The bioavailability of heavy metals in a sewage sludge amended soil as modelled with PHREEQC

(1)

The bioavailability of heavy metals in a

sewage sludge amended soil as modelled

with PHREEQC

(2)

Abstract

As traditional mineral fertilizers are often unsustainable, Bio Based Fertilizers (BBF) are an important upcoming sustainable fertilizer. BBF’s are fertilizers made from nutrient rich side streams such as sludge, compost and digestate. Nutrient rich urban sewage sludge can be converted to a Bio Based Fertilizer. BBF’s come from different sources so a concern is that they may contain pollutants such as heavy metals in varying amounts, depending on the source. The contamination of soil by heavy metals can pose risks to humans and the ecosystem. This happens through direct ingestion, contact with the contaminated soil, the food chain, via phytotoxicity, drinking of contaminated soil water, food quality, and reduction in land usability for agriculture. In this thesis, the potential bioavailability of heavy metals from BBF’s made from Urban Sewage Sludge (USS) are modelled using the speciation model PHREEQC. USS based fertilizers form great potential if the risks are managed. The case study for this thesis is a potatoes field and a grass field in the Flevo polder in the Netherlands. Samples were taken and chemical analysis was done to be used as base parameters for the model. Heavy metal concentrations of USS based fertilizers were retrieved from literature as input for the model. A sensitivity analysis of pH and ionic solution strength is done. Results showed that with maximum concentrations of heavy metals in sewage sludge, a fertilizer regime of more than 6.9 t ha-1_{exceeds EU regulations for the potatoes field. Bioavailability by speciation was calculated to} be most sensitive to pH for Pb and Cu. Clay adsorption was calculated to be the highest for Pb and Cu, with an overall highest variability to pH for Cd and Zn. Excessive use of assumptions and the lack of other complexations in the model may affect model results. Further research with time series, supplementary models, and validation with field experiments is recommended.

(4)

1. Introduction

1.1 Background and problem formulation

Traditional mineral fertilizers can have a large negative impact on the environment (Zhu & Chen, 2002). In the Netherlands, 60% of the total nitrogen and 40-50% of the total Phosphorus emissions to the surface waters are caused by agriculture. However, sustainability can be improved when fertilizers are made of nutrient-rich side-streams (Bashan & Bashan, 2002). An important nutrient-rich side-stream which is often seen as potentially harmful is Urban Sewage Sludge (USS) (Protano et al., 2020). USS is a nutrient rich residue from sewage treatment plants. As urbanization and industrialization continues in many parts of the world, the volume of USS will only increase. The reuse of these biosolids are cost-effective and environmentally beneficial (Chen et al., 2012). Land application is seen as the preferable option, due to its relatively low economic and energy costs. However, risk should be managed, as USS contain toxic heavy metals and other organic pollutants (Singh & Chen, 2002). Processes like ashing and hydrolyzation remove organic pollutants, while heavy metals survive such treatments. Many heavy metals are not harmful to the human body in small amounts, but exceeding certain concentrations can become harmful. The USS based fertilizers generally contain high amounts of heavy metals, something that has been concluded in studies as early as 1991 (Alloway & Jackson). One of the most important factors of this pollution is the bioavailability of these heavy metals. Bioavailability is the amount of an element that is accessible for adsorption or uptake through the membranes of a plant. Bioavailability of heavy metals is determined by the fraction of free heavy metals present in the soil solution in relation to the total content of heavy metals in the solid phase (Takáč et al., 2009). Pollution happens through direct ingestion, contact with the contaminated soil, the food chain, via phytotoxicity, drinking of contaminated soil water, food quality, and reduction in land usability for agriculture (Wuana & Okieiman, 2011).

These problems led to this research, where several objectives are present. Together with four bachelor thesis students, research into bio-based fertilizers in two agricultural fields in the Netherlands was done. The specific sites are a potatoes field for human consumption and a grass field for dairy production. These theses are based on a larger, EU-wide project called LEX4BIO (LEX4BIO, 2019). The goal of this project is optimization of Bio Based Fertilizers to provide a knowledge basis for policy regarding sustainable agriculture and in turn a sustainable future. The five theses are divided into three main subjects; the first is about the available N, P, and K and the amounts of BBF needed for the right concentrations for agriculture. The second subject is about the potential of commercially available test kits for citizen science. The third subject, where this thesis falls under, is about modelling the use of BBF and its potential for heavy metal pollution.

Modelling the amendment of sewage sludge to soil may give a better understanding on the expected bioavailability and thus toxicity. A model that is widely used to find the equilibrium of all reactions in the soil is the PHREEQC model. The bioavailability of the heavy metals for crop uptake is the focus in this study. Bioavailability is often determined by biological, chemical, and physical factors in the soil (Violante et al., 2010). One of factors that affects bioavailability is the adsorption and desorption on soil mineral surfaces (Violante et al., 2010). Heavy metal adsorption to clay minerals is an important mechanism controlling bioavailability, and this mechanism is pH dependent (Ashworth & Alloway, 2008). Complexation with dissolved organic matter (DOM) affects bioavailability as well, but is not available in the PHREEQC model. Within the model, complexation, precipitation, and leaching is simulated by many equilibrium equations. These are mechanisms that are controlling the solubility, and thus in large part the bioavailability (Halim et al., 2005). The heavy metals from sewage sludge most prone to increase concentrations in soil are Pb, Ni, Cd, Cr, and Cu (Elloumi et al., 2016). In this research, Pb, Cd, Cu, and Zn are the focus as Cr is not a readily available parameter within PHREEQC.

(5)

Lastly, EU regulation state different soil threshold concentrations for human consumption and for animal consumption. The reasoning for this is that potatoes are directly consumed by humans, and grass takes a longer path to human consumption, following the forage-cattle-human food system. The path of toxins in grass to humans is possible via dairy or meat consumption from the cattle that eat the grass. For example, Cr and Pb concentrations in milk were found to be higher when fed with sewage wastewater irrigated grassland (Chary at al., 2008). Path to human consumption, uptake-availability by crop type, and the bioavailability of heavy metals determine the risk to humans, with the last being the main focus of this thesis.

1.2 Research questions

The main research question for this thesis states: “To what extent does the PHREEQC model predict potential heavy metal pollution of an Urban Sewage Sludge (USS) Bio Based Fertilizer (BBF) in two agricultural fields in the Netherlands?”

In order to answer this research question, four sub questions are explored. The first sub question states: “How does Urban Sewage Sludge compare to traditional mineral fertilizers regarding N, P, and heavy metal concentrations?” This question is paramount for use in the model. The average concentrations of N, P, and more importantly the heavy metal concentrations in USS BBF’s are determined and compared to those found in traditional mineral fertilizer.

The second sub question states: “How much speciation and sorption to clay particles of heavy metals are calculated in relation to pH?” . The different inorganic heavy metal species and their fractions are assessed, as well as the fraction of adsorption of heavy metals. Together with pH dependency, these determine the bioavailability of these metals.

The third sub question states: “How sensitive is the model and at what levels does the model predict harmful concentrations?” This is used because different land management with different applications of BBF’s may change pH and ionic solution strength. What’s more, a sensitivity analysis helps with predicting which parameters have the largest effect on the results.

The last sub question states: “What assumptions in the model are made and what are the possible consequences?” For this question, the focus lies on the assumptions made and the consequences of these assumptions as mere modelling is used. Evaluating the assumptions which led to the results, together with a sensitivity analysis, will help to explain certain outcomes.

1.3 Hypothesis

The hypothesis for this study is that the heavy metal concentrations in USS are much higher compared to traditional fertilizers which will heighten the concentrations of heavy metal in the soil. Furthermore, the state and complexation of the metals is expected to indicate the bioavailability. However, it is also expected that heavy metal threshold concentration are not exceeded that easily. Adsorption of heavy metals to clay particles is expected to increase mobility and decrease sorption in soils (Ashworth & Alloway, 2004). pH is expected to largely influence the bioavailability of heavy metals, with a higher availability at a lower pH (Kim et al., 2015). Lastly, it is expected that the potatoes field will be more sensitive to potential pollution as the path to human consumption is shorter.

(6)

2. Data and Methods

2.1 Data collection

2.1.1 Literature

Literature was used to find the difference between mineral fertilizers and USS based fertilizers. The traditional mineral fertilizer used in the site farm had a nitrogen percentage of 46.52 as measured in unpublished data from personal communications with Bram Ebben. Heavy metal concentrations in mineral fertilizers were retrieved from Mortvedt (1996) where measurements were done in different parts of the world. Average Pb concentrations were 10 mg kg-1 _{fertilizer or 48}_{µM kg}-1_{. Average Cd concentrations}

were 25 mg kg-1_{or 222.4 mg kg}-1_{µM kg}-1_{. Average concentrations of Ni were 29 mg kg}-1 _{or 257.9 µM}

kg-1_{. These heavy metals were the only metals that matched de sewage sludge based research.}

Next, the data for the sewage sludge based fertilizer was retrieved from Protano et al. (2020). The measured USS comes from Italian sewage treatment plants. Concentrations of heavy metals, N & P, and carbon in urban sewage sludge were determined in this study with ICP, spectroscopy, and the Walkley and Black (1934) method, respectfully. Min-max and median values are used in the model. The concentrations are available in mg per kg dry weight but are translated to molar per kg dry weight (table 1). The fertilization regime in the study of Protano et al. (2020) was on average 5 tons per hectare per year.

Table (1). Lowest, median and highest heavy metal concentrations in dry sewage sludge (Protano et al., 2020). dw = dry weight. *These heavy metals are not included in the PHREEQC model.

Min Median Max

Heavy metal µM kg -1_dw _{µM kg}-1_dw _{mM kg}-1_dw Cd 0.89 7.12 4.36 Cr* 73.08 775.06 265.41 Cu 744.34 5067.20 1222.74 Hg* 0.25 4.99 4.29 Ni* 88.60 490.69 327.13 Pb 51.64 331.08 91.22 Zn 490.98 12174.98 3628.02 M kg -1 _{M kg}-1 _{M kg}-1 Ntotal 1.07 4.00 19.99 Ptotal 0.13 0.45 4.46 Corganic 22.31 32.05 50.79

The maximum threshold concentrations of heavy metals in topsoil used in this study are the limits for food consumption by the European Directive 86/278/EC (table 2). However, limit values of Cu, and Zn may be permitted to be exceeded by a member state on two conditions; when commercial food is grown

exclusively for animal consumption; or when the pH is consistently higher than 7, but with a maximum exceedance of 50%. Since the field sites have an average of at least 7, the limit values of Cu and Zn are increased with 50%.

(7)

Table (2). Maximum threshold values for heavy metals in soil as set by European Directive 86/278/EC. *Original values are increased with 50% as average pH of field sites > 7.

EU limit values

Heavy metal mg kg -1_{dw soil} _{µM kg}-1_{dw soil}

Pb 300 1448

Cd 3 27

Cu* 210 3305

(8)

2.1.2 Fieldwork

Samples were gathered at a farm in the Flevo polder located in the Netherlands, see figure (1) for a schematic view of the fields and sample location. The soil in these parts of Flevoland was classified as marine clay. This farm is focussed primarily on dairy production but has vegetable fields as well. Two fields were researched and compared, a grass field for hay production for cattle, and a potatoes field for human consumption. The grass field was located directly next to a highway. Both fields were assigned 20 random sampling points with ArcGIS software. At the site, the farmer asked to use a grass field (A) next to the planned grass field which led to an improvisation of a random selected point cloud in the new field. However, as spatial variability was not assessed in this thesis, this circumstance was neglectable. The potatoes field (B) remained as planned. From every data point, two full auger samples at a depth of 25-35 cm were taken. The reason for duplication of each sample point is to account for robustness and a check for consistency of the measurements. Including duplication, this accounted to 80 samples with a mean of around 100 grams per sample. The weather conditions were windy, around 10°C, and with some rainclouds. One sample of field A was lost in the field.

Figure (1): A site map showing the sampling locations on the farm. The windmill construction locations were added as this may be important for results of the measurements. 1. Schematic overview of the field sites. The potatoes field is indicated by A and the grass field is indicated by B. 2. Location of field sites in relation to the province of Flevoland. 3. Location of field sites in relation to the Netherlands.

(9)

2.1.3 Sample preparation

Prior to chemical analysis, 78 samples were prepared in a lab at Science Park at University of Amsterdam. Sample 9 was lost in the field. First, the 200 g samples were split in duplicates to account for sudden outliers that may have occurred due to contamination or technical errors. A solution of 30 g wet sample with 75 mL demineralized water was made and the samples were homogenized with a horizontal shaker table for 120 minutes. The samples were then centrifuged to separate the solution from the particles, before being infiltrated using a vacuum assisted infiltration box with membrane filters of 0.45 µm. This diameter ensures no large particles enter the final solution. This method resulted in a diluted solution of the soil samples in Milli-Q water and were ready for the chemical analysis. Four blank test samples were made to test for contamination.

2.1.4 Chemical analysis

A chemical analysis of the soil samples was done to measure the base concentrations of the soil to be used as input for the model. Standard procedures were used for the following five methods of chemical analysis. These analyses were conducted to determine the most important compounds and elements in the soil for this specific research.

Total Organic Carbon (TOC) analyser. Vario TOC cube, from Elementar GmbH. This analysis method measured the total organic- and inorganic carbon. Determination was done via high temperature combustion.

Inductively Coupled Plasma (ICP) analyser. Optima-8000 ICP-OES, from Perkin Elmer. An ionization source fully decomposed samples into its elements. These elements were transformed to ions. Argon gas was used in this process. This method measured Pb, Ni, Cd, Cr, total P, and total K. It must be stressed that only the free available elements were measured because of the water solution. The metals that were strongly adsorbed to solid matter did not enter the solution. More aggressive extractive methods were not made due to laboratory limitations.

Auto Analyser. San++, from Skalar. This machine used the continuous flow analysis method to measure the following compounds. Chemical reactions with specifics reagents emit different colours on the UV spectrum. It measured PO4, NH4, NO3, NO2, Ntotal.

Electrical Conductivity Meter. Consort C831. Four electrodes were used, with alternating current being applied on the outer electrodes. The potential between the inner pair was measured. This measurement was done before filtration.

Potentiometric pH Meter. Consort C831. The electrical potential between a reference electrode and a pH electrode was measured. This was done before filtration.

(10)

2.2 Modelling

2.2.1 PHREEQC model

The PHREEQC model was used for the chemical speciation modelling. PHREEQC is a free to use product of the U.S. Geological Survey (USGS, 2020). This model, written in the C++ programming language, implements several aqueous models including ion-association models. This model was used to simulate the addition of fertilizers to the soil, and finds an equilibrium of all chemicals introduced into the model. Addition of Urban Sewage Sludge (USS) Bio Based Fertilizers (BBF’s) to the base parameters which were retrieved from the samples were simulated. The means of all the concentrations of both fields were taken. Each field was seen as fully homogenous as this made modelling more manageable. Outliers were taken out if a logical explanation was deduced. For this research, the ion association model and the surface complexation model in PHREEQC were used. See the appendix for the model in script format.

Multiple scenarios with different fertilizing regimes were run in order to determine when threshold values are exceeded. The model results indicated if and when pollution is possible and whether heavy metal concentrations exceed EU-regulations. An important aspect of the model were speciation and complexion of the metals. These determined the bioavailability of the metals, which in turn influenced the concentration in the crops and heavy metal washout, i.e. pollution.

The precipitated heavy metals as calculated by the ion association model together with the adsorbed metals as calculated by the surface complexation model were combined. The fraction of these specific combined metal concentrations relative to the total amount of specific metal concentrations determined the fraction non-bioavailable heavy metals.

2.2.2 Parameter specification

Since all concentrations were presented per kg dry weight, a few assumptions were necessary to run the model. The model was simplified to one hectare of soil. The soil was assumed to be uniformly distributed throughout the plough depth of 30cm. This created a total volume of 3000 m3_{. Following the average moisture} content of the soil samples of 25%, this resulted in an estimated total water volume of 750,000 L per hectare. This was used in combination with a fertilization regime to calculate to molal. The highest concentrations of each compound in the sewage sludge of Protano et al. (2020) was used since the risk of sewage sludge is assessed, therefore the possible maximum should be taken.

Additionally, a surface complexation model was used in PHREEQC and was calculated after the ion association model. The PHREEQC database contains thermodynamic data for a diffuse-double-layer surface which was derived from Dzombak & Morel (1990). The specific diffuse layer used was the Donnan diffuse layer. Taken from the example in the PHREEQC database, the specific values to represent clay minerals were a debye length of 3.4 with a limit DDL of 0.9 (USGS, 2020). A site density of 1 site/nm2_{was used, which is} widely used in literature, and falls in the range of 0.54-1.16 nm2_{/site from Nadeau (1985). Vos & Groenwold} (1986) classified the soil in the Flevo polder, at the same depth of the samples in this study, as clay loam with a clay (<2 µm) percentage of 29%. Equation (1) by Knadel et al. (2018) was used to calculate the specific surface area. For this percentage, the estimated specific area per gram is 75.5 m2 _g-1_{. A mass of solid of 3000} g L-1_{was deduced from the Flevoland field sites average water content of 25%. This mass of solid times the} fraction of clay content of 29% resulted in a mass of clay of 870 g L-1_.

(11)

2.3 Sensitivity analysis

2.3.1 Method

Here, the method of sensitivity analysis is discussed. The method used was the one-at-a-time (OAT) method, derived from Morris (1991). There were three main steps in the OAT method:

1. Change one input variable and keep the other values the same as the baseline. 2. Save the results.

3. Change the variable to its original value and change the next variable.

Before any of the values were changed, a table with variables and their values was made. This is the sensitivity analysis plan and contained only the variables for which was decided that their effects were the most paramount in the model.

The simplicity of the OAT method ensured a few things. Firstly, this method was fairly time-effective as cross-variability of parameters was not accounted for making the amount of model runs manageable. Secondly, since only one variable was changed at a time, model results could be immediately explained by the adjusted input. Thirdly, all model results were fixed to the baseline values which increased the comparability of the results. As with every method, the OAT sensitivity analysis method had its shortcomings. These are discussed in the discussion chapter.

A sensitivity analysis of pH on the fraction of free ions was used since free metal ion concentrations is a key factor to determine bioavailability and toxicity (Yi et al., 2007). The fractions of free metal ions were calculated using equation (2), and was done for each individual heavy metal. Note that Free Ions and Ions Introduced represent the concentrations for each specific heavy metal. A sensitivity analysis on the adsorption on clay particles was used as well. The same equation was used, but Free Ions was replaced with Clay Adsorbed Ions.

𝐹𝑟𝑒𝑒 𝐼𝑜𝑛𝑠 (%) = 𝐹𝑟𝑒𝑒 𝐼𝑜𝑛𝑠𝑡𝑜𝑡𝑎𝑙 (𝑀)

𝐼𝑜𝑛𝑠 𝐼𝑛𝑡𝑟𝑜𝑑𝑢𝑐𝑒𝑑_{𝑡𝑜𝑡𝑎𝑙} (𝑀)∗ 100% Eq. (2).

2.3.2 Parameter input range

There were two main parameter ranges which contributed to the sensitivity analyses. These were pH and ionic strength. The pH for free ions was directly used as input for the model with the values 6 to 8.5 with steps of 0.5. The pH analysis on the clay adsorption had a wider range to determine patterns, specifically 3 to 8. The ionic strength changed as the input followed different fertilization regimes. The average fertilization regimes of Protano et al. (2020) varied from 3.89 t ha-1_{to 6.16 t ha}-1_{per year. A maximum of 15 t ha}-1_{for a} period of 3 years was in agreement with Italian legislation. In PHREEQC, fertilization regimes of 2.5 t ha-1_, 5 t ha-1_{and 7.5 t ha}-1_{were used. A regression line was fitted to calculate at what point threshold values were} reached. Three values were used to account for non-linearity in the model results.

(12)

3. Results

3.1 Chemical analysis results

See table 3 for the results of the chemical analysis. All sample results were recalculated into M kg -1 _dry weight soil or its milli or micro equivalent. All sample measurements were averaged out for field A and field B. Limit of detection values were taken up as 0.

Since the PHREEQC model does not include non-specific organic carbon, the total organic and total inorganic carbon were combined. The potatoes field (field A) and the grass field (field B) showed an average of 9.40 and 9.32 mM kg -1 _{dw soil, respectfully.}

Native heavy metal concentrations in the soil of both fields were low. The limit of detection was often not reached. Overall, NPK concentrations were all higher in field B. N concentrations for field A and field B were 31.74 and 47.67 µM kg -1 _{dw soil, respectfully. P concentrations were 1.28 and 1.59 µM kg}-1 _{dw soil,} respectfully. K concentrations between Field A and Field B were 5.15 and 20.4 µM kg -1 _{dw soil,}

respectfully. All can be found in table (3). Specific PO4, NH4, NO3, and NO2 are not included in the results as these cannot be specified in PHREEQC when total N is used.

The average pH of field A and field B were 7.76 and 7.50, respectfully. The average water content of both fields are close to 25%. See table (3) for the results.

Field A (Potatoes) Field B (Grass)

Total Carbon mM kg -1 _{dw soil} _9.40 _9.32

N µM kg -1 _{dw soil} _31.78 _47.67

P µM kg -1 _{dw soil} _1.28 _1.59

K µM kg -1 _{dw soil} _5.15 _20.45

pH -log[H3O+] 7.76 7.50

Water Content % 23.9 26.9

Table (3). All concentrations are means of all field samples. Ni, Pb, Cu and Cd averages were all < 0.01 µM kg -1 _{dw soil as the limit of detection was often not reached. dw = dry weight.}

(13)

3.2 Model results

3.2.1 Sewage sludge based fertilizer

The following non-bioavailable species resulted from the ion-association model. For lead the species PbCO3 and Pb(OH)2 were calculated to form. For copper, the species CuCO3 and Cu(OH)2 were calculated to form. ZnCO3 was the only non-bio-available zinc species to form, and for cadmium, the species CdCO3 and Cd(OH)2 were expected to form. The results of the ionic strength analysis can be found in table (4). Table (4). Results from modelling a sewage sludge based fertilizer regime with the maximum

concentrations (see table 1) from Protano et al., (2020). *These forms are used as the only other form calculated is the non-bioavailable gas N2.

Regime of 2.5 t ha-1 Regime of 5 t ha-1 Regime of 7.5 t ha-1 EU Maximum threshold concentration mM L-1 _{mM L}-1 _{mM L}-1

Ionic Strength Field A (Potatoes) 82.7 163.1 243.3

Field B (Grass) 85.5 168.3 250.9 µM kg-1_dw soil µM kg-1_dw soil µM kg-1_dw soil µM kg-1_dw soil Pb Field A (Potatoes) 16.00 45 81 1448 Field B (Grass) 14.77 42 77 1448 Cd Field A (Potatoes) 3.60 7.09 10.79 27 Field B (Grass) 3.61 7.21 10.82 27 Cu Field A (Potatoes) 1019 2038 3058 3305 Field B (Grass) 1019 2037 3057 3305 Zn Field A (Potatoes) 2209 4835 7591 6883 Field B (Grass) 2207 4821 7549 6883 NH4_{+ &} NH3_* Field A(Potatoes) 13775 26700 39450 Field B (Grass) 14560 28300 41825

All heavy metal concentrations followed a very close to linear pattern with fertilization regime. Non-linear patterns were neglectable. The linear formulas 3.1-3.4 were derived from table (4), which were used to calculate the maximum fertilization regimes per heavy metal. The EU maximum threshold concentrations were used to find at what regime these are exceeded (European Directive 86/278/EC). Y is the

concentration of heavy metal in µM kg-1_{dw soil, and X is the fertilization regime, in t ha}-1_.

Regimes of 45 t ha-1_{, 18.8 t ha}-1_{, 8.1 t ha}-1_{, and 6.9 t ha}-1_{are needed to reach the EU maximum threshold} concentrations for Pb, Cd, Cu, and Zn, respectfully.

Pb: ŷ = 32.12𝑋 + 2.43333 Eq. (3.1)

(14)

3.2.2 pH Sensitivity analysis

Firstly, the sensitivity analysis of pH on the bioavailable heavy metal ions can be found in figure (2). The ion association model was the only model used. The fraction of free ion species relative to the total concentration of introduced heavy metal was calculated and plotted. Cu and Pb showed the most

variability, with a low of ~40% availability for Pb at a pH of 6.5 and a low of ~39% availability of Cu at a pH of 7. Cd showed the least variability and the overall highest availability. Overall, a pH around 7 results in the lowest availability of free Pb, Cu, and Zn ion species.

Figure (2). Sensitivity analysis of percentage of bioavailable heavy metal ions at pH 6 - 8.5. Only calculated with the ion association model.

A pH sensitivity analysis on the adsorption of the heavy metals on clay is also conducted. The results can be found in figure (3). Pb showed an adsorption of ~99% at all pH levels. Adsorption of Cu was

consistently high, beginning at ~68% at a pH of 3 and forming a steady adsorption from pH 5 onwards at ~93%. Zn had a highest adsorption at a pH of 7 with a decline in both directions, with the lowest adsorption at pH 3 at around ~3%. Cd showed a consistent increase in adsorption from its low of ~1% at a pH of 3 to a high of ~99% at pH 8, following an s-curve pattern.

30 40 50 60 70 80 90 100 6 6 . 5 7 7 . 5 8 8 . 5 P er cent age of bi oav ai labl e ions (%) pH Pb Cd Cu Zn 0 10 20 30 40 50 60 70 80 90 100 3 4 5 6 7 8 9

P

erc

entag

e

adosorpe

d

by

c

lay

(%

)

pH Pb Cd Cu Zn

(15)

4. Discussion

4.1 Findings

In this paragraph the findings are discussed by answering the sub questions in chronological order as found in subchapter 1.2. Regarding the types of fertilizer, the first difference is that the composition of traditional mineral fertilizer that is currently used on the field sites contain ~46% N, compared to ~28% N at a maximum in sewage sludge (Protano et al., 2019). The second difference is that mineral fertilizer contains considerably less heavy metal concentrations for at least Pb, Cd, and Ni (Protano et al., 2020; Mortvedt, 1996). However, the fractions of heavy metal introduced as organic or inorganic forms may differ and could have an impact on the bioavailability once introduced in the soil. What’s more, heavy metal

concentrations in inorganic NPK fertilizers vary significantly with N:P:K ratio (Milinovic et al., 2015). On average, N and P are found to be lower and heavy metal concentrations to be higher in sewage sludge compared to traditional mineral fertilizers in all found literature (Protano et al., 2020; Siebielec & Stuczynski 2008; Mortvedt, 1996).

Next, the speciation of heavy metals and sorption to clay minerals are one of the factors that determine the bioavailability. The non-bioavailable species PbCO3, Pb(OH)2, CuCO3, Cu(OH)2, and ZnCO3 were expected to form. The fractions of bio available Pb and Cu species were the most depended on pH. The placement of the non-bioavailable Pb and Cu species curve on the pH axes was not in line with literature, but the form of the curve was (Powell, 2009). This may be due to the surface charge, ionic strength, the simplification of compounds introduced into the model, the presence of other heavy metals, or the lack of organic matter to form complexes in the model. Cd speciation was predominantly bioavailable, as in line with literature, which found free Cd+ do be predominant in pH 2-9 (Ozel, 2012). The high concentration of free Zn+ ions is at least partly in line with literature, as Knight et al (1998) found a free Zn+ ion fraction of higher than 80% on a pH scale from 4 – 6.9. However, it should be noted that these fractions do not include sorption of the heavy metals to either clay minerals, solid forms, or organic matter complexes. For the clay adsorption, Pb was overall higher than 95% adsorption at all pH values, which is in line with Lukman et al. (2013) where they measured adsorption of a multicomponent heavy metal solution on natural clays in relation to pH. Cd formed a curve (figure 3) that was in agreement with results from this same study. Hydrogen ions are more adsorbed to the clay surface at a lower pH, which supresses the Cd adsorption (Lukman et al. ,2013). The curve of Zn was in agreement with their measurements as well., but percentages differ. Cd and Zn had overall the lowest adsorption by clay, which is suggested to be the result of their relative low electronegativity (Futalan et al., 2011). Cu showed a slight difference, but was consistently high as well. The discrepancies may be explained by the difference in competition since the total ionic solution strength, concentration, and the variation of heavy metals differs with the study by Lukman et al. (2013).

Concerning the sensitivity of the model, the results showed that ionic solution strength greatly determined the concentrations of bioavailable heavy metals in the soil. In its turn, the ionic solution strength was determined by the fertilization regime, water content of the soil, and the concentrations of compounds per kg fertilizer. Tonnage fertilizer divided by the total amount of water in all the soil steered the ionic solution strength. However, the ionic solution strength did not influence the fraction of non-bioavailable species substantially, as modelled by PHREEQC. This was indicated by the linearity that resulted from the

sensitivity analysis of ionic strength and fraction bioavailable heavy metals. In literature, ionic strength was found to determine the intensity of competition by other cations for the bonding sites, but followed close to

(16)

respectively. It should be noted that these are calculated with the maximum concentrations of heavy metals in USS as measured by Protano et al. (2020). Assuming these maximum concentrations are present, a recommended dose of maximum 6.9 t ha-1 _{is advised. However, maximum concentrations are presumably} not often present, as median concentrations were much lower.

4.2 Limitations and weaknesses

In this subchapter the limitations of this research and the known weaknesses are discussed.

4.2.1 Data Collection

The first weakness in the data collection was the disappearance of one of the samples of field B. It was assumed that the sample bag was lost in the field. This made the total sample points 9 instead of 10 for this field. Furthermore, one of the four blank test samples had K, P, Ni, and Pb present. Although in very low concentrations, this may have indicated contamination.

4.2.2 Model and Sensitivity Analysis

Here, the last sub question is answered. Firstly, it was assumed that there exists a perfect solution in the tilled topsoil of one hectare with a depth of 30 cm. This decision may have greatly affected the total concentrations as in reality, added fertilizer may not form a perfect solution, may not distribute uniformly, and is not limited by boundaries. Secondly, other compounds in the soil that are not measured and not used in the model may in reality alter the chemical speciation and concentrations of the heavy metals, as well as the surface charge. Thirdly, organic adsorption and complexation was not used in the model, which is seen as a major contributor to the bioavailability of heavy metals (Ashworth & Alloway, 2008).

For the clay surface model, aggregates and preferential flow paths were not taken into account. In the model it was assumed that every clay surface was directly available for adsorption. In reality, weather conditions, DOM, and root growth affect aggregation of soil particles which may limit the specific surface area (Chenu et al., 2000). Preferential flow path is another phenomenon which may greatly influence adsorption potential. >90% of vertical flow may exist within fractures within the soil and only a few percent within the clay matrix (Jorgensen et al., 2002). Furthermore, the variables used for the donnan diffuse layer parameters were taken from an example of clay adsorption in PHREEQC’s users guide. The differences between this study and that specific example are unknown. It should be noted that with extensive literature exploration the information needed was not found.

Regarding the one-at-a-time method used for the sensitivity analysis, the major disadvantage was that interactions could not be estimated. As only one variable was changed with each model run, potential effect by interactions was undetectable. What’s more, optimal settings in between variable changes could have been missed. This could change, for example, the curves of the pH analyses.

4.3 Recommendations

A recommendation for further research is to calculate multiple years of fertilization using a time series model. Additionally, inclusion of models about adsorption to organic matter and ferric oxides should give a better understanding about adsorption. This in combination with models of heavy metal uptake by plants and washout to groundwater may give a better understanding of not only the entering of heavy metals into the system, but the leaving of heavy metals out of the system as well. It is also recommended to use a more aggressive extraction method with a different solution to extract adsorbed metals from the soil particles. Furthermore, field experiments with USS amendment could help with validation of the model.

(17)

5. Conclusion

In this chapter a conclusion is formulated by answering the main research question. This question states: “To what extent does the PHREEQC model predict potential heavy metal pollution of an Urban Sewage Sludge (USS) Bio Based Fertilizer (BBF) in two agricultural fields in the Netherlands?”. While the PHREEQC model had some shortcomings when it comes to organic complexation of heavy metals, it was still quite useful to predict speciation by the ion association model. With the data and models available, PHREEQC predicted heavy metal pollution if the maximum concentrations of heavy metals in USS by Protano et al. (2020) are present, together with a regime of at least 6.9 t ha-1_{. Amendment to grass fields for dairy production} may be higher as established by the EU, but specific threshold concentrations were unknown. (European Directive 86/278/EC). Heightening fertilization regime followed pollution thresholds by the European union for Zn > Cu > Cd > Pb, in that order. Fractions of bioavailable heavy metal species followed a linear pattern with ionic solution strength, with Pb and Cu being the most pH dependant. Clay adsorption was strongly pH dependent for Zn and Cd. However, the model may have been over-simplified by the use of assumptions and lack of solid-solution and organic complexation models. Further research with chemical analyses with more aggressive extraction of heavy metals, a model that includes organic complexation, and field experiments with USS amendment to be used as validation is recommended.

(18)

Literature

Alloway, B. J., & Jackson, A. P. (1991). The behaviour of heavy metals in sewage sludge-amended soils. Science of the Total Environment, 100, 151-176.

Ashworth, D. J., & Alloway, B. J. (2008). Influence of dissolved organic matter on the solubility of heavy metals in sewage‐sludge‐amended soils. Communications in Soil Science and Plant Analysis, 39(3-4), 538-550. Bradl, H. B. (2004). Adsorption of heavy metal ions on soils and soils constituents. Journal of colloid and interface science, 277(1), 1-18.

Calabi-Floody, M., Medina, J., Rumpel, C., Condron, L. M., Hernandez, M., Dumont, M., & de la Luz Mora, M. (2018). Smart fertilizers as a strategy for sustainable agriculture. Advances in agronomy, 147, 119-157.

Chary, N. S., Kamala, C. T., & Raj, D. S. S. (2008). Assessing risk of heavy metals from consuming food grown on sewage irrigated soils and food chain transfer. Ecotoxicology and environmental safety, 69(3), 513-524.

Chenu, C., Le Bissonnais, Y., & Arrouays, D. (2000). Organic matter influence on clay wettability and soil aggregate stability. Soil Science Society of America Journal, 64(4), 1479-1486.

De-Bashan, L. E., & Bashan, Y. (2004). Recent advances in removing phosphorus from wastewater and its future use as fertilizer (1997–2003). Water research, 38(19), 4222-4246.

Chen, H., Yan, S. H., Ye, Z. L., Meng, H. J., & Zhu, Y. G. (2012). Utilization of urban sewage sludge: Chinese perspectives. Environmental Science and Pollution Research, 19(5), 1454-1463.

Dzombak, D. A., & Morel, F. M. (1990). Surface complexation modeling: hydrous ferric oxide. John Wiley & Sons.

Elloumi, N., Jerbi, B., Zouari, M., Abdallah, F. B., Ayadi, H., & Kallel, M. (2016). Effects of sewage sludge fertilizer on heavy metal accumulation and consequent responses of sunflower (Helianthus annuus).

Environmental Science and Pollution Research, 23(20), 20168-20177.

Halim, C. E., Short, S. A., Scott, J. A., Amal, R., & Low, G. (2005). Modelling the leaching of Pb, Cd, As, and Cr from cementitious waste using PHREEQC. Journal of Hazardous Materials, 125(1-3), 45-61.

Intawongse, M., & Dean, J. R. (2006). Uptake of heavy metals by vegetable plants grown on contaminated soil and their bioavailability in the human gastrointestinal tract. Food additives and contaminants, 23(1), 36-48.

Kim, R. Y., Yoon, J. K., Kim, T. S., Yang, J. E., Owens, G., & Kim, K. R. (2015). Bioavailability of heavy metals in soils: definitions and practical implementation—a critical review. Environmental geochemistry and health, 37(6), 1041-1061.

Knadel, M., Arthur, E., Weber, P., Moldrup, P., Greve, M. H., Chrysodonta, Z. P., & De Jonge, L. W. (2018). Soil specific surface area determination by visible near‐infrared spectroscopy. Soil Science Society of America Journal, 82(5), 1046-1056.

LEX4BIO, (2019). LEX4BIO project. Funded by the European Union Horizon 2020 research and innovation program. Retrieved from https://www.lex4bio.eu/#

Lukman, S., Essa, M. H., Mu'azu, N. D., Bukhari, A., & Basheer, C. (2013). Adsorption and desorption of heavy metals onto natural clay material: influence of initial pH. Journal of Environmental Science and Technology, 6(1), 1-15.

Merdy, P., Gharbi, L. T., & Lucas, Y. (2009). Pb, Cu and Cr interactions with soil: sorption experiments and modelling. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 347(1-3), 192-199.

Milinovic, J., Lukic, V., Nikolic-Mandic, S., & Stojanovic, D. (2015). Concentrations of heavy metals in NPK fertilizers imported in Serbia.

(19)

Mortvedt, J. J. (1996). Heavy metal contaminants in inorganic and organic fertilizers. In Fertilizers and environment (pp. 5-11). Springer, Dordrecht.

Nadeau, P. H. (1985). The physical dimensions of fundamental clay particles. Clay Minerals, 20(4), 499-514.

Ozel, H. U. (2012). Biosorption of Cd (II) ions by nordmann fir cones. Fresenius Environmental Bulletin, 21(9), 2527-2535.

Powell, K. J., Brown, P. L., Byrne, R. H., Gajda, T., Hefter, G., Leuz, A. K., ... & Wanner, H. (2009). Chemical speciation of environmentally significant metals with inorganic ligands. Part 3: The Pb2++ OH–, Cl–, CO32–, SO42–, and PO43–systems (IUPAC Technical Report). Pure and Applied Chemistry, 81(12), 2425-2476.

Protano, G., Baroni, F., Di Lella, L. A., Mazzoni, A., Nannoni, F., & Papale, A. (2020). How do properties and heavy metal levels change in soils fertilized with regulated doses of urban sewage sludge in the framework of a real agronomic treatment program?. Journal of Soils and Sediments, 20(3), 1383-1394.

Siebielec, G., & Stuczynski, T. (2008). Trace metal aspects of biosolids use. CD Proceedings of Protection and Restoration of the Environment IX, Kefalonia, Greece, 7.

Singh, R. P., & Agrawal, M. (2008). Potential benefits and risks of land application of sewage sludge. Waste management, 28(2), 347-358.

Takáč, P., Szabová, T., Kozáková, Ľ., & Benková, M. (2009). Heavy metals and their bioavailability from soils in the long-term polluted Central Spiš region of SR. Plant, Soil and Environment, 55(4), 167-172.

Tóth, G., Hermann, T., Da Silva, M. R., & Montanarella, L. (2016). Heavy metals in agricultural soils of the European Union with implications for food safety. Environment international, 88, 299-309.

USGS, (2020). PHREEQC model version 3. Retrieved from https://www.usgs.gov/software/phreeqc-version-3

Violante, A., Cozzolino, V., Perelomov, L., Caporale, A. G., & Pigna, M. (2010). Mobility and bioavailability of heavy metals and metalloids in soil environments. Journal of soil science and plant nutrition, 10(3), 268-292.

Vos, J., & Groenwold, J. (1986). Root growth of potato crops on a marine-clay soil. Plant and Soil, 94(1), 17-33.

Wuana, R. A., & Okieimen, F. E. (2011). Heavy metals in contaminated soils: a review of sources, chemistry, risks and best available strategies for remediation. International Scholarly Research Notices, 2011.

Walkley, A., & Black, I. A. (1934). An examination of the Degtjareff method for determining soil organic matter, and a proposed modification of the chromic acid titration method. Soil science, 37(1), 29-38.

Yi, L., Hong, Y., Wang, D., & Zhu, Y. (2007). Determination of free heavy metal ion concentrations in soils around a cadmium rich zinc deposit. Geochemical Journal, 41(4), 235-240.

Zhu, Z. L., & Chen, D. L. (2002). Nitrogen fertilizer use in China–Contributions to food production, impacts on the environment and best management strategies. Nutrient Cycling in Agroecosystems, 63(2), 117-127.

(20)

Data Repository

The results from the chemical analysis can be found with the following hyperlink. The specific PHREEQC scripts can be found in the appendix.

https://drive.google.com/drive/folders/1cXW2pIPpi-ADnX5AANAkhv1Y2szb3TnX?usp=sharing

Appendix

1. Field A, 5 t/ha SOLUTION 1 temp 15 pH 7.76 pe 4 redox pe units umol/kgw density 1 C 338.5785252 mMol/kgw Cd 0.029060285 mMol/kgw Cu 8.151575237 mMol/kgw K 2.12 N 133.2695543 mMol/kgw P 29.70233099 mMol/kgw Pb 0.608108108 mMol/kgw Zn 24.18680534 mMol/kgw -water 1 # kg SURFACE 1 -sites DENSITY Hfo_sOH 1 75.5 870 -donnan debye_lengths 3.4 limit_ddl 0.9 END 1. Field B, 5 t/ha SOLUTION 1 temp 15 pH 7.50 pe 4 redox pe units umol/kgw density 1 C 338.5785252 mMol/kgw Cd 0.029060285 mMol/kgw Cu 8.151575237 mMol/kgw K 3.18 N 133.2695543 mMol/kgw P 29.70233099 mMol/kgw Pb 0.608108108 mMol/kgw Zn 24.18680534 mMol/kgw -water 1 # kg SURFACE 1 -sites DENSITY Hfo_sOH 1 75.5 870 -donnan debye_lengths 3.4 limit_ddl 0.9 END

(21)

Field A, 2.5 t/ha SOLUTION 1 temp 15 pH 7.76 pe 4 charge redox pe units umol/kgw density 1 C 169.2892626 mMol/kgw Cd 0.014530143 mMol/kgw Cu 4.075787618 mMol/kgw K 1.06 N 66.63477717 mMol/kgw P 14.85116549 mMol/kgw Pb 0.304054054 mMol/kgw Zn 12.09340267 mMol/kgw -water 1 # kg SURFACE 1 -sites DENSITY Hfo_sOH 1 75.5 870 -donnan debye_lengths 3.4 limit_ddl 0.9 END Field B, 2.5 t/ha SOLUTION 1 temp 15 pH 7.50 pe 4 charge redox pe units umol/kgw density 1 C 169.2892626 mMol/kgw Cd 0.014530143 mMol/kgw Cu 4.075787618 mMol/kgw K 1.59 N 66.63477717 mMol/kgw P 14.85116549 mMol/kgw Pb 0.304054054 mMol/kgw Zn 12.09340267 mMol/kgw -water 1 # kg SURFACE 1 -sites DENSITY Hfo_sOH 1 75.5 870 -donnan debye_lengths 3.4 limit_ddl 0.9 END

(22)

Field A, 7.5 t/ha SOLUTION 1 temp 15 pH 7.76 pe 4 charge redox pe units umol/kgw density 1 C 507.8677879 mMol/kgw Cd 0.043590428 mMol/kgw Cu 12.22736286 mMol/kgw K 3.18 N 199.9043315 mMol/kgw P 44.55349648 mMol/kgw Pb 0.912162162 mMol/kgw Zn 36.28020801 mMol/kgw -water 1 # kg SURFACE 1 -sites DENSITY Hfo_sOH 1 75.5 870 -donnan debye_lengths 3.4 limit_ddl 0.9 END Field B, 7.5 t/ha SOLUTION 1 temp 15 pH 7.50 pe 4 charge redox pe units umol/kgw density 1 C 507.8677879 mMol/kgw Cd 0.043590428 mMol/kgw Cu 12.22736286 mMol/kgw K 4.7670225 N 199.9043315 mMol/kgw P 44.55349648 mMol/kgw Pb 0.912162162 mMol/kgw Zn 36.28020801 mMol/kgw -water 1 # kg SURFACE 1 -sites DENSITY Hfo_sOH 1 75.5 870 -donnan debye_lengths 3.4 limit_ddl 0.9 END

The bioavailability of heavy metals in a sewage sludge amended soil as modelled with PHREEQC