VoL. 17, No. 2 53

S~ 2516 A B 2498 A

FIG. 2. MOVING PLATE SPECTROGRAM

Twenty sec intervals

Figure 1 gives a comparison of the aluminum hne at 3961 A using silver chloride alone (Figure 1A) with the same line obtained with the NK-2 mixture (Figure 1B). The same sort of comparison was observed for sihcon and other refractories. As shown in Figure 1, suppression of the line begins at the 60-ppm level when silver chloride is

v

the carrier. A moving plate exposure, shown in Figure 2, revealed that boron at the 100-ppm level is completely volatilized from a 50 mg charge in 20 sec. Sihcon at the 400-ppm level is almost completely volatilized m 40 sec. Table II shows the high and low limits of determination of the forty elements obtained by the procedure. Note that niobium is determined at 4 ppm whde the zirconium limit extends to a low of 10 ppm.

The precision of the method under discussion was com- pared with the silver chloride carrier method and an addi- tional method used by the laboratory for samples with higher impurity levels. The results of this comparison are shown in Table III. The mixed carrier results are, in gen- eral, significantly more precise than those obtained by the other two methods. The data from the new carrier show no biases.

The authors thank Mr. E. E. McCombs of UCNC, ORNL, for preparing the high-grade silver fluoride used in this work. Submitted June 11, 1962

Spectroscopic Tricks Nuclear Magnetic Resonance in Metal Powders

at Low Temperatures* Rex J. Snodgrass and Lawrence H. Bennett

Notional Bureau of Standards, Washington, D. C

An improved method for low temperature nuclear mag- netic resonance (NMR) measurements on metal powders is presented. A Varian Wide-Line NMR Spectrometer of the crossed-coil type is used to observe the resonance in metallic powder samples at liquid nitrogen and liquid helium temperatures. The spectrometer probe, dewar and sample are shown in Figure 1. Powders having particle sizes less than the skin depth are used to permit sufficient pene- tration of the sample by- the radio-frequency magnetic field. Motion of the sample caused by the bubbling refrig- erant introduces electromagnetic noise and represents the limiting factor in making accurate measurements. We find that the most satisfactory method for holding the sample stationary in the probe and thereby eliminating this noise is to glue it to the bottom of the dewar with paraffin oil.

The NMR signal voltage induced in the receiver coil of the probe by the precessing nuclei of the sample apt~ears as a modulation of the magnetic couvlin~ (leakage) be- tween the transmitter and receiver coils. The leakage can be controlled by couplin~ devices called paddles. The r)ad- dles must be adjusted with the sample in the probe, since the metallic samvle itself affects the coupling. It is im- portant that both the phase and amplitude of the leakage remain constant during the time of passage through a resonance curve. Incorrect leakage phasin~ results in the detection of a mixture of absorption and dispersion modes of the signal. The mode-mixin~ is reflected in d~stortions and asymmetries of the recorded resonance line and is a severe hindrance to accurate measurements of centers, widths, and line shapes.

~Note added in proof: Since the submission of this trick we noticed that Dr T J. Rowland, in his PhD. thesis (Harvard, 1954, un- published) mentions, without details, using paraffin oll to avoid the vibration accompanying the boding of liquid air in a single coll spectrometer.

When the probe is balanced, the magnetic flux lines near the coils have a definite and stable configuration. Probe unbalance is essentially a result of disturbing these lines of force. With conducting powder samples, the main unbalancing factors are the changing conductivity of the sample due to temperature changes and any change m the physical orientation of the sample in the receiver coil. The former may be largely overcome by holding the sample at a constant temperature. The latter is a general problem faced by experimentalists who have liquid mtrogen or helium in contact with the sample. The constant bubbling of the refrigerant imparts a small motion to the sample, which disturbs the flux lines and appears as a major source of noise.

The nitrogen dewar is designed so that its lower part (the finger) fits into the Varlan rf probe (Figure 1). The finger is 17 mm o.d., and 13 mm i.d. The dewar is 25 cm long, and its upper reservoir is 13 cm o.d. It is silvered

FIG. 1. SPECTROMETER RF PROBEr NITROGEN DEWAR, AND ENCAPSULATED SAMPLE

54 APPLIED SPECTROSCOPY

with the exception of the finger (since this would cause undesirable eddy-currents) and a strip for viewing the nitrogen level. The nitrogen lasts about two and one-half hr without refilling.

Observation of the resonance at 77°K was first at- tempted by pouring liquid nitrogen directly on the metalhc powder sample which was in the finger of the dewar. A large amount of noise from the motion of the sample was detected. High pressures built up in "pockets" with the results that the sample "exploded" out of the dewar.

The metallic powder in which the resonance is to be observed is now encapsulated in glass, care being taken not to melt any part of the sample when sealing off. The glass capsule is about 38 mm long, about 11 mm o.d., and is about one-half filled with sample. The capsule is placed in the bottom of the finger of a dewar and nitrogen is poured around it. A number of unsuccessful methods were tried to hold the capsule stationary with respect to the receiver coil.

These included: 1) Glass wool or Styrofoam was wedged around the encapsulated sample. This reduced but did not eliminate the electrical noise. In addition, the nitro- gen was not always able to contact the sample uniformly. 2) Duco cement was allowed to dry around the sample. The coefficient of thermal expansion of the cement is suffi- ciently different from that of glass to break the dewar finger when nitrogen is poured in. 3) GE low temperature cement was used to glue the sample in place. The con- ductivlty of this cement is apparently high enough to per- mit eddy-currents to flow and prevent balancing of the probe. 4) A glass rod was securely fastened to the sample and clamped at the upper part of the dewar. This provided too long a "lever arm" between the sample and the point

FIG. 2. FINGER OF DEWAR SHOWING GLASS CAPSULE GLUED WITH FROZEN PARAFFIN OIL NEAR THE BOTTOM

The metal powder sample fills about half the capsule Nitrogen bubbles can be seen near the top of the capsule.

of clamping to hold the sample perfectly still. 5) Nitrogen bubbling may be suppressed by introducing a stream of a low-boiling-point gas such as helium into the liquid nitro- gen (1). Although this method was found to be partly successful for reducing noise, it had several deficiences: a) The temperature of the sample is reduced below 77°K re- quiring control of the rate of flow of helium and measure- ment of temperature; b) It cannot be used at hquid hehunl temperature; and c) It is more complicated than the method described below.

We find that the best method to hold the sample sta- tionary is to glue it by pouring in paraffin oll until the part of the capsule filled with sample is about one-half covered (See Figure 2). Paraffin oil, a liquid at room temperature, solidifies around the lower half of the sample at 77°K or lower, thus providing the low temperature "glue". The volume change of the paraffin oll upon freez- ing is insufficient to break the glass dewar finger. I f the receiver coll is placed around that part of the sample not covered with paraffin oil, the presence of the frozen oil does not reduce the thermal contact between the sample and the refrigerant. A false bottom on the capsule would serve the same purpose and conserve the sample.

It is believed that this extremely simple and convenient way of holding small samples stationary, whde in ~ood thermal contact with the refrigerant, will be applicable to both wide-line NMR requiring the measurement of shifts small compared to the line width and to electron spin reso- nance at low temperatures.

Literature Cited (1) G. J. Minkoff, F. I. Scherber, and A. K. Stober,

NATURE 180, 1413 (1957). We thank Dr. E. D. Becker for calling our attention to this reference.

Submitted March 15, 1962

A Variable Three-Step Sector Rotating Filter R. M. Kennedy and A. Paolini, Jr.

Campbell Soup Company, Camden, New Jersey

Many spectrographic procedures require the use of filters in the optical path, some calling for two trans- mittance steps and others for three. The filtering of the transmitted light has normally been accomplished with the use of calibrated metallized quartz or, in some in- stances, with fixed aperture sector wheels. Laboratories that handle a considerable variety of materials for analysis would, out of necessity, need an assortment of these filter- ing devices. With this in mind, a variable three-step ro- tating sector has been devised. It permits the recording of 100% and any two combinations of 50% of the trans- mitted light (i.e., 100%/0 to 50%/0 to 50%) , with step three less than step two. The filtering arrangement is inexpensive and relatively easy to construct. I t consists of three discs mounted on a shaft, which is rotated through the transmitted light at the appropriate position on the optical bench. The shaft can be mounted on a motor used for step sector rotation or on a conventional laboratory stirrer.

The discs are constructed of 1 mm aluminum or 0.5 mm stainless steel. Two discs have o.d. of 90 mm; one disc has an o.d. of 88 mm. All other dimensions of the three discs are identical and are shown in Figure 1. Two 90 ° segments of each disc are cut away, leaving an i.d. of 65 mm. A center of 20 mm diam. is cut out for