Reactivity of silicate liming materials from Northern Europe assessed by Soil Incubation and two pH Stat methods

Karl-Jan Erstad Consultative Agronomists, Lutelandet, N-6964 Korssund, Norway, e-mail: karl-jan.erstad@raadgivande-agronomar.no Nikolai Y. Konovalov Regional Research Computer Centre, Lanskoe shosse 59-108, R-197343 St. Petersburg, Russia, e-mail: nyk@comset.net Jukka Putro Rautaruukki Steel, Primary Products, PO Box 93, FIN-92101 Raahe, Finland, e-mail: jukka.putro@rautaruukki.fi Martin Rex Versuchsanstalt Kamperhof, Mintarder Straße 264, D-45481 Mülheim an der Ruhr, Germany, e-mail: rex@th-duenger.de Elsa Luukkonen Plant Production Inspection Centre, Agricultural Chemistry Department, PO Box 83, FIN-01301 Vantaa, Finland, e-mail: elsa.luukkonen@kttk.fi


Introduction
The production of hot metal and steel produces different types of slags (silicate materials).Hot metal production under reduced conditions results in a blast furnace slag with 30-40% SiO 2. The previous converter slag from the manufacture of pig iron was the famous Thomas slag, which contained about 4% P. The current slags in Northern Europe contain only one tenth of this.By air-cooling, a slag will be a homogenous crystalline product, and by water-cooling it will be amorphous and will consist of porous granules.
A range of experiments has been carried out to test the efficiency of slags as agricultural liming materials.The glassy ferro manganese slag from Øye Smelteverk was tested by Simán (1984) in two fractions, 0-0.1 mm and 0-3 mm, and he found that the finer fractions had the greater liming effect, but were still inferior to chalk from Skåne.The Mg and in particular the Mn effects made it valuable as a special fertilizer, a fact that runs parallel to the recommendations for certain blast furnace slags made by Chichilo et al. (1954).
Other ferro slags with different features have been tested in Russia.Ogorodnikov and Novikov (1986) tested different liming materials in an acid derno-podzolic clay loam in the Ural region, adding lime to full hydrolytic soil acidity.Only a finely ground limestone raised the soil pH slightly more effectively than the applied ferro chromium slag.Sonina and Melnikova (1971) performed greenhouse experiments on an acid derno-podzolic loamy sand with pH KCl 4.2-4.4,adding lime to full hydrolytic acidity.Without lime the pH dropped to 3.9-4.0,but increased to 5.8, 4.9 and 5.9 when ferro chromium slag, blast furnace slag and ground limestone, respectively, were added.
In another liming experiment on a derno-podzolic soil using limestone, blast furnace slag and ferro alloy slag, Deryugin and Kultyshev (1973) found small differences.The ground limestone gave the best results for the yields of pea.
Blast furnace slag was better for winter rye and barley, and ferro alloy slag for potatoes.Myhr (1987Myhr ( , 1988) tested a 0-3 mm converter slag obtained from the Linz-Donavitz (LD) process at Norsk Jernverk AS.Due to its coarseness, its liming effect was inferior to finely ground limestone, but there were no differences in yield responses.On organic soils Myhr and Erstad (1996) found that the same slag was superior in maintaining pH in the long term, and P, micronutrients and silica had an additional fertilizing effect.Penninck et al. (1986) proved that carbonate liming materials and finely ground LD slag from Belgium had an equal liming effect.Comparing the effect of different metallurgical slags Yagodin et al. (1994) observed that converter slag had a particularly stable effect on soil acidity even eight years after application.In the long term this was an even better effect than other slags and agricultural limes.
The particle size distribution of slags is the most important factor of their efficiency as liming materials.Jaakkola (1988) pointed out that air-cooled blast furnace slag required very fine grinding to achieve high pH levels in soil, whilst the water-cooled product was more effective.The ground converter slag was the most efficient in raising the soil pH.Avdonin (1976) considered that blast furnace slag was most effective if its particle size was less than 0.25-0.5 mm.Kulakova et al. (1988) found that in the second year after application of converter slags, the highest increase in soil pH and base saturation, and simultaneous reduction in hydrolytic acidity, was obtained with a finely ground product.Particles coarser than 1.0 mm would only slightly influence yields during the first year after application, and dissolution of coarse particles above 2.5 mm would take many years (Dovgopol 1980).Russian technical specifications for the metallurgical slags used in agriculture require that the content of CaCO 3 equiv.(equivalent) should be at least 80%, maximum water content 2%, 100% of particles <2 mm, 90% <0.5 mm and 70% <0.25 mm (Avdonin 1976).In Finland (Statue Book of Finland 1994) the requirements for slags concern only NV (neutralizing value), Vol. 9 (2000): 333-348.and the minimum levels are 30% Ca for converter slags and 25% Ca for blast furnace slags.The previous Norwegian standard 2885 (Norges Standardiseringsforbund 1987) required only that slags should contain at least 30% CaO equivalent.Moreover, Dovgopol (1980) compared the effects of slag with the usual liming materials.He mentions the additional benefits of the multilateral effects of slags due to their additional acid soluble content of phosphorus, sulphur, micro-and macronutrients and other silica elements.Volk et al. (1952), Vetter (1974), Rex and Munk (1989) pointed out that crop response to slags is not predictable from soil reactions alone, but could also be related to the silica effect.Carter et al. (1951) found that a US agricultural grade air-cooled blast furnace slag containing boron was a highly valuable fertilizer on boron deficient loamy soils.
The purpose of the present experiment was to determine the liming efficiency of particle size fractions of different slags in order to classify them in the forthcoming European standards.In addition, conversion factors between different analytical methods are needed for reference in the production of standards and the subsequent advisory work.This study augments the work by comparing the results obtained from two quick laboratory methods.These results are compared with a soil incubation method.The soil incubation method is designed to show the actual liming effect in the soil.A preliminary report related to short term effects of the slags in this research work has been published (Erstad et al. 1994).

Material
The silicate liming products chosen for the experiments were the most relevant silicate products (slags) from the participating companies Øye Smelteverk, Tinfos Jernverk AS (Kvinesdal, Norway), SSAB Merox AB (Oxelösund, Sweden) and Rautaruukki OY (Raahe, Finland).Breivik (Norway) coarsely crystalline calcite was included as a reference.This product is well known from several other reactivity tests.Table 1 shows the average values of Ca and Mg content in these liming materials, determined according to prEN 13475 (2000) andEN 12947 (2000) respectively, and calculated figures of % CaO equiv.= [% CaO] + [% MgO x 1.4], a term which is very close to NV (neutralizing value) according to prEN 12945 (2000).There were very small differences in chemical composition between fractions of each material.The converter slags had a higher content of reactive calcium oxide and some magnesium oxide compared with the blast furnace slags.The ferro manganese slag from Øye Smelteverk contained only half the level of CaO equivalent compared to the others,  1987), and ombrogenous peats.The silty loam soil (Hodgson 1974) was brought from Strømmen in Askvoll, Sunnfjord in Norway, and consisted of 11.3% sand, 73.2% silt and 15.5% clay.The peaty soil was a purchased natural product (Hydro Huminal), and was chosen to investigate the particular effects of acid organic matter on lime dissolution.The soil chemical analyses are given in Table 2.
The soil density values were obtained by following the procedure described by Bondorff (1950).Loss on ignition was measured at 823 K for 12 h, and the organic matter calculated by subtracting the hygroscopic water held in the clay minerals of the soil and obtained from standard tables of correction, referenced by Scheffer and Schachtschabel (1979).pH(H 2 O) was measured in distilled water using a soil to water ratio of 1:2.5 v/v.Exchangeable H, Ca, Mg, K and Na were extracted using the NH 4 OAc method pH 7.00, as described by Thomas (1982).The acidity was determined by back titration to pH 7.00, and Ca, Mg, K and Na by atomic absorption spectrometry.Base saturation was determined by calculation.

Terminology of reaction of liming materials
Reactivity is the rate of reaction of a liming ma-terial with acid (soil or solution) under specified conditions (prEN 12944-3 2000).
Neutralising value is the ability of a liming material to neutralise soil or solution acidity, measured in accordance with prEN 12945 (2000).It is a number, which represents the amount in kilograms of calcium oxide (CaO) which has the same neutralising effect as 100 kg of the product under consideration (prEN 12944-3 2000).
The Effectiveness of a liming material is its ability to dissolve in soil or solution, measured relatively to its neutralising value, given at a fixed time of reaction.
The Efficiency of a liming material is its effectiveness over a period of time.
Effectiveness and efficiency data are generally given for each particle size fraction of the liming material being evaluated.
The Effective neutralising value (ENV) of a liming material is equal to its neutralising value multiplied with its effectiveness in soil at fixed times, normally 1 and 5 years.
were thoroughly mixed into the soil at two levels, corresponding to 3000 and 6000 kg CaO equiv.ha -1 in a 20 cm plough layer.Each limed treatment had two replicates, and each zero treatment had four replicates.One litre of normally compacted dried soil and the liming materials to be tested were mixed and water was added to a 70% water holding capacity for mineral soils, and 60% for organic soils.The incubation was performed on a barren soil.The pots were covered with a parafilm during the incubation period of 130 weeks.It was removed 5-7 days before each sampling.Samples were taken at 1, 3, 6, 12, 24, 52, 78, 104 and 130 weeks after liming.
On each sampling date, 10 cm 3 of soil was collected from each pot separately with a miniauger, then mixed and stirred with distilled water at a soil:water ratio of 1:2.5 v/v for 5 minutes and the pH measured after one night.The equipment used for pH measurements was a METROHM 654 digital pH meter with separate electrodes.
The pH effects on each soil were calculated by subtracting the average pH value of the zero treatment from those of the limed treatments.
Because there is linearity between liming and pH in the range 4.5-6.5(7.0), the conversion of the pH increase to the effectiveness of liming materials could be established.The finest fractions of calcite (<0.063 mm) are known to dissolve immediately, and as pH dropped for this material due to acidifying reactions in soil, the peaks of slightly coarser calcite fractions reached their maximum, and this maximum was set to 100% dissolution, as pH development was closely studied.This implied that the finer fractions of calcite were set to 100% dissolution, too, because these were known to have had their effect already.Correspondingly the finer fractions of silicates were set equal to the peak dissolution value within each liming material.The reactivity of a fraction was calculated as the effectiveness of the fraction compared to the completely dissolved fractions of the given slag type.
The data for 24 and 130 weeks were equivalent to the ENV (effective neutralising value = NV x reactivity in soil) over the course of 1 year and 5 years, respectively, according to Erstad (1992).By multiplying the given data of effectiveness with results from analyses of CaO equivalent and by performing the experiment by particle size fraction, the reactivity of any liming material can be calculated by multiplying the reactivity fractionwise with its particle size distribution, and subsequently summarised.

The Sauerbeck/Rietz method
The currently adopted German method was a potentiometric titration of aqueous carbonate suspensions with HCl at pH 2.0 according to Sauerbeck and Rietz (1985).The pH stat titration period was 10 min.Results were given as per cent dissolution of CaCO 3 equiv.for each particle size fraction.These CaCO 3 equiv.values were calculated from analyses of Ca and Mg, and were a little higher than neutralising values determined by titration according to prEN 12945 (2000).

The modified Finnish pH Stat method
The unpublished Finnish method was described by the Plant Production Inspection Centre as a 'Fast acting neutralising capability of liming materials with the pH Stat method.'The method was an automatic potentiometric titration, where each sample of liming material was leached for 24 hours in a water suspension, while pH was constantly kept at 5.0 by adding 0.4 M nitric acid (HNO 3 ) into the titration vessel, as soon as the pH of the solution tended to rise.The titration solution was stirred with a magnetic blender to prevent the finest particles settling out on the bottom of the beaker.The sample of liming material was put into a net basket cylinder made of an acid proof steel net (openings 0.15 mm).This cylinder was attached to a lateral stalk, and moved up and down in the solution at 30 strokes/ min (length of stroke 50 mm).
The effectiveness was assessed by measuring the consumption of HNO 3 during the analysis.The result was given as % dissolved Ca (pHstatFIN), and it did not take impurities into account.To establish a correlation with the re-

Statistical methods
All analyses of variance were made using the Statistical Analysis System (SAS), Release 6.03 (SAS Institute 1985, 1987, 1988).To test the differences of pH increases after 24 and 130 weeks of incubation, comparisons were performed by means of the Ryan-Einot-Gabriel-Welsch Multiple Q Test (REGWQ) at the 5% level (type I experimentwise error rate).The procedure SUMMARY was used to calculate the conversion factors (convf n ) fraction by fraction applying different methods to determine the reactivity of liming materials.Linear correlation equations were established as follows: where f 1 -f 11 = relative mass of each particle size group, from the finest to the coarsest.

Results and discussion
The Soil Incubation method

Erstad, K.-J. et al. Reactivity of silicate liming materials
quite slowly during the first weeks, especially in the case of the BF-Rauta slag, but the effectiveness increased during 130 weeks.The coarsest particle size fractions, in particular 3.15-5.0mm, demonstrated a very low effectiveness, and with exception of the BOF-Rauta slag, this was the case for the entire incubation period.
In general the highest pH values were obtained with Breivik calcite, at least in the short term, and the lowest pH values with the Mn-Øye slag.The converter slags (K-Merox and BOF-Rauta) were superior to the blast furnace slags (M-Merox and BF-Rauta) in their ability to raise soil pH during the first weeks, but the differences were gradually reduced.The am-M-Merox appeared to be slightly more effective than the ordinary M-Merox, and this was particularly true for the finest particle size fractions.
Table 3 shows the mean effects for both liming levels, and the REGWQ Test results for the silicate liming materials, tested by selected fractions and measured relative to Breivik calcite <0.063 mm.A combination of the two liming levels proved to be reasonable, because the low level overestimated the effects of coarse particles in an acid soil, whilst the high level was an over-liming which underestimated the liming effect of the same coarse fractions.The chosen times of measurements, at 24 and 130 weeks of incubation, correspond to ENV values at 1 and 5 years.It was apparent that the converter slags in general were superior, especially their coarser fractions in the last period.For the finer fractions no significant differences could be found between blast furnace and converter slags after 130 weeks.Am-M-Merox achieved its peak effectiveness in the fraction below 0.2 mm.The Mn-Øye was generally inferior to the other slags.In the long term the most effective slags did not differ from the Breivik calcite.
In the Soil Incubation experiment the reactivities of the finer fractions (<0.6 mm) of the blast furnace and converter slags were slightly lower than those of the calcite (Table 4).The coarser fractions (>0.8 mm) of the converter slags were superior to the other slags of similar particle size in increasing pH levels.This was even the case for the blast furnace slag M-Merox for the first period.The Mg-BOF-Rauta was a mixture of BF-Rauta and BOF-Rauta, and its reactivity was intermediate between its parent materials.The reactivity of Mn-Øye was very low, even lower than that reported for an unground 'quenched' slag (glassy calcium silicate) by MacIntire et al. (1946), and this is probably related to its extremely low porosity and dense surfaces.

Tests with the peaty soil
In the Soil Incubation experiment with peat (Table 5) there were fewer statistically significant differences between the liming materials when compared with the incubation in the silty loam.Larger variations could be traced back to spot effects around the liming material particles in the aggressive peat.There were scarcely any differences between the blast furnace and converter slags in this acid peat.The higher content of reactive oxides in the converter slags gave them no clear advantage.The coarser particles of Breivik calcite seemed to be more effective in the long term, and it was evident that its coarse crystallinity did not hamper its dissolution in this acid peaty soil.Even the Mn-Øye slag increased soil pH quite effectively in this experiment, although it was clearly inferior to the other slags.
Based on the Soil Incubation reactivity data in the peat (Table 6), the amorphous M-Merox exhibited almost the same dissolution as the Breivik calcite, followed by the ordinary blast furnace slags.The converter slags were somewhat lower in reactivity, and the Mn-Øye was the least reactive.The finer particles (<1.0 mm) of the mixed Mg-BOF slag gave results which are difficult to explain.It was expected that they would be an average of the BF and BOF slag parent materials.For coarser materials the data were as expected.
Previous investigations were mainly performed on mineral soils, with exception of parts of the experiments of Jaakkola (1988), which, identically with these results, showed that coarser particles of all liming materials tested were more efficient in the peat compared with the mineral soil.This effect was most evident for crystalline carbonates and slags.During a 9 years' soil incubation experiment with mineral and peat soils, Barkoff (1961) found that blast furnace slag had almost the same neutralising effect as finely ground limestone.The results of these experiments coincided very well with the Russian data (Avdonin 1976, Dovgopol 1980, Kulakova et al. 1988) and East German results (Ebert 1970).

The Sauerbeck/Rietz and the modified Finnish pH Stat method
Table 7 shows the test results for the silicate liming materials by fractions for the Sauerbeck/ Rietz and the Finnish pH Stat method.The Breivik calcite was included in this investigation with data from a previous investigation (Erstad et al. 1996).The lower values of the modified Finnish method occur because this method indicates the percentage of dissolved Ca in a material adjusted to be absolutely pure CaCO 3 (adj.% Ca), whereas the Sauerbeck/Rietz method measures the amount of dissolved CaCO 3 equivalent as percentage of the total analysed CaCO 3 in the sample.Thus the maximum value, by full reactivity, in a calcite would be 40.08 by the modified Finnish method, and 100 by the Sauerbeck/Rietz method.For both methods there were close correlations between the fineness of the materials and the reactivity, and the converter slags (K-Merox and BOF-Rauta) were It had previously been found (Erstad et al. 1996) that when testing carbonates with the Finnish method, the finest fraction (<0.2 mm) of CaCO 3 normally achieved full dissolution (adj.40.08% Ca), whereas the crystalline dolomites achieved approximately 40% dissolution (adj.15% Ca).The same dolomite reactivity was found by Lindroos (1986), and he stated that pH 5 of this solution presents a very weak acid, and by decreasing pH to 4 and 3 the blast furnace slag increased most rapidly in solubility.The underestimation of the liming effect of dolomites and silicate liming materials by the Finnish meth-od should be taken into account when considering performance in soil.The Sauerbeck/Rietz method and in particular the Soil Incubation method reflects the dissolution in soil more accurately for these materials.Similarly, Crane (1930) showed that on the basis of CaCO 3 equivalents, determined by titration with HCl (the official US AOAC method) using phenolphthalein as indicator, blast furnace slag and limestone, both finer than 0.15 mm, had nearly the same effectiveness.This investigation confirms that there are only small differences in the ability to raise pH between carbonate and silicate liming materials of the same particle size when added to the soil.

Correlations between the three reactivity methods
There appeared to be the same pattern of reactivity by materials and fractions between the Soil Incubation method in the silty loam, the Sauerbeck/Rietz and the modified Finnish pH Stat methods.The Sauerbeck/Rietz method, with some minor modifications, will be the forthcoming European standard for testing reactivity (prEN 13971. 2000).However, ENV will be widely used for comparison of liming materials under true agricultural conditions.The Finnish method is well accepted and widely used within its country of origin.
A general linear correlation equation for conversion of the results between the methods is given in the statistical part of materials and methods.The conversion factors and their standard deviations for all transformation of data between the methods were established by fraction (Table 8).The data of the Soil Incubation method are considered to be close to true values in agricultural soils, and it was very interesting to observe that the Finnish pH Stat method probably has an advantage compared with the Sauerbeck/Rietz method, because it allows a longer and less rough reaction and contact time between a less aggressive acid and the liming material.In spite of underestimation of some materials, it separates Erstad, K.-J.et al.Reactivity of silicate liming materials the fractions very clearly, making the tests sensitive.This is shown as smaller SD values when data are transformed from the modified Finnish method to the Soil Incubation method.
The SD values of the conversion factors appeared partly to be quite large, and revealed different behaviour of the reactivity methods.The magnitude of the uncertainty and the variations for a liming material is discussed in Erstad et al. (1996).It was found that they would be lower than indicated in Table 8 by each fraction, because these deviations went in arbitrary directions, and as vectors counteracted each other.

Conclusions
This investigation made a ranking of seven Northern European silicate liming materials possible, internally and compared them with a reference calcitic liming material.Correlations between data from three important reactivity meth-ods were established by conversion factors and with data of statistical reliability for these silicate liming materials.It was shown that converter slags with their high proportion of free oxides are superior in raising the pH in slightly acid mineral soils, whilst blast furnace slags are equally as effective in acid peaty soils.In the fine particle range, amorphous blast furnace slag normally has a higher reactivity than its nonamorphous variant.In the short-term converter slags were as efficient as calcite, and in the longterm blast furnace slags also showed the same efficiency.Silikaattisten kalkitusaineiden reaktiivisuutta tutkittiin norjalaisella ferromangaanikuonalla, kolmella ruotsalaisella ja kolmella suomalaisella masuuni-ja konvertterikuonalla.Vertailumateriaalina kokeissa oli kiteinen kalsiitti.Kolmea reaktiivisuuden määritysmenetelmää käytettiin kuonien luontaisen kalkitusvaikutuksen selvittämiseksi ja standardisoinnissa tarvittavien matemaattisten riippuvuuksien laskemiseksi.

%
Erstad, K.-J.et  al.Reactivity of silicate liming materials sults of the two other methods, the chemical purity of the liming materials was taken into account by applying the following correction fac-Ca equiv.= [% Ca] + [% Mg x 1.65] Fig. 1. pH effects for particle size fractions (mm) of Breivik calcite at highest level (6000 kg CaO equivalents per ha) in a silty loam from Askvoll, Norway, during 2.5 years (130 weeks) of incubation.

Table 1 .
Average concentrations of Ca and Mg and calculated CaO equivalents in the liming materials.

Table 2 .
Chemical analyses of the mineral soil from Askvoll, Norway, and the Norwegian natural peat, used in the soil incubation experiment.

Table 3 .
Mean effect values (%) and REGWQ tests for silicate liming materials, tested by fractions and measured relatively to Breivik calcite <0.063 mm, 24 and 130 weeks after start of incubation in a silty loam from Askvoll, Norway.The results of all the particle size fractions and the two reaction times were tested separately.Erstad, K.-J.et al.Reactivity of silicate liming materials

Table 4 .
Reactivity in the course of 1 year and 5 years according to the Soil Incubation method for particle size groups (mm) of the Breivik reference calcite and the silicate liming materials in a silty loam from Askvoll, Norway.Average values for liming levels 1 and 2.

Table 6 .
Reactivity in the course of 1 year and 5 years according to the Soil Incubation method for particle size groups (mm) of the Breivik reference calcite and the silicate liming materials in a natural Norwegian peat.Average values for liming levels 1 and 2.

Table 7 .
Reactivity by the Sauerbeck/Rietz and the modified Finnish pH Stat method for the Breivik reference calcite and the silicate liming materials.

Table 8 .
Conversion factors (convf.)and their standard deviations (SD, %) by fraction for silicate liming materials, obtained by comparing the Soil Incubation, the Sauerbeck/Rietz and the modified Finnish pH Stat method.7 liming materials tested.