Nitrogen Gas Purging for the Deoxygenation of PolyaromaticHydrocarbon Solutions in Cyclohexane for Routine FluorescenceAnalysis

TODD PAGANO,* ADAM J. BIACCHI, and JONATHAN E. KENNY*Department of Science and Mathematics, Laboratory Science Technology Program, Rochester Institute of Technology, National TechnicalInstitute for the Deaf, 52 Lomb Memorial Drive, Rochester, New York 14623 (T.P.); Cabot Corporation, 157 Concord Road, Billerica,

Massachusetts 01821 (A.J.B.); and Pearson Chemistry Laboratory, Tufts University, Medford, Massachusetts 02155 (J.E.K.)

During routine fluorescence analysis, the presence of dissolved oxygen in

solutions can result in the dynamic quenching of a fluorophore’s emission

through collisional deexcitation of the fluorophore’s excited state. In order

to avoid this type of fluorescence quenching, dissolved oxygen is often

removed from solutions by an inert gas purging procedure. However, the

details and quantification of this purging process are often limited in

fluorescence studies. In this work, standard 10 mm 3 10 mm fluorescence

cuvettes are filled with polyaromatic hydrocarbon (PAH) solutions in

cyclohexane and purged using nitrogen gas, and the experimental purging

parameters (nitrogen flow rate, amount of volatile solvent loss, and rate of

oxygen removal) are measured and analyzed. For experimental conditions

similar to those used in this study, we are able to provide useful guidelines

for the deoxygenation of solutions, specifically the purge times required

for solutions of fluorophores with various fluorescence lifetimes.

Enhancement factors, or F0 /F values (the ratio of fluorescence intensity

of a completely deoxygenated solution to the fluorescence intensity of an

aerated solution), for chrysene, phenanthrene, naphthalene, and pyrene

solutions in cyclohexane were found to be 3.61 6 0.02, 4.17 6 0.02, 7.63 6

0.07, and 21.81 6 0.35, respectively.

Index Headings: Fluorescence; Deoxygenation; Quenching; Purging;

Polyaromatic hydrocarbons; PAH.

INTRODUCTION

Fluorescence spectroscopy is well established as a sensitiveand selective technique for the qualitative and quantitativeanalysis of PAHs, many of which are known carcinogens andEPA priority pollutants. PAH solutions in nonpolar solventsare susceptible to dynamic quenching of fluorescence bydissolved oxygen. This form of quenching can result in adecrease in the PAH’s quantum yield. The mechanism fordynamic quenching involves the deexcitation of a fluorophore,without emission of photons, via collisions with molecularoxygen (O2) during the fluorophore’s excited-state lifetime.The ratio of the fluorescence measured in the absence ofoxygen, F0, to the fluorescence measured in the presence ofoxygen, F, is given by the following Stern–Volmer relation-ship:1

F0

F¼ 1þ s0kO2 ½O2� ð1Þ

where s0 is the lifetime of the fluorophore measured in anoxygen-free solution, kO2

is the oxygen quenching rateconstant, and [O2] is the solution’s oxygen concentration.F0 /F values (also known as intensity ratios, enhancementratios, L0/L, and I0/I ) give an indication of the impact of

molecular oxygen on a given fluorophore’s quantum yield andfluorescence lifetime. Typically fluorophores with longerfluorescence lifetimes will generate larger F0 /F values.Solutions can be deoxygenated in order to maximize thefluorescence intensity and improve quantitative analysis.

At equilibrium, the concentration of dissolved oxygen insolution is given by Henry’s law:2

P ¼ KS ð2Þ

where P is the partial pressure of oxygen above the solution, Kis the Henry’s law constant, and S is the solubility of oxygen inthe solution. Thus, oxygen quenching is more severe innonpolar solvents such as cyclohexane in which it is moresoluble than in aqueous solutions. At 20 8C, under anatmosphere of dry air, the equilibrium concentration of oxygenis about 2.9 3 10�4 M in water3 and 2.3 3 10�3 M in cyclo-hexane.4 By reducing or removing the oxygen above thesolution, one can reduce the oxygen concentration in thesolution, thus optimizing fluorescence intensity. This is doneby replacing the air above the solution with an inert (non-quenching) gas such as nitrogen or argon.

In practice, the inert gas is often bubbled through thesolution to speed the approach to equilibrium. If the solvent isvolatile, the purging process can gradually sweep away solventvapor as the headspace gas is displaced by the flow of inertgas. Cyclohexane, a common solvent for PAHs, has a room-temperature vapor pressure of about 78 torr.5 Loss of solventwould result in an increase in the concentration of the lessvolatile PAH solute. Since measured fluorescence intensity isoften used to quantitatively determine concentration; thisincrease, if significant, should be taken into account.

Berlman’s Handbook of Fluorescence Spectra of AromaticMolecules, a standard reference in the field, reports F0 /F valuesfor many aromatic molecules.6 The reliability of the reportedvalues depends on the complete removal of dissolved oxygenas well as compensation for solvent loss, if needed. Berlman’sdescription of the utilized nitrogen purging technique is terse,and the only quantitative detail provided is the purging timeused—2 minutes.6 This time is generally shorter than that usedby other workers. If his solutions were not completely purgedof oxygen, his F0 /F values would be too small; if, perhaps, heachieved full oxygen removal in two minutes using larger flowrates, the solvent loss problem would be exacerbated and hisreported F0 /F values might be too high. Therefore, we decidedto perform a more quantitative investigation of oxygen purgingfor three PAHs for which Berlman provided F0 /F values:chrysene, phenanthrene, and naphthalene, as well as one

Received 2 October 2007; accepted 4 December 2007.* Authors to whom correspondence should be sent. E-mail: tepnts@rit.edu and jonathan.kenny@tufts.edu.

Volume 62, Number 3, 2008 APPLIED SPECTROSCOPY 3330003-7028/08/6203-0333$2.00/0

� 2008 Society for Applied Spectroscopy

molecule, pyrene, whose spectra Berlman reported, but whoseF0 /F value he did not.

EXPERIMENTAL

Sample Preparation. Four PAH solutions were made incyclohexane. Chrysene (Acros, 98%), phenanthrene (Acros,98%), naphthalene (Aldrich, 99%), and pyrene (Acros, 98%)were dissolved without further purification in cyclohexane(Acros, HPLC grade, 99%) and made to the followingconcentrations: 8.2 3 10�5 M chrysene, 3.5 3 10�5 Mphenanthrene, 2.0 3 10�4 M naphthalene, and 1.7 3 10�5 Mpyrene.

Cuvette Preparation. The PAHs were analyzed in astandard 10 mm 3 10 mm cuvette (Starna Spectrosilt far-UVquartz) with a fused 60 mm long Pyrex graded seal neck. Thecuvette was submerged in 50% nitric acid (Fisher) overnightand subsequently rinsed with deionized water, 90% reagentalcohol (Fisher), and the cyclohexane solvent prior to use.After the cuvette was air-dried, 3 mL of sample was pipettedinto the cuvette, which was immediately capped with a size 8rubber septum (Aldrich).

Aeration. The following procedure, whose adequacy wasdetermined in preliminary experiments, was performed on eachsolution to ensure that it was fully aerated. A 5 mm diameterhose, fitted to a 20-gauge HPLC-type needle, was connected tothe ‘‘house’’ air valve. The flow rate was adjusted to 5 mL/minwith an Alltech Digital Flowcheck flow meter. A 25-gaugeHPLC-type needle was inserted into the cuvette system throughthe septum to act as a pressure release for the purge gas andwas kept in the cuvette headspace above the solution. Theneedle that was attached to the air valve was also pushedthrough the septum and submerged in the solution, and thesample was aerated for 7 minutes. After this aeration process,both needles were quickly removed from the septum.

Aerated Solution Absorbance and Fluorescence Mea-surements. Following aeration, the absorbance and fluores-cence spectra of the solution were measured. The absorbancewas measured with a Cary 300 dual beam ultraviolet–visible(UV–vis) spectrophotometer. A 10 mm 3 10 mm cuvette(Starna, Spectrosilt far-UV quartz), cleaned using the sameprocedure as above and filled with the cyclohexane solvent,was used as the absorbance reference. A baseline correctionwas run with both reference and sample cuvettes filled withcyclohexane. The aerated sample was scanned with a spectralbandwidth of 2.5 nm.

The fluorescence of the aerated sample was measured with aCary Eclipse fluorometer. The fluorescence emission wascollected with excitation and emission slits set to 2.5 nm,matching that of the UV-vis spectrophotometer. The excitationand emission wavelengths used for each of the PAH solutions

are reported in Table I. A background correction wasperformed by subtracting the spectrum of a cyclohexane blankmeasured under the same conditions from those of the PAHsolutions.

Purged Solution Absorbance and Fluorescence Mea-surements. After measurements of the aerated solution weremade, the solution was purged with nitrogen in order to removethe dissolved oxygen. The purging process was performed bythe same technique as the aeration, except the 20-gauge needlewas attached to a compressed cylinder of nitrogen gas(Northeast Airgas, Ultra High Purity) and the flow rate wasagain set to 5 mL/min. The solution was then purged for apredetermined amount of time. Once the purging process wascomplete, the needles were quickly removed and theabsorbance and fluorescence spectra of the purged solutionwere measured as described above. This process was repeatedfor different intervals of purging time. Each sample wasscanned three times, background corrected, and the meanabsorbance and fluorescence values and standard deviationswere determined.

RESULTS AND DISCUSSION

Because of its relatively long fluorescence lifetime, theeffects of dynamic quenching (and purging) on pyrene areespecially dramatic. Figure 1 shows the fluorescence intensityat 384 nm of the pyrene solution when measured at intervals ofpurge time. The increase in fluorescence intensity at shortertimes is caused primarily by the loss of oxygen. As can be seenin Fig. 1, the greatest increase in fluorescence emission occurswithin the first 6 minutes of purging.

The appearance of the measured fluorescence data in Fig. 1and the constant flow rate of the purging gas suggest that theO2 concentration in solution might follow a first-order decaylaw, [O2](t) ¼ [O2](t ¼ 0)exp(�krt). In this model, the Stern–Volmer relationship (Eq. 1) becomes

F0=FðtÞ � 1 ¼ s0kO2 ½O2�ðt ¼ 0Þexpð�kr tÞ ð3Þ

where kr is the rate constant for oxygen removal and t is time.Consequently, we explored the linear fit of ln(F0 /F� 1) versustime, which is also shown in Fig. 1. The optimum value of F0

was found to be slightly larger than the F value at 10 minutesby a trial-and-error maximization of R2 for the linear fit. Theslope of �0.78 6 0.04 corresponds to a half-life for oxygenremoval of 53 seconds, and the R2 value is 0.99, indicating avery good fit to the model.

Figure 2 shows the absorbance of the pyrene solution at 320nm over a 10 minute purge time, showing a gradual increase inabsorbance over time. The plot has a slope of 3.3 3 10�4 6 0.43 10�4 absorbance units/min.

TABLE I. Experimental parameters, absorbance data, and F0 /F values for the purging of the selected PAHs in cyclohexane.

Solutionkex, kem

(nm)Purge

time (min)

% change insolvent volume

(1/min.) [std. dev.]F0 /F at

kem [std. dev.]Lit. valuesfor F0 /F

Chrysene 306, 381 12 �0.065 [0.200] 3.61 [0.02] 3.2a

Phenanthrene 281, 364 12 �0.078 [0.006] 4.17 [0.02] 3.8a

Naphthalene 268, 336 12 �0.072 [0.007] 7.63 [0.07] 6.4a

Pyrene 320, 384 10 �0.068 [0.008] 21.81 [0.35] 18.6b

a Ref. 6.b Ref. 2.

334 Volume 62, Number 3, 2008

The results shown above for pyrene have been reproducedwith the three other PAHs mentioned. Nitrogen gas purging ofchrysene, phenanthrene, and naphthalene solutions in cyclo-hexane increases the average fluorescence yield several fold,with slight increases in absorbance, and therefore, analyteconcentration. The slopes of the plots similar to Fig. 2 forpyrene and the other three PAHs are converted to the morerelevant % change in solution volume per minute and reportedin Table I. Using rate values from the data for all four PAHs, anaverage of 2.0 3 10�5 6 0.2 3 10�5 moles of cyclohexane arelost per minute. This agrees well with the value of 2.1 3 10�5

moles of cyclohexane lost per minute calculated from the purgegas flow rate (5 mL/min) and the assumption that cyclohexanevapor maintains its equilibrium vapor pressure of 78 torr at20 8C, when an approximate solution volume of 3 mL is used.(At this rate of solvent loss, the loss of the heat of vaporizationrepresents a negligible decrease in the sample temperature, lessthan 2 mK.) Therefore, for the purge rates and times used here,the solvent loss is comparable to the typical size of the errorbars for the fluorescence measurements, but for differentpurging conditions and longer purge times the solvent losscould be more significant.

The fluorescence measurements for air-saturated and fullypurged PAH solutions in cyclohexane were utilized to calculateF0 /F values. Our F0 /F values are compared to those from theliterature in Table I. The F0 /F values for chrysene (lifetime ¼

45 ns6) and phenanthrene (lifetime¼58 ns6) have low standarddeviations and are only about 10% larger than Berlman’svalues. Naphthalene (lifetime ¼ 96 ns6), on the other hand,gave an F0 /F value with a substantially larger standarddeviation and is almost 20% larger than Berlman’s value.Finally, pyrene (lifetime ¼ 405 ns7) shows the largest F0 /Fvalue and the largest standard deviation. Our value for pyreneexceeds the equivalent ‘‘enhancement factor’’ reported byRollie et al. (using a novel chemical deoxygenation method,not purging) of 18.6 by more than 15%.2 These observationsare consistent with the expectation that molecules with longerlifetimes would be far more sensitive to residual oxygen.

Our results suggest that Berlman’s solutions may not havebeen fully deoxygenated. If we take our F0 /F values as correct,we can use Eq. 1 to estimate the amount of residual oxygenremaining in his solutions during the measurement of F0. Forchrysene, phenanthrene, and naphthalene, the results are 5%,3%, and 3% of [O2] (t¼0), respectively. It is interesting to notethat a review paper on purging suggests that 95–99% oxygenremoval is considered adequate in many applications.2 Clearly,F0 /F values can be very sensitive to a few percent of residualoxygen, especially for long-lived fluorophores.

We may combine our analysis of Eq. 1 with our observationof near-exponential decreases of O2 concentration in solutionto arrive at guidelines for purging time required for a givenanalyte. If we express the fractional loss in potentialfluorescence intensity as y, the fraction of oxygen remaining,x [ [O2]/[O2]sat, is given by:

x ¼ y

ð1� yÞðF0=F� 1Þ ¼y

ð1� yÞs0kO2 ½O2�sat

ð4Þ

where [O2]sat is the oxygen concentration of the initially air-saturated solution. Since x is equal to exp(�kr t ), we may writefor purging time required

t ¼ � 1

kr

ln y= 1� yð Þ½ � � ln s0kO2O2½ �sat

� �� �ð5Þ

The second term in brackets may be estimated if necessary,as maximum values of kO2 reported in the literature for PAHsare in the 2 to 3 3 1010 M�1s�1 range. In Table II, we providesample values of required purging times for y¼ 0.01 (i.e., 99%of potential fluorescence intensity achieved) and y ¼ 0.001

FIG. 1. (Top) Fluorescence intensity of pyrene at 384 nm collected at intervalsof purge time and (bottom) the corresponding linear fit of ln(F0 /F � 1).

FIG. 2. Absorbance of pyrene at intervals of purge time, with regression lineand error bars. R2¼ 0.95 for the regression.

APPLIED SPECTROSCOPY 335

(99.9% fluorescence intensity) for a range of lifetime valuesfrom 5 to 500 ns, and our estimated oxygen removal rateconstant.

We have provided a quantitative analysis of fluorescenceintensity, solvent loss, and oxygen removal rates for anexperimental setup of interest along with a methodology thatmay be adapted to similar purging situations. While we haveprovided details for the purging of solutions with a specificpurging setup and cyclohexane as the solvent, any changes insolvent or equipment used could give somewhat differentresults. Shorter purge times may be sufficient for analytes withshort fluorescence lifetimes, but those with lifetimes in the

hundreds of nanoseconds require substantially longer purge

times in order to reach their maximum fluorescence intensities.

If fluorescence studies of molecules in cyclohexane require

purge times that exceed those reported in this work, or

excessive purging is used, corrections for solvent loss may be

needed.

ACKNOWLEDGMENTS

The authors would like to thank J. Thomas Brownrigg for valuable

suggestions and encouragement during the preparation of this manuscript.

1. J. R. Lakowicz, Principles of Fluorescence Spectroscopy (Springer

ScienceþBusiness Media LLC, New York, 2006), 3rd ed., p. 280.

2. M. E. Rollie, G. Patonay, and I. M. Warner, Ind. Eng. Chem. Res. 26, 1

(1987).

3. T. L. Brown, H. E. Lemay, and B. E. Bursten, Chemistry the CentralScience (Prentice Hall, Upper Saddle River, NJ, 2000), 8th ed., p. 482.

4. J. D. Wild, T. Sridhar, and O. E. Potter, Chem. Eng. J. 15, 209 (1978).

5. R. C. Weast, Ed., CRC Handbook of Chemistry and Physics (CRC Press,

Boca Raton, FL, 1984), 65th ed.

6. I. B. Berlman, Handbook of Fluorescence Spectra of Aromatic Molecules(Academic Press, New York, 1971), 2nd ed.

7. N. I. Nijegorodov and W. S. Downey, J. Phys. Chem. 98, 5639 (1994).

TABLE II. Purging time required under our experimental conditions fora given lifetime and given fluorescence loss y, assuming kO2 [O2]sat¼ 4.6 3107 s�1 (typical for PAH in air-saturated cyclohexane at 20 8C) and first-order rate constant for oxygen removal kr ¼ 0.78/minute.

Lifetime, ns y ¼ 0.01 y ¼ 0.001

5 4.5 min. 7.5 min.50 7.5 min. 10.5 min.

500 10.4 min. 13.4 min.

336 Volume 62, Number 3, 2008