3.1.2. Powder X-ray Diffraction
All studied samples showed diffraction patterns typical for layered double hydroxides with the hydrotalcite- and hydrocalumite-like structures.
The pattern for sample MA-fluo2 (
Figure 1, top) shows a first peak at 10.3° (2θ) (d-spacing = 8.5 Å), which was caused by reflections of the (003) planes in the hydrotalcite-like structure with a rhombohedral 3R stacking of the main layers [
27,
44]. Here, the lattice parameter
c corresponded to three times the d-spacing value of this peak (
c = 25.63 Å). Lattice parameter
a corresponded to twice the d-spacing value of the peak at 60.7° (2θ), which was caused by reflections by the (110) planes. Although the intensity of the diffraction due to (110) planes seemed to be abnormally intense, there are previous reports in the literature where this peak was also very intense [
46,
47,
48]. Spacing of peak (110) corresponded to one half of the average distance between two neighbouring atoms in the main layers. Compared to data published for carbonate intercalated hydrotalcite [
8,
44], the natural analogue, the value obtained here for lattice parameter
a was in very good agreement, and lattice parameter
c was slightly larger than that of the carbonate analogue indicating a slightly increase in the interlayer spacing.
The peaks recorded in this diffraction pattern fit rather well to those of a nitrate-intercalated Mg,Al-LDH; only the very weak and broad peak around 5° (2θ) could not be ascribed to this phase. The literature [
27,
29] reported an interlayer spacing for fluorescein intercalated LDHs of 12 or 17 Å, corresponding to first peaks at 7° (2θ) or 5° (2θ), respectively.
The fluorescein molecule is almost planar, and has a triangle shape, with a base length of 10.87 Å and a height of 8.64 Å. The width of a brucite-like layer is ca. 4.8 Å [
44,
49,
50] and the spacing found for the diffraction by the (003) planes was somewhat larger than the sum of both values, so we could tentatively conclude that the peak at 5° (2θ) might be caused by a phase with intercalated dianionic fluorescein. The most probable orientation of dianionic fluorescein within the interlayer of LDHs is still a point of discussion and different theories were recently proposed by various research groups [
27,
29]. They report an interlayer distance of 12 Å [
29] or 16.5 Å [
27] caused by intercalated dianionic fluorescein arranged in a bilayer. If the peak at 5° (2θ) is caused by a dianionic fluorescein intercalated phase, the resulting interlayer spacing of 17.5 Å would be in very good agreement with the value reported by Lee et al. [
29]. Due to the very small concentration of fluorescein in the solid mixture, only the most intense peak, caused by diffraction on (003) planes at 5° (2θ), was recordable.
PXRD patterns of the fluorescein containing Ca,Al-LDHs (
Figure 1) are very similar to each other. All of them showed a first peak at ca. 10° (2θ), (d-spacing = 8.6 Å) which here corresponded to reflections by the (006) planes of hydrocalumite-like structure with a rhombohedral 6R stacking of the main layers [
45,
51]. Consequently, lattice parameter
c here corresponded to six times the spacing of this first peak. The gathered values of the studied Ca,Al-LDHs were very similar indicating no significant structural differences due to the presence of different amounts of fluorescein in the sample. Lattice parameter
a of the synthesised Ca,Al-samples differed significantly from that of the Mg,Al-sample. Due to the insertion of the larger calcium cation, the main layer of the resulting LDHs was slightly deformed as the distances between neighbouring atoms were increased [
45,
51]. The studied Ca,Al-samples showed a value of ca. 5.75 Å, which was in very good agreement with values reported in the literature (5.72 Å) [
45,
52]. It must be mentioned that all Ca,Al-samples showed a broad peak in the range of the first peak indicating stacking faults in the samples [
53]. Similarly to the Mg,Al-samples, the
c values of the Ca,Al-samples were slightly increased compared to the literature values reported for natural analogues (11.4° (2θ), 7.7 Å) [
45,
52] indicating an increased extension of the interlayer space. It should be noticed again that the values calculated here for the interlayer spacings do not fit to those published by Costantino et al. [
27] and Lee et al. [
29] for fluorescein intercalated hydrotalcite-like materials. On the contrary, those values are in very good agreement with values reported by Tanaka et al. [
28] for Mg,Al-LDHs with very low amounts of intercalated fluorescein. Tanaka and colleagues postulated that low amounts of fluorescein can either be intercalated as the dianionic form and oriented as a monolayer in the interlayer of the LDH or be adsorbed on the external surface of the LDH [
28]. Both variations would lead to interlayer spacings similar to that of the compared natural analogues or nitrate LDHs, as observed in the Ca,Al-samples studied here. Due to the similarity between the position of the main peak (001) of a nitrate intercalated Ca,Al-LDH with externally adsorbed fluorescein, and a Ca,Al-LDH with low content of fluorescein in the interlayer, as described by Tanaka et al. [
28], a clear distinction between the two types is not possible with PXRD.
The PXRD pattern of the synthesised samples showed rather broad peaks for sample MA-fluo2, but very sharp, more intense ones, for Ca,Al-LDH samples, indicating different crystallite sizes. Calculated crystallite sizes of the studied sample are included in
Table 2. Calculations were based on the Scherrer equation [
54] and carried out on the first peak of each sample, as this corresponds with the maximum spacing of the LDH in
c-direction. In other words, the values reported in this table correspond to the length of the crystallites in the direction perpendicular to the layers.
To achieve sufficiently crystalline, easily characterizable materials, longer aging times were selected based on information given in the literature and our previous studies regarding the synthesis of these materials [
8,
14,
32,
33]. In case of a potential injection, shorter aging times should be selected to produce particles of smaller sizes; those are supposed to be more reactive and less time consuming with respect to the economic aspects regarding an upscaling of the synthesis procedure. The different crystallite sizes of the samples are supposed to be a result of different aging times used for the Mg,Al-sample (one day) and the Ca,Al-samples (six days).
3.1.3. Thermal Analysis and Evolved Gas Analysis
All synthesised samples were studied regarding their thermal stability using thermogravimetry (TG) coupled with differential thermal analysis (DTA). Mass spectrometry (MS) was simultaneously used to analyse the gases released during the heating process. All curves are included in the
Supplementary Material Section (Figures S2–S5), and those for sample MA-fluo2 are given as an example in
Figure 2.
Results of the synthesised samples revealed a stepwise thermal decomposition resulting in four mass loss events, as expected for layered materials related to hydrotalcite- and hydrocalumite-like structures [
55].
Table 3 summarizes the recorded mass losses as well as the evolving gases of each thermal event for the studied samples.
For all studied samples, the first mass loss could be assigned to the removal of physisorbed water on the external surface of the LDH and of the interlayer water. The DTA curves revealed very intense endothermic peaks for the Mg,Al-sample and medium intensity endothermic peaks for the Ca,Al-samples in the same temperature range. Surprisingly, the peak for the mass-to-charge-ratio (m/z) of 18 (H2O+) was only recorded in the MS spectra of the Ca,Al-LDH samples, but only as a very weak peak for sample Mg,Al-LDH. Upon increasing the temperature, the behaviour of the Mg,Al-LDH and Ca,Al-LDH samples was quite different and will thus be described separately.
A second mass loss was recorded between 195 ° and 420 °C for sample Mg,Al-LDH, due to the decomposition of the interlayer phases and the condensation and loss of hydroxyl groups from the main layer as water molecules, as confirmed by the MS spectrum. MS peaks were recorded at
m/
z = 18 (H
2O
+), CO
2 (
m/
z = 44, CO
2+), and NO
2 (
m/
z = 46, NO
2+). The NO
2 was caused by the decomposition of interlayer nitrate. The peak due to CO
2 might be caused by the decomposition of the small amount of fluorescein existing in this sample, as under oxidic conditions, this would lead to the evolution of CO
2 and water vapour, which was detected as well. This finding agrees with the results of the PXRD (
Section 3.1.2) and FTIR (
Section 3.1.4) measurements. The MS spectra together with the DTA diagram, which showed two endothermic events in the discussed temperature range, indicated a stepwise process involving several reactions. Upon increasing the temperature up to 640 °C a third significant mass loss occurred which was, as seen in the MS spectrum, due to the continuing release of the intercalated nitrate species. There, a slightly larger peak for NO
2 was recorded compared to that in the previously discussed temperature range. A last, and very small, mass loss was recorded above 640 °C. It can be assumed that this corresponds to a release of gases, which were already formed at the beginning of the decomposition but occluded within the solid and released only at very high temperatures [
56]. Decomposition of fluorescein in a combustion process should give raise to an exothermic effect, but due to the small amount of fluorescein in the sample and probably, to the fact that the process should take place in a broad temperature range, such an exothermic effect was not recorded, or it might be cancelled by the stronger endothermic effects.
For the Ca,Al-LDH samples a second mass loss was recorded up to 350 °C, which, in contrast to the aforementioned Mg,Al-LDH sample, was only caused by condensation of the main layer hydroxyl groups leading to water vapour formation; weak peaks for m/z = 18 (H2O+) were recorded in the MS spectra. The DTA curves showed medium intensity endothermic events in this temperature range. Upon increasing the temperatures up to 625 °C, a third mass loss occurred, which could be assigned to the release of the intercalated nitrate species as evidenced in the MS spectra by a rather large peak for NO2 (m/z = 46, NO2+). Contrary to the discussed release of the nitrate species from sample MA-fluo2, both the MS spectra as well as the DTA curves revealed that the nitrate species were released from the Ca,Al-LDH samples in a single event. During the fourth and last mass loss event, at temperatures above 620 °C, the MS spectra of the studied Ca,Al-samples showed peaks for CO2 (m/z = 44, CO2+), which were most likely a result of the decomposition of the fluorescent dye, although the expected peak due to water release was not recorded.
The calculated formulas of the studied compounds, which were based on the results of the TG-DTA-MS analyses, are given in
Table 4. The content of magnesium, calcium, and aluminium were calculated based on the results of the ICP-OES analyses (see the raw data in
Table 1). The water content in the interlayer was calculated from the results of the thermal analyses of the solids. The nitrate content included was calculated based on the electrical balance between the main layers and the interlayer.
3.1.4. Fourier Transform Infrared Spectroscopy
All synthesised samples were studied using FTIR spectroscopy. All spectra showed absorption bands that are typical for layered hydrotalcite- and hydrocalumite-like materials. These bands are caused by vibrations of the hydroxyl groups in the main layers and the interlayer, the intercalated anions, and lattice vibrations in the main layers [
57,
58,
59,
60,
61]. The spectrum of sample MA-fluo2 is shown in
Figure 3; spectra of all samples are included in the
Supplementary Materials section (Figures S6–S9).
For the Mg,Al-sample a rather broad absorption band was recorded between 3630–3380 cm
−1, due to the stretching vibrations of the hydroxyl groups, both in the interlayer and the main layers, of the studied LDH and the water molecules. The broadening of the band is due to hydrogen bonds between these species [
58]. Vibrations of the hydroxyl groups in the main layer corresponded to the first band at 3640 cm
−1 and those of the interlayer water resulted in the second, broader, band between 3500–3425 cm
−1. All spectra showed intense absorption bands at 1380 cm
−1 which were most likely caused by the vibrations of intercalated nitrate species [
57,
60,
61,
62,
63]. Absorption bands below 750 cm
−1 could be assigned to bending vibrations of the metal hydroxyl groups in the main layers [
57,
59,
61,
62,
63]. Metal-hydroxyl translation modes in the crystal lattice of the studied LDHs were recorded as absorption modes below 700 cm
−1 [
57,
58].
All FTIR spectra of the studied samples were compared to that of the unintercalated fluorescein disodium salt, showing a variety of absorption bands ranging from 1600 to 500 cm
−1, in order to easily detect absorption bands caused by the fluorescent dye. The spectra of the studied fluo-LDHs showed only very weak absorption bands doubtlessly caused by a divalent fluorescein anion, as expected due to the very low fluorescein content of the samples. The most intense peak in this region at 1380 cm
−1 is due to a υ
3 stretching vibration mode of intercalated nitrate [
63]. The spectra closely resemble those of the reference samples, Mg,Al- as well as Ca,Al-LDHs, with intercalated carbonate (grey lines). The doublet recorded in some of the spectra around 2300–2250 cm
−1 resulted from a miscancellation of the atmospheric CO
2 band.
3.1.7. BET Measurements
To determine the specific surface areas of the synthesised samples, adsorption–desorption studies of N
2 adsorption at −196 °C were carried out. The isotherms are included in the
Supplementary Material Section (Figures S10–S13) and the values determined for the BET specific surface areas are given in
Table 5.
All studied samples showed isotherms corresponding to type II in the International Union of Pure and Applied Chemistry (IUPAC) classification [
64] indicating adsorption on non-porous or macroporous surfaces, with unrestricted monolayer–multilayer adsorption [
65]. The isotherm of sample MA-fluo2 exhibited a rather broad shaped hysteresis loop type H2, which is commonly found for disordered materials with an ill-defined distribution of pore size and shape [
64,
65]. The specific surface area shown by this sample is surprisingly small, bearing in mind the rather soft ageing treatment to which the sample has been submitted. Such a small specific surface area has not been found even for samples submitted to hydrothermal treatment for several days after synthesis [
58].
Isotherms of the Ca,Al-samples showed very narrow, nearly absent, hysteresis loops of H3 type, which is typical for the N
2 adsorption on non-porous or microporous materials formed by non-rigid aggregates of plate-like particles with slit-shaped pores [
64,
65]. The specific surface areas calculated for these three samples are in the expected range for this sort of material. Probably the ageing treatment of six days has favoured an Ostwald ripening [
66,
67], removing the very small, tiny particles and thus avoiding blocking of the pores of the large particles.
It can be assumed that the surface texture of the samples was most likely a result of the different aging times and slight alterations in the drying process, which can result in a stronger aggregation of the particles. Overall, there are no significant differences among the values determined for the Ca,Al-LDH samples, suggesting that the fluorescein content has no significant effect on the surface texture of the samples studied here. The observed changes are probably caused by minor variations during the drying process.
3.1.8. Particle Size Distribution
The particle size distribution was evaluated for samples MA-fluo2, CA-fluo2, and CA-fluo0.2 using dried sample material (subsequently labelled with the suffix dry) as well as the freshly prepared and still wet LDH slurry (subsequently labelled with the suffix wet) to determine the influence of the drying process as well as the amount of fluorescent dye on the resulting particle size distribution. Single curves are included in the
Supplementary Materials section (Figures S14–S19).
Measurements of the sample MA-fluo2-dry,
Figure 5, revealed a maximum at ca. 340 μm with a sharp decrease in the curve progression for values above this maximum. For values below this maximum the curve progression was somewhat more steady, even revealing small shoulders at ca. 35 and 3 μm. Upon increasing times of ultrasound treatment, a new maximum was developed at around 40 μm with a shoulder at 6 μm, while the intensity of the first maximum was strongly decreased. The curves of the measurements on sample MA-fluo2-wet differed significantly from those of the dry material. For the untreated sample, a maximum at about 7 μm was revealed with a sharp decrease in the curve progression for values above and below this maximum. Additional rather weak maxima were revealed at 60, 340, and 1180 μm. Application of up to 15 min ultrasound treatment resulted in the decrease in the first maximum (7 μm) accompanied by the formation of a shoulder at ca. 3 μm. The second maximum was shifted to ca. 85 μm, while the third and fourth merged forming a new maximum at ca. 510 μm; a small shoulder was formed at 1350 μm.
These results suggest that the primary crystallites aggregated during drying and that their separation by ultrasound treatment might be tricky. Even after 15 min of ultrasound treatment, two new maxima developed at ca. 6 and 40 μm, but that formed at 340 μm still remained with a large intensity. However, if the sample is kept wet in a slurry and the particle size distribution straightforwardly studied with no drying step, a monomodal size of 7 μm could be found. Application of ultrasounds resulted in the development of many peaks, indicating a rather random distribution of particle sizes. Particles with sizes above 7 μm increased their contribution, while those with an original size of 7 μm split showing a new contribution by particles of ca. 3 μm.
The curve of the untreated sample CA-fluo2-dry,
Figure 6, showed two maxima at ca. 65 and 290 μm with a rather sharp decrease in the curve progression for values above the maximum. The curve progression for values below this maximum was slightly steadier. After 5 min under ultrasound treatment the maximum at 65 μm disappeared and the maximum at 290 μm was shifted to ca. 440 μm, with a small shoulder at 1170 μm. A large peak developed at ca. 37 μm with weak shoulders at 5 and 0.5 μm. Upon increasing times of ultrasound treatment, the maxima above 100 μm disappeared and the maximum at 37 μm shifted to ca. 30 μm, still showing the aforementioned shoulders. The curve of the particle size distribution of the undried LDH slurry of sample CA-fluo2-wet differed significantly of that of the dried sample material. It showed a unimodal particle size distribution with a maximum at ca. 7 μm, which is shifted towards smaller values (ca. 6 μm) upon increasing times of ultrasound treatment, but without development of new maxima. These data again support the importance of the drying step to form larger particles, hardly disaggregated under ultrasound treatment.
Finally, the curves for sample CA-fluo0.2-dry,
Figure 7, showed a somewhat different behaviour. The curve of the untreated sample showed a rather broad maximum at around 200 μm with steady decrease in the curve for values above and below the maximum. A shoulder was revealed at ca. 8 μm. Upon extending ultrasound treatment until 15 min the first maximum is shifted towards a smaller particle size (85 μm) decreasing very sharply its intensity, the shoulder evolved into a second maximum at ca. 10 μm, this now being the most intense peak in the diagram, and a third was formed at about 1 μm, with a medium intensity. The curves of the particle size distribution of the undried LDH slurry of sample CA-fluo0.2-wet resembled closely those for sample CA-fluo2-wet. Once again, the curves expressed a unimodal particle size distribution with a maximum at ca. 5 μm, but now with a very small shoulder at 28 μm. Upon increasing times of ultrasound treatment, the shoulder disappeared resulting in an increase in the absolute maximum, showing a rather good monomodal particle size distribution.
Table 6 includes the d(0.5) values, which are defined as the median particle size in μm and divide the population exactly into two equal halves, of the particle size distribution of the samples with 2% and the sample with 0.2% fluorescent dye. On comparing the three samples submitted to this study, we can see that sample Mg,Al-LDH has a contribution by rather large particles, but also a contribution by very small particles, which might be blocking the pores of the large ones, thus accounting for the abnormally small specific surface area of this sample. However, for the Ca,Al-LDH samples studied, the peak corresponding to the larger particles is broader, while the contribution by very small particles is rather small, and in any case lower than for the Mg,Al-LDH sample. Consequently, blocking of the pores in the Ca,Al-LDH samples is avoided (or at least lowered), these samples showing a larger specific surface area than sample Mg,Al-LDH, where a contribution by very small pores is not negligible. As mentioned above, this might be a consequence of Ostwald ripening [
65,
66].
Table 6 reveals that the particle size distribution of a sample is heavily influenced by the drying procedure, leading to the formation of large agglomerates. Upon ultrasound treatment those express significantly smaller median particle sizes for the dried sample materials, although disaggregation never reaches the values for the undried samples. This is in good agreement with data previously published by the authors, in which the impact of the drying time on the particle size distribution was discussed [
8].
For Ca,Al-LDH samples the measurements on the freshly prepared wet sample material revealed no differences in the median particle size on un-treated and treated material. An exception must be made for sample MA-fluo2, which revealed a significantly larger median particle size after ultrasound treatment compared to that of the untreated, freshly prepared sample. This was most likely possible as ultrasound energy can, in some sort of way, help to re-agglomerate very small particles which were previously released during the ultrasound treatment. It is worth mentioning that the median particle sizes of the dried and wet Ca,Al-LDH samples are significantly smaller than that of the Mg,Al-LDH sample. This finding is in very good agreement with previously published results [
14], which confirmed that the median particle size also depends on the nature of the main layers of the studied material.
The finding that drying of the samples gave rise to the agglomeration of the particles and that such an agglomeration is less intense (or is even absent) if the samples are kept in a slurry, is very important regarding the potential use of the samples studied here as tracer particles in injection strategies. Using undried sample material is more cost effective and easier to apply, although storing of the slurry is more spacious and therefore more expensive than storing dried solids. Particles can directly be synthesised and applied without prior treatment.
Based on the overall goal of the study, to develop traceable particles that are able to mimic co-injected reactive particles, thus showing similar mobility in the subsurface after injection, comparative studies regarding their particle sizes, one of the most important features of particles when it comes to injections, were carried out to compare the fluorescent LDHs presented here with recently studied reactive LDHs. These reactive LDHs proved to be the best candidates in studies previously carried out and published by the authors [
8,
14], addressing the removal of chlorinated organic solvents. These potential remediation particles were intercalated with different interlayer anions to cover different reaction pathways of the LDHs with the target contaminants, including long-chain organic anions with various functional groups for an easy retention of the chlorinated organic solvents [
8] as well as oxidising agents [
14] for their chemical decomposition.
Figure 8 compares the samples studied here, MA-fluo2 and CA-fluo2, to Mg,Al-LDHs intercalated with octadecanoate (MA-17COO), octadecane sulfonate (MA-18SO
3), used to adsorb chlorinated organic solvents, and permanganate (MA-MnO
4), as well as a synthesised peroxydisulphate intercalated Ca,Al-LDH (CA-S
2O
8), used for oxidation in previous studies [
8,
14].
For the dried samples, the maxima of the fluorescent particles in the particle size distribution curve are in quite good agreement with that of previously studied reactive particles, only those for samples MA-17COO-dry and CA-S2O8-dry differ, the latter showing the highest deviation. Maxima of the fluorescent particles, generated using wet sample material, resemble quite nicely those of previously studied reactive LDHs. Here, sample MA-18SO3 revealed the largest deviation leading to the assumption that none of the fluorescent particles studied here would be suitable for this kind of reactant. Nevertheless, several pre-tests proved that the particle size of all studied LDHs can easily be tuned by adjusting the synthesis parameters and the aging time to match those of the fluorescent particles with that of the target reactants.
Table 7 summarizes the median particle sizes of the previously studied reactive samples used here for the comparative studies. It can be seen that the aggregated mean particle sizes of the wet samples are in good agreement with the fluorescent LDHs studied here; sample MA-18SO
3 needs to be excluded from this statement as further research is required to better match the particle size of this sample with that of the fluorescent LDHs studied here.
It can be proposed that the fluorescent LDHs studied here could be used as tracer particles in in situ remediation strategies. As the particle sizes of both the fluorescent LDH as well as the studied potential reactants are rather similar, it can be assumed that all those particles might show a similar behaviour after injection. If reactants were co-injected with the fluorescent particles presented here, potential movement of the reactants, most likely caused by groundwater flow, away from the target point of reaction could be traced and a spreading rate of these could clearly be estimated due to the presence of the fluorescent particles.