Trace Moisture Contamination in Ultra-High Purity Phosphine: Techniques for Measurement and Control M. Raynor, H. Funke, J. Yao, T. Watanabe and R. Torres, Matheson Tri-Gas, Advanced Technology Center, 1861 Lefthand Circle, Longmont, Colorado K. Bertness, Optoelectronics Division, National Institute of Standards and Technology, Broadway, Boulder, Colorado Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, March 7-12, 2004, Chicago, Illinois
Introduction Gas purity is critical to the production of III-V compound semiconductor devices Trace O from impurities such as H 2 O, are incorporated into epitaxial layers during MOCVD processes cause lattice defects and undesirable oxide formation decrease luminescence of optoelectronic components Objectives: Investigate methods for measurement of trace H 2 O impurity in PH 3 Demonstrate moisture control through point-of-use purification
Measurement Techniques Requirements of Trace Analysis Instrumentation Sensitivity and selectivity for H 2 O in PH 3 at ppb levels Rapid and accurate response Rugged and capable of withstanding reactive gas Long term stability Ease of use and applicability to on-line monitoring Techniques Investigated Chilled mirror hygrometry Cavity ring-down spectroscopy FTIR spectroscopy
Chilled Mirror Hygrometry Frost point is defined as the temperature to which a volume of gas must be cooled such that it becomes saturated with respect to ice. Exponential relationship between frost point and H 2 O concentration is well documented in N 2 Fundamental measure of H 2 O concentration Capable of long-term accuracy and stability Calibration required in PH 3 Condensation temperature of the PH 3 matrix gas limits the ultimate sensitivity of the technique
Chilled Mirror Hygrometer Cryogenic Sensing Chamber and Mirror Assembly Cooling Unit PH 3 in The mirror temperature is cooled until a frost layer is established. The mirror temperature is adjusted to maintain a constant condensation layer on the mirror. The layer is monitored optically and displayed as a balance voltage PH 3 out Buck Research CR-3 Chilled Mirror Hygrometer
Setup for Evaluation of Chilled Mirror Hygrometer Purified N 2 in N 2 in dry N Moisture Generator To hood Moisture Permeation Tube use to establish frost layer MFC Pressure transducer Mass Flow Meter To Scrubber PH 3 sample cylinder Purifier Chilled Mirror Hygrometer Gas Cabinet FTIR Purge N 2
Calibration Curve for Moisture in Phosphine using the Chilled Mirror Hygrometer -84-82 -80-78 -76-74 -72-70 -68 Frost Point Centigrade 2000 1800 1600 1400 1200 1000 800 600 400 200 0 ppbv Water Calibration points from 77-1730 ppbv moisture in PH 3 (1 slpm, 14.7 psia) Percentage of N 2 in PH 3 kept below 1% for all calibration points System equilibrated for ~4 hours at each moisture concentration
Moisture Detection in Phosphine Cylinder Sources Using Chilled Mirror Hygrometer 278 ppb H 2 O in PH 3 59 ppb H 2 O in PH 3 [H 2 O] (ppbv) 300 250 200 150 100 50 0-78 -79-80 -81-82 -83-84 -85-86 -87 0 60 120 180 240 300 Time (min) Frost Point ( C) [H 2 O] (ppbv) 120 100 80 60 40 20 0 Frost layer established 0 50 100 150 200 Time (min) -82-83 -84-85 -86-87 -88 Frost Point ( C) PH 3 from cylinder source, (N5.7 purity) at 1 slpm Long equilibration time PH 3 from cylinder source, (Ultima 6N purity) at 1 slpm Once established, frost layer maintained during analysis
Detection Limit [H 2 O] (ppbv) 70 65 60 55 50 45 Develop frost layer Lost layer Develop frost layer Lost layer 0 50 100 150 200 Time (min) Develop frost layer Lost layer -84.4-84.6-84.8-85 -85.2-85.4-85.6-85.8-86 -86.2 Temperature ( C) Mirror could not be cooled below minimum mirror set-point of -86 C due to potential condensation of liquid PH 3 H 2 O concentration in PH 3 less than detection limit
Observations with Chilled Mirror Hygrometer Detection of H 2 O in PH 3 possible down to ~55 ppb Reestablishing frost layer is time consuming and requires user intervention Mirror contamination occurs over time and results in noisy readings Regular cleaning Long equilibration time
Principle of Cavity Ring-Down Spectroscopy Light Light Time a Time Light is injected at a frequency, matched to an absorption frequency of the impurity being measured When the cavity and the input frequency are in resonance, the cavity fills with light The output light increases, triggering the shutter on the input light The light in the cavity recirculates and decays exponentially
Schematic of CRDS System at NIST Trigger Electronics AOM 936 nm Laser Scope/Digitizer Mode Matching Optics Cavity PH 3 (toxic) HeNe Cavity Length Servo Loop PZT (shutter) (freq. shifting) AOM
Ring-down Signals Normalized Light Power (arb. units) 1 8 6 4 2 0.1 8 6 4 2 0.01 8 6 4 0 100 200 A B 1.0 0.8 0.6 0.4 0.2 0.0 0 300 400 Time (µs) A B 200 500 400 600µs 600 700µs τ ( ν ) a = c L tot t = ring-down time a = cavity length c = speed of light (in cavity medium) 1 ( ν L tot = total loss per pass ) Ring-down time depends on base losses (frequency independent at this scale) absorption losses (frequency dependent)
CRDS Results: Dry cavity nitrogen Loss per pass x10 6 12.45 12.40 12.35 Water in Nitrogen at ν ~ =10697.42 cm -1 27 ± 2 nmol/mol -4000-2000 0 2000 4000 Frequency Shift (MHz) τ ( ν ) a = c L tot Water concentration is directly proportional to the peak area. 27 ± 2 nmol/mol water in nitrogen (27 ppb) n = 1 ( ν ) Model absorption lines as Voigt profiles on a linear baseline. Number density is then: Area a S a = cavity length S = absorption line strength
PH 3 Interference at 10687.36 cm -1 Loss per pass x10 6 22 20 18 16 Water in Phosphine at ν ~ =10687.36 cm -1 600 ± 350 nmol/mol Measured spectrum Fit to Voigt profiles H 2 O absorption PH 3 absorption -4000-2000 0 2000 4000 Frequency Shift (MHz) For low concentrations of H 2 O, there is too much interference from unmapped PH 3 absorption to accurately deconvolve the absorption lines Detection limit in N 2 at this absorption line < 15 nmol/mol (15 ppb)
Survey of Strong H 2 O lines near 935 nm 30 25 20 PH 3 at 6.65 kpa (50 Torr) 10667.76 cm -1 10687.36 cm -1 10697.42 cm -1 10700.67 cm -1 Loss per pass x10 6 15 10667.6 10667.8 10687.2 10687.4 10687.6 10697.2 10697.4 10697.6 10700.6 10700.8 10701.0 Wavenumber (cm -1 )
CRDS Spectrum: 6.65 kpa PH 3 at 10667.76 cm -1 15.5 14.0 H 2 O mole fraction = 590 ± 160 nmol/mol Pressure-broadening coefficient = 28 ± 7 MHz/kPa Background also changes with pressure Loss per pass x10 6 15.0 14.5 14.0 13.5 13.6 13.2 12.8-8000 -6000-4000 -2000 0 2000 4000 Measured spectrum Fit to Voigt profiles H 2 O absorption PH 3 absorption 13.0-2000 0 2000 4000 Frequency Shift (MHz) Estimated detection limit = 50 nmol/mol (50 ppb)
CRDS Summary Potential of CRDS for trace H 2 O detection in PH 3 demonstrated by NIST researchers Challenge to find a moisture line in 935 nm region where PH 3 absorptions are minimal Estimate 50 ppb detection limit using line at 937.40350 nm (10667.76 cm -1 ) Future Investigate other spectral ranges to lower detection limits
FTIR Spectroscopy Based on absorption of IR radiation by moisture impurity in a flowing gas stream within a gas cell Non-destructive technique Gas wetted parts can be selected for compatibility with phosphine Demonstrated sensitivity with long path-length cell Requires optimization of instrument design and measurement conditions
Infrared Spectrum of Dry PH 3 and Moisture in N 2 H 2 O detected in 1520-1560 cm -1 region with minimal interference from PH 3 absorptions Absorbance -0.118-0.119-0.120-0.121-0.122-0.123-0.124 990 ppb moisture in N 2 3.0 2.5 2.0 1.5 Dry Phosphine -0.125-0.126-0.127-0.128 1580 1570 1560 1550 1540 1530 1520 Wavenumber (cm-1) Dry phosphine 1510 1500 1490 Absorbance 1.0 0.5 0.0-0.5-1.0-1.5-2.0-2.5 4000 3500 990 ppbv Moisture in N 2 3000 2500 2000 Wavenumber (cm -1 ) 1500 1000 500 Calibration spectra input into classical least squares software for quantitation 512 spectra, 4 cm -1 resolution to maximize signal to noise Quantpad analysis program
FTIR Response IR absorption of moisture follows Beer s law (Log I/I o = -abc) Linear response for H 2 O standards diluted in dry PH 3 matrix LOD REG : 33 ppbv Dry N 2 bench purge required for long term stability Air must be excluded from sampling system to prevent PH 3 decomposition products affecting results CLS Response [ppmv] 1.2 1 0.8 0.6 0.4 0.2 0 y = 1.0783x + 0.0063 0 0.2 0.4 0.6 0.8 1 1.2 Moisture Concentration [ppmv] LOD REG based on ordinary least squares method (SEMI C10-0698)
FTIR Detection of Moisture in PH 3 0.1 Dry PH 3 0.1 PH 3 Ultima 6N Grade from Cylinder Source Moisture [ppmv] 0.08 0.06 0.04 0.02 0 Standard Deviation of Moisture data in Dry PH 3 is ± 10 ppbv Moisture [ppmv] 0.08 0.06 0.04 0.02 Moisture Equilibrates at 50 ppbv -0.02 0 60 120 0 0 50 100 150 200 Time (min) Time (min)
FTIR Observations FTIR currently method of choice for quantification of trace H 2 O in PH 3 Careful selection of H 2 O bands required to optimize sensitivity Detection to ~30 ppb possible with low resolution instrument, an MCT detector and a 10 m pathlength cell Improved sensitivity (10-20 ppb) likely with dedicated FTIR analyzers optimized for on-line gas measurements
Moisture Control Dry down of delivery systems exposed to atmosphere difficult due to adsorptive properties of moisture Gas phase moisture concentrations may vary over cylinder life, due to enrichment of water in the liquid phase Point-of-use purification provides a solution for consistent gas purity Efficiency of purifier materials evaluated using FTIR
Removal of H 2 O in PH 3 by Point of Use Purifier Removal of 5 ppm H 2 O in PH 3 by Nanochem PHX Purifier at 0.4 slpm -0.115-0.116-0.117 ~5 ppm H 2 O in PH 3 (Purifier Inlet Challenge) <100 ppb H 2 O reached in 10 min H 2 O level less than FTIR detection limit after 80 min Absorbance -0.118-0.119-0.120-0.121-0.122 0.25-0.123-0.124 0.2-0.125-0.126 Purified PH 3-0.127 H 2 O (ppm ) 0.15 0.1-0.128 1560 1 ppm H 2 O in N 2 1550 1540 1530 1520 Wavenumber (cm -1 ) 1510 1500 0.05 0 0 20 40 60 80 100 Time (min) IR spectra showing features of 5 ppm inlet H 2 O challenge and absence of features in purified PH 3
Summary and Outlook Current techniques allow H 2 O detection at tens of ppbs Performance of some III-V devices may be effected by H 2 O at much lower levels than can be measured analytically Point-of-use purification important for consistent gas purity Next generation analyzers should provide single digit and sub-ppb level sensitivity Laser based systems have greatest potential of filling sensitivity requirement