1. Introduction

Composites formed by embedding nanoclusters in metal, semiconductor and insulator matrices exhibit unique mechanical, optical, electrical and magnetic properties that can be tailored by varying the cluster size. Such nanostructured materials have potential applications in opto-electronics enabling efficient circuitry on nanometer dimensions [1], all optical switching in integrated optical devices due to enhanced nonlinear properties [2], charge storage [3], memory [1], strengthening of steel [4], spintronic devices due to large spin dependent transport and magneto-optical properties [5], biology [6] and green technologies like solar cells [7]. Nanoclusters are also of fundamental interest due to their quantum-size effect wherein the energy of optical transition depends on the cluster size. This effect in a matrix leads to photolumi-nescence of such clusters, which in some instances can be called as quantum dots. Nanoclusters are embedded in materials employing techniques such as ion implantation [8, 9], RF co-sputtering [10, 11], thermal decomposition of thin grown layers [12], sol-gel [13], ion-beam mixing [14], annealing of ion-exchanged glasses [15], electron beam irradiation [16], selective oxidization [17] and by depositing nanoparticles onto a surface and over growing the deposited area with additional matrix material [18]. Laser irradiation is known to alter optical properties of materials [19] and has also been employed to induce nanoclusters in dielectric materials.

Excimer lasers were used to form metal nanoclusters by photo-decomposition of organometallic crystals embedded in Poly(methyl methacrylate) [20]. The unique capability of ultrafast lasers to localize the interaction to micron dimensions was exploited to pattern functional microcrystals in 3D inside glass [21, 22] and to induce silver nanoparticle formation in silver containing glasses [23, 24] and polymers [25] by photoreduction and subsequent post baking. Si nanocrystals and amorphous Si clusters [26] have also been shown to form when silicon-rich nitride films were treated with femtosecond laser. Several studies have also focused on laser induced transformation of nanoparticles embedded in a matrices [27, 28]. In almost all cases light interacted with matrices rich with specific atomic species leading to their precipitation into nanoclusters after post baking.

In this paper, we demonstrate nanocluster formation by localized modification of the matrix structure itself. We show the formation of photoluminescent carbonaceous and siliconaceous nanoclusters within the laser modified region of a polymer after its backbone structure undergoes rearrangement. The clusters embedded in the polymer are characterized using energy dispersive X-ray spectroscopy, micro-Raman, UV-visible and fluorescence spectroscopy. Studies on surface irradiation of polymers, diamond-like carbon films and bulk silicon have demonstrated generation of carbon [29, 30] and silicon clusters [31] in a laser ablation process. Although cluster formation mechanisms still remain unclear it is widely accepted that bond breaking occurs due to light-matter interaction resulting in formation of radicals that are active and act as cluster centres [32]. Similar underlying photochemical processes can be expected to occur when light interacts with a bulk material in spite of its differences with surfaces. In fact, refractive index modification of fused silica by femtosecond lasers is associated with a rearrangement and subsequent reduction of SiO₂ ring structure. Moreover, extensive studies conducted on ion irradiation of polymers indicate structural rearrangement of the polymer backbone resulting in formation of carbonaceous clusters along the ion tracks [33–35]. However, the primary difference being ultrafast lasers enable spatial control of localized material modification in 3D and hence the nanocluster formation that can be randomly distributed within the modified region.

2. Experimental methods

In our experiments, 800 nm light from a Ti: Sapphire laser system operating at a repetition rate of 1 kHz and producing 45 fs pulses with a maximum energy of 2.5 mJ was focussed 300 μm below the surface of PDMS by a 0.25 NA (16X) aspheric microscope objective. The back aperture of the microscope objective (5.5 mm) was slightly over filled to minimize alignment errors. The position of the laser focus relative to the surface of bulk PDMS sample was accurately determined by imaging the back reflected light with a CCD camera at very low pulse energies below the ablation threshold. The PDMS sample was mounted on three-axis translation stages with a resolution of 50 nm along the lateral dimensions (X,Y) and 100 nm along the axial direction (Z). Modified lines in PDMS were fabricated by translating the sample at 0.1 mm/s speed. A combination of a half-wave plate and polarizer were used to vary the pulse energy. The incident pulse energies were measured after the microscope objective and monitored by a calibrated fast photodiode operating in linear regime. A single-shot autocorrelator continuously monitored the pulse duration. The pulses were not pre-chirped and the duration was measured to be 70 fs at the back aperture of the objective after passing through all the optics. The use of aspheric objective, where a single asphere replaced the compound lens system of a standard microscope objective, reduced dispersion and spherical aberrations. The laser focal spot size was measured to 1.65μm using a knife edge method, in close agreement with the diffraction limited spot size.

PDMS samples were prepared by mixing Dow Corning Sylgard 184 silicone elastomer base with curing agent in 10:1 ratio. The PDMS mixture was stirred for 10 minutes and degassed by placing it in a vacuum desiccator for 30 minutes. The mixture was then poured between four glasses slides that were taped onto a silicon wafer that was cleaned by methanol. Another glass slide was used to gently spread the PDMS mixture avoiding any air bubbles. A weight of 500 g was positioned on the top of the glass slide to ensure flat top and the bottom surfaces of PDMS. The mixture was cured in the oven at 80°C for two hours. Large sheets of ~25 mm × 75 mm ×1mm PDMS samples were prepared in this manner that were subsequently cut into smaller samples of ∼ 10 mm × 10 mm × 1mm.

After laser modification of PDMS the samples were cleaved and gold coated for characterization with scanning electron microscope (SEM) and energy dispersive X-ray spectroscopy (EDS). The spatial resolution of EDS was ∼1 μm. To further characterize the laser-modified regions in PDMS, confocal micro-Raman and photoluminescence spectroscopy was carried out using laser excitation sources with wavelengths of 488 nm or 532 nm. Laser power at the sample was kept to a minimum (5 mW) to avoid further photo-induced damage of the modified regions. Micro-Raman and photoluminescence spectra were collected using the back-scattering geometry with laser light focussed on the modified regions of the sample to a 5 μm spot using 50X (0.42 NA) or 100X (0.80 NA) long-working distance objectives. The spectral resolution of the system was 1 cm⁻¹ (FWHM) for micro-Raman and 0.1 nm for photoluminescence. To enhance the signal to noise ratio and to ensure signal is collected only from the modified region several closely spaced lines were fabricated in PDMS in 3D such that the cross section consisted of 20 laser modified regions within an area of 15 μm².

For absorption and fluorescence spectroscopic studies a large laser modified area of 3 mm × 3 mm was fabricated 300 μm below the surface consisting of 1500 lines spaced by 2 μm. The UV-visible absorption spectra were recorded in the transmission mode with a resolution of 2 nm. Both the fluorescence and excitation spectra were recorded using a slit width of 15 nm and spectral resolution of 1 nm using PerkinElmer LS 50 fluorescence spectrometer. Water Raman and a reference fluorescent dye were used to calibrate the spectrometer. A confocal microscope, Nikon A1si RMP, was used to obtain images of embedded modified regions by focussing 488nm light with a 0.45 NA microscope objective.

3. Results and discussion

Figure 1(a) shows a SEM image of a cross-section of the laser modified region consisting of a void in the middle surrounded by material that is ablated from the centre. High temperatures and pressures generated within the focal volume due to the optical breakdown process transformed the material into a new form consisting of an agglomeration of nanometer sized clusters with marked faceting. This suggests that on a microscopic scale bond breaking and subsequent rearrangement of atoms/molecules has occurred within the polymer backbone. A semi-quantitative elemental analysis of different spots within and outside the laser modified region in PDMS was obtained using EDS as shown in Fig. 1(b). EDS spectra consist of three peaks corresponding to the characteristic k_α emission lines of carbon, oxygen and silicon at 0.28, 0.53 and 1.73 keV, respectively. The elemental analysis of unmodified PDMS was obtained by averaging the spectra recorded from different locations as shown by the black circles in Fig. 1(a). EDS spectra obtained from different regions on the rim of the modified region (identified by coloured circles in Fig. 1(a) show the carbon and oxygen concentrations to increase by 17–70% and 24–80%, respectively, relative to the unmodified regions. However, the Si concentration from the same spots remained unchanged within the measurement uncertainties.

 

Fig. 1 (a) SEM image of cross-section of a modified region fabricated 300 μm below the surface with a speed of 100 μm/s and a pulse energy of 200 nJ. (b) EDS spectra from different spots within and outside the modified region, identified by coloured circles in (a). The spectrum shown in black represents an average over three regions.

Download Full Size | PDF

For low density polymeric materials, the spatial resolution of the EDS probe is very high (∼ 1 μm) at low accelerating voltages enabling to map elemental composition over the entire modified region. The colour maps illustrated in Fig. 2 show the distribution of the carbon and oxygen concentrations in and around the modified region. As the silicon concentration is nearly constant both in the modified and unmodified regions, it is used to normalize the carbon and oxygen counts. The elemental concentrations of unmodified PDMS are rescaled and used as a reference to study the concentration variations of carbon and oxygen across the modified region. Figure 2 shows carbon and oxygen concentrations to randomly vary across the laser modified regions. Negative (positive) numbers correspond to low (high) concentrations relative to that of the unmodified PDMS. There are regions where their concentrations are higher while they are lower in others. The results of the EDS mapping suggest clustering has resulted from laser irradiation. Since EDS is sensitive to surface topography, data was collected only from the edges of the modified region avoiding the central void like region where depth variations are significant. Data collected from several different modified regions fabricated under identical conditions show similar results.

 

Fig. 2 Element maps of (a) carbon and (b) oxygen within and outside the modified region, outlined by the dashed circle. The scale represents elemental concentration normalized to the unmodified region.

Download Full Size | PDF

Our EDS analysis shows carbon and oxygen concentrations to vary in tandem. In addition, the C/Si and O/Si ratios are always less than unity although carbon and oxygen concentrations have increased by up to 70–80% in some regions. This is in contrast to the EDS and XPS analysis of surface irradiation of PDMS carried out with a Xe₂ excimer lamp at 172 nm [36], where the surface layer was found to be enriched in oxygen and depleted in carbon resulting in O/Si ratio of almost 2 corresponding to transformation of PDMS to SiO₂ like structure. In our experiments the femtosecond laser induced modification occurs in the bulk, 300 μm below the surface, whereas on the surface photochemical oxidation of PDMS can occur resulting in differences in carbon and oxygen concentrations and their ratios with respect to Si. Micromodification of elemental distribution has also been observed in doped glasses irradiated with light from a femtosecond laser [37,38].

Presence of carbonaceous clusters within the laser modified region, as suggested by the EDS analysis, should lead to photoluminescence (PL) [10]. Figure 3(a) shows a PL spectrum obtained from laser modified PDMS excited with light at 488 nm. The PL spectrum peaks at ∼585 nm and is broad suggesting a wide distribution of cluster sizes. The measurements were made on a laser modified cross-section of ∼15 μm² and the PL was intense enough to be seen by the naked eye. No PL signal was observed from the pristine PDMS. The inset of Fig. 3(a) shows the variation of PL intensity as a function of laser pulse energy used to modify PDMS. Increasing PL intensity with pulse energy is an indication of rise in the number of clusters. However, we did not observe any spectral shifts of the PL peaks with pulse energy suggesting that the cluster size distribution remained unchanged with laser pulse energy. Change in PL emission energy is known to vary with cluster size as the bandgap of the cluster materials changes [39, 40]. Figure 3(b) shows confocal microscope image of an array of laser modified spots obtained at an excitation wavelength of 488nm. Each row of spots is irradiated by 1, 2, 5, and 10 pulses, respectively from top to bottom with pulse energy varying from 355nJ – 430nJ in steps of 10nJ from left to right. Fluorescence signal from the laser modified spots increases with pulse energy (along the row) and with number of pulses (along the column), consistent with PL measurements.

 

Fig. 3 (a) Photoluminescence from the laser modified PDMS induced by coherent light at a wavelength of 488 nm. Inset shows the variation of the PL intensity with laser pulse energy. (b) Confocal microscope image, obtained at an excitation wavelength of 488nm, of an array of laser modified spots under different irradiation conditions. See text for details.

Download Full Size | PDF

Visible photoluminescence has been reported in Si based polymers [40–42] including PDMS [43] upon ion irradiation. PL was associated with segregation of diamond like carbon in an amorphized structure of the polymer induced by ion irradiation. PL in visible spectral range was also observed from carbon clusters embedded in SiO₂ [10, 44]. The green PL is often associated with carbon clusters whereas the blue PL is associated with silicon clusters such as those formed by Si-O, Si-C and Si-O-C complexes [45]. However, the PL bands from Si are known to shift to red wavelengths with increasing cluster size [46] due to quantum confinement effect. Consequrntly, the green PL observed in laser modified PDMS can also be associated to silicon clusters although EDS analysis indicated no change in Si concentration.

To differentiate the contributions of carbon and silicon containing clusters, we carried out micro-Raman measurements on the laser modified and pure PDMS. Figure 4(a) shows the Raman spectra of pure PDMS (black) and in laser modified PDMS (red) at an excitation wavelength of 532 nm. The intense peaks in pure PDMS at 2965 cm⁻¹ and 2907 cm⁻¹ correspond to the stretching modes of the methyl group. The peaks at 484 and 708 cm⁻¹ correspond to Si-O-Si stretching mode and Si-C symmetric stretch respectively [47, 48]. Raman spectra of laser modified PDMS show a broad luminescence without any features with a maximum intensity at 503 cm⁻¹. Any signatures of carbonaceous clusters corresponding to graphitic [49, 50], C=C [51] and C=O modes which would appear in the range of 1300–1700 cm⁻¹ of frequency modes are likely be hidden under the intense luminescence of the modified region making it difficult to observe them. Excitation at longer wavelengths such as 785 nm did not reduce the background to observe signatures of carbonaceous clusters.

 

Fig. 4 (a) Raman spectra of laser modified (red) and pristine (black) PDMS at an excitation wavelength of 532 nm. Note the strong photoluminescence superimposing the Raman scattering arising from the modified region. (b) Raman peak shift (around 484 cm⁻¹) in laser modified PDMS at different pulse energies relative to pristine PDMS with an excitation wavelength of 488 nm.

Download Full Size | PDF

The Raman line at 503 cm⁻¹ in laser modified PDMS is shifted by 19 cm⁻¹ relative to the Si-O-Si stretching mode of pristine PDMS. Moreover, its position lies midway between that of crystalline Si at 520 cm⁻¹ and amorphous Si at 480 cm⁻¹. This suggests structural rearrangement has occurred in PDMS upon irradiation with a femtosecond laser leading to formation of quasi-crystalline Si nanocrystals. Figure 4(b) shows the evolution of the Raman peak corresponding to Si nanocrystals as a function of pulse energy. For pulse energies of 50, 100, 200, and 300 nJ the observed shifts are 9, 15, 19, and 8 cm⁻¹ respectively. The peak shift to higher frequencies with laser pulse energy indicate transformation of silicon state from amorphous to quasi crystalline [52]. At higher pulse energies the clusters are probably destroyed resulting in a reduced size distribution and smaller peak shift. Such a behaviour has also been observed in PL measurements of ion irradiated polymers at higher fluences [40]. The intensity of the peak also increases with the pulse energy suggesting an increase in the number of the siliconaceous clusters.

The size of the silicon nano-crystalline clusters can be obtained from the bond polarizability model [53, 54]. In crystals with well-defined geometries and cluster sizes below 4 to 5 nm this model provides a better agreement with experimental Raman shifts compared to the well known phonon confinement models [55]. According to this model, the Raman shift in a nanocrystal with diameter D is given by

Laser induced structural changes to PDMS and the presence of carbonaceous and siliconaceous clusters alters light absorption compared to pristine PDMS, as shown in Fig. 5. The band edge of pristine PDMS is ∼240 nm corresponding to a bandgap of ~5 eV [56, 57]. However, owing to a slight increase in absorption around 280 nm PDMS is transparent from 300 nm onwards. The absorption edge has shifted from UV - visible (at around 350 nm) under femtosecond laser irradiation resulting in a reduction in bandgap by about 30%. Such a change in bandgap upon laser irradiation has also been observed in other materials like fused silica [58]. Also, the overall absorption of laser modified PDMS has increased significantly but more so in the visible wavelength range rendering the modified region a brown-black colour. The cluster size can in principle be estimated from the absorption edge of UV-visible spectra of irradiated polymers [33] by correlating the band gap with the number of the carbon bonds per linear or cyclic structure. We estimate 102 atoms per cluster.

 

Fig. 5 The light absorption spectrum of the pristine (black) and laser modified PDMS (red). Inset shows the repeating monomer unit of PDMS.

Download Full Size | PDF

Chemical analysis of the defects induced by the femtosecond laser in PDMS was investigated using fluorescence spectroscopy. Figure 6(a) shows fluorescence emission spectra from modified PDMS excited at different wavelengths. Pristine PDMS is known to exhibit minimal fluorescence among plastic materials [57] and it decreases as the laser excitation wavelength is increased. In modified PDMS, emission intensity is maximum when excited with UV light at 250 nm and peaks at 400 nm with a shoulder at 445 nm (not shown in the figure). As the excitation wavelength increases the emission not only decreases in intensity but also shifts to longer wavelengths due to red-edge excitation. However, for excitation wavelength of 400 nm the emission intensity is higher than at 350 nm and then subsequently decreases with increasing excitation wavelength. Also, two new additional peaks appear at 530 nm and 608 nm when excited at 400 nm (peaks identified as b and c in Fig. 6(a). Such an emission behaviour suggests fluorescence emission from two distinct species.

 

Fig. 6 (a) Fluorescence emission spectra at different excitation wavelengths. (b) Red-edge excitation effect in laser modified PDMS for the three peaks identified in (a). Inset shows variation of fluorescence intensity of the three peaks as a function of excitation wavelength.

Download Full Size | PDF

Figure 6(b) shows the variation of the emission peaks with excitation wavelength. All three visible emission peaks (identified in Fig. 6(a) shift to longer wavelengths as the excitation wavelength is increased to the red edge of the absorption band. The linear relation between emission peaks and the excitation wavelength indicates red-edge excitation (REE) effect in PDMS. REE is related to excited state inhomogenuity due to dipole-dipole interactions between an ensemble of excited fluorophores and molecules in their surroundings [59]. It arises when the rigid matrix in which the fluorophores are embedded does not allow them to undergo relaxation on the time-scale of fluorescence decay. In condensed media, where energy of every state becomes distributed, absorption of a high energy photon causes excitation of fluorescence of all possible configurations which differ in interaction energy. When the system is excited with a lower energy photon (red edge of the absorption band) only the fluorophores constituting a part of the distribution will then be selectively excited. If there are no relaxations in the medium, then the emission energies of these fluorophores will also be lower. As a result, their emission spectra will be shifted towards longer wavelengths, compared to those excited at the band maximum. REE is most pronounced in polar solvents but has also been observed in polymer matrices [60–62].

The inset of Fig. 6(b) shows the variation of fluorescence intensity of the three emission peaks with the excitation wavelength. The emission peak at 420 nm in Fig. 6(a) is very broad and has been observed to shift up to 645 nm for excitation at 540 nm. The emission peak at 530 nm has a narrower bandwidth and it declines fast as the excitation wavelength increases, while the peak at 608 nm has broader bandwidth and decreases slowly.

Fluorescence excitation spectra enable to identify electronic transitions in functional groups of polymers. Transitions between occupied (one non-bonding n type and two bonding σ and π type) and unoccupied (two anti-bonding σ* and π* type) molecular orbitals are responsible for absorption and fluorescence [63]. Figure 7 shows excitation spectra measured for different emission wavelengths. We observed four distinct absorption peaks that shift towards higher wavelengths with increasing emission wavelengths due to the red-edge excitation effect. The first peak, represented by a in Fig. 7, appears at 340 nm, undergoes a red shift until 415 nm and is responsible for fluorescence emission from 400 nm to 500 nm. The second peak, represented by b, starts at 370 nm and persists until 480 nm. It is responsible for fluorescence emission from 480 nm to 600 nm. The third peak, represented by c, emerges at 460nm and disappears after 560 nm. It is responsible for fluorescence emission from 540 nm to 600 nm. The fourth peak, represented by d, originates at 380 nm and ceases after 580 nm. It is responsible for fluorescence emission from 570 nm to 730 nm. Each of these absorption bands represent specific electronic transitions within a functional group of modified PDMS. They contribute to different regions of the broad photoluminescence observed in Fig. 3(b).

 

Fig. 7 Fluorescence excitation spectra at different emission wavelengths. Dashed lines trace the red shift of absorption peaks, identified a-d, with excitation wavelength.

Download Full Size | PDF

Origin of the peaks in the excitation spectra can be associated to the byproducts of photochemical reactions that occur when polymers are irradiated with energetic ions or photons (UV or intense infrared light). Bonds can be broken and subsequent structural rearrangement results in formation of new bonds. Although the photon energy is less than the bond energies (4.8 eV for Si-O, 3.3 eV for Si-C and 4.2 eV for C–H) multiphoton interaction of intense light with PDMS can lead to bond-breaking of the Si-O main chain or side groups Si-C and C–H. Subsequent reactions can lead to formation of carbonaceous and siliconaceous clusters.

In UV irradiation of PDMS surface in air, silanol groups were shown to form when Si and methylene radicals reacted with O forming peroxy radicals that subsequently rearranged and also when Si radical reacted with hydroxyl radical (formed by UV absorption in air) [36]. Similarly, Si-O-Si bridge was formed when oxygen radical attacked a Si-C bond [36]. Carbonyl groups were also detected in plasma modified PDMS [64]. Though the nature of interaction of light is different in bulk we anticipate similar rearrangement to occur in PDMS when irradiated with femtosecond light pulses.

Based on this we can now assign the four peaks observed in excitation spectra. The first pair of bands are associated with carbonaceous clusters involving carbonyl (C=O) or methylene groups (C=C). The first band can be ascribed to π to π* transition of carbonyl or methylene groups while the second band can be associated with n to π* transition of carbonyl group. This particular transition in both groups occurs at nearly the same wavelength [63]. These bands agree with the reported values in literature [65,66]. The last pair of bands can be associated with electronic transitions in silicon clusters containing defects such as =Si(O₂) and non-bridging oxygen hole center ≡Si-O• and =Si-O-O^•. The third excitation band, represented by peak c in Fig. 7, can be associated with non-bridging oxygen hole center ≡Si-O• or =Si-O-O^• with an absorption peak at 2.2 eV [67, 68]. The fourth band, represented by d, can be associated to the formation of dioxasilirane =Si(O₂) whose excitation energy was estimated to occur at ~3 eV [69–72].

4. Conclusion

We demonstrated formation of embedded and localized carbonaceous and siliconaceous clusters in PDMS irradiated by femtosecond laser. These confined clusters exhibited broad photo-luminescence in the visible region and contributed to the observed red-edge excitation effect. They are formed due to structural rearrangement induced by the femtosecond. The silicona-ceous clusters also include quasi-crystalline Si as suggested by Raman. Spectroscopic studies suggested two distinct electronic transitions in each of these clusters that contribute to the observed luminescence. Such photoluminescent clusters embedded in a solid matrix could find applications in high density 3D data storage, display technology and cell migration studies.