提出了热脱附-气相色谱-质谱法测定稀土萃取分离车间固定污染源废气中18种挥发性有机化合物（VOCs，包括1，1-二氯乙烷、1，2-二氯乙烯、1，1，2-三氯乙烷、甲苯、1，2，4-三甲基苯、苯、三氯乙烯、四氯乙烯，C₇~C₁₆等10种正构烷烃）含量。将采样器与2只串联的吸附管（装载有2种石墨化碳黑和1种碳分子筛）连接，于固定污染源排气口以50~65 mL·min^-1采样速率采集样品50 mL，密封吸附管，置于热脱附仪上，按照优化的仪器工作条件测定收集的气体中的18种VOCs含量（苯和四氯乙烯含量为2只吸附管吸附量的加和）。吸附管和聚焦冷阱的脱附温度均设置为350℃，脱附时间均设置为10 min。结果显示，10种正构烷烃标准曲线的线性范围为50.0~400 ng，其他8种VOCs的质量在20.0~300 ng内与其对应的峰面积与内标物峰面积的比值呈线性关系，检出限（3s）为0.001 5~0.026 mg·m^-3。对空白样品进行3个浓度水平的加标回收试验，18种VOCs的回收率为73.5%~110%，测定值的相对标准偏差（n=5）为1.2%~22%。4家实验室同时采用以上方法分析空白加标样品，偏差的绝对值均小于参考HJ 168-2010计算所得的再现性限。

A method for determination of 18 VOCs (1,1-dichloroethane, 1,2-dichloroethylene, 1,1,2-trichloroethane, toluene, 1,2,4-trimethylbenzene, benzene, trichloroethylene, tetrachloroethylene and 10 n-alkanes such as C₇~C₁₆) in the exhaust gas from stationary polution source of rare earth extraction separation workshop was proposed by gas chromatography-mass spectrometry after thermal desorption. The sampler was combined with 2 adsorption tubes (loaded with 2 types of graphitized carbon black and 1 type of carbon molecular sieve) in series to collect 50 mL of the sample at sampling rate of 50-65 mL·min^-1 from the exhaust port of the stationary polution source. The adsorption tube was sealed and placed on the thermal desorption instrument, and the contents of 18 VOCs in the collected gas were determined according to the optimized working conditions of the instrument (the contents of benzene and tetrachloroethylene were the sum adsorption capacity of 2 adsorption tubes). The desorption temperature of the adsorption tube and the focused cold trap was set to 350 ℃, and the desorption time was set to 10 min. As shown by results, the linear range of standard curve for 10 n-alkanes was 50.0-400 ng, and the masses had linear relationships with ratios of peak areas between the other 8 VOCs and internal standards in the range of 20.0-300 ng, with detection limits (3s) in the range of 0.001 5-0.026 mg·m^-3. Test for recovery by addition standard method was made on blank samples at the 3 concentration levels, giving results of recovery in the range of 73.5%-110%, and RSDs (n=5) of the determined values ranged from 1.2% to 22%. The above method was used for analyzing blank spiked samples in 4 laboratories, and absolute values of the deviation obtained were less than the reproducibility limits calculated with referring to HJ 168-2010.