1 Introduction

Combustion processes, e.g. coal burning in industry, constitute a major source of air pollution today. Nitric oxide (NO) and sulfur dioxide (SO2) can be converted into HNO3 and H2SO4, respectively, thus constituting the dominant precursors responsible for acid precipitation, which can cause serious damage to human health, agriculture and buildings. In order to reduce harmful exhaust emissions, China has promulgated/enforced a series of laws and standards such as the Emission Standard of Air Pollutants from Thermal Power Plants (GB13223-2003), the Emission Standard of Air Pollutants for Coal-burning Oil-burning Gas-fired Boilers (GB13271-2001), the Emission Standard of Air Pollutants for Cement Industry (GB4915-2004) and the Law on Prevention and Control of Atmospheric Pollution (http://www.stats.gov.cn/). These types of industrial facilities constitute important sources of air pollution and are required to comply with a set of emission regulations. While the SO2 concentration is governed by the sulfur content in the fuel, the NO concentration depends on the flame parameters and emissions and can be reduced by optimizing furnace design to avoid flame regions with high temperatures. Thereby monitoring of pollutant gas concentrations is of particular interest in these industries [1–5].

Currently, the optical monitoring techniques for NO and SO2 pollution emissions mostly include nondispersive infrared (NDIR) spectroscopy [6–10], laser-induced fluorescence (LIF) [11, 12] and tunable diode laser absorption spectroscopy (TDLAS) [13, 14]. In the NDIR and LIF techniques the concentrations are obtained based on light intensity change and not by spectral analysis, thus being easily influenced by variations of light source intensity. The TDLAS technique can achieve high measurement precision and accuracy; however, it requires expert handling and expensive equipment, which adds considerable operational complexity and cost, therefore being less suitable for, for example, thermal power plants in China. A potential solution to the above difficulties is to use a direct spectral absorption technique in the UV range to detect NO concentrations in the exhaust, since NO has strong absorption lines within this wavelength region. However, the NO concentration measurements are hampered by the influence of SO2 on the absorption spectrum. Sensors based on differential absorption spectroscopy in the ultraviolet (UV) wavelength region have previously been described, but the technique is grossly affected by the variation in the reference light intensity [15–21].

In this paper, we introduce a new data evaluation method for measurement of the NO concentration and elimination of the spectral influence of SO2. The NO concentration was calculated by the absorption peak at 226 nm, thereby minimizing the interference from SO2. The instrument based on this measurement technique is expected to have a low relative cost since the SO2 and NO concentrations can be simultaneously measured using cheap light sources and only one compact spectrometer.

2 Theoretical considerations

The Beer–Lambert law defines the relationship between absorbance and concentration in absorbing species. The expression between received and incident radiation intensities can be written as

(1)

where I(λ) is the received radiation intensity at the wavelength λ, I 0(λ) is the incident radiation intensity at the wavelength λ, \(\sigma_{\mathrm{SO}_{2}}(\lambda)\) (cm2/molecule) is the absorption cross section of SO2, σ NO(λ) (cm2/molecule) is the absorption cross section of NO, \(N_{\mathrm{SO}_{2}}\) (molecule/cm3) is the concentration of SO2, N NO (molecule/cm3) is the concentration of NO, L (cm) is the absorption path length, and α(λ) is the absorption coefficient of other gases.

Since the broadband gas absorption has little spectral structure, it is difficult to differentiate it from scattering in water vapor and particles. Therefore, this technique is only taking into account the narrowband structure, ignoring all broadband absorption. The absorption cross section is affected by pressure and temperature due to the collision broadening effect and the Doppler effect. However, since the present measurements are performed at atmospheric pressure and room temperature, the variation of the absorption cross section on account of these effects is negligible. The absorption cross section \(\sigma_{\mathrm{SO}_{2}}(\lambda)\) can be described as the sum of the broadband absorption cross section σ 1(λ) and the narrowband absorption cross section σ 2(λ), because SO2 has an undulatory absorption band. The gas concentrations can be deduced from Eq. (1):

(2)
(3)

where S(λ) is the polynomial-fitted slow-variation broadband intensity and can be written as

$$S(\lambda) = I_{0}(\lambda)\exp \bigl[ - \sigma_{1}(\lambda)NL -\alpha(\lambda)L\bigr].$$

The concentration of SO2 can be obtained by the method which was introduced in [22], because the absorption cross section σ NO(λ) of NO is equal to zero in the ultraviolet wavelength range from 277.4 to 311.5 nm. The concentration of NO is acquired using Eq. (3) in the ultraviolet wavelength range from 224.6 to 227.2 nm, yielding

$$ N_{\mathrm{NO}} = \frac{\mathit{OP}_{\mathrm{NO}}}{AL},$$
(4)

where the optical parameter of NO (OP NO) is the sum of all the absorption signals and A is a constant in this certain wavelength range. OP NO and A can be expressed as

(5)
(6)

In Eq. (5), OP NO expresses the difference between the integrated narrowband absorption spectrum of SO2 and NO and the narrowband absorption spectrum of SO2. The integral narrowband absorption spectrum of SO2 and NO can be calculated by the received radiation intensity from 215.0 to 239.6 nm. The narrowband absorption spectrum of SO2 from 215.0 to 239.6 nm can be calculated by the known σ 2(λ) and the SO2 concentration obtained in the wavelength range from 277.4 to 311.5 nm. When both the integral narrowband absorption spectrum of SO2 and NO and the narrowband absorption of SO2 are obtained by the received radiation intensity and the concentration of SO2, the absorption spectrum of NO can be calculated by subtracting the narrowband absorption of SO2 from the integral narrowband absorption spectrum of SO2 and NO from 215.0 to 239.6 nm.

3 Experimental

The experimental setup is shown in Fig. 1. The light source was a high-pressure deuterium lamp with the output power of 30 W. The lamp was mounted in the focal plane of a plano-convex quartz lens with a 75-mm focus in order to produce a narrow beam of light. The collimated light was transmitted across two gas cells equipped with quartz windows used for sampling the SO2 and NO concentrations. The length of both gas cells was 50 cm. The transmitted light was collected by another quartz lens and coupled into an optical fiber to be guided away from the sensing zone prone to disturbances. The light was then entered into a high-resolution spectrometer with a range from approximately 209 to 313 nm (resolution of 0.1 nm), which was composed of a monochromator and a 2048-element charge-coupled device array detector (Ocean Optics Inc. HR2000+). The data analysis and the control of the spectrometer were performed automatically in a personal computer using software written in Visual Basic. A bank of gas cylinders provided standard test gas with NO concentrations from 8 to 1000 ppm and SO2 concentrations of 308 and 450 ppm. Conventional piping was used to transport the gas into the sampling cells. The gas filling procedure in these experiments was as follows: the gas cells were first evacuated using a vacuum pump until the pressure was below 100 Pa; then the standard gas was injected into the cells at one atmosphere pressure so that the sample gas concentrations were consistent with the nominal standard concentrations in the cylinders. Since the gas signals resulted from the integration over an absorption path length, the mixing of SO2 and NO was performed artificially by adding the gases separately into the light beam, which at atmospheric pressure yields an identical result as they were mixed into the same cell.

Fig. 1
figure 1

Experimental setup for SO2 and NO measurements

Full size image

4 Results and discussion

Recorded single absorption spectra of SO2 and NO from 215.0 to 312.8 nm are shown in Figs. 2a and b, respectively. As can be seen, SO2 has two absorption bands in the wavelength ranges from 215 to 232 nm and from 272 to 313 nm, while NO has an isolated absorption peak with a center wavelength of 226 nm. In Fig. 2c a mixture of the two gases was sampled and the absorption cross sections of SO2 and NO overlap from 224.6 to 227.2 nm. Thus, when determining the NO concentration by spectroscopic evaluation, it is important to effectively eliminate the influence from SO2 on the spectrum.

Fig. 2
figure 2

(a) Transmission spectrum of SO2 from 215.0 to 312.8 nm. (b) Transmission spectrum of NO from 215.0 to 312.8 nm. (c) Transmission spectrum of a mixture of 450 ppm SO2 and 310 ppm NO from 215.0 to 312.8 nm

Full size image

According to [22], the linear relationship between the path-integrated gas concentration and the optical parameter for SO2 \((\mathit{OP}_{\mathrm{SO}_{2}})\) is given by

$$ C_{\mathrm{SO}_{2}}L = 4.56( \pm 0.03)\mathit{OP}_{\mathrm{SO}_{2}} -0.15,$$
(7)

where \(\mathit{OP}_{\mathrm{SO}_{2}}\) is the integrated absorption signal obtained from 277.4 to 311.5 nm and \(C_{\mathrm{SO}_{2}} L\) is the path-integrated SO2 concentration (in ppm × m units). Thus, the concentration of SO2 can be obtained directly by calculating the \(\mathit{OP}_{\mathrm{SO}_{2}}\) value of the sensor in this wavelength range.

To calculate the concentration of NO, firstly OP NO can be obtained according to the data evaluation procedure described in detail above. The data processing is illustrated in Fig. 3, where Fig. 3a exhibits how the spectral absorption data was first fitted by a polynomial function from 218.1 to 239.6 nm. Since the presence of NO would affect the polynomial fit for SO2, the intensity data from 224.6 to 227.2 nm was excluded from the fitting procedure. Secondly, the value of the natural logarithm for the ratio between the spectroscopic data and the fitted polynomial (compare with Eq. (2)) is given in Fig. 3b. This yields the narrowband absorption spectrum for the combined SO2 and NO signals. From the figure, it can be seen that the combined narrowband absorption spectrum in the region unaffected by NO is strongly consistent with the SO2 spectrum. Finally, the pure absorption spectrum of NO was obtained by removing the absorption value of SO2 from the combined absorption of SO2 and NO, as can be seen in Fig. 3c. The measured absorption spectrum of NO was largely coincident with the theoretical absorption spectrum of NO from 224.6 to 227.2 nm, as can be seen in the small residual value in Fig. 3d. The integrated signal from 224.6 to 227.2 nm is proportional to the concentration of NO.

Fig. 3
figure 3

(a) Transmission spectrum of SO2 and NO from 215.02 to 239.6 nm, together with a third-order polynomial fit (red dotted line). (b) Combined narrowband absorption of SO2 and NO, together with a fitted narrowband absorption spectrum of SO2. (c) Measured and theoretical narrowband absorption spectra of NO. (d) Residual of the measured and theoretical narrowband absorption spectra of NO

Full size image

To evaluate the relationship between the gas concentration and the optical parameter for NO, we measured a series of NO standard gas mixtures with concentrations from 8 to 1000 ppm. The dependence between OP NO and the gas concentration is shown in Fig. 4, where a calibration curve is fitted according to

$$ C_{\mathrm{NO}}L =267.9\mathrm{e}^{\mathit{OP}_{\mathrm{NO}}\mathrm{/20}.5} - 266.9,$$
(8)

where C NO L is the path-integrated NO concentration (in ppm × m units). At low concentrations, the fit becomes linear, as shown in the inset of Fig. 4. The square of the correlation coefficient R 2 between the fitted curve and recorded data is 0.999, which indicates a high degree of correlation.

Fig. 4
figure 4

The optical parameter versus the 30 different path-integrated concentrations of standard NO gas mixtures. Inset: low-concentration recording

Full size image

The performance of the sensor to evaluate NO concentrations was estimated with respect to measurement precision and detection limit. The precision was derived from the standard deviation of a series of concentration measurements. Figure 5 shows 290 successive concentration measurements recorded using an optical path length of 50 cm during a 10-min acquisition time. An average concentration of 465.6±1.5 ppm was obtained, which indicates a measurement precision of 0.3 %. The detection limit was gained by analyzing the statistical fluctuations of the signal and the root mean square (RMS) noise that is essentially independent of concentration. Eight different NO concentrations from 8 to 650 ppm were measured in the presence of 308 ppm SO2 and for each measurement 100 data samples were acquired at an interval of 2 s. The average RMS noise of all NO concentrations was found to be about 0.8 ppm. The relationship between the signal-to-noise ratio (SNR) and the NO concentration is shown in Fig. 6, where the sensitivity limit was estimated to be 1.5 ppm with SNR=2.

Fig. 5
figure 5

Plots of 290 successive concentration measurements recorded using an optical path length of 50 cm during a 10-min experiment time. An average concentration of 465 ppm was obtained

Full size image
Fig. 6
figure 6

The relationship between measured SNR and corresponding concentration of NO. The average of RMS noise for all NO concentrations is about 0.8 ppm

Full size image

The cross sensitivity to other gases should be considered for industrial applications. In fact, around 226 nm, the absorption cross sections and concentrations of most gases are much smaller than those of NO; thus, their interference with NO can be ignored. But SO2 has both strong absorption cross section and high concentration; hence, the influence of SO2 on NO must be duly considered. In this paper, an approach to eliminate the influence of SO2 on the NO concentration evaluation was introduced. In order to verify the validity of the method, the measurements of both 450 ppm SO2 and 351 ppm NO are presented in Fig. 3c. The optical parameters \(\mathit{OP}_{\mathrm{NO}}^{1}\) and \(\mathit{OP}_{\mathrm{NO}}^{2}\) denote the values in absence and presence of SO2, respectively, and are given by

As can be seen, the relative error between \(\mathit{OP}_{\mathrm{NO}}^{1}\) and \(\mathit{OP}_{\mathrm{NO}}^{2}\) is less than 1 %; moreover, the optical parameters were approaching zero when 450 ppm SO2 was measured in absence of NO, which clearly shows that the method can eliminate the influence of SO2 on NO. A series of standard test gases, which include 9.8, 44, 108, 183 and 485 ppm SO2 and 9.7, 45, 195 and 507 ppm NO, were measured successively in a cell and the results are shown in Fig. 7. It is clear that there was no change in the NO concentrations evaluated when SO2 was introduced, which shows that the sensor could successfully distinguish between the gases so that the accurate concentration of the gases present was given.

Fig. 7
figure 7

Averaged concentration from the sensor for a series of standard test gases in the cell in turn

Full size image

A field measurement was carried out in the winter of 2008, at a thermal power plant located in Weihai in Shandong Province, China. The sensor was tested for simultaneous measurements of SO2 and NO in flue gas emitted from an industrial coal-fired boiler. A continuous 15-h recording of the SO2 and NO is shown in Figs. 8a and b. From the figure, we estimated the average SO2 and NO concentrations to be 301 and 112 ppm, respectively. This field measurement demonstrates that the sensor based on differential absorption spectroscopy has the capacity to accurately monitor the concentrations of SO2 and NO from power plants.

Fig. 8
figure 8

Fifteen hours recording of the NO and SO2 concentrations emitted from a thermal power plant in Eastern China. The average concentrations of NO and SO2 were 112 and 301 ppm, respectively

Full size image

5 Conclusions

A novel method for the measurement of NO by the spectral separation technique based on differential absorption spectroscopy has been described. The concentrations of SO2 and NO were measured within two absorption bands: 277.4 to 311.5 nm and a region around the NO absorption peak from 224.6 to 227.2 nm, respectively, using a single spectrometer. Cross sensitivity between the gases is a potential issue due to their overlapping absorption spectra. However, the influence of SO2 on the NO absorption was successfully eliminated in the ultraviolet spectral range using the proposed data evaluation method. The calibration expression of NO was obtained using standard gases of NO. The measurement precision of the system was evaluated to be about 0.3 %. The detection limit for NO was 0.8 ppm with SNR of 2 per meter of path length, as derived from the root mean square noise. Field measurements were conducted on a thermal power plant, and the results show that this sensor technology is suitable for industrial NO and SO2 emission monitoring. With better spectral resolution of the spectrometer used, the SNR and the precision of the method may be further improved. This study demonstrates that our detection technique should be promising for industrial NO and SO2 emission monitoring.