ESSP reported fi rst time in this study can be attributed to a completely new phase of mono calcium phosphate. However, anhydrate CaSO 4 and gypsum -CaSO 4 .2H 2 O also can be identifi ed as major byproducts. Table 1 shows the grain yield obtained in the experiment conducted at RRDI Batalagoda. There was no signifi cant yield diff erence among four (4) treatments in the 1 st season, but a signifi cant yield reduction was observed in the 2 nd and 3 rd seasons in the control plot. Even in the 4 th season there was no signifi cant diff erence in yield between plots treated with TSP and ESSP at any rate. TSP treated plot produced the highest yield in 1 st , 3 rd and 4 th seasons followed by ESSP treated plots at the rates of 30 and 45 kg P 2 O 5 /ha, respectively. The lowest yield was recorded in the no-P plot. The initial soil P content of the site was 6 mg/kg. The results are in agreement with the fi ndings of IRRI (2000) that when the initial P level in the soil is below 10 mg/kg soil, there is a response to P application. In the 2 nd season, ESSP 45 kg P 2 O 5 /ha rate had a higher yield than TSP 45 kg P 2 O 5 /ha rate. It is clear from these results that there was a gradual reduction June 2020

Performance of ESSP in comparison to TSP at RRDI Batalagoda
Journal of the National Science Foundation of Sri Lanka 48(2) Means followed by the same letter in a column are not signifi cantly diff erent at 5 % level

Soil P content after harvesting
There was a tendency to increase the soil P levels at the time of harvesting when plots were treated with P fertiliser (Table 4). Plots treated with ESSP at the rate of 45 kg P 2 O 5 /ha had the highest soil P at the time of harvesting followed by TSP 45 kg P 2 O 5 /ha. Plots treated with no P fertiliser recorded the lowest soil P at the time of harvesting.

Plant P content at maturity
Plant P content varied among treatments in the same location and also in diff erent locations as well. The location where the lowest soil P content was recorded (Hingurakgoda) had the lowest plant P content (Table 4).

Field experiments in low country wet zone (LCWZ)
Chemical characteristics of three diff erent locations in the LCWZ given in Table 5 show that a considerable amount of available P present in the soil is available at all three locations. The eff ect of addition of ESSP on yield of rice in LCWZ is given in  The eff ect of application of ESSP on dry weight of foliage is given in Table 6. Crops planted at RARDC, Bombuwala and Keenagasmanana locations did not show any signifi cant increase in dry weight of plants due to addition of P fertilisers. However, application of TSP at the rate of 30 kg P 2 O 5 /ha and ESSP at rates of 14.4 and 30 kg P 2 O 5 /ha resulted in signifi cant increase in plant dry weight over the control. The highest plant dry weight was recorded in plots treated with ESSP at the rate of 30 kg P 2 O 5 /ha. However, application of TSP at 30 kg P 2 O 5 /ha ESP at 14.4 and 30 kg P 2 O 5 /ha showed signifi cantly similar plant dry weight.
( Figure 2). However, the highest root dry weight was recorded in plots treated with ESSP at the rate of 30 kg P 2 O 5 /ha followed by ESSP at the rate of 14.4 kg P 2 O 5 / ha; TSP at the rates of 30 kg P 2 O 5 /ha; ESSP at rates of 10.8 and 7.2 kg P 2 O 5 /ha, respectively. Rice crop at Palayangoda and Keenagasmanana did not show any signifi cant increase in root dry weight with the addition of P fertilisers.
There was no signifi cant response in plant height as well as root length to addition of P fertilisers at all three locations.

Eff ect of P sources on yield
The eff ect of P sources on the yield of rice is shown in    Eff ect of P fertiliser sources on available soil P Soil analysis after each crop season showed that the available P (Olsen's) content of all the P treated plots was higher than that of the control plot. However, the increase of soil P to ERP application was insignifi cant compared to that of HERP, ESSP, TSP and 50 %TSP + 50 % HERP. In general, higher P contents were recorded in plots treated with TSP and ESSP.

Pilot scale farmer fi eld demonstrations
The results of 146 fi eld demonstrations conducted throughout the country clearly showed ( These results clearly show that ESSP could be used as a source of P fertiliser for rice cultivation in Sri Lanka. Sulpher content in ESSP may have a positive eff ect in increasing the crop yield particularly in S-defi cient sandy soils, mainly in the eastern region of the country.

CONCLUSIONS
Application of TSP and ESSP has shown similar results in rice cultivation. Direct application of ERP and HERP did not show any positive eff ect on increasing rice yield. However, application of TSP, ESSP, 50 % TSP + 50 % HERP, and 75 % TSP+25 % HERP has shown a positive eff ect in increasing rice yields in the LCWZ.
Farmer fi eld demonstrations have shown that ESSP performance was on par with that of TSP.
These results clearly indicate that ESSP could be used as a source of P fertiliser for rice cultivation in Sri Lanka. In addition, S content in ESSP may have a positive eff ect in increasing crop yield, particularly in S defi cient sandy soils mainly in the eastern region of the country.