III. RESULTS AND DISCUSSION

Ion implantation of PI results in both the disruption of phenyl rings and degradation of imide groups. The ether linkage degrades first and imide groups are gradually converted into amide ones with CO being the major gaseous emitted product. ^{Reference Sviridov17} Further evolution yields iminic- and pyridinic-like groups as well as tertiary amines, ^{Reference Sviridov, Odzhaev, Kozlov, Wise, Wnek, Trantolo, Cooper and Gresser18} thus, gradually converting the polymer exposed to high-fluence implantation toward the structure with extended polycondensed systems as proposed in Ref. Reference Popok, Azarko, Khaibullin, Stepanov, Hnatowicz, Mackova and Prasalovich28 and shown in Fig. 1. The chemical structure of PMMA is presented in Fig. 2. Ion implantation of this polymer is known to cause scission of chains, for instance, at the oxygen-containing group and between the carbon in the backbone and in CH₂ group yielding low molecular weight species that can easily evaporate. ^{Reference Kondyurin and Bilek29} Thus, free radical formation and degassing promote cross-linking and depolymerization transforming PMMA into a structure with heavily carbonized areas consisting of aromatic rings (as shown in Fig. 2) having minor inclusions of residual PMMA. In other words, the implanted PMMA layer evolves into pyrolytic carbon-like structure. By comparing the degradation schemes for PI and PMMA, it can be concluded that structure and composition of both polymers significantly evolve under the ion implantation. However, PI is known to have higher resistance against radiation compared to PMMA. ^{Reference Popok and Norris15,Reference Popok, Azarko, Odzhaev, Toth and Khaibullin30}

FIG. 1.

(a) Chemical structure of pristine PI and (b) final evolution of PI under high-fluence ion implantation. For more details about intermediate transformation stages see the text.

FIG. 2.

Chemical structure of PMMA and final evolution of PMMA under high-fluence ion implantation. For more details about intermediate transformation stages see the text.

AFM images of the PI and PMMA surfaces are shown in Figs. 3 and 4, respectively. Typical roughness of unimplanted samples is found to be around 2 nm for both polymers and 2 × 2 μm areas. Sometimes it is possible to see trenches or ridges as shown, for example, for PMMA in Fig. 4(a). These features can locally increase the roughness. For this scanned area, the root mean square value (RMS) is found to be 2.7 nm. Implantation with F = (1.0–2.5) × 10¹⁶ cm⁻² does not lead to significant changes in the PI surface morphology compared to the pristine polymer except appearance of a few nanometer high bumps as can be seen in Fig. 3(a). Fluence increase up to (5.0–7.5) × 10¹⁶ cm⁻² causes significant restructuring of the PI surfaces. They become much rougher (see RMS values in the panels) and formation of numerous small bumps that cover entire implanted surface is observed. Along with small bumps larger features with a height of 20–40 nm and lateral basal dimensions of 100–300 nm are observed on the PI surfaces [see Figs. 3(b) and 3(c)]. Modification of the surface morphology is qualitatively very similar for PMMA as one can see by comparing panel (b) of Fig. 3 with panel (b) of Fig. 4 corresponding to the same fluence. For PI implanted to higher fluences there is no qualitative change in the surface topography as can be seen by comparison of Figs. 3(c) and 3(d) except that the roughness is increased. However, for PMMA, number of large bumps is decreased with fluence and they are not observed for F = 1.2 × 10¹⁷ cm⁻² [compare Figs. 4(b)–4(d)]. This tendency leads to appropriate decrease of RMS.

FIG. 3.

3D AFM images of PI implanted by Cu⁺ ions with fluences of (a) 2.5 × 10¹⁶ cm⁻², (b) 5.0 × 10¹⁶ cm⁻², (c) 7.5 × 10¹⁶ cm⁻², and (d) 1.0 × 10¹⁷ cm⁻². RMS represents root mean square values for surface roughness.

FIG. 4.

3D AFM images of (a) pristine PMMA and PMMA implanted by Cu⁺ ions with fluences of (b) 5.0 × 10¹⁶ cm⁻², (c) 1.0 × 10¹⁷ cm⁻², and (d) 1.2 × 10¹⁷ cm⁻². RMS represents root mean square values for surface roughness.

It is well known that low-energy implantation to fluences above approximately 1.0 × 10¹⁶ cm⁻² leads to nucleation of metal NPs in thin polymer layer. ^{Reference Stepanov10,Reference Popok13} Earlier studies showed that thickness of the modified layer slightly varies depending on the polymer type and metal ion species, but for energies of 30–40 keV typically does not exceed a few tens of nanometers with the maximum metal concentration located at very low depth. ^{Reference Popok13,Reference Stepanov, Abdullin, Petukhov, Osin, Khaibullin and Khaibullin31,Reference Yuguang, Tonghe, Yawen, Gu, Huixing and Xiaoji32} Assuming that NP growth occurs by successive joining of single metal atoms one can conclude that the nucleation process is governed by the local copper concentration. Therefore, NPs are formed at the depth corresponding to the highest copper concentration, i.e., in the very shallow layer. However, high-fluence ion implantation leads not only to the NP nucleation but also causes sputtering of the polymer surface. ^{Reference Popok13} Sputtering rates of polymer materials used in these experiments are few times higher compared to copper atoms incorporated into substrate during the implantation. Since metal NPs are very shallowly located, sputtering of thin polymer layer at later stage of implantation may originate the towering of already formed (at earlier stage of implantation) NPs. This scenario was proved for a number of materials like SiO₂, sapphire, etc. implanted by different metal ions. See for example Ref. Reference Popok, Gromov, Nuzhdin and Stepanov33. Thus, a number of tiny bumps with a height of few nanometers visible in Fig. 3(a) can be a consequence of such sputtering and appearance of some NPs on the surface.

Increase of implantation fluence leads to the increase of particle density, and many NPs become partly towered above the surface as can be seen in Figs. 3(b) and 4(b). Along with small bumps representing individual NPs, the larger features are observed for both polymers. They are shown in Figs. 3(b) and 3(c), Figs. 4(b) and 4(c) and can be assigned to agglomeration or coagulation of copper NPs. The association of the bumps with NPs is confirmed by the agreement between the topography and phase AFM images. They are presented for selected fluences and PMMA in Fig. 5 to be able to make direct comparison for the exactly same area. One can clearly see that the bumps associated with NPs in topography are well reproduced by the phase change. Further support is obtained using SEM study where clear contrast between the metal-rich (bright) areas and the rest of the surface is observed (see Fig. 6 as an example). Size of bright regions in SEM images is found to be in good agreement with lateral dimensions of the bumps observed by AFM for the corresponding implantation fluence [compare Figs. 3(b) and 6]. Unfortunately, AFM does not allow detailed statistical study of NP sizes because it is not known for how much a particle is towered above the surface. It can be ¼, 1/3, ½, or other fraction of the diameter.

FIG. 5.

2D AFM images of PMMA implanted by Cu⁺ ions. Panels (a) and (b) correspond to fluence of 1.0 × 10¹⁷ cm⁻² showing height (topography) and phase scans, respectively. Panels (c) and (d) correspond to fluence of 1.2 × 10¹⁷ cm⁻² showing height (topography) and phase scans, respectively.

FIG. 6.

SEM image of PI implanted by Cu⁺ ions with fluence of 5.0 × 10¹⁶ cm⁻². Image size is 2 × 2 μm.

One can see that the morphology of the sample surfaces is very similar for both types of polymers at moderate fluences (∼10¹⁶ cm⁻²), i.e., one can suggest quite similar mechanisms for NP formation in PI and PMMA. However, at very high fluences, large bumps associated with coagulates of NPs start disappearing in PMMA and they are not observed for the highest fluence of 1.2 × 10¹⁷ cm⁻². Such change can be explained through the consideration of properties of the pristine polymers. PMMA has much lower T _g compared to PI. High fluences mean longer implantation time compared to moderate ones, i.e., rise of mean temperature of the polymer surface can be assumed. As mentioned in the section “Experimental details”, the mean sample temperature under the very high-fluence regimes reaches the values very close to T _g of PMMA while it is much below of T _g for PI. Thus, heating of the PMMA samples can lead to softening of the material and change its viscosity. It was shown elsewhere that metal implantation into polymers with high viscosity affects the diffusion mobility of metal atoms, thus, leading to more homogeneous distribution that hampers coagulation into large granules. ^{Reference Abdullin, Stepanov, Osin, Khaibullin and Khibullin34} This may very well be the reason in our case. Softening of PMMA under very high fluences prevents Ostwald ripening and formation of large agglomerates of NPs while for PI the softening is lower and the agglomerates can form.

Visually, both polymers change to brown color under the implantation and acquire metallic luster at the highest fluences. The change in color is related to the carbonization accompanied by the formation of conjugated bonds due to the above-described structural evolution. Optical transmission spectra of the Cu-implanted PI and PMMA samples are presented in Figs. 7 and 8, respectively. Ion implantation drastically decreases transmittance of the samples in the UV–visible interval of spectrum that is related to the evolution of polymer structure and composition, in particular causing a decrease of optical gap. This is a well-known phenomenon extensively studied for different polymers and various ion species (see for example Refs. Reference Popok and Norris15,Reference Sviridov17,Reference Sviridov, Odzhaev, Kozlov, Wise, Wnek, Trantolo, Cooper and Gresser18,Reference Mackova, Malinsky, Miksova, Hnatowicz, Khaibullin, Slepicka and Svorcik25 and references therein), and therefore, we do not address it further.

FIG. 7.

Optical transmission spectra of pristine and ion-implanted PI. Fluences are shown in the panel.

FIG. 8.

Optical transmission spectra of pristine and ion-implanted PMMA. Fluences are shown in the panel.

Our focus is on the study of plasmonic properties of NPs. In Figs. 7 and 8, an absorption band with minimum of transmittance between approximately 610 and 640 nm for both implanted PI and PMMA is seen. The band becomes pronounced for PI at F ≥ 5.0 × 10¹⁶ cm⁻². For PMMA, it is present only for the highest fluence. The band can be assigned to LSPR on copper NPs. ^{Reference Ganeev, Ryasnyansky, Stepanov and Usmanov35,Reference Stepanov, Popok, Hole and Khaibullin36} However, this LSPR band is relatively weak and broad that is related to the NP environment in the radiation-damaged polymers, i.e., to the dramatic change of dielectric function under the implantation due to the conversion of shallow polymer layers into amorphous carbon-like material. The irradiation causes significant carbonization that leads to dumping of the LSPR extinction and causes the band broadening. ^{Reference Stepanov, Nicolais and Garotenuto9,Reference Stepanov, Abdullin, Petukhov, Osin, Khaibullin and Khaibullin31}

From AFM study, it was suggested that NP formation under the implantation with moderate fluences was very similar in both polymers. Hence, one can expect similar optical properties related to the copper NPs. However, by comparing the spectra presented in Figs. 7 and 8, one can see that LSPR band is observed in PI for lower fluences, starting from 5.0 × 10¹⁶ cm⁻², compared to PMMA where the band is found for the highest fluence only. Reason for the quenching of LSPR in PMMA for moderate fluences is not very clear. It can be related to the difference in electronic properties of the radiation-modified PI and PMMA. For example, the formation of high-conductive medium (due to more efficient conjugation) surrounding the particles in PMMA and, thus, leading to the dumping of the resonance can be suggested as a possible reason. However, this phenomenon requires further detailed investigation. On the other hand, the band observed for the implanted PMMA is sharper compared to the PI cases. This fact is most probably related to PMMA softening causing more homogeneous metal distribution and lower size dispersion of NPs as suggested while discussing the AFM results.