Metallized DNA

Chapter 4.2.7, Discussion

Comparison with Theory

The experimental values shall now be compared to theoretical calculations done by Slistan-Grijalva et al They applied Mie theory up to the tenth multipole for silver nanospheres in water. Although the metallized DNA structures are not assumed to be spherical particles, the experimental and theoretical values shall be compared to get a first insight about the processes happening during metallization. The theoretical values, that are shown in fig. 4.11 and 4.12, agree with experiments I did with commercial silver spheres in a range from 20 to 80nm in diameter, shown in fig. 4.13.

For small particles, the peak position remains approximately constant. Small red shifts with increasing size are due to a radius-dependent dielectric function. Very small particles show broad peaks: In bulk medium the mean free path of electrons is 52nm. Thus for particles with a radius shorter than 26nm an additional damping mechanism, surface scattering, becomes important and increases the peak width. According to the electrostatic calculation in 2.2.1, the peak intensity is proportional to the volume. This is at least quantitatively in accordance with Slistan-Grijalva's results for small particles. Theory predicts a significant red shift in the plasmon spectrum for larger particles since retardation effects, such as screening of the incoming wave, shift the plasmon peak to the red. The excitation of higher order modes, such as the quadrupole mode, broadens the spectrum.

One aspect Slistan-Grijalva does not discuss, is polydispersity of the particle sizes. Thus experimental spectra will certainly show broader spectra than Slistan-Grijalva predicts.

Another difference between my measurement conditions and Slistan-Grijalvas calculations is the refractive index of the surrounding medium. Slistan-Grijalva chose water and therefore a refractive index of 1.33. In the UV-metallization experiments, the aqueous solution contains various amounts of salt, the refractive index is therefore slightly increased. Moreover the influence of the DNA, namely its refractive index and the core-shell character of the resulting nanoparticles must not be ignored. The refractive index influences the absolute values of the peak position, the maximum absorbance as well as the bandwidth. The result of the UV metallization is, that the metallization process worked for all DNA samples. All spectra showed plasmon peaks in the same range and with similar shapes and widths. There were slight, reproducible differences, indeed, that shall be concluded and discussed in the following: All measurements show plasmon peaks in the range between 405 and 425nm. As the refractive indexes of neither the buffer solution nor the DNA are known, the values cannot be quantitatively compared to the theoretical values. It can be only concluded that the particle radius is probably below 20nm. Taking the peak position as the sole indicator for particle size, the smallest particles, corresponding to the shortest peak wavelengths, occurs for ssDNA, for high silver concentrations and long irradiation times. The former can be understood as ssDNA has a shorter persistence length than dsDNA and therefor forms bundles that are at least in one direction smaller than the dsDNA. The latter cannot be understood with this simple theory and will be explained below. A possible explanation for the third one is that the DNA strands break during irradiation and therefore the particle size decreases. An alternative explanation will be given below. The largest particle sizes occur for long DNA, short irradiation times and very high GC-contents. The large particle size for long DNA matches the expectations as long DNA means more seed positions and therefore larger nanoparticles. The large size for short irradiation times cannot be explained with this theory, only conjectures can be made about its origin. The large particle size for high GC-contents shows that the metallization is, indeed, sequence dependent.

The FWHM is mostly between 80 and 100nm. Assuming complete monodispersity, which certainly does not hold in reality, this means diameters below 9 or above 70nm according to Slistan-Grijalva. The smaller size is more in accordance with the results of the peak positions and therefore more probable. In reality, however, polydispersity, the deviations of the refractive index from that of water as well as shape effects may influence the FWHM. Most peaks were quite symmetric, exceptions are the 75%GC-sample, samples that were irradiated for a very short time and samples with long DNA. According to Henglein an asymmetric peak shape indicates a broad size distribution of particles. The reverse, however, is not always true. A slight shoulder is detectable for the 75% GC sample which might be indicative of a quadrupole mode as predicted for large (_60nm)particles. On the other hand it might reflect the particle asymmetry, a definite assignment is not possible at this stage. The background extinction, that can be assigned to extinction by unspecific silver deposition is between 2 and 30% and depends e.g. on the amount of added silver ions. There might be a tendency for a red-shift and a broadening of the plasmon peaks with the length of the DNA. To prove this, it is necessary to build up similar DNA sequences of various lengths, as the sequence strongly influences the metallization process. When increasing the amount of silver the absorption intensity as well as the bandwidth increases, when increasing the irradiation time, the absorption intensity as well as the bandwidth increases. Regarding the calculations of Slistan-Grijalva, nor of them can be explained by simple Mie theory as the absorption intensity and the bandwidth always show opposite trends. Hence a more advanced theory concerning about particle shapes, as will be discussed later, has to be applied. As a conclusion, the comparison of the measured values and the Mie theory calculated by Slistan-Grijalva show that small nanoparticles have been produces. The peak position depends on the length and sequence of the DNA as well as the measurement conditions. The results cannot be completely explained in a quantitative way and thus another theory has to be applied.

Rodlike Character

As we assume the metallized DNA to have either rod-like (whole DNA is metallized) or chain-like (nanoclusters have grown on DNA) character, two extinction peaks should appear in the spectrum. The electrostatic calculation for nanorods from 2.3.1 predicts a low intensity peak at short wavelengths around 400nm as well as a high intensity peak at long wavelengths. The latter one can be anywhere in the visible or near infrared range and strongly depends on the axis ratio. A similar behavior is also expected for chains of nanoparticles, as they might have grown on the DNA strands. Experiments with nanorods in solution, as for example done by C. Soennichsen's group [BKRS06] are in great accordance with this theory.

Recorded absorption spectra of the metallized DNA, however, have not shown a dominant long-wavelength peak, even when the scan range was extended up to 1200nm. Literature research, however, shows that not all plasmon peaks obtained for nanorods are in the wavelength range specified by the electrostatic calculation as mentioned above:

E.g. in 1989, Goudonnet analyzed silver spheroids with a length of 12nm and a diameter of 3.6nm on a quartz substrate. He identified two peaks in the spectrum, one at 468nm and one at 338nm. Qu found a 350nm plasmon peak for silver nanoparticles grown on a copper foil by silver mirror reaction. In his PhD thesis, M. Noyong calculated the plasmon spectrum for different aggregates of nine silver spheres with a diameter of 40nm. He found one peak around 350nm and another one between 450nm and 500nm.

Possible low-intensity absorption peaks of metallized DNA in a range below 400nm might be covered by the absorption of the DNA and the nitrate. Thus the spectra were deconvoluted to eliminate all peaks that are not related to plasmon oscillations. The result of such a deconvolution can be seen in fig. 4.14. Indeed, there are two peaks visible: One at 300nm and one at 420nm, which can be assigned to the transverse and longitudinal plasmon respectively. In the spectra the long wavelength peak is more intense and more sensitive to changes in the synthesis parameters than the short wavelength peak. This is in accordance with the electrostatic calculation. The following table lists the peak positions found for deconvoluted data.

The longer the DNA, the more separated the peaks are, i.e. the more elongated the particle is. This is again not true for the 48mer DNA. The 48mer sequences with different GC-values also exhibit two peaks. There is, however, no trend observable. The longer the samples are irradiated, the closer the peaks approach. This indicates a decrease of the particles' axis ratio: The maximum length of the nanoparticles is determined by the length of the DNA and therefore does not change during metallization. With increasing irradiation time, more silver is deposited on the DNA, the particles become rounder. A similar explanation can be used to explain the peak shifts dependent on the silver-to-base amount: The more silver is added, the more spherical the particle becomes. Thus, both peaks could be identified in the spectrum but contrary to the electrostatic calculation, they are tremendously blue-shifted: One possible reason for the blue shift of the long wavelength plasmon is the charge of the nanoparticles. According to Mulvaney the injection of electrons to the surface shifts the longitudinal Plasmon 50nm/V to the red. This effect is more pronounced for rods than for spheres, as the enhanced oscillator strength makes the particle be more prone to surface perturbations. It also leads to a damping of the peak.

Concerning the literature mentioned above, the blue shifts of both plasmon peaks mainly occurs for particles in contact with a surface. The 'surface' in this experiment is the DNA. Unfortunately there is no theory in literature explaining this effect up to now.

Summing up it can be said, that small rod- or chainlike metallic nanoparticles have grown on DNA templates. Their axis ratio can be tuned by varying the length, by the added amount of silver and by the irradiation time. Unfortunately no quantitative theory is applicable to extract the axis ratios from the optical data.