electron micrographs of spectroscopic gratings

ELECTRON MICROGRAPHS OF SPECTROSCOPIC GRATINGS Randall C. Brooks Abstract From the discovery of spectral lines in the Sun's light by Joseph Fraunhofer to today's Hubble Space Telescope, one of the most fundamental tools of scientific enquiry is the spectroscopic grating. Traditionally created by ruling fine lines in the surface of speculum metal or aluminum-coated glass, these elegant devices create spectra of bright objects in direct or reflected light. The quality of gratings improved dramatically with very discrete steps. This paper presents the results of electron micrographic analysis of several gratings dating from 1875 to about 1962 and graphically illustrates those advances. The Origins of Gratings The shape of science in the last 150 years has been sculpted by the delicate man-made gossamer surfaces of spectroscopic gratings. The purpose of this paper is not to give a history of the engines employed to rule spectroscopic gratings, the most precise mechanical devices made until ca. 1965, but some background relevant to the American experience is desirable. David Rittenhouse is said to have observed spectra with gratings made of parallel wires in 1785.l In Germany Joseph Fraunhofer measured and identified several lines in the solar spectrum which still carry his name. His observations mark the beginnings of spectroscopy in 1814-1815. He also experimented with fine wire gratings. John Barton of London then patented the process for making his so-called Barton's buttons — rulings made on round, metal discs for decoration. They were the by-product of his attempts, along with Henry Maudslay, to make a perfect screw. Subsequently Barton made a 2000 lines/inch grating for David Brewster, though the ruling diamond broke before the grating's completion.2 Back in the USA, Joseph Saxton reportedly made a grating in the early 1840s which was used by John W. Draper to make the first photograph of a diffraction spectrum. Charles Fasoldt, a German immigrant to the US, made a few test plates for microscope testing on a machine which is preserved in the Smithsonian's National Museum of American History.3 The only gratings previously imaged with electron microscopes, of which I am aware, are examples of F.A. Nobert's 10 and 20 band test plates 27

~

Vol. 17(2003)

designed to assess microscope optics, though the finest bands were beyond the resolving power of even the best modern optical microscopes. This imaging was done by Saville Bradbury and published by Gerard Turner.4 Another was imaged by J.J. McNeill of a grating made by Henry Grayson in Australia and made about 1917.5 However, most of McNeill's paper is devoted to interferogram analysis so will not be discussed here. Friedrich Adolph Nobert (1806-1881) was one of the most talented instrument makers of the 19th century. The son of a German clockmaker, Nobert attended the Technical Institute in Berlin, and in his first week learned scale dividing techniques. With this knowledge he made his own dividing engine, presumed to be the one he used to rule diffraction gratings and test plates. After his training in Berlin, Nobert became a technician and instrument maker at the University of Greifswald (1835) and later moved to Earth (1846). Nobert's interests included testing microscope optics. It occurred to him

Fig. la (above) Electron-micrographs of a replica of Nobert's 19 band (top) and 20 band (lower) microscope test plate made about 1873. It was tested by Bradbury in 1965. Each band was ruled with about 9090 lines per mm and was beyond resolution of optical microscopes - even modern ones. Fig. Ib (right) Higher resolution micrograph of a Nobert grating shows that one side of the lines was straight with slight raised edge while the other shows chipping of the glass. This problem resulted in asymmetry of the spectral lines. RlTTENHOUSE

28

in 1845 that the technology of dividing engines could be used to make test objects engraved on glass. His 10-band test plate had line spacings in the range of 1/1000 -1/4000 Paris line (1 Paris line = 2.25583mm) in a geometrical progression. This was the first target with known spacing with which one could objectively test the quality of microscope optics. The properties of the diamonds used by Nobert were discussed by Mayall, who noted that Nobert used a force of 3 to 30 grams to produce his rulings and that after 1869 he used 'mild' (i.e. soft) glass.6 Nobert made seven progressively finer test plates on glass culminating with his 'new' 20-band plate in 1873. Unfortunately we do not know much about how he made his gratings but they were presumably made on the same dividing engine with similar techniques to those used to rule the test plates. The electron micrographs of a replica (made with a mould and coated to allow imaging in the electron microsope) of a Nobert test plate (Fig. la) show two of the bands (19 and 20) defined by 54 and 57 ruled lines resp. These gratings were subjected to electron microscope analysis with the objective of assessing the quality and nature of the grooves in the gratings. The finest band (20) has average interline spacings of 0.11 - 0.12 um or almost 9000 lines/mm. In the closeup (Fig. Ib), the rulings show chips on one side but the overall quality of the lines is remarkably good and the line spacing looks even. However, flaws in the surface, visible between the grooves, had an effect on the quality of the ruling and the interline spacing of the grooves visible in the lower resolution image (left) shows that the diamond was not ruling the lines perfectly straight. The chipping of one side of the grooves would have made one side of the observed spectrum lines brighter than the other and the resolution of closely spaced lines would have been decreased. Lord Rayleigh (1874) experimented with replicas of Nobert gratings using a collodio-chloride process in which he used gratings of 3000 and 6000 lines per inch, and a third by American Lewis Rutherfurd of 6000 lines per inch.7 He found Nobert's "much superior", though the 3000 line version demonstrated three distinct zones which Rayleigh attributed to use of three different diamond points though it was possibly due to mechanical shifts of the ruling arm and/or diamond. He also found that these three zones were reproduced in the copies and were visible to the unaided eye. Rayleigh was able to observe almost all of the lines Angstrom had plotted in his solar spectrum with both the 3000 line original and its copy; he also noted that the third order spectrum was the brightest - the cause and significance escaped him but was the basis of an important innovation in the early 20th century. What really stimulated interest in observing spectra was the work of Gustav Kirchoff and Robert Bunsen. In 1859 they delineated the three 29

Vol. 17(2003)

Fig. 2 Rutherfurd's ruling engine as it appeared in 1877 (Warner, 1971).

fundamental rules that explain the origins of emission and absorption spectral lines in gases - rules that became the backbone of chemical analysis. This, in part, spurred the wealthy New York amateur scientist, Lewis M. Rutherfurd (1816-1892), to make a ruling engine in 1863 as a means of improving the efficiency of spectral investigations.8 Spectrographs with trains of prisms achieve high dispersion but absorbed too much light. It was recognized, though not initially fulfilled, that gratings could overcome this. Rutherfurd had some success with this engine but a second made more to his satisfaction was completed in 1872 but was later modified several times over the years (Fig. 2). Gratings produced on this machine were given to friends and scientists who he knew would make good use of them. For the next ten years, Rutherfurd's gratings were sought after as the best available - certainly for the price (free!). Nobert's gratings were occassionally superior to Rutherfurd's but difficult and expensive to acquire. The quality of gratings is primarily dependent on a high precision screw free of periodic errors, with no gradual change in pitch and perfectly straight. Seeing the limitations of Rutherfurd's gratings, i.e. ghost images of lines due to periodic errors of the lead screw in the grating engine, Harvard's William Rogers tried to make a perfect screw (and his were probably the best made in the 19th century) as the basis of a new ruling engine.9 However, physicist Henry Rowland's achievements at Johns Hopkins University made Roger's efforts superfluous, though a number of superior gratings were made on RlTTENHOUSE

30

Rogers' two engines constructed ca. 1873 and 1880. Trained as an engineer himself, Henry Rowland (1848-1901) lured Theodore Schneider from W. & L.E. Gurley in 1876. Rowland's work, and their subsequent success built on the work of Rutherfurd and Rogers. Rowland was well funded and he and Schneider had access to high quality machine tools. Rowland recognized that they were unlikely to build a perfect apparatus, but was clever enough to design apparatus that overcame several fjaws — in particular, eccentricity and periodic errors of the screw as well as friction and wear. Lord Rayleigh had pointed out that for reflecting surfaces an error less than 1/4 wavelength of the light being studied is required. Once the engine was completed, and with superior quality blanks (both flat and concave) from John Brashear, Rowland's gratings soon received acclaim.10 Brashear handled distribution of Rowland's gratings produced at Johns Hopkins under Schneider's watchful eye. By 1901 some 250-300 gratings had been distributed, excluding those given away. Those gratings sold were sold at cost, as Rowland had agreed with the University. Rowland's most important achievement (1882) was recognizing that concave gratings would focus the spectra without the need for two large collimating / measuring telescopes as were required with flat gratings.11 This improved efficiency tremendously. Brashear provided the concave blanks for

Fig. 3 Henry Rowland with one of his grating engines. Photo courtesy of the Johns Hopkins University Special Collections. 31

Vol. 17(2003)

I

I

50

Rutherfurd not blazed Rowland/Schneider not blazed Anderson/Jacomini not blazed unknown 16° Horace Babcock unknown Horace Babcock? not blazed unknown 36° 52' Unknown 4° 45'

1875 c.1887

1924 ca. 1935

1951 ca. 1950

1960 ca. 1965

Bausch & Lomb aluminum

Bausch & Lomb aluminum

Cal Tech? aluminum

fused quartz MgF2 coating

glass no coating

glass no coating

glass no coating

glass no coating

Bausch & Lomb aluminum Cal Tech aluminum

speculum no coating

speculum no coating

speculum no coating

12000/11370

15.3 x 7.3 930316 / 524 A4

7.5 x 6.5 cm 930315 / 371A4B4

7.9 x 4.2 cm 930321

-/5830 12000 / 12800

12.7 x 10.0 cm 960247

12.9 x 7.3 cm 960246/124

11.0x10.4 cm 960245 /19A

unknown - broken N/A

2.7 x 1.7 cm N/A .

CSTM#/SER.#

RULED SIZE

5930 / 5880

2130 / 2090

5900 / 5850

- / 8100

3400 / 3475

SUBSTRATE GROOVES/CM' ANTI-REFLECTANCE?

Mt. Wilson Obs. no coating

Johns Hopkins no coating

private no coating

INSTITUTION COATING

GRATING CHARACTERISTICS

21 ft. (Mackenzie)

flat (see description)

flat

= 175000 lines 2998.3 mm NRCC (Gerhard Herzberg)

« 96000 lines flat NRCC (Gerhard Herzberg)

~ 74000 lines flat NRCC (Gerhard Herzberg)

= 75000 lines Dominion Obs.

~ 27000 lines flat Dominion Obs. (Ralph DeLury)

~ 65000 lines flat Dominion Obs. (Ralph DeLury)

unknown Johns Hopkins

5761 Alvin Clark

TOTAL # LINES f/l2 USER'S INSTITUTION (SCIENTIST)

Notes: 1 The first number is that stated on the grating (if provided), the second is the number of lines measured from the electron micrographs. It should be noted that calibration of the SEM's scale bars has not been checked and may be systematically in error. 2 Focal length of curved gratings. 3 This grating has a layer of varnish or lacquer added after the grating was broken to help preserve the surface.

MAKER BLAZE ANGLE

DATE

c °- en S. ft

&: ft , ~ r> a. Lft _ p. g. ft -a ff. 3 •_ w cr o ft ft . era Ln o ft 3

w sr $• r±

variable-pressure SEM proved more successful.16 The working distance for these images varied from 5.3-10.2 mm at a voltage of 5 kV to 20 kV and with magnifications ranging from 1500 to 8000 times. A selection of the images obtained are included in this paper. Rutherfurd grating (1875): Currently in the collection of John W. Briggs, this Rutherfurd grating was acquired at auction from the collection of Burton L. Fitzgerald of Milton, MA. Fitzgerald had told Briggs years before that he had acquired it from the assets of the Clark Corporation (originally Alvan Clark & Sons of Cambridgeport, MA) when the firm had closed. The grating is signed Lewis M. Rutherfurd, dated 1875 and is marked with the number of ruled lines per inch (5761) and total number of lines (8640). Warner lists a number of known Rutherfurd gratings one of which he gave to Harvard astronomer Edward C. Pickering and one to Wolcott Gibbs; both match the line spacing of the one tested though the dates of both are given by Warner as 1874.17 Of a Rutherfurd grating, tested and compared to two by Nobert, Lord

Rayleigh stated "...I may observe that in respect of brightness they [i.e. solar spectra] fall far short of Nobert's".18 However, in 1882 Rowland observed "....Young, Rutherfurd, Lockyer, and others, [have] done much good work for science. Many mechanics in this country and in France and Germany have sought to equal Mr. Rutherfurd's gratings, but without success".19 Rowland also noted that only 1 in 4 of Rutherfurd's gratings was of good quality and only a rare one that could be considered first class. The electron micrograph of the 1875 grating (Fig. 4a) shows a substrate (speculum metal) that appears to be scratched, i.e. it was insufficiently polished prior to ruling but it is also crossed by prominent scratches made after ruling. The grooves are very shallow and narrow relative to the spacing, with some variability in width (== 0.70 urn with =5-7% variation) and interline spacing of about 2.75 ± 0.15 urn. In the image the flaws of the diamond point are very clear having left 3 fine lines within each groove. Most importantly, in ruling each line, the diamond was throwing up a burr on each side with the one on the upper side being wider and more prominent (note that the ruling was executed from left to right in this illustration as shown by the effect of the underlying scratches on the burrs). Because of the scattered light such burrs caused, Charles S. Pierce had, in desperation, coated one such Rutherfurd grating with 'black lead' and then polished it as one would a speculum blank. Once flat, the black lead was dissolved and the grating performed much better producing more evenly bright spectra side to side.20

Fig. 4 The Rutherfurd grating tested is dated 1875 and has 5761 lines (3475 lines/cm). The ruled portion appears as the lighter gray rectangle in the centre. Fig. 4a (right) Electron micrograph of the Rutherfurd grating made at 10 KeV, a working distance of 7 mm and magnification of 8000 times. This and the following electronmicrographs were produced at CCI.

Rowland grating (ca. 1887. (Fig. 5, 6} Rowland's first machine could rule areas of just over 6 x 4 inches. According to Rowland, the screw was nearly perfect; he was unable to detect any periodic errors or errors of run to 1/100,000 part of an inch. With this lead screw he ruled "gratings with 43,000 lines to the inch, and have made a ruled surface with 160000 lines on it, having about 29000 lines to the inch." 21 It was also in this paper that he describes his idea and first attempt at using a spherical surface to avoid the need to use large telescopes to bring a spectrum to focus, but still need to collect the light. This discovery was particularly important and useful for photography of the UV and IR (or "ultra-red" as he referred to it) at a time when photographic emulsions were far less efficient. The fact that the lines were parallel on the concave surface was not a problem. He recorded in his paper that four gratings were ruled on flat surfaces and four ruled on concave surfaces with the latter having focal lengths ranging from 5 ft. 4 in. to 21 ft. The example that was available for study was, according to a note with it, broken by Rowland near the end of Stanley MacKenzie's PhD research

34

35

,

RlTTENHOUSE

Vol. 17(2003)

Fig. 5 Note with the grating broken by Rowland himself that ended a nearly completed PhD study.

Tuuaa.