AMC monitoring by laser spectroscopy: Introduction and Recommendations Document information Owner Lead Name: Nils Lüttschwager Lead Organisation: PTB E-mail: [email protected] Context Author(s): Nils Lüttschwager (PTB) Olav Werhahn (PTB) Volker Ebert (PTB) Stefan Persijn (VSL) Kaj Nyholm (VTT MIKES) Timo Rajamäki (Aalto) Work Package: 1 Task: 4 Deliverable Number 1.4.4 Recommendations on AMC monitoring by existing optical Deliverable Name methods - 1 - Version: 1.0 (28 April 2016) Report Status: PU Public - 2 - Version: 1.0 (28 April 2016) Report Status: PU Public Scope This document shall serve as an introduction to optical detection methods for AMC monitoring and shall discuss their strengths and weaknesses. It shall thus facilitate the selection of a suit- able detection method by clean room operators and furthermore provide recommendations on the usage of these methods. Specifically, commercially available laser spectroscopic analys- ers based on optical cavity enhancement spectroscopy and photoacoustic spectroscopy will be discussed. Abbreviations (OA)-ICOS (Off-axis) integrated cavity output spectroscopy AMC Airborne molecular contamination CEAS Cavity-enhanced absorption spectroscopy CRDS Cavity ring-down spectroscopy DI Designated institute GC Gas chromatography HITRAN High-resolution transmission molecular absorption database IMS Ion mobility spectrometry ITRS International Technology Roadmap for Semiconductors NDIR Non-dispersive infrared sensor NMI National metrology institute PAS Photoacoustic spectroscopy ppb Parts-per-billion, used here synonymous for nmol mol-1 as a measure of amount fraction TDLAS Tunable diode laser absorption spectroscopy UV Ultra violet (light) VOC Volatile organic compound - 3 - Version: 1.0 (28 April 2016) Report Status: PU Public Content 1 Introduction ....................................................................................................................... 5 2 What laser-spectroscopic analysers can and cannot do .................................................... 6 3 Measurement principle of laser absorption spectroscopy .................................................. 7 3.1 What is measured? .................................................................................................... 7 3.2 Cavity-enhancement and photoacoustic detection ................................................... 10 3.3 Interpretation of absorption spectra .......................................................................... 11 3.4 From spectrum to compound amount fraction (concentration) .................................. 13 4 Recommendations on instrument selection ..................................................................... 13 5 Recommendations on instrument operation .................................................................... 16 References ............................................................................................................................. 18 - 4 - Version: 1.0 (28 April 2016) Report Status: PU Public 1 Introduction Airborne molecular contamination (AMC) is a collective term for chemical contaminants which can be harmful to a variety of processes running in clean rooms. AMC covers a wide range of different chemical compounds like inorganic acids, inorganic bases, or volatile organic com- pounds (VOCs), which can harm manufacturing processes for example by corrosion, or the formation of haze on optics. Yield loss and low product quality can follow from elevated AMC levels in clean room air, such that sometimes high efforts to control and monitor AMC are prof- itable. Control of AMC may be accomplished by the use of specialized filters or use of mini- environments. Monitoring the clean room air quality is necessary to identify source of AMC and ultimately eliminate these sources, as well as assessing whether AMC control measures meet the required performance. A well controlled and well characterized environment also helps to more easily reproduce processes from site to site and to push boundaries to smaller scale fabrication. Several monitoring techniques for clean room air exist, the most interesting ones probably be- ing the “real-time” or “online” techniques which enable immediate corrective action in case of a contamination. Here, we consider a technique to be online or in real-time when it provides measurement results at least every few minutes. This excludes techniques which trap the AMC that shall be monitored (e.g. impingers) over time for later offline analysis, for example by ion chromatography or mass spectrometry. Online techniques may be based on chemilu- minescence, UV fluorescence, cavity-ring-down spectroscopy, or ion mobility spectrometry, to name a few common techniques. Depending on the user requirements, one of these techniques may be better suited than the others. In this Guide, we focus on techniques based on laser spectroscopy (such as cavity- ring-down spectroscopy). We want to shortly introduce these techniques to readers which are not familiar with laser spectroscopy and to explain why it may be preferred over other tech- niques. Also, we provide recommendations what should be considered when purchasing and running a laser spectroscopy-based analyser. The outline of this Guide is as follows. We start with a description of the measurement capabil- ities of laser spectroscopic analysers, followed by a short and very basic theoretical descrip- tion of laser absorption spectroscopy and three common and more advanced spectroscopic detection techniques: Cavity-enhanced absorption spectroscopy (CEAS), cavity-ring-down spectroscopy (CRDS) and photoacoustic spectroscopy (PAS). We believe that a basic under- standing of the main concepts will greatly enhance the reader’s ability to judge whether optical detection is a good option for his measurement needs and also helps to make better use of - 5 - Version: 1.0 (28 April 2016) Report Status: PU Public spectroscopic devices. Based on these discussions we conclude recommendations what to consider when purchasing and using a spectroscopic instrument. 2 What laser-spectroscopic analysers can and cannot do AMC monitoring using laser spectroscopy provides many features which are sought by the user: Fast measurements, high sensitivity and selectivity, high dynamic range, accuracy in the low percent-range, low cost of ownership and little maintenance (we discuss how these bene- fits come about later in the text). Special sample preparation is not required and the AMC con- centration is measured from air samples without in between steps like catalytic conversion (e.g. chemiluminescence measurements of ammonia) or mixture with a doped carrier gas (ion mobility spectrometry) which lowers the probability of malfunction and erroneous measure- ments. Even sampling-free measurements of air not contained in a gas cell using an open- path configuration (Miller et al., 2014) are possible. In some cases, absolute laser- spectroscopic measurements can be done which do not require the calibration of the analyser using a calibration gas (calibration-free operation). Easy-to-operate, mobile spectrometers are now widely available. Besides all these positive properties, laser spectroscopy is not the universal solution to AMC monitoring. The analysers we discuss here work best only for small molecules composed of up to about 5 atoms, which excludes most VOCs and other larger AMCs. What is left are the small (but important) typical AMCs like ammonia (NH ), hydrogen chloride (HCl) or hydrogen 3 fluoride (HF), which are well suited for laser spectroscopic detection. Laser spectrometers often detect only a single species. The price for multi-species analysers is considerably higher because they include several lasers (lasers for gas sensing can be ex- pensive). Laser spectroscopy is technically more demanding than, for example, NDIR, chemi- luminescence, or UV fluorescence, and the price for a spectroscopic analyser may be in the range of several 10k EUR, which can be too high for a point-of-use device. - 6 - Version: 1.0 (28 April 2016) Report Status: PU Public To provide an orientation on the approximate performance of laser-spectroscopic analysers, we summarize specifications which were collected from overall 18 commercial instruments in Table 1. (Note that response time and accuracy of HCl and HF analysers was often not speci- fied.) Table 1: Typical specifications of commercial laser-spectroscopic analysers, currently available on the market (based on a web-search yielding 18 instruments) ammonia hydrogen chloride hydrogen fluoride AMC NH HCl HF 3 detection limit / ppb 0.1−10 0.1−2.5 0.1−1 dynamic range 1:10000 1:10000 1:10000 time resolution / s <1−5 1−5 1−5 response time / min 0.2−5 n/a n/a accuracy / % 1−10 n/a n/a dimensions / cm3 30 x 40 x 60 30 x 40 x 60 30 x 40 x 60 3 Measurement principle of laser absorption spectroscopy 3.1 What is measured? The laser-based analysers we are discussing in this Guide measure, in various ways, how much infrared light from a laser is absorbed when the light passes through a gas sample. The light that is passed through the gas sample excites gas molecules to vibrate if the light fre- quency (wavelength) matches very precisely one of the molecule-specific vibration frequen- cies. Light with a frequency that matches a molecular vibration frequency is partly absorbed, depending on the concentration of the excited species, and the resulting light attenuation can be measured. Figure 1 shows the schematic of a basic laser absorption spectrometer which consists of a laser, a gas cell that holds the gas sample and provides stable measurement conditions, a photodetector which measures the intensity of the laser light that passes through the sample, and electronics to analyse the signal generated by the photodetector, e.g. an os- - 7 - Version: 1.0 (28 April 2016) Report Status: PU Public cilloscope. The laser is repeatedly scanned in a small wavelength interval and absorption of light is seen as a drop in the detector signal. Since the laser wavelength can be scanned very rapidly, measurements with a time resolution well below a second are possible. Figure 1: Illustration of a laser absorption experiment. Typically, light absorption is measured in terms of the absorbance, which is the logarithm of the ratio of the light intensity before (I ) and after it passes through the gas sample (I): 0 A = ln(I /I). 0 Figure 2 shows how a laser absorbance spectrum of air could look like under typical meas- urement conditions if it would contain a trace of hydrogen fluoride. One sees different peaks, commonly called absorption lines, each belonging to a particular molecular vibration excitation from one of the species present in the sampled air. Most of the absorption lines belong to wa- ter from the air humidity, but the peak highlighted in blue is caused by the excitation of hydro- gen fluoride molecules. The peak absorbance of the HF line is proportional to the amount of HF in the measured gas sample over a wide range of concentrations (which is the basis for the high dynamic range of laser spectrometry). A basic analyzer for HF could thus be realized by using a laser which is tuned to an emission wavelength of about 2475.88 nm (4038.9 cm-1), - 8 - Version: 1.0 (28 April 2016) Report Status: PU Public the peak centre of the HF absorption line in our example, measuring the absorbance at this single wavelength and calibrating against known HF concentrations. However, more often the laser will be scanned repeatedly in a small wavelength window to measure not only the height but also the shape of the absorption line. This allows a more sophisticated data evaluation that will be more robust to potential cross interference which may result from overlapping absorp- tion lines. Through their use in telecommunications, tunable diode lasers have reached particular tech- nical maturity and are often employed as light sources, not least because their emission wave- length can conveniently be controlled by varying the laser temperature and/or supply current. Spectroscopy which uses a diode laser as light source is often termed TDLAS, for tunable di- ode laser absorption spectroscopy. While TDLAS is simple, robust and well developed, detec- tion limits are commonly in the range of several parts-per-million (ppm) which is not sensitive enough for use in clean room AMC monitoring. Figure 2: Absorption spectrum of hydrogen fluoride in air (HITRAN simulation, see below) - 9 - Version: 1.0 (28 April 2016) Report Status: PU Public