It was found that the 'dark circle' shown in the height image and the phase image (Figure 5c, c') was close to the nanopores shown in the TEM image. So, the morphology and the phase separation observed in TM-AFM images could be verified.
The phase image contrast revealed in Figure 5 can be interpreted according to the phase shift-rsp curve. For rsp > 0.85, the phase shift is negative both on PB and matrix, and the phase shift difference between PB and matrix is small, which accounts for the obscure phase contrast obtained at rsp = 0.95 (shown in Figure 5a'). As rsp is in the region of 0.77 to 0.85, the phase shift becomes larger on the matrix than on PB, and the largest phase shift difference between PB and matrix was observed. So, in the phase images obtained at rsp = 0.77 (Figure 5c') and at rsp = 0.85 (Figure 5b'), the PB-rich region shows to be darker than the matrix region, and the phase images show maximum contrast. While the rsp decreases to 0.1 (Figure 5d'), a clear phase contrast is also obtained, and the PB-rich region remains to have a dark contrast. Additionally, it was found that the size of PB-rich domains decreases compared with that shown in Figure 5b', c' in both height and phase images. In the blend, the cross-linked epoxy resin of the matrix is a thermoset polymer, and the PB polymer of the domain materials is liquid at room temperature . The two materials are quite different in stiffness. So, it is necessary to take the indentation depth of the probe tip into consideration. The indentation depth-rsp curve (Figure 7b) reveals that the indentation depth of the PB-rich region is larger than that of the matrix when rsp is more than 0.8, and it is reverse when rsp is less than 0.8. For larger rsp, the tip-sample interaction on the matrix of PCL mixed with epoxy will be more affected by the attractive capillary force than on the PB domains because the matrix of epoxy resin is more hydrophilic than the PB domain due to the oxygen atoms. So, the stiffer matrix shows larger indentation depth. For smaller rsp, the indentation depth is nearly identical for the PB-rich domain and for the matrix, and the indentation depth is just about several nanometers. Considering that the radius size of the PB domain (approximately 10 nm) is close to the tip curvature radius (approximately 10 nm), when rsp is dramatically decreased, the tip contacts both the PB domain and the matrix simultaneously. The stiff edge of the domain will prevent the tip from indenting into the compliant PB-rich region. So, the size of the domain obtained is just the size of the tip, which results in the shrinkage of the measured size of the PB domain.
Many studies showed that to obtain images which describe the 'true' topography of a sample surface, the images should be recorded using sufficiently high set-point amplitudes and high rsp values. In this study, at rsp = 0.95 to 1.0, significant tip indentation in the PB region and the matrix did not take place (see Figure 7b). We thought that at this condition, the true topography is obtained (Figure 5a), but at rsp = 0.85, the height image contrast shown in Figure 5b was completely reverse. We found that in the phase shift-rsp curve, at rsp = 0.85, the interaction force between the tip and the sample varied from the attractive interaction on the PB into the repulsive interaction on the matrix (Figure 7a), leading to a large decrease in amplitude. Thus, the PB region looks higher than the real topography in the height image as far as the feedback mechanism is concerned. So, a height artifact is observed in Figure 5b, instead of the height image contrast inversion. While rsp ≤ 0.77, the tip-sample interaction force is repulsive for both the PB and the matrix, so the height images did not undergo reverse but influenced by the effect of tip indentation.