大氣PM2.5監測及採樣正確性的研究

A study of monitoring and sampling accuracy of PM2.5 in the ambient air

劉俊男(Chun-Nan Liu)

指導教授 ： 蔡春進

國立交通大學/工學院/環境工程系所/博士(2014年)

https://doi.org/10.6842/NCTU.2014.00063

摘要

由於許多流行病學的研究顯示大氣PM2.5濃度對人體有負面的影響，使得PM2.5這項議題已越來越受重視。為了維護人民的健康，許多國家也紛紛制定了PM2.5的空氣品質標準。目前用來量測大氣PM2.5濃度進而評估空氣品質是否合乎這些標準的方法主要分為手動採樣與自動即時監測兩種方法。然而由於採樣干擾的影響，使這些方法的測值和大氣中實際的PM2.5濃度常有差異。因此本研究主要的目的為評估這些大氣PM2.5儀器之量測結果的正確性。
在PM2.5採樣技術方面，本研究首先開發了一部可同時採集4組PM10與4組PM2.5樣本的多濾紙PM10-PM2.5採樣器(Multi-filter PM10-PM2.5 sampler, MFPPS)。MFPPS的操作流量為33.4 L/min，在經過進口端的PM10衝擊器之後，主要採樣氣流會分成兩道16.7的次要採樣氣流。一道會進入4組PM10濾紙匣，另一道則是在通過一個PM2.5衝擊器之後再進入4組PM2.5濾紙匣。故每道濾紙匣的採樣流量為4.17 L/min。實驗室內的校正及現場比對結果均顯示，MFPPS可正確地採集大氣PM樣本。
本研究也接著利用MFPPS進行濾紙採樣器可能產生之採樣干擾的評估。研究結果顯示，在採樣過程中PM2.5的揮發損失相當嚴重，此揮發量佔修正揮發量後之PM2.5濃度的比例為5.8至36 %，且會隨著濾紙面速度的增加以及濾紙上的微粒負荷量的減少而增加。在24小時的採樣過程中，揮發掉的NH4+、NO3- 及Cl-離子濃度佔修正揮發量後之PM2.5濃度的比例分別為9.5 ± 6.2、5.4 ± 3.7及2.0 ± 1.3 %，佔個別離子濃度的比例則分別為46.4 ± 19.2、66.9 ± 18.5及74.4 ± 14.0 %。
除了實驗評估之外，本研究也建立了一個模式來推算在採樣過程中PM2.5的揮發損失，結果顯示在樣本上微粒酸鹼程度呈中性的情況下，以此模式所算出的PM2.5揮發量和實驗數據相當接近，兩者之間的平均誤差在10 %以內。本研究也進一步利用此模式來探討PM2.5濃度、環境溫度與相對溼度對於微粒在濾紙採樣過程中揮發程度的影響。
在PM2.5的監測技術方面，本研究評估一部可自動修正採樣干擾的儀器-錐狀微量震盪天平與濾紙動態量測系統(Tapered element oscillating microbalance – filter dynamic measurement system, TEOM-FDMS)的量測結果。研究結果顯示由於上述濾紙採樣揮發損失的影響，使得WINS PM2.5採樣器(WINS)、雙道採樣器(Dichot)及MFPPS所量測到的PM2.5濃度會分別較TEOM-FDMS的測值平均低了16.6 ± 9.0、15.2 ± 10.6及12.5 ± 8.8 %。然而當MFPPS測得的PM2.5濃度以多孔金屬片固氣分離器採樣器(porous-metal denuder sampler, PDS)測得的揮發量加以修正後，其量測結果會和TEOM-FDMS接近。
最後本研究也評估了目前台灣空品測站所使用之貝他計(Beta attenuation monitor, BAM)的PM2.5量測正確性。研究結果顯示BAM在新莊及竹東測站的PM2.5濃度監測值(PM2.5,B)會分別較雙通道採樣器所量測到的PM2.5濃度(PM2.5,D)高出29.78 及 28.42 %。研究結果發現，BAM和Dichot之間的採樣差異主要係由BAM所使用之玻璃纖維濾紙吸附酸性氣體造成的正向干擾所導致。本研究也另外估算了採樣期間的大氣含水量以及BAM和Dichot於採樣過程中樣本上揮發性硝酸銨的揮發量，結果顯示兩者對PM2.5,B和PM2.5D之間採樣誤差並無顯著的影響。

關鍵字

大氣細微粒 ； 濾紙採樣器 ； PM2.5即時監測器 ； 採樣干擾 ； 半揮發性微粒

並列摘要

Ambient PM2.5 concentration has been attracting great concern because of its adverse effect on human health as revealed by many epidemiological studies. To protect public’s health, PM2.5 air quality standard was promulgated in many countries. Currently, manual sampling and automated real-time monitoring are the two major methods for measuring ambient PM2.5 concentration to determine the compliance with these standards. However, the difference in PM2.5 between these measurements and ambient actual level often exists due to the effect of sampling artifacts. Therefore, the objective of this study is to evaluate the measurement accuracy of the PM2.5 sampling and monitoring techniques.
For PM2.5 sampling, a multi-filter PM10-PM2.5 sampler (MFPPS), which enable the collection of four PM10 and four PM2.5 samples simultaneously, was first developed. MFPPS operates at a total flow rate of 33.4 L/min, and after a PM10 impactor assembled at the sampling inlet, the main flow rate is separated into two minor flows of 16.7 L/min. One of them is directed into four PM10 filter cassettes while the other is directed to a PM2.5 impactor followed by four PM2.5 filter cassette. Therefore, the sampling flow rate for each sampling channel of the MFPPS is 4.17 L/min. Both laboratory calibration and field validation show that the MFPPS is able to measure the ambient PM concentration accurately.
After that the MFPPS was used to evaluate the sampling artifacts in filter-based sampler. Results showed that the evaporation loss in PM2.5 was severe during sampling, accounting for 5.8 to 36.0 % of the corrected PM2.5 concentration and the percentage increased with decreasing loaded particle mass and increasing filtration velocity. During 24-h sampling, the evaporated NH4+, NO3- and Cl- concentrations accounted for 9.5 ± 6.2, 5.4 ± 3.7, and 2.0 ± 1.3 % in corrected PM2.5 concentration, respectively, or 46.4 ± 19.2, 66.9 ± 18.5, and 74.4 ± 14.0 % in the concentration of each species, respectively.
Besides experimental study, a model for predicting the evaporation loss during filter sampling was also developed. Results show that when the collected particles are nearly neutral with a pH equals to 7 to 7.5, the evaporated concentrations predicted by the present model agree well with the experimental data with an average difference less than 10 %. The theoretical model was then used to predict the effect of PM2.5 concentration, ambient temperature and relative humidity on the extent of evaporation loss during filter sampling.
In PM2.5 monitoring, the measurement performance of a PM monitor so called tapered element oscillating microbalance – filter dynamic measurement system (TEOM-FDMS), which is able to correct sampling aritfacts, was evaluated. Results show that due to the evaporation loss, PM2.5 concentrations measured by WINS PM2.5 sampler (WINS), Dichotomous sampler (Dichot), and the MFPPS were lower than those the TEOM-FDMS by 16.6 ± 9.0, 15.2 ± 10.6 and 12.5 ± 8.8 %, respectively. However, when the MFPPS PM2.5 concentrations were corrected for the evaporated loss determined by the porous-metal denuder sampler (PDS), good agreement with those by the TEOM-FDMS was achieved.
Finally, the measurement accuracy of the beta attenuation monitor (BAM) which is used in Taiwan air monitoring stations was evaluated. Results show that the PM2.5 concentrations measured by the BAM (PM2.5,B) are consistently higher than those by the dichot (PM2.5,D) by 29.8 ± 20.2 and 28.4 ± 19.0 %, respectively, at Sinjhuang and Judong stations. It is found that the overestimation is mainly caused by the positive artifacts due to acid gas adsorption by the glass fiber filter tapes used in the BAM. Other influencing factors such as aerosol water content and volatilization loss of inorganic semi-volatile species are found to be less important to the overestimation.

並列關鍵字

ambient fine particles ； filter-based sampler ； PM2.5 real-time monitor ； sampling artifacts ； semi-volatile particles

參考文獻

[3] Ashbaugh, L.L., Eldred, R.A., 2004. Loss of particle nitrate from Teflon sampling filters: Effects on measured gravimetric mass in California and in the IMPROVE network. Journal of the Air & Waste Management Association 54, 93-104.

[4] Aggarwal, S.G., Mochida, M., Kitamori, Y., Kawamura, K., 2007. Chemical closure study on hygroscopic properties of urban aerosol particle in Sapporo, Japan. Environment Science and Technology 41, 6920–6925.

[5] Appel, B.R., Tokiwa, Y., Haik, M. Kothny, E.L., 1984. Artifact particulate sulfate and nitrate formation on filter media. Atmospheric Environment 18, 409–416.

[6] Arends, B.G., Nell, J., Rutten, S.M., 2000. Field comparison of four PM10 samplers in a polluted area in the Netherlands. Journal of Aerosol Science 31, 512–513.

[7] Chang, C.T., Tsai, C.J., Lee, C.T., Chang, S.Y., Cheng, M.T. Chein, H.M., 2001. Differences in PM10 concentrations measured by ß-gauge monitor and Hi-vol Sampler. Atmospheric Environment 35, 5741–5748.