An Instrument for Measuring the Hydrogen Content of Metals J. B. Condon, R. A. Strehlow, and G. L. Powell Union Carbide Corporation, Nuclear Division, Oak Ridge, Tenn. 37830 An instrument has been developed to measure the hydrogen content of metals in the range from 0.01 to 100 weight parts per million (wppm). The instrument requires a 0.1-gram to 5.0-gram sample size and records, as a function of time, both the rate at which hydrogen is thermally evolved from the sample and the total amount of hydrogen evolved.
Low CONCENTRATIONS of hydrogen in metals can have a dramatic effect on their physical properties. Hydrogen is known to contribute significantly to hardness and embrittlement of stainless steels and transition metals in general. Stress corrosion cracking in titanium and other transition metal alloys has been attributed to hydrogen charging. A determination of low concentrations of hydrogen in the one weight-part-permillion (wppm) range and lower has become increasingly important to the study of this phenomenon. Often along with a knowledge of the hydrogen content, the diffusion properties and distribution of hydrogen in the metal or alloy is of vital concern. This would be especially important in the progression of stress corrosion cracking induced by hydrogen charging. A knowledge of the hydrogen diffusion rate also makes possible the establishment of reasonable control specifications in metal and alloy processing. High sensitivity instruments for hydrogen analysis have been reported in the literature (1-7). Most have relied on the vacuum hot extraction or fusion method. The most sensitive (8-11) utilized the high sensitivity of a gas chromatograph o n a collected volume of gas obtained by vacuum fusion. A few methods utilized a mass spectrometer and hot extraction to obtain both high sensitivity and instantaneous readout (12-14). The instrument described in the present article was designed to perform the following functions: To measure the hydrogen content of a metal in the range from 0.01 wppm to 100 wppm using a specimen size ranging from 0.1 gram to 5.0 grams; to measure the hydrogen evolution rate from the metal; to analyze for gases other than hydrogen; and to be operated under routine analytical laboratory conditions. Hydrogen evolution rate measurements could be used to determine the diffusion rate of hydrogen in metals as well as (1) C. R. Masson, Mefallurg. Rev.,12, 147, (1967). (2) M. W. Mallett, Talanfa,9, 133-144 (1962). (3) J. Fischer, Fresenius’ Z . Anal. Chem., 209, 163-176 (1965). (4) W. Baum and S. Eckhard, ibid., pp 188-197. ( 5 ) N. A. Gokcen and E. S. Tankins, J. Mefals, 14, 584-87 (1962). (6) W. G. Guldner, Talanfa, 8, 191-202 (1961). (7) S. Matsuo, T. Hirata, Nippon Kinzoku Gakkaishi, 31, 590-3 (1967). (8) F. Sperner and K. H. Kock, Metall, 18, 701-704 (1964). (9) Richard Lasser and H. Gruber, Z . Metallk, 51,495-501 (1960). (10) Richard Lasser, Trans. Nat. Vac. Symp., 8, 782-787 (1961). (11) J. J. Schmidt-Collerus and A. J. Frank, WADD TR, W482, Denver Research Institute, University of Denver, Denver, Colo., 1961. (12) H. Hintenberger, Fresenius’ 2. Anal. Chem., 209, 176-187 (1965). (13) L. A. Ferguson, D. E. Seizinger, and C. H. McBride, U. S. At. Energy Rept. MCW-1462 (1961). (14) V. I. Fistul, Kom Anal. Khim.Akad, Nauk SSSR, Inst. Gcokhim Anal. Khim., 10, 71-81 (1960) (URRGTrans-1377). 1448
to time-resolve “surface hydrogen” from hydrogen distributed uniformly throughout the bulk of the metal. “Surface hydrogen” could be atomic or molecular hydrogen adsorbed on or near the surface of the metal or hydrogen combined in water or organic compounds adsorbed on the metal surface. For most transition metals, these compounds would react with the metal to produce molecular hydrogen during the early stages of the hot-extraction process. This “surface hydrogen” could contribute up to one wppm to the total hydrogen content of a one-gram metal specimen. EXPERIMENTAL
Principles of Operation. The principle of operation of the hot-extraction method used here was as follows. A sample was allowed to drop into a hot zone in a high vacuum system. The hydrogen and other gases were evacuated through an orifice into a detection chamber which was equipped with a residual gas analyzer (a small mass spectrometer or RGA). Proper control over the pumping speeds of the extraction chamber and the detection chamber allowed a calculation to be made of the total amount of hydrogen evolved as a function of time. In this system there was a constant pumping speed, SI,determined by a n orifice from the furnace chamber. The detection chamber was evacuated by a constant pumping speed, SZ. According to kinetic theory, the pressure in the detection chamber (P2) was related to the pressure in the furnace chamber (PI) by the following equations: dpi
dt
SI - RTdni - (Pl - PZ) Vi dt VI
+ Lv,1
and
I sz _ _ - (PI - Pz) S - - - Pz
dpz
vz
dt
vz
+ L2v-z
where n designates the amount of hydrogen evolved, L designates the inherent leak rate of the system which is constant and shows up as a base pressure, and V is the volume of the respective chambers. Upon integration and rearrangement, the following is obtained : RT VI
= (1
+ +
t)
Pz
+
Adt
+
-
+
If SZ >> SI and Vz E V I , then 1 Vz/V1 S2/SlS S a l . The last two terms in Equation 3 are extremely small and the third term is the base pressure correction. Letting Pz = Ai (Vl/RT), and subwhere i is the RGA signal, N = A (&/SI) stracting out the base pressure correction Equation 3 reduces to: n = N [i
+ S 1 / V , J idtl
(4)
Equation 4 is also the solution to Equation 1 for the case where i is prOpOrtiOM1 to PI and Pz >> PI. Under these conditions, the mass spectrometer could be used to measure PI in LL very passive manner.
ANALYTICAL CHEMISTRY, VOL. 43, NO. 11, SEPTEMBER 1971
EA1 QUAD 150 Output
a p Electrometer
R1
Signal Hydrogen Evolution Rate
Inverter b
Total Hydroaen ( w p m ) b
Channel 1 Channel 2 Recorder
Figure 2. Analog device for computing total hydrogen (wppm) from mass spectrometer output Philbrick-Nexus operational amplifiers powered by a PR-30C power supply were used. All fixed resistors were rated 1%and all variable resistors were rated 3%, 0.2% linear. Component designations were as follows:
Figure 1. High Vacuum Hydrogen Analyzer 1. Sample feed mechanism 2. Windows 3. Gas injection calibration system (Nupro Type-H valves) 4. Optically dense baffles 5. Bakeable ultra-high vacuum straight-through valve (Varian) 6. Kovar-to-borosilicate glass graded seal I. Borosilicate glass-to-quartz graded seal 8. Stainless steel mesh baffles to reduce sample impact on quartz baffles 9. Quartz baffles 10. Quartz furnace chamber 11. Quartz thermocouple well with thermocouple 12. Standard 1-inch4.d. combustion tube furnace (element insulated with firebrick contained in SS) 13. Reentrant-type liquid nitrogen dewars 14. Aluminum foil orifice mounted between Mini-ConFlat flanges 15. Orifice observation window 16. Nude ionization Gauge 17. Mass spectrometer (EA1 QUAD 150) 18. Bakeable ultra-high vacuum L-valve (Varian) 19. Liquid nitrogen cold trap (Granville-Phillips) 20. Refrigerated (-33°C) baffle (Torr Vacuum Prod) 21. Mercury diffusion pump (Torr Vacuum Prod) 22. Temperature controls for bake-out mantle 23. Bake-out mantle with 5000 watts heating capability 24. Bake-out mantle cover (removed during operation of instrument) 25. Liquid nitrogen fill tube 26. Color Ceran table top (Johns-Manville) All components made of 304 Stainless Steel with exception of those noted. All flanges above the liquid nitrogen cold trap were of the ConFlat type with which OFHC copper gaskets were used
Apparatus. The instrument was constructed mostly from ultrahigh-vacuum components readily available from commercial vendors. Figure 1 is a complete diagram of the instrument. Evacuation of the system was accomplished by a diffusion pump and a mechanical forepump. Use of ion pumps was unsatisfactory since the pumps re-evolved hydrogen and were easily saturated. The orifice, SI,consisted of a thin aluminum foil barrier containing a hole. A hole size of 0.030-inch diameter was found to be optimum. For convenience, this foil was mounted on the inside of the standard (1 ‘/*-inch i d . ) ultrahigh-vacuum “T” fitting by welding a Mini-ConFlat (Varian Associates) flange on the
Integrator-SP-65AU All other amplifiers P-35AU RI - 50KR R1 - 100Kn RB - 1 KQ Rd - 1 MO c - 20pf inside. The foil was then fixed in place with a mating MiniConFlat flange. To admit the sample to the hot zone, a mechanical manipulator (a bellows-type rotary-linear feed-through) was used to shove samples off the end of a platform. To ensure separation, a sliding board with holes in it was used to contain the samples. When a sample was admitted, it fell through a set of stainless steel baffles, through a straightthrough valve, and through another set of quartz baffles into the quartz tube hot zone. The function of the latter set of baffles was to prevent sublimation pumping. The potential gettering action that some metals will produce by the sublimation of the metal to form a n active thin film in the cooler portions of the extraction tube could significantly reduce the evolution of gases from the furnace tube. Titanium is especially effective in this respect and each metal must be examined on an individual basis to determine the severity of the problem. In most instances the problem can be solved by a series of baffles leading into the hot zone. Generally, metals will condense at a temperature sufficiently high that the sticking coefficient for hydrogen is very low, and pumping action will not occur with successive layer deposition. The purpose of the former set of baffles was to prevent premature heating of the samples by radiation from the quartz tube hot zone. The straight-through valve was extremely convenient for it isolated the sample chamber from the rest of the vacuum system. Sample loading could therefore be done without disturbing the rest of the vacuum system. A separate sorption pump was used on this area for pumpdown after loading. The quadrupole residual gas analyzer used with this instrument was a n EA1 QUAD 150 (Electronics Associates, Inc.). The signal was amplified with an electrometer (Keithley 417, Keithley Instruments, Inc.) which had a current suppressor to substract out the background signal. The signal was then routed through a n analog device whose schematic is shown in Figure 2. The analog device carried out the calculations indicated in Equation 4, taking into account the weight of the sample, the time constant, SI/VI, for the vacuum system, and displayed the total amount of hydrogen evolved from the specimen in weight parts per million as a function
ANALYTICAL CHEMISTRY, VOL. 43, NO. 11, SEPTEMBER 1971
1449
-
loo0
800
10
-
600
400 200 0
f x n
100 80 60
40 20 10 8
Hydrogen Admitted through the Calibrated PrerrureVolume Chamber (micrograms)
Figure 5. Experimental determination of accuracy of the hydrogen analyzer
6
4 2 1
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 Time (seconds)
Figure 3. Pump-down curves for hydrogen through two different orifices in the hydrogen analyzer Orifice 1
Orifice 2
S / V = 0.172 sec-l A S / V = 0.169 sec-l 0 SjV = 0.170 sec-l 0 SjV = 0.174 sec-l SjV av = 0.171 sec-l u = 0.002 or 1.3%
S/V = 0 . 3 4 9 ~ ~ - ’ A SiV = 0.342 sec-l 0 SjV = 0.316 sec-l 0 SiV = 0.338 sec-l S/V av = 0.336 sec-l
0
volume was attached to the furnace chamber through a n all-metal, bellows-type valve (Nupro SS-4H-SW, The Nupro Company). A second valve (Nupro SS4H-SW) connected this volume to a hydrogen source and a quartz Borden Gauge (Texas Instrument). Absolute pressure measurements in the range of 1.00 torr to 50.0 torr could be made with 1% accuracy. The Oak Ridge Y-12 Plant Standards Laboratory determined the “trapping” volume to be 2.10 & 0.05 cm3. This calibration was traceable to the National Bureau of Standards.
0
u
=
0.014 or 4.2%
of time on a strip-chart recorder. A two-channel (Brush 220, Gould Brush, Inc.) recorder was used. The second channel was used to display the mass spectrometer output which was approximately the hydrogen evolution rate from the sample. The analogue device had three variables that could be programmed. Two of the variables were N and (S,/Vl) found in Equation 4. The third variable was a scaler for the weight of the sample to give direct readout in wppm. The value of N and (Sl/V1> was determined by expanding a known amount of hydrogen from a “trapping” volume. A calibrated, “trapping”
RESULTS AND DISCUSSION The Sl/Vl term could be evaluated experimentally from the pumpdown rate. Typical pumpdown curves determined by this method can be seen in Figure 3, showing S J V , values of 0.34 0.02 sec-l and 0.171 i 0.002 sec-’ for two different orifices. After the value of S,/Vl was programmed into the analog device, the value of N was determined by adjusting the analog device to read directly on the recorder the amount of hydrogen in wppm released from the trapping volume for an assumed sample weight. Figure 4 shows typical results of the calibration procedure. Before each analysis the base pressure correction was adjusted until the analog device drifted at a rate of less than 0.001 wppm per minute. Repeated calibrations of the system over a large range (Figure 5) confirm that the precision was better than 10%. A typical hydrogen analysis of a 0.50-gram uranium cube containing 2.5 wppm hydrogen is given in Figure 6.
*
,5 U F CE I
0
I
; I
I
l
l
I
I
I
l
l
I
!
n
) c t = o
I
I
4 L
I
6x ‘
205 Seconds
1
-
$2
I
Integrator Reset
p - t = o
V
I
L
I
I
I
I
9
I
I
IO
I
I VI I
Seconds
I
1
I
I
I J
Figure 4. Typical calibration measurement (Trapping volume pressure, 15.6 torr, i.e., 3.5 weight, 1.00 gram) 1450
pg;
ANALYTICAL CHEMISTRY, VOL. 43,
assumed sample Figure 6. Hydrogen analysis of a 0.4-gram uranium specimen
NO. 11, SEPTEMBER 1971
.-w 6 A0 5
Time (seconds)
Figure 8. Effect of various treatments on hydrogen content and distribution in a uranium alloy (all samples were 0.50-gram
cubes) A. 30
60
90
120
Time (seconds)
-- - - -D. - - - - - -
B.
C.
Figure 7. Gaseous evolution observed on the RGA for uranium alloy rods Mass 2 (without trap) Mass 2 (with trap) Mass 15 (without trap) Mass 28 (without trap) Mass 18 (no increase without trap) Another factor taken into account is the formation of hydrogen by the reaction of the hot metal with the residual water or hydrocarbon vapor pressure. At the high temperature of extraction, the interdiffusion of metal and metal oxide could be very rapid and therefore affords exposure of free metal atoms at the sample surface to react rapidly with the water vapor to produce hydrogen. Since it was convenient in many cases to run the system at or lo-' Torr rather than ultrahigh vacuum, the production of hydrogen from this reaction for an extraction of a 1-gram sample for a 1-hour duration would be nearly 2 X mole which corresponds to a determinate error of f 4 ppm. The problem was effectively eliminated by exposing the extraction chamber to a liquid nitrogen cold finger with a pumping speed 1000 times greater than the orifice. This reduced the above error to j~0.004 PPm. The liquid nitrogen trap in the extraction chamber was a reentrant dewar with approximately a ten-square-inch cold area exposed to the vacuum. With a background pressure of Torr the analyzer without the trap cooled yielded an answer that was at least twice as high as it should have been for 32 wppm Ti standards. Even with a low background pressure of Torr or better a high answer was obtained. The size of the error was, of course, a function of the metals analyzed and the speed by which the surface was poisoned; it was best, however, to eliminate the error by use of the cold trap. Another error which may arise is from the cracking patterns of low molecular weight hydrocarbons. The origin of these hydrocarbons is probably from specimen handling and should, therefore, not be counted as part of the hydrogen content. Figure 7 presents the time dependence of masses 2 (hydrogen), 15 (methane and other hydrocarbons), 18 (water), and 28 (CO, N B ,and hydrocarbons) for some uranium alloy rods of nearly identical weight and treatment which were dropped in the hot zone without the liquid nitrogen dewar cooled. An overnight pumpdown of the system was used in an effort to remove surface water. Another trace of mass 2 was made with the re-entrant dewar cooled showing a marked decrease in the first hydrogen peak. The difference could be accounted for by the assumption that it was the mass 2 fragment from
Machined uranium alloy (32 wppm hydrogen) (attenuated X20) Machined uranium alloy (1.76 wppm hydrogen) Machined uranium alloy etched with 10% aqueous HNOJfor 10 min (2.2 wppm hydrogen) Machined uranium alloy high vacuum degassed at loo0 "C and water quenched (0.82 wppm hydrogen)
Table I. Typical Set of Hydrogen Analyses as Run HYdrogen Specimen content weight, total, specSpecimen description grams imen WPPm 1 5.0 pg Hzgas (47 torr cc) 1.oOoa 5.1 1 .oooa 2 5.0 pg Hzgas (47 torr cc) 4.8 3 5 . 0 pg H2gas (47 torr cc) 1 .OOoa 5.1 0. 054b 35 4 Titanium NBS Standards Rated, 32 wppm 0 . 047b 5 Titanium NBS Standards Rated, 98 98 wppm 6 Uranium Alloy No. 1 0,3786 2.7 7 Uranium Alloy No. 2 0.091b 42 Uranium Alloy No. 1 0.435* 1.9 8 Uranium Alloy No. 2 9 0.0986 44 Titanium NBS Standards Rated, 0 .O5Ob 41 10 32 wppm 107 11 Titanium NBS Standards Rated, 0 ,036b 98 wppm Assumed. 6 Extraction temperature IO00 "C.
hydrocarbons. With liquid nitrogen dewar pumping, no increase was observed for masses 15, 18, and 28. Without the trap, n o water evolution was observed indicating that the overnight high-vacuum pumpdown was sufficient to remove superficial water from the sample surfaces. As can be seen from Equation 4, a time plot of i was very nearly proportional to a time plot of the instantaneous gas evolution (dnldt) from the sample. Given a sufficiently high pumping speed from the furnace chamber, these plots would be proportional. In practice, the only place where i (or HP pressure in the detection chamber) and dn/dt differ significantly was where the pressure was changing rapidly. Since the plot of i was generally used to qualitatively evaluate the distribution of hydrogen in the sample, this difference between i and dn/dt was of little consequence. When a rigorous curve shape analysis of i was used to determine diffusion coefficients and activation energies, the mass spectrometer output (i) was taken digitally and corrected to dnldt by computer. Figure 8 demonstrates the effect of sample history on the rate that hydrogen was evolved from the furnace chamber. Hydrogen generated by surface effects significantly contributed to the total amount of hydrogen determined for a specimen containing less than 10 wppm hydrogen.
ANALYTICAL CHEMISTRY, VOL. 43, NO. 11, SEPTEMBER 1971
1451
Table 11. Hydrogen Determinations in wppm for Uranium and Uranium Alloy Specimens. Samoleb Specimen 1 2 3 4 Uranium-le 1.6 2.3 1.5 ... Uranium-2e 2.5 2.5 2.2 ... Uranium Alloy-ld 33 31 30 33 Uranium Alloy-2d 32 28 28 27 Uranium Alloy-3d 2.3 2.0 2.4 2.0 Uranium Alloy-4d 3.0 2.9 2.4 2.2 a Extraction temperature IO00 “C. * Each sample was cut from an adjacent position along a lia-inch by lis-inch rod. Uranium specimens were from different billets of high-purity derby uranium. Uranium alloy specimens were from different batches of similar materials except uranium alloys 3 and 4 had been vacuumheat treated.
Some hydrogen determinations for titanium, uranium, and uranium alloys were performed. Tables I and I1 demonsirate the precision of the instrument when used as a routine analytical device. During these analyses, the samples were de-
greased in’acetone, and the samples were allowed to accumulate in the furnace chamber until all eight samples were analyzed. Samples from a given specimen were analyzed on two different days in order to demonstrate the stability of the instrument. CONCLUSIONS
The instrument described in this article has been satisfactory for the determination of hydrogen in metals over the concentration range from 0.01 wppm to 100 wppm with a precision of 10%. Diffusion parameters, surface reactions, and the distribution of hydrogen in the sample can be evaluated using the hydrogen evolution rate data and the thermal history of the sample after it is dropped into the furnace. Micromole quantities of other gases can also be determined; thus, the instrument can be used to study a great variety of thermally activated gas evolution processes.
RECEIVED for review March 16, 1971. Accepted June 7, 1971. Work performed at Oak Ridge Y-12 Plant under contract W-7405-eng-26 with the U. S. Atomic Energy Commission.
Effect of Atmosphere on Spectral Emission from Plasmas Generated by the Laser Microprobe William J. Treytl, Kenneth W . Marich, James B. Orenberg, Peter W. Cam,’ D. Craig Miller, and David Glick Division of Histochemistry, Department of Pathology, Stanford University School of Medicine, Stanford, Calg. 94305 The effects of various atmospheres on laser-induced optical emission of plasmas from solid samples have been investigated. Atmos heres of argon, air, oxygen, nitrogen, helium, andalso vacuum, were used. Samples of iron in steel and in iron oxide-coated tape, and of magnesium in aluminum foil, human serum, and liver were employed. Laser energies of approximately 1.2, 3.6, and 8.0 mJ were used. Signal-to-background ratios were not found to vary systematically with atmosphere and laser energy, but significantly larger values occurred in vacuum at 1.6-mJ laser energy. Signal intensities were greater in denser atmospheres at 3.6 and 8 mJ, but approximately the same at 1.2 mJ. Signal intensities varied directly with these laser energies, except in vacuum where the signal was independent of laser energy. Metallic and nonmetallic targets behaved similarly.
SINCETHE INNOVATIONS in laser microprobe instrumentation reported by Peppers et al. (I), efforts have been directed to optimization of conditions for analysis of elements. The effect of composition of the sample on the laser-induced emission signal has been studied (2). The present investigation concerns the effects of various atmospheres on signal intensity from laser-generated plasmas. Effects on emission spectra of atmospheres in which plasmas Present address, Department of Chemistry, University of Georgia, Athens, Ga. 30601 ( I ) N. A. Peppers, E. J. Scribner, L. E. Alterton, R. C. Honey, E. S. Beatrice, I. Harding-Barlow, R. C. Rosan, and D. Glick, 40, 1178 (1968), ANAL.CHEM., (2) K. W. Marich, P. W. Carr, W. J. Treytl, and D. Glick, ibid.,
42, 1775 (1970). 1452
are generated by a dc arc have been reported (3-7). Vallee (3, 4 ) demonstrated that an argon-atmosphere enhanced spectral line emission without significantly increasing the background. In contrast, increased signal-to-noise ratio was achieved with helium by suppression of the background (3). Undesirable cyanogen bands were reduced by use of noble gases. Rates of volatilization of elements in a sample varied with the composition of the atmosphere. Vallee (3) concluded that, for particular metals, certain atmospheres could be used to enhance the sensitivity of analysis. The use of inert gases with the Stallwood jet increased signals, diminished background, reduced selective volatilization, and produced a more stable dc arc (8). Stabilization of the ac arc was also achieved by Sukhnevich (9) with inert gases. More recently, use of controlled atmospheres in spark excitation showed improved precision and accuracy, higher sensitivity, and reduced inter-element effects (IO, I I ) . (3) B. L. Vallee, C. B. Reirner, and J . R. Loofbourow, J . Opt. SOC.Amer., 40, 751 (1950). (4) B. L. Vallee and S. J. Adelstein, /bid., 42,295 (1952). (5) B. L. Vallee and S. J. Adelstein, Specrrochini. Acta, 6 , 134 (1954). ( 6 ) Z. L. Szabo and I. Toth, Mugy. Kem. Foly., 74, 394 (1968). (7) C. K. Matocha and J. Petit, Appl. Sprcrrosc., 22, 562 (1968). (8) A. J. Mitteldorf, “Trace Analysis: Physical Methods,” G. H. Morrison, Ed., Interscience Publishers, New York, N. Y., 1965, p 200. (9) V. S. Sukhnevich, Z. Prikl. Spektrosk., 9 , 199 (1968). (IO) R. R. Boyd and A. Goldblatt, Appl. SpPcrrosc., 19, 22 (1965). (11) J. Kashirna and M. Kubota, Rep. G r s r . Res. Lrrb., W m 4 Utiio., 1967, No. 18, pp 9-19.
ANALYTICAL CHEMISTRY, VOL. 43, NO. 11, SEPTEMBER 1971