SMITHSONIAN MISCELLANEOUS COLLECTIONSVOLUME 93, NUMBER 7
THE CHRISTIANSEN LIGHT FILTER: ITSADVANTAGES AND LIMITATIONS(With Two Plates)
BYE. D. McALISTERDivision of Radiation and Organisms, Smithsonian Institution
(Publication 3297)
CITY OF WASHINGTONPUBLISHED BY THE SMITHSONIAN INSTITUTIONAPRIL 2, 1935
^^e £ovi> Q0afttmore {pvceeBALTIMOIIE, MD., V. S. A.
THE CHRISTIANSEN LIGHT FILTER: ITS ADVANTAGESAND LIMITATIONSBy E. D. McALISTERDivision of Radiation and Organisms, Smithsonian Institution(With 2 Plates)INTRODUCTIONSince the Christiansen Hght filter is Httle known in this country,it is believed that a brief description of the filter and a discussion ofits possibilities may be useful. The object of the present paper isthreefold: l, to report an improvement in the construction of thefilter, which allows its use in an intense beam of light; 2, to discussthe advantages and limitations of these filters for general usage
;
and 3, to give some " practical suggestions " concerning the construc-tion of these filters. The improvement mentioned has arisen from aneed (in our laboratory) for an extensive beam of reasonably mono-chromatic light intense enough to produce an easily measured amountof photosynthesis in a higher plant. The second and third purposesof the paper are to answer numerous inquiries the writer has receivedduring the past year.REVIEW OF LITERATUREIn 1884 C. Christiansen discovered that a mass of glass particlesimmersed in a liquid transmitted freely that color for which the liquidand glass particles had the same refractive index. He pointed out intwo papers (1884, 1885) that any desired color could be obtained andthat a color complementary to the one directly transmitted was seenat oblique angles. He also showed that the wave length of the trans-mitted ray decreased rapidly with an increase in temperature. Aftera paper with comments and improvements by Lord Rayleigh in 1885,the subject lay dormant for nearly 50 years with the exception of adescriptive paragraph in all editions of R. W. Wood's " PhysicalOptics." In a series of three papers F. Weigert and collaborators(1927, 1929, 1930) show the necessity of accurately controlling thetemperature of the filters and the advantage of a refined optical system,and also describe a single filter that transmits red light when at 18° C.Smithsonian Miscellaneous Collections, Vol. 93, No. 7
2 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 93
and blue light when at 50° C. Konrad von Fragstein, in 1932 and1933, describes a filter for the near ultraviolet. One filter covers therange from 3000 A to 3700 A by temperature variation. E. Knudsen.in 1934, discusses all the various ways of making these filters and]:)oints out the possibility of making a filter of particles of low-dis-]:)ersion glass in combination with jjarticles of high-dispersion glassfused together, both having the same index of refraction for thedesired wave length. lie has made such a filter but gives no detailsof its performance.DESCRIPTION OF THE FILTER AND DISCUSSION OFITS ACTIONIn their commonest form these filters are made up of a solid packof optical glass particles (0.5 to 2 mm in size) in a glass cell, withthe spaces between filled with a liquid having the same index of refrac-tion as the glass for the wave length desired. (The present paper isnot concerned with the various emulsions and colloidal preparationsexhibiting " Christiansen colors." Readers interested in these are re-ferred to Knudsen, 1934.) Figure i gives the curves—index of re-fraction plotted against wave length—for a low-dispersion (borosili-cate) crown glass and a suitable liquid—10 percent (by volume) car-bon disulphide in benzene at 20° C. (both anhydrous). Rememberingthe laws of refraction and reflection at an interface, we see that forthe wave length where both liquid and glass have the same index ofrefraction, the filter acts as a solid plate, and the rays of this wavelength are transmitted without deviation or reflection loss within thefilter. All other rays of shorter and longer wave lengths are deviatedand reflected in an amount dependent upon the difference in the in-dices at the interfaces—glass to liquid and liquid to glass. Examiningthese curves in figure i more closely, we see that they depart from eachother more rapidly on the blue side of the crossing than they do onthe red side. This is typical of most suitable glasses and liquids. Thisshows that the filters will have a sharper blue " cut-off " than the red.Also a filter made for blue light will transmit purer colors than onemade for longer wave lengths. These two characteristics are evidentin the curves shown in figures 2 and 3. Obviously, it is desirable to usea glass of the lowest possible dispersion in combination with a licjuidhaving the highest possible dis])ersion.The refractive index of a liquid changes rapidly with its tempera-ture in comparison with that of the glass. Hence the color transmittedby the filter will vary with its temperature. Thus to maintain a givencolor, the temperature of the filter must be held constant. For use
NO. CHRISTIANSEN LIGHT FILTER McALISTER
with light of low intensity, such as in visual work, a carefully thermo-stated water hath is sufficient. For use with intense light, such asdirect sunlight, other more direct means of cooling (discussed below)are necessary. Weigert (1929), making use of this temperature co-
I.5-4O0O800600
•4O02001.53000800600400200 I-
Gl/JSS.
Z /.520008006004002001.51000800 4000
Fig. 5000 6000WAVE LENGTH-Index of refraction curves of the components of a filter. 7000
efficient, constructed an ingenious filter using methyl benzoate withcrown glass particles. This filter transmits red light when at 18° C.and blue at 50° C.These filters are not used like the ordinary colored-glass ones. The" undesired " colors are not absorbed as in the case of colored glass,but are scattered symmetrically in a halo about the center line through
4 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 93
the filter. The angular position of a given " undesired " color aboutthe axis of the filter depends upon two factors: i, the difference inthe indices of the liquid and glass for that wave length ; and 2, the
4500 5000 5500 <S000 65O0 700O 7500WAVE LENGTHl"iG. 2.—Transmission curves of a set of five filters at 20° C.number of interfaces through which the beam passes (i. e., particlesize and thickness of the filter). Also, since the interfaces are orientedin a random or probability manner, there exists only a " most proba-
NO. 7 CHRISTIANSEN LIGHT FILTER McALISTER 5ble " angle for a given " undesired " color and this color is in evidencein varying amounts at all angular positions about the axis of the filter.Some means of intercepting these " undesired " wave lengths is neces-sary. The simplest means is to use an optical system consisting of twolenses with the filter placed between them in parallel light. In this casethe " undesired " wave lengths are cut out by a diaphragm placed atthe image of the source of light. It is imperative that the opening inthis diaphragm should conform with the shape of the source of light.If a filament is the source, the opening in the diaphragm should be cutto conform to the shape and size of the image of the filament. If thisis not done, the maximum purity of color is not attained. Obviously,the purity of the color obtained increases with an increase in focallength or a decrease in numerical aperture of the optical system used.(See, for instance, von Fragstein, 1933, pp. 33 and 34.) Thus it isnecessary to measure the transmission of the filter with the particularoptical set-up to be used. Another way of using the filter is in parallellight—direct sunlight, for instance—employing a series of diaphragmsto intercept the halo of " undesired " wave lengths. In this case itis necessary to place the filter at a considerable distance from the ob-server, since the purity of color obtained increases with distance fromthe filter.Owing to the fact that the " undesired " colors are not absorbed butare scattered at various angles about its axis, the Christiansen filtercannot be used in any optical system where sharp images are desired.For instance, it cannot be used before the lens of a camera in pho-tography. The only way it could be used in this respect is with itsown optical system to illuminate the object (necessarily a small one)to be photographed.Figure 2 shows the transmission characteristics of a set of fivefilters at 20° C. They are all made of borosilicate crown glass particles(i to 2 mm in size) immersed in mixtures of carbon disulphide andbenzene. The blue filter has about 4 percent (by volume) carbon di-sulphide, the red one 20 percent, and the others have percentagesbetween these limits. These filters are 50 mm in diameter, 18 mm thick,and the windows are fused on optical flats. Two of them are shown inplate I. The transmission curves (fig. 2) were measured with thefilters in parallel light between two 20-cm focal length lenses. Adouble monochromator and vacuum thermocouple were used to makethe measurements. This purity of color is obtained only in the imageof the filament used as a source.A battery of 10 such filters ranging from ultraviolet to infrared,each selecting a spectral region about 150 A wide (at half maximum),
SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 93
was used by Messrs. Abbot, Stebbins, and Aldrich on Mount Wilsonin 1934 to measure the distribution of radiation in the spectra of starsat the Conde focus of the lOO-inch reflector. The filters were mountedwithin a constant-temperature box upon a squirrel-cage device, so asto be successively introduced into a collimated beam. The selected raywas brought to focus with a 19-cm focus lens. All this part of theexperiment worked well, and owing to the short-focus objective lens
NO. 7 CHRISTIANSEN LIGHT FILTER McALISTER^
O SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 93direct sunlight. The observations were made 66 feet from the filterwith the filter temperature held at 20° C. This purity of color is ob-tained over a 6-inch circular area at that distance.AN IMPROVEMENT THAT PERMITS THE USE OF AN INTENSEBEAM OF LIGHTThe necessity of an optical system and the disadvantage of thetemperature coefficient have been pointed out in detail in previouspublications. However, for any exacting use of the filter where muchlight energy is used there is still another trouble which proves to bevery serious : when a strong beam of light is forced through the filter,some energy is absorbed, and the center reaches a higher temperaturethan the edges, owing to poor heat conduction. The color transmittedis no longer pure, even when the filter is in a water bath. The writerhas finally overcome this difficulty by inserting aluminum vanesthrough the body of the filter so as to cut off the least amount oflight and carry ofif as much heat as possible. Details of a satisfactoryfilter equipped with these vanes are shown in figure 4. The body ofthe filter is cast aluminum, machined as shown, to take the glasswindows and the necessary gaskets. No cement is known to the writerthat will satisfactorily withstand the benzene and carbon disulphidemixture on the inside and the water on the outside. For this reasonthe windows were clamped on, as shown, with a soft lead gasket (-^inch thick) underneath the glass. Ridges were machined on thealuminum face to press into the lead gasket and improve the seal. Aje-inch rubber gasket is placed between the glass window and thebrass clamping ring. The vanes are of g^-inch aluminum assembledso that their extremities press firmly against the inner wall of thealuminum case, thus providing a path of good heat conduction fromthe inside of the filter to the surrounding water bath. The holesshown in the vanes allow the cell to be filled with the glass particlesafter it is assembled. Without these vanes, the center of this filterrose 9° C. above the temperature of the water bath when the rays froma 1,000-watt lamp were concentrated on the filter. With the vanesinstalled, the temperature at the center of the filter rose only 0.25° C.above that of the water bath under the same conditions.Plate 2 figure i, is a photograph of this filter in its water bath.The filter is filled with glass particles and a liquid (about 9 percentcarbon disulphide in benzene), and gave the transmission curve shownin figure 3 under the conditions previously mentioned. Plate 2, fig-ure 2, is a photograph of a 12 X 14-inch filter (not filled) with its waterbath. When in use the water of the bath is thermostated and stirred.
NO. 7 CHRISTIANSEN LIGHT FILTER McALISTER QAluminum was chosen as the metal for the cell and vanes because it isleast attacked by the various liquids used. The outside of the alumi-num case must be carefully covered with several coats of waterproofpaint.PRACTICAL CONSIDERATIONS IN THE CONSTRUCTION ANDUSE OF THESE FILTERSThe Christiansen filter is little known in this country. For thisreason the writer believes it will not be amiss to pass on to those in-terested some practical points concerning the construction of thesefilters and their uses. In this connection the writer is drawing on theliterature cited and his own experience with these filters.The type of cell chosen to hold the components depends upon theuse to be made of the filter. For visual work and other uses whereonly moderate intensities are necessary, a glass cell with parallel win-dows fused on is suitable. If a permanent filter is desired, a smallexpansion chamber should be provided on the filling " neck," andafter filling, the cell should be sealed ofif above the expansion chamber-in a flame. To do this safely the expansion chamber should be packedin carbon dioxide snow. When high intensities of light are used, suchas direct sunlight, the cell needs to be of the type detailed in figure 4.To be sure, a thin glass cell may be used for high intensity work, butthe purity of color will be very inferior to that obtained with a metal-cased filter equipped with vanes.The glass particles for the filter should be of the best optical qualityobtainable—preferably low-dispersion borosilicate crown glass. Fusedquartz is also suitable and of course necessary for ultraviolet work.However, the quartz should be free of bubbles and inclusions, as theselower the transmission of the filter and give it a muddy appearance.In preparing the glass particles, the writer has used the following pro-cedure. If the glass or fused quartz is in large fragments, it is groundup with an iron mortar and pestle until the larger particles are 2 or3 mm in size. This should be done with a minimum of grinding. Adamp towel should be wrapped about the top of the pestle and drapedover the top of the mortar to prevent the " dust " from flying. Theoperator should use some protection over his nostrils to avoid breath-ing the dust. The glass particles are graded by running them throughseveral sizes of sieves. Before using, the particles must be carefullycleaned. This is best accomplished by boiling in chromic acid cleaningsolution. The particles are then washed many times in clear water,then in distilled water, and finally dried completely. The particle sizefound suitable by previous workers and the writer ranges from 0.5
lO SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 93
to 2 mm (usually liracled closer
—
i.e., 0.5 to I mm, or i to 2 mm).A cell 1 5-mm thick using the 0.5 to i mm particles gives approximatelythe same results as a 30-mm cell using the i to 2 mm particles. Witha given optical system, reducing the particle size or increasing the cellthickness gives a narrower transmission curve with lower percentagetransmission at the " peak." In an " ideal " filter the glass particleswould be perfectly homogeneous as to refractive index, and these par-ticles and the liquid surrounding them would be all at exactly the sametemperature. It is because these two conditions can never be realizedthat the percentage transmission at the " peak " decreases as the num-ber of interfaces in the filter is increased.In filling the cell it is best to put the liquid in first—enough to fill thecell about half full. The glass particles are then poured in slowly so thatair bubbles are not carried down. It is difiicult to free the cell of airbubbles after it is packed solid with the glass particles. The liquid orliquids used must be anhydrous and of the highest purity. The mixingof carbon disulphide and benzene—originally suggested by Chris-tiansen in 1884—to obtain a liquid of any desired index of refraction(between that of pure benzene and pure carbon disulphide, of course)has been found very satisfactory by the writer in spite of its relativelyhigh temperature coefficient. Methyl benzoate, used by Weigert ( 1929,1930), in combination with crown glass particles makes a remarkablyvariable filter. Von Fragstein (1932, 1933) uses a mixture of 44 per-cent alcohol (ethyl) and benzene with fused quartz particles for anultraviolet filter. This filter, with suitable optics, transmits a narrowband of wave lengths in any desired part of the region 3000 A to3700 A. The wave length of maximum transmission is shifted asdesired in this region by temperature variation, just as in Weigert'smethyl benzoate cell.Various optical systems have been described in the literature. Wei-gert's (1929) autocoUimator is of considerable interest as it passesthe rays twice through the filter. The writer has shown that the filtercan be used successfully without an optical system (other than planemirrors and diaphragms) in direct sunlight, or, of course, in any beamof similar parallelism of rays. In using an optical system it is againemphasized that the diaphragm at the image of the source of light mustconform in size and shape to this image. Any change in this diaphragmwill change the transmission characteristics of the set-up. This isshown clearly by Weigert (1929, fig. 13, p. 159).In the use of the filter for studying the wave-length effect of somephotochemical phenomena it is necessary to allow for or take intoaccount the effect of the " undesired " colors—i. e., those wave lengths
NO. 7 CHRISTIANSEN LIGHT FILTER McALISTER II
shorter and longer than the wave length of maximum transmission.In any case care must be exercised in determining the combined efifectof the shape of the transmission curve of the filter, the wave lengthversus sensitivity curve of the phenomena under investigation, and,in the case where the energy content of the beam from the filter ismeasured with a photocell, the sensitivity versus wave-length responseof the detector. For instance, if the energy in the beam from an ultra-violet filter is measured with a photocell that has its maximum sensi-tivity in the blue, considerable error may come into the final resultowing to the long-wave-length " tail " on the transmission curve ofthese filters. (See von Fragstein, 1933, fig. 8, p. 33.)The writer believes that these filters will be found of considerablevalue as a source of monochromatic light for rough visual measure-ments of refractive index, rotation of plane of polarization, etc., be-cause one can set cross hairs on the wave length of maximum trans-mission within ±10 angstroms. With a sealed filter and accuratetemperature control this wave length of maximum transmission issharp and reproducible.The large filter shown in plate 2, figure 2, will be used with sunlightto irradiate a growing plant in an experiment to determine the wave-length efifect of photosynthesis. At great distance it will yield a trans-mission curve comparable to that shown in figure 3. Two filters areto be used—one to cover the range 4000 A to 6000 A, the other from5500 A to 8000 A. The wave length of maximum transmission ismoved through these ranges by temperature variation.The possibility of substituting a high-dispersion glass for theliquid—i. e., making the filter of a high-dispersion glass flowed aroundthe particles of low-dispersion glass—is interesting. Knudsen (1934)has accomplished this, but gives no details. The resultant filter willhave only a very small temperature coefficient, which will considerablyenhance its usefulness. The writer has in his possession two suitableglasses, but has not yet had an opportunity to complete the filter.The writer is grateful to A. N. Finn, of the United States Bureauof Standards, who has kindly furnished the borosilicate crown glass,and to L. B. Clark, of the Division of Radiation and Organisms ofthe Smithsonian Institution, w^ho constructed the glass cells with fused-on optical windows. LITERATURE CITEDChristiansen, C.1884. Untersuchungen iiber die optischen Eigenshaften von fein vertheiltenKorpern. Ann. Phys. Chem., vol. 23, pp. 298-306.1885. Idem, vol. 24 pp. 439-446.
12 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 93Knudsen, E.1934. Chromatische disperse Zweiphasensysteme unter besonderer Bertick-sichtigung ihrer Verwendung als Lichtfilter. Koll. Zeitschr., vol.66, pp. 257-266, March.KoHN, H., and von Fragstein, K.1932. Ein Christiansenfilter fiir ultraviolettes Strahlung. Phys. Zeitschr.,vol. 33, PP- 929-931.Ravleigh, Lord1885. On an improved apparatus for Christiansen's experiment. Phil. Mag.,vol. 20, pp. 358-360.VON Fragstein, K.1933- Ein Christiansenfilter fiir ultraviolettes Licht. Ann. Phys., vol. 17,pp. 22-40.Weigert, F., and Staude, H.1927. tjber monochromatische Farbfilter. Zeitschr. Phys. Chem., vol. 130,pp. 607-615.Weigert, F., Staude, H., and Elvegard, E.1929. tJber monochromatische Lichtfilter. IL Zeitschr. Phys. Chem., ser.B, vol. 2, pp. 149-160.Weigert, F., and Shidei, J.1930. Photodichroismus und Photoanisotropie. VIL Ibid., vol. 9, pp. 329-355. (See pp. 33I-334-)Wood, R. W.Christiansen's experiment. In all editions of Wood's Physical Optics.