FTIR conformity analysis and performance testings of fresh, aged and expired polymeric paints under different storage conditions

FTIR conformity analysis

FTIR conformity analysis was carried out using Nicolet iS10 (Thermo Scientific, Madison, USA) equipped with ATR diamond crystal window and Nicolet iS5 (Thermo Scientific, Madison, USA) equipped with ATR diamond crystal window. Two spectrophotometers used were labeled as Instrument_in-house (iS10) and Instrument_3rd-party (iS5). One drops of polymeric paints was placed onto the ATR crystal until completely covered, no pressure was applied while the spectrum was measured. In reference to the procedures in Standards Malaysia MS2736L2022 [48], the spectra were recorded over the range of 700–4000 cm⁻¹ by averaging 32 scans at a maximum resolution of 4 cm⁻¹. Triplicate analyses were carried out for each sample with a background infrared spectrum every time sample was changed.

Quality control of the FTIR analysis

A total of forty-two polymeric paint samples were tested: 18 fresh and 24 retained (properly and improperly stored) samples. Three fresh samples from each paint were evaluated for its consistency across different locations (Bottom, Middle and Top) in the mixing tank. The average FTIR spectra from Bottom, Middle and Top of each paint were further used to generate a Reference spectrum. The spectra of FTIR were analyzed by OMNIC Software Suite (Thermo Scientific, Madison, USA) at both (i) whole wavenumber region (700–4000 cm⁻¹) and (ii) specific fingerprint region (900–2000 cm⁻¹) without spectral correction. The degree of similarity (r) of the samples was obtained by comparing the spectra of the samples to that of the Reference spectrum using high sensitivity compare incorporated to the FTIR software. This numerical compare algorithm employs squared derivative algorithm which applies an exponential scaling function on both wavenumber and absorbance vectors of the sample and Reference spectra [37, 49]. The r value (0–1) represents the degree of sample spectrum matching to that of the Reference spectrum, where r→1 indicates high degree of similarity. The Reference spectrum was generated by averaging nine spectra (triplicates from Top, Middle and Bottom of the mixing tank) for each paint sample. The acceptance criterion for qualification and batch-to-batch monitoring was set at r ≥ 0.900 ± 0.002, which was outlined based on the considerations of the sample size, types of sample and sampling uncertainties [50], [51], [52].

During a tender qualification (Fig. 1), the end user shall receive a Reference FTIR spectrum that has been cross-validated between in-house laboratory (owned by paint manufacturer) and independent 3rd-party laboratory. Before cross-validation, the r values of individual fresh samples collected by both (a) in-house laboratory and (b) 3rd-party laboratory are estimated by cross-reference to the received Reference spectrum. The acceptance criterion for qualification of the fresh paint sample was determined using eq. 1:

 [15]

Equation 1 is defined as the quantity of a product of similar variables that is used predominantly to describe the relationship between two or more observations across specific periods. Quantity rRef, in other words, can be distinguished as the geometric mean of the nth root product of n numbers [53], [54], [55]. For this study, eq. 1 was utilized to determine the degree of similarity of FTIR Reference spectrum from in-house laboratory and 3rd-party laboratory of the same paint. rRefin−house was estimated for entire wavenumber and specific fingerprint region by referencing in-house Reference spectrum to 3rd-party Reference spectrum and vice versa for estimation of rRef3rd−party. The qualification passed criterion was set at r_Ref ≥ 0.90, as the r values were reduced to two significant figures after mathematical function (multiplication and square root) [56, 57]. As an example, for the qualification of Epoxy part A Reference spectrum at whole wavenumber region was shown below.

After the Reference spectrum was qualified, the r values were then estimated for aged and expired samples which were collected from Bottom location of the mixing tank. Paired Student’s t-test was then used to compare the difference between r values estimated for retained paint samples stored in two conditions. The null hypothesis (H₀) was created with the assumption that there was no significant difference between the r values of properly and improperly stored retained paint samples while the alternate hypothesis (H_a) showed there was significant difference. The significance level used in this assumption test was p = 0.05. The t-statistic of a given set of data was calculated according to the following equation:

 [58], [59], [60]

where, sD‾/n = SE(D‾) is the standard error of average differences (D‾), μ_D is the population mean of different values, n is sample size and sD‾ is the standard deviation of the differences. Null hypothesis was rejected if these two conditions were met: (1) t-statistic > t-critical (t-critical value was extracted from the statistic table using the degree of freedom and pre-selected level of significance); and (2) p-value < significance level used in this study.

Adhesion pull-off test

Adhesion pull-off test measures the pulling force applied to remove dolly from the coated test coupons or until the coating system fails. Two test coupons were evaluated: before and after salt-spray test. During the pull-off test, five detachable aluminium loading fixtures (dolly) with 20 mm in diameter were glued perpendicularly onto each two test coupons using Araldite Standard (blue). The cohesive and bonding properties of this adhesive were greater than the three-layer coating, to remove the dolly from the test coupons completely. The average DFT for two test coupons used in this test was 349 and 336 µm before and after abrading, respectively. This thickness surpassed the minimum thickness requirement for multi-layer coating set at 200 µm by the end user. A perpendicular force by hydraulic pull-off Elcometer 108 (Elcometer Limited, England, United Kingdom) was exerted to the surface in an effort to remove both the dolly and coating from the test coupons at atmospheric condition (26 °C, 65 % relative humidity).

Salt-spray test

For salt-spray test, one of five test coupons was used as a control while the other four test coupons were placed 30° from the vertical inside a Q-fog chamber (Q-lab, Westlake, USA) containing an atomized salt solution at 35 °C for 2000 h. The salt reservoir was maintained at 5 % (w/v) of NaCl, pH of 6.8 and 1.3 mL/h to ensure uniform fog delivery during the test. Before that, two out of four test coupons were X-scribed manually with a sharp cutter so that it penetrated the protective coating exposing the metallic substrate. The creepage distance (max and min) of the scribed test coupons was measured using a vernier calliper, photographed and classified according to ASTM D1654-08(2016)E1 [64] rating scheme.

FTIR conformity analysis

The FTIR spectra of all the studied paint systems are illustrated in Fig. 2. All the absorption bands around 700–4000 cm⁻¹ are an accumulation of the different ingredients used to formulate paints. EPZ part A contains primarily C–O–C epoxide (1232 and 1294 cm⁻¹), –C = benzene (745 cm⁻¹) and –OH (3419 cm⁻¹) functional groups. These groups can be assigned to the absorption bands of bisphenol A, xylene and 1-methoxy-2-propanol, respectively. Similar absorption bands can also be observed in the spectra of Epoxy part A. Both EPZ and Epoxy part A contain a common organic binder such as epoxy resin in large quantities (∼10 wt %). The difference between the two lies in the presence of zinc dust. The zinc dust can only be detected using Far IR at 454 cm⁻¹ [65] because it is an inorganic metal. As for the curing agent, both EPZ and Epoxy part B exhibit the absorption bands of C–N (1042 cm⁻¹), –NH (1249 cm⁻¹) and –C= (732 cm⁻¹). These bands can be assigned to 2,4,6-tris(dimethylaminomethyl)phenol, triethylenetetramine and ethyl benzene, respectively.

Fig. 2:

Overlay of FTIR spectra for both part A and B of the paints of 2-pack three-layer maintenance coating system. Solid line – properly stored retained paints while dotted line – improperly stored retained paints.

PU part A consists of mainly polyacrylate resin. The absorption bands of this resin are located at 1164 and 1732 cm⁻¹, which can be assigned to –C–COO– and –C=O of n-butyl acetate respectively. PU part B contains active hexamethylene diisocyanate-based polymer, which is exhibited by the absorption bands at 2261 cm⁻¹ (NC=O) and 1735 cm⁻¹ (–NCO). Detailed assignments of FTIR wavenumbers for all paints are tabulated in Table 3.

A general outlook of FTIR spectra for all paint systems with different conditions (fresh, aged and expired) shows that these spectra are quite similar except for the intensity differences between the spectra of properly stored and improperly stored retained paints. As the paint spectrum generally exhibits absorption bands from the resin, any changes occur to the resin will be reflected in the spectrum. The higher absorption bands of atmospheric water vapor (3000–3800 cm⁻¹) is estimated due to the photooxidation of unsaturated groups of the resins, while carbon dioxide (2000–2300 cm⁻¹) is attributed to carbonyl group formation which includes ketones and lactones [29].

The properly stored paints at different aging duration did not show any signs of discoloration, pungent smell (fishy-like), lumpy and grainy bits, while the improperly stored paints show early signs of paint defects. All the improperly stored paints show solvent leakage underneath its containers, probably due to expansion caused by heat. This may eventually affect paint viscosity, more importantly it also affects base material (part A) that contains solid particles. During the sampling of improperly stored paint, a soft “pop” sound was heard and this noise is a result of the air pressure escaping (gassing was formed inside the retained paint containers).

Besides gassing, the solid content of EPZ part A settled down at the bottom of the container and this resulted in lumpy paints and hard to be homogeneously stirred. Other defects include color change (brown-purplish to dark brown) and sharp smell (fishy-like) were exhibited by amine-based curing agents of Epoxy paint. These changes are most likely caused by the chemical changes after absorption of atmospheric carbon dioxide [66]. The reaction of isocyanate-based PU with moisture in the air may also be accelerated when the paint is exposed to humidity, heat and presence of sunlight for a certain period of time. This effect can be identified when there is a formation of gas bubbles during stirring [1].

Fresh sample

The distribution of r values across different sampling locations in the mixing tank for different paint systems is illustrated in Fig. 3. Results from in-house laboratory (open markers) show that the r values for all the fresh paint samples in referenced to the respective Reference spectrum are well above 0.900 ± 0.002 for both (i) whole wavenumber region (4000–700 cm⁻¹) and (ii) specific fingerprint region (2000–900 cm⁻¹). This indicates that all paints at Top, Middle and Bottom of the mixing tank were well-stirred. The variation of r values is less than ±0.010 when analyzed in the in-house laboratory. When the samples were sent out for verification in a 3rd-party laboratory (solid markers), the standard deviations are slightly higher (±0.030).

Fig. 3:

Estimation of r values against the respective Reference spectrum generated from in-house (open markers) and 3rd-party (solid markers) laboratories for fresh paint samples collected from Top, Middle and Bottom of mixing tank. Blue dash line marks the acceptance criterion set at r = 0.900 with tentative tolerance of ±0.002.

Qualification between in-house and 3rd-party Reference spectra of different paints is tabulated in Table 4. All the paints pass the qualification test, calculated after eq. 1, with an average r_Ref of 0.97 ± 0.02 for (i) the whole wavenumber region and (ii) r_Ref = 0.98 ± 0.01 for specific fingerprint region. These values suggest that the Reference spectra generated by the paint manufacturer in-house laboratory can now be used for other in-house production checks such as routine batch check, retained sample check and on-site sample verification. This geometric mean is suitable for consistency analysis as the generated r value needs to be close to and not lower than 0.80 even if the other one is 0.99 [74].

In the previous study, good reproducibility for routine batch has been successfully demonstrated through the consistent r values of Epoxy paint that are above 0.900 that was elucidated at three different fingerprint regions in accordance to IMM FP01:2020 (before migrated to MS7236L2022) [13, 75]. For the purpose of simplicity for on-site verification performed by coating inspector, a newly proposed “universal” specific fingerprint region (900–2000 cm⁻¹) has been statistically demonstrated for different paints, either for part A or part B [36]. Therefore, in the continuation to previous study, the retained samples check was conducted with a newly proposed specific fingerprint region in accordance to MS2736L2022 [48] using different paints stored at different condition.

Retained sample

Figure 4 shows the r values of retained paints (aged and expired) that were properly and improperly stored in reference to the qualified fresh paint. The open square marker refers to the properly stored aged and expired samples while the open triangle marker refers to the improperly stored retained paint samples. Result shows that the r values of properly stored aged and expired samples are well above the acceptance criterion at r ≥ 0.900 with tolerance ±0.002 for both (i) whole wavenumber region and (ii) specific fingerprint region. On the contrary, the r values of improperly stored retained paint samples are below the acceptance criterion for both regions. High r values would have been expected since the aged paints are still within the shelf life. This is only true for properly stored aged paints. By duration, the properly stored expired paints unexpectedly show high values of r (r > 0.920) despite exceeding their shelf life of more than 6 months (PU part B), 9 months (PU part A), 12 months (EPZ part B, Epoxy part A) and 24 months (EPZ part A, Epoxy part B). Since all properly stored paints pass the acceptance criterion, these suggest that the retained paints, both aged and expired, have insignificant chemical reactions (such as degradation) when they are properly stored in the warehouse. The accuracy of measurement can be improved by only allowing the paint containers to be opened right before sampling analysis. In terms of data uncertainty, a small deviation of r values (0.001–0.009 for aged paint and 0.001–0.012 for expired paint) can be observed across replications and it is within the tolerance limit of sampling error.

Fig. 4:

Estimation of r values against the Reference spectrum generated from in-house for properly stored (square markers) and improperly stored (triangle markers) retained samples. Blue dash line marks the acceptance criterion set at r = 0.900 ± 0.002.

A paired Student’s t-test was further conducted in order to differentiate a mean population of retained paint sample stored under different conditions. Based on the average difference (D‾) between the r values of properly and improperly stored shown in Table 5, the calculated t-statistic values for both regions are far above t-critical while the p-values are below 0.05. Both unfulfilled criteria of null hypothesis (H₀) (t-statistic < t-critical and p > 0.05) indicate that the alternative hypothesis (H_a) may be accepted, which translates to the r values of properly stored retained paints are significantly difference from improperly stored retained paints. Lower r values with r in the range of 0.744–0.878 are obvious for paints that display physical changes as mentioned earlier. Further tests related to adhesion and salt spray was conducted to investigate the performance if the dried coatings of these expired paints perform equivalently to that of the dried coatings of fresh paints.

Adhesion pull-off test

Table 6 shows the area of test coupons after removal of dollies for expired 2-pack three-layer maintenance coating system. The detached area reflects an interface strength between the carbon steel substrate, coatings and adhesive. Comparison was made for the coupons (a) before and (b) after salt spray. When the coating residue on selected test coupons is compared against the paint color shown in Table 1, there are two distinct colors. The dark grey color exhibits primer coating while the light grey color exhibits midcoat layer. Coupon-A1 has light grey coating left at the detached area, which implies that the topcoat was successfully removed along with the dolly. On the contrary, the dark grey coating on Coupon-A2 indicates that both midcoat and topcoat were removed along with the dolly.

Table 6:

Pull-off (dolly) adhesion test on two coupons (-A1: before and -A2: after salt-spray exposure).

Test coupon	Detached area on the test coupons after removal of dollies
	1	2	3	4	5
Coupon-A1
Coupon-A2

The dry (before salt spray) and recovery (after 2000 h of salt spray) adhesion strength exerted by dollies from the coated test coupons are summarized in Table 7. Average dry pull-off strength of Coupon-A1 is 10 646 ± 1551 kPa. This substantial strength indicates a series of good industrial practices: that the surface of metal substrate had been adequately prepared, each coating layer had been appropriately cured and each paint still possessed good physical properties. High adhesion strength is expected for fresh paint but not for expired paint. The amount of cohesive failure on the affected coating layer (midcoat) is 5 %, with an adhesive failure of 85–90 % between midcoat/topcoat and 5–10 % between primer/midcoat.

After exposure to salt solution, the Coupon-A2 has a recovery pull-off strength of 4447 ± 351 kPa. This is lower by almost half the strength of Coupon-A1, which then suggests that the adhesion strength weakens after 2000 h of exposure in the simulated environment. Despite having lower adhesion strength, the strength of sprayed coupons is still above the passing remark (pull-off strength > 2069 kPa). The cohesive failure takes place on the midcoat and it ranges from 20 to 30 %. A large adhesive failure was detected on the primer/midcoat (70–80 %) due to excessive hydration of the substrate – high humidity results in greater cohesive failure rate with lower rate of adhesive effectiveness and durability.

Salt-spray test

Figure 5 shows the photographs of pre-blast surface cleaned test coupons before and after salt spray test. On the unscribed coupons (Coupon-B2 and -B3), there are no coating defects related to blistering, rusting, cracking, flaking, and wrinkling. The coatings on both plates appear to be dull and slightly yellowish when compared to the control Coupon-B1. On the scribed coupons (Coupon-B4 and -B5), the white coating remains intact near the X-cut despite being immersed in corrosive environment for 2000 h. Close visual examination on the coupons reveals that the creepage distance between scribed coupons remains unchanged. There is no expansion on the affected area of protective coatings. This proves that the zinc dust in ZRP (primer) sacrifices itself to protect the metal substrate from corroding. All the test coupons of expired coating system are rated 10 based on the absence of failure site on tested coupons. The unscribed coupon shows barrier corrosion protection by topcoat (PU) while the scribed coupon shows cathodic protection by primer (EPZ). Both the rating outcome and visual observations suggest that the expired coating system could perform within the end user specifications up to at least 8 months of service. Both the rating outcome and visual observations suggest that the expired paints (at least 8 months exceeding the shelf life) could perform within the end user specifications.

Fig. 5:

Photographed images of test coupons before (-B1: control) and after (-B2, -B3: scribed and -B4, -B5: unscribed) 2000 h of exposure in corrosive environment.