Talanta 207 (2020) 120283

Contents lists available at ScienceDirect

Talanta
journal homepage: www.elsevier.com/locate/talanta

A flame photometric detector with a silicon photodiode assembly for sulfur T

detection
Lanxiu Nia,b,1, Xuhui Genga,1, Shenghong Lia, Haijing Ninga, Yan Gaoa, Yafeng Guana,∗
a
CAS Key Laboratory of Separation Sciences for Analytical Chemistry, Dalian Institute of Chemical Physics, Chinese Academy of Sciences, 457 Zhongshan Road, Dalian,
116023, China
b
University of Chinese Academy of Sciences, Beijing, 100039, China

ARTICLE INFO ABSTRACT

Keywords: A flame photometric detector with a silicon photodiode assembly instead of a photomultiplier tube for sulfur
Sulfur detection detection was developed and evaluated. The photosensitive area of photodiode, the optical design, and band-
Photodiode assembly pass filters, were optimized. It was found that the optimal photosensitive area of the photodiode was 100 (mm)2,
Flame photometric detector and three focus lenses combined with a broad band-pass filter of 378/52 nm and a QB21 glass yielded the best
Band-pass filter
result. This design fully utilized the wide emission spectrum of S2*, the response characteristics of silicon
Low cost
photodiode, and effective absorption of strong emission spectrums of OH* at wavelength around 310 nm by
QB21 glass. The limits of detection for nine kinds of sulfur containing compounds were between 5.8 × 10−12 to
9.5 × 10−12 g s−1. This mode provided a linear response of 3 orders of magnitude for compounds being tested
and a selectivity of sulfur over carbon of 105. It is demonstrated for the first time that the overall performance of
the flame photometric detector integrated with a silicon photodiode assembly work at room temperature was
comparable to a conventional detector coupled with a photomultiplier tube, with advantages of short equili-
bration time, robust to electromagnetic interference and vibration, and low cost. The new detector can find wide
application in gas chromatography and on-line monitoring instruments for sulfur measurement.

1. Introduction Compared to PMT, Si photodiode (PD) has advantages such as fast

Flame photometric detector (FPD) is one of the major selective
detector in gas chromatography (GC) and on-line monitoring equip-
ment for quantitative measurement of sulfur-containing compounds
both in gases and liquid samples [1], such as in petroleum [2,3] and
natural gas [4], environmental [5–7], pesticides and preservative
[8–10], and chemical warfare agent [11,12]. Several other types of
sulfur-selective detectors are also employed in GC, such as pulsed flame
photometer detector (PFPD) [13,14], sulfur chemiluminescence de-
tector (SCD) [15], and atomic emission detector (AED) [16].
To our knowledge, all these sulfur detectors employ photomultiplier
tube (PMT) as photo detector since it provides fast response and ex-
tremely high sensitivity to photons emitted from S2*. However, the
PMT has some disadvantages [17,18], including relative high cost, re-
quirement of high polarization voltage supply (450–900 V) and a con-
ditioning circuit [19], long equilibration time (≥45 min), fragile to
strong light radiation during working, and susceptible to electro-
magnetic interference and vibration.
equilibrium, small in volume, resistant to strong light, tolerance to
electromagnetic interference and strong vibration, and lower cost. The
disadvantage of Si PD is substantially low sensitivity in comparison
with PMT. In recent years, however, PD integrated with pre-amplifier
(PDA) has been applied to fluorescence detection. Our previous work
developed a facile and highly sensitive micro fluorimeter by using a
light emission diode (LED) as light source and a PDA as photo detector.
The limit of detection for fluorescein isothiocyanate (FITC) was
0.004 μg L−1 [18], which is similar to that of PMT as photo detector.
Backhouse et al. [20] reported a fluorescence detector for microchip
electrophoresis by using a PD and a pre-amplifier with 1GΩ feedback
resistor for photo detection. The limit of detection for Alexa Fluor 532
end-labeled DNA was 0.15 nM. Li and coworkers [21] integrated a LED
and PDs in a compact pressure hull for measurement of the chemilu-
minescence of ammonia nitrogen in aquaculture with a limit of detec-
tion (LOD) of 0.16 μg L−1.
In this work, we explore what is possible if one attempts to use a
PDA as photo detector for FPD instead of a PMT. To compensate the

∗
Corresponding author.
E-mail address: guanyafeng@dicp.ac.cn (Y. Guan).
1
These authors contributed equally to this work.

https://doi.org/10.1016/j.talanta.2019.120283
Received 11 June 2019; Received in revised form 12 August 2019; Accepted 20 August 2019
Available online 21 August 2019
0039-9140/ © 2019 Published by Elsevier B.V.
L. Ni, et al.
Fig. 1. Schematic diagram of the FPD (a) integrated with a PDA and Nexis GC-2030 FPD integrated with a PMT.
weakness of PDA in sensitivity, a broad band-pass filter of 378/52 nm
and a QB21 glass were used for collection of the wide-wavelength
emission of S2* (350–410 nm), and absorption strong emission of OH*
around 310 nm, respectively. The optical design of the FPD was de-
tailed discussed and optimized. The different bond-structures of the
sulfur atoms (S]P, S–C, S–S, S–H, etc) would produce different che-
mical reactions in the hydrogen-rich flame, and greatly affect response
of sulfur containing compounds on the same FPD [22–24]. Therefore,
nine sulfur containing compounds were used to evaluate the perfor-
mance of the proposed FPD.
2. Materials and methods
2.1. Apparatus
An FPD model without photo detector was supplied by Focused
Photonics Inc (Hangzhou, China), and reconstructed by mounting op-
timized optical elements and PDA in our laboratory as shown in Fig. 1a.
An Agilent 6890 GC was used as a platform for study of the new FPD.
The original detector on the GC was dismantled, and replaced by the
home-made FPD. The signal from the PDA was acquired by a chroma-
tographic workstation Sepu 3010 (Puhui Science and Technology Ltd.,
Hangzhou, China). The actual flow rate of hydrogen and air that flow
through the FPD was measured by a soap film flowmeter. The hy-
drogen-rich flame was ignited by a heated wire igniter which was fixed
in the top chamber of the FPD.
2.2. The photo detection assembly
Lenses used in the FPD included a hemispherical lens (ϕ 25 mm,
focal distance 22 mm) and a planoconvex lens (ϕ 20 mm, focal distance
27 mm) from Focused Photonics Inc. (Hangzhou, China), and a bi-
convex lens (ϕ 16 mm, focal distance 16 mm) from Beijing Golden Way
Scientific Co. LTD (Beijing, China). Optical filters include a narrow
band-pass filter 394/20 nm/OD 4 from Focused Photonics Inc, a broad
band-pass filter 365/30 nm/OD 4 and a QB21 glass (ϕ18 mm, δ 2 mm)
from Bodian Optical Technology Co. LTD (Beijing, China), and a broad
band-pass filter 378/52 nm/OD 7 from Semrock (New York, USA). The
PDA used in the experiment was an AccuOpt 2003 with photosensitive
area of 100 (mm)2 from Dalian Scien & Tech Instrument Inc. (Dalian,
China). The arrangement of all optical parts and the PDA were shown in
Fig. 1a.
2
Talanta 207 (2020) 120283
2.3. GC conditions
The GC conditions were as follows: injector and detector tempera-
ture, 250 °C; detector gas flow，75 mL min−1 H2 and 110 mL min−1
Air. A 30 m × 0.32 mm i.d. × 0.40 μm, SE-54 capillary column (Dalian
Scien & Tech Instrument Inc., Dalian, China) was used as separation
column. Before evaluating the performance of FPD, 20 μg mL−1 of
methyl-parathion was used for optimization: splitless injection, purge
time 1 min and purge flow 50 mL min−1; column temperature, 210 °C,
carrier gas flow rate, 5 mL min−1 N2. For measurement of benzene, n-
dodecane and other eight sulfur-containing compounds, the GC condi-
tions were as follows: column temperature, 40 °C for 2 min, program to
60 °C at 10 °C min−1, then to 220 °C at 40 °C min−1; carrier gas flow
rate, 3 mL min−1 N2. For benzene and n-dodecane, splitless injection
was used, while for other eight sulfur-containing compounds, split in-
jection (ratio 10:1) was used.
2.4. Chemicals and reagents
Standard methyl-parathion sample of 1 mg mL−1 in methanol was
purchased from Aladdin (Shanghai, China). Stock solution was prepared
by diluting the standard sample with ethanol (HPLC grade, Merck) at the
concentration of 20 μg mL−1 unless otherwise specified, and stored in
darkness at −20 °C. Working solutions were prepared daily from this stock
solution by diluting it in ethanol unless otherwise specified.
Tetrahydrothiophene of purity 99%, 2-propanethiol of purity 96%, 1-bu-
tanethiol of purity 98%, isopropyl disulfide of purity 96%; dipropylsulfide
of purity 98%; and thianaphthene of purity 98%, were purchased from
Energy Chemical (Shanghai, China). Diethyl sulfide of purity 98% was
from J&K (Beijing, China). 1-Methyl-1-propanethiol of purity > 95% was
from Innochem (Beijing, China). Standard sulfur mixtures were prepared
by dissolving each of the chemicals at 1000 μg mL−1 in n-hexane (HPLC
grade, TEDIA) and stored in darkness at −20 °C. Working solutions were
prepared daily from the stock solutions by dissolving it in n-hexane. To
obtain the selectivity of sulfur over carbon, benzene (> 98%; kermel,
Tianjin, China) and n-dodecane (> 98%; Fluka, USA) dissolved in hexane
at concentration of 10, 50 and 100 mg mL−1.
3. Results and discussion
3.1. Selection of photosensitive area of the PDA
The initial optical setup included an ϕ 25 mm hemispherical lens
(11 in Fig. 1a), a narrow band pass filter of 394/20 nm (13 in Fig. 1a),
L. Ni, et al.
Fig. 2. The relation of SNR vs distance (a) between PDA and planoconvex lens, and distance (b) between PDA and a biconvex lens.
and a planoconvex lens (14 in Fig. 1a).
In order to know the size of the focal facula on the photosensitive
surface of PDA, we simulated the light emission in the FPD by using a ϕ
3 LED with 532 nm wavelength and divergence angle of 45° and placed
in the chemiluminescent zone along the central axis of the quartz
combustion chamber with head pointing to the nozzle, since the che-
miluminescence of S2* is impossible to see by naked eye. A screen was
positioned at the location of the PDA window to observe the size of the
focal facula, as shown in Fig. S2a. It was found that the smallest dia-
meter of the focal facula was about 12 mm, as shown in Fig. S2b. We
then selected an AccuOpt 2003 as the PDA since its photosensitive area
is 10 × 10 (mm)2, the largest available to us.
It showed that the signal-to-noise ratio (SNR) value declined
monotonously as the distance between PDA and planoconvex lens (d1)
increased (Fig. 2a). The result proved that the total light intensity of S2*
irradiated on the detection window sharply decreased as the distance
increased. When the PDA was placed as near as possible to the plano-
convex lens, the SNR of 20 μg mL−1 methyl-parathion reached 557,
corresponding to LOD of 1.9 × 10−11 g s−1.
3.2. Optimization of optical design
At the location close to planoconvex lens, the diameter of facula was
found to be about 16 mm, which was the same as the effective diameter
of the lens. The utilization rate of light by the PDA was 10 × 10/
[π × (16/2)2 ] = 49.76%. A biconvex lens of effective diameter 16 mm
was then added between the PDA and planoconvex lens, in order to
reduce the size of focusing facula to about 10 mm to match the pho-
tosensitive area of PDA. However, the biconvex lens brought additional
light loss of 10% because of surface reflection and absorption by the
glass. Fig. 2b showed the relation of SNR vs the distance between PDA
and a biconvex lens (d2) of this optical configuration, which demon-
strated the highest SNR of 20 μg mL−1 methyl-parathion reached about
698 at d2 = 2 mm. The addition of a biconvex lens brought 25% in-
crease on SNR and lowered the LOD down to 1.7 × 10−11 g s−1.
3.3. Optimization of band-pass optical filter
The chemiluminescent spectrum of excited S2* is distributed be-
tween 330 and 430 nm [25–28], and mainly in 350–410 nm region
[27,29]. Since hydrocarbons are usually the main constitute in sample
and sulfur is in trace level, the selectivity of sulfur over hydrocarbons is
an important parameter for the FPD. The main emission spectrum of
CH* is between 430–530 nm, but there is a weak emission peak around
387 nm, which is located in the S2* spectrum. Traditionally, the band-
pass filter employed in FPD was 394/20 nm or 384/20 nm narrow
3
Talanta 207 (2020) 120283
band-pass filter, and a PMT was used as photo detector. The filter
should block the interference spectrum of both CH*and OH*. Though a
broad band-pass filter could increase the S2* signal, the background
noise came from the emission light of CH* and the stray light was also
increased. In addition, the baseline noise of a PMT is proportional to the
incident light intensity owing to the intrinsic property of electron
emission materials coated on the photocathode and the electron mul-
tiplication principle within the PMT. Therefore, a narrow band-pass
filter could achieve a high SNR value for use with a PMT as photo-
detector [19]. In contrary, the photosensitivity of Si PD is only
0.1–0.2 A W−1 at wavelength around 300–400 nm (shown in Fig. S1),
and the baseline noise is independent of incident light intensity, though
its photosensitivity is much lower than a PMT. Therefore, a broad band-
pass filter shall be the first choice for use with a PD. However, we found
that the absolute baseline level was 2060 mV and 658 mV, when the
band-pass filter of 365/30 nm and 378/52 nm was used alone, re-
spectively. This phenomenon indicated that the high density of OH* in
the hydrogen-rich flame generated a strong 310 nm emission, which
passed through the filter from different angle and yielded a strong stray
light with wavelength within the filter wavelength window, causing a
high background level. We then selected a QB21 glass with 2 mm
thickness and placed in front of a band-pass filter to absorb more than
99% of OH* emission light (shown in Fig. 3). With the QB21 glass, the
absolute baseline level was then down to 1.27 mV for use with 365/
30 nm band-pass filter, and 1.3 mV for 378/52 nm band-pass filter. The
S2* emission wavelength, including 354 nm, 364 nm, 374 nm, 384 nm,
394 nm and 403 nm, were all utilized by using the 378/52 nm band-
pass filter, yielding SNR value of 1555; while the SNR value was 978.6
and 698.6 by using the 365/30 nm or 394/20 nm band-pass filter, as
demonstrated in Table 1.
Fig. 3. The transmittance vs wavelength of QB21 glass.
L. Ni, et al. Talanta 207 (2020) 120283

Table 1 and noise level was 15 μV. The peak height of 200 μg mL−1 sulfur-
Detector response by using different optical filters. containing compounds was above 1250 mV with attenuation (1/8). To
Optical filtera Hc/μV Ad/μV·s RSD/%(H, RSD/%(A, S/N broaden the linear response of the FPD, the acquired signal of the PDA
n = 3) n = 3) was further attenuated 11 times before Sepu 3010. The signal to noise
Hattenuation (11) × 11
was calculated by SNR = 15
.
394/20 nm 69865 404876 0.8 1.0 698.7
As mentioned earlier, the molecular structure of sulfur containing
365/30 nmb 97863 580160 1.5 1.5 978.6
378/52 nmb 155535 910654 1.3 0.7 1555.4 compounds could make the response of FPD various. The performance
of the FPD, besides methyl-parathion, was evaluated by using eight
a
Center wavelength-FWHM. sulfur containing compounds, including mercaptan, thioether, disulfide
b
In combination with a QB21 glass. and thiophene since they represent different types of compounds. To
c
Peak height. assess the linearity of the FPD, samples at 9 concentrations (0.5, 0.7,
d
Peak area. 1.0, 5.0, 10, 50, 100, 200, 400 μg mL−1 of each) were measured in
triplicate. Fig. 4 illustrated the results of logarithm SNR varied with
logarithm grams of sulfur per second for the nine sulfur containing
compounds. Their corresponding log (g/s) and log (SNR) values were
listed in Table S2. The linear response of sulfur containing compounds
spanned about 3 orders of magnitude with a slope close to 2, verifying
that the signal was proportional to the square of concentration of sulfur-
containing compounds, as shown in Fig. 4. The LODs for different sulfur
containing compounds were ranged from 5.8 × 10−12 to
9.5 × 10−12 g s−1, as illustrated Table 2. A performance comparison of
various FPDs, including commercially available ones and literature
reported ones, were listed in Table 3. The sensitivity, linearity and se-
lectivity of the present FPD was in the same order as commercial ones.
The selectivity of sulfur over carbon was measured by using benzene
and n-dodecane as sample, and the data were shown in Fig. 4. With
benzene as the hydrocarbon tested analyte, the calculated values of
sulfur/carbon (S/C) were between 1.4 × 105 to 4.1 × 105 depending
on specific sulfur compounds being compared (Table 2). Fig. 5 showed
the chromatogram of eleven tested analytes. Because the injection mass
was far beyond the capacity of the separation column, the peaks of two
hydrocarbons were forward. Even the sample concentration was
50 mg mL−1, S/N of benzene was only ~12.
Fig. 4. Logarithm (SNR) varied with logarithm (grams of element) per second.

3.4. Evaluation of the FPD
According to the principle of FPD response, the signal was propor-
tional to the square of concentration of sulfur-containing compounds.
The peak height of 20 μg mL−1 methyl-parathion was ~155 mV. By
calculating, the peak height of 100 μg mL−1 methyl-parathion should
be around 3875 mV, which was far beyond 1250 mV of the input upper
limit of the chromatographic workstation. This would suppress the
dynamic range of the FPD. Considering the baseline noise of AccuOpt
2003 is about 100 μV and the maximum output voltage is 10 V, there is
a room for PDA output attenuation in order to meet the dynamic range
of FPD responses, since the noise level of the chromatographic work-
station is less than 2 μV. A circuit with attenuation 8 times was con-
nected between the PDA output and the chromatographic workstation.
The SNR value after attenuation was 1347.4, a little drop from 1555.35,
4. Conclusion
An FPD with a PDA as photo detector instead of an expensive PMT
was successfully developed by fully utilization of S2* emission spectrum
and the intrinsic characteristics of Si PD. A PDA with 100 (mm)2 pho-
tosensitive area and a broad band-pass filter of 378/52 nm in combi-
nation with a QB21 glass were found to achieve the best performance.
The LOD for different sulfur containing compounds was from
5.8 × 10−12 to 9.5 × 10−12 g s−1, the linear response for sulfur was 3
orders of magnitude, the selectivity of sulfur over carbon was 105. The
overall performance of the new FPD was comparable to commercially
available FPDs, while the cost and robustness of the FPD were ad-
vantageous than others. This FPD was applicable for quantitative
measurement of sulfur containing compounds both in gases and liquid
samples in various disciplines.

Table 2
The performance of the new FPD.
Compounds Linear fitting equation R2 P Response range/μg·mL−1 LOD/g·s−1 S/Ca

2-propanethiol y = 26.3 + 2.3x 0.9942 6.0 × 10−8 0.5–400 9.5 × 10−12 1.4 × 105
1-methyl-1-propanethiol y = 25.4 + 2.2x 0.9979 3.2 × 10−9 0.5–400 6.4 × 10−12 2.3 × 105
diethyl sulfide y = 24.2 + 2.1x 0.9979 3.1 × 10−9 0.5–400 6.5 × 10−12 2.5 × 105
1-butanethiol y = 28.5 + 2.5x 0.9973 6.1 × 10−9 0.5–400 9.1 × 10−12 1.8 × 105
tetrahydrothiophene y = 24.5 + 2.1x 0.9974 5.5 × 10−9 0.5–400 6.5 × 10−12 3.0 × 105
dipropylsulfide y = 24.9 + 2.1x 0.9945 5.3 × 10−8 0.5–400 6.7 × 10−12 2.2 × 105
isopropyl disulfide y = 24.1 + 2.0x 0.9950 3.9 × 10−8 0.5–400 6.7 × 10−12 4.1 × 105
thianaphthene y = 24.6 + 2.1x 0.9952 3.5 × 10−8 0.5–400 5.8 × 10−12 3.2 × 105
methyl-parathion y = 24.5 + 2.1x 0.9997 4.3 × 10−8 1–200 8.4 × 10−12 2.0 × 105

a: S/C was calculated with benzene as the hydrocarbon test analyte.

4
L. Ni, et al. Talanta 207 (2020) 120283

Table 3
Performance comparison of sulfur selective detectors.
Detector LOD (g s−1） Selectivity(S/C) Linear dynamic range (decades) Ref.

−12 6
Agilent FPD+ 2.5 × 10 10 >3 [30]
SS mFPDa 9.0 × 10−12 104 4 [25]
Our work 8.4 × 10−12 > 105 3
Pulse photometric detector (PFPD) 10–13 107 – [13,14]
Sulfur chemiluminescence detection (SCD) 10–13 106–107 3–4 [15]
Atomic emission detection (AED) 10–12 104 3–4 [16]

a
Stainless steel multiple flame photometric detector.

Fig. 5. The chromatogram of all eleven tested analytes. Sample concentration:
5 μg mL−1 for sulfur containing compounds, 50 mg mL−1 for hydrocarbons.
Acknowledgment
This work was supported by National Key R&D Program of China
(No. 2017YFF0106102).
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://
doi.org/10.1016/j.talanta.2019.120283.
References
[1] M.J. Wang, L.J. Liu, J.C. Wang, L.P. Chang, H. Wang, Y.F. Hu, Sulfur K-edge XANES
study of sulfur transformation during pyrolysis of four coals with different ranks,
Fuel Process. Technol. 131 (2015) 262–269.
[2] C.T. Yue, S.Y. Li, H. Song, Simulation experiments on the generation of organic
sulfide in the Shengli crude oil, Geochem. Int. 53 (2015) 1052–1063.
[3] Y.H. Wang, R.T. Yang, J.M. Heinzel, Desulfurization of jet fuel JP-5 light fraction by
MCM-41 and SBA-15 supported cuprous oxide for fuel cell applications, Ind. Eng.
Chem. Res. 48 (2009) 142–147.
[4] X. Li, L. Shen, Q. Luo, L. Zhou, F. Yang, Z. He, W. Yang, Standard methods and
related techniques for on-line detection of the total sulfur content in natural gas,
Nat. Gas. Ind. B 5 (2018) 99–104.
[5] S. Jang, K.-T. Park, K. Lee, Y.-S. Suh, An analytical system enabling consistent and
long-term measurement of atmospheric dimethyl sulfide, Atmos. Environ. 134
(2016) 217–223.
[6] B.F. Yu, Y.H. Song, H.B. Yu, L. Han, H.L. Liu, Optimizations of large volume-direct
aqueous injection-gas chromatography to monitor volatile organic compounds in
surface water, Anal. Methods 6 (2014) 6931–6938.
[7] B.F. Yu, Y.H. Song, L. Han, H.B. Yu, Y. Liu, H.L. Liu, Optimizations of packed
sorbent and inlet temperature for large volume-direct aqueous injection-gas chro-
matography to determine high boiling volatile organic compounds in water, J.
Chromatogr. A 1356 (2014) 221–229.
[8] M.M. Rahman, A.M.A. El-Aty, S.-W. Kim, S.C. Shin, H.-C. Shin, J.-H. Shim, Quick,
easy, cheap, effective, rugged, and safe sample preparation approach for pesticide
residue analysis using traditional detectors in chromatography: a review, J. Sep. Sci.
40 (2017) 203–212.
[9] X.S. Zhao, W.J. Kong, J.H. Wei, M.H. Yang, Gas chromatography with flame pho-
tometric detection of 31 organophosphorus pesticide residues in Alpinia oxyphylla
dried fruits, Food Chem. 162 (2014) 270–276.
[10] E. Engel, J. Ratel, P. Blinet, S.T. Chin, G. Rose, P.J. Marriott, Benchmarking of
candidate detectors for multiresidue analysis of pesticides by comprehensive two-
dimensional gas chromatography, J. Chromatogr. A 1311 (2013) 140–148.
[11] T.P. Logan, J.S. Graham, J.L. Martin, J.E. Zallnick, E.M. Jakubowski, E.H. Braue,
Detection and measurement of sulfur mustard offgassing from the weanling pig
following exposure to saturated sulfur mustard vapor, J. Appl. Toxicol. 20 (2000)
S199–S204.
[12] W.K. Fowler, J.J.E. Smith, Solid sorbent collection and gas chromatograph de-
termination of bis(2-chloroethyl)sulfide in air at trace concentration, J.
Chromatogr. Sci. 28 (1990) 118–122.
[13] A. Amirav, H.W. Jing, Pulsed flame photometer detector for gas chromatography,
Anal. Chem. 67 (1995) 3305–3318.
[14] S. Cheskis, E. Atar, A. Amirav, Pulsed-flame photometerr: a novel gas chromato-
graphy detector, Anal. Chem. 65 (1993) 539–555.
[15] M.A. Khan, M.E. Whelan, R.C. Rhew, Analysis of low concentration reduced sulfur
compounds (RSCs) in air: storage issues and measurement by gas chromatography
with sulfur chemiluminescence detection, Talanta 88 (2012) 581–586.
[16] D.D. Link, J.P. Baltrus, K.S. Rothenberger, P. Zandhuis, DonaldMinus,
R.C. Striebich, Rapid determination of total sulfur in fuels using gas chromato-
graphy with atomic emission detection, J. Chromatogr. Sci. 40 (2002) 500–504.
[17] X. Geng, M. Shi, H. Ning, C. Feng, Y. Guan, A compact and low-cost laser induced
fluorescence detector with silicon based photodetector assembly for capillary flow
systems, Talanta 182 (2018) 279–284.
[18] X.H. Geng, Y. Gao, C.B. Feng, Y.F. Guan, A facile and high sensitive micro fluori-
meter based on light emitting diode and photodiode, Talanta 175 (2017) 183–188.
[19] S.S. Brody, J.E. Chaney, Flame Photometric Detector Jounal of Gas
Chromatography, (1966), pp. 42–46.
[20] G.H. Hall, D.M. Glerum, C.J. Backhouse, Light emitting diode, photodiode-based
fluorescence detection system for DNA analysis with microchip electrophoresis,
Electrophoresis 37 (2016) 406–413.
[21] C. Wang, Z. Li, Z.l. Pan, D.L. Li, Development and characterization of a highly
sensitive fluorometric transducer for ultra low aqueous ammonia nitrogen mea-
surements in aquaculture, Comput. Electron. Agric. 150 (2018) 364–373.
[22] T.C. Hayward, K.B. Thurbide, Quenching-resistant multiple micro-flame photo-
metric detector for gas chromatography, Anal. Chem. 81 (2009) 8858–8867.
[23] Y.S. Toshiaki Sugiyama, Tsugio Takeuchi, Intensity characteristics of S2 emission
for sulfur compounds with flame photometric detector, J. Chromatogr. Sci. 11
(1973) 639–641.
[24] K.B.T. Taylor, C. Hayward, Characteristics of sulfur response in a micro-flame
photometric detector, J. Chromatogr. A 1105 (2006) 66–70.
[25] A.G. Clark, K.B. Thurbide, An improved multiple flame photometric detector for gas
chromatography, J. Chromatogr. A 1421 (2015) 154–161.
[26] A.G. Clark, K.B. Thurbide, Properties of a novel linear sulfur response mode in a
multiple flame photometric detector, J. Chromatogr. A 1326 (2014) 103–109.
[27] A.G. Clark, K.B. Thurbide, Spectral examination of a multiple-flame photometric
detector for use in chromatography, Can. J. Chem. 92 (2014) 629–634.
[28] T.C. Hayward, K.B. Thurbide, Quenching-resistant multiple micro-flame photo-
metric detector for gas chromatography, Anal. Chem. 81 (2009) 8858–8867.
[29] R.M. Dagnall, K.C. Thompson, T.S. West, Molecular-emission spectroscopy in cool
flames part I. the behaviour of sulphur species in a hydrogen-nitrogen diffusion
flame and in a shielded air-hydrogen flame, Analyst 92 (1967) 506–512.
[30] https://www.agilent.com/zh-cn/products/gas-chromatography/gc-systems/
7890b-gc-system.