Fund Project:Project supported by the National Key Research & Development Program of China (Grant No. 2016YFC0201104) and the National Natural Science Foundation of China (Grant Nos. 51676105, 11972213, 51906120)
Received Date:11 July 2019
Accepted Date:19 August 2019
Available Online:01 October 2019
Published Online:20 October 2019
Abstract:Continuous wave cavity ring down spectroscopy (CW-CRDS) method with using cavity length scanning is ideal for accurately characterizing the low pressure spectra and measuring the small spectral parameters (such as the Dicke narrowing coefficient and the speed dependent collision broadening coefficient). However, the laser of any wavelength can be coupled to the cavity due to the cavity scan, so the spectral noise caused by the laser wavelength fluctuations cannot be ignored. This noise is non-uniformly distributed in the spectrum (especially on both wings on the spectral line) and is difficult to eliminate even with long-term averaging. Unlike the complex laser frequency locking techniques or the optical frequency combs or the better lasers, in this paper, a simple, easy to operate, fast wavelength-scanned CRDS method is proposed based on Fourier transform. The laser wavelength is continuously tuned across the absorption line to measure the periodic ring-down time. A reconstruction algorithm is developed to precisely recover the absorbance by extracting the characteristic frequencies of the periodic ring-down time after the Fourier transform. An etalon, instead of the wavelength meter, is used to calibrate the relative laser wavelength. This method effectively eliminates the non-uniform spectral noise caused by laser wavelength fluctuation in traditional CW-CRDS and significantly improves the measurement accuracy of spectral line parameters (especially line parameters in complex line shapes, such as speed dependent Voigt line shape) at low pressure. In addition, the measuring system, in which no wavelength meter is used, is simpler, more economical than CW-CRDS. The smaller residuals of the Galatry profile fit to the measured CO transitions at R(5) 6371.299 cm–1 and R(6) 6374.406 cm–1 show that the noise on both wings of the spectra, caused by laser wavelength fluctuation, is effectively reduced and the spectral SNR is then improved. The measured N2 perturbed collision broadening coefficient of the Voigt profile fit for CO is consistent with that from the classical CW-CRDS method and is in good agreement with the HITRAN2016 database. The measured N2 perturbed Dicke narrowing coefficient of the Rautian and Galatry profile and speed dependent collision broadening coefficient of the speed dependent Voigt profile have very good linear relationship with pressure, and have smaller uncertainties than the results from the CW-CRDS method. Keywords:cavity ring down spectroscopy/ Fourier transform/ wavelength scanned/ collision broadening coefficient
全文HTML
--> --> -->
3.FTS-CRDS理论CW-CRDS[7—9]采用固定激光波长和扫描腔长的方式, 使腔模式与激光耦合, 两者耦合时腔的透射光强可达到最大, 快速关闭激光即可得到单指数衰减的透射光强信号, 对该信号进行拟合可得到衰荡时间, 同时利用波长计测量得到激光绝对波长. 然后慢速改变激光波长以扫描分子整条吸收谱线, 从而可得到气体吸收光谱, 并采用多次平均方式来提升光谱信噪比. 但是, 由于腔长扫描时任意波长的激光均可耦合进腔, 波长波动导致的光谱噪声不可避免[14]且该噪声会非均匀分布在光谱上, 采用平均的方式难以进一步提升光谱信噪比[13—17]. 因此, 为了减小或消除激光波长波动等带来的噪声, 本文提出的FTS-CRDS方法在CW-CRDS基础上进行了如下改进. 激光电流(图2(a)和图2(b)中蓝色曲线)周期性扫描, 使得激光波长可扫描分子整条吸收谱线, 同时高速扫描腔长以保证衰荡信号的连续采集, 由于腔内气体分子吸收导致的损耗会随激光波长变化而周期性改变, 因此衰荡时间也随之改变, 从而得到蕴含气体吸收信息的周期性衰荡时间, 如图2(a)中黑色曲线所示. 与此同时, 采用法布里-珀罗(Fabry-Parot, F-P)标准具标定激光相对波长, 测量信号如图2(b)中红色曲线所示, 其中黑色曲线为拟合得到的激光相对波长(本文采用激光器的波长与电流呈负相关). 随后, 对图2(a)中黑色曲线所示的周期性衰荡时间信号进行傅里叶变换, 并从频域上滤除激光波长波动等噪声, 其频谱如图3所示. 最后, 通过提取特征频谱以复现气体吸收光谱. 该方法在测量方式上采用了波长连续扫描, 可有效减小波长波动等影响, 且无需波长计实时测量激光绝对波长, 可避免波长计引入的仪器噪声. 图 2 (a) 激光电流(蓝色)及周期性的衰减常数(黑色); (b) 激光电流(蓝色), 激光波长(黑色), 标准具信号(红色); 图中仅展示了100个周期中的4个 Figure2. (a) Instantaneous laser current (blue) and instantaneous ring-down time (black), τ(t); (b) wavelength (red) and etalon signal (black). Note that the (a) and (b) figure shows an example with only 4 of 100 the circles.
图 3τ(t)的傅里叶幅值谱及周期性噪声(~5.5, ~11, ~19和~21 Hz) Figure3. The amplitude spectrum of τ(t) and the periodic noises (~5.5, ~11, ~19 and ~21 Hz).
实验中采用CW-CRDS和FTS-CRDS方法分别对CO分子的6374.406 cm–1谱线进行了测量, 气体温度、压力和CO浓度分别为288 K, 18 kPa和0.1%(背景气N2). 在CW-CRDS方法中, 激光电流从70 mA以50 μA的间隔步进至90 mA, 在每个电流点测量200次, 总测量点数为8 × 104个, 用时约27 min, 测得的衰荡时间如图4(a)所示. 相应地, FTS-CRDS采集了200个锯齿波周期, 其扫描频率为1/8 Hz, 每个周期400点, 电流扫描幅度(70—90 mA)、总点数和时间与CW-CRDS相同, 测量结果如图4(b)所示. 与图4(a)相比, 图4(b)谱线中心两侧(蓝色与绿色区域)衰荡时间波动较小, 测量数据整体更加平滑. 由于FTS-CRDS采用快速波长扫描的方式, 测量中不需固定激光电流, 对激光电流和温度的稳定性依赖较小, 因此可有效减小激光波长波动等影响. 图 4 (a) CW-CRDS测量的衰荡时间, nd为每个电流点的测量次数, 白色点表示超出色阶范围; (b) FTS-CRDS方法测量的衰荡时间, np为所测周期数; 横轴均为激光电流, 为了更清晰地显示谱线中心区域, 这里仅显示了74?86 mA的数据 Figure4. (a) Ring-down time measured by CW-CRDS, nd is the number of measurements per current, white points represent out-of-range data; (b) ring-down time measured by FTS-CRDS, np is the number of cycles; the x axis represents laser current with the range of 74?86 mA.
图5(a)为CW-CRDS方法测得的吸收系数, 以及采用Voigt线型(VP)和Galatry线型(GP)拟合得到的最优结果. 从VP拟合结果可知, 残差中存在“w”形的精细结构, 其大小约为峰值吸收的1.7%, 其原因在于VP线型未考虑Dicke收敛效应或者速度依赖的碰撞展宽效应[11,12], 这种精细结构也验证了本文CW-CRDS测量结果的可靠性. 相比之下, GP线型考虑了Dicke收敛效应, 拟合时可消除Dicke收敛导致的“w”形的残差. 从GP拟合残差可知, 谱线中心两侧斜率较大位置存在较大的噪声, 其幅度最大约为4 × 10–8 cm–1, 相当于峰值吸收的1.6%, 这与Kassi等[14] (1.6%), Mondelain等[15](1.4%)测得的结果一致, 也即验证了此噪声主要来源于激光波长波动. 图 5 在相同条件下, 两种方法测量的CO吸收光谱(黑色虚线)及VP(红线)和GP(蓝线)拟合 (a) CW-CRDS; (b) FTS-CRDS; 两幅图的x轴和y轴的尺度相同 Figure5. The absorption spectra (black dotted line) of CO measured by the two methods and the best fit of Voigt profile (red) and Galatry profile (blue): (a) CW-CRDS; (b) FTS-CRDS; the x and y-axes scales of the residuals obtained by the two methods are the same