HHS Public Access

Author manuscript
Appl Opt. Author manuscript; available in PMC 2019 November 21.
Author Manuscript

Published in final edited form as:

Appl Opt. 2017 April 10; 56(11): 3155–3161. doi:10.1364/AO.56.003155.

Interference filter tilting to detect a polycyclic aromatic

hydrocarbon at the second harmonic of wavelength modulation
frequency
Dmitri Lanevski1, Koit Mauring1, Eric Tkaczyk1,2
1Institute of Physics, University of Tartu, Wilhelm Ostwaldi tn 1, 51014 Tartu, Estonia
Author Manuscript

2Departmentof Medicine, Vanderbilt University Medical Center, One Hundred Oaks Suite 26300,
719 Thompson Lane, Nashville TN 37204

Abstract
We present a practical implementation of the wavelength-modulation spectroscopy technique with
second-harmonic detection (WMS-2f) for selective detection and concentration measurements of
polycyclic aromatic hydrocarbons (PAHs) - common fossil fuel pollutants. The method is based on
excitation light wavelength modulation around the maximum of a narrow peak in the absorption
spectrum of a PAH and lock-in detection of fluorescence at the second harmonic of the excitation
wavelength modulation frequency. A violet LED is used as the excitation source and wavelength
modulation is performed via narrow-band interference filter tilting. Feasibility is demonstrated
with a carcinogenic PAH, benzo(k)fluoranthene. Model computations are presented to estimate the
Author Manuscript

influence of background signal on the sensitivity of the method.

OCIS codes:
(280.4788) Optical sensing and sensors; (300.6380) Spectroscopy, modulation; (300.6280)
Spectroscopy, fluorescence and luminescence; (230. 7408) Wavelength filtering devices

1. Introduction
For several decades, it has been generally accepted that polycyclic aromatic hydrocarbons
(PAHs) have mutagenic and carcinogenic effects [1]. They have received considerable
attention and presently different restrictions and contamination standards imposed on PAH
Author Manuscript

contamination in food [2], consumer products [3], water [4] and air [5] have been
established. However, due to forest fires, combustion of fossil fuels and wide usage of other
fossil products that are the main sources of PAHs [6], PAH pollution is constantly increasing
and negatively affects our environment and health [7–9]. Therefore, there is a great need for
accurate PAH detection and concentration measurements.

Traditionally, chromatography and mass-spectrometry [10, 11] techniques are used for PAH
contamination analysis and concentration determination, which provide results of high

Corresponding author: [email protected].

 Lanevski et al. Page 2

precision and sensitivity but require complex instruments, an expensive laboratory

Author Manuscript

environment, and long processing times. Therefore, these techniques are not suited for real-
time measurements that are necessary for effective PAH contamination monitoring and fast
pollution-prevention actions. In the present work, we propose an in-solution (methanol,
water) PAH detection method based on wavelength modulation spectroscopy (WMS)
techniques.

The method utilizes narrow absorption peaks of PAHs and their natural fluorescence. We
excite the sample around certain PAH absorption peaks and measure fluorescence with a
photodiode. The excitation light wavelength is modulated around a PAH absorption peak at
known frequency via tilting of an interference filter. The photodiode signal is fed to a lock-in
amplifier in order to extract fluorescence signal at the second harmonic of excitation light
wavelength modulation frequency, which is high only in of the presence of PAH. This
enables detection of PAHs and estimation of their concentrations.
Author Manuscript

Our approach has two major advantages. Firstly, high sensitivity even to weak signals results
from lock-in detection that removes most background and noise, allowing detection of low
concentrations of studied substances. Secondly, implementation is simple, relying on filter
tilting for modulation. As true for WMS methods in general, one must simply pass this
wavelength-modulated light through the sample and measure the signal at the frequency of
interest. Coupled with fast electronics, this method is ideally suited for real-time PAH
detection, as demonstrated herein for benzo(k)fluoranthene (BKF), a typical PAH.

2. Theory
A. Dependence of fluorescence on absorption spectrum and concentration
Author Manuscript

WMS for detection of PAHs in solution is made possible by their intrinsic fluorescence and
relatively narrow absorption peaks. The fluorescence intensity Ifl depends linearly on the
amount of absorbed light energy. This can be calculated in terms of the PAH molecules’
absorbance A by:

I f l = φ0I 0 1 − 10− A (1)

Here I0 is incident light intensity and φ0 is quantum efficiency.

In practice PAHs are relatively insoluble in polar solutions, resulting in very small
absorbance. Thus, only the first Taylor component of (1) contributes appreciably and the
fluorescent signal is directly proportional to absorption:
Author Manuscript

I f l = φ0I 0 A . (2)

Detecting the fluorescence signal is greatly advantageous, as it provides intrinsically higher

signal-to-noise ratio (SNR) than absorption measurements. In particular, fluorescence+
background depends only on the part of background absorption that also results in
fluorescence, which is typically minimal.

Appl Opt. Author manuscript; available in PMC 2019 November 21.

 Lanevski et al. Page 3

Quantitative information about PAH concentration in homogeneous solutions can be

Author Manuscript

obtained with the Beer-Lambert law:

A = ε(λ)cl (3)

where ε(λ) is the extinction coefficient at wavelength λ, l is the excitation light optical path
length within the sample and c is the concentration of PAH. Substituting this relation into (2)
we obtain equation

I f l = φ0I 0ε(λ)cl (4)

which reveals the fluorescent signal dependence on concentration of absorbing molecules.

We can use this direct proportionality to conduct a quantitative analysis on PAH
Author Manuscript

contamination.

B. WMS signal near and away from absorption peaks

Fig. 1 shows an absorption peak of BKF in methanol as a function of wavelength, which can
be locally approximated by a Gaussian:

2
λ − λ0
−
2σ 2
(5)
A(λ) = A0e

A0 is absorbance at absorption spectrum maximum λ0. Constant σ is related to the peak full-
width-half-maximum (FWHM) by FWHM = 2 2ln 2 σ.
Author Manuscript

On the upslope of the PAH absorption peak (Fig 1, case 1), curvature is negligible, and the
absorbance can be approximated linearly with slope k:

A(λ) = A0 + kλ (6)

This describes the situation where nonlinearities, such as absorption maxima, are
insignificant within the wavelength modulation range. Though this assumption may not
always hold, it is true in the case of broad absorption backgrounds in natural waters [12].

The excitation light wavelength modulation can be described by Fourier series:

N i2nπ f 0t
∑
Author Manuscript

λ(t) = Δλne (7)

n= −N

Here t is time, Δλn is the amplitude of n-th Fourier component, N is number of modulation
components and f0 is the fundamental frequency of modulation. We employ the simplest
case of N = 1 - a pure sine wave modulated around the peak center λ0 with modulation
amplitude Δλ at angular frequency ω0 = 2π f 0:

Appl Opt. Author manuscript; available in PMC 2019 November 21.

 Lanevski et al. Page 4

λ(t) = λ0 + Δλsin ω0t

Author Manuscript

(8)

Accordingly, the time-dependent absorbance of BKF in the presence of symmetric

wavelength modulation around the absorption peak is:

2
Δλsin ω0t
−
2σ 2
(9)
A(t) = A0e

with corresponding fluorescence per (4):

2
Δλsin ω0t
−
Author Manuscript

2σ 2
(10)
I f l(t) = φ0I 0 A0e

By contrast, away from a peak the fluoresce intensity is:

I f l(t) = φ0I 0 A0 + kλ0 + kΔλsin(ωt) (11)

Figure (2) shows the fluorescence power spectrum for wavelength modulation frequency f0
= 1 Hz calculated by a numerical Fast Fourier Transform (FFT) of equations (10) and (11).
Near the peak (Fig. 2.A), +only even Fourier components appear in the output, with the most
significant component at the second harmonic of the wavelength modulation frequency (2
Author Manuscript

Hz). By contrast, the power spectrum of fluorescence in the linear regime away from a peak
(Fig. 2.B) contains only one component, which is the first harmonic of the fundamental
excitation light wavelength modulation frequency (f0= 1 Hz).

Accordingly, lock-in measurement of fluorescence intensity at the second harmonic of the

fundamental will yield a significant signal only when an absorption maximum is present
within the modulated wavelength range. If other substances in the sample fluoresce into the
detector, the second harmonic signal 2ω, will be negligible as long as their absorption
features are broad enough to change linearly in the wavelength modulation range Δλ.

C. Fourier and Taylor generalization of WMS

As a generalization of the WMS scheme described above, consider the fluorescence
intensity signal Ifl(t) that is the overlap integral over λ in each time instant of absorbance
Author Manuscript

A(λ) and the excitation spectrum E(λ, t).

I f l(t) = ∫ dλA(λ)E(λ, t) (12)

We expand the absorption spectrum around the peak λ0 as a Taylor series with constants An
(calculated in the usual way from derivatives):

Appl Opt. Author manuscript; available in PMC 2019 November 21.

 Lanevski et al. Page 5

A(λ) = ∑ An λ − λ0 n
Author Manuscript

(13)
n

Now expand the excitation light spectrum with periodicity ω0 as a Fourier series with
coefficient functions Em(λ) (calculated in the usual way from one period):

imω0t
E(λ, t) = ∑ Em(λ)e (14)
m

Accordingly, let us decompose Ifl(t) into the same Fourier components in terms of harmonics
of ω0:

imωot
Author Manuscript

I f l(t) = ∑ I m(λ)e (15)

m

Comparing this to (12), (13), and (14), it is evident that Im is simply the summed
mathematical correlation (at λ0) of functions λn and Em:

Im = ∑ An∫ dλ λ − λ0 nEm(λ) (16)

n

Therefore, in the case when the excitation spectrum is symmetric around λ0, it becomes
clear that for each even harmonic m (in particular m = 2, the second harmonic), there is no
Author Manuscript

contribution from any odd components n of the absorption spectrum (in particular for n = 1
in the case of linear component of absorption spectrum) due to cancellation of the integral’s
negative and positive halves. In the case of broken symmetry (as for example if E(λ, t) is
detuned from λ0), this may not in general hold.

C. Interference filter incidence angle adjustment for wavelength modulation

Traditional options for wavelength modulation methods typically complicate experimental
setups with additional optical components. For example, modulation can be achieved via
oscillation of a slit, mirror, diffraction gratings or prism of a monochromator [13]. In the
present work, we add the approach of interference filter tilting, which has the advantage of
relying on an optical component necessary in most excitation-detection schemes.

Interference filter transmission central wavelength is known to depend nonlinearly on angle

Author Manuscript

of incidence θ according to:

sin2θ
λ θ = λc 1 − (19)
K2

Appl Opt. Author manuscript; available in PMC 2019 November 21.

 Lanevski et al. Page 6

where λc is the central transmission wavelength at normal incidence and K is the effective
Author Manuscript

refraction index of the interference filter. However, our detection scheme requires
wavelength modulation with only the first Fourier component (i.e. a pure sinusoid) to avoid
second harmonic signals of background fluorescence.

In order to create the desired wavelength modulation waveform expressed in equation (8),
we must accordingly tilt the filter with angle time dependence:

2
Δλsin ω0t + λ0
θ(t) = asin K 1 − (20)
λc

This was used to calculate the control voltage waveform for the rotating mechanism, which
has a linear voltage-angle dependence. In order to obtain a clean second harmonic signal, the
Author Manuscript

modulation amplitude Δλ in equation (8) is selected as broad as permitted by the filter

transmission bandwidth and range of PAH absorption band symmetry about λ0.

2. Experimental designs
A. Optical setup
To experimentally test the proposed method, we selected benzo(k)fluoranthene (BKF) as a
representative PAH because of its high fluorescence quantum yield and narrow absorption
peaks. We select the longest wavelength absorption maximum of BKF, which occurs at
399.5 nm.

Fig. 6 shows the experimental setup. An excitation light beam from an near-UV LED
Author Manuscript

(UV5TZ-400–30, Bivar Inc.., USA) is collimated onto a narrow band interference filter
(Omega Optics, 405BP3, effective refractive index K = 2.223, FWHM = 3 nm, and normal
center wavelength λc = 404.6 nm) that is nearly matched to the BKF sample peak λ0 =
399.5 nm. This combination permits a modulation amplitude Δλ = 3.5 nm (Fig. 1).

LED is powered by laser-diode controller (LDC501, Stanford Research Systems, USA) that
is electronically regulated with an external voltage applied to its modulation input. For
wavelength modulation of the transmission band, the interference filter is tilted with a high-
speed (20 kilopoints per second (Kpps) galvo-scanner (PT-20K, TE-LIGHTING,
Guangzhou, China).

The galvo-scanner control voltage waveform was calculated using the galvo-scanner linear
transmission function and the equation (20). The obtained waveform was applied via digital-
Author Manuscript

analog converter (NI USB-6251, National Instruments, Austin, TX, USA) controlled by a
LabVIEW (National Instruments, Austin, TX, USA) custom program. For the full range of
excitation wavelength from λmin = 396 nm to λmax =403 nm, the corresponding excitation
light angles of incidence extrema are θmin = 11.5° to θmax = 27.3°.

Wavelength modulation sinusoidality was checked with spectrometer (FSD-9 mini, Art
Photonics, Berlin, Germany) by recording interference filter transmission spectra at 10s time

Appl Opt. Author manuscript; available in PMC 2019 November 21.

 Lanevski et al. Page 7

interval during 1.75 periods (T=6min) of interference filter tilting. The result, depicted in
Author Manuscript

Fig. 3, confirms high accuracy sinusoidal modulation of wavelength.

As interference filter oscillated the LED current was regulated to compensate excitation light
intensity changes. The regulation was done by another LabVIEW program (described in
next section below) that automatically found the required control voltage and provided it to
the LED power supply (SRS LDC501) through DAC of NI USB-6251 analog-digital
converter.

The resultant wavelength modulated and intensity stabilized excitation light was directed
onto the sample within a quartz cuvette with a 10 × 10 mm2 cross section. Fluorescence was
gathered at 90 degrees focused with a lens onto the preamplified photodiode
(ODA-6WB-500M, Opto Diode, Camarillo, California, USA) through a bandpass
interference filter (D430/30M, Edmund Optics, Barrington, NJ, USA). The photodiode
Author Manuscript

signal was measured at the second harmonic of excitation light wavelength modulation
frequency with a lock-in amplifier (SR850, Stanford Research Systems, USA).

B. Compensation of intensity changes

Although interference filter tilting allows to achieve an accurate sinusoidal wavelength
modulation, variation in the excitation light intensity, reaching the sample, can generate
unwanted second harmonic signals of background fluorescence. Two major potential sources
of this variation in our setup are the well-known [14] phenomenon of transmissivity changes
during interference filter tilting (Fig. 4.A) and the wavelength-dependent spectrum of
excitation light source emission spectrum (Fig. 4.B). Therefore, it is essential to compensate
excitation light intensity changes during wavelength modulation.
Author Manuscript

Excitation light intensity diminished 63% as we changed incidence angle from θmin = 11.5°
to θmax = 27.3° (narrow line in Fig. 5). We compensated this variation by regulating the
output current of the excitation source power supply via a control voltage waveform. This
waveform was automatically determined by a LabVIEW program. Firstly, it recorded raw
excitation intensities (Iraw) during excitation filter tilting, as measured at the detecting
photodiode without the emission bandpass filter (narrow line in figure 5). For each angle of
interference filter tilt, the LED power supply modulation input voltage was adjusted to a
constant photodiode signal, yielding the initial LED control voltage (Uinit) waveform.
Proportional–integral–derivative controller (PID) principles were used to optimize and
accelerate this process. However, stability of excitation intensity during application of initial
control voltage was insufficient. Therefore, a simple correction algorithm was developed
using the following equation:
Author Manuscript

U init * I perfect − I raw

U corr = (21)
I stab − I raw

where Ucorr is corrected control voltage, Iperfect is the theoretically perfectly constant
intensity and Istab is measured intensity stabilized by Uinit control voltage.

Appl Opt. Author manuscript; available in PMC 2019 November 21.

 Lanevski et al. Page 8

We found that the time dependence of LED supply control voltage was bell-shaped and
Author Manuscript

could be fitted by a Voigt function. We used it to eliminate the statistical noise of corrected
control voltage. Thus we achieved excitation light stability up to 0.5% (bold line in Fig. 5)
that we used in further experiments with PAH samples.

C. Preparation of PAH dilution series

To estimate WMS techniques suitability for PAH detection we prepared a series of BKF
solutions from 1⋅10−9M to 2.30⋅10−5M by serially diluting an initial solution with
concentration of 2.30⋅10−5M Initial solution was prepared from pure, crystalline
benzo(k)fluoranthe dissolved in methanol. The concentration of initial solution was
determined using benzo(k)fluoranthene extinction coefficient and absorption spectrum
measurements conducted with a spectrophotometer (V-570, Jasco Inc., Easton, MD, USA).
Dilution was performed with the use of variable pipettes (Genex beta, Genexlab,
Author Manuscript

Northampton, UK) with volume ranges of 20–200 μl and 100–1000 μl.

4. Results
Several measurements were conducted using the 2.30⋅10−5M BKF solution. The time-
dependent fluorescence signal was visualized by digital oscilloscope (TDS 524A, Tektronix
Inc., Beaverton, Oregon, USA) and compared with the galvo-scanner control voltage (which
varies at the modulation frequency ω). For excitation light wavelength modulation around
the absorption maximum, the fluorescence signal was, as predicted, at the second harmonic
2ω (panel A in Fig. 7). When wavelength modulation is centered at the upslope of the
absorption peak (panel B in Fig. 7), both signals have the same first harmonic frequency.
The finite response rate of the photodiode preamplifier (cutoff frequency = 130 Hz) results
in the visible phase shift in the fluorescence signals. Lock-in measurements confirmed the
Author Manuscript

second harmonic fluorescence signal prevailing over first harmonic with ratio 6:1 during
wavelength modulation around the absorption maximum, whereas first harmonic signal
prevailed during upslope modulation.

To estimate the sensitivity of WMS techniques for PAH detection we recorded the second
harmonic signal by SR850 lock-in amplifier for a series of concentrations (1⋅10−9M to
2.30⋅10−5M mol/L) of benzo(k)fluoranthe solutions (Fig. 8).

Due to direct proportionality of concentration and fluorescence intensity, we expected also a

linear relation between lock-in amplifier signal and benzo(k)fluoranthene concentration
changes. To check this hypothesis, we applied a linear regression to acquired data. We used a
weighted linear regression because of concentration heteroscedasticity. For the weighting
factor, we chose a 1/x2 as was suggested by Huidong Gu et al. in the case of direct
Author Manuscript

proportionality between variable and its standard deviation [15]. The obtained linear
approximation (Fig. 8) goes through almost all the points, and lies within concentration
uncertainties. Only one point deviates significantly from the linear regression line, what is
most probably the result of the limited precision in sample preparation. The minimum
detectable concentration, we achieved, was 5⋅10−9M. Lock-in amplifier signal for smaller
concentrations was indistinguishable from background signal.

Appl Opt. Author manuscript; available in PMC 2019 November 21.

 Lanevski et al. Page 9

The uncertainty of the measurements stem from two basic sources: the sample preparation
Author Manuscript

uncertainty and fluorescence intensity measurement uncertainty. The sample preparation

uncertainty originates from Jasco spectrophotometer instrumental error of optical density
measurements and from volume measurement uncertainty of Genex variable pipettes. These
uncertainties were taken into account and are depicted in Fig 7.

The fluorescence intensity measurement uncertainty results from the finite optical density
(OD = 5) of the fluorescence emission filter, which permitted some excitation light bleed-
through onto the detector. This resulted in a background signal about 50 μV during lock-in
measurements (while smallest measured fluorescence signal is 200 μV), that distorted the
fluorescence signal at concentrations smaller than 5⋅10−9M. The SR850 lock-in amplifier
instrumental uncertainty is quoted by the manufacturer in the range of 0.2 nV, and therefore
should not significantly affect these measurements.
Author Manuscript

5. Discussion
The results of the conducted experiments indicate that wavelength modulation spectroscopy
can be successfully applied for sensitive PAH detection. Our experience shows that the PAH
detection at the limit concentration of 5⋅10−9M can be achieved. However, we don’t see any
conceptual obstacles for diminishing this detection limit even further.

The detection limit concentration corresponds to a signal of 0.2 mV. The minimum signal,
which can be measured with SR850 lock-in amplifier, is 0.2 nV at low-pass filter bandwidth
of 10 mHz (corresponding to a time constant of 10s). This indicates that, in respect of
sensitivity of electronic equipment, several orders of magnitude lower concentrations could
be detected.
Author Manuscript

In our case it couldn’t be achieved due to imperfection of our experimental system. As we

found out, used fluorescence filter optical density at excitation light wavelengths wasn’t high
enough and caused excitation light leakage. This significantly contributed to background
during lock-in measurements and distorted the fluorescence signal at concentrations smaller
than 5⋅10−9M. Also due to excitation light residual intensity modulation (remained after
intensity stabilization) at level of 2–4% contained higher harmonics, we weren’t able to filter
excitation light leakage with the lock-in amplifier at second harmonic frequency. In
principal, the intensity stabilization of higher accuracy could be achieved, if we would build
a feedback system to control LED current instead of calculating control voltages. Or if we
could implement normalization of second harmonic signal (S2) by first harmonic signal (S2/
S1) within the lock-in amplifier as it was done by G.B.Rieker et al. [16]. In our case, this
was not possible because our lock-in amplifier (SR850 DSP) could not simultaneously
Author Manuscript

measure two harmonics. Our method has the advantage of a constant signal to noise ratio
during wavelength modulation.

In the real world pollution incidents the oil contaminants certainly contain several PAHs. In
the case of overlapping absorption bands the selectivity for single PAH detection could be
insufficient. Nevertheless, there is a possibility to enhance our method to further improve the
selectivity. In addition to excitation wavelength modulation we can selectively filter out the

Appl Opt. Author manuscript; available in PMC 2019 November 21.

 Lanevski et al. Page 10

fluorescence band with a narrowband interference filter. To further increase the selectivity,
we can modulate its transmission wavelength around the fluorescence peak center λfl while
Author Manuscript

registering the fluorescence signal at the second harmonic of the difference of the excitation
and fluorescence modulation frequencies 2(ω0 − ωfl). This excitation/emission double
selection method allows the selectivity to be greatly enhanced.

6. Conclusion
Sensitive detection of the polycyclic aromatic hydrocarbon benzo(k)fluoranthene by
application of the WMS method has been realized. The sensitivity and selectivity is achieved
by the lock-in detection of the second harmonic 2ω of the fluorescence signal under
sinusoidal modulation (at frequency 1ω) of excitation wavelength around the narrow BKF
absorption peak. The sinusoidal excitation wavelength modulation was achieved by precisely
tilting a narrowband interference filter in the excitation path. Variation of the filter
Author Manuscript

transmission was compensated by controlling the LED current, achieving an excitation

intensity variation below 4%. We detected benzo(k)fluoranthene to a concentration limit of
5⋅10−9M.

Acknowledgment.
We thank the Estonian Environmental Investment Centre for provided funding. We thank the Laser Spectroscopy
and Sensor Technology laboratories of Institute of Physics of University of Tartu for the use of their equipment. We
express our gratitude to Indrek Renge for providing benzo(k)fluoranthene samples and help in sample preparations.
We are thankful to Agu Anijalg for help with electronics of our experimental setup.

Funding Information. Estonian Environmental Investment Centre –EIC (project nr. 4352). Dr. Tkaczyk is grateful
for support from NIH K12 CA090625.
Author Manuscript

7. References
1. United States Environmental Protection Agency, “Polycyclic Aromatic Hydrocarbons” (Office of
solid waste, Washington, DC, January 2008).
2. European Commission, “Polycyclic Aromatic Hydrocarbons – Occurrence in foods, dietary
exposure and health effects” (Health and Consumer Protection Directorate, Scientific Committee on
Food, Brussels, Belgium, 2002).
3. European Commission Regulation No 1272/2013, OJ L 328 (2013), p. 69–71.
4. World Health Organization, “Polynuclear aromatic hydrocarbons in Drinking-water”, 2nd ed.
Addendum to Vol. 2. Health criteria and other supporting information (Geneva 1998).
5. Directive 2004/107/EC of the European Parliament and of the Council of 15 December 2004
relating to arsenic, cadmium, mercury, nickel and polycyclic aromatic hydrocarbons in ambient air,
OJ L 23 (2005), p. 3.
6. Kafilzadeh F, Shiva AH, Malekpour R, “Determination of Polycilyc Aromatic Hydrocarbons in
Water and Sediments of Kor River, Iran”, Middle-East Journal of Scientific Research 10, 1–7
Author Manuscript

(2011).
7. Soceanu A, Stanciu SD, Popescu V, “Polycyclic Aromatic Hydrocarbons in vegetables grown in
urban and rural areas”, Environmental Engineering and Management Journal, Vol. 13, No. 9, 2311–
2315 (2014).
8. Maliszewska-Kordybach B, “Sources, Concentrations, Fate and Effects of Polycyclic Aromatic
Hydrocarbons (PAHs) in the Environment”, Polish Journal of Environmental Studies, Vol. 8, No. 3,
131–136 (1999).
9. Larsson M, “Integrated Bioassay- and Chemical Analysis in Risk Assessment of Remediated PAH-
Contaminated Soils”, Licentiate Thesis, Öbero University, (2009).

Appl Opt. Author manuscript; available in PMC 2019 November 21.

 Lanevski et al. Page 11

10. Wenzl T, Simon R, Kleiner J, Anklam E, “Analytical methods for polycyclic aromatic
hydrocarbons (PAHs) in food and the environment needed for new food legislation in the European
Author Manuscript

Union”, Trends in Analytical Chemistry, Vol. 25, No. 7, 716–725 (2006).

11. Sakuma T et al., “Analysis of Polycyclic Aromatic Hydrocarbons (PAH), Alkylated Derivatives,
and Photo-degradation Products in Environmenta and Food Samples using LC-FLD-MS/MS with
Q TRAP Technology”, AB Sciex Food & Environmental brochure, p. 1–7 (2011).
12. Erlandsson M, Futter MN, Kothawala DN and Köhler SJ, “Variability in spectral absorbance
metrics across boreal lake waters”, J. Environ. Monit, 14, p. 2643–2652 (2012). [PubMed:
22885538]
13. O’Haver TC, “Derivative and Wavelength Modulation Spectrometry”, Analytical Chemistry vol. 51
No. 1, p. 91–100 (1979).
14. Lisseberger FH, Wilcock WL, “Properties of All-Dielectric Interference Filters. II. Filters in
Parallel Beams of Light Incident Obliquely and in Convergent Beams”, Journal of the Optical
Society of America, Vol. 49, No. 2, 126–130 (1989).
15. Gu Huidong et al., “Selecting the Correct Weighting Factors for Linear and Quadratic Calibration
Curves with Least-Squares Regression Algorithm in Bioanalytical LC-MS/MS Assays and
Author Manuscript

Impacts of Using Incorrect Weighting Factors on Curve Stability, Data Quality, and Assay
Performance”, Analytical Chemistry, 86, 8959–8966, (2014). [PubMed: 25157966]
16. Rieker GB, Jeffries JB, Hanson RK, “Calibration-free wavelength-modulation spectroscopy for
measurements of gas temperature and concentration in harsh environments”, Applied Optics, Vol.
48, No. 29, 5546–5560 (2009). [PubMed: 19823239]
Author Manuscript
Author Manuscript

Appl Opt. Author manuscript; available in PMC 2019 November 21.

 Lanevski et al. Page 12
Author Manuscript
Author Manuscript

Fig. 1.
Drawing showing part of the benzo(k)fluoranthene absorption spectrum and two possible
cases that can occur during PAH detection. Nr 1. describes the case when the wavelength
modulation is performed symmetrically about the absorption maximum. Nr 2. corresponds
to the case when maxima are not present in the selected wavelength modulation range Δλ.
Here, sinusoids under the absorption peak depict excitation light wavelength modulation
with period T. Sinusoids in the right upper corner show fluorescence intensity waveforms in
Author Manuscript

case of presence (2) or absence (1) of absorption maximum.

Author Manuscript

Appl Opt. Author manuscript; available in PMC 2019 November 21.

 Lanevski et al. Page 13
Author Manuscript

Fig. 2.
Power spectra of excitation light wavelength modulation obtained from equation (7) in the
case of an absorption maximum (A) and equation (9) in the case of a linear absorption
background (B).
Author Manuscript
Author Manuscript
Author Manuscript

Appl Opt. Author manuscript; available in PMC 2019 November 21.

 Lanevski et al. Page 14
Author Manuscript
Author Manuscript

Fig. 3.
Excitation light wavelength modulation waveform measured (dots) and approximated with
sinusoid (line).
Author Manuscript
Author Manuscript

Appl Opt. Author manuscript; available in PMC 2019 November 21.

 Lanevski et al. Page 15
Author Manuscript

Fig. 4.
Experimentally measured interference filter (Omega Optics, 405BP3) transmission
dependence on angle (A) and excitation light source (UV5TZ-400–30, near-UV LED, Bivar
Inc.) spectrum (B).
Author Manuscript
Author Manuscript
Author Manuscript

Appl Opt. Author manuscript; available in PMC 2019 November 21.

 Lanevski et al. Page 16
Author Manuscript
Author Manuscript

Fig. 5.
Excitation light intensity changes during one period of wavelength modulation without
(narrow dash line) and with (bold line) LED current adjustment.
Author Manuscript
Author Manuscript

Appl Opt. Author manuscript; available in PMC 2019 November 21.

 Lanevski et al. Page 17
Author Manuscript
Author Manuscript

Fig. 6.
Experimental setup schematics.
Author Manuscript
Author Manuscript

Appl Opt. Author manuscript; available in PMC 2019 November 21.

 Lanevski et al. Page 18
Author Manuscript

Fig. 7.
Waveforms of fluorescence intensity signals during wavelength modulation around
absorption maximum (A) and at the downslope of absorption peak (B).
Author Manuscript
Author Manuscript
Author Manuscript

Appl Opt. Author manuscript; available in PMC 2019 November 21.

 Lanevski et al. Page 19
Author Manuscript
Author Manuscript

Fig. 8.
Measured SR850 lock-in amplifier signal vs benzo-(k)fluoranthene probe concentrations,
approximated by weighted linear regression with weighting of 1/x2. Data is expressed in log-
log scale due to the broad range of concentrations. Error bars show uncertainty of
concentrations at a level of confidence of 95% (see text for details).
Author Manuscript
Author Manuscript

Appl Opt. Author manuscript; available in PMC 2019 November 21.