Agronomic efficiency and heavy metal contamination of phosphorus (P) recycling products from old sewage sludge ash landfills

Agronomische Effizienz und Schwermetallkontamination von Phosphor (P)-Recyclingprodukten aus mehrjährig deponierten Klärschlammaschen

2 Material and Methods

Fig. 1. Modified Neubauer trial with summer rye at day 8 after sowing.

Test products included a monoincineration ash produced from chemically precipitated sewage sludge and stored as landfill for about 10 years, subjected to various recycling steps in order to strip it of heavy metals and make its P content plant available (SSA 1–4; SSA 1: raw monoincineration sewage sludge ash from the storage site „Raffelberg“ (North Rhine-Westphalia, Germany); SSA 2: SSA1 recycled by stepwise dissolution in 20% H₂SO₄, and subsequent precipitation of P from the filtrate by addition of KOH (pH 3) and filtration, removal of heavy metals based on the different solubility of P and heavy metals at different pH; SSA 3: SSA 1, subjected to the same dissolution process as SSA 2, but precipitation of P as Ca-phosphate to further reduce Al and Fe contents; SSA 4: SSA 2 dissolved in acid again, then treated with sodium based ion exchange resins to remove Al, Fe and Ca and thus improve P solubility). For comparison, a thermo-chemically treated sewage sludge ash produced from recently incinerated sludge (SSA-TC, a monoincinerated SSA, subjected to a thermo-chemical treatment in a rotary kiln at a temperature of 1000°C and with the addition of Na₂SO₄ to form plant available Ca-Na-phosphates and remove heavy metals) was investigated, representing the fertilizer efficiency of treated ash at the current stage of technical development. A water soluble commercial fertilizer (single super phosphate, SSP) was used as control. A zero control variant without any P addition was also included. P contents and P solubility of the fertilizers used in the experiment are described in Table 1.

The experiment was set up with three different P levels, which should represent a deficient, an optimum and a luxury supply with P (10, 20 and 40 mg P per pot), based on previous results on P uptake (Kratz et al., 2010) and an assumed maximum fertilizer use efficiency of 20%. P additions were calculated based on total P contents of the fertilizers after extraction with aqua regia (Table 1). Due to the small amounts of fertilizer to be applied to the limited substrate volume, the test products had to be applied in fine-ground form in order to allow their homoge­nous distribution in the substrate. All other essential macro- and micro-nutrients except P (i.e. N, K, Ca, Mg, S, Fe, Mn, Zn, Cu, B, Mo) were supplied equally to all pots in liquid form as lab grade chemicals and at optimum rates to make sure no other nutrient deficiency would influence the experiment (fertilization details see Table 2).

450 g of fertilized substrate were filled into transparent polystyrol vessels (diameter 12 cm, height 6 cm) and seeded with 100 dressed summer rye seeds of the variety „Arantes“ (1000 grain weight: 41.2 g, germinability: 93%). P-free quartz sand [grading: „fine“ sand: 0.1–0.4 mm, coarse sand 0.7–1.2 mm] was used as substrate. Quartz sand was filled into the vessels in three layers: The bottom layer consisted of 50 g coarse and 100 g fine sand, which was mixed with the ground test fertilizers. This layer was also sprinkled with the supplemental fertilizer solutions. On top of this, a layer of 150 g fine sand was placed. Rye seeds were evenly spread on top of the second layer and covered with a third layer of 150 g fine sand. 4 replicates were prepared for each variant.

All pots were brought to approximately 65% of field capacity with de-ionized water. During the trial, water was supplied individually according to pot weight.

The first harvest of above ground biomass took place 16 days after sowing (BBCH 12/13). A second cut was carried out 30 days after sowing (BBCH 11/12), and the third and final cut, including a separate harvest of roots, 51 days after sowing (BBCH 13/14). BBCH-growth stages were defined according to Stauss et al. (1994). At each harvest, the fresh weight yield was recorded.

For the determination of P contents and solubilities in fertilizers, standard procedures were applied: The original fertilizers were conditioned at 40°C and fine-ground to a particle size of < 60 μm in a Retsch RS 1 vibrating disc mill with zirconium oxide grinding equipment. Total P: Aqua regia (P-AR) digest according to DIN EN 13346/2001–04. Soluble P: Extraction with neutral ammonium citrate (P-NAC) according to EU-method 3.1.4 (EC Regulation 2003/2003), extraction with water (P-water) according to VDLUFA II 4.1.7 (2007). P determination was done colorimetrically (Molybdenum-Blue) according to Murphy and Riley (1962) with (NH₄)₆Mo₇O₂₄ × 4H₂O, using a spectral photometer (Specord 50, Analytik Jena).

Heavy metal contents of the fertilizers were determined in the aqua regia digests by ICP-OES (Thermo icap 6300) and SF-ICP-MS (Thermo Element XR).

Plant available P contents in the substrate after the experiment were determined by P-CAL (calcium acetate lactate) extraction according to Schueller (1969), which is the standard method for the determination of plant available P in soils in Germany (VDLUFA I A 6.2.1.1, 1991). P determination was conducted colorimetrically (see above).

P efficiency was calculated as P uptake in % of P supplied with fertilization, also accounting for the amount of P supplied to the pot with plant seeds:

Relative agronomic efficiency (RAE) was calculated for each variant regarding both DMY and PU. Single superphosphate (SSP) was used as a reference. The calculations were performed as follows:

Differences between variants were statistically tested by one-way ANOVA with a Tukey post-hoc test at a level of significance of p < 0.05, using the SPSS 17.0 software package (SPSS Inc., USA).

3 Results and Discussion

Fig. 2. Cumulative dry matter yield (DMY) of rye for the different tested pro­ducts. Different letters indicate significant dif­ferences between vari­ants at a level of p < 0.05, tested by one­way ANOVA with Tu­key post-hoc test.

The small variation in DMY of the different tested fertilizers is also reflected by the RAE-DMY which did not show large variation between products in most cases, ranging from 84% for the raw SSA 1 up to 91% for SSA 2 at the optimum P level (20 mg) (Table 3). The highest RAE-DMY was achieved at the lowest P level in most cases, indicating that a fertilizer level of 10 mg/pot was already high enough to cover the P demand of the crops under the conditions of the current experiment.

Fig. 3. Cumulative P uptake by rye for the different te­sted products. Different letters indicate significant differences between variants at a le­vel of p < 0.05, tested by oneway ANOVA with Tukey post-hoc test.

Accordingly, P efficiency of the test fertilizers was best for SSA 4, which achieved a similar efficiency as the SSP control at P levels 10 and 20 (Fig. 4).

Fig. 4. P efficiency (% of P supplied with fertilizer and seeds) of the tested products at different P-levels

It is important to mention that the „Neubauer trial“ is a closed system where negative compounds affect plant growth directly without the buffering capacity of a soil. Fertilizer products cannot really interact with the substrate as this is pure sand. Therefore reactions can take place which would not occur under field conditions. This is especially true for element imbalances which can directly affect plant growth. This way the drastic decrease in plant growth observed for SSA 4 at P level 40 can be explained by Na toxicity affecting the plants. SSA 4 had been treated with NaOH after ion exchange, leading to Na toxicity in the small pots at a high fertilization rate. As shown in Table 7, Na-concentrations of SSA 4 exceeded those of the other products by one order of magnitude, or even two orders of magnitude when looking at the P-related concentrations. While the thermo-chemically treated SSA-TC also contained high amounts of Na, this product also had a much higher P concentration than SSA 4, resulting in a total Na load per pot which was about seven times lower than that supplied with SSA 4. Thus, the SSA-TC variant neither suffered from visible toxicity symptoms, nor from such drastic decreases in DMY as observed for SSA 4.

In comparison to RAE-DMY, a larger variation was displayed when relative agronomic efficiency regarding P uptake was calculated (Table 3). At the supposed optimum P level (20 mg), RAE-PU ranged between 53.5% for raw SSA (SSA 1) up to 99% for SSA 4. At the lowest P level (10 mg), which seemed to be already high enough to cover the crop demand for P in the current trial, SSA 4 (treated with ion exchange resin) even reached RAE-PU of 105%, strengthening the assumption that P levels may have been calculated too high, in other words, 20 mg P may have been above optimum in this experiment. Thus, regarding P uptake, SSA 4, which was the only product with not only a high NAC-solubility, but also a high solubility in water (see Table 1), clearly distinguished itself from the non-water-soluble products.

However, as shown in Table 1, P concentrations in SSA 3 and SSA 4 were also considerably lower than in the raw ash and in SSA 2. Therefore, in order to evaluate the contamination of these materials as P fertilizers, their heavy metal concentrations were also calculated in relation to their P contents (Table 5). As the calculations show, all three treatments (SSA 2–4) resulted in a considerable reduction of P-related heavy metal concentrations. Heavy metal concentrations in the thermo-chemically treated SSA-TC were more or less in the same order of magnitude as in the other treated ashes, however, their Ni and Pb concentrations were somewhat higher. Whether this is a result of the different raw ashes used or of a more successful treatment in SSA 2–4 cannot be decided as the raw material from SSA-TC was not available.

The treatments of SSA 1 also reduced the concentrations of Cu and Zn (Table 6). Cu and Zn are no longer regulated in the German Fertilizer Ordinance, because it was decided to focus more on their quality as essential micronutrients than as potential contaminants. However, in larger quantities, they can also become toxic to living organisms. Therefore it is still relevant to look at the effects of various treatments on Cu and Zn concentrations in recycling fertilizers.

Uranium is not yet regulated in the German Fertilizer Ordinance as potential contaminant, however, there has been a lot of discussion about it recently, since it often occurs in rather high concentrations in phosphate rock based P fertilizers (Kratz et al., 2016), and as a heavy metal might have toxic effects on living organisms in the long run. The treatments did not reduce the U concentrations in the products. However, all sewage sludge ash based products displayed considerably lower U concentrations than the commercial SSP (differing by at least one order of magnitude, see Table 6). This should be kept in mind when discussing additional advantages of using recycling fertilizers instead of phosphate rock based products.

An important goal of the Ca treatment (SSA 3) was to further reduce the high concentrations of Al and Fe in the raw ash resulting from chemical P-precipitation with Fe- and Al-salts in the original sewage sludge. However, as the data in Table 7 shows, this attempt was not fully successful. While SSA 3 showed lower Al/Fe concentrations in relation to its dry matter, the Al concentration related to P contents was almost the same for both SSA 2 and SSA 3, and the amount of Fe (per kg P₂O₅) was only reduced by approximately one third in SSA 3. Obviously, the decrease in concentrations related to dry matter was mainly an effect of dilution of the material caused by the Ca-treatment, which reduced absolute P concentration together with the concentrations of Al and Fe.

The limited ability of standard extraction procedures for P fertilizers to estimate the P plant availability of recycling fertilizers has been noted recently by other researchers (e.g. Duboc et al., 2017; Steckenmesser et al., 2017). They suggest re-assessing the current standards and turning to alternative methods which better mirror the processes taking place at the plant-soil-fertilizer interface, particularly sink-based methods such as Fe-oxide strips, Fe-oxide resin bags or the DGT (diffusive gradient in thin films) method.

Fig. 5 shows the amounts of plant available P (P-CAL) left in the substrate after the end of the experiment, indicating to the amount of available P which could not be used by the plants. At the 10 mg P level, the amount of P which was extracted by CAL was in the same order of magnitude as in the zero P control in case of SSP and SSA-TC, indicating to a more or less complete consumption of available P by the plants at this fertilizer level. Even without P fertilization, plants were not able to use up the complete amount of P supplied by the seeds (on average, P uptake of the zero P variant was 9.8 mg P, while on average 12.2 mg P per pot were added with seeds) during the course of the experiment, causing some P-CAL (0.24 mg/pot) left in the substrate in the zero P control (Figures 5 and 6). An obvious explanation for this is the fact that seed germination capacity was also not 100%, but only 93% (determined by a pre-test). The incomplete utilization of seed P is in line with Nanzer et al. (2014b) who report that utilization of seed P depends on P fertilization rate and plant P uptake, with plants under non-limited P supply growing more vigorously and thus being able to utilize all available P sources (including seed P) better than under limited P conditions.

Fig. 5. Plant available P con­tents (P-CAL) left over by the tested products in the substrate after the fi­nal harvest.

The different variants varied considerably with regard to the amounts of P-CAL left over. The most efficient use of P was observed for the SSP control, which had no P-CAL left at the end of the experiment at P levels 10 and 20. When 40 mg P were supplied, however, about 18% of it could not be used up by the plants during the experiment, indicating to an over-supply with P. For the variants fertilized with SSA 1, SSA 2, SSA 4 and SSA-TC, the amount of P-CAL left in the substrate after the experiment was also low at a P level of 10 mg/pot. At the higher P levels (20 and 40), increasing amounts of P-CAL were left in the substrate. While SSA 1 and SSA 2 both show „left-over“ P-CAL in the same order of magnitude as SSP, SSA 3, SSA 4 and (to a lower extent) SSA-TC, display much higher P-CAL contents after the experiment. This is particularly true for the highest P level (40 mg). Apparently, less than 50% of the P supplied with the fertilizer was used up in the SSA 4 variant at that level, and even less than 20% in the SSA 3 variant. The high P-CAL values are in line with the high plant-availability of SSA 3 and SSA 4 indicated by their high solubility in NAC (Table 1). The NAC-solubility of SSA-TC was clearly lower, which explains the lower amount of „left-over“ P-CAL in the latter case. Obviously, however, not all of the P added in plant available form could be utilized by the plants and was thus left over at the end of the experiment. This is illustrated more clearly when looking at a simplified balance for the system, comparing the input of total and potentially available P with the sum of P used by plants (P uptake) plus P left in the substrate (P-CAL), termed here as „transferred P“ (Fig. 6). Theoretically, the amount of available P added to the system should be found as P taken up by the plants or left over in the substrate in plant available form, i.e. both columns should have more or less the same height. If this is not the case, it can be assumed that the unaccounted-for rest was immobilized in the substrate in some form (and should be found as insoluble soil P, not analyzed in this study). In the case of SSA 3 and 4, indeed both columns are more or less the same heights at all fertilizer levels. In contrast to this, in the SSA 2 variant, the plants clearly did not use up the complete amount of P offered in „available form“, which can be seen by the difference in height between the two columns. This strengthens once more the assumption that a large portion of the P extracted by NAC was not plant available due to its chemical form.

Fig. 6. Comparison of total (= seed plus fertilizer) P input (dotted green line), the share of available P (= available seed P, reflected by P-uptake of the zero P variant, plus NAC-soluble fertilizer P; shown in blue bars) and „transferred P“ (= P uptake by plants plus soluble P (P-CAL) left over in the substrate af­ter the end of the experiment; shown in red/orange bars) for the different fertilizer variants.

While the SSA 4 variant showed a clear decrease in DMY as a result of Na toxicity at the higher P levels (see above), no such effects in DMY were observed for SSA 3. However, as shown in Table 7, SSA 3 (like SSA-TC) had considerably higher Na concentrations (and, in consequence, was associated with a much higher Na load) than the commercial SSP. This was also reflected in elevated Na concentrations in the plant material of all three variants (data not shown here). It is assumed that this increase in Na uptake by the plants also had an important impact on the uptake of other nutrients, such as P, as the Neubauer trial is a closed system with nearly no buffering capacity due to the sandy substrate.

In contrast to SSA 3 and SSA 4, the untreated raw ash (SSA 1) only displayed a NAC solubility of 36.5%. Accordingly, less usable P was available to the system, which was, however, used up by the plants to a higher degree than in the case of the treated ashes. The low amount and share of P-CAL left over after the experiment is in line with this. It can be assumed that higher shares of non plant available P forms are still left in the substrate. SSA 2 also displayed only a low share of soluble P left over in the substrate (= P-CAL), despite of the fact that the fertilizer showed a high solubility in NAC (86.9%). Furthermore, at the higher P levels (20 and 40 mg), a considerable share of the input of available P remains unaccounted for in the „transferred P“ (Fig. 6). This suggests a low stability of the P forms added by SSA 2, which seem to have re-precipitated as unavailable P minerals again when used as fertilizer.

The present investigations with sewage sludge ash that had been deposited for several years confirm this finding. The employed acid leaching procedure with subsequent stepwise raising of the solution pH was able to concentrate the phosphates up to the range of commercially available mineral P fertilizers and to reduce heavy metal concentrations below the limit values of the DüMV.

However, results of the reported trial show that there are still some difficulties to overcome in order to produce viable P fertilizers: The acid leaching procedure alone was not able to reduce the high Al/Fe contents of the products, which caused the preferential formation of non-bioavailable Al-/Fe-phosphates. P uptake achieved by the resulting products remained significantly below that of the commercial mineral P fertilizer used as control.

While an exchange of the cations Al and Fe against Na was able to enhance the plant availability of P, resulting in a P uptake in the same order of magnitude as a commercial P fertilizer, it induced other difficulties related to the resulting increase in Na content, which proved harmful to the plants at the higher fertilization levels. As discussed above, the severe effects observed in the Neubauer experiment were most likely related to the experimental design, using a small volume of a non-buffering sand substrate with a large number of plants, Such a closed, non-buffered system is of course very sensitive to nutrient imbalances, which would probably not occur under field conditions, where plants are growing in a large volume of soil with better buffering capacity. However, these observations highlight the fact that for recycling products – as of course for any other fertilizers as well – it is important to adjust application rates to the actual crop demand and to prevent over-supply in order to minimize the risk of other components to cause imbalances when added to the system at higher concentrations.

It can be concluded that, to make the ash-based recycling fertilizers more attractive, further research is needed to find (or optimize) a procedure which will make the ash-P more plant available and at the same time result in a product which is low in or almost free from growth-reducing contaminants such as sodium or heavy metals, and to deliver a stable plant available P fertilizer product.

Another important question raised by this experiment is the suitability of standard extraction procedures for P recycling fertilizers to estimate their P plant availability. Future research should focus on additional options to describe P solubility of fertilizers more accurately, particularly P sink methods such as Fe-oxide resin bags or DGT.

Finally, the data also revealed that even the untreated ash contains a certain amount of plant available P, which implies that specific measures are necessary when long-term storage of mono-incinerated ash is planned to prevent unwanted P losses to adjacent soils, the groundwater and surface water bodies such as lakes or rivers.