3.1. Mechanical Tests on Test Coatings
Determination of the catalytic power of Fe-diiPc was examined in solvent-borne (S471 and S622) and high-solid alkyd binders (FP07 and TI870) using standardized mechanical assays on test coatings applied on glass substrates. They included the determination of drying times by the Beck–Koller method [
31], determination of the relative hardness by Persoz pendulum [
32], and determination of chemical resistance by the MEK test [
33]. The collected experimental data were compared with reference data obtained for coatings treated by the commercial primary drier Co-2EH.
Unusually high catalytic activity of Fe-diiPc was documented on short total dry times (τ
4), observed for solvent-borne formulations (
Table 1). In case of the medium oil-length alkyd resin S471, low values of τ
4 were obtained at 0.003–0.01 wt.% of metal in dry matter content. In this concentration range, tack-free time (τ
2) of the formulations Fe-diiPc/S471 decreased with the increasing concentration. Inverse dependence, observed for τ
4, was due to the formation of a crosslinked polymeric layer on the coating surface at high metal concentration, which is responsible for slower oxygen diffusion and pure through-drying, as previously documented on formulations of Co-2EH [
25]. The metal concentration 0.003 wt.% seems to be the optimal dosage for Fe-diiPc in the absence of secondary driers; these can strongly influence the process of through drying. Commercial Co-2EH, used as the reference, shows optimal performance at considerably higher dosage (0.1 wt.%), which is in line with recommendation of suppliers and literature data [
17]. At this concentration of Co-2EH, τ
2 is comparable but τ
4 is significantly longer than that observed for Fe-diiPc at 0.003 wt.%. Comparison with commercial cobalt-free primary drier Borchi OXY-Coat is given in
Table S1 in the Supplementary Materials. Such a compound is highly active at concentrations even lower than Fe-diiPc (0.003–0.0003 wt.%). We note that the inverse dependence of τ
4 is observed here as well.
Formulation of long-length alkyd resin S622, treated by Fe-diiPc, exhibits rapid drying at a concentration of 0.01 wt.%, as evident from low values of tack-free time (τ2 = 0.9 h) and total dry time (τ4 = 3.8 h). Lowering of metal concentration leads to the prolongation of drying times. Nevertheless, even at 0.003 wt.%, performance of Fe-diiPc is still acceptable because τ4 is shorter than that observed for formulations treated with Co-2EH.
Long-term activity of Fe-diiPc in the test coatings was evaluated by the measurements of their relative hardness.
Table 1 summarizes the data obtained for solvent-borne resins S471 and S622. The values obtained for the formulations Fe-diiPc/S471 ten days after application (
Hrel;10d) are comparable to Co-2EH/S471. Lower final hardness (
Hrel;100d) of the coatings treated by Fe-diiPc, measured one hundred days after application, is attributed to a lower density of crosslinking. Data collected for the binder of longer oil-length (S622) show similar trends. In this case, however, significantly slower hardening of the coatings treated by Fe-diiPc is evidenced already on values of
Hrel;10d.
Fe-diiPc exhibited a very good performance in the high-solid binder FP07 at metal concentration 0.01 wt.%, as evident from the short tack-free time (τ
2 = 3.9 h) and dry-hard time (τ
3 = 8.3 h); see
Table 2. Lowering of the concentration led to prolongation of the drying times, but they were still acceptable at 0.006 wt.%. It should be noted that total-dry time is less important in the case of high-solid binders, owing to their generally purer through-drying and lower relative hardness, which often leads to values higher than 24 h. The optimal dosage of Co-2EH is 0.06 wt.% due to its acceptable dry-hard time (τ
3) and very low values of τ
1 and τ
2. Inverse dependence of τ
3 on concentration, observed for Co-2EH, suggests considerably worse through-drying at high metal concentration than in the case of Fe-diiPc.
Formulations Fe-diiPc/TI870 were cured slower than the aforementioned Fe-diiPc/FP07 at the same dosage. The optimal concentration recommended for potential application is 0.01 wt.%. It is noteworthy that formulations Co-2EH/TI870 exhibited a higher tendency to overdose, as evident by the high values of τ2 and τ3 at concentrations of 0.01 wt.%. This is ascribed to different fatty acid patterns, which results, in the case of cobalt-based drier, in faster drying.
Test coatings of high-solid binders, treated by Fe-diiPc, show significantly lower final hardness (
Hrel;100d) than those treated by Co-2EH (
Table 2). Initial fast rise of the coating hardness, documented by the
Hrel;10d value, is followed by stagnation while coatings treated by Co-2EH continue in slow hardening up to final value (
Hrel;100d). Such difference documents the lower long-term activity of Fe-diiPc. Nevertheless, lower hardness of the films containing Fe-diiPc is not necessarily an obstacle for particular application because it only has a minor effect on their chemical resistance, as evidenced by the MEK test (
Table 3). Results of the testing, performed on the coatings cured for 105 days, reveal only a minor effect of drier composition and drier concentration on the resistance time. The experimental data, summarized in
Table 3, further prove that thickness of the film has a considerably stronger effect as on MEK resistance that the drying agent.
3.2. Characterization of the Binders by Vibration Spectroscopy
Infrared and Raman spectroscopy was used for detailed characterization of the alkyd resins S471, S622, FP07 and TI870. The spectra were measured from samples of fresh binders and from samples cured for 100 days at room temperature. The drier composition and drier concentration have only a minor effect on the spectrum pattern of the cured samples; therefore, only spectra of those treated by Fe-diiPc are presented below.
In the binder S471, 21 characteristic absorption bands were observed in the spectra and unambiguously agreed with prior results in the literature [
35,
36]. They are denoted a–u and are summarized in
Table 4. Full spectra are presented in
Figure 2. O–H stretching mode (a) of unesterified carboxylic and hydroxyl groups appears in the IR spectrum of the fresh binder as a very broad band at 3520 cm
−1. Upon the curing process, it moves to lower wavenumbers due to the formation of hydroperoxide by autoxidation of the unsaturated fatty acid chains and appearance of the OH-containing side products [
26]. Disappearance of the absorption bands c (IR: 3008 cm
−1; Raman: 3010 cm
−1) and i (Raman: 1658 cm
−1) is the most characteristic change in the vibration spectra of the binder S471.
The bands were assigned to C–H stretching and C=C stretching of the C=C–H moiety in unsaturated fatty acid tails, respectively. We note that both bands can be used for monitoring the unsaturation degree upon the curing process. The absorption bands related with aromatic rings (b, j, k, r, s and t) stayed almost unchanged upon the curing. Only band u, assigned to out-of-plane bending of the C–H bond in the aromatic ring, decreased in intensity due to the superposition of similar vibration modes of the reactive
cis-C=C–H moiety [
37]. Frequency of aromatic ring breathing, intense and well separated in Raman spectra, enables distinguishing between phthalate (band r: 1042 cm
−1) and isophthalate functions (band s: 1004 cm
−1). We note that the aforementioned bands b, j and k are very typical for binders containing aromatic rings.
Saturated parts of the hydrocarbon chains give very characteristic C–H stretching bands b–g assigned to methyl and methylene groups. Their decrease in intensity is attributed to oxidative degradation, responsible for the emission of volatile side products, and to a lower flexibility of the fatty acid chains related with a decrease in given extinction coefficients. Ester functions of the polyester backbone give characteristic stretching bands h, o, p and q. Broadening of the C=O stretching band h, which is very strong in infrared spectra and well separated, is attributed to the formation of aldehyde and ketone functions upon the oxidative degradation.
Vibration spectra of fresh and cured samples of the resins S622, FP07 and TI870 are presented in
Figure 3. Assignments of characteristic absorption bands were performed in line with the binder S471; detailed assignments are given in
Tables S2–S4 in the Supplementary Materials.
The infrared and Raman spectra, given in
Figure 3, accurately document an effect of oil length on the spectrum pattern. The increasing oil-length of the alkyd binder resulted in a lower relative intensity of the absorption bands related with the fatty acid tails (c–g and i). These bands increased in the order S471 < S622 < FP07 ~ TI870. Lower intensity of the bands b, j and k in the spectra of high-solid binders FP07 and TI870 reflects a lower content of phthalic or isophthalic acid, respectively. As mentioned before, phthalate (S622 and FP07) and isophthalate (TI870) can be distinguished in the binder by breathing of the aromatic ring (ν
s(C=C, arom.)) active in the Raman spectra (bands r and s). Different patterns in the region of δ(C=C, arom.) are also typical features for this type of modification.
We note that curing of the resins S622, FP07 and TI870 had the same effect in vibration spectra as described for S471. In all cases, the bands c and i were suitable for monitoring the curing process, and composition of the drier had only a minor effect on the intensity of infrared and Raman absorption bands of the cured samples. The only noticeable effect was a higher infrared intensity of bands e and g in the samples treated by Fe-diiPc, when compared with samples treated with Co-2EH. It is ascribed to a higher flexibility of the fatty acid chains resulting in increased extinction coefficients. Suggested lower density of crosslinking is in line with lower relative hardness of the coatings cured by Fe-diiPc documented by the aforementioned mechanical tests.
3.3. Investigation of the Autoxidation Kinetics on Test Coatings
Time-resolved infrared spectroscopy was used for a detailed investigation of the curing process. The study was conducted on alkyd binder S471 treated by Fe-diiPc, and the obtained kinetic data were compared with samples treated by Co-2EH. We note that the measurements were performed on thin samples of the formulation using the ATR sampling technique, which prevents formation of inhomogeneities due to limited air-oxygen diffusion into whole paint layer [
25].
Consumption of the reactive double allylic moiety, shown in
Figure 4, was evidenced in the infrared spectra as the intensity decrease in the absorption band at 3008 cm
−1, attributed to ν
a(
cis-C=C–H) as mentioned in the preceding section (band
c in the
Figure 2a). It should be noted that registered reaction profiles do not reflect the appearance of crosslinks; hydroperoxides are formed as relatively stable intermediates. These, however, are usually decomposed by the action of primary driers as well [
11].
In line with previous studies, the peroxidation step was treated as a reaction of pseudo-first order because the concentration of molecular oxygen is expected to be constant upon the curing process [
17,
25]. Estimated maximal rate coefficients (
kmax), induction times (
tind) and half-lives (
t1/2) for formulations Fe-diiPc/S471 and Co-2EH/S471 are listed in
Table 5. Corresponding reaction profiles are given in
Figure 5.
In the concentration range 0.01–0.003 wt.% (runs A–C), diiPc gives very short induction times and high rate coefficients, which correlates well with the short tack-free times (τ
2) estimated by the mechanical tests. It is noteworthy that commercial Co-2EH gives similar rate coefficients at concentrations of about one magnitude higher; see runs A–C in
Table 5. Another important feature of the formulations treated by diiPc is much slower increase in the induction time when the metal concentration decreases. Indeed, very long induction times are not observed even at the concentration 3 × 10
−4 wt.% (run F), while the limiting dosage of the cobalt-based drier Co-2EH was estimated to be 0.001 wt.% (run D).
Reaction profiles of the formulations Fe-diiPc/S471 and Co-2EH/S471 in linear scale (
Figure 5a,b) well document the effect of concentration on the induction time. Furthermore, they show that the coatings containing Fe-diiPc reached significantly lower conversions after 20 h of curing even at the high dosage. Such a phenomenon is related with a deviation from the pseudo-first order kinetics appearing at lower conversions, as became evident from nonlinearity in the logarithmic plots given in
Figure 5c (cf. with
Figure 5d).
Different shapes of reaction profiles led us to analyze the experimental data in more detail. It was found that composition of the drier and its concentration have a strong effect not only on the magnitude of maximal rate coefficients (
kmax) but also on the conversion at which
kmax is reached.
Figure 6 displays the development of the pseudo-first order rate coefficient on the allylic moiety conversion for two driers under study, together with two other promising primary driers, namely Borchi OXY-Coat and Mn(acac)
3, which will be helpful for better understanding of the phenomenon. We note that the kinetic parameters for Borchi OXY-Coat are given in
Table S5 in the Supplementary Materials, while those for Mn(acac)
3 have been published elsewhere [
17].
Figure 6 documents that these four primary driers are able to reach high values of
kmax (>1.4 h
−1), but each of them at different metal concentrations. From this point of view, the efficiency of the driers increases in the order Mn(acac)
3 ~ Co-2EH < Fe-diiPc < Borchi OXY-Coat. At high metal concentrations (0.01–0.003 wt.%), Fe-diiPc reaches the maximal value of the rate coefficient at a low conversion (~30%) and the following decrease to a half-maximal value is also relatively fast; it appears at ~60% conversion. At a concentration lower than 0.003 wt.%, the maximal and half-maximal rate coefficients appear at even lower conversions. Such behavior differs considerably from Co-2EH, which, at 0.1 wt.%, exhibits the maximal and half-maximal values of the rate coefficient at ~50% and 75% conversion, respectively. Furthermore, lowering of the metal concentration has an opposite effect on these parameters. The unusual behavior of Fe-diiPc can be ascribed to lower stability of the active species, which is responsible for earlier deviation from pseudo-first order kinetics. Such reasoning also fits the concentration dependence observed for Fe-diiPc. Indeed, slower peroxidation process at a lower metal concentration should lead to the decomposition of higher fractions of catalyst at a given conversion if stability of the active species is the issue. Considering the shape of the plots given in
Figure 6, we suggest the stability of the active species increases in the order Fe-diiPc < Borchi OXY-Coat ~ Mn(acac)
3 ~ Co-2EH.