Measurement of Trace Levels of Moisture in UHP Hydride Gases by FTIR Spectroscopy

Measurement of Trace Levels of Moisture in UHP Hydride Gases by FTIR Spectroscopy

The authors discuss a long pathlength Fourier Transform Infrared (FTIR) spectroscopy method for direct on-line measurement of low ppb to ppm moisture levels in hydrides.

By S. Salim, A. Gupta

Measurement of ppm to ppb levels of moisture in ultra high purity NH3, AsH3, PH3 and SiH4 is crucial for quality assurance and quality control in the microelectronics industry. In this work, a long pathlength Fourier Transform Infrared (FTIR) spectroscopy method for direct on-line measurement of low ppb to ppm moisture level in the hydrides was successfully developed. FTIR is the method of choice, since the infrared beam does not decompose the gases and the sampling system can be constructed of materials which are compatible with the hydride gases. It, therefore, allows direct on-line analysis with ease of operation. Detection limits as low as 2 ppb are achievable for moisture in these hydrides. This is significantly better compared to the currently available commercial moisture analyses techniques, which allow only 1 ppm detection limit for NH3 and 0.1 ppm for AsH3, PH3 and SiH4.

Moisture impurities in Ammonia (NH3), Arsine (AsH3), Phosphine (PH3) and Silane (SiH4) are parasitic in chemical vapor deposition applications for the microelectronics industry NH3 and SiH4 are used as sources for nitrogen and silicon in the deposition of Si3N4 layers. The presence of moisture impurity has a well-known destruction effect to the formation of “haze” on the surface of freshly nitrided silicon wafers. AsH3 and PH3 are sources of arsenic and phosphorous in the epitaxy of III-V compound semiconductor thin film [e.g. Gallium Arsenide (GaAs) and Indium Phosphide (InP)] used for optoelectronics applications. The presence of moisture can lead to undesired oxygen incorporation in the epitaxal layers, which significantly decrease the device performance. The level of moisture in these hydrides should be minimized to better serve the stringent needs of the industry. Development of a state-of-the-art analytical tool for moisture measurement is vital for assuring a low level of moisture in these hydride gases.

Moisture measurement in NH3 has traditionally been very difficult and no direct vapor phase measurement has been available for ppm and sub-ppm detection. [1] The accepted SEMI International Standard is based on an indirect analysis involving a dew point measurement of byproducts of thermal decoposition of NH3. [1] The detection limit obtainable by this technique is only low ppm. Current available commercial analytical measurements of moisture in SiH4, AsH3 and PH3 are limited to the detection limit of 100 ppb. [1].

This work pioneered the application of the Fourier Transform Infrared (FTIR) technique for measurement of sub-ppm and low-ppb moisture in hydrides. FTIR is the method of choice since:

1. The low energy IR source does not decompose the hydrides.

2. It allows direct gas phase analysis and is a non-destructive technique.

3. Gas-wetted parts are easily dried and compatible with the hydride gases.

4. Analysis time is short.

5. The theoretical limit of detection is much better than currently used techniques.

The scope of this work includes selection of the spectral region for moisture measurement in hydrides, optimization of FTIR parameters for improving the Limit of Detection (LOD), and demonstration of moisture measurement in hydrides.

Experimental setup

A schematic of the FTIR and the gas handling setup used in this work is shown in Figure 1.

The gas handling system was constructed of electropolished stainless steel piping, air actuated valves and mass flow controllers. The source of moisture for calibration was a Kintek Permeation tube kept in a Span Pac 61 oven, which was supplied with purified house Nitrogen. Praxair Phoenix-grade AsH3 and PH3 and UHP-grade NH3 and SiH4 were used in this study. Nanochem resin purifiers for NH3 and SiH4 and a Millipore resin purifier for AsH3 and PH3 were used to generated zero gas with less than 1 ppb moisture.

The exhaust of the system was connected to an appropriate scrubbing system. A resin-based ATMI scrubber was used to trap AsH3 and PH3 from the system effluent. For SiH4, a custom-designed scrubber containing 20 percent KOH was used with diluent Nitrogen on the head space. A water scrubber was used for capturing effluent NH3.

Both the gas handling system and the FTIR system were kept inside a working hood, due to the toxic and pyrophoric nature of the these gases. The system was designed to allow remote control and access to the valves and mass flow controllers. Three MDA toxic gas monitors were used during the experiments with AsH3 and PH3. They were placed inside the hood, at the effluent of the scrubber system, and just outside the hood.

Selecting the spectral region

Figure 2 shows the IR spectra of N2, H2O, NH3, PH3, AsH3 and SiH4. All the spectra were taken at 1 atm gas and 40&#176C. Moisture absorbs in two spectral regions: 1,900 to 1,300 cm-1 and 3,900 to 3,600 cm-1. NH3 absorbs weakly in the 3,650 to 3,900 cm-1, and therefore this regional may be used for moisture measurement. Moisture measurement is not possible in 1,900 to 1,300 cm-1 due to the strong Ammonia absorption. Moisture measurement in AsH3, PH3 and SiH4 can be performed as well in the 3,900 to 3,600 cm-1, since these hydrides do not absorb strongly in this spectral region.

Optimization of the FTIR

The optimization of the FTIR system was carried out by installation of an optical filter, optimization of spectral resolution, optimization of the scan time, and minimization of the background moisture.

The optical filter installed in combination with the Indium Antimonide (InSb) detector improves the signal-to-noise ratio by an order of magnitude. The filter allows light to pass only in the region of interest in the 3,900 to 3,600 cm-1 as shown in Figure 1. Without the addition of an optical filter, the typical light level is usually about 30 times lower than the usable level for the detector. [2]

A customized purging setup has been developed and used for removing background moisture. A long-term stability study shows that the build-in purge system in the Magna 750 was not sufficient for maintaining a constant level of moisture in the system. As an example, the moisture level changed by 50 ppb in the span of 5 minutes as the humidity of the room changed.

The long-term stability study, performed using the customized purging, indicates the system is stable and constant within the LOD for more than eight hours. This approach is better compared using a vacuum FTIR bench. [2, 5] or a double beam FTIR system. [2]

Infrared absorption of moisture follows the Beer`s law [3]:

Log (I/I&#176)= -a.b.c

where I/I&#176 is the ratio of the moisture absorption in comparison with reference signal, a is the moisture absorptivity, b is the gas cell path length, and c is the concentration of the moisture. Figure 3 shows the plot and the result of regression analysis of moisture absorbence as a function of calculated concentration. Moisture standards used were generated by a permeation device.

An LOD of 1.6 ppb was obtained when the system was operated optimally. This LOD was determined using three times standard of error estimate (s) obtained from the regression analysis. [4, 5] This limit is 500 times better than the currently accepted SEMI standard for NH3 and about 50 times better than those used for the other hydrides.

Moisture measurement in hydrides

The moisture measurement is performed by utilizing QuantPad software which is a quantitative analysis program from Nicolet. The software is equipped with classical least square algorithm for quantification of the spectra. [6]

Figure 3a shows single-beam IR absorption of less than 1 ppb moisture in NH3 in the 3,620 to 3,870 cm-1. This condition is achieved by flowing NH3 through a Nanochem purifier at 4 slpm. Figure 3c is the spectrum of 59 ppb of moisture in NH3 obtained by flowing 3.9 slpm of NH3 and 100 sccm Nitrogen containing 2.9 ppm moisture. The reference of the spectrum was Ammonia with less than 1 ppb moisture-free Ammonia flowing at the same flow rate. All the experiments were carried out at 780 torr and 40&#176C gas cell condition.

The level of moisture absorbence obtained from this measurement is the same to that of 59 ppb moisture in Nitrogen. Similar experiments were carried out in AsH3, SiH4, and PH3 matrixes and the results were similar to the NH3 case. These indicate that the moisture calibration in Nitrogen can be used for moisture measurement in all these hydrides. This result is similar to the conclusion drawn by Pivonka for moisture measurement in HCl7, which indicated that for resolution less than 1 cm-1, band broadening and absorbance attenuation due to moisture interaction with the matrix gases are not detectable.

Figure 3d shows a steady-state spectra of moisture present in Ammonia that is delivered from a cylinder. The level of moisture observed was 50 ppb.


An FTIR technique for quantification of low ppb moisture levels in NH3, AsH3, PH3, and SiH4 is successfully developed. The measurement is based on moisture absorption in the 3,600 to 3,900 cm-1 region. The optimal system incorporates an InSb detector with an optical filter, Ge-coated Potassium Bromide (KBr) beam splitter, and operates at 39 second scan using 4 cm-1 resolution. A standard 10 m white gas cell is used to obtain LOD of greater than 2 ppb. This technique enables a direct, non-destructive measurement and is suitable for routine on-line analysis. n


1. Semi C3. 12-94, “Book of SEMI Standards 1995, Gases Volume,” SEMI, Mountain View, CA, 1995, p. 15-19.

2. B.R. Stallard, L.H. Espinoza, and T.M. Niemczy, “Trace Water Determination in Gases by Infrared Spectroscopy,” Proceedings, The Institute of Environmental Sciences, 1995, p. 1-8.

3. N.B. Colthup; L.H. Daly; S.E. Wiberly, “Introduction to Infrared and Raman Spectroscopy,” Academic Press, San Diego, CA, 1990.

4. T.J. Bzik; G.H. Smudde, Jr.; J.V. Martinez de Pinilos, “How Good is Your Limit of Detections,” Proceedings of Microcontamination `94 Conference, Canon Communications, Santa Monica, CA, p. 653-671.

5. P.B. Henderson; G.L. Auth; and S. Werschke, “An Optimized FTIR Spectrometer for the Determination of Moisture in Corrosive Gases,” 1996 Pittsburgh Conference, paper no. 1402, New Orleans, LA, 1995.

6. D.M. Haaland; R.G. Easterling, “Improved Sensitivity of Infrared Spectroscopy by Application of Least Square Methods,” Applied Spectroscopy, Vol. 34, 1980, p. 539.

7. D.E. Pivonka, “The Infrared Spectroscopic Determination of Moisture in HCl for the Characterization of HCl Gas Drying Resin Performance,” Applied Spectroscopy, Vol. 45, 1991, p. 597.

Amitabh Gupta is manager of the electronic technology and analytical development group at Praxair. This group performs research and development to create new products, analytical techniques, equipment and systems for Praxair Electronic Gases.

Sateria Salim is a development associate in the electronic technology and analytical development group of Praxair. He is an author and co-author of approximately 20 publications and presentations.

Click here to enlarge image

Figure 1. A schematic diagram of the gas handling system.

Click here to enlarge image

Figure 2. A) Infrared spectrum of trace moisture in Nitrogen; B) Background spectrum of Nitrogen; C) Background spectrum of Nitrogen plus the optical filter; D) Background spectrum of NH3; E) Background spectrum of PH3; F) Background spectrum of AsH3, G) Background spectrum of SiH4.

Click here to enlarge image

Figure 3. A) Background spectrum of NH3; B) IR spectrum of 0 ppb of moisture in NH3, C) IR spectrum of calibrated 59 ppb level of moisture in NH3; D) IR spectrum of moisture from an NH3 cylinder.


Easily post a comment below using your Linkedin, Twitter, Google or Facebook account. Comments won't automatically be posted to your social media accounts unless you select to share.