OVER THE PAST DOZEN YEARS this department has presented several designs for spectroscopes, spectrophotometers and related instruments for spreading light into its component colors (wavelengths) so that it can be analyzed. The instruments were of two basic types, one for studying the light absorbing properties of a transparent liquid, solid or gas and the other for analyzing a source of light such as a lamp or a star. A new design for a special type of spectrophotometer was sent to me recently by Dean Morelli of Rye, N.Y. Its primary advantage over the instruments described here previously is its speed in scanning the visible spectrum: it can obtain a complete set of measurements in as little as a thirtieth of a second.

I can best set the stage for describing Morelli's instrument by reviewing the two basic types of spectroscope and spectrophotometer. In the first of them white light from a lamp is directed through a cell containing the liquid, solid or gas to be investigated. The sample absorbs the light at certain wavelengths and so diminishes it. The transmitted light is dispersed (separated according to wavelength) by a prism or a diffraction grating before it reaches a detector. In this way the detector can determine which colors have been removed from the initially white light by the sample. With reference works on atomic and molecular spectroscopy the observer can identify the constituents of the sample and can distinguish the sample from others that might look similar to the unaided eye even though they differ in composition.

The second basic type of instrument is the one most frequently employed in astronomical work. The spectroscope is attached to a telescope so that the observer can identify the colors in the light from a celestial object. Sunlight has been analyzed in this way by many amateurs. The sun radiates light throughout a range of wavelengths, with the peak intensity falling near the middle of the visible spectrum. When that basically white light passes through the sun's atmosphere, the elements in the atmosphere absorb the light at their characteristic wavelengths. The absorbed wavelengths appear as thousands of dark lines (the Fraunhofer lines) superposed on the colors of the visible spectrum. With the appropriate reference works the observer can identify the lines with the elements in the sun's atmosphere.

A spectroscope relies on the human eye for examination of the spectrum. If a light-sensitive device such as a

photomultiplier is substituted for the eye, the instrument is a spectrophotometer. The experimenter may still have to take readings of the intensity of the light point by point across the spectrum. A more convenient system is one where the photomultiplier is coupled to a chart recorder, so that as the photomultiplier moves across the spectrum its response is automatically recorded on paper as a curve of the intensity of the light at each s wavelength.

Recording the intensity of light point by point across the spectrum can take a good deal of time. Wavelengths are measured in angstrom units, and the visible spectrum is 3,000 angstroms wide. Therefore if a measurement every 100 angstroms is needed, it will take 31 measurements and at least that many minutes to cover the spectrum. Morelli's rapid-scan spectrophotometer does the job in a fraction of a second and so gives the experimenter the opportunity to investigate phenomena occurring much faster than the ones that can be investigated with slower instruments.

As an example of the potential of rapid scanning, Morelli told me how he examined the color oscillations of the chemical oscillator I described in this department for July, 1978. Although the period of the oscillations depends on the concentrations of the chemicals, it is usually less than a minute. With a conventional spectrophotometer one would have to be content with following the oscillations at a single wavelength, since the instrument would never be able to keep pace with the reactions.

Although the design for Morelli's instrument can be modified according to the experimenter's needs, it basically consists of a lamp, a curved diffraction grating, a photomultiplier tube and an oscilloscope. Light from the lamp is passed through a cell containing the sample that is to be investigated. The diffraction grating disperses the light into its component wavelengths. The grating is oscillated horizontally to sweep the visible spectrum across the photomultiplier tube, and the intensity of the light is displayed by the oscilloscope. The horizontal axis of the oscilloscope trace represents wavelength and the vertical axis intensity. One therefore sees a graph on which the intensity of the light is plotted against the wavelengths in the visible range. When the sample cell is absent, the oscilloscope shows a straight horizontal line. When the cell is in place, the resulting curve quickly identifies the wavelengths at which the sample has absorbed light.

For a lamp Morelli used an automobile light bulb (such as the bulb for a ceiling light) with a straight filament. The lamp is mounted in its housing with the filament oriented vertically. Although the lamp is rated for operation at 12 volts, Morelli runs it at about 21 volts in order to obtain the necessary intensity and uniformity across the visible spectrum. (The higher voltage does, of course, shorten the life of the lamp.)

The housing for the lamp is fashioned out of thin-gauge sheet steel cut and bent into-shape. Morelli made the housing in two pieces so that he could replace the lamp more easily. The rear part of the housing is fastened to the main base of the spectrophotometer with a U bracket. The front part has attached to it a short tube through which the light passes. At the far end of the tube is a flat plate with two razor blades fastened to it by four screws. The tube and the razor blades serve as a baffle to keep stray light from entering the photomultiplier. More baffles are set up inside the instrument for the same purpose. The entire instrument is also covered with a box to keep out room light.

Morelli made the sample cell by cutting a U-shaped piece out of quarter inch Plexiglas and sandwiching it between two

glass plates. He glued the pieces together with silicone sealant. As a support for the cell he suggests cutting and bending sheet metal into a slide and screwing it onto a wood block. The cell is of course positioned in front of the light output from the lamp. In the normal procedure of measuring the absorption characteristics of a solution two cells are needed, one cell for the solvent alone so that the spectrophotometer can be standardized and one for the solution that is to be compared with the standard.

The diffraction grating disperses the light transmitted through the sample cell. Morelli chose a curved diffraction grating of the kind found in certain astronomical spectroscopes. (For retailers of the gratings check the pages of an astronomy magazine.) Morelli's grating has a radius of curvature of 50 centimeters, a ruled area of 3.175 by 4.445 centimeters and a line density of 1,200 lines per millimeter. The lines are spaced so that light with a wavelength of 5,000 angstroms has its maximum reflection in the first order of the diffraction pattern. (Such a grating is said to be blazed at 5,000 angstroms.) Other curved gratings could be substituted, although Morelli warns that gratings with a higher number will make the instrument less sensitive. If a flat grating is used, a lens will be needed in order to project the image of the slit of the lamp housing onto the slit of the photomultiplier (by way of reflection from the grating).

Morelli's grating is mounted on an aluminum plate to which a hinge is riveted. The idea is to have the grating swing horizontally around the hinge in order to sweep the spectrum across the photomultiplier. The hinge is attached to an aluminum angle that is fastened to the main base of the spectrophotometer. The hinge should turn smoothly with no free play, but it should not be so firm that it offers too much resistance to being swung. Morelli says he tried several hinges before he found a suitable one.

For the highest possible scanning rate the grating should be rigged to oscillate around its center. This pivot point will not, however, provide the best focusing for all the wavelengths in the visible spectrum, because the best focusing distance for a grating varies with wavelength. For the optimum optical performance the pivot point should be offset by about six centimeters from the center of the grating. For such a pivot point the scanning rate is lower (because the grating has a greater moment of inertia than it would have with a central pivot point), but the improvement in focusing is worth the sacrifice.

The grating is made to sweep by means of a cam against which a small tab from the mounting plate of the grating is held. The tab is cut from right angle stock and firmly attached to the plate. One end of the tab is held against the cam by the tension from a spring attached to the plate. When the cam turns, the tab slides along the edge of it and forces the grating to sweep horizon tally. The spring ensures that the tab is held against the cam.

The cam is cut from 3/32-inch sheet brass: Its shape and size are first mapped out on paper. The paper mask is laid on the metal sheet and traced, after which the metal is cut and filed smooth around its perimeter. Ideally the cam should sweep the grating in a sawtooth motion, uniformly driving it horizontally until the end of the sweep is reached. This motion is obtained when the radius of the cam varies uniformly with the angle around its center. The center hole for the cam is drilled with a quarter-inch bit to fit on a motor shaft of the same diameter. Morelli soldered onto the cam a brass insert he had removed from a panel knob (Radio Shack Knob 274-415). The insert is normally inside the plastic covering. The screw into the insert enabled Morelli to attach the insert and the cam firmly to the motor shaft.

The upper limit to the sweep rate of the spectrophotometer is set by the oscillation rate of the grating. If the grating is run too fast, the spring cannot hold the tab against the cam and the sweep becomes erratic. Substituting a stronger spring may not improve the performance if the resulting increase in friction between the cam and the tab distorts the running speed of the motor.

Opposite the grating on the base of the spectrophotometer is the photomultiplier and its housing. The housing is made from thin-gauge sheet steel. A steel tube is fastened to the front of the housing with a weld or epoxy glue. Inside this tube is slid a smaller tube. Razor blades are attached to the smaller tube to make a narrow slit. Since the smaller tube is easily removed, the experimenter can readily install slits of different widths.

The photomultiplier tube is mounted on a shelf in the housing. The resistors necessary to provide the proper range of voltage to the photomultiplier are soldered directly onto the socket on the shelf. Like all the other exposed metal parts in the spectrophotometer, the housing should be painted a dull black to absorb stray light.

Morelli has tried several types of photomultiplier tube. The one you would buy may depend primarily on what is to be had on the surplus market. Model 931-A is often available. Morelli also has a lP21, which is the same type of tube except that it has a higher quantum efficiency and a lower dark current. (I recently saw an advertisement offering two of these tubes for $75.) The quantum efficiency has to do with how many electrons the photocathode of the tube will emit for each photon it absorbs. If the ratio is fairly high, the photomultiplier is said to be sensitive and to have a high quantum efficiency. Dark current is the current that flows from the tube when no light is falling on the photocathode. It should be low so that it does not mask the desired signal and prevent the measurement of low levels of light. The 931-A and IP21 tubes both have the type of photoelectric surface designated S-4, it is composed of cesium and antimony. Although the tubes have a relatively high quantum efficiency over the greater part of the visible spectrum, they are somewhat insensitive toward the red end, reaching a practical limit near 6,700 angstroms.

A photomultiplier tube must be biased properly, that is, it must have a suitable range of voltage. Morelli employs a regulated power supply to provide about 100 volts per stage for each of the nine stages in his 931-A photomultiplier. One should consult Radio Amateur s Handbook for information on power-supply designs.

Recently Morelli tested an RCA 4840 photomultiplier. Its multialkali photoelectric surface (designated S-20) had a

spectral range larger than that of his lP21 tube, being sensitive to wavelengths as long as 8,000 angstroms. The tube also has an envelope that passes ultraviolet wavelengths. With this tube Morelli was able to get full-scale vertical deflections on the oscilloscope as he measured the intensity of the light in the range from 3,400 angstroms to 7,000.

The output from the photomultiplier is connected to the vertical input of the oscilloscope, which should have a vertical amplifier for direct current. The oscilloscope trace should move across the screen from left to right as the grating sweeps the spectrum over the entrance slit of the photomultiplier. How the oscilloscope is triggered to begin each sweep across the screen depends somewhat on the type of oscilloscope. If it has an internal provision for a triggered sweep, the triggering signal can be provided by a small light, a slit disk and a silicon photocell. The disk is mounted on an extension of the shaft of the motor that oscillates the grating. The light is on one side of the disk and the photocell is on the other. When the slit in the disk is properly positioned, the light from the small bulb falls on the photocell each time the grating begins another sweep of the spectrum across the photomultiplier. The signal from the photocell then starts the sweep of the oscilloscope trace. (Morelli cautions that inexpensive oscilloscopes do not trigger reliably at the low sweep frequencies of the spectrophotometer.)

A better system for displaying the output of the photomultiplier can be set up if the oscilloscope has a horizontal amplifier for direct current. A small projector is mounted in a metal tube so that it projects an image of a slit onto a silicon photocell (SM4) on the other side of an appropriately shaped cam. The cam is similar to the one that drive the grating. When this cam turns, it causes progressively more light to fall on the photocell. The signal from the photocell is amplified at the horizonta1 input of the oscilloscope, sending the trace from left to right across the screen The second cam could be mounted o an extension of the shaft of the grating's motor to synchronize the horizontal sweep of the oscilloscope with the movement of the grating.

If the amplifier for the horizontal input of the oscilloscope is not sensitive enough for this system, the projector and the photocell can be replaced with a potentiometer, which will supply a larger signal to the input. In this role the potentiometer must turn continuously, so that the small indentation in its housing that keeps it from rotating fully must be removed. The potentiometer is mounted on the shaft of the motor with a short length of rubber tubing. A 1.5volt battery provides the voltage across the potentiometer, the output of which then varies continuously from zero to 1.5 volts in a sawtooth fashion as the shaft causes the inner connection of the potentiometer to rotate. The signal will drive the oscilloscope trace smoothly across the screen.

One may want to eliminate the return trace of the oscilloscope, which moves across the screen from right to left before each new measurement of intensity. To blank out the return trace Morelli put a switch on the drive shaft between the mount for the grating and the corrector plate for the photomultiplier. The switch consists of two metal strips that are pushed apart by a cam on the drive shaft during the period when the oscilloscope trace is to be blanked out. The switch is in series with the wire supplying the high voltage to the oscilloscope. One may, however, want to keep the return trace to serve as a reference line on the screen indicating the base line of zero light intensity. Such a base line may be particularly desirable if the oscilloscope has an unregulated power supply and so has a trace that tends to float.

If the oscilloscope has low sensitivity on its vertical input, a preamplifier is needed to boost the signal from the photomultiplier. Morelli employs an inexpensive uA74 1 operational amplifier chip in a simple feedback circuit powered by penlight batteries (enough to give + or - 15 volts).

Morelli uses his instrument as follows to determine the absorption characteristics of a solution compared with a pure solvent. He first gets an oscilloscope trace with the pure solvent in the sample cell. Even if the solvent transmits perfectly, the trace will not be flat, because the intensity of the light emitted by the lamp is not uniform at all wavelengths across the visible spectrum. Moreover, the response of the photomultiplier is not uniform across the spectrum. Hence the trace on the oscilloscope is approximately bell-shaped, with the top of the bell disappearing off the top of the screen.

A corrector disk is made to compensate for these nonuniformities (and any others that distort the response of the instrument across the spectrum). The disk is mounted on a shaft coupled through miter gears to the shaft from the motor that causes the grating to oscillate. The purpose of the disk is to selectively diminish the light falling on the photomultiplier at the wavelengths that give a higher trace on the oscilloscope. A suitably fashioned corrector disk will block the light at each wavelength in the spectrum in such a way that the trace is approximately horizontal across the top of the screen. With a pure solvent one then gets a flat trace on the oscilloscope.

Morelli cuts the corrector disk out of a piece of sheet brass. When the disk is mounted on the drive shaft, it blocks part of the light falling on the slit in front of the photomultiplier. Next Morelli places pieces of black electrical tape at the edge of the disk as he manually turns the shaft of the motor. For each position of the motor the grating casts a certain wavelength onto the slit, and the vertical deflection of the spot on the oscilloscope screen corresponds to the response of the photomultiplier at that wavelength. At each wavelength the spot is initially deflected off the top of the screen. Morelli adds a small piece of tape to the corrector disk to block enough light for the spot to be lowered to the top of the screen. Then he rotates the shaft on the motor slightly and repeats the process. He says that although the system is somewhat crude, the resulting trace across the top of the screen is flat to within 5 percent.

When the cell containing only the solvent is replaced with a cell containing a substance dissolved in the same solvent, the

absorption of light by the solution at certain wavelengths gives rise to spikes that extend downward on the oscilloscope screen. If the solution were able to totally absorb the light at a certain wavelength, the corresponding spike would reach the bottom of the screen.

The resolution of the spectrophotometer is governed partly by the quality of the grating and partly by the width of the slit in front of the photomultiplier. A narrower slit samples a narrower band of wavelengths in the spectrum the grating casts across the slit. At the position of the slit the width of the full visible spectrum is 125 millimeters. Morelli normally uses a s-lit with a width o about four millimeters, which therefore admits 4/125 (roughly 100 angstroms) of the full spectrum to the photomultiplier One can increase the resolution by replacing the four-millimeter slit with a narrower one, but a price must be paid in a loss of light intensity. Then the photomultiplier will respond only to light at wavelengths in its more sensitive range (for Morelli's 931-A tube approximately 3,700 angstroms to 6,000) and the red end of the spectrum will be lost. With a four-millimeter slit a larger range (3,500 angstroms to 6,300) can be monitored. The lower limit to the range is then set by the ultraviolet absorption of the glass in the lamp, the photomultiplier tube and the sample cell. Another problem with slits narrower than four millimeters is that they may call for off-center pivoting of the grating to prevent defocusing at the ends of the spectrum.

Morelli calibrates his spectrophotometer by putting a hydrogen Geissler 9 emission tube in place of the usual lamp. Several of the visible spectral lines emitted by hydrogen (the Balmer series) can be identified on the oscilloscope screen. When the four-millimeter slit is in place, the lines appear as sharp spikes. Since the wavelengths of the lines are known precisely, Morelli can calibrate the horizontal axis of the trace on the screen in angstroms.

I have purposely not been precise about the dimensions of the spectrophotometer because Morelli emphasizes the need for the experimenter to modify and adjust the basic design to suit whatever equipment he has gathered for building the instrument. The housings for the photomultiplier and the lamp should be as close together as possible. Morelli suggests a way to determine the proper distances of the lamp, the photomultiplier and the grating. Set up on a table a Geissler tube, the grating and a slit. With the room lights off and the tube on move the three objects on the table around until the spectrum falls on the slit in good focus. The distances so determined can serve as a guide for the construction of the spectrophotometer.

To use the spectrophotometer for examining the spectral characteristics of external sources of light, such as a lamp or a flame, a mirror is mounted in front of the lamp in the instrument. Light from the external source is admitted through a slit in the side of the box housing the spectrophotometer and is directed by the mirror to the grating. The source must be relatively bright or the red end of the spectrum will be lost because of the insensitivity of the photomultiplier at that end.

The possible applications of Morelli's rapid-scan spectrophotometer are numerous. In addition to the oscillating chemical reactions I have mentioned, there are almost countless other optical phenomena that change too rapidly for any spectrophotometer requiring a point-by-point measurement across the visible spectrum. Morelli himself, for example, wants to examine the phosphorescence of certain crystals.

Some amateur astronomers may be interested in analyzing the spectral composition of sunlight. Morelli employs a telescope and a mirror to project an image of the sun onto a pinhole that he has installed temporarily in place of the slit in the side of the box housing the spectrophotometer. He moves the image of the sun across the pinhole and thereby gets tracings on his oscilloscope for spectra of different areas on the sun. Similar observations could be made of other celestial objects, but because of the much lower intensity of their light the spectrophotometer of Morelli's basic design would respond primarily to the wavelengths at which the photomultiplier is the most sensitive.

Morelli has not stopped tinkering with his instrument. His next big change will be to replace the mechanical scanning system of the present design with an entirely electronic scanning system based on a Vidicon (an image-storage tube). Such a system will be more sensitive because the oscilloscope trace will be built up through the storage of information from many scans. It would therefore be better for the spectral analysis of light from faint sources such as phosphorescing crystals. With the electronic system Morelli will also be able to examine ultrarapid phenomena such as the flash from a high-speed xenon flash lamp. The Vidicon is also more sensitive at the red end of the spectrum, so that the instrument will not have one of the shortcomings of an instrument based on a photomultiplier. Morelli is now building a rapid-scan spectrophotometer with a Vidicon tube that is employed in closed-circuit black-and-white television cameras.

Bibliography

ANALYTICAL ABSORPTION SPECTROSCOPY: ABSORPTIMETRY AND COEORIMETRY. Edited by M. G. Mellon. John Wiley & Sons, Inc., 1950.

The Society for Amateur Scientists (SAS) is a nonprofit research and educational organization dedicated to helping people enrich their lives by following their passion to take part in scientific adventures of all kinds.