In this paper, the continuous online measurements of isoprene in the atmosphere have been carried out by using differential optical absorption spectroscopy (DOAS) in the band of 202.71–227.72 nm for the first time. Under a zero optical path in the laboratory, different equivalent concentrations of isoprene were measured by the combination of known concentrations of gas and series calibration cells. The correlation between the measured concentrations and the equivalent concentrations was 0.9995, and the slope was 1.065. The correlation coefficient between DOAS and the online volatile organic compound (VOC) instrument observed from 23 d of field observations is 0.85 with a slope of 0.86. It was estimated that the detection limit of isoprene with DOAS is approximately 0.1 ppb at an optical path of 75 m, and it was verified that isoprene could be measured in the ultraviolet absorption band using the DOAS method with high temporal resolution and a low maintenance cost.
Isoprene, named as 2-methyl-1,3-butadiene (C
Isoprene produced by plants is a byproduct of photosynthesis; its emission
intensity directly relates to the abundance of plants, leaf area index and
plant species. Meteorological parameters, such as temperature, radiation
intensity and humidity, can also affect isoprene emissions (Bai, 2015). In
the daytime, the chemical process oxidized by OH is the main sink of
isoprene. Due to the existence of multiple double bonds, the additional
reaction with OH will lead to the formation of a variety of products and the
formation of RO
In recent years, with the increase in urban vegetation diversity, the emission intensity of urban BVOCs has shown a significant upward trend. The monitoring and control of isoprene in urban ecosystems have also attracted increasing attention. Because the isoprene concentration in the atmosphere is low and its lifetime is short, highly precise and accurate methods are needed for monitoring. Currently, general methods, including gas chromatography–mass spectrometry (GC-MS), proton transfer reaction mass spectrometry (PTR-MS), and chemical ionization mass spectrometry (CIMS), have been introduced to measure isoprene.
GC-MS utilizes the high separation ability of gas chromatography to separate the components of environmental samples and then measures the different compounds with the mass spectrometer. With the advantages of high precision and stability, GC-MS can distinguish most VOCs qualitatively and quantitatively; however, it is difficult to maintain and operate due to the complex requirements of power, temperature control and special carrier gas. GC-MS measurement generally requires sampling, preservation and pre-treatment before analysis. During this process, the sample may change to some extent, resulting in inaccurate results.
PTR-MS involves the chemical ionization of a gas sample through proton
transfer in a drift tube. The proton source is usually H
Comparison of different online methods for isoprene measurement.
CIMS (Leibrock and Huey, 2000) retains the qualitative ability of mass spectrometry and couples the traditional air sampler with mass spectrometry technology. However, this method is not sensitive to low concentrations of isoprene. In addition, the VOC composition in the atmosphere is complex, and an unknown composition may react with the benzene reagent to interfere with the measurement results. Table 1 lists the comparison of the performance of these three methods for isoprene measurements together with the differential optical absorption spectroscopy (DOAS) method in this study.
In addition, a portable gas chromatograph (iDirac) equipped with a photo-ionization detector to measure isoprene was proposed by Bolas et al. (2020) at Cambridge University. The instrument is an improved technology for GC-MS that can work independently for weeks to months in the field environment. Previous studies have rarely mentioned the measurement of isoprene by spectral methods. Brauer et al. (2014) measured the infrared spectrum of isoprene by Fourier transform spectrometer and found that isoprene has a strong absorption near 11 000 nm, which provides a new possibility for the measurement of isoprene by spectral technology. So far, however, few people have mentioned the measurement of isoprene by ultraviolet spectroscopy. In this paper, an online measurement method with high temporal resolution for isoprene in the atmosphere is proposed by using DOAS technology in the far ultraviolet band.
DOAS technology was initially proposed by Platt et al. (1979, 1980) in the 1970s. The principle of the instrument has been detailed in other literature (Platt and Stutz, 2008), so the following is a description of deep UV-DOAS. The system is mainly composed of a light source, transmitting telescope, receiving telescope, spectroscope, and computer, etc. (see Fig. 2). The transmitting and receiving telescopes are located at both ends of the measuring optical path with a distance of 75 m. Since the measurement of isoprene detects the absorption in deep ultraviolet light, we choose a deuterium lamp (L6311-50, Hamamatsu, 35 W) as the light source. The aperture of the transmitting telescope is 76 mm, with a UV-enhanced spherical mirror with a focal length of 304 mm. The aperture of the receiving telescope is 152 mm with a UV-enhanced spherical mirror with a focal length of 608 mm. A spectroscope (B&W TEK Inc. BRC741E-1024) with a spectral range of 185–400 nm, a spectral resolution of 0.75 nm FWHM (full width at half maximum), and a 1024-pixel photodiode array was used as the detector to record the spectrum. In the measurement routine, the light emitted by the light source is collimated by the transmitting telescope and then sent out. After a certain distance of transmission, it is collected by the receiving telescope and focused on the incident end of the optical fiber. The optical fiber feeds the light into the spectroscope, which detects the light signal and sends it to the computer for spectral analysis.
The measured atmospheric spectrum contains the absorption information of molecules in the atmosphere. After removing the Rayleigh scattering and Mie scattering, as well as the broadband absorption of molecules by high-pass filtering, the so-called differential absorption spectrum is obtained. This high-pass filtering is performed by a high-pass binomial on the spectrum using the 500 iterations twice to eliminate the broadband structures. The concentration of the corresponding atmospheric components can be retrieved by fitting the differential absorption spectrum with the differential absorption cross section of the measured molecules. The reference spectrum during laboratory experiments was recorded by receiving a light beam close to the transmitting device, suggesting a zero light path and no absorption of isoprene. In the field measurements, the measured atmospheric spectrum collected at 00:00 LT on 1 July 2018 was used as the reference spectrum considering it is “clean” without isoprene absorption.
Isoprene has strong absorptions between 200.0–225.0 nm, among which there
are relatively obvious absorption peaks (Martins et al., 2009) near 210.0, 216.0 and 222.1 nm, as shown in Fig. 1a. After high-pass
filtering, the differential absorption spectrum (1 ppb km) of isoprene is
shown in Fig. 1b. According to its differential absorption
characteristics, the fitting band of isoprene is 202.71–227.72 nm. Within
this band, there are also absorptions of NH
The absorption cross section and differential absorption spectrum of isoprene in 1 ppb km.
Example of the spectral fitting of an actual atmospheric spectrum (measured on 8 July 2018 at 12:47 LT).
To verify the accuracy of the measurement results, isoprene gas with a known
concentration was used to calibrate the instrument in the laboratory. The
method is to close the emitting telescope and receiving telescope (close to
zero optical path) in the laboratory, and then a series absorption cell is
placed between the telescopes. Isoprene gas (10 ppm) was injected into the
cells at a constant flow rate of 100 mL min
The scheme of the calibration system.
The absorption cell group is composed of one 2 cm and two 4 cm long cells in
series. When using different combinations of cells, different equivalent
concentrations (
The calibration results in different gas cell combinations.
Figure 4 shows the linear fit of the calibration results. The ordinate in
the figure is the equivalent concentration, and the abscissa is the measured
concentration. For six measuring points, including the zero point, the
linear fitting correlation coefficient
The linear fitting of calibration results for isoprene measurement.
To further verify the reliability of the DOAS method in actual atmospheric
measurements, in July 2018, the field measurement results of the DOAS were
compared with the online VOC (TH-300B online VOC monitoring system)
analyzer (Zhu et al., 2020), which is based on the GC-MS technology. The
DOAS instrument is installed on the seventh floor of the Environmental Science
Building (31.344
Field measurement sites of DOAS and online VOCs, A is the transmitting telescope, B is the receiving telescope, and C is the online VOCs, and the yellow arrow is the light path of the DOAS. This map is sourced from © Baidu.
The comparison experiment was carried out from 1 to 23 July
2018. The temporal resolution of the DOAS was 1 min, while that of the
online VOCs was 1 h. To match the temporal resolution, the DOAS data were
averaged hourly. Moreover, the measured spectra with low light intensity and
high integration time were excluded from the spectral fitting and data
processing, which were mainly due to the unfavorable weather conditions
influencing the measurements. The spectra were also corrected for offset
before introducing fitting. Figure 6a shows the time series of the
isoprene data measured by these two instruments, which are in good
agreement. The average values of DOAS and online VOCs were 0.325 and
0.217 ppb, respectively, and the standard deviations (SDs) were 0.254 ppb
(
The comparison of hourly isoprene measured by DOAS and online VOCs during the field measurement.
The main reason for the difference in DOAS and online VOC results is that
the sampling and measurement heights of the two instruments are different.
The light path of the DOAS is about 25 m above the ground, while the
sampling height of online VOC instrument is about 10 m. In addition to the
500 m distance between these two sites, the air sampled by the VOC analyzer
or penetrated by the DOAS light beam is completely different. The
inhomogeneous spatial distribution of isoprene will lead to different data
results between the two instruments. Considering that the sampling of
online VOCs occurs through the sampling tube, isoprene will be more or less
lost during the sampling process, which could account for up to 10 % in
some high-carbon VOCs (EPA, 2019). To ensure the authenticity and accuracy
of the observed data, the working status and response of the TH-300B
monitoring system were inspected every day. Daily calibrations were
performed automatically at 00:00 to 01:00 LT. In addition, the external
standard method for the FID (flame ionization detector) and the internal standard method for the MS were
adopted. Implementing the daily calibration at midnight could move the
online VOC-observed value close to the zero point, which may deviate from
the actual abundance. Since the observation is in summer, there is also a
very high temperature at night during the observation period, i.e.,
27.1
It can also be seen in Fig. 6b that when the isoprene concentration is
higher than 0.5 ppb, the measurement results of the two instruments show
large scattering. The different measurement principles, especially the
difference in sampling time, can also cause scattering of the results of the
two instruments. Online VOCs only have about 50 % of the time (1 h) to be
used for sampling, while the rest of the time is used for analysis. However,
DOAS is an almost continuous measurement with just a small part of the time
to be used for analysis (about 1 s min
The detection limit of DOAS mainly depends on the signal-to-noise ratio of the spectrum. Under the condition of a zero light path in the laboratory, the zero noise (standard deviation of the results) of isoprene is 0.005 ppb, and the detection limit can be defined as 2 times the zero noise so that the detection limit of the system is 0.010 ppb (HJ 654-2013, 2013). However, in real atmospheric measurements, it is difficult to determine the actual detection limit due to the varied environment and the interference of other gases. The detection limit of DOAS in a real atmosphere is mainly determined by the residual of spectral fitting. This residual mainly comes from the absorption of interfering substances, the change in lamp spectral intensity and structure, the spectral shift caused by the change of ambient temperature of the spectrometer, and the noise of the detector. Since the stability of the light source and spectrometer will influence the fitting residual and instrumental performance, temperature control was adopted for the spectrometer and operating ambient environment. To reduce the influence of these factors on the measurement, during the spectral fitting process, the absorption of interfering substances and the spectral structure of the lamp must be considered together with the isoprene absorption spectrum. The lamp spectrum will also be introduced into the fitting process if an obvious lamp spectral structure was observed in the residual. At the same time, it is also necessary to calibrate the spectral drift. However, some residuals remain after spectral fitting due to possible imperfect reference spectra. Overall, the averaged measurement errors of isoprene were estimated to be lower than 20 %.
In the fitting band of isoprene, the absorption of NO, benzene and toluene
are the main interference factors. The reason for the influence of NO is
that there are three obvious absorption peaks of NO in the fitting band.
After high-pass filtering, there is a component in the differential
absorption cross section of NO similar to the variation frequency of
isoprene's differential absorption spectrum. After an analysis of the
measurement results, the impact of NO on isoprene is about 0.3 % of its
concentration. However, the effect of NO mainly occurs in the morning and
evening rush hours. The influence of benzene and toluene is mainly due to
their strong absorptions in the fitting band of the spectrum. Their presence
will lead to a significant reduction in the spectral intensity in this band,
resulting in a reduction in the signal-to-noise ratio of the spectrum.
During the comparison experiment, a high concentration of benzene or toluene
occasionally occurs, resulting in a large fitting residual. Other aromatics,
such as xylene and styrene, also absorb strongly in the fitting band, but
because of their lower concentration in the natural atmosphere, their
impacts on isoprene are significantly smaller than that of benzene and
toluene. Although NH
The absorption cross sections of benzene, toluene and isoprene
Benzene, toluene, or NO, SO
This paper introduces, for the first time, the continuous online measurement of isoprene in the atmosphere by means of DOAS in the band of 202.71–227.72 nm. Although the current measurements of isoprene mainly consist of GC-MS, PTR-MS and CIMS methods, the DOAS method has the characteristics of high time resolution, rapid temporal response and simple operation. It is especially suitable for long-term online measurement in fields or forests where the travel is inconvenient, and the low cost of instrument is also conducive to building monitoring networks.
Under the condition of zero optical path in the laboratory, several equivalent concentrations were measured by using series absorption cells and known concentrations of isoprene gas. The correlation coefficient between the measured concentrations and the equivalent concentrations was 0.9996, and the slope was 1.065, indicating that the instrument has good linearity and accuracy. After 23 d of field comparisons, there was a good correlation between the results of the DOAS and online VOC instrument, with a correlation coefficient of 0.85 and a slope of 0.86. Considering the differences in measurement principles and the sampled air, the comparison results show good agreement between these two instruments.
To evaluate the detection limit of the DOAS instrument under actual
atmospheric measurements, this study proposes to calculate the standard
deviation of all the data when the measured concentration of isoprene in the
ambient air is close to zero (
Data are available at
The study was designed by SG and BZ. Laboratory and field experiments were performed by YG, RZ and YY. Spectral analysis and data processing were done by BZ, JZ and CG. The paper was written by BZ, SW and SG, with contributions from all authors.
The authors declare that they have no conflict of interest.
We would like to thank the Shanghai Environmental Monitoring Center for supporting the online VOC analyzer measurement. We thank the National Key Research and Development Program of China and the National Natural Science Foundation of China for their financial support.
This research has been supported by the National Key Research and Development Program of China (grant nos. 2017YFC0210002, 2016YFC0200401) and the National Natural Science Foundation of China (grant nos. 21777026, 41775113, 21976031, 42075097).
This paper was edited by Jochen Stutz and reviewed by two anonymous referees.