If organic fluorescent molecules are to be effectively used in such applications, they must be able to withstand repeated excitations and the large amounts of energy that will be cycled through them. Unfortunately, upon repeated absorption, the dye molecules begin to photooxidize, and they consequently lose the ability to fluoresce (Mackey 2001).
Cresyl violet (C₁₆H₁₂N₃O₅Cl), technically known as 5,9-diaminobenzo[a]phenoxazonium perchlorate (or oxazine 9 for short) and commonly referred to in the trade as LC6700 or CV670, is an efficient emitter at far red wavelengths (Castelli 1975). It is stable under ambient conditions, with minimal power saturation even at peak intensities as high as 100 MW/cm² (Moore 1978). It has a molecular weight of 361.74 g/mol, and its absorption spectrum is a good match to the emission from an Rh6G dye laser — the laser dye most widely used today.

Our experiment consisted of irradiating cresyl violet (CV) doped in thin solid slabs of poly(methyl methacrylate) with progressively higher intensities of Rh6G-laser light tuned to an absorption peak of the CV molecule. We expected the molecules would eventually photooxidize during this process of repeated excitation and relaxation of the electronic energy levels responsible for the violet color of the CV dye. This occurs because there is a weak but measurable probability that an excited molecule will make a transition into a chemically altered state as a result of a reaction with oxygen or water (originally dissolved in the dye and polymer solvents), rather than simply relaxing back to the ground state of the unreacted molecule (Schäfer 1992). By measuring the rate at which this occurs for different laser intensities, our goal was to quantify the quantum efficiency or probability for such a chemical transformation to occur. This can be accomplished by continuously monitoring the photoluminescence emitted by the relaxing CV molecules and fitting the decrease in this signal to a theoretical model describing the rate of photooxidation.

Figure 1 . The optical setup. The computer controls both the CCD detector and the monochromator (MONO). Other symbols used in the drawing: M = mirror, A1 = attenuators before the sample, A2 = attenuators at the entrance slit, L = lens, and S = sample.

Initial alignment of the optics was performed using a multi-color helium-neon laser. This provides a simple method to excite samples for use in student laboratories in which a dye laser is not available or convenient, albeit at low powers and discrete wavelengths.
It is important to take precautions to avoid spurious effects in such a setup. The pump laser must strike the sample near the edge closest to the entrance slit to the monochromator. Otherwise the fluorescence may be reabsorbed or scattered by the parts of the sample between the pumped volume and the slit. The alignment of the sample and optics must not be disturbed during a run. The sample should be monitored for burning of physical holes at high pump intensities due to the thermal load from nonradiative de-excitations. This load can be estimated from the quantum yield (i.e., the number of photons emitted per photon absorbed). As a baseline, the quantum yield for cresyl violent in methanol solution is 0.54 (Magde 1979).

Figure 2. Chemical structure of the repeat unit of poly(methyl methacrylate), abbreviated PMMA, and of cresyl violet perchlorate (CV).

Figure 3. A liquid dye solution inside the sample chamber being excited by the pump beam entering through the lens from the rear. The excited spot is near the left-hand edge of the cuvette, in the direction of the entrance slit (cf. Figure 1) off the left side of the picture.

The solid blocks were made using acetone as the solvent. High molecular weight polymer (i.e., long, unbroken chains) improves the optical quality of the final samples. We used starting PMMA material with a weight of about one million. Approximately 10 g of polymer was dissolved in 25 mL of acetone in a wide-mouthed beaker to which about 1 g of dye was added. The beaker was left under a fume hood until all the acetone had evaporated (slowly so as to inhibit bubble formation), leaving a smooth piece of flat, colored plexiglas doped at 10 wt%. This was then cut into small pieces for spectral measurement. Pictured in Figure 4 are various dye-doped polymer samples. Each is about 1 cm² in area and about 1–2 mm thick.

Figure 4. A variety of pieces of PMMA doped with different laser dyes: styryl 9M, Rh6G, pyrromethene 567, and cresyl violet from left to right.

Figure 5. Fluorescence spectra of a cresyl-violet-doped PMMA sample that is bleaching with increasing total incident laser energy (as labeled on the curves).

Figure 6. The decay in the fluorescence peak signal of CV:PMMA with pump exposure time. The blue diamonds are the background-corrected measured values, with error bars about the size of the plot symbols. The attenuators were changed at the two cusps, where the optical density (OD) in front of the sample is labeled in green. The red curve is a fit to Equation (8).

There is both a broadening and blue shifting of the unbleached fluorescence spectrum of cresyl violet in a solid sample of PMMA, compared to that of the same dye in ethanol solution. Specifically, λ_F = 625 nm is the peak emission wavelength in the polymer, which is about 5 nm lower than in the liquid. This host dependence comes about because of the interactions of the dye molecules with its neighbors. Presumably the dye in the polymer can be “frozen” into strained configurations that alter the energy levels of the optically active π-orbital electrons.

The excitation intensity (where is the pump laser power with OD the optical density of the attenuators prior to the sample) is small compared to the saturation intensity (where λ_P = 585 nm is the pump wavelength). Here the absorption cross section at the pump wavelength is estimated to be using the published extinction coefficient in ethanol (Drexhage 1990). Since , stimulated emission can be neglected so that the rate of absorption must balance the spontaneous emission under steady state cw excitation,
(2)
where N is the number density of unbleached dye molecules. This number can be related in turn to the transmittance of the sample,
, (3)
neglecting the reflectance loss from each face (where the refractive index of plexiglas is n = 1.49). Substituting Equation (3) into (2), and then that into (1) and simplifying leads to the compact result
, (4)
where and are the fluorescence and pump photon fluxes (in photons/s), respectively.

The time-dependent transmittance of the sample as it bleaches during cw laser exposure is given by (Kaminow 1972)
, (5)
where T₀ is the initial transmittance of the sample and the bleaching rate constant is with B the so-called bleaching number (i.e., the average number of excitation-relaxation cycles a dye molecule undergoes before it bleaches). That is, B is the reciprocal of the bleaching quantum efficiency. Equation (5) assumes that the sample fully bleaches ( ) at long times ( ). In contrast, at any fixed pump intensity, Figure 6 indicates that the bleaching saturates at some lower transmittance , so that Equation (5) should be modified to
. (6)

This might arise from thermal effects (Fork 1972), stemming from the fraction of the incident light which heats the sample rather than being reradiated away, or from a site dependence of the photobleaching quantum efficiency (Higuchi 1983), assuming that the orientations or polymeric neighbors of the dye molecules strongly influence B.
Substituting Equation (6) into (4) and expressing and in terms of the corresponding fluorescence photon fluxes and using Equation (4) gives
. (7)

However, the monochromator only collects a portion of the fluorescence. Using Equation (4) once more, one finally deduces
(8)
where T_U and are the transmittance and the fluorescence photon flux, respectively, of the unbleached sample. Since the fluorescence flux only appears as a ratio in Equation (8), the unknown geometrical factor representing the fraction of light collected drops out, and Φ_F can be directly taken to be the number of CCD detector counts at λ_F = 625 nm.

The unbleached transmittance of the sample was measured to be by exposing it to a low-intensity pump beam and creating a ratio of the readings of a power meter placed before and after the sample. The unbleached peak fluorescence flux was deduced to be by extrapolating the signal in Figure 6 back to t = 0.

Equation (8) was then separately fit to each of the three pump intensities in Figure 6, restarting at each knee (when an attenuator was moved). Note that for one range of intensities becomes in the successive range. This leaves two independent parameters for each data sequence, namely the bleached fluorescence flux and the photooxidation rate constant β. The best-fit values are listed in Table 1. From these, we computed after substituting the known values of the various parameters and constants. The bleaching number was found to vary from 1.1 to 8.4 million as the pump intensity was increased from 17 to 170 W/cm² (somewhat beyond the damage threshold of the sample, thus preventing us from continuing to yet higher laser powers). This compares favorably to 1.4 and 2.5 million measured for Rh6G in PMMA assuming isotropic and axial molecular geometries, respectively (Kaminow 1972). In liquid hosts, cresyl violet and rhodamine 6G have similar values for B (Beer 1972) and the present result suggests that the same is true of polymer hosts.

Table 1 . Photobleaching of Cresyl Violet in PMMA *Total optical density of the attenuators in front of the sample †Fit value of ‡Fit value of 1/β ˆBleaching number calculated from β **Total fraction of dye molecules in the pumped volume of the sample that has been oxidized at long exposures, as calculated using Equation (9)

Since , we can also calculate the cumulative fraction of dye molecules that have been bleached,
. (9)
As Table 1 indicates, about 80% of the cresyl violet molecules get photooxidized by the time the sample is experiencing substantial thermal damage. This correlation between dopant oxidation and host damage is probably not a coincidence, since both during and after bleaching the photoproducts may release chemical energy or absorb laser radiation and convert it into thermal energy.

Conclusion

In conclusion, we have measured the quantum efficiency for photooxidation of cresyl violet in solid poly(methyl methacrylate). The key experimental data and theoretical fit are graphed in Figure 6, and the important parameters are summarized in Table 1.

Since the photooxidation irreversibly destroys the optical properties of the dye molecules, knowledge of this quantum efficiency is important in determining the lifetimes of the molecules in a given optical environment, and hence the possible technological applications of a given dye-host system. In particular, the photodegradation is minimal at low intensities, and thus this material can be used under ordinary brightness levels, such as in computer displays or electrical indicator lamps. On the other hand, this dye-doped plastic would have severely limited longevity in laser amplifiers, detectors, optical fibers, and other applications in which high intensities are required.

Other variables not investigated in this study that may affect photostability of the dye molecules include temperature of the sample, nature of the polymer, and degree of alignment of the long-chained polymer molecules. Further research into these issues could open up or rule out other possible applications of dye-doped organic materials.

Acknowledgements

We thank the Research Corporation and the Office of Naval Research for their generous support.