热光法测量扬尘碳组分的激光分割点偏移

为探讨热光法测量扬尘碳组分的激光分割点偏移（LSPS）的特点和原因，并提出偏移数据的处理方法，选择西宁市、沈阳市、淄博市、十堰市和新疆乌昌石地区共607个扬尘（道路扬尘、城市二次扬尘、土壤扬尘）样品，通过再悬浮采样器获取PM_2.5滤膜样品，使用DRI2001A型碳分析仪结合透射法测量碳组分.结果表明：道路扬尘和城市二次扬尘在碳分析过程中会有不超过10%的样品发生前偏现象，该现象主要受碳酸盐碳（CC）和元素组分的影响；当CC占总碳（TC）的百分比大于9%时，加酸方法可解决前偏问题.土壤扬尘中有40%~90%的样品会发生后偏现象，原因是其元素碳（EC）含量过低（EC/TC<；10.7%），导致灼烧过程残留的耐热物质Fe₂O₃对激光的干扰大于EC对激光的吸收.针对以上偏移问题，采用归零法和替代法对偏移数据进行处理.基于与参比值比较，归零法和替代法得到的OC相对偏差分别为0.1%~10.3%和0.1%~2.9%，EC相对偏差分别为2.1%~52.5%和0.1%~20.6%，远低于偏移引起的相对偏差（0.6%~13.8%和66.6%~149.6%），且替代法处理效果优于归零法.

Laser split point shift in measuring carbon components of fugitive dust by thermal-optical method

In order to study the characteristics and causes of laser split point shift(LSPS) of Organic Carbon(OC) and Elemental Carbon(EC) in fugitive dust by thermal-optical method(IMPROVE_A TOT), and to propose two data processing methods, we collected 607 fugitive dust samples(road dust, urban dust, soil dust) from number of cities, including Xining, Shenyang, Zibo,Shiyan, and areas of Xinjiang Wu-Chang-Shi. We used resuspension chamber to collect PM_2.5 filter samples from the fugitive dust.The results showed:road dust and urban dust samples mainly occurred early split of EC/OC, and the proportion was less than 10%.Carbonate carbon(CC) and metal/mineral elements were the main influencing factors. When the proportion of CC to total carbon(TC) was more than 9%, CC could be removed by adding acid. The soil dust samples mainly occurred late split of EC/OC, and the proportion was between 40% and 90%. The reason was that its EC was low(EC/TC<10.7%), resulting in the interference of laser by the remaining heat resistant Fe₂O₃ was greater than the absorption of laser by EC. We proposed two data processing methods to correct LSPS:(1) Setting OPC to zero(OPC₀); or(2) a replaced value(OPC_re). Based on the comparison with the reference values,the relative deviations of OC were 0.1%~10.3% and 2.1%~52.5%, and EC were 0.1%~2.9% and 0.1%~20.6%. They were much lower than the relative deviations caused by LSPS:0.6%~13.8% and 66.6%~149.6%. And the OPC_re was better than OPC₀.