Full Text
NEW INSTRUMENTS AND METHODS OF MEASUREMENT
NEW METHODS IN MICROSCOPY
E. V. Shpolsky
I. THE PROBLEM OF CONTRAST IN MICROSCOPY
Microscopy has made significant progress over the last 15 years. The increase in the limit of resolving power achieved through the use of electron microscopes made it possible to move to enormous magnifications and opened up the broadest new field for research. However, resolving power is not the only factor that limits the possibility of using a microscope. Another no less essential limiting factor is insufficient contrast or the complete absence of contrast. Until recently, even the resolving power of optical microscopes with visible light was used far from fully precisely because of the difficulties caused by the poor contrast of many microscopic—especially biological—objects.
Indeed, the possibility of distinguishing the structure of the object under examination in optical microscopy is connected with the different absorption of light by the structural elements. But biological objects, in the overwhelming majority of cases, are practically completely transparent within the visible spectrum. This forced microscopists to resort to differential staining of their preparations, which was by no means always successful, quite apart from the fact that staining did not permit observation of living objects.
It should be noted that an analogous difficulty was also encountered in electron microscopy, although for a different physical reason. If contrast in the case of the optical microscope is due to differences in the absorption spectrum of the elementary regions of the preparation bordering one another, then in electron microscopy the contrast of a preparation “illuminated” by a beam of electrons is produced by the unequal scattering of electrons in different parts of the object. But the scattering power of the light atoms of which organic molecules consist is in general small. It is therefore clear that the limit of distinguishability of details in electron microscopy is determined not so much by the resolving power of the microscope as by contrast, i.e., ultimately
in counting—the amount of material in the individual structural elements. It is necessary, moreover, to bear in mind that the specimen is placed on a supporting film (collodion or a plastic mass—the so-called Formvar), which also contains a certain amount of material that scatters electrons. It follows from this that a detail cannot be distinguished if it scatters electrons only slightly in comparison with the substrate.
One way of overcoming this difficulty in electron microscopy consists in using a technique analogous to the staining of biological specimens, with the difference that in this case the “stains” are substances with a high scattering power. Another, apparently especially fruitful method, consisting in coating the specimen with the thinnest layer of metal in an oblique direction (“metallic shadow casting” in the terminology of American authors), will be considered in Section IV of the present article.
As for optical microscopy, various ways of improving contrast have repeatedly been indicated. The simplest of them consists in reducing the aperture of the condenser diaphragm, insofar as the decreasing illumination permits this. This method gives very limited results and is disadvantageous because it is associated with a lowering of the resolving power of the microscope.
The second possibility consists in the use of oblique illumination. In this case a peculiar shadow effect arises: the observer receives an impression of relief; he sees, as it were, slight elevations on a white surface. In principle, the well-known Töpler “schlieren method,” used to detect phenomena connected with nonuniformity of the refractive index in macroscopic objects, is identical with this method of oblique illumination. In microscopy the method of oblique illumination was known long before Abbe. Its diffraction justification, however, was given only in 1934 by Zernike¹ (a few words about this justification will be said later). In Fig. 1a is shown the profile of an object that is perfectly homogeneous and colorless, and in Fig. 1e—the microscopic image calculated by Zernike under oblique illumination*). As can be seen, the image does not correspond exactly to the object; its contours are too rounded.
Among other methods of the same type let us indicate: a) the method of defocusing the objective and b) the dark-field method.
From consideration of Figs. 1e, c, f, and g it is seen that the best results are given by the dark-field method with central illumination. However, it too has substantial shortcomings. For example, as is seen from Fig. 1g, the contrast obtained is greatly exaggerated. In some
*) Zernike’s work also contains a diffraction justification of the schlieren method. As Zernike indicates, the interpretation of this method as a kind of dark-field method is erroneous and could have appeared only as a consequence of the absence of a diffraction theory.
) F. Zernike, Zschr. techn. Physik 16**, 454 (1935).
in some cases this is required, whereas in others an exaggeration of contrast may lead to errors.
An entirely different method of increasing contrast is based on the use of the ultraviolet part of the spectrum. This use is based on the fact that (apart from a certain increase in resolving power owing to the decrease in wavelength) it is precisely in the ultraviolet that the absorption bands of organic—and especially biologically important—substances lie.
Fig. 1. Calculated intensity distribution in the image of a phase grating: a) grating profile (phase difference between elevations and depressions 30°); b) central illumination (grating in front of the focus); c) the same, 8× farther beyond the focus; d) strip method; e) relief under oblique incidence of light; f) dark field, lateral illumination; g) the same, central illumination; h) positive phase contrast; i) negative phase contrast. The same scale of ordinates in b), c), d), h), and i); in f) and g) it is enlarged 15× (after Czernike).
All who have worked in the field of absorption spectroscopy are also well aware that these ultraviolet bands of organic substances are extremely strong. One may therefore say that a biological specimen in the ultraviolet has its own coloration, not inferior in strength to the artificial coloration achieved with the aid of dyes. The difficulty of ultraviolet microscopy and its limited prevalence until recently were connected with the impossibility of constructing, for ultraviolet, optical systems achromatized over a broad spectral interval. In the ultraviolet microscope produced by Zeiss according to Köhler, the quartz-fluorite achromatic systems employed...
are corrected only for one wavelength, in view of which a necessary accessory of the microscope is also a quartz monochromator. It goes without saying that this complicates the apparatus to an extreme degree and, moreover, does not permit the most advantageous use of the distribution over the spectrum of the absorption bands of the specimen.
An exceedingly ingenious way out of these difficulties of ultraviolet microscopy was found by E. M. Brumberg\(^3\) at the State Optical Institute in Leningrad. E. M. Brumberg, together with Gershgorin,\(^4\) first of all calculated and constructed a reflecting objective for the ultraviolet with an aperture of 0.5. Since chromatic aberration is absent in reflecting systems, the difficulty of constructing an achromatic objective over a wide interval of the ultraviolet spectrum was thereby overcome. However, the contrast due to the ultraviolet absorption spectrum of the specimen is used in Brumberg’s method in an especially ingenious way: by means of specially selected light filters (which in the ultraviolet is already in itself a difficult task), microphotographs are made at three different wavelengths. The three microphotographs obtained are, when viewed, placed in a projector—a chromoscope with three light filters, red, green, and blue; the chromoscope projects three differently colored images onto one place—the screen—and thus the contrast present in the ultraviolet is reproduced in the form of a brightly colored picture (of course, in arbitrary colors) in the visible part of the spectrum. The effect produced by this technique is striking, and it undoubtedly has a great future.
II. ABBE’S THEORY
Alongside the methods for increasing contrast in optical microscopy considered in the preceding section, in recent years a technique has begun to come into use called the method of phase contrast, or phase microscopy. From the theoretical side, this method is based on certain aspects of Abbe’s theory which until recently had escaped the attention of investigators. The merit for substantiating this method belongs to Zernike,\(^{1,2}\) and its further development was carried out by Köhler and Loos,\(^{5,7}\) while the Zeiss firm, not long before the beginning of the war, issued technically finished apparatus for the application of this method.\(^6\)
Let us briefly recall the basic features of Abbe’s theory of the microscopic image. According to this theory, in the form in which it was quantitatively developed by Lummer and Reiche\(^*\), the image of a non-self-luminous—
\(\ ^*\) Abbe developed the theory of the microscopic image, but did not publish it. After his death, Lummer and Reiche gave a completed exposition of Abbe’s theory on the basis of a lecture delivered by him in 1888 in private for a narrow circle of listeners. In doing so, Abbe, for calculating the intensity in the image plane, used the Huygens–Fresnel principle.
...objects in a microscope is regarded as a double diffraction pattern, arising successively at the object and at the objective. This leads to a twofold application of the Huygens–Kirchhoff principle, with the sources in the first case lying in the plane of the object, and in the second—in the aperture of the objective. As a result, in order to calculate the amplitude in the image plane one has to carry out a double integration, extended successively over both of these planes.1
A physical interpretation of this computational procedure can be given in two ways. The first was indicated by Rayleigh,[^10] the second is, in fact, the physical substance of Abbe’s theory. These two interpretations, however, are entirely equivalent and, as Zernike[^2] noted, are different visual interpretations of one and the same double integral. To explain the method of phase contrast that interests us, it is expedient to use Abbe’s interpretation. The general scheme of this interpretation, as is known, consists in the following: first the diffraction of the light sent by the source at the object is considered. The diffraction pattern that arises is called the primary image. The image of the object itself (the secondary image) is then explained as the result of interference of coherent waves issuing from different points of the primary image.
Fig. 2 explains this for the example usual in Abbe’s theory—the object being a diffraction grating. The source here is the aperture of the condenser diaphragm \(EP\); it is located at the principal focus of the condenser, which therefore sends a parallel beam onto the object. On passing through the object there arise diffraction beams \(0, +1, -1, +2, -2\), etc., which give, in the exit pupil \(AP\) of the objective, a series of diffraction images of the source \(S_0, S_1, S_{-1}\). These diffraction images are regarded as new sources of coherent waves, whose interference gives, in the plane conjugate with the object, the image of this object. The accuracy of the image, according to this theory, depends on the number of diffraction beams entering the objective (in Fig. 2 only the spectra of the zero and first orders enter the objective).
The fundamental law of the theory in Lummer’s formulation is as follows: “if all the rays diffracted by the object that still have appreciable intensity enter the aperture of the imaging system (objective), then the system produces, in the plane conjugate with the non-self-luminous object, an image exactly similar to the object in structure and phase.” If, however, the aperture does not encompass all
of beams of sufficient intensity, or if part of the beams entering the aperture is deliberately stopped down, then the exact similarity of the image to the object is lost to a greater or lesser degree.
In Abbe’s theory, the center of gravity is concentrated precisely on these questions of similarity and on the related question of the limiting resolving power of the microscope. Cernike, on the contrary, drew attention to another side of the theory, namely to the question of phases. As for the question of resolving power, for the theory of phase contrast it is not essential, and one may even, together with Cernike, assume that the ideal case obtains as a first approximation, i.e., that the aperture captures all the diffraction beams and an image arises that is exactly similar to the object.
Fig. 2. On Abbe’s theory.
In conclusion of this section, let us note that the diffraction theory of contrast under oblique illumination mentioned above is based on the following consideration. If the illuminating beam is directed so obliquely that it can only just still pass through the object, then only spectra located on one side of the central diffraction image enter the aperture. As a result, an asymmetry arises, which gives the picture shown in Figs. 1d and 1e. In exactly the same way, according to Cernike, the microscopic picture in a dark field in the case of oblique illumination is explained by the disappearance of diffraction spectra lying on one side, and under central illumination by the disappearance of the central image.
III. THE PHASE-CONTRAST METHOD
At the beginning of this article it was indicated that a feature of biological objects unfavorable for microscopy consists in the fact that these objects, generally speaking, do not exhibit selective absorption of light in the visible part of the spectrum—they are colorless. The only difference that exists in the light passing through different places of the specimen is due either to a difference in the thickness of these regions, or to a difference in their refractive indices. This difference, therefore, consists in the fact that the phase
light vibrations in individual parts of the specimen undergo unequal change. We may, in general, quite roughly and schematically divide microscopic objects into two classes. The first class includes objects in which the individual structural elements have unequal absorption of light. In passing through such an object, the amplitude of the light wave changes in different ways. The second class includes objects that do not absorb light, i.e. that do not change its amplitude, but do change its phase. An ideal representative of objects of the first class will be, for us, a grating with alternating lines of greater and lesser transparency. Following Zernike, we shall call such a grating an amplitude grating. As an ideal representative of objects of the second class we shall consider a phase grating, i.e. a grating made of a homogeneous, perfectly transparent substance with alternating regions of greater and lesser thickness. The profile of such a grating is shown in Fig. 1a.
If we place a phase grating on the stage of a microscope, then in the exit pupil of the objective there will arise the same Fraunhofer diffraction spectra as in the case of an amplitude grating. Thus the primary image in both cases will be the same. The secondary image, however, cannot be the same, since, according to the Abbe–Lummer law of similarity, the ideal secondary image must coincide with the object in structure and in phase, i.e. in the case of an amplitude grating it must everywhere have the same phase but different amplitude, while in the case of a phase grating it must have the same amplitude but different phase. It is evident that this is precisely connected with the fact that, in the case of an amplitude grating, we distinguish gradations of illumination, whereas in the case of a phase grating we see a uniformly illuminated field. The fact that, as a result of the interference of spectra having exactly the same form (the primary image), such different secondary images are obtained is undoubtedly connected with the inequality of phases in the primary image in the case of the phase grating. Recognition of this circumstance is just what enabled Zernike to indicate a method for obtaining contrast in the case of a phase grating.
For a qualitative explanation of the phase-contrast method, Zernike uses vector diagrams analogous to those employed in the theory of alternating currents. In their paper, Köhler and Loos\(^6\) give a very clear drawing constructed by them, which we reproduce with the corresponding explanations in Fig. 3.
In Fig. 3 (Ia and Ib) two gratings are shown—an amplitude grating (Ia) and a phase grating (Ib). Each of them is superposed on a plane-parallel homogeneous, perfectly transparent plate. Above the gratings are shown the corresponding vector diagrams, representing the state of the light vibrations after passage through the grating—
grid. In the case of an amplitude grating, the vectors over places of lower transparency have a smaller length, but the same phase as the vectors over places of greater transparency; in the case of a phase grating
Fig. 3. Explanation of phase contrast by means of vector diagrams (after Zernike-Köhler and Loos).
Ia. Section of an amplitude grating. After passing through the grating, the vector over the strip (on the right) indicates the change in amplitude while the phase is preserved.
Ib. Section of a phase grating. The vector over the strip preserves the amplitude but changes the phase.
Ic. Diffraction spectrum of an amplitude or phase grating. The zero-order maximum is decomposed into two components, one of which, for clarity, is displaced somewhat upward. \(P_{90^\circ}\) is a phase plate that changes the phase of the zero maximum by \(90^\circ\).
IIa and IIb. Vectors representing the state of the light oscillations after passage through an amplitude (IIa) and a phase (IIb) grating, decomposed into two components, of which one is equal to the vector over the slits.
IIIa and IIIb. The larger components of the preceding decomposition, shown separately. Beneath them is the section of a plane-parallel plate which creates the state of light oscillations represented by these vectors.
IIIc. Diffraction spectrum of this plate, consisting of a single zero-order maximum.
IVa. The smaller components of the decomposition IIa. Below is a section of the amplitude grating and of the phase plate \(P_{180^\circ}\), which, acting together, create the state represented by the smaller components of the decomposition IIa.
IVb. Analogous representation for the smaller component of the decomposition IIb. Below is a section of the corresponding amplitude grating and of the phase plate \(P_{90^\circ}\).
Note: \(P_{180^\circ}\) and \(P_{90^\circ}\) differ in that IVa and IVb differ only by phase plates (\(P_{180^\circ}\) in one case and \(P_{90^\circ}\) in the other).
IVc. Diffraction spectrum created by these amplitude gratings and phase plates. Vb. The vectors IIIb are rotated by \(90^\circ\) by the phase plate \(P_{90^\circ}\), placed over the zero maximum.
VIb. Results of adding the vector diagrams IVb and Vb. The resultant vectors represent the state of the light after passage through the phase grating plus a phase plate placed at the position of the zero diffraction maximum. Below is the amplitude grating corresponding to these vectors. Negative phase contrast.
VIIb. Analogous to the preceding: realization of positive phase contrast.
of the grating, the vector representing the state after passage through the thickening (or the place with a different refractive index) is somewhat rotated, but its length remains unchanged. We shall call the places of lower transparency in the case of an amplitude grating, and the thickenings in the case of a phase grating, strips, and the intervals
between them—the slits. Let us now decompose the vectors over the strips of the amplitude and phase gratings into two components in such a way that one of them, in magnitude and phase, is equal to the vectors over the slits. This is done in Fig. 3, IIa and IIb. In the case of the amplitude grating (IIa), the vector representing the difference differs in phase from the corresponding vector of Fig. Ia by \(180^\circ\), and in magnitude is equal to the difference of the lengths of both vectors in Fig. Ia. In the case of the phase grating (Fig. IIb), the difference vector will be the smaller in magnitude, and in direction the closer to \(90^\circ\), the smaller the phase shift upon passage of light through the strips of the phase grating. As for the vectors over the slits of both gratings, they remain unchanged, or, in other words, are decomposed into two components, one of which is equal to the original vector and the other is zero. In Fig. 3, IIIa and IIIb, both large vectors are once again represented, and in Figs. IVa and IVb—both smaller ones. Comparing these Figs. III and IV with Fig. I, we see that both gratings—the amplitude and the phase grating—may be regarded as composed, in both cases, of identical homogeneous plane-parallel plates, which do not in general cause diffraction, and of two gratings in which the places corresponding to the slits are completely transparent, while the places corresponding to the strips absorb light strongly. In order, however, that the state of the light after passing through these gratings be represented by the vector diagrams shown above them, it is also necessary to change the phase—by \(180^\circ\) in the case of grating IVa and by \(90^\circ\) in the case of grating IVb. In Figs. IVa and IVb this is represented by means of plates \(P180^\circ\) and \(P90^\circ\), producing the corresponding phase change.
These drawings are very instructive; they show that the whole difference between the gratings—the phase and the amplitude grating—reduces, in the final analysis, precisely to an inequality of phases. If in case IVa, in front of the grating, instead of the plate \(P180^\circ\), one were to place a plate \(P90^\circ\), and in case IVb, instead of the plate \(P90^\circ\), a plate \(P180^\circ\), then the amplitude grating would be transformed into a phase grating, and conversely.
The problem of obtaining contrast in the case of a phase grating thus consists in either changing by \(90^\circ\) the phase of the smaller vector (over the strip), or the phase of the larger vector (over the slit). In this case the vectors over the slits and over the strips would have the same (or approximately the same) phase, but different lengths, and the phase grating would be transformed into an amplitude grating.
Zernike showed that this is in fact possible. This is seen from the following. Both gratings—the amplitude and the phase grating—as already stated, give in the exit pupil of the objective a Fraunhofer diffraction spectrum. This latter must contain, as component parts, spectra that are given by plane plates (IIIa and IIIb) and by dark amplitude gratings with phase plates—
each separately. But a plane plate generally does not give a diffraction spectrum, or—better said—gives only a zero maximum (IIIc), a colorless, undeviated image of the source. In Fig. Ic, for clarity, the zero maximum produced by the plane plate and the weaker diffraction spectra of the amplitude grating are shown separately. In the place where the diffraction spectra are obtained, the phase plate is placed; it covers only the zero maximum, imparting to it a phase retardation of \(90^\circ\), but leaving completely unchanged the light passing through the maxima \(\pm 1, \pm 2\), etc.
Since all the light represented by the large vectors (Fig. 3, IIIb) passes through the maximum 0, these vectors are rotated by \(90^\circ\) counterclockwise and are arranged almost in the direction of the small components (Fig. Vb). Above the slits the vectors rotated by \(90^\circ\) remain unchanged (VIb); above the bars still smaller vectors are added (VIb, right). Fig. VIb also shows the construction of the resultants from the vectors above the slits and above the bars: they have almost the same phase, but different length, just as in an amplitude grating, and the image must therefore represent precisely the latter. The bars appear brighter than the slits, as is shown by the hatching in Fig. VIb.
VIIb shows that the opposite result is obtained if the phase plate \(P\,90^\circ\) advances the maximum \(0^\circ\), or retards the maxima \(\pm 1, \pm 2,\ldots\) by \(90^\circ\), since only the phase difference is essential. In this case, however, the bars appear darker, and such contrast should be preferred.
The explanation given is not quite exact. For example, it does not take into account the circumstance that the phase plate also retards or advances by \(90^\circ\) the phase of the zero maximum of the diffraction spectrum belonging to the dark amplitude grating (Ic). Nor is the deviation of the phase angle from \(90^\circ\) taken into account in the case of a phase grating. This simplified explanation correctly conveys the contrast, but quantitatively it is not quite exact.
The error consists in the fact that the sum of the vectors above the slits and bars proves to be greater (VIb) or smaller (VIIb) than it can be according to Ib. It thus becomes unclear why the intensity of the light transmitted by the grating in the first case, which Zernike called “negative phase contrast,” generally increases, while in the second—with positive phase contrast—it decreases. But Zernike showed that this difference in fact does not exist; phase contrast is based on the fact that the intensity, remaining unchanged as a whole, is distributed nonuniformly over the surface of the object, in accordance with the nonuniform distribution of the phase shift at different points of the surface of the object. A small difference in the phases between the vectors above the bars and above the slits means nothing essential. The majority of mi-
microscopic specimens containing absorbing elements are, in general, not pure phase gratings, since absorbing elements also have a refractive index different from that of the surrounding medium, so that the corresponding vectors exhibit both rotation and shortening.
A more detailed consideration shows that, for phase microscopy, the shape of the condenser stop is also of essential importance. Namely, it proved advantageous to replace the usual diaphragm, which is a circular opening in an opaque screen, by an annular diaphragm, in which the middle is an opaque screen. The reasons for this are twofold.
Fig. 4. Dependence of the magnitude of the overlapping areas on the distance from the center in the case of two circles (a) and two rings (b).
First, in order to transform a phase grating into an amplitude grating, as we have seen, it is necessary that only the light of the zero maximum, by means of the use of a phase plate, change its phase by 90°. However, the lateral maxima of the diffraction spectra are completely separated from the zero maximum only in the case when the grating constant is sufficiently small. Otherwise, as is the case for most microscopic specimens, the lateral maxima partially overlap the zero maximum. It turns out, however, that this overlap is considerably smaller in the case when the diaphragm has the form of a ring. Fig. 4, illustrating this, is clear without further explanation.
The second reason why it is advisable to give the condenser diaphragm an annular form is of an entirely different nature. Namely, Loos showed that, by using an annular diaphragm, which is, as it were, a negative image of the ordinary circular diaphragm, positive phase contrast can be converted into negative contrast and vice versa (Fig. 7).
The essence of the explanation that Köhler and Loos give for this fact may be formulated as follows. In Section II (p. 372) it was already indicated that, in order to compute the amplitude in the image plane, two integrations must be carried out, of which the first extends over the surface of the object, and the second over the exit pupil of the objective. In doing so, however, it is assumed that the light source is a point source. If, as is in fact the case, the source is an extended one, then a third integration must be carried out, namely, integration over the surface of the source (the aperture of the diaphragm). Since the individual points of the source give incoherent trains of waves, it is not amplitudes that are summed in this process, but intensities. One may therefore imagine that the wide diaphragm is divided in an arbitrary manner into parts, each of which gives a “partial” image of the object, while in the image plane the brightnesses of all these partial images are summed (i.e., no interference arises).
Next, it should be recalled that, with a wide illuminating cone, a phase grating gives in the image plane a uniformly illuminated field, whereas a more or less narrowed “positive” diaphragm (a circular aperture in an opaque screen) gives a gradation of illuminations corresponding to the structure of the object. One may therefore imagine that the wide diaphragm is divided into two parts, one of which corresponds to the aperture of the “positive” diaphragm, and the other to the “negative” diaphragm. The partial images produced by these two diaphragms must sum to the “image” obtained with the wide diaphragm, i.e., to a uniformly illuminated field. But this can be so only if the brightnesses at corresponding points of both partial images give the same sum. It follows from this that the images with the positive and negative diaphragms are related to one another as a positive to the corresponding negative.
In Fig. 5 there is given (after Köhler and Loos) an interesting example of such a “reversal” of contrast. The object here consisted of droplets of Canada balsam on glass. Photograph 5a was obtained with phase contrast using a “positive” diaphragm. Here the droplets appear as black spots on a light background; photograph 5b was obtained with a “negative” diaphragm—the contrast is completely reversed.
Let us now return to the phase plate itself.
Since it must exactly coincide with the image of the condenser diaphragm formed in the exit pupil of the objective, the phase plate must be placed precisely at this location and must have the corresponding shape. In practice, the layer that changes the phase is placed on one of the lens surfaces of the objective, so as not to introduce a special plate (the exit pupil is often inside the objective). The phase plate may be placed either in the cement between two lenses or on the surface bounding
E. V. Shpolsky
with air (Fig. 6). It may constitute either the positive or the negative of the condenser diaphragm. In accordance with this, either a positive or a negative image of the specimen is obtained (see Fig. 6).
a b
Fig. 5. Influence of the shape of the diaphragm on phase contrast. Drops of Canada balsam on glass. a) circular diaphragm—positive contrast; b) annular diaphragm—negative contrast (after Köhler and Loos).
In some cases it is convenient or necessary to be able to pass continuously from the ordinary bright field to phase contrast. In such cases one may use a “polarization stop” of the following construction. From a single-crystal layer of herapathite (a Polaroid manufactured by Zeiss under the name “Bernotar”) a circle is cut out in the middle; from another layer a ring is cut out. The areas of the two figures are the same. The resulting Polaroids are cemented together so that the directions of the vibrations transmitted by them are mutually perpendicular. Such a compound polarization filter has a transparent middle, followed by a dark ring and again by a transparent ring (Fig. 7). If one more Polaroid is added to this filter, then, depending on its position, either an ordinary circular stop or an annular stop is obtained (see Fig. 7).
Interesting example of phase contrast is given in Fig. 8, which is a microphotograph of fresh blood. In an ordinary microphotograph, in bright field or in dark field, erythrocytes appear as rings—dark in the first case and light in the second. In a microphotograph obtained with pha-
Positive
contrast
Negative
contrast
Microscopic
image of the object
Phase
plate
in air
Phase
plate
in cement
Object
Fig. 6. Various possible arrangements of the phase plate in the objective and their relation to the character of the microscopic image (after Köhler and Loos).
Fig. 7. Double diaphragm for transition from bright field to phase contrast: I — arrangement for bright field; II — arrangement for phase contrast. The hatching indicates the direction of the vibrations transmitted by the polaroid.
Fig. 8. Microphotograph of fresh blood; positive phase contrast (after Köhler and Loos). Magnification 100:1.
Fig. 9. Fixed unstained treponemes. Left — bright field, right — phase contrast.
with positive contrast, it is clearly seen that the erythrocytes are not rings, but disks.
In Figs. 9 and 10*) two more examples are given illustrating the role of phase contrast. Figure 10 especially shows how many details, completely invisible even in a well-stained preparation, are revealed by phase contrast.
Fig. 10. Glands of the human duodenum (stained preparation). Left—bright field; right—phase contrast. Despite the fact that the preparation was well stained, the gland cells were stained poorly. Therefore in the bright field the details cannot be examined. Phase contrast reveals with complete clarity the boundaries of the cells and the nuclei.
The field of application of phase contrast is quite broad. Of course, not all objects are equally suitable. The best results are obtained with unstained thin biological preparations. In the article by Köhler and Loos there is an indication of the possibility of studying, by an analogous method, unetched surfaces of metals.
IV. METHOD OF SHADOW CONTRAST
In conclusion let us dwell briefly on one more special method, namely the method of “shadow contrast,” which in recent times has achieved considerable success chiefly in electron microscopy. This method consists in the fact that the preparation, or the replica taken from it
) Borrowed from the catalogue Zeiss Phasen Kontrasteinwuchtung Mikro 577*.
Fig. 11. Aluminum-carbon replica of the treated surface of brass. Optical microscope. Magnification 10:1 (after Williams and Wyckoff).
Fig. 12. Typhoid bacilli. Shadow contrast with chromium. Electron microscope (after Williams and Wyckoff).
the replica is coated in an oblique direction with the thinnest layer of metal. Owing to the fact that the coating is carried out precisely in an oblique direction, “shaded” regions arise on the specimen, where the metal does not fall, or falls in a smaller quantity. The presence of these shaded regions enhances the contrast.
Fig. 13. Hemocyanin molecules contrasted with gold on a collodion film. The structure of the collodion film, contrasted simultaneously with the hemocyanin molecules, is visible (after Williams and Wyckoff).
Williams and Wyckoff, who developed this method in especially great detail for electron microscopy, point to the possibility of applying it also in optical microscopy. For coating they used aluminum, which was deposited to such a thickness that the layer proved partially opaque to visible light. Fig. 11 shows a microphotograph of an aluminized collodion replica of a surface of brass machined on a lathe, at only 10-fold magnification.
Williams and Wyckoff point out that the contrast of such “shadowed” specimens is so great that the microscope condenser can be used at the full aperture required to attain the maximum resolving power.
If, in the field of optical microscopy, the method described has not yet found wide application, in the field of electron microscopy it has already yielded a number of valuable results. Although a survey of the advances in electron
Fig. 14. Gold replica of hemocyanin molecules on the surface of glass (after Williams and Wyckoff).
microscopy is not part of the task of the present article, we shall nevertheless present several interesting photographs obtained by Williams and Wyckoff with the aid of the shadow-contrast method described. Comparatively large objects, such as certain bacteria, are well contrasted by coating with chromium. As an example, Fig. 12 shows a microphotograph of typhoid bacilli. Although the organism itself in this photograph is clearly not well preserved, it is nevertheless possible to make out certain details on its surface.
Photographs of particles of influenza virus and fibrils of the tobacco mosaic virus, obtained by Williams and Wyckoff, have shown the applicability of the shadow method to objects so small that they are already inaccessible to optical microscopy.
Substantial advances have been made in the microphotography of large molecules. For example, with the aid of the shadow-contrast method, successful photographs were obtained of hemocyanin molecules on a collodion film when contrasted with gold. Since, however, the granules of the substrate (collodion) are also contrasted simultaneously with the hemocyanin molecules, in some cases it is difficult to distinguish the molecules
Fig. 15. Gold replica of fragments of polystyrene molecules on the surface of glass (after Williams and Wyckoff).
from the collodion on which they rest (Fig. 13). In view of this, Williams and Wyckoff developed a refined technique for obtaining shaded all-metal replicas of molecules. The technique is as follows. A drop of a solution of the substance under study is placed on the clean surface of polished glass and left until completely dry. After this, the glass, with the dried material adhering to it, is coated obliquely with a layer of gold 8 Å thick. The resulting film is too thin to be removed directly. In view of this, an additional layer of collodion is applied to the gold layer.
or Formvar, and the whole is then removed from the glass. Since the molecules of the underlying film were not shaded, the hemocyanin molecules stand out with remarkable clarity, as can be seen in Fig. 14.
In an analogous way, the fine structure of collodion and other high-polymer substances was studied. Fig. 15 shows a photograph of a gold replica of fragments of polystyrene molecules, likewise removed from glass (the polystyrene had previously been dissolved in ethyl bromide). From the length of the shadows one can estimate the thickness of these fragments; it proves not to exceed 15 Å. But this is not the limit of what is possible. Williams and Wyckoff point out that, with the aid of electron microscopy, one can detect the existence of sharply bounded regions 5–10 Å high.
Of the various new methods of microscopy mentioned in this article, the most promising are: (a) the method of ultraviolet microscopy with Brumberg’s contrast rendering in arbitrary colors; (b) the phase-contrast method; (c) the shadow-contrast method. When the apparatus needed for applying these methods becomes sufficiently widely available, they will undoubtedly bear rich fruit.
References
- F. Zernike, Physica 1, 689 (1934); Monthly Not. of R. A. I. 49, 377 (1934).
- F. Zernike, Z. techn. Physik 16, 454 (1935).
- E. Brumberg, Nature 152, 357 (1943).
- Brumberg and Gershgorin C. R. Acad. Sciences.
- A. Köhler u. W. Loos, Naturwiss. 29, 49, (1941).
- Special catalogs of the Zeiss firm — Micro G II-304-I; Zeiss Phasen—Kontrast—Einrichtung (Micro 572 u. 577).
- W. Loos, Klinische Wochenschr. 20, 849 (1941).
- Lummer u. Reiche, Die Lehre von der Bildentstehung im Mikroskop von Ernst Abbe. — Braunschweig (1910).
- Müller—Pouillet’s Lehrbuch der Physik II. Auflage, vol. II, p. 867, Braunschweig (1926).
- Lord Rayleigh, Scientific Papers, vol. 4, 235, 241.
- Robley C. Williams and Ralph W. G. Wyckoff, J. Appl. Physics, 15. 712 (1944).
- R. C. Williams and R. W. G. Wyckoff, J. Appl. Physics 17, 23 (1946).
- R. C. Williams and R. W. G. Wyckoff, Proc. Soc. Exp. Biol. Med. 59, 265 (1945).
- R. C. Williams and R. W. G. Wyckoff, Science 101, 594 (1945).
-
Lummer and Reiche in their exposition used the more exact Kirchhoff principle and the electromagnetic theory of light (see 8). An abridged exposition was given by Lummer: Müller—Pouillets Lehrbuch de Physik, 11 Auflage, Band II, I, ss. 841—877, Braunschweig, 1926. See Müller—Ponillet, Band II, 1, s. 867. ↩