Challenges of Spectrofluorometry Part 3: Sample-Specific Concerns

The first installment of this three-part series provided a procedure for quickly recording acceptable fluorescence spectra with classic commercial spectrofluorometers for most samples in common 1-cm pathlength cuvettes. Part 2 described how to correct for instrument response so data can be compared properly between laboratories. This part describes sample-specific concerns that could require modifications of our basic procedure to prevent our data from being inconsistent.

In parts 1 and 2 of this article (1,2), we described a simple procedure to select conditions for spectrofluorometry in 1-cm cuvette samples, and how to determine and correct data for instrument artifacts. These procedures are adequate for comparing spectrofluorometry between laboratories, as well as for producing data that is consistent within the same laboratory over time or with different instruments, in most cases. The remaining complications of conventional spectrofluorometry can be ascribed to sample-specific concerns that cause the sample fluorescence itself to be inconsistent. Part 3 of this series discusses reasons that samples might give inconsistent results even when you have chosen what appear to be appropriate instrument settings, and considered every potential artifact and defect of your measurement system. We primarily focus on measurements with continuous (not pulsed) light sources.

Molecules that emit light can be sensitive to their environments, adversely affecting interlaboratory comparisons if environmental factors are not understood and controlled. Conversely, many chemical or physical phenomena that cause irreproducibility in spectrofluorometry are also potential targets for fluorescence sensing applications, so it is advisable for an analyst to be aware of the mechanisms of these effects. We consider the most obvious elements of sample condition first:

• Solvent
• Sample purity
• Temperature
• Pressure
• Sample concentration

Solvents

Solvents greatly exceed solutes in concentration, and the effect of a solvent on spectrofluorometry can therefore be substantial. Solvents (even “spectroscopic grade” solvents) may have fluorescent impurities (3); these could have come in the fresh solvent, or have been introduced through poor laboratory technique, detergents, or contaminated glassware (4). Some solvents may also have non-fluorescent stabilizers intentionally added (5,6). For example, diethyl ether, an aprotic solvent, is sometimes stabilized with ethanol, a protic solvent. Also, fresh solvents, left exposed to air, will dissolve atmospheric gases including moisture. Traces of water in a non-aqueous solvent can have significant effects on fluorescence compared to a scrupulously dry non-aqueous solvent (for example, chlorophyll fluoresces more strongly in protic solvents). Traces of acetone sometimes found in solvents are also known to affect fluorescence (7).

Even clean solvents interact with solute molecules through their dipole moments and polarizabilities in ways that cause shifts in spectra, changes in apparent intensity, changes in the quantum yield of luminescence, and changes in bandwidths (8). These effects are fairly common to all emit- ting molecules. In some cases, however, solvents can have additional effects on the equilibrium distribution of molecular conformations, the protonation state of molecules, or the ordering of excited states (9). Behera, Park, and Gierschner have recently provided a systematic organization of the various types of dual-emitting samples (10), and we can anticipate specific solvent effects for nearly all these cases. It should be clear to the reader that the choice of solvent is one of those measurement conditions that must be recorded and reported to have any hope of repeatability.

Sample Purity

We refer to the molecule on which we are performing spectrofluorometry as the “target compound,” to distinguish it from other fluorescent species in the sample. The purity of target compounds used for spectrofluorometry is a potential source for concern when comparing results. The lower the quantum efficiency of the target compound, the more concerning any potential strongly fluorescent contaminant will be (11).

If the target compound is purchased, the purity is usually specified as a minimum percentage by weight, giving an upper limit to the amount of all impurities. If the target compound is synthesized locally, quantitative nuclear magnetic resonance (NMR) appears to hold promise for detecting all impurities (including water), as long as they are amenable to NMR study (12). Fluorescent impurities (such as chemical relatives of the target compound that have not been adequately separated) have historically been detected by some form of chromatography (for example, from simple thin layer chromatography to high performance liquid chromatography [HPLC]) using fluorescence detection (including visually inspecting for fluorescent bands).

Target compounds stored in solution in clear bottles or for long periods of time are likely to be thermally degraded over time, leading to more contaminants than were originally present. For instance, many organic dyes degrade when exposed to oxygen in solution, so even a dark bottle is not sufficient to avoid making fresh solutions for calibrations. If you want to keep your target compound as nearly pure as possible, keep it in the solid form, in its original container, in the dark, and avoid heat. If you refrigerate it, wait for it to warm before opening to avoid condensation.

Temperature

Increasing temperature ordinarily decreases the yield of luminescence in the absence of other effects by increasing non-radiative decay processes for isolated molecules in solution. This change is typically in the range of 1% per ^oC, although some systems can show up to 5% per degree (4) There can be exceptions with the opposite response to temperature, at least over limited temperature ranges, because of changes in thermal equilibrium distributions of conformations, association equilibria or thermally accessible excited states, and other reasons.

Pressure

Pressure effects are usually much less pronounced than temperature effects under near-ambient conditions. However, one significant effect of pressure is to change the equilibrium concentration of dissolved atmospheric gases (especially oxygen) in the solution. This effect has been used as the basis for fluorescence-based (actually phosphorescence-based) oxygen and pressure sensors (13).

Concentration

Concentration of the target compound affects spectrofluorometry in three important ways. The first of these is through the inner-filter effect (18). The inner-filter effect can itself be divided into primary and secondary types, the former caused by absorption of the excitation beam by the sample, and the latter caused by re-absorption of the fluorescence before it emerges from the sample. If the procedure described in Part I is followed, these effects should be real, but small and correctable (8).

The second sample concentration effect is through the possible formation of dimers and aggregates (14) in the ground state, excited state, or both. Carboxylic acids (15) and laser dyes (16) are well known to exhibit dimerization in solution in their ground states, whereas pyrene is the classic example of a molecule that can form excited dimers (also known as excimers) (17). Dimerization can result in a change in the absorption spectrum, a change in the fluorescence spectrum, concentration quenching, or a change in the fluorescence quantum yield of molecules in solution.

The third effect of sample concentration is through excited state energy transfer between different identical molecules that can depolarize fluorescence (14), a factor that is only important in situations where the fluorescence is not already depolarized for other reasons, or where it has not been corrected for in instrumentation (see Part 2).

We called these first five sources of irreproducibility in spectrofluorometry obvious because solvents, sample purity, temperature, pressure, and concentration are the state variables that many of us learn about in introductory courses in chemistry. However, there are some less obvious sources of irreproducibility that we’d like to mention. These include polarization of fluorescence, long-lived phosphorescence, quenching, photochemical decay, and dissolved ion concentration.

Polarization

We mentioned in Part 2 that samples may exhibit polarized fluorescence, and, if they do, this needs to be accommodated in the instrument with polarizers placed appropriately. We did not, at that point, explain why molecules might exhibit polarized fluorescence.

Because of the symmetry of the dipole moment operator that enables most light absorption and emission, both excitation and emission processes of target compounds are guaranteed to be inherently polarized along one or two directions in the molecule. The only exceptions are optical transitions involving molecular species of very high symmetry—molecules of exactly tetrahedral, octahedral, or even higher symmetry. In a 90-degree instrument configuration as discussed above, even unpolarized excitation can result in partially polarized emission (8,18). Macromolecules like proteins, or small molecules bound to macromolecules and macromolecular assemblies like micelles or cyclodextrins, can rotate in solution so slowly that polarized emission is preserved. In this case, the presence of certain surfactants or macromolecules in solution can change the polarization state of emission even for small molecules with normal fluorescence lifetimes. There are also cases where the fluorescence lifetime of a small molecule is so short that polarization of emission is still detectable, even without viscous solvents or association with a macromolecule (19). Any residual polarization of the emission from a target compound can change the relative intensity of observed fluorescence, and it can vary with excitation wavelength because different optical transitions can be excited across the excitation wavelength range. Part 2 provides the means of correcting for this potential problem.

Phosphorescence

Phosphorescence is distinguished experimentally from fluorescence by the lifetime of its exponential decay following a pulse of light, typically being orders of magnitude slower than fluorescence. In rare cases, phosphorescence can have lifetimes on the order of seconds. This gives rise to two important concerns.

First, when working with long-lived phosphorescent molecules, emission intensity does not change promptly when the excitation is turned on or shuttered, but requires multiple half-lives to reach a steady-state intensity. An excitation spectrum can be “blurred” if it is scanned faster than the sample phosphorescence equilibrates. Likewise, repeated excitation scans may include moments where the excitation wavelength changes at the maximum slew rate of the excitation monochromator to start the next run, and the start of a subsequent spectral measurement could be affected by the sample’s “memory” of recent excitation conditions. For the same reason, emission spectra can show artifacts at the start of a new measurement because they begin too quickly after the excitation is unblocked or the wavelength of excitation is reset to the measurement condition. Fortunately, only a fraction of luminescent molecules possess decay rates long enough (>1/5th of the dwell or integration time) for long-lived phosphorescence to cause a problem like this.

However, nearly all phosphorescent molecules are subject to quenching of their emission. This is because even short-lived phosphorescence is orders of magnitude longer than most fluorescence, giving ample time for molecules in solution to find and interact with other molecules that can quench their emission. Just as important, there is a very common molecule that is an excellent quenching agent for phosphorescence: molecular oxygen.

Quenching

Most of the emitting excited states of phosphorescent molecules have unpaired electron spins. Molecular oxygen has a number of low-lying excited states that could accept energy from an excited molecule, and also has unpaired electron spins. Because of the relatively long-range magnetic coupling between the unpaired spins of molecular oxygen and those of other phosphorescent molecules, a small amount of dissolved oxygen can be very effective at quenching the energy from excited phosphorescent molecules before they can emit light (11). Because the analyst has probably not intentionally introduced oxygen to their spectrofluorometer sample, the oxygen concentration and thus the phosphorescence intensity may vary significantly between samples—or even for exactly the same sample from the moment it is freshly made to when it equilibrates with the atmosphere.

Quenching by molecular oxygen is a very important type of quenching for phosphorescent molecules. However, all excited molecules can show quenching phenomena under the right conditions. The presence of contaminants in the solvent, or in an impure target compound, can quench the fluorescence of the target compound. The presence of dimers of the target compound can also quench fluorescence from isolated molecules.

Photochemical Decay

Photochemistry is another of the possible non-radiative decay pathways for excited molecules, and it can result in conversion of our target compound to other chemical structures or con- formations. In rare cases, photochemical reactions may be reversible (such as, for example, spiropyrans [20]), but usually not. The products of photochemical decay may be colorless or strongly colored; they may be fluorescent themselves, or not fluorescent. They may in some cases act as quenching agents when they build up.

Photochemistry is nearly always more important in fluid solutions than in solids; this is another reason target compounds are best stored in a solid form if not immediately being used. Multiple photochemical products of a target compound may be possible, and the yield of each may vary differently with wavelength (21). More photochemical pathways will probably be opened with high energy (such as UV) than with low energy excitation, which will not surprise the sunburned among us.

However, one absolute requirement for photochemistry is that photons be involved. Photochemistry cannot happen in the absence of light absorption, so the traditional cure for photochemistry that might occur during a study is to keep the target compound in the dark, or only illuminate it with long wavelengths of light, both before spectrofluorometry and whenever possible during a study.

Dissolved Ions

Some molecules are particularly susceptible to changes in their absorption and fluorescence spectra due to changes in surfactant, H⁺ and other ion concentrations, or other trace molecules. In addition to possible polarization effects, surfactant micelles can form local environments and local concentrations of a target compound, and give rise to dimerization or other effects (22). Target compounds with acidic or basic groups may show substantial changes in spectroscopy and quantum yield, depending on their protonation state. Complexation with metal ions can also produce changes in fluorescence.

Procedure: Prepare → Evaluate → Measure

When publishing or transmitting your data to others, it is important to specify as much of what you have done as possible to enable others to reproduce your work.

Preparing for a Study

Cuvette

Begin with a clean, fluorescence-free cuvette with low absorbance. When buying cuvettes, manufacturers may describe the materials from which their cuvettes are made in detail (by specific composition), and this is useful information to know. Quartz or fused silica are often the materials of choice for fluorescence-free cuvettes. However, after settling on high quality cuvettes, you will need to confirm the lack of measurable fluorescence in your instrument, and will need to keep them clean to maintain that performance. Manufacturers of cuvettes and spectrofluorometers can provide one or more protocols for cleaning cuvettes to avoid contaminants. Protocols may include, among other steps, soaking in nitric acid, rinsing with deionized water, or cleaning with a detergent such as Alconox. HNO₃ protocols are particularly effective at removing organic contaminants with little residue. Chromic acid solutions are often used for cleaning laboratory glassware, but they absorb UV, and so have never been recommended for cleaning optical materials. Phosphoric acid, HF, and strong bases are not suited to cleaning optical materials because they etch glass and silica. Also, be aware that quartz or fused silica cuvettes are not usually intended for high temperature or ultrasonic processing.

Solvent

Ensure you have an appropriate, fluorescence-free solvent of high purity. An appropriate solvent would be one whose evaporation rate is low or can be controlled, that dissolves the sample well, and whose likely effect on the solute is understood. Regardless of the purity of the solvent, or whether it says “spectroscopic” or “HPLC” grade on its label, a blank cuvette of the solvent should be run to confirm low absorbance across the spectral window of interest and low fluorescence. The blank solvent spectra can be used as a background for the eventual measurement of the solute. If the purity of the solvent is of concern, it has been observed that filtering a solvent through activated carbon followed by distillation is effective for removing many organic contaminants. Molecular sieves can reduce moisture levels to sub-10 ppm levels (23); other more active techniques can reduce moisture even lower. Electrochemistry texts provide excellent procedures for removing water from solvents by vacuum line techniques if this is an issue (24). Note that benzene and related compounds may sometimes found as contaminants in several solvents otherwise suitable for UV studies (25), so it is a good idea to check for any deep-UV-excited fluorescence in a solvent.

The solvent should have a well-defined concentration of known ions, and a known pH if using aqueous solutions with solutes that have acid-base properties. Most fresh solvents have been assayed for their major metals and ionic species. Samples taken from nature may require clean-up steps to remove unwanted species, or steps to ensure the pH is maintained at a desired level without introducing new chemical species to the solution.

Sample Purity

The target compound you intend to study should ideally be of a known, high purity. One common approach to checking purity is to see if the fluorescence excitation spectrum of sample is similar to the absorption spectrum of the target compound (11), and that the fluorescence spectrum itself is independent of the excitation wavelength. These procedures are sometimes misleading and are not very sensitive to low levels of contaminants, so if you’re not sure, high-performance liquid chromatography (HPLC) with fluorescence detection is a very effective way to test for fluorescent impurities.

Evaluate Your Sample

Long-lived Phosphorescence

Long-lived phosphorescence of the sample under these measurement conditions can either be an inherent property of the sample, or evidence of an impurity (such as a free ligand in the solution of a coordination complex). We assume here that the time between measurements in the measurement protocol is longer than ~200 ms, because of the combination of dwell time at each point and the time required to change wavelengths. For the purpose of detecting luminescence with a lifetime long enough to require slowing data acquisition further, detection can often be achieved with the naked eye, as long as the emission is visible. One can hold the emission filter from the spectrometer in front of the unaided eye, and see the emission coming from the sample, while using an opaque card to the block the excitation beam. If there is a discernible lag between blocking the beam and the end of emission to the naked eye, then there is likely a long-lived phosphorescent component to the emission. If you worry your eyes are fooling you, many of us have an optical detector with ~30 ms time resolution in our pocket (our cell phone cameras), and, if the emission is bright enough, a cell phone video can be used to confirm the observation. If still in doubt, the sample can be subjected to a fluorescence lifetime measurement, which usually requires another instrument. If a long-lived emission is present, an increased delay between measurements may be needed to allow the emission to equilibrate. In the case that there is a phosphorescent contaminant in the sample that cannot be removed, and when the target compound itself is not phosphorescent, it may be possible to eliminate the contaminant’s phosphorescence, and retain the desired target compound fluorescence by saturating the sample with oxygen and sealing it. This will not, however, be helpful if the target compound has.

Sensitivity to Oxygen

To determine whether the emission of the target compound is sensitive to oxygen at the selected concentration, we can follow a published protocol (26). We begin by setting up the fluorimeter for a measurement with your planned spectral bandwidth (SBW) and other instrument parameters, only without the sample in place. Set the excitation and emission wavelengths to their optimum positions based on preliminary studies, and prepare to record a time-scan using constant excitation and emission wavelengths. While the instrument is sitting ready, place a septum over your cuvette containing about 3 mL of your experimental solution (a typical 1-cm cuvette sample). Place a syringe needle through the septum into the air space above the sample as a vent, and another that passes to the bottom of the sample. This latter syringe should then be connected to a low flow rate of dry nitrogen gas so that bubbles of N₂ are observable in the sample. Continue purging with N₂ for 7 min to degas oxygen from the sample, and then remove both needles. Immediately transfer the sample to the fluorimeter and measure the fluorescence intensity for a period of time (about a minute) to ensure it is stable, or at least changing very slowly and consistently. Leaving the cuvette in the instrument, place the syringe needles back into the cuvette and bubble fresh air into the solution using a bulb until enough has been added to displace the gas above the sample, taking care not to suck the sample back into the bulb. Remove the needles and observe the intensity of the sample relative to the purged sample. Small changes can be ascribed to changes in sample position, but large changes suggest significant oxygen sensitivity. The changes in intensity for four sensitive compounds reported by Pagano (26) ranged from a factor of 3.6 to a factor of 21.8. If your sample shows oxygen sensitivity, it should be prepared for study by purging with N₂ and leaving it sealed during analysis. If you need to purge the sample, be aware that a correction to the concentration because of solvent evaporation will be required, if your solvent is at all volatile.

Photochemical Decay

If your target compound shows no immediate loss of signal from introduction of oxygen in the preceding test, continue to record data for a period of minutes—at least as long as you expect your measurement to be—to observe the decay of fluorescence because of any photochemistry of the target compound. It is also a good idea to change the excitation wavelength to the shortest you plan to use, and repeat this experiment to see how quickly it decays with short-wavelength excitation. If your target compound does exhibit oxygen sensitivity, then this photochemical study should be repeated with a purged sample. If photochemical decay is too rapid, the measurement protocol may need to be modified to limit the exposure of the sample to light during analysis. This can be done by blocking the excitation beam when not making measurements, and also by using shorter dwell times and taking larger wavelength steps. If available and practical, a flow cell can be used to decrease the amount of light exposure of the sample by constantly replacing it.

Concentration Dependence

To test for a concentration dependence in your sample, dilute the target compound by a factor of 2, and compare the fluorescence excitation spectra of the original and diluted samples when scaled together. Their scaled profiles should be the same and not show the appearance or disappearance of peaks, while their raw intensities should differ by the dilution factor. If you are unsure whether the profiles are sufficiently alike, dilute again to see if there is a systematic change occurring in the spectra with concentration. If there is no systematic variation, the original concentration is fine to work with. If there is a systematic variation, then use a lower concentration that shows no further systematic variation.

If the sample concentration that is selected by the procedures of Part I and the preceding paragraph give a sample whose intensity exceeds the linear range of the instrument for the intended SBWs, the sample concentration can be reduced as necessary to remain in the linear range.

Temperature Dependence

If your instrument has a temperature-regulated sample holder, then you can skip this step as long as the holder is working properly, and sufficient time is given for the sample to equilibrate. If your instrument has an unregulated sample, then prepare a cuvette with a septum covering it to protect it, and cool it in an ice bath. Set up the spectrofluorometer so it is ready for an immediate rapid measurement (short dwell times, larger steps in wavelength). Ensure the sample compartment of the spectrofluorometer is purged with dry gas to avoid condensation, and return the sample to the holder. Record a rapid fluorescence excitation spectrum and a rapid fluorescence emission spectrum, returning to the ice bath between measurements if needed. Then, block the excitation beam and give the sample a few minutes to warm to room temperature. When it has equilibrated, unblock the excitation beam and repeat the excitation and emission scans. These measurements will give you a semi-quantitative sense of how sensitive the sample is to temperature changes. When the actual sample measurements take place, ensure the cuvette has equilibrated and record the temperature of the sample or sample compartment during the measurement.

References

(1) C. English, Z. Kitzhaber, J. Williams, A. Humphries, and M.L. Myrick, Spectroscopy 36(10), 44–46 (2021).

(2) C. English, Z. Kitzhaber, J. Williams, and M.L. Myrick, Spectroscopy 36(11), 41–45 (2021).

(3) C.E. White, Anal. Chem. 40, 116R–135R (1968).

(4) R.T. Williams and J.W. Bridges, J. Clin. Path. 17, 371–394 (1964). https://doi.org/10.1136/jcp.17.4.371

(5) R.H. Anderson, J.K. Anderson and P.A. Bower, Integr. Environ. Assess. Manage. 8, 731–737 (2012). https://doi.org/10.1002/ieam.1306

(6) W.L. Archer, Industrial Engineering Chemistry and Product Research and Development 18, 131–135 (1979). https://doi.org/10.1021/i360070a011

(7) B.L. van Duuren, Chem. Rev. 63, 325–354 (1963). https://doi.org/10.1021/cr60224a001

(8) J.R. Lakowicz, Principles of Fluorescence Spectroscopy (Springer, Boston, Massachusetts, 3rd ed., 2006), https://doi.org/10.1007/978-0-387-46312-4

(9) C. Reichardt, Chem. Rev. 94, 2319–2358 (1994). https://doi.org/10.1021/cr00032a005

(10) S.K. Behera, S.Y. Park and J. Gierschner, Angewandte Chemie International Edition 60, 2–17 (2021). https://doi.org/10.1002/anie.202009789

(11) B. Ciesielska, A. Lukaszewicz, L. Celewicz, A. Maciejewski, and J. Kubicki, Appl. Spec. 61, 102–112 (2007). https://doi.org/10.1366/000370207779701389

(12) B.U. Jaki, A. Bzhelyansky, and G.F. Pauli, Magnetic Resonance in Chemistry 59, 7–15 (2021). https://doi.org/10.1002/mrc.5099

(13) J.R. Bacon and J.N. Demas, Anal. Chem. 59, 2780–2785 (1987). https://doi.org/10.1021/ac00150a012

(14) D.R. Lutz, K.A. Nelson, C.R. Gochanour, and M.D. Fayer, Chem. Phys. 58, 325–334 (1981). https://doi.org/10.1016/0301-0104(81)80068-7

(15) G. Allen and E.F. Caldin, Q. Rev., Chem. Soc. 7, 255–278 (1953). https://doi.org/10.1039/qr9530700255

(16) O. Valdes-Aguilera and D.C. Neckers, Acc. Chem. Res. 22, 171–177 (1989). https://doi.org/10.1021/ar00161a002

(17) J.B. Birks, Rep. Prog. Phys. 38, 903–974 (1975). https://doi.org/10.1088/0034-4885/38/8/001

(18) A.C. Albrecht, J. Mol. Spectrosc. 6, 84–108 (1961). https://doi.org/10.1016/0022-2852(61)90234-X

(19) J. Paoletti and J-B. Le Pecq, Anal. Biochem. 31, 33–41 (1969). https://doi.org/10.1016/0003-2697(69)90238-3

(20) R. Klajn, Chem. Soc. Rev. 43, 148–184 (2014). https://doi.org/10.1039/c3cs60181a

(21) J.P. Menzel, B.B. Noble, J.P. Blinco, and C. Barner-Kowollik, Nat. Commun. 12, 1691 (2021). https://doi.org/10.1038/s41467-021-21797-x

(22) R. Vonwandruszka, Crit. Rev. Anal. Chem. 23, 187–215 (1992). https://doi.org/10.1080/10408349208050854

(23) D.B.G. Williams and M. Lawton, J. Org. Chem. 75, 8351–8354 (2010). https://doi.org/10.1021/jo101589h

(24) Laboratory Techniques in Electroanalytical Chemistry, P. Kissinger and W.R. Heineman, Eds. (CRC Press, Boca Raton, Florida, 2018), https://doi.org/10.1201/9781315274263

(25) S. Okada and Y. Iida, J. Mass Spectrom. Soc., Jpn. 28, 263–268 (1980). https://doi.org/10.5702/massspec.28.263

(26) T. Pagano, A.J. Biacchi, and J.E. Kenny, Appl. Spec. 62, 333–336 (2008). https://doi.org/10.1366/000370208783759696

Caitlyn English, Zechariah Kitzhaber, Joshua Williams, and M.L. Myrick are with the University of South Carolina, in Columbia, South Carolina. Direct correspondence to: MYRICK@mailbox.sc.edu ●