3.1. Experiment 1—Adsorption Behavior of Six Phosphonates in the Presence of and Absence of CaII
Figure 2 enables a comparison of the adsorption behavior of six different phosphonates at molar Ca:phosphonate ratios of 0:1 and 2:1, with a contact time of seven days and different pH values. Additionally, the removal rates in the absence of GFH are shown, in order to determine the relevance of precipitation. The phosphonates in
Figure 2 were sorted in the order of the increasing number of phosphonate groups (PG): HPAA (1 PG), PBTC (1 PG), HEDP (2 PG), NTMP (3 PG), EDTMP (4 PG), DTPMP (5 PG).
For all phosphonates, except HEDP, the removal rates decreased with increasing pH to a similar extent (removal rates at pH
target 5 and 12): HPAA from 67% to 27%, PBTC 65% to 23%, NTMP 74% to 23%, EDTMP 59% to 24%, DTPMP 66% to 24%. The difference in the removal rate between the solutions in the presence of and absence of Ca
II was always less than 13%. Therefore, no major influence of Ca
II on phosphonate removal was observed. This stands in contradiction to several previous studies. Nowack and Stone [
21] investigated the adsorption of HEDP, NTMP, EDTMP, and DTPMP on goethite and found that excess Ca
II concentrations significantly increased the maximum loading. Even at an equimolar Ca:NTMP ratio, the maximum loading for NTMP almost doubled. Similarly, Boels et al. [
4] observed a near doubling of the maximum loading of GFH at a molar Ca:NTMP ratio of 2:1. At a molar ratio of 60:1, it was possible to increase the loading even further. Rott et al. [
27] also found a positive influence of Ca
II on the adsorption of NTMP and DTPMP on magnetic adsorbent particles (ZnFeZr-oxyhydroxide). In addition, Chen et al. [
17] found a positive influence of the hardness ions calcium and magnesium on NTMP adsorption on GFH. The aforementioned studies attribute this behavior mainly to the potential formation of ternary complexes. Ternary complexes can build a bridge between the GFH surface and phosphonates, leading to an increased adsorption [
4,
19,
20].
In the batches without GFH, no removal could be detected. Thus, for this experiment, precipitation can be excluded as the cause of phosphorus removal. HEDP, however, deviated clearly from the behavior of the other phosphonates. In the presence of GFH, it showed a removal of 60% (pHtarget 5) to 53% (pHtarget 12), whereas the removal at pHtarget 8 to 10 was > 90%.
A closer look at the behavior of HEDP reveals two particularly noteworthy aspects. First, with a molar Ca:phosphonate ratio of 2:1, phosphorus was removed even in the absence of GFH, and second, in the presence of GFH, phosphorus was removed without the addition of Ca
II. In the latter case, about 67% of the phosphorus was removed at pH 8.0, about 81% at pH 9.1 and 10.1, and about 76% at pH 12.0. The removal of phosphorus even without the dosing of GFH suggests precipitation of CaCO
3 or Ca-HEDP complexes. According to calculations performed with PHREEQC, at a calcium concentration of 156.2 µmol/L, more than 7.2 mmol/L of (hydrogen-)carbonate would be necessary to precipitate CaCO
3 at pH 8.0. These calculations do not consider HEDP, which complexes at least parts of the Ca
II and additionally inhibits precipitation of CaCO
3 [
12,
28]. Since deionized water was used in the experiments, and the samples were rotated in closed centrifuge tubes (closed system), the (hydrogen-)carbonate content (from airborne CO
2) of the solutions is expected to be low. Therefore, the formation of CaCO
3 is unlikely.
Another explanation for the observed HEDP elimination in the absence of GFH could be the precipitation of Ca-HEDP complexes. Several researchers have investigated various different calcium-phosphonate precipitates, such as Ca-HEDP [
29], Ca-NTMP [
16,
30], and Ca-DTPMP [
31]. However, these studies were conducted using relatively high phosphonate concentrations. In another study, Zhang et al. [
32] found more Ca-phosphonate precipitation for HEDP as compared with other phosphonates. The authors concluded that the order of solubility of the Ca-phosphonate complexes was as follows: PBTC > DTPMP > EDTMP > NTMP > HEDP. This corresponds with the results of the current study and the findings of Amjad et al. [
33], who found a calcium ion tolerance of PBTC >> HEDP.
Interestingly, between pH 8 and 10, a peak in phosphorus removal was observed without the addition of CaII. The reason for this could be the dissolution of calcium from the GFH. The re-dissolved CaII could then lead to precipitation. To investigate this in more detail, a 0.01 M CAPSO solution at pH 9 with 0.2 g/L GFH was rotated with a contact time of seven days (without a phosphonate). A Ca2+ concentration of 11.0 ± 1.1 mg/L could then be observed in the membrane-filtered supernatant. According to the manufacturer’s statement specifying ≥12–19% calcium content in the GFH, a dosage of 0.2 g/L GFH would result in a maximum calcium concentration in the solution of 24 to 38 mg/L, if completely re-dissolved. Therefore, it can be assumed that approx. 29% to 46% of the CaII was re-dissolved.
Those measurements may explain why no positive effect of CaII on the removal of phosphorus was observed with the other five phosphonates. A calcium concentration of 11.0 mg/L (as re-dissolved from the GFH) already corresponds to the following molar Ca:phosphonate ratios: Ca:HPAA 2.7:1, Ca:PBTC 4.6:1, Ca:HEDP 3.5:1, Ca:NTMP 5.1:1, Ca:EDTMP 7.5:1, and Ca:DTPMP 9.9:1. Therefore, although no CaII was added, the positive effect of CaII had probably already been achieved by re-dissolution from the GFH. This shows the important role of the CaCO3 content of the GFH and also explains the deviation between these results and those found in previous publications.
In conclusion, Experiment 1 yielded two findings: First, the re-dissolved calcium from the GFH had a positive effect on phosphonate adsorption, possibly attributable to the formation of ternary complexes. Second, at the given conditions, HEDP was found to precipitate presumably as Ca-phosphonate complexes, which also increases its elimination rate.
3.2. Experiment 2—Adsorption Behavior of NTMP and DTPMP in the Presence of CaII in Higher Concentrations
The aim of Experiment 2 was to investigate the influence of Ca
II in higher concentrations than those in Experiment 1 on the adsorption of NTMP and DTPMP on GFH.
Figure 3 shows the results of batches with NTMP and DTPMP in the presence of and absence of GFH over different pH values from 5 to 12.
According to
Figure 3b, in the batches without adsorbent, no removal of NTMP took place up to a molar Ca:NTMP ratio of 7.33:1. A different behavior was observed when Ca
II was added in a molar Ca:NTMP ratio of 18.33:1. NTMP was removed at particular pH values: ~89% at pH 8.1, ~93% at pH 9.0, ~98% at pH 9.8, and ~28% at pH 12.0. A possible explanation for this behavior could be CaCO
3 or Ca-NTMP precipitation. According to calculations performed with PHREEQC, more than 0.8 mmol/L of (hydrogen-)carbonate would be necessary at a calcium concentration of 986.3 µmol/L to precipitate CaCO
3 at pH 8.1. These calculations do not consider NTMP, which complexes at least parts of the Ca
II and inhibits precipitation of CaCO
3 [
12,
13]. The (hydrogen-)carbonate content of the solutions should be low due to the use of deionized water in the experiments, and due to the rotation of the samples in closed centrifuge tubes (closed system). Thus, the formation of CaCO
3 is unlikely.
In the case of added GFH adsorbent (
Figure 3a), the adsorption of NTMP was nearly identical at molar Ca:NTMP ratios of up to 5:1. At a molar ratio of 7.33:1, between pH 8.5 and 10.0, and at 18.33:1, between pH 8.1 and 11.6, increased removal occurred. Interestingly, when comparing the batches with (
Figure 3a) and without (
Figure 3b) GFH, a discrepancy can be observed. If precipitates were the reason for an increased removal, a removal of phosphorus should have also been seen at a molar Ca:NTMP ratio of 7.33:1 from pH 8.5 to 10.0 and at a molar ratio of 18.33:1 at pH 12, as observed in the batches without GFH. However, this can be explained again by the re-dissolution of calcium from the GFH, which may have led to a higher availability of Ca
II in the batches with GFH than indicated by the molar ratios in
Figure 3.
In
Figure 3c,d the behavior of DTPMP is shown. In the batches without GFH (
Figure 3d), no removal of DTPMP took place throughout all pH values and all molar Ca:DTPMP ratios tested. Therefore, a precipitation of CaCO
3 or Ca-DTPMP can be excluded. It is known that DTPMP, unlike other organophosphonates such as HEDP, NTMP, and EDTMP, does not precipitate as 1:1 complexes [
34]. The batches with DTPMP and GFH (
Figure 3c) show the typical behavior of phosphonates on iron-containing surfaces, namely a decreasing adsorption capacity with increasing pH [
3,
18,
21,
35]. The differences in phosphorus removal among the different molar Ca:DTPMP ratios do not show any pattern; therefore, they must have been due to inaccuracies that commonly arise when conducting experiments. Certain deviations in the results of these experiments are to be expected, since some of the input variables may already differ slightly, such as the weighed adsorbent mass, homogeneity of the GFH, pH
end value, and so on. Repetitions of selected batches produced similar results (not shown in the figures for better readability). However, despite these deviations, tendencies can still be clearly identified.
In conclusion, the batches with NTMP showed precipitation, which presumably consisted of Ca-NTMP complexes, whereas the batches with DTPMP did not show any precipitation. This corresponds with data found in the literature: Zhang et al. [
32] observed that Ca-NTMP complexes have a lower solubility than Ca-DTPMP complexes. Furthermore, Gledhill and Feijtel [
36] stated that Ca-NTMP complexes have higher stability constants than Ca-DTPMP complexes at any given pH. Taken together, these conclusions indicate that there are more Ca-NTMP than Ca-DTPMP complexes at any given pH, but the solubility of the Ca-NTMP complexes is lower.
3.3. Experiment 3—Investigations on NTMP and DTPMP Precipitation
Since the previous experiments had shown that precipitation is responsible for increased elimination, the aim of Experiment 3 was to investigate this phenomena in more detail. The experiment was necessary because it does not seem to be possible to predict the precipitation of Ca-phosphonate complexes on the basis of the existing data. Depending on the experimental conditions (molar Ca:phosphonate ratio, pH, ionic strength, temperature) different complexes can precipitate [
30,
31], but reliable solubility products have been published only for some of them. In addition, the presence of calcium can have an effect on the pK
a values of the phosphonates [
29]. Additionally, a critical evaluation showed that the stability constants of DTPMP are not reliable, due to difficulties in synthesis and purification [
34].
Different phosphonate concentrations and molar Ca:phosphonate ratios were applied (
Table 4 and
Table 5). The pH values 8 and 9 were examined, as the previously described increased removal of phosphonates started in this pH range. Higher pH values were not investigated, because lower pH values are recommended for the adsorption of phosphonates on GFH [
18]. In
Table 4 and
Table 5, removal rates higher than 90% are highlighted in dark gray, and removal rates ≥5% and ≤90% are highlighted in light gray. Removal rates of less than 5% were considered to be measurement inaccuracies, as were negative values (which were within the 5% inaccuracy range). Standard deviations are shown in
Tables S1 and S2. Calcium concentrations that are discussed separately are highlighted in bold type.
Table 4 shows the results of the batches with NTMP at different calcium concentrations. In the batches without Ca
II and with a molar Ca:NTMP ratio of 1:1, no removal of phosphorus was observed. The same applies to the batches with 3.22 mg/L NTMP at all molar Ca:NTMP ratios. At higher NTMP concentrations, phosphorus was removed almost completely above particular calcium concentrations. At pH 9, the removal started at lower molar Ca:NTMP ratios than at pH 8 (e.g., with 80.5 mg/L NTMP at pH 9, a molar Ca:NTMP ratio of 2:1 was sufficient for an elimination of 91%, whereas at pH 8 no removal occurred).
The three batches with the same Ca
II concentration of 0.54 mmol/L highlight an interesting aspect (highlighted in bold type). The NTMP removal at pH 9 increased with increasing NTMP concentration from 64 ± 0.4% at 16.1 mg/L, and 78 ± 1.9% at 32.2 mg/L, up to 91 ± 0.2% at 80.5 mg/L. This indicates a precipitation of Ca-NTMP complexes, since a possible precipitation of CaCO
3 would have been inhibited by increasing concentrations of NTMP, due to its scale inhibition effect [
13].
In
Table 5 the results of the DTPMP batches at different calcium concentrations are shown. At molar Ca:DTPMP ratios of up to 2:1 and at 3.22 mg/L DTPMP, no removal of phosphorus was observed. At higher DTPMP concentrations, however, phosphorus was at least partially removed at particular Ca
II concentrations. Similar to NTMP, at pH 9, DTPMP was eliminated at lower molar Ca:DTPMP ratios than at pH 8.
At pH 8 and 16.1 mg/L DTPMP, a removal of 54% was observed at a molar Ca:DTPMP ratio of 60:1. At a concentration of 12 mM Ca and 0.73 µM DTPMP (0.42 mg/L) at 70 °C, Kan et al. [
31] found a crystalline Ca-DTPMP precipitate. Yan et al. [
37] also assumed precipitation of a Ca-DTPMP complex in their experiments with residual concentrations of 0.06 mM DTPMP and approx. 0.8 mM calcium.
An increased removal rate of DTPMP was observed at a constant Ca
II concentration of 0.70 mmol/L, paralleled by increasing DTPMP concentrations (
Table 5) (i.e., at pH 9; no elimination was found at 16.1 mg/L DTPMP, but at 80.5 mg/L DTPMP, 62 ± 1.9% was eliminated). Moreover, at pH 8, at a Ca
II concentration of 1.40 mmol/L and 32.2 mg/L DTPMP, 56 ± 1.4% was removed, whereas at 80.5 mg/L DTPMP, a higher removal rate of 74 ± 0.3% was observed. This indicates that precipitation of Ca-DTPMP complexes occurred, as it was observed for NTMP (
Table 4).
The results of Experiment 3 match the results from Experiment 2, in which the same concentration of 16.1 mg/L NTMP and DTPMP was used. In conclusion, NTMP and DTPMP demonstrated similar behavior. However, at comparable molar Ca:phosphonate ratios, NTMP showed a higher removal than DTPMP. This corresponds with the findings of Zhang et al. [
32], who found the solubility of Ca-DTPMP complexes to be higher than that of Ca-NTMP complexes.
To gain more knowledge as to which substances are precipitated, further batches were investigated with the five highest Ca
II concentrations used in Experiment 3 in the absence of phosphonate. In this case, the calcium concentration was measured in the membrane filtrate after seven days of rotation (
Table 6). Ca
start deviated only ±3.4% from the target calcium concentration (Ca
target). Therefore, this deviation is assumed to represent the degree of measurement inaccuracy inherent in this calcium determination method.
The initial and final pH values are nearly identical. A precipitation of CaCO3, however, would have resulted in a pH decrease (calculations performed with PHREEQC) that should have been noticeable, even though a buffer was used. The deviation between the initial and final concentration of Ca2+ varied between −3% and +3%. The negative deviations are attributable to batches in which the measured final concentration was above the measured initial concentration. As this is not possible, the negative values indicate the measurement error of the analysis method. As the positive deviations are of the same order of magnitude, it is assumed that the deviations within this range are due to measurement inaccuracies, which is also in line with the deviations between Castart and Catarget as mentioned above. It can be concluded that neither CaCO3 nor Ca(OH)2 precipitated. Since precipitation only occurred in the presence of the phosphonates, this is another indication that the precipitates were Ca-phosphonate complexes.