1470
ANALYTICAL CHEMISTRY
Sovotny ( 5 ) proposed that the silver in the precipit:ito I)(, ( i t \ rerniined. T h e disadvantages of using silver ion are mainly due to t h r interference of any halides, cyanide, sulfides, and traces (down t80 0.01%) of aldehydes in the sample. These interferences vonsume 4lvei. ion, the halides and sulfides by precipitation of t,htJsilver d t s , the cyanides by complex ion formation, and the aldehydes I)y reduction of the silvtar ion to niet:tllic silver. Traces of aldt~liydes affect the deteruiinntion of the nitric acid liberated t)y the 4 v e r nitrate (1, Z), Foi, tlie nietallic silver formed obscurrs tht' r ~ n point d completely. Sriittll :+mount,sof aldrhydes (up io 0.59 :L.* formddehyde) cttn lw t o k m t d in the procedurc 1)c~low. 1 !nwver, because aldr.hydc+ arc' oxidized by the potassium mt?rvuric iodide, amounts larger than 0 . 5 9 will begin t'o affect th(h rtcetylenic hydrogen result P. Wellers (10) absoi,bed thc #:in in slightly alkaline solutions of ium mercuric iodides. .-\wtylene has also been estimated by lining the mercury in tlir :icet,ylide precipitate ( 4 ) by tit,rating with ammonium thiocy:tn:rte by the method of Volhard. FIX purposes, Shriner and Fuson (8)reconiin mercuric iodide t o form the mercuric. $(lerivative of those c~ornpouritis whirti rontain the grouping -C'=CH, according to the formula:
2lfor hoth opaqiir and clear liquids. The sample stream flowwntiriiiousl), and ma? he under pressure. Good stability has been achieved w i t h a maximum sensiti\it> of *0.00005 refractive index unit. The action depends on intensity of internal reflection. near the critical angle, in a prism i n routact with the sample streani. iist.,
R
I efle(*tiwi t echiiiquc. The aiiigle-rc~flc~ri io11 pnn(vpk m:&y be t~:E'RACTIVh;iritirx has long been recognized :ts it funclautilizrtl by a trapezoidal prism as shown in Pigirrt* 1 and the mental propert>-of niiittrr, but in spite of its long usage as multiple-reflection one by a glass rod as shown in Figurr 2 ( 2 ) . :i itleasure of chemical purity it has fuund little :ippIiratinn HS This papcr is concerned with the application of tlie single-reflw:I meiiiis of process control in clic.mica1 industry. Thrrc. are basically t v Y J types of refr:ictivtl index nie~murenient~ t itjri technique to n continuously recording refr:rrtoinrtrr in gcnwil use: One depends on refraction of light :it the h o u n d a ~ y THEOHl OF I N S I R U M E h T Iwtween the sample and N rrfrrcnce medium, rind t h t s ot1ic.r tic ] ) t w d s upon the form:triori of' iiitc~rfcwrricefringes t o rstim:itr thv 111 I'igui.c. 1 :I parttllcl beam of light i3 reflected f r o i i i ,I gldsb1.fftvtive optical path length. -nniple boundary. If the :ingle of incidence is less th:tn the It has heen long knowri :ind recent.ly emphasized (3)t h t rvmiticd, the amplitude of tlir reflected light will depend on the I r:tctive indexes of substan *re correlated with the iiiitwsity 01' dirrctiori of polarization o f the incident light, on t h r :tngle of Itle reflected compontwt of light impinging on a pri,sm-s:tnipl(~ incidence, aiid on the ratio of refractive Indexeq of the #la I)oundary. For a fixed angle o f i n c i d e n c ~the intrwsilg of tlie ribsninple. 1 - s ~rnn be ni:tde of Fresnel's equations ( 1 ) where the ficlcted light is dependent upon the ratio of refractivcr indtsxcas of t h r amplitude of :L reflected ray of light, the incident light being I h n i and the liquid. If the reflected light is converted into an polarized normal t o the plitnp of incidrncc arid of unit :implitude, (.lectric:il signal, this principle Imniiir~siisahle in continuously rili.; in the notation nf Figurr 1 .(.I )rding instrument s. This method may t)r t:xploitc~l1 ) ~ either :t single- or niultipleI
I'ri.+nt
address, Comiuercinl So1rc~nt.iCorporation, Terre Hantc. Ind.
1471
V O L U M E 2 1 , NO. 1 2 , D E C E M B E R 1 9 4 9
lteferriiig t,o Figure 1, at m:tximuiii sensitivity p = !!O', S ~ that from Equation 2 sin 01 = l / n . It, is apparent, however, that under these conditions the prism :tirgIo should be exactly eqii:iI t o 01. The authors' application required the control of a liquid :it .I refractive index of about 1.39. L-sing glass of refractive i i i i I v \ 1.51, the piisni was constructed ivith :L prism angle = siii-' 1.39/1.51 = 67'. By making the angle of incident light i i i i i l tlil. position of the detector adjustable, it \vas found possiblr t r i n r : t k i ~ n e a s ~ ~ r e n iover e ~ i ta range of from 1.36 to 1.42. Til(. romplete optical systeni eniplogrd :appears in Figurtx I
Figure 1. Trapezoidal Prism f o r C tilizatinn hf Single-Reflection System
Light
Figure 2. Cylindrical Prisni for IJtilization of R1 irltipleReflection Sy-stem
T h v i i i t cwsity oi the rcfirct t d light I i o ~ v ~ v eisr , proportio1ial t o d thtr square of the aniplitutie. TXng uripo1:irized light ~ n defining t h r reflrctivity, l', :IS tllr. r:ttio o f tho inttJnsitie.qo f t,he reRe(-tcd:ind iucidi,nt light, ~
The light source consi of it tungsten filament projectioii l : i t i i i i , arid collimator, the collimator being provided with a holdrr f o i filters or polarizing elemmt,s. Inasmuch as red light i, thfb primary source of drift in barrier layer cells a light blue filter wac made a part of the collimator. A Wratten'ko. 43 filter was l'ouiitl effective in removing the red elid of the spectrum without seriousl! reducing the intensity at, the shorttir wave lengths. AlthougIi provisions were made for the use of' polaroid in the light, p i l i , insufficient data have been ohtitined to date t o determine the effect of polarized light on the operation of thcs instrument. T h r h w i n iy l i t into two components, one passing through the samplr prismt the othrr through a reference prisiii. The reflected light frwni the. prism-liquid boundaries is finally cmiverted t o an e1ectrir:il signal t>ysuitable detectors after it p:~ssrsout of the prisms. Detection and Recording. 1:it hci, vacuum phototutltts ( 1 1 \):iri,iei,liiyer cells are suitable as thr photosensitive elements. 13:ii.Tier layer crlls are easier to use, t)rcause they require no powt'i $uplily, arid when prop?rly converted into it null balancr ciwiiil 1 tit, difference in the output of t h r t w o cells may be r~c*orclcdl i ~ :tny standard recording €"'tt,nt,ionirter, a-hicli has hem i i i o ( l i t i t ? O for this purpose. 100-
.80 -
, l
o!=eoo
.60n
l)lh
O F Sl?i(;LE-KEFLE(X'ION KEFK.iC'I'O\IETER
.'iI)t:cific conditions iinposecl on a cont,rol instrument mi'e: (1) c*oiitinuous flow, ( 2 ) recording features, (3) pressures up to 150 1)oulidsgage, and ( 4 ) operation with opaque liquids; three added pr,iriie technical consideratioris which had to be kept in mind in the cLt4gn of t,he plant instruniririt m r e ( 5 ) sensitivity, ( 6 ) stahilit,y, :riicI ( 7 ) ternperature coinpens:tt,ion. .in instrumeiit !vas built ut~ilizirig thr! above principles arid i i i c 4 n g thew contlitions. Sufficient data were ohtaiued with this iiiodrl t,o prove t h r fca.sihility of such ii rofr:irtometw fnr plant cwntrol. Optical System. Because the range of optimal sensitivity of R rcflectrd light refractometer is limited, it is necessary in the tlesign of t,he optical system t o knoiv the range of refractive index to he measured in a specific application. The refractive iudex and geometry of the prism best suited for that range can then be dcterminrd. It is desirable that the light. pass into and out of the Ijrisni a t near noma1 inciderice. This retfured bending of tlie light, :tt t h r fares of the prism and simplified the positioning of the cltitrctirig c~leniriit. I n order that t,he light may pass into t,hr prism :it iicwly noimiwl incidence :tiid still he within the region of most seiisitiw oper:ttion, the refractive index of the liquid to h iiieasui~ediiiust htl knon-n. The prism angle can then be deteriiiineti i n twins nf the rrfractive i n d m of the glass and t h r liquid.
I
I',v 2 c .40 -
20 -
1lie circuit sholvri in Yiguith .i \v:i. uheJ mccessfully \vir ti : I iiiudified Brown Electronik recorder. If vacuum photot,uhrs : ~ I Y * used as the detecting elements, it special type of recorder i i needed to match the high impedance of the tubes. -4 recorrlw mariufaetured by Leeds dz Sort,hrup for pH measurements wah found to be well suited for phototube measurements. Essentially it is an electronic milliroltmrtrr rnlihrated t o record from -209 I .
ANALYTICAL CHEMISTRY
1472
For example, using the circuits of Figure 5 let Rt = R, = 1000 ohms and the slide-wire = 300 ohms. Initially assuming 15 pa. as the current generated by each ecll, theslidewill be st mid-scale, and the potentials developed a t A and B will be 1150 ohms X Mechanical Details. One of the difficult problems associated 1 5 ~ =. 17.3 mv. Assuming the reference cell current t o remain unchanged with the slide a t the extreme left of the scale, the p" with the construction of the refractometer was t h a t of pressurtentid a t B will be 13W ohms X 15 pa. = 19.5 mv. For balance izing the prisms and making i t possible to remove them easily for the current output of cell A must increase to 19.5 mv. + 1000 cleaning. Figure'lisa photographof theinstrument,and FigurPX ohms = 19.5 pa. Similarly, with the slide a t the extreme right the qhows the construction details of t,hn prism assembly. current output of cell A must he reduced to 11.5 pa. for balance, the balance voltage having changed to 15mv. Hence, for full-wale deflection the current output of A has changed Somple P r i m by 8 pa. and the balance voltage by 4.5 mv. Assuming I 1 the current, OUtDUt,of the oells to be direetlvDroDortional to the reflected liiht intensity, an S-+. char;& i d t h e output of cell Aisappronimatelyequivalenttoa.50%ehangeinT. Using this information, the range of refractive index Pbol covered by a full-scale deflection can be approximated. ClCCll Assuming no = 1.51 and n, = 1.40, i t can be seen from devicL. an expanded scale of Figure 3 thitk in the region of most Filter Collimator sensitive operation a change of 50% in T corresponds to Light a refractive index change of about 0.005. If a recorder %".Le with a minimum resuonse of *0.03 mv. is used. the expected recorder acc&cy will be 0.03 mv. i4.5 mv. = *0.7% of full scale or, in terms of refractive index, better t'han *0.00004. Figure 4. Optical S y s t e m for Single-Reflection Refraotomerer to +ZOO millivolts. The null balarrce circuid of Figme 6 inoorporating No. 929 phototubm operated very well when coupled directly to the recorder.
c-
a
The prism housing is entirely of siainless steel. The four edges of the base of t.he prism rest on a ledge bordering the liquid compartment. A ganket, cut to fit the ledge, prevenbs any leakage from the sample or reference chambers. The top of the prism is covered hy a strip of gasket material and above that is a metal plate. Three bolts, which can be reached st the top of the housing, seat on the metal plate; when these bolts are tightened the prism is forced againat the lower gasket to accomplish the seal. By loosening the bolts and removing the face plate, which supports the detecting element, the prism can he easily removed from .he housing. Reference Cell
,
Slide Wire
Figure 9 represents the results obtained when the above conditions were imposed experimentally on blends of benzene and butyl alcohol. The sensitivity is somewhat less than predicted, full-scale deflection representing a refractive index change of about 0.01 or a sensitivity of 0,0001 per chart division. I n practical applications i t was found that th? sensitivity did not change Stondord
Reference
2Y+TuL
I
F i g u r e 6. Electrical C i r c u i t Utilizing Vanuum-Type Phototubes and Leeds 8: N o r t h r u p pH Recorder
V O L U M E 21, NO. 1 2 , D E C E M B E R 1 9 4 9
1473 appreciably over a range of 0.01 refractive index. I n general, t,he sensitivity can be increased by increasing the intensity of the light source or by increasing the resistors, R, and R 2 . Lnfortunately. however, increasing the light intensity alters the internal characteristics of t h r cells, so that a high external resistance becomes impractical. It is generally preferable t o keep the output of the cells as low as possible, consistent with the minimum response of the recorder, while increasing Rr and Rz to obtain the desired sensitivity. If phototubes are used instead of barrier layer rells, the recording mechanism becomes somewhat more complicated, although much more effective. The maximum sensitivity possible with the circuit shown in Figure 6 is almost unlimited, for several volts can be developed bl- the t.uhes from light wurces of moderate intensity. For practical purposes a filtered light, source of IOQ- inknsity results in wcellent operation and adequate sensitivity. Fxperimental results are illustratd in Figure 10, obtained on benzene in butyl alcohol solutions! For various load resistors. From these results and from subsequent stability tests an over-all accuracy of *0.00005 refractive index unit w a p determined. Stability. Twenty-four-hour stability runs were made using barrier erlls and phototubPs as the detecting elcmente. T h r sensitivity n m set a t 0.0001 refradive index unit per chart divi~ionin both cases. Runs were niadr on a rontinuoudy circulating sample and also wit,h air in t h r sample and reference chambers. With the liquid chambers empty, no dotectable drift o c w r r d during the 23
PE RS PECTl V E VIEW
__
Seating Set Screws Seating Plate Teflon Gosket
Photocell Seat _ _ -
Rubber
S E C T I O N A L VIEW Figure 8.
Construction Details of Prism Assembl?
80
70 -
70
60
01
5 50
u
0 0)
c
12:
5 40
r
L
0 L 0)
8 30
0 0,
n
20
0
2
I
0
1
10
4 6 8 R . I. Difference x IO3 I
I
I
I I
2 4 6 8 10 Per Cent Benzene in Sufyl A l c o h o l (by volume)
Figure 9.
Experimental Data Obtained with Circuit Shown i n Figure 5
2
0
I
1
0
4 6 8 R. I. Difference x IO3 I
I
IO I
I
2 4 6 8 IO Per Cent Benzene in Butyl Alcohol ( b y volume)
Figure 10.
Experimental Data Obtained with Circuit Shown in F i g u r ~h
1474
A N A L Y T I C A L CHEMISTRY
hours of opwatioii. \Vhen liquids w ~ r vir(-ulat,ed t ~ in thcx samplv and reference systems, no drift occurred, but some slight instabilities were caused by small bubbles passing along the interface. ;\t no time, however, wrre there deviations of niorp than 0.5 chart tlivision (0.00005 1t.I.) from the initial setting. Temperature Compensation. The main advarit age of the null balance system used in this instrument is to eliminate effects due l o tc~inperat~urr~ rhanges in the sample. By means of a suitable h w t cischanger the sample and reference liquid can be maintained :tt the same temperat,ure, which thereby reduces the degree of t enipcwt.ure control nrcessary. I f T I and , t t , are the refractive indexes of the sarnple :inti reference liquids, respectively, t hrrr thc. ctriitlit.iniisfw good rompensat,ion a r ~ : I!,
-
TI,
f 1.8 !.nI1,1,IP
20 -- ! l , . ~ r c f t w 1 w
20
(:OY(:LzIsIo\ r .
lht! refractomet,er provides a coiitiiiuous and accurate n i i ~ t l i o i l of iwording the refractive index of any liquid which does not h v i . :i tendency to deposit a film on a glass surface. Even a veq- t h i l i film will seriously interfere wit.h operation of the instrument. It i. independent of any coloration in the liquid. Very little n i : t i i i tenance is required, except for periodic cleaning of prisms ant1 I ' ~ L st.andardization, and it will operat,e for long periods of t,imr \vitlt ;it1 :icrur:iry nf better than +0.0001 refractive index unit. 4 C K N O W LEI)GAI ICNT
.ippreciatio1i is expressed to J. li;. Glarsuii mid Walter Sch\v:irtz K. Bowman for valuable comments: t i , I\'. A . Hariior and G . €1. Young for frimdly encouragemrwt : to t'w operative data; to J.
f I , l{, Stevenson of the Procter and Gamble Company, whn riiiiiiltanr~oudyTvorked n-jth reflrcted light type refractometers; : i n ( l t o John I*nvrtl, Jr., of the LTnertlOptical Company, Pittsburgh, Pa.. for. ni:iny helpful merhanic:tl s u g g e ~inn.: t (luring the construe-t i o i l of bhr instrument.
I)c.ttLctoi. ~:liiti~i~(~t~~i.i~tiCs must be the sanit~.
TIit. greater the deviat,ions from these conditions the poorer this twinpensation. The t,hird condit,ion can be satisfied closely, anii :tlso the first condition, depending upon specific conditions. The second and fourt,h conditions, however, can only be approximated, so that, in general, absolute temperature variat,ions of nior(1 th:in 1 ' C. will seriously :tffec.t t8hracciir.:iry of thr. instrunic~ri(
L I T E H ~ T O K Kcrrm
I: H I ) I , I IMax, , "Optik," Ani1 Arbor, Mieh., Edwards Brothels. I !W3. (2) ( ' a r y . €1. H., Applied Ph;\,sicsc:orporation, Pasadena, Calif.. 1111yuhlished report of use of this method. (::i)Fehruary 16, 1949. Contribution of niultiple industrial f d l o w . .hi], o f r h t , KPronst,rurtion F i n a n r ~Corliorntion. Offive of Rubhpr R z ' s i . r ~ c '
Pressure-Tight Infrared Absorption Cel for liquids Infrared spectroinetric anal?sis of mixtures of an) or 1111 of the CI olefins and paraffins in the liquid phase at rooin temperature has heeri impracticable because thin ahsorption cells could n o t he made tight enough to retain, without leakage, liquids haling high vapor pressures. itial?sis in the liquid phase has lieen considered as a means of a\ oiding inaccu racies resulting from 5anlplilIg a liquid mixture for anal?sis in the \apiir phase. The design and con-
r
r
HE an:ilysis 01 mixtures of an). or all of the (', olefins iind paraffins in the liquid phase at room temperature by infriii,ed sprvhomcbtrir niethods has heretofore been imprscticwhlt. I)(\( ~ u s thin r :iI)sorption cells could not be made tight enough t o ret win, without leakage, liquids having high vapor pressurc.~. in the liquid phase is a possible means of avoiding thc inaccwracies rc=sulting from the difficulty of securing it saniplr, in the vapor ph:tse which has the same composition :is thix liqiiiil in t h e container from which it \vas withdrawn ( 3 ) . I)ifficultic~sare also encouritercd in preventing t,he loss fi,oni liquid absorption cells of volatile solvents such B S ciirl)on tetrarhloritie and carbon disulfide with consequent change o f (*onoentration of solutions while recording their spectra. -1number of workers (.$, 7 , 8) have designed cells to hold volatile liquids, but, of the cells thin tmough to permit the study of' pure liquids without dilution in a solvent, onlj. those of Gildart :mi JVright ( 5 ) ,Benning, Ebert, and Irwin ( I ) , and Kiveiison (61 are tight at pressures appreciably above atmospheric. Of thest,. t8hc1:ittc.r two iised quartz windows arid sealiiig mpthods not ap-
strurtion of tight cells are descrihetl, ab well as the technique used in assembling arid testing them. The cells h a \ e rock salt windows and Mere sealed w i t h .irnalgam. The) ha\e been used at pressureu p t o 3.5 pounds per square inch gage for recording the spectra, from 1 to 1.5 microns, of the hutenes and 1,3-hirtaclieiie. The same cell serkes also to prevent loss o f tolatile solvent- from iolutions during the rec*ordingof spectra.
l)lir:il)l(~ t o rock salt. :tnd tht. first did iiot liiive connections s u ~-t :it)lr. for tilling the cell and confining tht. sample under prwsurr. This piper describes the design, ronstruction, and technique n i :Lbwnit)ling cells using rock salt wintion-s, h s e d on the principlw 1)y (iildart and 11-right ( 5 ) . The cells have thickirrsstss ~lwrrit)c~:l I J t~" i n i 0.427 to 0.018 nmi. a n d have twen used a t pressures 111) t o :%.?Ipouricls per square inch (2.5 kg. J F I ' s ~c.m . ) gage to rerciril I h r d q)ec.tix of the liquid butenes and I,:3-hiitatli~ne from 1 t o 1.5 iniri,ons :it room tmiperatiiw ( 2 ) . DESIGN A Y I ) COXSTHUCTION
The c . t s l l Ir:tnies ivew desigiied to fit t h e cell holder of l'wkiiiLlmer bpectronieters. The)- \\-tire made of steel and somt~\vliat heavier than nolmnl t o nlinimizc~ possiblc deformation r u i i l c ~ r . pressurta.
Prrliniitiaq- trials indicated that leaks wuld be made Irslikely by inrreasing the length of thr potential leak path. To provide for this and also to minimize stresses in the wintlo\w. the filling holes were drilled thrnugh one of the windows nornix1 to its suiTace. Because it was drsiFed not to intercept any of t h ( , iwnc of radiation whirh rould pass the usual 21-mm. dianir~tc~i
