PROBLEMS AND METHODS OF VACUUM SPECTROSCOPY1

Gertrud Schoner.

Introduction

In all fields of natural science at the present time attempts are being made to find elementary processes and to understand them. Physics teaches us that spectral lines are a reflection of processes that take place in individual atoms or molecules. Therefore spectroscopy, with all its manifold applications, is now the most important auxiliary tool in the investigation of processes in the smallest particles of matter. The range of wavelengths in which spectral lines are found is vast,

Fig. 1.

but only an insignificant part of this range is perceived by our eye as light; in the remaining part we resort to optical, electrical, and thermal methods, which are entirely independent of the peculiarities of our eye. The accompanying drawing gives a survey of the various regions of the spectrum.

The whole spectrum embraces all possible natural frequencies of various configurations of the particles of matter. It is divided into four principal parts: the region of Hertzian waves, infrared waves, ordinary light waves, and X-ray waves. This division may be made according to two different principles: either according to the methods employed in the different wavelength regions, or according to the mode of origin

spectral lines, is not the same in different regions. Fortunately, both principles give almost coincident results. The portions in which two regions overlap deserve special attention.

Hertzian waves, the shortest of which have been obtained by Glagoleva-Arkad’eva and measured up to the present to \(0.1\ \mathrm{mm}\), arise in the oscillations of microscopic oscillators. The infrared part of the spectrum arises as a consequence of the vibrations of atoms and groups of atoms. This includes the intrinsic vibrations of crystals, caused by the vibrations of individual atoms and ions forming the lattice, and of bound lattices; it also includes the vibrations and rotations of molecules of polyatomic gases and liquids. It follows from the diagram that both regions—Hertzian and infrared waves—overlap. Waves of \(0.1\ \mathrm{mm}\) are the shortest electromagnetic waves that have so far been measured, and Rubens was able to trace the wavelength region of a quartz mercury lamp down to \(400\ \mu = 0.4\ \mathrm{mm}\). Here we first encounter a case in which identical frequencies can arise in different ways. The vibrations of bound lattices in a crystal—the lattice of anions relative to the lattice of cations—lie in the region \(30\text{–}150\ \mu\) \((0.03\text{–}15\ \mathrm{mm})\); vibrations of individual ions in the region \(2\text{–}40\ \mu\), rotational bands of the molecule approximately around \(100\ \mu\), and rotational-vibrational spectra in the region of several \(\mu\). We can thus establish the boundary on the side of short waves in such a way that those spectral lines which, though they lie in the region of infrared waves, arise in transitions between excited states of the atom, we shall not assign to the region of infrared waves. Thus we again encounter an overlap of the regions of infrared and ordinary light waves. In the same way, the methods applied in both regions also intersect. In the region of infrared waves the long waves are investigated by the method of residual rays and quartz lenses, the shorter ones by prisms and gratings; the intensity is determined by thermal methods.

We now pass to the region of ordinary light waves. I have divided it into four large parts: the infrared, the visible, the ultraviolet, and the region of vacuum spectroscopy. In all these parts, spectral lines or bands arise in transitions of electrons between different states of the atom or molecule. The states may be normal or excited, i.e. the electron may be located in an orbit of rest or in one of the outer orbits. In the case of a molecule, transitions of the electron may be connected with vibrations of the nuclei of the molecule or with rotations of the entire molecule. The infrared part of ordinary light waves is not specially marked in the diagram. Here again there occurs an overlap of two regions—different modes of production give rise to spectral lines of the same wavelength. In the visible region, in the only portion perceived by our

by the eye as light, lie, for example, the numerous lines of the iron spectrum and of other spectra of the iron group, as well as the lines of all elements that correspond to transitions between higher excited states. The first members of the principal series of the alkali metals are located, for example, also in the visible region, although, generally speaking, transitions between excited and normal states lead to lines of ultraviolet frequencies. The region of visible waves is small: it comprises only one octave, whereas infrared waves, for example, comprise approximately 12 octaves.

Spectroscopists work with glass optics; the detection of lines and the determination of their intensity are carried out photographically. However, beginning with the wavelength \(3600\ \text{Å}\), ordinary glass no longer transmits rays at all. Down to the wavelength \(3000\ \text{Å}\) ultraviolet glass is used; from there on one has to work with quartz optics. In this way the region of wavelengths is extended to \(1850\ \text{Å}\). In order to detect lines by the photographic method, special sensitized plates are used. In the ultraviolet region there are located, for example, the higher members of the principal series of the alkali metals, the subordinate series of the alkaline-earth metals, and many bands that are now well known, such as the bands \(N_2\), CN, CO. Beginning with \(1850\ \text{Å}\), on the side of short waves, experimentation becomes difficult. In this very region the oxygen of the air already begins to absorb \([^{2}]\), so that work by ordinary methods becomes impossible. It is necessary to place both the light source and the entire light path in a vacuum. Since we shall later speak about this region in detail, we shall now indicate only the meaning of the three subdivisions in our diagram. The interval usually called the Schumann region extends to \(1250\ \text{Å}\)—Schumann reached this point in his investigations; Lyman extended this region to \(510\ \text{Å}\), and the shortest wavelengths that have at present been attained by vacuum-spectroscopic methods, waves of \(136\ \text{Å}\), were obtained by Millikan and Bowen \([^{3}]\).

We now pass to the fourth division of our diagram, to the region of X-rays. Whereas the spectral lines in the part designated as the region of ordinary light waves are due to transitions of electrons between outer unoccupied orbits, X-ray lines arise because electrons from the inner filled shells of the atom are raised to the periphery of the atom, and the vacated places are filled by electrons from the neighboring shell lying farther outward. The region of X-rays investigated up to the present extends approximately from \(20\ \text{Å}\) to

0.1 Å. Since air again becomes transparent to X-ray waves, the need for laborious work in a vacuum disappears. The spectroscopically unexplored region from 136 Å to 20 Å is accessible for approximate measurements by the electron-impact method. Work with X-rays of longer wavelength is limited by the fact that at present no crystals are known whose constants would be suitable for the study of wavelengths above 13 Å. On the other hand, the fabrication of artificial gratings that would give distinct spectra for such short waves as 100 Å and less is very difficult. We shall return to this later ¹).

After this brief survey of the various regions of wavelengths, let us return once again to the subject of our article—vacuum spectroscopy.

Problems of vacuum spectroscopy.

The reason why at the present time attention is being directed more and more often to this experimentally difficult region is that it is hoped here to obtain answers to many questions concerning the structure of atoms and molecules, the transparency of solids, the photoelectric effect, and photochemistry.

The first attempt to penetrate into the region of wavelengths shorter than 2000 Å was undertaken by Schumann in Leipzig. In a number of excellent works [4] he systematically extended the accessible region of wavelengths down to 1250 Å. To him we owe the first study of numerous emission and absorption spectra of atoms and molecules in this region. The continuation of the work he began is due to Lyman. At the present time vacuum-spectroscopic work is carried out almost exclusively in America; I shall name above all Lyman, Millikan, Bowen, McLennan, and Tophild. The circumstance that in Germany Schumann had no successors is due chiefly to the fact that we are now compelled to avoid works that involve the expenditure of large technical means and financial outlays.

For the knowledge of the structure of the atom and the molecule it is necessary to study emission and absorption spectra. In the region accessible to study only by vacuum-spectroscopic methods lie, for example, the absorption series of the noble gases. At present the excitation potentials—

¹) L. Meitner and Ellis succeeded in extending the investigated region of wavelengths down to 0.02 Å by studying the velocity distribution of β-rays and, from this, with the aid of quantum theory, drawing a conclusion about the primary γ-rays. In the figure, however, these waves are not plotted, since they were detected by nonspectroscopic methods.

excitation and ionization of all the noble gases are known thanks mainly to the investigations of Franck \[16\] and Hertz \[17\], who used the method of electron impacts. Although their results led to very important conclusions—for example, in the case of helium to the discovery of metastable helium—nevertheless their numerical data cannot fully replace spectroscopic results, since the accuracy of these data is naturally less than spectroscopic accuracy. For the exact establishment of the fundamental states, optical observations are therefore inevitably necessary.

The exceptional experimental skill of Lyman \[8\] made it possible for the first time to find spectroscopic fundamental states in helium and to photograph resonance lines in the extreme ultraviolet region. His method will be discussed further on. Recently he extended his investigations to neon and argon \[9\]. At the same time Hertz \[10\] photographed the resonance lines of neon. From these investigations, in connection with other spectroscopic data concerning neon (Paschen, Meissner, Dorgelo, Goudsmit, Jordan \[11\]), there followed, as the fundamental term, a \(p\)-term with inner quantum number \(j=\frac{1}{2}\), and the next higher terms—four \(s\)-terms, of which two are metastable. The ultraviolet lines 735 and 743 Å, arising from the non-metastable terms, are the resonance lines of unexcited neon. Lyman indicated two corresponding lines in argon, and he assumes that spectrum A is constructed analogously to the Ne spectrum analyzed by Paschen. He also found lines that corresponded to the excitation potential of argon, indicated by Hertz, at 14.0 volts, so that now the task of bringing order into the argon spectrum, rich in lines, is already less uncertain. For Kr and Xe, corresponding vacuum-spectroscopic measurements are not yet available.

Just as the study of the line spectra of atoms, the study of spectra built on normal states, led to the establishment of fundamental states and absorption series, so too measurements of absorption bands may lead to analogous results for molecules. In what region do the absorption bands of molecules generally lie? Here it should first of all be established that a considerable majority of the bands that we know so far lie in the visible part and in the ultraviolet, and that they, as a rule, arise in quantum transitions between excited states of molecules. Few of them originate from the normal state, as, for example, the negative bands of nitrogen, which are built on the fundamental state of the molecular ion. With respect to the violet cyanogen bands, it has been suggested, for example, that the initial state for them is the normal state \[12\]. Probably,

a large portion of the bands that originate from the normal state is situated in the vacuum-ultraviolet region. These include, for example, the absorption bands of nitrogen and oxygen. Although it is of the highest importance first to analyze those bands that originate from the normal state, corresponding to resonance lines of the atom, and only after this to turn to bands depending on higher terms, precise measurements of absorption bands in the vacuum-ultraviolet region have as yet been carried out to only a very slight extent. Although Schumann, and subsequently Lyman and others \[13\], investigated the absorption of various gases in this region, they lacked a light source with a truly continuous spectrum in this region. Recently Hopfield and Leifson \[14\] measured the absorption bands of oxygen in the region 1240–1850 Å, using as a light source the continuous spectrum of hydrogen. Indeed, at Lewis’s institute (E. P. Lewis) it proved possible to obtain such a spectrum with sufficient intensity down to 1200 Å, so that it could be used as a source of a continuous spectrum in the vacuum-ultraviolet region. In view of the importance which this question has for vacuum spectroscopy, it is necessary briefly to indicate the possibility of obtaining other analogous light sources. Foote, Meggers, and Chenault \[15\], in a short note, report that by bombarding metallic films of copper, platinum, carbon, and iron with 1000-volt electrons they obtained a bremsstrahlung spectrum that could be traced through the visible and ultraviolet regions. It is quite possible that this spectrum in the vacuum-ultraviolet part possesses sufficient intensity for it to be used as a light source. Lilienfeld \[16\] described the gray-blue glow of the focal spot of an X-ray tube, as well as the visible part of the spectrum of electron bremsstrahlung. Apparently there exists still another possibility for realizing a source of a continuous spectrum for the vacuum-ultraviolet region. Anderson \[17\] was the first to point out that, in the explosion, for example, of a thin iron wire as a result of the discharge of a large capacitance through the wire, the iron lines appear as absorption lines on a continuous background. The latter arises as a result of the glow of the incandescent parts of the metal, extends to the limit observed in these experiments, 1850 Å, and in explosions in air is sufficiently intense, but is weak in explosions in vacuum. This is due to the fact that an explosion in an evacuated space entails rapid scattering of the individual particles, whereas high pressure confines the discharge. It would therefore be necessary to carry out the explosion in a vessel with a pressure of one or two atmospheres of He, Ne, A, or H₂, in order to obtain a continuous spectrum also in the vacuum-ultraviolet region. Since the temperature

during the explosion was estimated approximately at \(10000^\circ\) abs., it is possible that a larger number of explosions could give a sufficiently intense continuous spectrum in the vacuum region. Recently Lyman pointed out \([18]\) that, by discharging a capacity of approximately a quarter of a microfarad through a Geissler tube with helium, he obtained a continuous spectrum extending from 1900 to 900 Å.

From what has been said it is clear why investigations of emission spectra in the vacuum-ultraviolet region are far more numerous. The arc and spark spectra of many elements have been studied more or less in detail by numerous investigators (for example, McLennan and collaborators, L. and E. Bloch, Fowler, and others). Especially extensive results were obtained by Millikan and Bowen \([19]\). In a series of systematic investigations they set themselves the task of finding the most important series relations for atoms which, besides a shell of the noble-gas type, have only one electron or a larger number of them. In other words, they compare, for example, atoms with a \(K\)-shell and only one electron in the \(L\)-shell, such as \(\mathrm{Li}^{\mathrm{I}}, \mathrm{Be}^{\mathrm{II}}, \mathrm{B}^{\mathrm{III}}, \mathrm{C}^{\mathrm{IV}}, \mathrm{N}^{\mathrm{V}}\), or atoms with a \(K\)-shell and with 2, 3, and 4 electrons; or they considered atoms with completed \(K\)- and \(L\)-shells and with one, two, and a larger number of electrons in the \(M\)-shell (Stripped atoms). They confirmed for doublets with splittings into \(p\)-, \(d\)-, and \(f\)-terms the relativistic formula for X-ray doublets \(\Delta \nu = h (Z - s)^4\), where \(h\) and \(s\) are constants, and \(Z\) is the number of charges of the nucleus. Subsequently, with the aid of this formula, they drew the reverse conclusion from the doublets they had found concerning the carriers of the spectral lines. In the case of triplets they found that, for the frequency difference of the most widely separated pairs, a corresponding formula with somewhat larger values of \(s\) is applicable. Doublets with \(p\)-, \(d\)- and \(p\)-, \(s\)-terms they found, in accordance with the systematics of X-ray spectra, to correspond to irregular X-ray doublets. They succeeded, for example, in detecting \(pp'\)-groups in the spectra of the following elements with two valence electrons:

\[ \mathrm{Mg}^{\mathrm{I}},\ \mathrm{Al}^{\mathrm{II}},\ \mathrm{Si}^{\mathrm{III}},\ \mathrm{P}^{\mathrm{IV}},\ \mathrm{S}^{\mathrm{V}},\ \mathrm{Cl}^{\mathrm{VI}} \]

\[ \mathrm{Be}^{\mathrm{I}},\ \mathrm{B}^{\mathrm{II}},\ \mathrm{C}^{\mathrm{III}},\ \mathrm{N}^{\mathrm{IV}},\ \mathrm{O}^{\mathrm{V}}. \]

They also found the corresponding \(pp'\)-groups in the following atoms with three valence electrons:

\[ \mathrm{Al}^{\mathrm{I}},\ \mathrm{Si}^{\mathrm{II}},\ \mathrm{P}^{\mathrm{III}},\ \mathrm{S}^{\mathrm{IV}},\ \mathrm{Cl}^{\mathrm{V}} \]

\[ \mathrm{C}^{\mathrm{II}},\ \mathrm{N}^{\mathrm{III}},\ \mathrm{O}^{\mathrm{IV}}. \]

There is no place here to enter into a consideration of the numerous results of these works; it should be emphasized that our knowledge of atomic spectra has been greatly expanded thanks to these investigations. The methods that were used in this connection will be discussed in the following part.

With regard to the emission and absorption spectra of liquids and solids in the region of wavelengths shorter than \(2000\ \text{\AA}\), extremely little is still known. Even in the visible and ultraviolet parts these questions have not yet been clarified, owing to the insufficient number of experimental results and the great complexity of the phenomena. Likewise, little data are known on the transparency of solids that could be used theoretically.

In the field of photoelectricity there are also questions that can be solved only at wavelengths accessible to vacuum-spectroscopic methods. First of all, two questions should be mentioned. First, the question of the number of photoelectrically liberated electrons per unit of absorbed energy, and connected with it the second question, concerning the direction of the emitted electrons. We now know that, in the visible region, for the photoelectric surface effect the ratio of the number of liberated electrons to the number of absorbed light quanta is considerably less than unity (approximately about \(1/1000\)), and that it increases as the wavelength decreases. On the other hand, Gudden and Pohl [20], in the photoelectric effect inside crystals, found that this ratio is equal to unity. Investigations in the vacuum-ultraviolet region would make possible the corresponding measurements for gases, in which one may expect the absence of side effects. The question of the dependence of the direction of electron emission on the direction of the light vector in the photoelectric effect can be solved in exactly the same way, only by using extreme ultraviolet light, since for carrying out the measurements the magnitude of the light quantum must greatly exceed the work of liberation of the electron. When x-rays were used, where this condition is fulfilled, it turned out that the direction of emission coincides with the direction of the electric vector.

In conclusion of this section it should be briefly mentioned that in photochemistry there exist problems whose solutions may perhaps be brought about by investigations in the extreme ultraviolet region. It is enough to recall only the gas reactions of such gases as are transparent up to this region, above all the reactions of nitrogen.

These examples are probably sufficient to give some idea of the importance of experimentation in the vacuum-ultraviolet region. It is now necessary to dwell on the methods used in this region.

Methodology.

Experimentation in the extreme ultraviolet region is considerably more difficult than in other regions of the spectrum, for, as already mentioned, the light must travel the entire path in a vacuum or in the gas under investigation. Spectrographs—here we are speaking of fluorite spectrographs with a diffraction grating—are therefore subject to such stringent requirements. They must not be too large, so that they can be evacuated quickly; they must not admit air; their adjustment devices must, as far as possible, be such that the spectrograph need not be opened for every small change; plates must be replaceable as conveniently as possible; finally, the light source must be capable of being placed as close as possible to the slit. In the various designs these requirements are met in different ways. First of all we shall dwell on fluorite spectrographs. It is expedient to use them when large light-gathering power is needed and when one works with wavelengths not less than 1250 Å. In this connection it may be mentioned that, according to Abbe’s indication [12], all fluorite that is actually transparent down to 1250 Å comes from only one locality, near Lake Brienz. In 1832 these were rich deposits, where pieces weighing more than 100 centners were found. At the present time, unfortunately, we possess only small pieces of fluorite—the greater part of the precious material was bought by chemists for the manufacture of hydrofluoric acid. After, as a result of intensive searches, the old deposit was rediscovered, the Zeiss firm carried out investigations which showed that the preceding extraction had indeed been very thorough, since only a few small pieces suitable for the purpose could be found. Lyman [22] mentions in his book a light-green fluorite from Westmoreland (New Hampshire, U.S.A.), which is almost as transparent as the aforementioned colorless fluorite. The use of fluorite spectrographs also determined the limiting wavelength to which Schumann reached in his investigations. His spectrograph was technically so well designed that any adjustment could be made without admitting air into the spectrograph. In what follows, as an example of a new design, McLennan’s fluorite spectrograph [23] is described.

The metallic casing of the apparatus is made of brass, which at one end is closed by a cassette (Fig. 2). The latter has the form of a flat little box, into which a plate can be pushed from the side, after which the box is closed with a brass cover. \(A\) and \(B\) are metal tubes, fitted precisely to the main tube, with \(A\) ...

carries a washer with a fluorite lens, and \(B\) is a slit. The washer is, of course, drilled through so that pumping is possible. The connection of the light sources to the spectrograph is shown in the figure. In later work MacLennan \([24]\) described a somewhat different model, since the adjustment of the spectrograph just described each time required a great deal of time.

Fig. 2.

In the new design the prism, the second lens, and the cassette are placed inside one and the same cylinder of cast bronze, \(7.5\) cm high, with an internal diameter of \(30\) cm and wall thickness of \(1\) cm (Fig. 3). The cover has to be removed at every adjustment and when changing the plate. In the collimator tube \(K\), which is attached to the large vessel by means of a piece of cast bronze \(S\), a second tube \(H\) is inserted concentrically; it carries the slit \(G\) and the collimator lens \(E\). For pumping, the tube is provided with openings. To the ground joint \(T\) one may attach a vacuum lamp for exciting spark spectra, shown in Fig. 4. The arm \(P\), on which the lens \(D\) and the cassette \(AB\) are fastened, can rotate about an axis passing through the center of the prism. After adjustment the arm can be fixed by a setting screw. The supports for the lens \(D\) and for the cassette can, in addition, be displaced along \(P\). Moreover, the plate can be moved in its own plane. The small dish \(O\) contains \(P_2O_5\) and is covered with glass wool.

Fig. 3. Fluorite spectrograph of MacLennan.

Fig. 4. Spark discharge tube of MacLennan.

The arrangement for a spark in vacuum is shown in Figure 4. It is a glass sphere \(10\) cm in diameter with aluminum leads \(L\) and \(M\). \(A\) and \(B\) are pieces of the corresponding element whose spark spectrum is being investigated; \(K\) is a fluorite win-

rite. During the experiment the apparatus is filled with helium at atmospheric pressure.

The construction and adjustment of spectrographs with a grating is simpler. The slit, the photographic plate, and the grating lie on the circumference of one and the same circle, whose diameter is equal to the radius of curvature of the grating. The adjusting devices are usually arranged so that the grating can be rotated about a horizontal and a vertical axis and displaced along the axis of the spectrometer. Likewise, the slit and the holder for the plate are not fixed immovably. The position of the light source varies very considerably depending on the purpose in view. Milliken \[25\], who investigated spark spectra in high vacuum, placed the light source in the spectrograph itself in front of the slit. MacLennan’s vacuum spark, already shown in Fig. 4, was used by MacLennan not only for the fluorite spectrograph but also for a vacuum spectrograph with a grating \[26\], but without the fluorite window; in this case the relatively large distance of the light source from the slit was more inconvenient than in the former. The vacuum spectrograph with a grating recently described by Lyman \[27\] eliminates this drawback. It is given as an example in the following figures (Figs. 5 and 6).

Fig. 5. Holder for photographic plates of a vacuum spectrograph.

The grating, measuring \(5 \times 3\) cm, has a radius of curvature of 50 cm and is placed in a brass tube with an internal diameter of 9 cm, which is provided with flanges at both ends. The side on which the grating is located is closed with a round plate and sealed with soft glue. This side is not shown in the drawing. The grating can be installed in the usual manner. On the other side the joint between the flange and the closing plate is sealed with Khotinsky cement. The holder for the plate consists of a special tube \(C\) (Fig. 6), which is fitted to tube \(L\) (Fig. 5). At one end it is provided with a handle, which—

Fig. 6. Same. Detail.

... allows it to be moved, and at the other end there is a plate \(C\) with cutouts for photographic plates measuring \(4 \times 1\) cm. In order to make it possible to give the plate various inclinations, \(C\) permits rotation about a vertical axis. The cylinder \(L\) (Fig. 5) is clasped at the top by the ring \(W\), and the intervening space between \(L\) and \(W\) is filled with a certain amount of sealing wax. The cover \(K\) has a cylindrical attachment, which fits exactly into the space between \(L\) and \(W\). By heating, the sealing wax is melted and, after cooling, closes the gaps so as to be impermeable to air. The attachment \(T\) contains the slit, and the light source is connected directly to it, so that it is situated close to the slit. The advantage of this spectrograph, which Lyman emphasizes, consists in the fact that, without particular difficulty, it can be made absolutely impermeable to air.

Another construction is used by Gopfield [28]. This is a spectrograph, built by Lewis, with a grating having a radius of curvature of 50 cm, which is placed not in a long tube, as usual, but in a cast-brass box in the form of a sector. Thus more room remains for adjustment devices than with the usual form of spectrograph in the shape of a tube. In the vacuum spectrograph constructed by Hilger [29], the plate holder is fastened by means of a ground joint in a special tube which emerges from the front plate. Two slits are located in the same tube beneath the plate holder. Recently G. Hertz [30] constructed a vacuum spectrograph with a grating, using very simple means. The details of the construction will be published by him in the near future.

A few words should also be said about gratings. Gratings used in the visible and ultraviolet regions are, generally speaking, unsuitable for the vacuum-ultraviolet region, since small irregularities in the grooves of the grating are so harmful that light of very short wavelength is scattered diffusely. For longer waves these irregularities are not significant, since they are small in comparison with the wavelength. Thus, for gratings used in the vacuum-ultraviolet region, accurate ruling and good material are important. With the usual number of lines—500–1100 lines per mm—the ratio of the distance between the lines to the wavelength for very short wavelengths is approximately \(100 : 1\). Millikan, Bowen, and Sawyer [31] scratched their grating very lightly and superficially, so that part of the surface remained undamaged. With this method it is much easier to obtain perfectly uniform lines. They found, at a certain ratio of the surface of the grating to the surface of the grooves, that the principal intensity falls on the central image and on the first order. Recently, using the same method, Anderson [32], with very lightly scratched gratings...

obtained excellent spectra in the 7th, 8th, and even 10th order. With these gratings Millikan and Bowen obtained their extensive results concerning the spectra of multiply ionized atoms. Lyman also finds the method of shallow ruling of gratings very advantageous for the vacuum-ultraviolet region.

For spectral photographs of the vacuum-ultraviolet region, Schumann plates are usually used (plates free of gelatin, since gelatin, beginning at 2000 Å, absorbs ever more strongly as the wavelength decreases). If one adheres exactly to the method indicated by Schumann [33], then it is not difficult to prepare these plates oneself in the laboratory. The method indicated by Duclaux and Jeantet [34], consisting in the use of an extremely thin layer of oil applied to an ordinary photographic plate, is also very convenient and, owing to its simplicity, may be especially recommended. Hopfield [35] uses in his work specially prepared films, which have proved to be very good.

We shall now consider, by means of several examples, how investigations in the extreme vacuum-ultraviolet region are actually carried out. Lyman [36] investigated the spectra of helium in a series of papers. In his first investigations the light source and the vacuum spectrograph were filled with helium at a pressure of approximately 1 mm and connected to one another by a slit. An ordinary Geissler discharge served as the source. With this method of excitation many lines appeared which evidently were due to contamination. This is not surprising, since a noble gas is extraordinarily sensitive under such a method of excitation. Therefore, as the light source there was used a tube with hollow cylindrical electrodes, which, according to Paschen, is advisable in the present case [37], and the observation was made in the longitudinal direction. The tube was supplied with direct current. With this arrangement it was possible almost completely to avoid lines due to contamination, and it proved possible to assign to series certain far-lying lines of helium. Since the apparatus was entirely filled with helium, the absorption lines of helium along the long path from the light source to the plate were again absorbed and therefore were not observed. To avoid this, Lyman continuously admitted helium into the space occupied by the light source. The gas diffused through the narrow slit into the spectrograph, but from there it was continuously pumped out by a fast-acting pump: a method which Wien repeatedly used in his investigations of canal rays.

The arrangement chosen by Lyman is ultimately shown in the following figure (Fig. 7). The spectrograph consisted of a tube of hea-

made of brass with an internal diameter of 14.9 cm. At one end, closed by a brass plate, a grating was placed; this end is not shown in the drawing. At the other end there was a fitting with a plate \(A\), into which the tube \(B\), containing the slit, entered. Further, a brass box \(C\) with a holder for the plate was connected to \(A\); the holder was mounted on the brass plate \(D\). The open end \(C\) terminated in a flange with a closing plate \(E\), and, for better air-tightness, the connecting ring was provided with soft glue. \(W\) is a glass plate which during the exposure was tightly closed. The holder for the plate could be moved in the horizontal and vertical directions, and rotated about a vertical axis. The vertical displacement was effected by means of an electromagnet located above \(C\), so that several photographs could be taken on one plate without opening the spectrograph each time. Light from the grating fell on the plate through the slit \(F\), 2 mm wide and 7.8 cm long. The plate \(A\) had a corresponding slit as well. This slit was connected with a rectangular tube \(H\), which in turn was connected with a second similar tube with a diaphragm \(K\). The slit \(S\) was 0.16 mm high and usually 0.04 mm wide. The fitting \(Y\) led to a pump so that the gas entering through the slit was immediately pumped out.

Fig. 7. Lyman spectrograph.

The remaining part of the spectrograph was evacuated through a tube symmetrically placed in its middle. In order, as far as possible, to avoid absorption in the vessel with the light source, a capillary tube was introduced, which approached as close as possible to the slit. For evacuation there was a special connection with the pump. During the discharge the gas flowed in through a very fine capillary. In this way a pressure of approximately 1 mm was maintained throughout. The discharge was in most cases produced by a direct current, whose strength was 15–20 milliamperes. For the investigation of the spark spectrum a hollow cathode of Paschen was used. Since, in studying a new region of the spectrum, the appearance of false lines (so-called “ghosts,” due to the periodic repetition of irregularities of the grating) was especially dangerous, Lyman photographed the same spectra with several gratings, which

which had been made by different methods. By then comparing the spectra obtained with these gratings, he was able to distinguish true lines from false ones.

A similar method of avoiding absorption by maintaining a high vacuum in the spectrograph was used by Herz in the investigation of the absorption lines of neon [38].

Millikan and his collaborators chose a different arrangement for studying the spark spectra of the elements. As already mentioned, they placed the light source in the spectrograph directly in front of the slit, i.e., put it into the high vacuum. In doing so they used a spark of very great energy, which jumped between electrodes placed at a distance of 1 mm or less. The leads were introduced into the spectrograph. During the exposures, vigorous pumping was carried out continuously. Since the sparks were very hot, the leads, electrodes, and insulation became so heated that they gave off much gas. Therefore, after a one-second exposure it was necessary to wait 5′ in order to pump out again the gas that had been liberated. For taking a photograph, a full exposure of 5–20′ was required, depending on the electrodes. From this alone the great luminosity of the spark is evident. It was very difficult to avoid fogging of the plates. This was alleviated above all by allowing only direct light to act on the plate by means of diaphragms. The greatest difficulty, however, was presented by the fact that during the exposures the grating became covered with a deposit, as a result of which its quality was greatly diminished. By prolonged experimentation the authors established that the deposit on the grating consisted of sulfur and sulfur compounds, which were released from the ebonite insulation and ebonite rings as a result of heating. After the insulation of the leads was made of synthetic amber, and the rubber gaskets of a material free of sulfur, the lifetime of the grating increased greatly, the evolution of gas decreased, and at the same time clear negatives were obtained.

Since we cannot dwell here on the various methods employed by different authors, it remains to mention, by way of example, the investigation of McLennan and collaborators [40], as well as the work of Hopfield [41], in which he describes an oil-cooled discharge tube fed by high voltage, the use of which greatly reduced the exposure time.

What has been set forth in this article gives an idea of the methods used in this region, so difficult to access experimentally, and, perhaps, makes it possible to understand why this region is attracting ever-increasing interest.

Bibliography

1. A. Glagolewa-Arkadiewa, Zeitschr. für Phys. 24, 153, 1924.
2. V. Schumann, Smithsonian Contribution to Knowledge 29, No. 14, 1903. L. et E. Bloch, Comp. Rend. 158, 1161, 1914; E. Schmidt, Zeitschr. für Physik 31, 475, 1925.
3. R. A. Millikan and J. S. Bowen, Phys. Rev. 23, 1, 1924.
4. For the literature see the review by Franck, Phys. Zeitschr. 22, 1921. Also: Compton and Mohler, Bull. of the National Research Council, Critical Potentials, 1924.
5. For the literature see Th. Lyman, The Spectroscopy of the Extreme Ultraviolet, 1914.
6. J. Franck und P. Knipping, Phys. Zeitschr. 20, 481, 1919; Zeitschr. für Phys. 1, 320, 1920. J. Franck und F. Reiche, Zeitschr. für Phys. 1, 154, 1920.
7. G. Hertz, Zeitschr. für Phys. 22, 18, 1924; G. Hertz und K. R. Kloppers, Zeitschr. f. Phys. 31, 463, 1925.
8. Th. Lyman, Phil. Mag. 41, 814, 1921; Science 56, 167, 1922; Astroph. Journ. 60, 1, 1924.
9. Th. Lyman and F. Saunders, Nature 116, 358, 1925.
10. G. Hertz, Naturwissenschaften 13, 489, 1925.
11. F. Paschen, Ann. d. Phys. 60, 405, 1919; K. W. Meissner, Ann. d. Phys. 76, 124, 1925; P. Jordan, Zeitschr. für Phys. 31, 877, 1925.
12. K. Mulliken, Nature 114, 858, 1924.
13. See 5.
14. S. J. Hopfield and S. W. Leifson, Phys. Rev. 25, 716, 1925.
15. P. D. Foote, W. F. Meggers and K. L. Chenault, Journ. of Opt. Soc. Am. 9, 541, 1924; 10, 287, 1925.
16. J. Lilienfeld, Phys. Zeitschr. 20, 280, 1919; 21, 249, 1920.
17. J. A. Anderson, Astroph. Journ. 51, 37, 1920.
18. Th. Lyman, Astroph. Journ. 60, 1, 1924.
19. J. S. Bowen and R. A. Millican, Phys. Rev. 24, 209, 1924; 25, 590, 1925; 26, 150, 1925.
20. B. Gudden und R. Pohl, Phys. Zeitschr. 23, 417, 1922.
21. Ernst Abbe, Ges. Abh. Bd. 1, p. 484, 1909.
22. Th. Lyman, Spectroscopy etc., p. 58, 1914.
23. J. C. McLennan, D. C. Ainslie and D. S. Fuller, Roy. Soc. Proc. A, 95, 316, 1919.
24. J. C. McLennan and A. C. Lewis, Proc. Roy. Soc. London, A, 98, 109, 1920.
25. R. A. Millikan and B. A. Sawyer, Phys. Rev. 12, 168, 1918.
26. J. C. McLennan, Proc. Roy. Soc. A, 98, 115, 1920.
27. Th. Lyman, Journ. Opt. Soc. Am. 7, 495, 1923.
28. J. J. Hopfield, Phys. Rev. 20, 573, 1922.
29. Simeon, Proc. Roy. Soc., A, 102, 484, 1923.
30. G. Hertz, Physica 5, 159, 1925.
31. R. A. Millikan, J. S. Bowen and R. A. Sawyer, Astroph. Journ. 53, 150, 1921.
32. J. S. Bowen and R. A. Millikan, Phil. Mag. 48, 259, 1924.
33. V. Schumann, Ann. d. Phys. 5, 349, 1901.
34. Duclaux et Jeantet, Science 58, 48, 1923.
35. Cf. 28.
36. Cf. 8.
37. F. Paschen, Ann. d. Phys. 71, 142, 1923.
38. Cf. 10 and 30.
39. R. A. Sawyer, Astroph. Journ. 52, 286, 1920.
40. McLennan and collaborators, Trans. Roy. Soc. Can. 1920—1925.
41. Cf. 28.

1. Naturwissenschaften, 14, 356, 1926. ↩