Characteristic Infrared Absorption Bands of Steroids with Reduced Ring A 3-Keto-5a- and 5p- Compounds HARRIS ROSENKRANTZ and PAUL SKOGSTROM Worcester Foundation for €xperimental Biology, Shrewsbury, Mass.
A band for band analysis has been made of the infrared spectra of 27 3-keto reduced ring A steroids, 12 related structures, and 11 acetylated derivatives. The 58configuration gave rise to a combination of bands near
tions was elucidated from an examination of 13 spectra of less complex molecules.
1337, 1299, 1103, and 1005 crn.-' Both all0 and normal arrangements appeared to have characteristic absorption near 1254, 1154, and 965 cm.-' 3-Keto-50- or 5pand 3-hydroxy-5a- or 5p- steroids were differentiated in the 1100 to IOOO-cm.-' region, as it has been demonstrated that ketones and hydroxyls at positions other than 3 have their effect usually at frequencies above 1050 cm.-' Contrariwise the identifying bands of 3hydroxy-5a- or 58- steroids were at frequencies below 1050 cm.-l
Since the earliest frequency correlation studies on steroids, good evidence has existed for the interpretation t h a t absorption bands in the 1100 to 1000-cm.-l region originate from C-0 bonds ( 2 ) . Although prominent bands in this region have been related primarily to hydroxyl groups a t position 3, the C-0 linkage of hydroxy and acetoxy functions a t other positions could be expect,ed t o have an influence between 1100 and 1000 cm.-* NO correlation of infrared absorption in this region and C-0 bonds of hydroxyl groups a t other positions of t'he steroid nucleus has as yet become apparent. If no significant interference of confusion can emerge from C-0 linkages other than those of hydroxy groups a t position 3, then perhaps alterations between 1100 and 1000 cm.-l could be useful for differentiating 3-hydroxy5a- or 5p- from 3-keto-&- or 5p- structures. Either a specific 3-keto effect or a spectral alteration arising from the absence of the 3-hydroxy group could suffice to differentiate 3-keto from %hydroxy reduced ring d structures in the 1100- t o 1000-cm.-' region. With this in view 3-keto-5a- and $3- and 3-hydroxy-5aand 5p- ( 4 ) spectra were compared. -\n examination of the spectra of ergostan-3-0ne~ cholestan-3one, androstan-3-one, etiocholan-3-one, and coprostan-3-one revealed no intense characteristic band between 1100 and 1000 em.-' (Table I ) . Removal of the 3-ketone substituent-e.g., androstane, etiocholane, pregnane, etc.-did not result in a spectral alteration which could be consistently assigned to the loss of the 3-ketone group. In the normal configuration, etiocholane and pregnane, a n intensification of bands near 1008, 1020, and 1064 cm.-l occurred which appeared to be characteristic of these molecules. In general, the 3-ketone function did not seem to exert a significant influence in the 1100- t o 1000-cm.-l region. A comparison of 3-ketone and 17-ketone spectra (androstan-lione, etiocholan-17-one, and 4 2 ( O r 3!-androsten-li-one) indicated that the 17-carbonyl grouping does have an effect in the region under observation. Irrespective of the cis/trans relationship of rings A and B, all three 17-ketone steroids gave rise to intense bands near 1009 and 1052 em.-' The corresponding 3-keto analogs were shown not t o interfere with these bands. On the other hand the lip-hydroxy analogs had a profound eflect in the 1100t o 1000-cm.-l region. The spectrum of etiocholan-lip-ol contained intense bands near 1031, 1039, 1052, and 1069, nhereas the relative intensity of the 1008-cm.-l band appeared diminished. The spectrum of androstan-lip-ol had strong bands near 1028, 1049, and 1069 em.-' -4lthough substituents a t position 17 in the carbon-19steroids have a significant influence in the 1100- t o
[email protected] region, this would not necessarily eliminate use of this spectral zone for distinguishing carbon-21-3-keto and 3-hydroxy reduced ring A st,ructures. Absorption bands arising from steroids Lvith two functional groups are also given in Table I. The two intense bands near 1010 and 1058 cni. -l in the spectrum of androstane-3,17-dione and at' least one of the bands near 999 and 1018 in addition t o that near 1050 crn.-' in the infrared curve of etiocholane-3,17-
I
ANALYSIS O F SPECTRA
?; A recent paper the infrared analysis of steroids containing a
completely saturated ring A (3-hydroxy-50- or 5p- compounds) were discussed ( 4 ) . I n t h a t investigation it was mentioned t h a t a useful approach to structural elucidation would be the application of a combination of characteristic frequencies which could permit assignment of a suspected steroid compound t o a specific structural species. In the hopes of distinguishing 501- and 5p- structures in the 3-keto series and differentiating 3keto-5a- or 58- and 3-hydroxy-5a- or 5p- compounds, consideration was given t o all consistently occurring bands irrespective of intensity. There is no intention t o assign specific interatomic vibrations to particular bands. The situation in the fingerprint region is complex, but a combination of characteristic bands would afford a high probability of assigning certain structural features t o steroids with unknown structures. Thus, certain relationships of infrared absorption to chemical constitution in reduced ring A steroids with a hydroxyl or acetoxy group a t position 3 were brought forth ( 4 ) . In the present study a second chemical species of reduced ring A steroids containing a ketone function a t position 3 was examined. Certain infrared frequencies appeared t o be characteristic for 3-keto-5a- and 56- compounds and permitted ready differentiation from the 3-hydroxy-50- and 5p- structures a hich also have spectral distinctiveness. The spectra of both groups of saturated ring A compounds xere easily distinguished from those of steroids with conjugated ketones by the absence of absorption near 1666 cm.-l Therefore in the isolation of metabolites of e-, p-unsaturated steroids subjected t o chemical or enzymic reduction, the intermediate (3-keto&- or 5p-) and terminal (3-hydroxy-50- or 5p-) reduced compounds can be identified. METHOD
The procedures and instrument employed TTere identical t o those described in the preceding communication ( 4 ) . D a t a from the pertinent spectra in Dobriner's atlas ( 1 ) have been incorporated with the findings from this laboratory. Both solution and solid film spectra were compared with the necessary precautions (5). The infrared spectra of fourteen 3-keto, 5a-, twelve 3-keto, 58-, five 3-keto-5a-acetate, and six 3-keto-5p-acetate steroids were analyzed. The influence of functional groups a t other posi31
ANALYTICAL CHEMISTRY
32
dione could be assigned t o a n effect by the 17-ketone function. The findings may be summarized as follows: A 17-ketone has The monoketo-monohydroxy steroids gave a more complex pican effect near 1008 and 1052, whereas ketone groups a t positions ture Androstan-Iip-ol-3-one was respon.cible for hands near 3,6,11, and 20 had no comparahle influence in the 1100 to 1000 1028, 1042, 1059, and 1078, whereaP etiocholan-17p-ol-3-one had em.-' regions; a 17p-hydroxyl funrtion was related t o several its major absorptions near 999, 1031, 1052, 1064, and 1072 em.-' bande near 1030, 1055, and lOi5, Tvhereas lib-, 17a-, and 21Differentiation of carbon-19-3-keto and 3-hvdroxy reduced ring hydroxyl substituents appeared to be reeponsihle for bands near A compounds becomes difficult again. 1088, 1085, and 1075 em.-', respectively. h band near 1008 As expected, the spectra of androetane-3,11J17-trioneand etioem.-' was also related to the ring . 4 / B stereoisomerism, the norcholane-3,11,17-trione contained strong bands near 1009 and mal configuration giving a n intensification of this hand. There1050 em.-' associated 17ith 17-ketone vibrations. It has already fore in the compounds studied here hydroxyl gr,oups at other been mentioned t h a t the spectra of 3-ketone compounds have posit>ionsthan 3 exerted their more intense effects at frequencies only weak bands a t inconsistent frequencies in this range; this above 1050 cm.-l In contradistinction 3-hydroxy-all0 or norwas also true for 11-ketone compounds. mal structures absorbed below 1050 em.-' ( 4 ) . Some overlap similar situation existed for the 20-ketone function. K e a k may occur in solid film spectra for the 38-, 50- configuration and bands at inconsistent frequencies occurred in the spectra of allo176-hydroxy or 17-ketone compounds. The latter could be pregnane, allopregnan-2O-one, allopregnane-3,20-dione, and allodistinguished by the characteristic pentacyclic ketone absorppregnane-3,6,20-trione. I n the latter compound little influence tion, but the l i p - compound would be troublesome. Although was related t o the 6-ketone grouping. The strong band near 3a->50-configurations absorbed between 1000 and 1009 em.-' ( 4 ) 1008 cm.-' associated with the normal configuration of _____ rings :l and B appeared in the spectra of pregnane, pregnancTable 1. Influence of Ketone and H>droxyl Functions of Dih?dro Steroids in 1100- to 1000-Cm.-' Region 3,20-dione, pregnane-3,11,20N o s t Intense Frequenci+ brrlsren 1100 and 1000 C m . 2 trione, and pregnan-lie-olCompound" 3,20-dione (Table I). A. N o Functional Groups I n the more complex moleHydrocarbon b Androstane cules with three, four, and five 1085 Etiocholane lO"O9 I022 Ergostane functional groups, assignments b b Cholestane of absorption bands to certain 1063 1017 Pregnane lOOG .. . . Allopregnane hydroxyl groups were apparent u. One Functional Group (Table I). 21-Hydroxy steroids Jfonoketone gave rise to a band between b 1022 Androstan-3-one 1070 and 1079 em.-' which did Etiocholan-3-one 1001 b 'd Cholestan-3-one not depend on the type of side b loot: : : Coprostan-3-one b Ergostan-3-one C chain-e.g., lie-hydroxy-alOi3 1009 b Androstan-l7-one ketol, a-ketol, or simply 211051 1006 b E t iocholan- 17-one 1044 inno b A2-Androsten-17-one hydroxy like allopregnan-21-01h b .. Allopregnan-20-one 3,20-dione. T h a t t h e l i e RIonohydroxy b b 1028 1049 10G!l Androstan-178-01 hydroxy grouping has a dis1069 1052 in06 .. loaid Etiocliolan-l7~-ol tinct frequency is shown b v Two Functional Groups C. the spectra of allopregnan17a-ol-3,20-dioneJ 1088, allo1058 inin b 10.50 Y99 1018 pregnan-17a-01- 3,i 420 - trione, b I005 b h 1085, pregnan-lia-ol-3,11,20b b trione, 1083, and pregnan-liaMonoketone-monohydroxy 1078 b ., 1028 10311 in42 Androstan- 178-01-3-one ol-3,20-dione, either 1081 or 1072 1064 1052 999 .. 1031 Etiocholan-178-01-3-one 1090 cm.-' I n 17eJ21-dihyD. Three Functional Groups droxy steroids merging of lieh hydroxy and 21-hydroxy vibra1049 1010 b h 1049 1007 1016 tions occurred. .. , . h 1072 io02 ibis T a a compounds were availSfonohydroxydiketone able \r.ith a n Ilp-hydroxyl sub, , 1081d 1062 100.5 Pregnan-17cr-ol-3.20-dione .. s t i t u e n t , dndrostan-1 1p-011005 Prepnan-1 la-ol-3,ZO-dionec 107:) b b . . Allopregnan-2 l-ol-3,20-dione 3,17-dione strongly absorbed 1088 h Allopregnan-l?a-ol-3,20-dione . . 1088 1063 in16 , , 1029 Androstan-1 1&01-3,17-dione near 1018,1065, and 1088 em.-' A shift t o higher frequencies E. F o u r Functional Groups has occurred for the 17-ketoncl Nonoliydroxytiiketone 1071 1083 1003 Pregnan-17ar-ol-3,11,2O-trione bands (1018 and 1065 em.-') in .. 1085 10r;l'l Allopregnan-17a-ol-3,11,2O-trione this solid film spectrum. AlloDihydroxydiketone b 1070 1088 h p r e g 11a ne- 1 1p , 2 1- d i o1-3,20Allopregnane-ll(3,21-dio1-3,20-dione 1073 b b Allopregnane-17a.21-diol-3,20-d10ne~ 1078 b dione gave rise to intense band8 lbb') 1009 Pregnane- 17a,2l-dio1-3,20-dionec near 1076 and 1088 em.-' The Five Functional Groups; lOiG-cm.-' band has been Diliydroxytriketone 1076 .. b 1055 shown to originate in part from Allopregnane-17a,2 1-diol-3.1120-trione C 1070 1041 1007 Pregnane-17a,21-dio1-3,11,20-trione the 21-hydroxy v i b r a t i o n s . Therefore the 1088 cm.-lwould a Most of spectra were studied in solution and can he found in Dobriner's atlas ( 1 ) . b Weak bands are present near these frequencies. zppear to represent some c Spectra obtained on solid films. d Another band is present a t a slightly hipher frequency. :tctivity of the 1lp-hydrox-1 __.___ ~function. I
~
.
12
I
~~
V O L U M E 2 8 , N O . 1, J A N U A R Y 1 9 5 6
33
and in the present study a 17-ketone group also absorbed near this range, the pentacyclic ketone vibrations could easily differentiate the structures. It would appear that the 1100- t o 1000cm. spectra range can be useful for distinguishing carbon-21-3 hydroxy-&- or 5p- and 3-keto-501- or 58- steroids. I n addition to this characteristic absorption between 1100 and 1000 cm.-' the 3-keto-50- or $3- steroids were differentiated from 3-hydroxy-5a- or 58- compounds by the consistently occurring bands of weak t o medium weak intensity near 1254 and 1154 cm.-'in both solid and solution spectra (Table 11). Furthermore other bands, which aided in distinguishing 5a- and 5p- configurations in the 3-keto series, were also useful in separating 3-keto and 8-hydrouy reduced ring A structures. A combination of characteristic absorption bands related t o the steric conformation of rings A and B occurred near 1333 (solid), 1340 (solution); 1302 (solid), 1297 (solution); and 1101 (solid), 1105 em.-' (solution) in the spectra of the $3- configuration (Table 11). Two steroids out of 22, etiocholane and pregnan-170-01-3,11,2O-trione,did not contain all three of this characteristic series of bands for the 56- structure. -4lthough many 5a- compounds had bands near these frequencies only the spectra of ergostane, A2-androsten-1ione and allopregnan-20-one (three out of 28 steroids) contained all three bands. I t can be seen in Table I1 t h a t another band appeared t o occur with consistency in the spectra of 3-keto 5a- and $3- steroids.
Although the intensity of this band varied from medium weak t o strong, it was observed near 963 (solution) and 967 cm.-1 (solid) in the spectra of all allo compounds. T h e situation was somewhat variable in the normal structures; 11 of 12 solution spectra containing a band near 965 while the band was shifted out of this range in solid film curves. I n agreement with the observation on free 3-hydroxy-5a- or 5p-21-desovy steroids containing no 17-hydrovyl group ( 4 ) was the finding that a n intensification of the barid near 1157 cm.-' occurred in the spectra of allopregnane-3,20-dione, pregnan-llaol-3,20-dione, and pregnane-3,11,20-trione. A serious interpretation of the 1250-c1n.-~region could not be undertaken for the acetylated structures because of the limitation of available compounds. However, some observations are worth noting. Most of the spectra of the acetates contained more than one band in the acetate region near 1258, 1225, and 1236 em.-', the latter being of greatest intensity, whereas the other frequencies sometimes appeared only as side inflections. Readily distinguishable from this combination of frequencies was the singular absorption of allopregnane-11~,21-diol-3,20-dione-21-acetate, and the two 20a-hydroxysteroids near 1233 cni.-' The other 118- hydrouysteroid, pregnane - 11p,l'ia,21-triol - 3,20- dione - 21 acetate, gave rise t o four bands, 1258, 1244, 1236, and 1230 cm.-' Furthermore the 21-ol-3,20-dione isomers seemed t o absorb differently, the allo form being responsible for frrquencies near 1253, 1238, and 1233, whereas the normal arrangement was near 1218 and 1230 em.-' Finally the spectrum of the only Table 11. Combinations of Characteristic Frequencies in Spectra of Reduced Ring 4 Steroids" a v a i l a b l e carbon-19-3-ke t o Frequency Combinations Correlated u i t h saturated acetate (etiocholanCompound b 3-Keto-5a- and 3-Keto-58- Configurations 174-01-3-one-17-acetat e) had its Allo major bands near 1255 and Androstanec 1302 S 958 S 1334 nI 1305 11 Ergostanec 960 S 1221 cni.-l, different from all Allopregnanec 1331 S 9 59 S other curves. 1338 S I291 W 1231 Cholestanee 958 S Ernostan-3-onea 958 A i 1306 w 1251 After this work was coniChhestan-3-onec 961 M 1335 ni .. 1253 Androstan-3-onec 1255 1340 W .. 968 1%pleted, a paper appeared by Androstan-17-one 1339 n i 1289 9 59 hi Jones, Herling, and KatzenelA~-Androsten-17-oneC 1338 S I295 31 1256 961 11 Allopregnan-20-one 1339 W 1292 31 1259 968 S lenbogen ( 3 ) discussing the inAndrostan-17B-olC 1331 31 1305 W 1259 958 M Androstane-3,17-dioned 1339 nI 1258 9 69 w frared spectra of ketosteroids Allopregnane-3,20-dione 1292 w 1250 964 W below 1350 cm.-l T h e findAndrostan- 178-01-3-oned .. 1258 960 w Androstane-3,11,I7-trioneC .. 1297 w 1255 967 w ings of this laboratory are in Allopregnane-3,6,20-trionec 0 0 0 958 h1 Androstan-lIB-ol-3,17-dioned 133-1 w 1297 W 1249 965 11good agreement with theirs. .4llopregnan-l7a-ol-3,20-dione C 1300 W 1255 968 wThe spectra of all 3-keto-58.4llopregnan-2 I-01-3.20-dionec 1297 nf 1255 963 11 Allopregnane- 118.2l-diol-3,20-dione d .. 1252 962 1T structures contained bands near Allopregnane- 17a ,2 l-dio1-3,20-dioned 1340 n i 1252 9 69 w Allopregnan- 17a-ol-3,1120-trioned 1247 1333 AI 970 n i 13317, 1295, 1104, and 1005 Allopregnane-l7a,2 l-diol-3,Il ,ZO-trioned 1250 971 A i em.-' Both 5p- and 5a-conAllopregnan-20a-ol-3-one-20-acetated 1335 W 1297 K O 966 W Allopregnan-21-ol-A,2O-dione-2l-acetated 1339 w 1297 11 1253 970 W figurations gave rise t o bands A l l o p r e g n a n e - l 1 ~ .l-diol-A,ZO-dione-2 2 1-acetate c 0 0 971 W Allopregnane-l7a,21-diol-3.20-dione-21-acetated .. 126.1 9 69 w near 1254, 1154, and 965 em - 1 Allopregnane-17a,21-diol-3,11.20-trione-21-acetated 1340 1233 968 n1 T h e li-ketone grouping was reKarma1 sponsible for strong absorption EtiocholaneC 1299 S 1253 1342 S 960 S PregnaneC 1336 31 1305 M 1250 962 3 f The near 1008and 1050 n Etiocholan-17-one 1292 ni 1260 968 W 20-ketone function did not have Etiocholan- 178.01 1338 hi 1308 b l 1250 958 JI E tiocholan-3-oneC 1340 hI 1299 11 12.56 959 hi intense absorption in the 1100Coprostan-3-onec 1340 n1 1297 11 1256 967 11 Etiocholane-3,I7-dioned 1337 nl 1302 31 1260 t o 1000-em-1 repion evcept for Pregnane-3,ZO-dioned 1335 M 1300 W 1260 the expected band near 1005 Etiocholan-17~-ol-3-oneC 1339 ni 1294 W 1283 970 ni Etiocholane-3,11,17-trionec 1340 ni 1295 JI 1265 968 IT cm.-' characteristic of 5pPregnane-3,ll ,ZO-trionec 1340 RZ 1299 31 1256 Pregnan- 11a-ol-3,20-dioned 1328 W 1307 JV 1252 0 structure.. C
Pregnan-l7cr-ol-3,20-dionec 0 Pregnan-l7a-ol-3,11.20-trionec 0 Pregnane- 17a,21-dio1-3,20-dioned 1335 Pregnane-17a,21-diol-3,11 ,ZO-trioned 1332 Etiocholan- 178-01-3-one-17-acetated 1339 Pregnan-20a-ol-3-one-20-acetatec 0 Pregnan-2 l-ol-3,20-dione-2 1-acetatec 1339 Pregnane-l7a,Z 1-diol-3,20-dione-21-acetated 1340 Pregnane-17a,21-diol-3,11,20-trione-21-acetated 1344 Pregnane-I 1~,17a.21-triol-3,20-dione-2l-acetated1337
n
W W XI 11 3T
ni ni
n 1299 w 1302 IT 1294 31 1292 W 1292 M . 1299 M 1299 nr 1300 W
0
n
967 ni 968 XI
..
0 0
972 R'
..
..
ACKNOWLEDGMENT
The authors wish to thank the following for samples of compounds investigated: A. S. Meyer for androstan-llp-ol3,17-dione, allopregnan-21ol-3,2O-dione, allopregnan-17001-3,11,20-trione, allopregnanelTa,21-diol-3,11,20-trionc,
34
ANALYTICAL CHEMISTRY
allopregnane-11fl,21-diol-3,20-dione-21-acetate,and allopregacetate; Kenneth Savard for pregnane-llp,l7,a-21-trio1-3,20nane-17a,21-dio1-3,11,20-trione-21-acetate;Alika Hayano for dione-21-acetate; and Lederle Laboratories Division, American allopregnane-1 1p,21-diol-3,20-dione; Enrico Forchielli for allopregCyanamid Co. for ergostan-3-one. nane-17a,21-diol-3,2O-dione and pregnane-17a,21-dio1-3,20LITERATURE CITED dione; Harold Levy for allopregnane-3,20-dione and allopregnan17a-ol-3,20-dione; Ciba Pharmaceutical Co., for androstan-3(1) Dobriner, K., Katzenellenbogen, E. R., and Jones, R. N., “Infrared Absorption Spectra of Steroids, An Atlas,’’ Interscience, one, androstane-3,17-dioiie, androstan-17P-ol-3-one, and alloNew York, 1953. pregnan-21-01-3,20-dione-21-acetate;R. I. Dorfman for andro(2) Furchgott, R. F., Rosenkranta, H., and Shorr, E., J . B i d . Chem. stane-3,11,17-trione, pregnan-1 la-ol-3,20-dione, and pregnan163., 375 .(1946). , 20a-ol-3-one-20-acetate; Merck & Co. for etiocholan-17fl-ol-3(3) Jones, R. N., Herling, F., and Katzenellenbogen. E., J . Am. Chem. SOC.77, 651 (1955). one and pregnane-17aJ21-diol-3,11,20-trione; Parke-Davis Co. (4) Rosenkrantx, H.. and Skogstrom, P., Ibid., 77, 2237 (1955). for pregnane-3,aO-dione; Glidden Co. for etiocholane-3,17-dione; 25, 1025 (1953). (5) Rosenkranta, H., and Zablow, L., ANAL.CHEW. Frank Ungar for allopregnane-17a,21-diol-3,20-dione-21-acetate, RECEIVED for review June 29, 1965. Accepted October 17, 1955. Supported et iocholan-l7p-ol-3-one- 17-acetat e, and pregnane-1 7a,2 1-diol-3, by a grant from Medical Research and Development Board, Office of t h e 20-dione-21-acetate; Kicholas Saba for allopregnan-20a-ol-3-oneSurgeon General, Department of .4rmy under Contract No. DA-48-007-M D20-acetate; John Davis for pregnane-l~a,21-diol-3,11,20-trione-21-310.
Flame Photometric Determination of Sodium, Potassium, Calcium, Magnesium, and Manganese in Glass and Raw Materials NORMAN ROY Research laboratory, Thatcher Glass Manufacturing Co., Elmira,
Use of the hydrogen flame attachment and photomultiplier tube with the Beckman Model B spectrophotometer in the analysis of soda-lime and borosilicate glasses, as well as various naturally occurring materials, such as sand, slag, and feldspar, is described. The range of concentration of the various components is wide and the accuracy obtained in the determinations varies, but in most cases is within 0.3 to 0.5% of the amount of the component present in the sample. A single disintegration of the material with no preliminary separations is employed in determining all of the components listed.
S
OME flame photometric techniques involve preliminary
separations from interfering elements, whereas other techniques call for comparison of the unknown solution with a standard ivhich incorporates some or all of the interferences present in the unknown solution. Some authors add known or excess amounts of the interfering cations t o both the unknown and the standard solutions, with or without first completely removing the interferent from the unknown solution by precipitation or by an exchange resin. Although the flame photometric determination of the alkali, alkaline-earth, and certain other elements has considerably shortened analysis time, the preliminary separation and/or addition techniques detract from the speed and simplicity which might ultimately be attained. For this reason and because it was felt that the best accuracy could be obtained by making the standards as similar as possible to the sample solution as proposed by Gilbert m d others ( 6 ) , no preliminary treatment other than solution of the sample was employed in the method adopted and here described. hlost of the flame photometric methods described in the literature employ a water or acid solution, or acid extraction of the sample for the determination of the alkali and alkaline-earth metals. Sintering is also used to effect eolution of the sample for flame analysis. Where the sample is water- or acid-soluble, this technique is perfectly feasible. iicid extraction has the disadvantage of leaving part of the sample undissolved, and sintering of the sample is relatively time-consuming. T h e method of solu-
N. Y.
tion described herein takes less than an hour for most glasses and raw materials. The powerful oxidizing action of boiling perchloric acid combined with the volatilizing action on silica and boron of hydrofluoric acid provides a sample solution technique which ie sufficient for all glasses known to the author and for most glass raw materials. The development of the photomultiplier tube for use in connection with the spectrophotometer has unquestionably extended the usefulness of flame analysis. N o w it is possible t o use the broad selection of sensitivities of the photodetector on the spectrophotometer in analyzing a single solution containing a number of components of tvidely varying emissivity. Although the Beckman 3Iodel B spectrophotometer with No. 9125 flame attachment was used in developing this method, the Reckman Model DL with flame attachment can be used, with no essential change in the method. The Model B is a direct-reading instrument and might be slightly faster in reading than the DC; the D U is undoubtably more sensitive and selective.. GENERAL CONSIDERATIONS
In all quantitative analyses, the qualitative make-up aiid approximate quantities of the components present in a giveii substance must be knowi, if the analysis is to have reasonable cert,aint.y of success. This information must also be knoll-n in making up a standard for a glass or other material which is to be analyzed by the method described. The needed information may be obtained from analyses of similar material, such as glass from the same tank, or examinat,ion of the flame spectrum of a totally unknown material will give a quick qualitative analysis for sodium, potassium, calcium, magnesium, and manganese, as \yell as lithium, strontium, copper, etc., and then a rough quantitative analysis may be run off immediately for the elements detected. In the latter case, it is most convenient to make an approximately 0.1N hydrochloric acid solution of the material t o be examined and compare i t with solutions of each of the chlorides of sodium, potassium, calcium, magnesium, and manganese following in outline the procedure for glass analysis. I n the case of a
