Ion Microprojector

In a previous note,^3 we described experiments in which, with the aid of an electron microprojector, images of relatively simple molecules of phthalocyanine copper were observed. In these images it was possible to identify the molecules by their shape, comparing the photographs with data obtained by stereochemical and X-ray methods.

In the electron microprojector,^1,2 the hemispherical surface of a metallic tip is imaged with large magnification on a fluorescent screen. Electrons are torn from the metal in a strong electric field, whose intensity is about \(5 \cdot 10^7\ \text{V/cm}\). With a radius of curvature of the tip of the order of \(10^{-5}\ \text{cm}\), an anode voltage of several thousand volts is sufficient for this. The resolving power of such a projector^1,3 is on the average \(20\ \text{Å}\), i.e., it is somewhat better than the resolution practically attainable in modern models of electron microscopes.

At a distance from the tip to the screen \(R = 10\ \text{cm}\), the magnification is about one million. The circle of confusion corresponding to a point on

surface of the tip, has a diameter of \(2\) mm. According to Müller’s data\(^4\), this diameter can be expressed as

\[ D = 2R \sqrt{\frac{u}{U}}, \]

where \(u\) is the tangential initial velocity of the electrons, independent of the accelerating voltage \(U\) (in volts); \(u\) is usually about \(0.5\) V. The resolving power is also affected by electron diffraction; in the case of extraction of electrons by a field, it is rather difficult to take this influence into account quantitatively\(^5\).

An increase in the resolving power by increasing the field strength is hindered by the extremely rapidly increasing emission-current density, reaching \(10^8\) A/cm\(^2\) and leading to melting of the tip.

A significant increase in resolution may be expected if the electrons are replaced by positive ions, with a shorter de Broglie wavelength and thermal initial tangential velocities.

As early as 1941 it was shown\(^6\) that, by reversing the sign of the potential of the tip of an electron projector, it is possible, by increasing the field strength to

Fig. 1.

\(0.8\)—\(1.2 \cdot 10^8\) V/cm, to cause barium atoms to be torn from the tip on whose surface they had been adsorbed. If the particles leave the surface of the metal in the form of positive ions, they must move in straight lines toward the fluorescent screen and form on it an image of the surface of the tip, i.e. of those regions on which they had been adsorbed.

The current density in this phenomenon, called by Müller “field desorption,” is very small. In a monoatomic layer of barium there are about \(10^{14}\) atoms per \(1\) cm\(^2\) of surface; and since the surface of the emitting part of the tip is less than \(10^{-9}\) cm\(^2\), the total charge transferred upon removal of the ions is only \(10^{-14}\) coulomb. For an observable image to appear on the screen, the intensity must be approximately a million times greater.

This can be accomplished by rapid repetition of the processes of adsorption and desorption in an electric field. For this, it is not at all necessary to apply an accelerating voltage in pulses. As soon as neutral particles occupy places on the surface, they give up their electron, are desorbed as positive ions, and fly to the screen. Their paths almost coincide with the trajectories of electrons in a micro-

in a projector, with the difference that the tangential component of the velocity corresponds, in order of magnitude, to the thermal energy. As a result, the sharpness of the image increases considerably.

At present it is difficult to indicate elements whose ions would give the best results in such experiments. For this purpose hydrogen is convenient; the sizes of its atoms are smaller than those of atoms of the metal forming the tip; adsorption in this case takes place when a small quantity of hydrogen is admitted into the projector flask. The field strength at which desorption occurs, however, is in this case much higher than for barium.

Figure 1 shows the scheme of an ionic (proton) microprojector.^5 The image on the screen was observed in a magnifying glass or photographed by means of a high-aperture objective. Owing to the low intensities, better results should be expected with in-vacuum photography, when the protons bombard the plate directly. It is natural, however, that rubber and other seals used in “sleeves” for the plates of electron microscopes are inapplicable in the present case.

At first, after etching, the instrument was used with a negative tip, i.e. as an electron projector. It was then possible to judge the perfection of the tip surface. Figure 2 gives, through the limited aperture of the annular electrode, an image of the surface of a tungsten tip. On the surface traces of adsorbed carbon are visible, as can be judged from the light ring surrounding the central plane 110, and from the appearance of 334 planes around the octahedron poles.^7 Simultaneously with the carbon, hydrogen present inside the instrument during observation and exposure was adsorbed on the surface. At an anode voltage of 3.3 kV, a current of only \(10^{-5}\) ampere was emitted from the tip; according to the Fowler theory, \(\varphi = 4.8\) eV, and this corresponds to a tip radius \(r = 940\) angstroms. The field strength is \(5.4 \cdot 10^7\) V/cm. On a screen at a distance \(R = 45\) mm there arises an image magnified approximately

\[ \frac{R}{r} = 480\,000 \]

times. The radius of the circle of blurring (which determines the resolution), which cannot be determined sufficiently well from the photograph given by the author, is equal to 1.2 mm and corresponds to 25 Å on the object.

After changing the sign of the tip potential, the voltage between the tip and the annular electrode must be increased considerably before an image appears on the screen. When hydrogen was admitted, at about \(10^{-3}\) mm Hg and a potential difference of 12 kV, i.e. a field strength of \(2 \cdot 10^8\) V/cm, a very weak but sharp image appeared.

With a further increase in voltage the brightness increased, while the sharpness remained unchanged. In addition, the brightness was proportional to the pressure, which in the apparatus described could be raised to \(6 \cdot 10^{-3}\) mm without the occurrence of a glow discharge. The photograph shown in Fig. 3 was taken at 17 kV, i.e. \(2.8 \cdot 10^8\) V/cm, on a high-sensitivity film, at \(F = 1.2\) and an exposure of 30 sec. In comparison with the electron image the sharpness is very great; the author asserts that the diameter of the scattering circle was \(0.1\)—\(0.2\) mm, i.e. corresponded to 2—4 Å on the object. The photographic image, owing to interference during the exposure and possible fluctuations of the potential of the grains of the fluorescent screen, is considerably less sharp. However, the photograph shows that the light edging of the 110 plane has broken up into “steps.” The concentric diffuse light spot of large radius in the center arises from secondary electron emission from the annular electrode, caused by the protons falling on it, as is proved by deflection in a magnetic field.

The ionic image was quite high-contrast. The edges, surrounding the plane $110$, were detected, according to the authors’ statement, when observed visually and with good adaptation of the eye in darkness, as small, regularly alternating protrusions and steps, the distances between which are close to the constant of the crystal lattice. In order to eliminate possible errors due to the granular structure of the screen, the observations were repeated in another instrument with a magnification equal to $10^6$ ($R = 10\ \text{cm}$). In addition to the mentioned “steps,” numerous bright scintillating points were observed, not oriented with respect to the crystal lattice. It is possible that at these points there were lattice defects at which desorption occurred with greater ease.

Fig. 2.

Fig. 3.

The surface of one and the same single-crystal tungsten tip when imaged by electrons (Fig. 2) and by protons (Fig. 3).

The results described, concerning resolving power, correspond to what may be expected on the basis of considerations applied to the electron microscope. The tangential velocity of the imaging particle in this case corresponds to the energy of the maximum of the Maxwell distribution, i.e. $u = 0.04\ \text{eV}$ at room temperature.

The accelerating voltage in the example under consideration was $17\ \text{kV}$. The diameter of the scattering circle was $D = 0.14\ \text{mm}$, or $2.9\ \text{Å}$, i.e. almost 10 times smaller than in imaging by electrons. Upon cooling the entire instrument with liquid air, the scattering circle would decrease to $1.6\ \text{Å}$. The influence of electron diffraction in this case is insignificant, since the de Broglie wavelength for a proton with energy $0.04\ \text{eV}$ is $1.42\ \text{Å}$; at a velocity of 2 volts, reached at a distance of $3\ \text{Å}$ from the surface (taking into account the “image force”), the wavelength is only $0.2\ \text{Å}$.

At the attained pressure of $6 \cdot 10^{-3}\ \text{mm Hg}$ on the hemisphere of the ra—

with a radius of 940 Å, \(5.2 \cdot 10^9\) hydrogen molecules strike it per second. The ion current corresponding to twice the number of protons, as a result of dissociation of the molecules, may reach \(1.66 \cdot 10^{-9}\) A. In reality, under the conditions described, the current was \(6 \cdot 10^{-8}\) A. Evidently, it should be considered that, in a strong electric field, considerably more molecules reach the tip than follows from the gas-kinetic theory. Many of the molecules flying past may be attracted, as dipoles, in the strongly nonuniform field.

An important question in the field of “microscopy of atomic dimensions” is the action of the imaging particles on the object under investigation. Molecules captured by the tip strike it with a velocity exceeding the thermal velocity. If a particle characterized by polarizability \(\alpha\) is drawn from a region of zero field into a region where the field strength is \(E\), its energy is \(W = \frac{\alpha}{2} E^2\). In a field of \(2.8 \cdot 10^8\) V/cm, if one takes \(\alpha = 0.8 \cdot 10^{-24}\ \text{cm}^3\) for the hydrogen molecule, the energy of collision with the tip is \(\frac{1}{6}\) eV, which is insufficient to cause destruction of the tungsten surface.

If the field strength exceeds \((2.9 \pm 3) \cdot 10^8\) V/cm, the image on the screen becomes indistinct. It should be considered that in this case the appearance of ions occurs not on the surface of the tip crystal, but in the space surrounding the tip.

At field strengths between 2 and \(3 \cdot 10^8\) V/cm, heating the tip to 400–500° K leads to complete loss of image sharpness. It is possible that in this case the adsorbed atoms cannot occupy definite positions on the surface as a result of thermal vibrations. At higher temperatures of the tip (from 800 to 900°), the crystal lattice of tungsten is destroyed in the strong electric field.

The results of the experiments described apparently indicate that, with the aid of an ion microprojector, a resolution has been attained that makes possible the direct “seeing” of the crystal lattice of a metal. Of course, the data obtained are preliminary in character, since, owing to technical difficulties associated with intravacuum photography, the observations were made by the visual method. The author believes that good results may be expected by cooling the tip sufficiently strongly and by using as imaging particles ions of lithium, the desorption of which should occur in weaker electric fields, or helium ions. The study of metals whose crystal lattice is bound more weakly than that of tungsten is apparently possible only in still weaker fields at the tip.

It should be considered that the method described will be successfully applied to the study of the structural features of real crystals.

V. S. Vavilov

CITED LITERATURE

1. B. M. Tsarev, Electron projector as a method of physicochemical investigations, UFN 36, 181 (1948).
2. E. W. Müller, Zeits. f. Physik 106, 541 (1937).
3. V. S. Vavilov, UFN 42, 580 (1950).
4. E. W. Müller, Zeits. f. Physik 120, 270 (1943).
5. E. W. Müller, Zeits. f. Physik 131, 136 (1951).
6. E. W. Müller, Naturwiss. 29, 533 (1941).
7. E. W. Müller, Naturwiss. Rundschau, 255 (1951).
8. R. Haefer, Zeits. f. Physik 116, 604 (1940).