*3.3. Morphological Caracterization*

Figure 13 shows SEM images of the etched surface of the EP-160-T, EP-160-DBA, and EP-160-M samples. The images give clear information on the size and distribution of the rubber domains in the hosting epoxy matrix. It is worth noting that to better observe the microstructure of the samples, before being analyzed by FESEM, the samples were subjected to an etching process, according to a procedure reported in literature [55].

**Figure 13.** SEM images of the etched surface of: (**a**,**b**) the EP-160-T sample; (**c**) the EP-160-DBA sample; (**d**) the EP-160-M sample.

Chemical composition and conditions chosen for the functionalization process allow obtaining elastomeric domains of dimensions that do not exceed 500–600 nanometers, as seen in Figure 13a,b for the sample EP-160-T (where also the dimension of rubber domains are indicated on the left image), in Figure 13c for the sample EP-160-DBA, and in Figure 13d for the sample EP-160-M. In the case of M filler, very small crystallites of this component (not completely solubilized in the resin) are also observed in the solidified matrix.

Extensive studies have been carried out on the solubility of self-healing fillers in the components of the epoxy formulation before and after the curing process. Experimental tests and results are reported in Section S3 of the S.M. (see Figure S3-1–S3-4).

It is worth noting that the morphological feature obtained made it possible to maximize the interphase area between the rubber domains.

#### *3.4. Evaluation of Self-Healing Efficiency*

Ep-R-160, Ep-R-160-DBA, Ep-R-160-T, and Ep-R-160-M samples were tested to evaluate the self-healing efficiency, as described in Section 2.3 Methods.

The results evaluated with the TDCB geometry at 25 ◦C are shown in Figure 14. In particular, Figure 14a–c show the behaviour of load as a function of the displacement for the virgin samples Ep-R-160-DBA, Ep-R-160-T, and Ep-R-160-M (continuous line) and the same healed samples (dashed line). Figure 14d depicts the histogram of the values of healing efficiency, calculated by Equation (1), using the values of *PCV* and *PCH*, reported in Table 5. The introduction of the self-healing fillers causes a recovery of mechanical properties. The healing efficiency value is higher than 69% for all fillers. The highest value was detected for the sample with DBA, for which a value of 88% was found. Statistically, the tests performed on different samples with the same composition and treatments showed values with standard deviations of around 5%

**Figure 14.** Load-Displacement curves for the samples (**a**) Ep-R-160-DBA; (**b**) Ep-R-160-T; (**c**) Ep-R-160-M; and (**d**) histogram illustrating the healing efficiency values.

**Table 5.** Critical fracture loads values of the analyzed samples.


Most likely, the high healing efficiency values are due to polar groups in the selfhealing fillers, such as O-H, N-H, and C=O functional groups. These groups act as hydrogen bond donors or acceptors with the polar groups of the functionalized epoxy-precursor, establishing cumulative effects of the attractive reversible interactions. The attractive reversible interactions are probably established much more effectively in the higher mobility domains of the polymeric chains, therefore, at the interface between the rubbery domains and the matrix resin (where the degree of crosslinking is reduced).

Notably, the peculiar morphology of these samples, with rubber domains of dimensions in the order of a few hundred nanometres (see Figure 13), contributes to increasing the areas at reduced crosslinking density.

An example of possible interactions based on reversible hydrogen bonds is shown in Figure 15, which depicts the H-bond interactions built between the hydroxyl groups, acting as H-bond donor sites, of the cured epoxy resin and the carbonyl groups, acting as H-bond acceptor sites, of the DBA filler. The resulting supramolecular network activates self-healing mechanisms in the formulated materials.

**Figure 15.** Picture illustrating the H-bond interactions between the hydroxyl groups of the epoxy resin (red highlighted) and the carbonyl groups of the DBA filler.

Hydrogen bonds are widely recognized to activate self-healing processes [70–72], it has mostly been applied in rubbery matrices [47,73–75]. In our samples, we have small rubber domains distributed throughout the whole sample. The employment of reversible hydrogen bonds in epoxy resins has only been reported in specially modified systems through the addition of hydrogen bonding moieties such as ureido- pyrimidinone (UPy) [76,77], barbiturate and thymine functionalized MWCNTs [48], and amide motifs [78]. Reference [48] refers to our previously published paper on self-healing resins. These papers describe self-healing resins, where barbiturate and thymine functionalized MWCNTs were embedded in a rubber-toughened epoxy formulation. The groups with hydrogen donor or acceptor sites lead to reversible MWCNTs-bridges through the matrix due to strong, attractive interactions between the rubber phase, finely dispersed in the matrix, and MWCNT walls. Healing efficiencies higher than 50% have been found for both functional groups. Dynamic mechanical analysis (DMA) evidenced an enhancement in epoxy chain movements due to the rubber phase's micro/nanodomains, enabling selfhealing behavior by recovering the critical fracture load. In this paper, molecules capable of establishing strong cumulative effects due to hydrogen bonds have been dispersed in the toughened thermosetting matrix.

The authors demonstrate that the effect of the functionalization temperature should not be underestimated to evaluate the healing ability of the materials. Figure S4 of Section S4 of S.M. shows the results of self-healing tests carried out for the samples Ep-R-120 and Ep-R-160 loaded with the DBA filler (Ep-R-120-DBA and Ep-R-160-DBA, respectively). The comparison between Ep-R-120-DBA and Ep-R-160-DBA shows that the functionalization temperature of 160 ◦C allows an increase in the self-healing efficiency, as expected considering the DMA results in Figure 11, where a lower value of Tg is observed for the sample functionalized at 160 ◦C.

DMA results of this last sample including the self-healing fillers (see Figure 16) show that the Tg values of the complete formulations are lower than the same sample without filler (132 ◦C, see Figure 11) of about 10 ◦C for Ep-R-160-DBA, Ep-R-160-T, and Ep-R-160-M.

**Figure 16.** DMA curves: (**a**) Tan δ vs. temperature; (**b**) Storage modulus vs. temperature of the samples Ep-R-160-DBA, Ep-R-160-T, and Ep-R-160-M.

The modulus drop (see Figure 16b) is slightly anticipated for the sample containing the DBA filler for which the highest healing efficiency value is recorded.
