Preparation Methods to Optimize the Performance of Sensor Discs for

Article pubs.acs.org/est

Preparation Methods to Optimize the Performance of Sensor Discs for Fast Chemiluminescence Ozone Analyzers M. Ermel,*,† R. Oswald,† J.-C. Mayer,† A. Moravek,† G. Song,† M. Beck,‡ F. X. Meixner,† and I. Trebs† †

Biogeochemistry Department, Max Planck Institute for Chemistry, P.O. Box 3060, 55020 Mainz, Germany Institute of Inorganic and Analytical Chemistry, Johannes Gutenberg-University of Mainz, Duesbergweg 10-14, 55128 Mainz, Germany

‡

S Supporting Information *

ABSTRACT: Fast ozone (O3) measurements (1−50 Hz) in the atmosphere are required for airborne studies and for the measurement of ground-based O3 fluxes by the eddy covariance technique. Fast response analyzers, based on heterogeneous chemiluminescence, need dye coated sensor discs on which the chemiluminescence is generated. In this study, we present three new preparation methods for those sensor discs. Currently available sensor discs exhibit a fast temporal decay of sensitivity, resulting in short duty times which is troublesome for many field applications. To produce sensor discs that provide more stable signals over time, three dyes and nine energy transfer reagents were tested (as well as different stoichiometric mixtures). The resulting optimal method saves 80% of the solid chemicals and shows a duty ozone dose that is prolonged by a factor of 3.5, revealing the same average sensitivity as currently available discs. In addition, we observed a strong effect of the adsorption matrix on the O3 sensitivity, although silica discs from the same manufacturer were used. Application of the new sensor discs during field measurements showed that the results are consistent with the laboratory data.

1. INTRODUCTION Ozone (O3) is a very important trace gas in the stratosphere, protecting the biosphere from harmful ultraviolet radiation. In the troposphere it is the major source of the hydroxyl radical (OH), which is the primary oxidant in the atmosphere.1 O3 is mainly produced by photochemical reactions involving nitrogen oxides (NOx) and volatile organic compounds (VOC).2 Besides also being regarded as an important greenhouse gas,3 O3 is a harmful pollutant in the troposphere due to its oxidative capacities. It has adverse effects on human health,4,5 plant functioning,6,7 and many surfaces, e.g., stones.8 Hence, the investigation of O3 formation and deposition processes is crucial to establish the atmospheric budget of this trace gas. Several field studies were performed recently to determine the dry deposition of O3 to ecosystems and to investigate the underlying uptake processes of O3 by plant cuticle and stomata.9−13 The state-of-the art method to measure areaintegrated (plot/stand scale) turbulent O3 fluxes is the eddy covariance (EC) method (e.g., ref 14). This is a micrometeorological method which correlates high-frequency fluctuations of vertical wind speed with those of the O3 mixing ratio. Since vertical wind speed fluctuations relevant to turbulent transfer range between 1 and 50 Hz, measurements of the O3 mixing ratio have to cover the same frequency domain. Typically, gas phase homogeneous15 and heterogeneous16,17 dry chemiluminescence techniques feature suitable response times for the EC method. Another application of these fast © 2013 American Chemical Society

chemiluminescence analyzers is airborne measurements of O3 mixing ratio.18,19 Particularly those on unmanned small aircrafts, smart or tethered balloons, or high performance kites have crucial demands on small size, light weight, and low power consumption of the corresponding sensor (e.g., ref 20). Heterogeneous solid phase chemiluminescence O3 sensors fulfill these demands. The dry chemiluminescence technique uses organic dyes, which are fixed on an adsorptive matrix (sensor disc). The reaction of the dye with O3 emits light, which correlates with the O3 mixing ratio. The first dry chemiluminescence sensor used luminol adsorbed on a silica gel disc21 to detect O3. Since this reaction needs water to take place, more recent studies used sensor discs with Coumarin 4716,22,23 and Coumarin 30724 in combination with gallic acid. The latter is used as antioxidant and energy transfer reagent according to the Förster-resonant energy transfer mechanism.25,26 O3 reacts with gallic acid and the excited reaction products transfer the energy to the dye, which then fluoresces: gallic acid + O3 → product* Received: Revised: Accepted: Published: 1930

(1)

October 4, 2012 December 9, 2012 January 23, 2013 January 23, 2013 dx.doi.org/10.1021/es3040363 | Environ. Sci. Technol. 2013, 47, 1930−1936

Environmental Science & Technology

Article

Table 1. Dyes and Energy Transfer Reagents (ETR) Used for the Desorption Method (DM) and the Efficient Adsorption Method (EAM)a dye ETR

DM

EAM1

EAM2

EAM3

EAM4

EAM5

EAM6

EAM7

EAM8

BAGUS

C47 C152 THB 2,4DHB

C47 C152 THB 2,4DHB

C47 THB

C47 2,5DMB

C47 VANA

C152A THB

C152A 2,4DHB

C152A 2,5DMB

C47 C152A THB

C47 THB

a

The dyes are Coumarin 47 (C47), Coumarin 152 (C152), and Coumarin (152A). The ETRs are gallic acid (THB), 2,4-dihydroxybenzoic acid (2,4DHB), 2,5-dimethoxybenzoic acid (2,5DMB), and vanillic acid (VANA).

product* + dye → product + dye + hν

2.2. Preparation of Sensor Discs. Coumarin 47 (C47, 99% purity) was purchased from Sigma Aldrich (Steinheim, Germany). Coumarin 152 (C152) and Coumarin 152A (C152A) were obtained from Radiant Dyes Laser (Wermelskirchen, Germany). The energy transfer reagents (ETR) gallic acid (THB), caffeic acid (CAFFA, ≥ 98.0% purity), and syringic acid (SYRA, ≥ 95.0% purity) were purchased from Sigma. 3,4,5-Trimethoxycinnamic acid (TMC, 97% purity), 2,5dimethoxybenzoic acid (2,5DMB, 98% purity), and 2,5dihydroxybenzoic acid (2,5DHB, 98% puritiy) were obtained from Aldrich. Vanillic acid (VANA, ≥ 97.0% purity) and bis(2,4,6-trichlorophenyl)oxalate (C3PO, ≥ 99.0% purity) were purchased from Fluka (Steinheim, Germany). 2,4-Dihydroxybenzoic acid (2,4DHB, 99% purity) was ordered from Acros Organics (Geel, Belgium). The liquids acetone (99.8% purity), dichloromethane (99.5% purity), and Rotisilon A were purchased from Carl Roth (Karlsruhe, Germany). For the support matrix of the discs thin layer chromatography plates (RP2-TLC) of the type Alugram RP-2/UV from MachereyNagel (Düren, Germany) made of silica were used. 2.2.1. Oswald Method (ROS). We prepared a mixture of 7.5 mL of acetone with 127.7 mg (0.55 mmol) of C47, 141.5 mg (0.55 mmol) of C152, 187.1 mg (1.1 mmol) of THB, and 175.9 mg (1.1 mmol) of 2,4DHB. A solution of 10% by volume Rotisilon A in dichloromethane was added and a 4 × 5.5 cm2 RP2-TLC disc was immersed in the solution for 5 min. Subsequently, the disc was vacuum-dried at 60 °C for 60 min. 2.2.2. Desorption Method (DM). We prepared a mixture of 7.5 mL of acetone with 127.7 mg (0.55 mmol) of C47, 141.5 mg (0.55 mmol) of C152, 187.1 mg (1.1 mmol) of THB, and 175.9 mg (1.1 mmol) of 2,4DHB. A 4 × 5.5 cm2 RP2-TLC disc was immersed in the solution for 5 min. Subsequently the disc was vacuum-dried at 60 °C for 60 min. Afterward the disc was immersed in 7.5 mL of 10% by volume Rotisilon A in dichloromethane for 3.5 min. The disc was dried again at the same conditions. 2.2.3. Efficient Adsorption Method (EAM). EAM was used with different dyes and energy transfer reagents (see Table 1) and here, a general preparation procedure is provided. A maximum of two substances of each type of compounds was used on a single sensor disc. We dissolved 0.11 mmol of each dye and 0.22 mmol of each energy transfer reagent in a mixture of 7.5 mL of acetone and 2.5 mL of 10% by volume Rotisilon A in dichloromethane. A 4 × 4 cm2 RP2-TLC disc was immersed in this solution for only 10 s. Subsequently the disc was vacuum-dried at 60 °C for 60 min. 2.3. Field Measurements. To test and compare the sensor discs under field conditions, we made highly time-resolved measurements of the O3 mixing ratio (20 Hz) at a nutrient poor steppe-like grassland ecosystem in Rhine Hessen (Mainz, Germany) from Sep 15, 2011 until Oct 21, 2011. Two fast response O3 detectors (Enviscope GmbH, Germany) were set up side-by-side with the inlets mounted at a height of 3 m

(2)

An enhancement of the sensitivity may occur due to interferences of water and sulfur dioxide (SO2). However, they do not affect the high frequency O3 measurements, as the response times are 22 s and 30 min, respectively.16,27 Generally, this method requires continuous calibration due to unstable sensitivities over time.28 The sensor discs prepared according to the method of Speuser 22 are currently commercially available (BAGUS Consulting GmbH, Speyer, Germany), but exhibit a fast decay of the sensitivity within 1−2 days, depending on the ambient O3 mixing ratio. We also found that the loss of sensitivity is often unpredictable since it varies from disc to disc, which is most likely a consequence of the varying composition of the adsorbed substances on the surface. In this study we will present three new preparation methods of sensor discs needed for the dry chemiluminescence technique. The discs were optimized by (i) varying parameters during the preparation procedures and (ii) combining different dyes with energy transfer reagents (ETR). The optimized discs exhibit a temporally more constant, and also higher, sensitivity. Consequently, this will allow more reliable measurements over long periods without frequently changing the discs. Furthermore, a higher sensitivity over time will result in data with a better signal-to-noise ratio.

2. EXPERIMENTAL SECTION 2.1. Measurement Setup. Measurements for optimizing the sensor discs were made using dried and cleaned compressed air (see Supporting Information (SI) for a schematic setup). The air first entered a membrane dryer combined with a filter for compressed air (Clearpoint and Drypoint M from BEKO Deutschland GmbH, Germany). Compressed air was then cleaned from trace gases (O3, NOx, hydrocarbons) by a homebuilt gas cleaning station which contained glass wool, silica gel, activated charcoal, and molecular sieves. This device is described in more detail by Yang and Meixner.29 Data were acquired by a CR3000 Datalogger (Campbell Scientific, Inc., USA) on a 10 Hz basis. The air flow of 20 L/min through the system was regulated by a mass flow controller (20 L/min, MKS Instruments, Munich, Germany). The overflow provided ambient pressure during the measurements. O3 was generated by an O3 generator based on a mercury lamp (OG-1, Ultra-Violet Products Ltd., USA) and measured by an O3 analyzer based on UV absorption (49C, Thermo Environmental Instruments Inc., USA). The latter was used to calibrate the two fast response O3 detectors (Enviscope GmbH, Germany), which were used to measure the sensitivity of the sensor discs. They were based on dry chemiluminescence to measure O3 at a frequency of up to 50 Hz. It used a photomultiplier to count the photons emitted by the sensor disc. The instrument was previously described and characterized in detail by Zahn, et al.30 and has also been used for EC measurements in the field. 1931

dx.doi.org/10.1021/es3040363 | Environ. Sci. Technol. 2013, 47, 1930−1936


Article

above ground. Simultaneously, O3 mixing ratios were measured with a time resolution of 30 s by a UV absorption analyzer (49C, Thermo Environmental Instruments Inc., USA) at a height of 4 m above ground to calibrate the fast response O3 detector (see ref 28). To ensure comparability between the two different fast response O3 detectors, we operated them with EAM8 discs from the same production batch from Sep 15 to 30, 2011. Periods with temporally constant O3 mixing ratios during this period were used to calculate a proportionality factor. Such conditions are required to ensure that changes in the O3 mixing ratio on larger time scales (minutes to hours) do not affect the calculation. The proportionality factor corrects for systematically different sensitivities of the photomultipliers in the fast response O3 detectors. This allows a subsequent comparison of the sensitivities from two different detectors when using different sensor discs (e.g., BAGUS versus EAM8, see below). The background noise of both analyzers was determined by closing inlet and outlet tubes while sensor discs were installed. The remaining background signal after O3 reacted completely with the chemicals was subtracted from the highly timeresolved O3 data. From Sep 30 to Oct 21, 2011 one fast response O3 detector was equipped with the commercially available discs (BAGUS Consulting GmbH) and the other O3 detector was equipped with discs prepared using the EAM8. Both disc types were exposed to elevated O3 mixing ratios using an ozonizer (GFAS Ozonizer PR-S-2, GFAS GmbH, Germany) prior to each individual application in the field for 1 h (BAGUS) and for 5 h (EAM8), respectively (for further explanation see below).

Figure 1. Sensitivity of sensor discs prepared by the desorption method (DM) with different batches of the RP2-TLC discs as a function of the O3 dose.

disc prepared with batch 2 shows a very flat and constant performance, whereas the use of batch 3 shows an optimum curve comparable to batch 1, although the slope is steeper. The manufacturing process of the RP2-TLC discs did not change among the three batches (Macherey-Nagel, Ms. Janiel, personal communication). However the particle size of the RP2-TLC discs is specified by the manufacturer as 5−17 μm, hence the resulting effective surface area can vary depending on the processed particles. The amount of dyes and energy transfer reagents in the solution was the same for all three batches. Consequently, the effective surface area has a large impact on the sensor disc properties, because it influences the number of adsorbate layers. As outlined in Section 3.2 the layers have a significant influence on the sensitivity of the sensor discs. Furthermore, TLC discs with large effective surface area can achieve higher adsorbate occupancies and, thus, they can be exposed to O3 longer before the sensitivity declines. 3.2. Comparison of the Preparation Methods. The sensitivity of the different sensor discs during O3 exposition is shown in Figure 2. To validate the performance of the newly developed preparation procedures, the sensitivity of the BAGUS discs is also shown. After a fast sensitivity increase a high maximum of 13.8 ± 5.9 mV/ppb at an ozone dose of 131.6 ppb·h is reached. The following decrease quickly yields low sensitivities. For comparison, the sensor discs by the Oswald method show a lower, but long lasting sensitivity at

3. RESULTS AND DISCUSSION In this section, the sensitivities of the sensor discs instead of the signal intensity are shown and discussed. Using the sensitivity of the discs (in mV/ppb) rather than the signal intensity itself (mV) has the advantage that the bias of temporal intensity fluctuations of the mercury lamp of the O3 generator is not considered. For the same reason the O3 dose, which is the product of the O3 mixing ratio and the exposition time, is used instead of the time of O3 exposure. Measured data were averaged to 5-min time intervals. Since the properties of the discs can vary within one production batch (see below), at least three replicates of each sensor disc were measured. To ensure comparability, always the sensor disc with the highest sensitivity of one preparation method is shown. To compare the performance of the sensor discs the following quantities are used: maximum sensitivity (Smax, mV/ ppb), average duty sensitivity (Savg, mV/ppb), and the duty ozone dose (Davg, ppb·h). The duty range is defined as the range with a sensitivity higher than 3 mV/ppb. This limit was chosen based on the duty time given by several authors17,28 for the BAGUS sensor discs prepared by the method of Speuser.22 The presented errors are standard deviations of three replicate measurements of the corresponding quantity. 3.1. Influence of the TLC Disc Batch. Three different batches of RP2-TLC discs were used during this study. Figure 1 shows the sensitivity of the sensor discs prepared by the DM as a function of the ozone dose. The change to batch 2 resulted in a very low sensitivity of only 18% compared to batch 1. Furthermore, the sensitivity of batch 2 never exceeded the threshold for the duty range of 3 mV/ppb. The use of batch 3 decreased the sensitivity to only 71% of batch 1. Besides the sensitivity, obviously the shape of the curve changed also. The

Figure 2. Sensitivity of sensor discs prepared by the Oswald method (ROS), the desorption method (DM), and the efficient adsorption method with energy transfer system 1 (EAM1) and 8 (EAM8). Sensor discs ROS, DM, and EAM1 use the same energy transfer system. EAM1 (hypoth.) and EAM8 (hypoth.) consider the impact of using silica discs from different batches. 1932



Article

relatively constant values of 1.5 ± 0.1 mV/ppb, which never exceeds the duty range threshold. The amount of adsorbate on the disc surface (Table 3) shows significant differences between the two types of discs. The BAGUS discs have only 5.3% of the total mass of the discs made by the Oswald method adsorbed. This very high surface occupancy explains the high duty time of the Oswald discs, as more reactants are available for O3. Miet, et al.31 have shown for the reaction of O3 with pyrene, which was adsorbed on silica particles, that the reaction rate depends on the solid reactant concentration on the disc surface. The rate declines with increasing concentrations of the adsorbed O3 reaction partners, which leads to a decreased sensitivity. Furthermore, not only is the surface concentration important, but a crucial factor is the development of mono or multi layers of adsorbate, with the mono layers revealing higher reaction rates. According to these findings the highest sensitivity and the highest duty ozone dose can be achieved by a fully developed mono layer (without forming additional layers). The BAGUS and Oswald discs show both extremes resulting in either a high sensitivity or a high ozone dose. These findings motivated us to develop the desorption method, which includes an additional desorption phase to reduce the surface density of the adsorbate and, thus, to reduce the amount of layers as well. Figure 2 shows that the maximum sensitivity increased to 7.3 ± 0.3 mV/ppb and the average duty ozone dose is 5103.5 ± 886.2 ppb·h (Table 2). Compared to

the BAGUS discs, 74.1% of the average duty sensitivity and 433% of the duty time was achieved. These results are consistent with the findings of Miet, et al.,31 which implies that less adsorbate on the surface yields a higher reaction rate. However, the desorption step during the preparation yields a horizontal concentration gradient on the silica plates. This is caused by the presence of a thin solution layer (3−4 mm) only above the plates and a large “reservoir” of solution at the margin of the plates, which can take up more adsorbate. Consequently, when sensor discs with sizes suitable for the O3 analyzer are punched out of a single silica plate the discs may have varying sensitivities and duty ozone doses. To avoid this problem, we developed the efficient adsorption method, where the silica plate is only immersed in the adsorption solution for 10 s. Hence, only a small amount of adsorbate (Table 3) was found on the surface. The maximum sensitivity of disc EAM1, which uses the same energy transfer system as the two discs discussed before, reaches 6.5 ± 0.3 mV/ ppb and average duty ozone dose of 3003.3 ± 106.1 ppb·h (Table 2). Since a silica plate from batch 3 was used, Figure 2 also shows a hypothetical sensitivity of disc EAM1 calculated from the sensitivity differences between silica plates of batch 1 and batch 3 (see Figure 1). This reveals that the EAM yields the best results and the sensitivity is higher than for the DM using silica plates from batch 1. The correction of the sensitivity for different batches is a simple approach to compare all methods, which is most likely erroneous, as the different substrate has a substantial impact on the chemistry of the energy transfer system. Furthermore, the EAM saves 80% of the solid chemicals compared to the former methods. All EAM discs show smaller standard deviations (see Table 2) than the BAGUS sensor discs. This is of significant importance to guarantee a reliable performance during field measurements. The efficient adsorption method showed the best results, reaching 82% (hypothetical 103%, when using a better silica plate as adsorption matrix) of the average duty sensitivity compared to the BAGUS sensor discs and a prolonged duty ozone dose by a factor of 2.8 (hypothetical 3.5). 3.3. Energy Transfer Systems. To improve the performance of the sensor discs, several dyes and energy transfer reagents were tested for their fluorescence properties. Figure 3 shows the emission wavelengths of different energy transfer reagents on the excitation−emission matrix (EEM) of C152A. This was also tested for the dyes listed in the Experimental Section. According to the theory of the Förster-resonance energy transfer, we were looking for energy transfer reagents whose emission wavelength overlaps with the excitation wavelength of the dye. Furthermore, the intensity of the

Table 2. Maximum Sensitivity (Smax), Average Duty Sensitivity (Savg), and Duty Ozone Dose (Davg) of Sensor Discs Prepared by the Desorption Method (DM) and the Efficient Adsorption Method (EAM)a namebatch DM1 DM2 DM3 EAM13 EAM23 EAM33 EAM43 EAM53 EAM63 EAM73 EAM83 BAGUS

Smax, mV/ppb

Savg, mV/ppb

Davg, ppb·h

± ± ± ± ± ± ± ± ± ± ± ±

5.5 ± 0.2

5103.5 ± 886.2

7.3 1.4 5.0 6.5 5.7 5.8 5.4 7.7 6.8 8.4 7.9 13.8

0.3 0.1 0.4 0.3 0.1 0.1 0.6 0.1 0.2 0.4 0.0 5.9

3.1 5.3 4.5 4.7 4.5 6.1 5.6 6.4 6.1 7.4

± ± ± ± ± ± ± ± ± ±

0.3 0.2 0.0 0.1 0.3 0.1 0.1 0.1 0.1 2.0

4283.1 3003.3 1130.0 957.3 830.9 2926.2 1735.7 1755.3 3297.5 1177.7

± ± ± ± ± ± ± ± ± ±

551.3 106.1 51.5 123.3 0.3 302.6 46.4 148.2 176.1 472.1

a

Subscripts indicate the batch of the silica discs, which yield the corresponding sensitivity. Note that Savg and Davg cannot be determined for DM2, as the sensitivity was always below 3 mV/ppb. The errors are standard deviations of three replicates.

Table 3. Concentration of Energy Transfer Reagents and Dyes Adsorbed on the Sensor Discs (Given in μg/mg Silica Gel) Determined by HPLC-ESI-MS (see SI)a ROS BAGUS EAM1 EAM8-A EAM8-B EAM8-C a

THB

2,4DHB

± ± ± ± ± ±

18.0 ± 3.4

39.0 2.05 9.33 7.86 7.46 3.68

2.3 0.19 0.64 0.58 0.57 0.40

5.85 ± 1.42

C47

C152

± ± ± ± ± ±

14.1 ± 1.3

12.4 2.35 4.60 5.98 3.29 0.40

1.8 0.52 0.90 1.13 0.87 0.71

C152A

5.02 ± 0.47 6.50 ± 1.20 5.44 ± 1.11 1.05 ± 0.77

total 83.5 4.40 24.80 20.34 16.19 5.13

± ± ± ± ± ±

8.8 0.71 5.49 2.91 2.55 1.82

EAM8 was exposed to different ozone doses (see text for details). Errors are calculated from the errors of the calibration function. 1933



Article

reaction partner. These results yield important insights to the reactions taking place on the sensor discs. The expected mechanism for the surface reaction22,24 is only valid after some time of O3 exposure. The direct reaction of ozone with C47 is the most important step at the beginning. The preparation procedure described in Section 2.2.3 is the result of optimizing (i) the ratio of the dyes to THB, (ii) the ratio of C47 to C152A, and (iii) the concentration of Rotisilon A. The Rotisilon A is used to obtain a more hydrophobic surface, which increases the ozone dose.23 Under these conditions the EAM8 shows an average duty sensitivity of 82% (hypoth. 103%) and a prolonged duty ozone dose by a factor of 2.8 (hypoth. 3.5) compared to the BAGUS discs. The change of sensitivity over time for the EAM8 discs is much less than for the BAGUS discs, which means that the accuracy of the calculated eddy covariance fluxes is improved. These are major improvements compared to currently available sensor discs according to the method of Speuser,22 which feature a high maximum sensitivity, but a fast decay over time combined with a highly variable performance from disc to disc. 3.4. Application in the Field. As described in ref 28, sensor discs were “pre-ozonized” prior to their application in the field, (i.e., they were exposed to enhanced O3 mixing ratios for 1 h). This treatment ensures maximum sensitivity at the start of the measurement. Moreover, it avoids the strong increase of the sensitivity of the BAGUS discs (see Figure 2), which usually causes high uncertainties in the O3 mixing ratio measurements. Due to the smaller slope of the sensitivity increase of the EAM8 discs, these were preozonized for 5 h. The results from the intercomparison of the EAM8 discs from Sep 15 to 30 are shown in Figures 4 and 5. Obviously, the

Figure 3. Excitation−emission matrix of Coumarin 152A, with emission wavelengths of different energy transfer reagents. Abbreviations are explained in the Experimental Section.

fluorescence correlates with the energy transfer25,26 and, thus, should be high. The energy transfer reagents with the best properties for C47 are THB, VANA, and 2,5DMB (not shown). The properties of the sensor discs (EAM2−EAM4) prepared with these energy transfer systems are shown in Table 2. The system of C47 and THB shows the best results, as its duty ozone dose is the longest although it does not feature the highest maximum and average duty sensitivity. The best reagents to form an energy transfer system for C152A are THB, 2,5DMB, and 2,4DHB (Figure 3). Accordingly, disc EAM7 has a very high maximum sensitivity, but only a slightly higher average duty sensitivity compared to EAM5. However, the duty ozone dose of EAM7 is only 60% of EAM5. It is important to note that the EEM of EAM5 (not shown) shows a peak of THB, thus the energy transfer is not or only incompletely working. To complete the energy transfer from THB to C152A, C47 was added (EAM8), as its absorption and emission properties fit in the gap between THB and C152A. This yielded higher maximum and average sensitivities as well as a longer average duty ozone dose (Table 2). To improve the understanding of the processes taking place on the sensor disc, the adsorbate concentrations on sensor discs exposed to different ozone doses were investigated (see Table 3). The disc EAM8-A was not exposed to ozone, EAM8-B was exposed to 562.5 ppb·h, and EAM8-C was exposed to 2043.9 ppb·h. Table 3 illustrates that the adsorbate concentration decreased depending on the ozone dose. In contrast to the assumption of the role of THB as an antioxidant,22,24 the Coumarin dyes react first. After applying an O3 dose of 562.5 ppb·h, about 45% of C47 and 15% of C152A are decomposed, whereas only 5% of THB were consumed. The faster loss of C47 compared to C152A can be explained by the different chemical structure of the compounds. The electrophilic ozone prefers the more electron rich double bond of C47. This period of dye decrease represents the fast increase of sensitivity after the start of the O3 exposition (Figures 1 and 2). EAM8C at 2043 ppb·h was sampled shortly after the maximum sensitivity. The concentration of the dyes had decreased drastically to 7% and 16% of the initial concentrations, whereas 46% of THB was left. In this second period, the reaction rate of THB with O3 is larger than for the dyes, probably due to the concentration differences and hence accompanied by the higher probability of O3 to find THB as

Figure 4. Correlation of the sensitivities measured during the side-byside comparison of the two fast response O3 detectors (2-h averages). Error bars denote standard deviations. The black line represents the 1:1 ratio and the red line is the result of the bivariate regression according to Cantrell32 (y = 1.29x − 1.95 mV/ppb, R2 = 0.937, N = 118, error: m = 0.04, b = 0.29 mV/ppb, significance 99.99% (for m and b)).

EAM8 discs, as well as the sensors, exhibited only a small deviation of the sensitivity from the 1:1 line (see Figure 4). A bivariate regression according to Cantrell32 was applied yielding a slope of 1.29. According to the t test, this slope is “highly significant”. The difference of 29% can be attributed either to the fast response O3 detectors or to the sensor discs. However, even if this discrepancy were fully attributed to the instruments, the measured difference between the sensitivities of BAGUS and EAM8 sensor discs is significantly larger (see below). 1934



Article

results found in the laboratory. The new type of sensor discs exhibit a significantly higher duty ozone dose and the decay of the sensitivity is comparable for different discs. Future studies to improve the dry chemiluminescence method should focus on the surface properties of the TLC silica discs that are used as adsorption matrix.

■

ASSOCIATED CONTENT

S Supporting Information *

Details on the measurement setup, the fluorescence spectrometry, and the HPLC-ESI-IT/MS; a figure illustrating the temporal behavior of BAGUS-1, 2, and EAM8-5 over one week during the field measurement. This information is available free of charge via the Internet at http://pubs.acs.org/

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Tel.: +4961313056406. Fax: +4961313056405. Notes

The authors declare no competing financial interest.

■

Figure 5. Two-hourly averaged sensitivities of the sensor discs EAM81−EAM8-7 and BAGUS-1−BAGUS-6 used during the field measurements at the grassland site. Error bars denote standard deviations.

ACKNOWLEDGMENTS We gratefully acknowledge financial support by the Max Planck Society. We are indebted to C. Pöhlker and U. Pöschl for allowing us to use the fluorescence spectrometer.

During the second part of the experiment three EAM8 discs (EAM8 5−7, see Figure 5) and six BAGUS sensor discs (BAGUS 1−6, see Figure 5) were used. A figure illustrating the temporal behavior of the sensor discs during the field measurements is supplied in the Supporting Information. The EAM8 discs were exchanged after 1 week, while the BAGUS discs were exchanged after 2−5 days, depending on the signal sensitivity. The temperature ranged from 23 °C (daytime) to −4 °C (nighttime) and the relative humidity ranged from 38% to 100% during the measurements. The mixing ratio of ozone varied between 61 ppb during daytime and 1 ppb during nighttime. The corresponding results are presented in Figure 5, showing 2-h averages of the sensitivities. The discs EAM8 1−5 and EAM8 6−7 were each from the same preparation batch and exhibited a uniform behavior of the sensitivity within their batch, although they were exposed to different environmental conditions. In addition, the sensitivity differences between the two preparation batches were small and the sensitivity was relatively constant over time. In contrast, the BAGUS discs 1−6 show very inhomogeneous properties, with large differences of the sensitivity maxima, as well as in the duty ozone dose (Figure 5c). The maximum sensitivity varies over more than 1 order of magnitude from 2.3 to 34.9 mV/ppb. These findings are consistent with the results from the laboratory experiments (Figure 2). Half of the BAGUS discs (2−4) show a lower sensitivity than the EAM8 discs at the start of each measurement. BAGUS-2 and -3 were below the sensitivity threshold of 3 mV/ppb throughout their operation. The large error bars (standard deviations, e.g., EAM8-1 at 2440 ppb·h) can be explained by very low O3 mixing ratios during this period, which leads to higher signal-to-noise ratios. The BAGUS sensor discs show large standard deviations at their maximum sensitivity (e.g., BAGUS-1 and 6), which is related to the fast decay of the sensitivity. Beside this, both types of sensor discs, BAGUS and EAM8, showed low standard deviations of the sensitivity for all other periods. The comparison between the BAGUS sensor discs and the EAM8 sensor discs during field measurements confirms the

■

REFERENCES

(1) Levy, H. Normal Atmosphere - Large Radical and Formaldehyde Concentrations Predicted. Science 1971, 173 (3992), 141−143. (2) Crutzen, P. J.; Lelieveld, J. Human impacts on atmospheric chemisty. Annu. Rev. Earth. Planet. Sci. 2001, 29, 17−45. (3) Solomon, S. Intergovernmental Panel on Climate Change. Working Group I. Climate Change 2007: The Physical Science Basis: Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change; Cambridge University Press: Cambridge; New York, 2007; p viii, 996 pp. (4) Uysal, N.; Schapira, R. M. Effects of ozone on lung function and lung diseases. Curr. Opin. Pulm. Med. 2003, 9 (2), 144−150. (5) Levy, J. I.; Chemerynski, S. M.; Sarnat, J. A. Ozone exposure and mortality - An empiric Bayes metaregression analysis. Epidemiology 2005, 16 (4), 458−468. (6) Ainsworth, E. A. Rice production in a changing climate: A metaanalysis of responses to elevated carbon dioxide and elevated ozone concentration. Global Change Biol. 2008, 14 (7), 1642−1650. (7) Paoletti, E. Ozone slows stomatal response to light and leaf wounding in a Mediterranean evergreen broadleaf, Arbutus unedo. Environ. Pollut. 2005, 134 (3), 439−445. (8) Massey, S. W. The effects of ozone and NOx on the deterioration of calcareous stone. Sci. Total Environ. 1999, 227 (2−3), 109−121. (9) Lamaud, E.; Loubet, B.; Irvine, M.; Stella, P.; Personne, E.; Cellier, P. Partitioning of ozone deposition over a developed maize crop between stomatal and non-stomatal uptakes, using eddycovariance flux measurements and modelling. Agric. Forest. Meteorol. 2009, 149 (9), 1385−1396. (10) Stella, P.; Personne, E.; Loubet, B.; Lamaud, E.; Ceschia, E.; Beziat, P.; Bonnefond, J. M.; Irvine, M.; Keravec, P.; Mascher, N.; Cellier, P. Predicting and partitioning ozone fluxes to maize crops from sowing to harvest: the Surfatm-O-3 model. Biogeosciences 2011, 8 (10), 2869−2886. (11) Dorsey, J. R.; Duyzer, J. H.; Gallagher, M. W.; Coe, H.; Pilegaard, K.; Weststrate, J. H.; Jensen, N. O.; Walton, S. Oxidized nitrogen and ozone interaction with forests. I: Experimental observations and analysis of exchange with Douglas fir. Q. J. R. Meteorol. Soc. 2004, 130 (600), 1941−1955. 1935



Article

(32) Cantrell, C. A. Technical Note: Review of methods for linear least-squares fitting of data and application to atmospheric chemistry problems. Atmos. Chem. Phys. 2008, 8 (17), 5477−5487.

(12) Rummel, U.; Ammann, C.; Kirkman, G. A.; Moura, M. A. L.; Foken, T.; Andreae, M. O.; Meixner, F. X. Seasonal variation of ozone deposition to a tropical rain forest in southwest Amazonia. Atmos. Chem. Phys. 2007, 7 (20), 5415−5435. (13) Klemm, O. M. A. Ozone deposition at a forest site in NE Bavaria. Water, Air, Soil Pollut.: Focus 2001, 1, 223−232. (14) Güsten, H.; Heinrich, G.; Mönnich, E.; Sprung, D.; Weppner, J.; Ramadan, A. B.; Eldin, M. R. M. E.; Ahmed, D. M.; Hassans, G. K. Y. On-line measurements of ozone surface fluxes 0.2. Surface-level ozone fluxes onto the sahara desert. Atmos. Environ. 1996, 30 (6), 911−918. (15) Eastman, J. A.; Stedman, D. H. Fast Response Sensor for Ozone Eddy-Correlation Flux Measurements. Atmos. Environ. 1977, 11 (12), 1209−1211. (16) Schurath, U.; Speuser, W.; Schmidt, R. Principle and Application of a Fast Sensor for Atmospheric Ozone. Fresenius’ J. Anal. Chem. 1991, 340 (9), 544−547. (17) Güsten, H.; Heinrich, G.; Schmidt, R. W. H.; Schurath, U. A Novel Ozone Sensor for Direct Eddy Flux Measurements. J. Atmos. Chem. 1992, 14 (1−4), 73−84. (18) Brenninkmeijer, C. A. M.; Crutzen, P. J.; Fischer, H.; Gusten, H.; Hans, W.; Heinrich, G.; Heintzenberg, J.; Hermann, M.; Immelmann, T.; Kersting, D.; Maiss, M.; Nolle, M.; Pitscheider, A.; Pohlkamp, H.; Scharffe, D.; Specht, K.; Wiedensohler, A. CARIBIC Civil aircraft for global measurement of trace gases and aerosols in the tropopause region. J. Atmos. Ocean. Tech. 1999, 16 (10), 1373−1383. (19) Güsten, H.; Heinrich, G.; Mönnich, E.; Nolle, M.; Weppner, J. Two automated ozone analyzers for use on civil aircraft operating in the tropopause region. J. Atmos. Ocean. Tech. 2003, 20 (2), 292−300. (20) Mao, H. T.; Talbot, R.; Troop, D.; Johnson, R.; Businger, S.; Thompson, A. M. Smart balloon observations over the North Atlantic: O-3 data analysis and modeling. J. Geophys. Res.-Atmos. 2006, 111, (D23). (21) Regener, V. H. On a Sensitive Method for the Recording of Atmospheric Ozone. J. Geophys. Res. 1960, 65 (12), 3975−3977. (22) Speuser, W. Eine Chemilumineszenzsonde zur Messung von Ozonprofilen in der Atmosphäre. Ph.D. Dissertation. Rheinische Friedrich-Wilhelms Universität Bonn, Bonn, 1989. (23) Sahand, S. Optimierung der Oberflächen-Chemilumineszenz für Ozonmessungen in der Atmosphäre. Ph.D. Dissertation. Rheinische Friedrich-Wilhelms-Universität, Bonn, 1989. (24) Ulanovsky, A. E.; Yushkov, V. A.; Sitnikov, N. M.; Ravegnani, F. The FOZAN-II fast-response chemiluminescent airborne ozone analyzer. Instrum. Exp. Tech. 2001, 44 (2), 249−256. (25) Fö rster, T. Zwischenmolekulare Energiewanderung Und Fluoreszenz. Ann. Phys. 1948, 2 (1−2), 55−75. (26) Periasamy, A.; Day, R. N. Molecular Imaging: FRET Microscopy and Spectroscopy; Oxford University Press: Oxford; New York, 2005. (27) Güsten, H.; Heinrich, G. On-line measurements of ozone surface fluxes 0.1. Methodology and instrumentation. Atmos. Environ. 1996, 30 (6), 897−909. (28) Muller, J. B. A.; Percival, C. J.; Gallagher, M. W.; Fowler, D.; Coyle, M.; Nemitz, E. Sources of uncertainty in eddy covariance ozone flux measurements made by dry chemiluminescence fast response analysers. Atmos. Meas. Tech. 2010, 3 (1), 163−176. (29) Yang, W. X.; Meixner, F. X. Laboratory Studies on the Release of Nitric Oxide from Subtropical Grassland Soils: The Effect of Soil Temperature and Moisture. In Gaseous Nitrogen Emissions from Grasslands, Jarvis, S. C., Pain, B. F., Eds.; CAB International: Wallingford, NY, 1997; pp 67−70. (30) Zahn, A.; Weppner, J.; Widmann, H.; Schlote-Holubek, K.; Burger, B.; Kuhner, T.; Franke, H. A fast and precise chemiluminescence ozone detector for eddy flux and airborne application. Atmos. Meas. Tech. 2012, 5 (2), 363−375. (31) Miet, K.; Le Menach, K.; Flaud, P. M.; Budzinski, H.; Villenave, E. Heterogeneous reactions of ozone with pyrene, 1-hydroxypyrene and 1-nitropyrene adsorbed on particles. Atmos. Environ. 2009, 43 (24), 3699−3707. 1936


Preparation Methods to Optimize the Performance of Sensor Discs for

Recommend Documents