Tungsten Thermionic Emission as a Gauge for Low Pressures of Cesium Vapor

Abstract

Heated metal filaments under electric fields and low pressures of alkali metal gas eject electrons by thermionic emission as a function of the pressure of the gas and the temperature of the filament. To explore this process in a program to develop large-area alkali metal photocathodes, we have designed and built a gauge following the studies of Taylor and Langmuir. We present proof-of-principle measurements of the thermionic emission of a tungsten filament in cesium vapor. We describe a second-generation design that corrects flaws in the first gauge.

1. Introduction

Alkali metals adsorb onto metal surfaces, with a concentration and structure dependent on the temperature of the surface and the pressure of the metal vapor. The adsorption increases the thermionic emission of the surface compared to bare metal. We show that, by measuring the thermionic emission as a function of the temperature of a tungsten filament, one can infer the pressure of cesium vapor around the filament.

According to Richardson’s law, the thermionic current is solely determined by the work function of the surface and its temperature:

$$J={\lambda }_R{A}_0{T}^2{e}^{-W/k_{B}T}$$

(1)

The work function changes with the coverage of adsorbed cesium atoms. For coverages between 0 and 0.66, the adsorbed atoms lose an electron, creating an ionic charge layer which diminishes the work function and increases thermionic emission [1].

For coverages above 0.66, the work function is low enough so that cesium ions that adsorb do not lose an electron, creating a neutral layer. To accommodate the larger radius of the neutral atoms, this layer is rotated 30° about the normal [2] relative to the ionized layer below. The work function approaches that of cesium metal, suppressing thermionic emission.

Taylor and Langmuir [3,4] describe a relationship between the temperature of a tungsten filament, the cesium vapor pressure, and tungsten thermionic emission. The dotted curves in Figure 1 show the relationship of the thermionic emission as the temperature of the filament changes for different cesium atom fluxes, and is reproduced from Figure 15 in their 1933 paper [3].

In a program to develop large-area alkali metal photocathodes, we have explored thermionic emission by building a gauge to measure the partial pressure of cesium [5]. A crucial step in the creation of Cs₃Sb photocathodes is the transfer of cesium onto a Sb surface via evaporation from a source far from the surface, followed by deposition [6,7]. The gauge was designed to monitor the cesium pressure during process, between ${10}^{-5}$ and ${10}^{-3}$ Pa, although it was ultimately not implemented.

Section 2 describes the gauge construction, assembly, and operation. The results are presented in the context of the Taylor and Langmuir model in Section 3. Section 4 describes lessons we have learned from this first generation gauge and describes specific solutions to the issues we encountered.

2. Materials and Methods

The proof-of-concept gauge is shown in Figure 2. It consists of a 4.8514 cm by 0.0762 cm by 0.0025 cm tungsten ribbon filament from Scientific Instrument Services (SIS) surrounded by a collector for the thermionic current. The collector comprises a flat base made with a ceramic rectangular plate overlaid by a flat copper plate, on which is mounted a semi-cylindrical copper sheet, both cut from stock sheet copper of 0.001" thickness. The tungsten ribbon is supported above the rectangular plate by isolating screws at each end which serve as terminals for the filament.

All components are inside a 2 3/4^′′ CF 4-way cross flange, and are individually connected by ceramic-isolated copper wires to a vacuum feedthrough at one end of the cross. The cross chamber connects to an SS4-BG-V51 Swagelok valve leading to a custom cesium source containing a glass vial of 1 g of 99.9% cesium from ESPI Metals. The cross chamber is also connected to a HiPace 30 turbo-molecular pump from Pfeiffer Vacuum and a Compact FullRange™ Gauge PKR 251 gauge from Pfeiffer Vacuum through a valve [8]. The CF manifold pieces were DN40CF flanges and were metal-sealed, and all valves but the one to the cesium source were all-metal 541 angle valves from VMT.

To begin data-taking, the system was pumped to a pressure of 1.3 × ${10}^{-5}$ Pa, the FullRange™ gauge was valved off and the glass vial containing the cesium metal was broken to release cesium inside the cesium source chamber [6]. The turbo-molecular pump was valved off during measurements, but allowed to pump otherwise to remove hydrogen outgassing.

To achieve specific partial pressures of cesium, the manifold was heated, its temperature measured by 10 type K thermocouples and the temperature of a deliberately created cold spot recorded. The partial pressure of cesium was assumed to equilibrate after 12 h with the saturation pressure of the vapor over the cold spot. The manifold was heated by means of external glass-fiber insulated heat cable under a jacket of fiberglass insulation and the cold spot was created by omitting insulation from part of the external surface.

The circuit used to measure thermionic emission is outlined in Figure 3. It comprises two subsystems, the heating of the filament, and collectors that measure thermionic emission. The heating circuit consists of a power supply with a built-in ammeter and voltmeter (A and V in Figure 3) connected to the screws at the ends of the tungsten filament, with the negative terminal set at manifold ground. The measurement circuit consists of the copper plate and semi-circular sheet collectors tied to a bias voltage relative to ground through a 10.03 kΩ resistor. The voltage from the collector current, typically 0.01–100 V, is measured across this resistor.

As the bias voltage increases from zero, more thermionic electrons are directed to the collectors. There is a critical bias voltage after which the current plateaus because all emitted electrons reach the filament. We reproduced Figure 21 in the Taylor and Langmuir paper [3] to find this point: 180–200 V as shown in Figure 4. All measurements were performed at or beyond this critical bias voltage.

A measurement with the gauge consists of changing the filament temperature by changing the current and measuring the collection current by the voltage through the 10.03 kΩ resistor. The collected electron current divided by the surface area of the filament yields the thermionic electron flux. The conversion between filament current to filament temperature invokes a temperature transport equation, as described in Appendix A.

The measurements can be fit to the model of Taylor and Langmuir, which is used to infer the flux of cesium atoms in the manifold, and hence the Cs partial pressure. Measurements were taken without the aid of automated logging using voltmeters with 0.01 V precision. An automated measurement algorithm could easily be implemented.

Before each measurement, the circuit was probed for shorts between circuit elements and/or ground from cesium condensation on the gauge surfaces. To eliminate the shorts, we applied 300 V between each element and between the elements and ground. The initial low resistances increased to over 10 kΩ once the bias was applied. If the system was left idle for more than two hours, even if heated, the shorts returned and the process had to be repeated.

3. Results

Measurements of thermionic emission were taken at source temperatures of 363 K (Measurement A) and 305 K (Measurement B) with the manifold temperatures at 503 K and 516 K, respectively. The measured pressures fit to $1.30\times {10}^{-4}$ Pa and $8.34\times {10}^{-4}$ Pa as shown in Figure 5. As expected, thermionic emission is greater at higher source temperatures. The results match a pure tungsten curve at high filament temperatures.

The measured data do not match Taylor and Langmuir’s (TL) predictions. The filament we used had a temperature gradient due to thermal conduction at the leads. In contrast, the TL results were obtained by using guard rings to suppress collection from filament segments that were not uniform in temperature: electronic thermionic emission was nearly uniform along the fraction of the filament measured. A correction based on the temperature gradient was computed, described in Appendix A. The corrected curve can be seen in Figure 1 as the solid lines, and in Figure 5 and Figure 6. The corrected model agrees much better with the data.

As filament temperatures decrease towards that of the manifold, heat conduction into the filament becomes significant, altering the relationship between filament current and temperature. The gauge was operated at elevated temperatures (>200 °C, above 500 K for the measurements) to minimize cesium condensation; results at filament temperatures close to that of the manifold are omitted.

The proof-of-principle gauge failed after commissioning and two series of measurements due to cesium condensation impervious to evaporation. After the system had been opened and exposed to air, we verified a darkening of essentially all ceramic surfaces. Between electronic terminals, lighter patches appeared. This coloration pattern can be explained by cesium deposition and later vaporization between electronic terminals. When exposed to air, the remaining cesium darkened. For a more in-depth discussion on operation precautions and leakage currents, see Appendix B.

4. Discussion: Proposed Second Generation Gauge

The proof-of-concept gauge demonstrated the Taylor and Langmuir model can be used to measure cesium pressures, but made apparent that we had made mistakes in the design. The condensation of cesium vapor shorting circuit elements, and the presence of a filament temperature gradient are the main issues. An example second-generation instrument is illustrated in Figure 7 and described in Section 4.1. Alternatively, Springer and Cameron had proposed another method using a Bayard–Alpert gauge (BAG) to measure partial pressures of cesium using its ions instead of electrons [9,10], as described in Section 4.2.

4.1. Improved Gauge Using Thermionic Emission

In order to counter leakage currents and electrical shorts, the gauge must minimize areas for cesium deposition around relevant conductors such as the collector or the leads. All circuit elements must avoid contact with support structures if possible. One solution is to have the circuit elements be self-supporting as in Figure 7.

To account for the filament temperature gradient, the collector may be split into two guard rings and a central collector ring (Figure 7 and Figure 8). All rings should be cylindrical and biased at the same voltage to create a uniform radial electric field. Only the central guard ring would be used to measure thermionic emission as the center is the hottest part of the filament with the smallest gradients. This was done by Taylor and Langmuir [3].

The effectiveness of the guard rings is improved if the ends of the filament are coiled, increasing the length of the filament and thus flattening the temperature gradient. Although the emissions from the coiled areas will not follow the Taylor and Langmuir model, they will be shielded by the guard rings and not measured by the central ring.

4.2. Using Ionized Cesium over Electrons as an Alternative

An alternative method is to measure the current of ionized cesium ions instead of electrons thermionically emitted from the metal surface. A commercial Bayard–Alpert gauge (BAG) is an ion gauge with an emitter filament, a collector filament, and a cylindrical grid along the collector length. Springer and Cameron heated the grid to eject electrons and ionize cesium and biased the collector filament negatively to collect the ions. Measuring the collected current yields the cesium partial pressure in a similar way to the method described in this paper.

We had originally intended for the proof-of-principle gauge to be usable in ionic and electronic regimes. However, given the difficulties with cesium condensation, the electron regime was preferred. A description of a different ionization-based gauge and an additional method based on spectroscopic techniques are documented by Pradel, Roussel, and Spiess [11].

5. Conclusions

From the relationship between the thermionic emission of a tungsten filament and its temperature in the presence of cesium vapor, a gauge can be built to determine the partial pressure of the cesium. We have shown a proof-of-concept gauge and measurements, documented its problems, and proposed a new design. We hope this paper is useful to those who wish to construct a similar device or to explore the principles of thermionic emission, metallic crystal growth on metal surfaces, and work-functions.