ULTRAVIOLET SPECTROPHOTOMETRY CONSIDERED FROM AN INSTRUMENTAL STANDPOINT
Document Type:
Collection:
Document Number (FOIA) /ESDN (CREST):
CIA-RDP78-04861A000400030023-1
Release Decision:
RIPPUB
Original Classification:
K
Document Page Count:
29
Document Creation Date:
December 20, 2016
Document Release Date:
June 6, 2006
Sequence Number:
23
Case Number:
Publication Date:
September 1, 1999
Content Type:
REPORT
File:
Attachment | Size |
---|---|
CIA-RDP78-04861A000400030023-1.pdf | 1.92 MB |
Body:
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
I
van Dranen -
Chemisch Weekblad 49 (1953) 482 - 491
(From Dutch)
II CRITICAL INVESTIGATION OF SIMPLE
SPECTROPHOTOMETERS
R. Schmidt
Chemisch Weekblad 49'(1953) 492 - 493
(From Dutch)
III. THE BEHAVIOUR OF THE SPEKKER
ABSORPTIOMETER
R. Schmidt
ULTRAVIOLET SPECTROPHOTOMETRY CONSIDERED
FROM AN INSTRUMENTAL STANDPOINT
IV.
SIMPLE SPECTROPHOTOMETERS' OF THE
(From Dutch)
PRECISION EXTINCTION MEASUREMENTS WITH
DEFLECTION TYPE
H. L. Zwiers
Chemisch Weekblad 49 (1953) 496.- 499
(From Dutch)
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
. Chemisch Weekblad 49 (1953) 494 496
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
25X1
11LTRAV I OLET SPECTROPHOTOMETRY CONS I DEREb. FROM AN
INSTRUMENTAL STANDPOINT
J. van DRANEN
Chemi sch Weekb l ad _Q (1953) 482-491
( From Dutch)
SUMMARY
This article deals with ultraviolet apectrophotome try from an instrumental standpoint,
and discusses in particular the Beckman and Unicam photo-electric quartz spectrophotometers.
The first section contains a short description of the construction and operation of these
instruments. Thereafter, the accuracy of the extinction measurements which may be obtained
with these instruments, is dealt with in detail.
INTR0DJCTI0N
In this article, dealing with ultraviolet spectrophotometry, from an instrumental
standpoint, we should like to discuss more particularly the Beckman and Unicam photo-electric
quartz spectrophotometers.
Firstly, however, the advantages and disadvantages of the different methods of
measurement will be briefly discussed. The radiometric methods, by which the intensity of
the radiation is determined by bolometers or thermopiles, are not of importance for ultra-
violet photometry, because the useful wavelength range falls within the infra-red region.
Photographic and photo-electric methods, however, are all used for the measurement of visible
and ultraviolet radiation. The properties of these methods are mentioned in Table 1 (1)*.
Although, from this table, it. appears that no one of. the methods is superior, the experience
of the last decade has shown that moderp.photo-electric instruments possess very considerable
advantages over photographic for the measurement of extinction curves in the ultraviolet
region. This Improvement in technique is stressed by examinin LP the difficulties discussed
in the dissertation by Dr. C.Fb F. Spiers (Amsterdam 136 ) in obtaining accurate measurement of
the absorption spectra of pyridine and its homologues, compared with the ease with which this can be
achieved, at present, with for Instance a Beckman Instrument.
MEASUREMENT OF INTENSITIES
METHOD
WAVELENGSI
RANGE
PANATOMIC
PROPERTY
CUMULATIVE
PROPERTY
LINEARITY
Visual
4000 - 7000 X
limited
none
very.poor
Photoelectric
to 30,0 CO X
none
fair
good
Photographic
to 12,00Q X
excellent
good
poor
The disadvantages of the photographic method compared with the photo-electric method
1. Difficulty In adjusting the spectrographic apparatus.
2. Establishing the extinction curve from photographic records:
(a) Optically inaccurate
(b) Photometric - slow.
3.. Expensive, especially if the extinction Is to be measured at single
wavelengths only.
4. Operators have to be highly trained.
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
2-
I. THE MODERN QUARTZ SPECTROPHOTOMETER
H.H. Cary and Arnold 0. Beckman, the designers of the "Beckmann, in an article in the
'Journal of the optical Society of America (2) ,have. thoroughly explained the motives and experiments
which led to the construction of their instrument. Their aim was to develop'an instrument,
which would satisfy the conditions of economy, accuracy and simplicity of handling. A compromise,
as favourable as possible, was sought between these somewhat contradictory requirements. The
photoelectric system was chosen; the following fundamental conditions must be satisfied if
accuracy is required:-
These requirements determine the selection of the four principal components of the
instrument:-
(a) Source of light;
(b) Optical system;
(c) The photo cells;
and
(d) lire electrical system. (See Figures i and 2).
(a) LIGHT SOURCE
A 25 watt, 6 volt tungsten lamp is used as a light source for the region between 10,000 X
and 3,20.0 $, the current being supplied by an accumulator. Below 3,200 R, the continuous spectrum
of a hydrogen discharge tube is used. In this case an electronic voltage stabiliser is used
(stabilisation 1 0.290 to supply the required 0.4 amp. From A the light falls on the focussing
mirror B (aluminized, focal length = 2.85 cm) and is reflected from mirror'C on, to the slit A
The slits (curved, bilateral, 13 mm high) can be accurately adjusted between 2.00 and 0.01 mm:
the entry slit is below the exit slit. Calculation and experiment have demonstrated that the
light on the entry slit has its origin in a small section 0.6 mm wide and 3.1 mm high of the
source of light, thereby satisfying the requirements for high intensity and small area. From
this It follows that with the aid of B and C a seemingly small source of light Is in a position
to supply all the light required by the monochromator.
(b) OPTICAL SYSTEM
in the construction of their monochromator (that part of the instrument D, B, F., which
receives light of a very definite wavelength from the original light source A) Cary End Beckman
were confronted with the difficult choice of using a grating or a quartz prism. The advantage
of a grating is in general, the greater dispersion, which means that the slits and the wavelength
scale neednot be made so accurately. Furthermore the price of a replica grating i,, lower than
that of a quartz prism. However, the principal reason for deciding to use a prism, lay in the
fact that a grating scatters much more light.
I t Is. most important that the monochromatic light should be free from stray light:
although, in a well constructed monochromator, the stray light, which originates from undesired
reflections, can be made negligibly small, the surface of a grating Is so much less optically
perfect than a prism that stray light, not originating from reflection Is much increased.
Extensive tests undertaken by these investigators with eight different gratings and two different
prisms in different installations, showed on the whole that scatter with a grating is ten times
greater than with a prism. For the optical system, the so-called Littrow installation was
finally chosen with mirror collimation, which requires a minimum in optical and mechanical
details: The collimator L+' (aluminized, focal length = 50 cm) in whose focus the slit D is
situated, reflects rays deviating 50 from the principal axis, whereby both astignatisn and
spherical aberration occur. The spectral width or the band of light emerging at D is thereby
Increased by an amount which would be obtained by increasing the width of the slit by 0.02 mm.
We shall return later to a simple formula for calculating the spectral band width for a given
slit width. From E the light falls on the prism F. This reflecting prism (refracting angle
30?, the aluminium layer on the back nas a diameter of 5'cm) is made from very good quality
crystal quartz. Compared with a monochromator with a 80? prism, only half the quantity of
quartz Is necessary. Furthermore, the prism need not consist of two halves of opposite
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
rotatory power in order to avoid doubling of the Image, due to the double refraction
of the quartz. The prism can be rotated by means of a long spindle, the other end at
which is connected to the wavelength scale. This scale consists of a quadruple spiral
one metre In length reading from 2,000 X. to 2 p. The light, which comes via F and .9 from
the exit slit at D is not blocked by the mirror C which is situated below it, but passes
directly 'through the cell G containing the absorbent substance. The. intensity of the
unabsorbed light is then measured by the photocell H. The wavelength of the light which
traverses the path D, G, H is indicated on the wavelength scale. By rotating the prism
it is possible to adjust to any desired.. wavelength.
The use of a mirror as collimator obviates the need for an expensive achromatic lens,
while focussing also becomes much simpler.
Two different photocells are used in the instrument; a red, sensitive cell and an
ultraviolet sensitive cell. The first is a caesium oxide cell, which must be used above
6,250 Though this cell is useful. down to 4,000 R, below 6,000 1, stray light, to which
this cell Is exceptionally sensitive, begins to play a. greater role, so that it is better
to use the ultraviolet sensitive caesium antimony cell.
(d) ELECTRIC SYSTExf
In an article of restricted length, it is of course not possible to go fully into
the question of the rather complicated circuits. We reproducer therefore, a simplified
circuit diagram given by Caster (4) with which to explain. the measurement of the photo-
current (Figure 2).
The photocurrent is measured by compensating the voltage drop across the resistance
R of 2 x 109 ohm with a potentiometer. This operation is controlled by a galvanometer
through an, amplifier. The amplifier has an amplification factor of 5 x 107, by means of
which full deflections of the galvanometer are possible. with a light influx of 2.5 x 10 8 watt,
corresponding to a photocurrent of 1013 amp. The resistance of the photocell varies according
to the intensity of the focussed light from 10 to approximately 1013 ohm. The photocell
compartment contains a dessicant to prevent leakage with these high resistances. By means of
P it is possible to switch from one photocell to the other.
A point between the photocell and the galvanometer Is maintained at + 20 volt. The
potential thus drops from + 20 to + 2 volt across the galvanometer and valve T. The current
passing through the galvanometer thus depends on the grid voltage of this valve. In the
other branch of the circuit, the same drop occurs across the photocell and resistance R.
Now If light fails on the photocell, the internal resistance of the cell is lowered, and thus
the grid voltage increases. In order to compensate for this change, the total potential
difference of the system can be adjusted by means of the potentiometer knob which has a scale
readable in percentage transmission (and optical density or extinction D, D = log T). The
variation in the photocurrent'is thus directly balanced by this potentiometer. The dark
current is the small residual current, which also occurs when the photocell receives no light.
In order to compensate for this, the cell and the potentiometer are disconnected;. after that
the auxiliary potentiometer is adjusted by means of the dark current control knob, so that
the galvanometer is at zero. .
The third potentiometer, the."sensivity potentiometer" Is used for compensating the
absorption of the solvent. By taking this as a zero level, this part of the:absorp tion for
Which the dissolved substance is responsible can be measured directly., The transmission
potentiometer can be disconnected by setting the switch at the "check" position. The solvent
is In the path of the light. By means of the sensitivity knob the galvanometer-is set at'
zero (to save time, the transmission potentiometer Is replaced by an equal. resistance, so
that the transmission need not first be set separately at 10050. Should the transmission
be less than 11%, it is better to set the switch at 0.1. The scale values will become ten
times as small, thereby increasing the accuracy of reading.
For biochemical purposes, it is important to be able to work with very small quantities
of liquid.. Lowry and Bessey (5) could indicate in. 0. 01-0.06 em3 serum, the .vi tamins A and C
in fractions of y's. .
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
OTHER APPLICATIO98
The Beckman spectrophotometer can also be made suitable with relatively few alterations,
for a number, of other purposes, besides the measuring of the extinctions of solutions, These
will be briefly described.
Greater quantities of liquid and gases (optical length of path of the gas cell. 10 cm) can
be measured in containers obtainable separately. The only alteration is, of course, that another
cell must be used for these containers.
In addition special apparatus can be supplied for fluorescent and flame photometry. The
difficulty which presents itself with this fluorescent apparatus is that while avoiding the
monochromator the light falls directly on the photocells. Thus the total intensity of the
fluorescent light is calculated, and because of this it is not possible to analyse an unknown
combination or to define a mixture of fluorescent substances. Dr. A.M.W. Indemans (1;8) drew
our attention to a method, by which the monochromator can be used for the determination of the
complete fluorescent spectrum. The fluorescent light, generated by a mercury lamp, passes via
a converging lens directly to the mirror C, Figure 1. Therefore it is not necessary to introduce
still further alterations (6). In -the case of very weak fluorescence, the photocells can be
substituted by photomultipliers.
In order to make the instrument suitable for flame photometry, the lampholder must be
substituted by a gas-oxygen flame. In this case the liquids are disposed in the customary way.
THE UNICAM PHOTOELECTRIC SPECTROPHOTOMETER
This instrument presents only a few minor items, which differ from the Beckman. It may
be said, In general, that the Unicam is designed to be somewhat more robust than the Beckman.
The control panel slants back for better visibility unlike the Beckman, where one must look up
at it. At present, In order to avoid this last disadvantage, we have fixed the light in the
Beckman in a position slanting towards the observer. In the Unicem, the lampholder contains the
tungsten as well as the hydrogen lamp. The condensor mirror B, Figure 1, can be focussed on the
required lamp by means of a small lever.
The Hilger Uvispek (ultraviolet and visible photoelectric spectrophotometer) and the
Zeiss Opton Spectral Photometer (with monochromator and an electron multiplier in place of
photocells) belong to the same group as the Beckman and the Unicam with regard to cost; and
capability.
In contrast to the Beckman and the Unicam, separate accumulators for these two last
mentioned instruments are not required, supply of an alternating current source being sufficient.
Furthermore, the differences in light source, optical system and suchlike, are not great. The
Uvispek also possesses the same compact construction as the Beckman. In the Zeiss, the separate
components are arranged on an optical bench. The wavelength scale, however, is read in an
entirely different way to that of the other instruments. In this case an image of the entrance
slit is shown on the scale. The centre of the image Indicates the wavelength, the width shows
the spectral bandwidth.
In order to give an opinion, as to which of the four instruments is the best, a very
extensive Investigation would be necessary. For this investigation, one ought to have at one's
disposal at least 10 specimens of each instrument. A priori, one may expect that these instruments,
obtainable practically at the same price and supplied by very good firms, would not show any great
differences In quality. Each apparatus can be delivered with quartz optics (range atproximately
2, 100 - approximately 101000 54 see also discussion) as well as with glass optics (range
approximately 3,800 $ .. approximately 11,000 $) and provided with the necessary accese,ories for
fluorescence, reflection measurements and flame photometry.
"AUTCHATIC" SPECTROPHOTOMETERS
In the instruments previously discussed the extinction curve must be recorded point by
point. By making use of modem electronic resources it is now possible to record the curve
automatically. In the Cary Recording Spectrophotometer (7), the light of the hydrogen or
tungsten lamp is split into two beams, of which one is the 0100% transmission" beam the
remaining beam falling on the material under measurement. The Intensity of both beans is
measured by separate photomultipliers. After amplification, the ratio of the two photoelectric
currents is registered directly on an "electronic recorder". The double monochromator with
two 30? quartz Littrow prisms gives the instrument a high resolving capacity, so that for
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
instance the rotation structure can be seen very well in the ultraviolet absorption spectrum.
The spectral bandwidth in the ultraviolet region, except at the boundaries, is approximately
i/A. Kaye and others (8) have also succeeded in making the ordinary Beckman apparatus
automatic, at the same time increasing the wavelength range considerably (to 2,700 m /4.
In this way the red sensitive cell is replaced by a lead sulDhide cell and the cell, sensitive
to ultraviolet, by a multiplier; the beam of light Is modulated in order to increase the
amplification. The resolution of the instrument thus modified Is ten times higher than that
of the original Instrument, so that one Is really able to get the mercury line split up at
2,541 Is, although the distance between the components is only 1.7 R. The saving of time is
considerable; in ten minutes, the whole infra-red spectrum of 700 m /l to 2,700 m A can be
recorded. These three advantages: greater wavelength range, saving of time and greater
resolution in the ultraviolet region, make this automatic operation very attractive, especially
as an existing instrument can be modified. Important details on the subject of the
instrumentation of spectrophotometers can be found In recent numbers of the Review of
Scientific Instruments and the Journal of the Optical ?Soc.ie.ty of America. We also refer to
the well known work compiled by Mellon: Analytical Absorption Spectroscopy (9)?
THE STRUCTURE OF THE EXTINCTION CURVE
Here we would like to mention briefly a few of the all. important facts from theory.
Naturally the most important application of the ultraviolet absorption spectra is the
identification and qualitative (quantitative) analysis of chemical compounds. The more
simple Infra-red or Raman spectra lend themselves better to the examination of the structures
of molecules.
The energy of a molecule, may be regarded as consisting of electron energy, vibration
energy and rotation energy. A change in the electron energy takes place due to absorption
of light by the molecule. As a result, the molecule moves from thebasic?position to a new
one. In a physical sense this means that a particular electron makes a Jump towards a band
with higher energy (this is entirely analogous to the absorption of light, as described with
the aid of the atom model of Bohr). Electronic bands are produced when there is a,?change
from one electronic level to another. The separate bands are caused by a jump in the
vibration energy which is superimposed on an electron dump.. The vibration energy is the.
energy of vibrations which the atoms execute with respect to one another.. With some reserve,
it can be said that ,, these vibration bands are often the peaks in the extinction curve, obtained
with a spectrophotometer., The structure of these bands is in turn caused by dumps
in the rotation energy..,: Experimentally, this rotation structure can be..indicated only for
the absorption spectrum of a gas. In the fluid and solid states, the rotation, is hindered
or even completely prevented by the molecular interaction force, in such a way, that the.
rotation energy conditions are spread out to a continuum. We can therefore state that the
energy of the electron jump determines the position of the centre of gravity of the band
system. The width of this depends on the vibration energies. .
Before formulating this somewhat more quantitatively, it must, at first, be mentioned
that In spectroscopy, it Is customary to express the energy of a vibration by the. Have number.
The wave number Is the frequency v divided by the velocity of light C . If the wave number
of a vibration Is v, then the corresponding energy, according to Planck's ratio Is.,
h . v = h . Y . C per grammolecule N . h . C , v or calculated .;.844 v cal/mol. The energy
E of an absorption at 5,000 I is thus: X = 5,000 R = 5 x iq 5 cm; v = CA or V = i/k = 20,000 cnf'
or E = 2.844 x 20,000 =57 Kcal/acl. (i Kcal/mol = 362 cm 1).
The construction of a band system is given below. The indicated boundaries are very
rough and are more applicable to the organic molecules, which absorb ultraviolet in this region.
As an aid to visualisation it can be said that the distances between.the bands consist mostly
of a few tens to hundreds of .';s; .the rotation ,fine structure contains. components of one
up to at the most a few s.
BAND SYSTEM'S
Position of band system determined by electron jump ti 3000 er i (3333 R)
% 300-3000 cm 1
in gaseous state: instrument with wide dispersion, separate
Fine structure lines at small distances (order of magnitude i R) ti 3-30 cm-i
rotation bands
in liquid state: due to disturbance the separate points are
linked together into a continuous line (extinction curve)
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
Let us discuss, for example, the well known absorption band system of benzene at 2*7600 L
Benzene consisting of 12 atoms, possesses 12 x 3 = 36 degrees of freedom. In order to discover the
number of vibrations of this molecule, one must subtract from this quantity the degrees of freedom
of the translation (3) and of the rotation (3,for a linear molecule 2). Therefore benzene possesses
3J vibrations. Due to the great symmetry of the molecule, there are only 20 different frequencies,
and of these, the so-called double damped vibrations coincide in pairs. However, in th:ls example,
we are dealing with only two of the 20 frequencies. In theory, the electron jump which occurs as
a result of adsorption, is a forbidden transition, which is in accord with the six-sided symmetry
of benzene. The pure electron jump at 2620 1 therefore rarely appears In the extinction curve,
but the band g (2590 X) will, because besides the absorption of the electron system, the g
vibration also absorbs a quantum. This vibration (R indicates that the vibration is double
damped) therefore passes from its basic state to Its first adopted state. Expressed in cm 1
the difference, between the rarely occurring 0 band and the g band of frequency of the E vibration
amounts to 606 cm-1. The other bands now come into existence, because the A vibration absorbs
one (R + A), two (g + 2 A) or three (E + 3 A) Quanta. Naturally the E vibration must always be
associated with one quantum. It is interesting that in the adopted state, the frequencies of A
and g are smaller. This is due to the fact that in the adopted state, the cohesion between the
atoms has become weaker, because the loss of the electron has decreased the cohesion in the new
state.
In the majority of cases, the absorbed light energy is transformed into heat. At first
the excited molecule transforms its surplus electron energy into vibration energy, because the
electron system is energetically linked with the vibration. By means of collisions between the
molecules of the solvent, this vibration energy, in the case under examination, becomes distributed
throughout the total fluid. Therefore, it appears that the chance of fluorescent radiation is
determined by the intensity of association between the electron system and the vibrations and the
number of collisions. Therefore, it is necessary for the occurrence of fluorescence, that the
absorption electron system is reasonably protected against interior and exterior disturbance, so
that the actuated molecule may remain for at least 16-8 sec. In an adopted state. in addition to
the wavelength of the absorption, the intensity, i.e. the value of the extinction coefficient, is
of great importance. Theoretically one can deduce intricate formulas for this coefficient.
The theoretical values, however, may vary by a factor of 2 or 3, from the experimental results which
even under unfavourable circumstances, are accurate to within 20%. For a book containing many
useful particulars on the subject of optics, emission and absorption of light, we would like to
refer to E.J. Bowen "The chemical aspects of light" (10). This work, written on an academic level,
does not demand any special pre-knowl dge of the subject. Friedel and Orchin have published a
collection of nearly 800 ultraviolet spectra of aromatic compounds (1i). This collection If
added to in future, as is the intention, will be of very great Importance to users of spectro-
photometers. The introduction contains a useful review of instruments and the use of ultraviolet
spectra for analytical purposes.
I.I. THE ACCURACY OF THE EXTINCTION MEASUREMENTS
Having discussed the instruments and a few theoretical aspects of the ultraviolet spectra,
we will now deal with the question of accuracy. Publications on this subject have not been
lacking during the last years, so now that the first enthusiasm for these spectrophotometric
instruments Is a thing of the past, we have made a more critical examination. It is a fact that
some investigators have a very high opinion of the accuracy reached. The literature contains
values such as 14704 and 4954 for the extinction coefficient. It is obvious this magnitude
cannot be determined so exactly, but, before going any further into this, we would like first of
all to mention the most important results of very fundamental research, which Gibson and. Balcom
have undertaken, for the National Bureau of Standards, regarding the properties of the Eeckman
spectrophotometer (12). This article is very highly recommended to anyone interested In
spectropho tome try.
MULTIPLE REFLECTIONS
These are due to light which is reflected back and forth between the cell and the quartz
slit. If one measures a solution against the solvent (transmission set at itb%) then this
error is generally negligibly small. However, if the transmission of a glassplate is determined
from the unweakened been, the transmission may be up to 1% too high.
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
STRAY LI($i2
Gibson and Balcom mention that one can observe the stray light directly by visual
means, i f with the monochromator set above 7, 0O R or below 4. 500. one looks directly
into the exit slit. According to these Investigators, practically all energy above
1,200 m fit, as measured by the photocells, can be accounted for by stray light. Therefore
the instrument cannot be used for wavelengths between 1, 200 m /.c and 2, 000 m /t. They
consider it necessary to use a Corning No. 9863 red purple filter, when using the tungstan
lamp below 4,000 1. This filter is placed between the slits and the cell. For measurements
between 4,OCO R and 5,000 R, it may be sometimes useful to fit a blue filter in the (third)
open space of the slide.
If the dissolved matter: shows any fluorescence, then naturally a small part of this
fluorescent light also falls on the photocell. In view of the fact that the intensity of
the fluorescent light is low this error may nearly always be neglected. In special cases
suitable filters can be used. From the small quartz plate, which covers the entrance slit,
the optics run parallel with the side-panels, and this may produce a polarisation effect.
In order to prevent errors in the measurement of polarising materials, one should therefore
remove the small plate, or else carry out the measurement with a different orientation.
THE WAVELENGTH SCALE
With the aid of the screw, which is situated on the left side panel and is connected
to the collimator, the wavelength scale is calibrated to the green mercury line (5461
with subsequent checking by the location of the spectral lines of hydrogen, helium and neon
sources on the scale. . up to 6,000 R, the error appeared to be not greater than 5 R: while
throughout the useful range the error nowhere greatly exceeded 10 R. Therefore if calibration
can be carried out by means of the green.mercury line, this error can be ignored in most
types of measurement. Only for very accurate measurements must the scale be calibrated at
other points as well. It then becomes necessary to fix to the wavelength scale a small
plate with an Indicator strip, in order to prevent parallax. Such a device may be useful
for everyday measurements also.
SPECTRAL BANDWIDTH
It may be said that the slit Is the most Important component of an optical instrument,
as the slit is finally the indirect source of light, from which the spectrum is obtained.
Every irregularity In the slit expresses Itself In the image obtained, so that very high
demands are required from 'the finish of the edges. It is very Important that monochromatic
light is used for, the correct determination of the extinction curve of a solution, since
Beer's Law only applies to monochromatic light. The most important maguitude for the
monochromator used in spectrophotometers,. is the spectral bandwidth, i.e. the narrow band
of frequencies, which for a predetermined setting of the prism leaves the exit slit and
falls on the cell.. If the extinction coefficient 6 is measured at a strict maximum or
minimum, .then, of course, the value of 6 is very dependent on the spectral bandwidth.
We give, by way of example, a number of observations of Ho Bless and others (13) for benzene
in iso-octane for the wavelengths 2,540 R (strict maximum) and 2, 626 R (strict minimum).
It will be noted from this small list, that, as the spectral bandwidth becomes greater, the
values Emax. and Emin. approach one another, because then in both positions, the monochromator
transmits the same spectral region.
EXTINCTION COEFFICIENT OF ; NZENE IN T SO-OCTANE
SPECTRAL BANK IDTH
2, 540 R (max)
2. 52 5 R (min)
5.6 R
212
45.3
8.8
204
45.3
16.0
190
45.5
32
154
50.0
80
104
62.5
120
89
69.0
160
77
72,2
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
Following up the work of Hardy and Young (14), Eberhardt (15) has derived correction
formulas, making use of this dependence of e on the spectral bandwidth, in order to define the
true value of E. However, a number of assumptions must be made, concerning the behaviour of
the extinction curve in the neighbourhood, of the maximum (for instance acute or parabolic, angle).
Much work can be saved, by the application of this formula. In the case of the absorption spectrum
of K2C r2O7 with flat maxima, the Influence of the spectral bandwidth is much smaller than, to
give an extreme example, rarely encountered absorption spectra with pronounced peaks. It follows
that it is very important to state the spectral bandwidth when giving absorption spectra data in
order to facilitate reproduction of the results. For the Beckman and the Unicam a simple formula
is available for the I calculation of the spectral bandwidth W, for a given slit aperture X (In mm's),
namely, WS = 2 W = 2 WD (X + 0.02). WD is the nominal bandwidth per 1 mm.? slit. aperture. .This
factor is given by the designers in the form of a curve as a function of the wavelength (the nominal
bandwidth is the wavelength range, for which the intensity of the light has dropped to one half of
the maximum value).
X (mm)
&D (R)
WS (R)
H2 191 p 2200 R
2.00
14.6
585
2500
0.80
23
285,
3000
0.40
41
34
3500
0.30
68
42
T lamp 3.500
0.10
68
18
4500
0.04
15
18
6000
0.04
34
40
8000
0.05
66
92
1000 M )U
0.50
88
90 m
As has already been explained, the term 0.02 WD is due to optical aberrations. The fact
that the same formula can be applied to both the Beckman and the Unicam, Is further proof of the
similarity of these instruments. Information is given concerning the spectral bandwidths for a
number of usable slit widths. For small wavelengths, it is possible to work with narrower slits,
but the quantity of light which falls on the photocells then becomes too small and the transmission
cannot be measured with sufficient accuracy.
THE MEASUREMENT OF EXTINCTION (OPTICAL DENSITY)
In order to know to what degree of accuracy the extinction can be measured It is necessary
to examine the method of measurement. In order to obtain the extinction of a solution at a
particular wavelength, the galvanometer must be set 3 times at zero.
the'"dark current" must be compensated.
(2) the transmission of the solvent Is set at 100% by adjusting the slit and the sensitivity
knob, or a fixed slit is used and regulation obtained entirely by means of the
sensitivity knob, or the sensitivity knob is turned three times to the right and
the indicator set at zero by adjusting the slit. The sensitivity control can then
be re-set with precision.
(3) the switch is then set to 1.0 and the solvent Is brought into focus by means of the
transmission control. The Indicator is then reset at zoro. A small error is
involved in each of these three steps. In addition there are fluctuations of the
light source, which,anount to 0.1% to 29, according to Edisbury (16), dark current
and sensitivity errors amount to 0.1%. A more detailed observation of the
fluctuations in the reading of the optical density potentiometer is given by
Edisbury.
The optical density potentiometer contains approximately one thousand turns giving an error of
approximately 0.1%, as the contact arm must jump from turn to turn and therefore there is always
a difference of a whole turn. For transmissions in the useful region, the following errors
are calculated: A T = 0.001 is constant. The error is defined as half the difference between
two successive values.
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
TRANSMISSION T
EXTINCTION D
(D = -log T)
ERROR IN D
(? L III) x 1009a
0.630
0.2007
0.631
0.2000
0.18%
0.370
0.4318
0.571
0.4306
0.14%
0.199
0.7011
0.200
0.6990
0.15%
As Is known, a small variation In the extinction coefficient gives the greatest
variation in T at 37% transmission. The minimum uncertainty in D is therefore approximately
0.14%, From the fact that the needle of the galvanometer moves continuously, instead of
swinging violently backwards and forwards as the control Is adjusted it follows that this
uncertainty is smaller than the limit of error In reading the galvanometer. According to
Edisbury, it may be concluded, from analogous considerations, that there is an error of the
same order of magnitude with regard to the "dark current" and the "sensitivity". Al thou gh
this reasoning does not appear to us entirely conclusive, it is, however, not easy to find
arguments against it. Naturally much depends on the accuracy, with which the galvanometer
Is. se,t.up. The pointer should always be adjusted from the same direction with regard to
the zero point and It is more important to obtain the same reading for all three calibrations
than to obtain an absolute setting of zero. Edisbury's observations lead to a total error.
of approximately one half percent in the "optical density" and the extinction coefficient,
an error, which cannot be avoided with the present design of instrument. It is interesting
to see what results are obtained in practice. Ewing and Parsons (17) have compared
observations on 10 Beckman instruments. The extinction coefficient of the acid K phtalate,
which can easily be obtained in a pure state, was measured at 281 m ? (maximum) and 264 m fL
(minimum). For log a (max.) the values varied from 6.45 to 6.17 (average value, 36
measurement with 10 instruments 6.314);; for log E (min.) from 4.27 to 4.06 (average value
4.175). The values for single instruments amounted to: 6.39, 8.38, 6.39 and 6.40 for
the "best" instruments; 6.26 and 8.31 for "average" instruments; 6.45, 6.41, 8.37 and 6.37
for the"worst"Instruments. These observations show that the agreement for readings from
a single instrument is much better than between different instruments (namely 1 1% against
t 4%). This is due to the fact that it is not entirely possible to avoid small variations
in the manufacture of such a sensitive Instrument as a spectrophotometer. Ewing and Parsons
suggest therefore that each apparatus should be compared with a statistical average obtained
from a large number of instruments and a correction table proviided for each Instrument.
Now if consideration is given to the diversity of the sources of error, it is more than
probable that the correction terms will be based on a degree of accuracy, which does not
in fact exist. The British,."Photoelectric Spectrometry Group" carried out in 1950'a
comparative test, In which the extinction of solutions of K 2 C r 2 0 7 in 0.01 N ll 4 was
measured with 63 instruments (35 Beckman, 15 Unicams and 13 Hilger Uvispeks).
The variation coefficient, that is, the value of the standard deviation expressed
In percentages of the accepted value, for the measurements of 8 extinctions (2 different
solutions with `4 wavelengths) amounted to 1.85.%. It was once thought that photoelectric
instruments tend to register low extinctions somewhat too high and the high extinctions
somewhat too low. This test shows that this is not the case. , Insofar as can be judged
.from, the results of the 3 photographic instruments examined, it appears that these instruments
did in fact suffer-from this fault since they measured the extinctions ,.few percent too low.
From statistics, It appears that a photoelectric reading is three times as accurate as a
photographic reading. The following table Indicates by how much actual measurements may
vary from the accepted value. The extinction Is expressed as 9(1% 1 cm)that is the extinction
obtained from a 1% solution with an optical pa th of. 1 cm, a method often followed for natural
products and polymer compounds, for which the molecular weight is not known. The accepted
value for G (1% i cm) is compared with the average of the 2 highest and of the 2 lowest results
for the 4 different wavelengths.
N.T. Gridgeman, has published the results for these 83 instruments statistically and
suggested the use of K20"207 as a standard (3 times crystallised from double distilled water
and dried at 40C?C; to this solution is added 0.25 ml. concentrated H2& 4 per liter). The
extinction coefficient for X20'207 is not too high or. too low. There are four useful
wavelengths (compare the small list) which cover the whole ultraviolet.. For these wavelengths
the extinction curve possesses apparently flat minima and maxima, so that the instrument can
be focussed satisfactorily.
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
E (1%, 1 cm)
K2C''207
WAVELENGTH
ACCT PIED
AVERAGE OF
AVERAGE OF
VALUE
TWO HIGHEST
TWO LOWEST
235D R (min)
124.8
131.6
114.9
2 570 (max)
145.2
150.0
138,0
3130 (min)
48.7
50.2
44.7
3500 (max)
108.7
108.3
105.2
K.F. Westerling and F. Hartog (Laboratory for General and Inorganic Chemistry and Laboratory
for Analytical Chemistry, University of Amsterdam) have measured the extinction of solutions of
K2C r207, CO Sr)4 and CU Sb4 with the aid of a Beckman and a Unicam. The provisional results of
this investigation are in agreement with the values ,riven by other workers. For seven different
solutions of X2Cr O,, COS()4 and CUSC)4 with extinctionsbetween approximately 0.1 and approximately
0.9 (21 solutions In all) the deviation in the values measured by both instruments, was on an
average a good 2
PRECISION COLORIMETRY "
The photometric analysis methods may be more sensitive and quicker than the classic
gravimetric methods, the disadvantage being that they are more inaccurate. Bastian et al (18)
Ayres (21) and Hiskey et al (22) have developed differential or precision method of colorimetry,
which reduces this disadvantage considerably. In principle precision colorimetry is based on the
fact that the extinction of an unknown solution is compared with that of a known solution, for
which the extinction coefficient differs but little in value. It will be shown that in this case
the accuracy of the calculation will be much greater. The extinction E (optical density) of a
solution = E C 1; an Identical Increase A C in C always gives the same augmentation in the value
for the extinction regardless of the value of the concentration C if the validity of Beer's law
is assumed. By using high concentrations, the error A C/C could be made smaller If it were not
for the fact that these high values for the extinctions can only be determined with much less
accuracy than lower values (24). According to the above mentioned authors, it is better not to
set the transmission- for the pure solvent at 100%, but, for a solution with a concentration
CO to
use a neighbouring value CC. if the difference - Co can be measured accurately at for examsle
0.5% then the accuracy obtained is equal to CX - C0/Cx x 0, 5%. The following table gives a few
results obtained.
COMPARISON SOLUTION
UNKNOWN SOLUTION
LE TERMINED WITH
(TRANSMISSION 100%)
PRECISION
COLORIMETRY
1.5000 g 01/100 ml
1.6471 g
1.6467 g
0. 10000 N W7
0.0989 3 N
0.09897 N
0. 10000 N 4
0.09891 N
0.09878 N
However,. the question is whether this precision method can be used universally. Naturally
it Is always possible to find concentrations in the neighbourhood of the unknown and interpolate.
The best way is to use a number of concentrations, both greater end smaller than the unknown.
This gives complete freedom from deviations of Beer's law and similar laws. The weakest solutions
obtained of a small concentration range are set at transmission 100% (E = 0). With high
concentrations large slit-widths must be used in order to enable sufficient light to fall on the
photocells. It is not only that the stray light is of more influence, but, with a comparison
solution, there are two possibilities for deviations of the Beer's law, namely (1) the high
concentration (Beer's law is in principle a boundary law applicable to monochromatic light and
an infinitely weak solution) and (2) when the width of the slit is great. the spectral bandwidth becomes
so great, that the light, can no longer be considered monochromatic. A narrower focus is then
desirable in order to reduce the width of the slit. It may be said that precision colorimetry is
an interesting development, but that all methods must meet the requirements of practical use.
Differences In cells can also play an important role and certain precautions must be taken.
Furthermore, it is apparent that reasonable reproducibility does not demand great accuracy.
Before concluding, mention will be made of some results of an investigation by Caster (41 into
the various factors which are responsible for the variations in the Beckman spectrophotometer.
This examination is important, because It draws attention to factors, which are generally ignored.
Failure to set the "dark current" exactly at zero, led to an error of approximately 0.4% per
scale division. Therefore In order to obtain reliable measurements the "dark current" must not
only be correctly compensated, but must also remain constant. To obtain this stable "dark current"
the accumulators must'be in good condition and the various contacts must be'clean, In view of
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
the fact that both photocells are usable at 6, 100 R, the differences in these were also
investigated. The red sensitive photocell gave at this wavelength a 3.7% greater extinction
for a 3% CU'04 solution, than the cell sensitive to ultraviolet. in the long run, the red
cell was also less stable than the ultraviolet cell. After a time lapse of eight months,
the differences were respectively 2% and 0.6%. Thus this factor can have a serious effect
on the accuracy of absolute measurements.
(1) it is very desirable that the spectrophotometric instrument should be checked at
regular intervals. This can be done by means of a standard solution of known
extinction, but it is better to wprk with standard glasses of accurately known
transmission (12) (25). The influence of the cells is then eliminated. It is
a disadvantage, that these standard glasses are expensive and difficult to obtain.
As a secondary standard use could be made of solutions of, for instance 12Cr207.
Nevertheless, standard glasses for different wavelengths should be available since
cell-corrections must be determined for different slit widths and wavelengths.
Preferably transmissions between 20 and 60% should be used.
(2) Deviations from Beer's law can also be caused as a result of non-linearity of the
instrument. It is clear that the fairly large unavoidable errors,maY suggest
a deviation in Beer's law. This in reality does not exist. If the ultraviolet
cell gives in practice a constant molar extinction coefficient for 1%, 3% and 9%
, then the red cell gives a difference of 2% (4). If extinctions smaller
than 0.1 (therefore in the inaccurate region) must be measured control of the law
is impossible in practice (26). In that case cells of different optical path
lengths should be available.
(3) Too little attention is paid to the error which arises because the solutions are
clouded. Particularly If "old" solutions are involved these should be centrifuged.
Also turbidity, which is not perceptiple to the eye, can be the cause of error
amounting to a few percent.
(4) By careful operation the error with comparative measurements can be brought dowry
to approximately 1%. This applies to methods of analysis in which comparison is
always made with a known standard, for instance a known solution.
(5) In the determination of absolute molecular extinction coefficients, errors of
at least 5.10% should be expected. A similar error can occur in an
analytical determination, in which calculation is made from the measured
extinction (E C 1) the concentration C and a value for e taken from the literature.
In reproducing extinction (absorption) curves, it is desirable to fulfil the
following conditions:
Integrate log E against . (in R's) where ? increases towards the right. Log E
gives a better representation because the variation in E is large in most cases.-
Often the wavelength is expressed in m ? since the wavelength is accurate approximately
to one m M. However, apart from the 1. It is perhaps better to use no other unit
for atomic dimensions, but to adhere to general spectroscope practice even though
this results in the last digit being generally a zero. The actual points of
measurement should be indicated on the curve.
A. the concentration relative to Beer's law,
B. the solvent. Cyclohexaae and methyl cyclohexane are good solvents for
aromatic compounds. For more polar materials, 95% alcohol may be used
(not the absolute alcohol, as this is often contaminated with traces of
benzene). Fo.r materials, which dissolve with difficulty; 1.4 dioxane
can be recommended, provided this is well purified. A list of solvents
that can be used and their degree of purity is given in reference V.
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
C. Temperature. The extinction curve varies with the temperature, at the rate of 0.1%
per degree centigrade. In this case the reduction In concentration as a result of
expansion of the fluid must be taken into account,
D. The width of the slit or, if possible, the spectral bandwidths.
E. The length of cell and the type of the instrument used.
LITERATURE
This article, without concentrating too much on details of operation gives a general
review helpful to the investigator working in the field of ultraviolet spectrographic analysis.
Among the literature previously mentioned, the articles by Hogness et. al. (13), Caster (4),
Gibson and Balcom (12) and Bastian at. al. (18) are specially important. Furthermore
tCOlorimetry and Spectrophotometry" by G. Kort&n can be recommended, although the modem American
instruments are dealt with too summarily (19). In the first 130 pages of the second part of
Physical Methods in Organic Chemistry, W. West deals with spec trophotome try in a way which Is
entirely in conformity with the general high level of this standard work (20). In the January
and February numbers of Analytical Chemistry (also obtainable separately) from 1949 onwards, there
are annual reviews of absorption and ultraviolet spe c tropho tome try. In these reviews very many
publications are represented. This applies also to the book by Sandell (23).
We are grateful for the co-operation of the Laboratory for Universal and Inorganic
Chemistry and for Analytical Chemistry, in particular to Professor Dr. J.A.A. Ketelaar and
Professor Dr. W. van Tongeren for their assistance in compiling this review.
DISCUSSION
J.A.A. K ETELAAR
The cleanliness and preparation of the cells are important factors. The relaCive accuracy
of measurements by i$.15 person with 2U Instrument of Q,I},$ material is certainly higher than that
which may be evident from the "test". As regards the problem of the choice of a material for
comparison, there are indeed only very few materials, for which the extinction is known with
satisfactory accuracy.
Experience teaches us, that the ultraviolet Beckman spectrophotometer can be used reasonably
well to 1,500 m /4 so that, the higher frequencies of carburetted hydrogen can be measured correctly.
The "automatic" Beckman discussed in the lecture is a very expensive instrument (price approximately
40,000 guilder).
P.B. ROTTIER
Should not a "test solution" such as K2C x207 be used regularly, in order to keep the user
of an ultraviolet spectrometer up to date, regarding the condition of his instrument?
As stressed in the present paper the regular checking of the ultraviolet instrument is
a constant requirement. However, the problem is, how can this best be done. In the author's
view the best method is to use a set of standard glasses or known transmission.
W. A IEY3
Has not the K2Cr2O7 been wrongly chosen as a standard for comparing the accuracy of
various Instrument (influence jai, easily reducible; 0.01 N H2T4 is also not buffered)?
K2C%07 is chosen principally because It possesses favourable physical properties, as
mentioned in the article. Indeed, one may expect that it will show some chemical instability,
so that care must be taken that any contact with easily oxydisable materials is avoided. At
the Laboratory for Analytical Chemistry, research is proceeding regarding the effect on the
extinction of K2Ct'2I7 contaminated with H 4 (20).
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
REFERENCES
(1) G. R. HARRISON, R. C. LORD and J. R. LOOFBOUROW.
Practical Spectroscopy (published by Prentice Hall, Inc. 1948)
ChaPter14 deals with absorption spec tropho tome try.
(2) H.H. CARY and O.A. BECKMAN. J. Opt, Soc. Am. 31 (1941) 682.
(3) C.J. F. BtTTCHER. Chem. Weekblad 44 (1948) 274.
(4) W.O. CASTER. Anal. Chem. 23 (1951) 1229.
(5) O.H. LOWRY and O.A. BESSEY. J. Biol. Chem. 163 (1946) 633; 160 (1945) 609;
166 (1947) 177.
(B) R.A. BURDETT and L.C. JONES. J. Opt. Soc. Am. 37 (1947) 554.
(7) New Instruments, Ref. Sol. Instr. 17 (1946) 568; Instrumentation Ind. Eng. Chem.
39 (April 1947) 75A compare also Harrison at. al. ref. 1.
(8) W. KAYE, C. CANON and H. G. DEVANEY. J. Op t. Soc. An. 41 (19 51) 668.
(9) M. G. MELLON (editor). Analytical Absorption Spectroscopy (Published by
John Wiley and Sons).
(10) E.J. BOWED. The Chemical Aspects of Light (Published by the Clarendon Press 1948).
(11) R.A. FRIEDEL and N. ORCHIN. Ultraviolet Spectra of Aromatic Compounds
(Published by John Wiley and Sons, Inc. New York 1951).
(13) T.R. HODGES, F.P. ZSCHEILE and A. E. SIDWELL. J.Phys. Chem. 41 (1937) 379.
(14) A. C. HARDY and F.M. YOUNG. J. Opt. Soc. An. 39 (1949) 265.
(15) W.H. EBERHARDT. J. Opt. Soc. Am. 40 (1950) 172.
(16) J. R. EDISBURY. Photoelectric Spectrometry Groups No. 5 Oct. 1952.
(17) G. W. EWING and T. PARSONS. Anal. Chem. 20 (1948) 423.
(i8) R. BASTIAN, R. WEBERLING and F. PALLILA. Anal. Chem. 22 (1950) 160; 21 (1949) 972.
(19) G. KORAN. Colorimetry and Spectrophotometry (Published by Springer-Verlag, Berlin 1948).
A. WEISSBERGER (editor). Physical Methods in Organic Chemistry.
(Published by Interscience Publishers Inc. New York 1946)..
(21) G.H. AYERS. Anal. Chem. 21 (1949) 652.
(22) C. F. HISKEY at. al. Anal. Chem. 21 (1949) 1440; 22 (1995) 1464; 23 (1961) 1196.
(23) F.& SANDELL. Colorimetric Determination of Traces of Metals.
I(Interselence Publishers Inc. New York 1950).
(24) N.T. GRIDGEMAN. Anal. Chem. 24 (195?) 445.
(25) K. S. GIBSON, O.K. WALKER and M.E. BROWN. J. Opt. Soc. Am. 24 (1934) 58.
(26) J.M. VANDENBELT, J. FORSYTH and A. GARRETT. Ind. Eng. Chem. Anal. Ed. 17 (1945) 235.
(27) A.M.G. RUTTEN. Chem. en Pharm. techniek. 8 (19M) 37, 57.
(28) A.W.M. INDEMANS. Thesis (Utrecht. 1983).
(29) From measurements made by Miss K. Janmaat at this Laboratory, it is evident that a
solution of K2Cr2"7 prepared in this way, retains a constant extinction for a period
of one week. The transmission, after three weeks, over the whole region has, on an
average, increased by 2 3%.
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
Approved For Release 2007/10/23: CIA-RDP78-04861AO00400030023-1
Fig.l: The optical system of the Beckman
spectrophotometer
Fig.2: Simplified circuit of the Beckman
spectrophotometer
ABSORPTION WRVE
(Schematic)
Fig-3: Band system for benzene at 2600 1
Ground state 992 cm-1 Ground state 606 cm-1
Excited state 923 cm-1 Excited state 520 cm-1
Approved For Release 2007/10/23: CIA-RDP78-04861AO00400030023-1
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
H. CRITICAL INVESTIGATION OF SIMPLE
SPECTROPHOTOMETERS
Chemisch Weekblad 42 (1953) 492-493
(From Dutch)
1. INTRODUCTION
This paper describes certain investigations which have been undertaken at the
Analytical Institute, T.N.09, for the purpose of obtaining some knowledge of the
reliability of simple spectrophotometers, wnich are universally in use in analytical
practice. The spectrophotometers, which are examined, are not very interesting as such,
unless they are examined in order to investigate certain principles of measurement. The
emphasis is upon the critical evaluation of the observations , a problem which generally
is more involved than the instrument itself. Similar investigations are at the moment
very much in vogue, witness the many "collaborative tests" which are undertaken. However
important such tests may become it is my impression that many of the individuals
participating have no thorough knowledge of the instrument, and worse still do not
maintain their instruments in good condition. This conclusion is in agreement with the
recent decision of the TAPPI (Technical Association of the Pulp and Paper Industry) to
be very wary of such large scale researches, since they produced more confusion than
clarification. in my opinion, it would be preferable for investigators to study more
seriously the properties and capabilities of the various Instruments and the correct
methods to be used in research.
The ultimate aim of this investigation is to answer questions regarding the best
method of measurement particularly with regard to accuracy in measuring the optimum
extinction (absorption).
2. NATURE OF POSSIBLE ERRORS
It is clear that an insight into the behaviour of an instrument can only be
obtained,by repeated measurements, under clearly prescribed circumstances. The instrument
must, as it were, Itself indicate what it can and what It cannot perform. The setting up
of the test must preferably adhere, as near as possible, to general practice. In addition
it is necessary to take into account the character of the sources of error, whether they
are systematic or fluctuate. It seems that there is also a hybrid form of error, namely
systematic errors, which build up and finally become subject to fluctuation. Although it
may sound absurd, we have been able with an objective; spectrophotometer, to ascribe such a
case to the subjective influence of the investigator. Eventually, these'-fluctuating
systematic errors become ordinary random errors (fluctuations). Itreally means that the
instrument, with which they are observed, are not sufficiently tested and calibrated.
When determining the magnitude of the fluctuations, in the results obtained with a
specific instrument, it is essential that the required measurements are undertaken at random,
i.e. In such a way that the sequence of the observations may in no way be connected with
the systematic continuity of the observations. As an example consider the setting up of a
standard line for a spectrophotometeic analysis. Assuming that we. wish to set up this line
on the basis of measurements, undertaken in triplicate, of five standard solutions, It is
incorrect to undertake the observations in the direction of Increasing or decreasing
concentrations of the standard solutions or to undertake the repetitions in succession with
one concentration. The correct method is to arrange the order of sequence by chance from
the total number of observations to be performed. The design standard V.1047, reproduction
of series of observations, gives the order of sequence required (Chapter 4).
When determining fluctuating systematic errors, it is essential to use a different
technique and to repeat the series of observations, singly and at random several times.
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
15-
3. TYPES OF SPECTROPHOTOMETER
Besides the chara;c:yer;af ti}a sources of errors, the method of measurement is of importance.
The method to be used depends on each individual case and no general rules can be laid down.
Spectrophotometers are divided into three groups:-
(a) deflection type (Coleman Jr.)
(b) compensation type (Nedoptifa)
(c) substitution type (Spekker absorptiometer).
The principles of measurement can be represented schematically as follows:
100 setting
o setting Measurement
Blank Blank
compensation
Proof
0 setting
Blank
+ Standard
Measurement
y 1 T
Comparison Comparison
SUBSTITUTIOT TYPE
The possibility of using a comparison solution with a final extinction (absorbance),
instead of a blank should be considered. This amounts to an "expansion" of the galvanometer
scale as regards the first type instrument. In this way, the optimum extinction becomes variable.
4. THE ROLE OF STATISTICS
In investigations of this kind, it is essential to take into account the fact that
fluctuations are superimposed on the systematic and the fluctuating systematic errors. It is
almost impossible to unravel this knot without making use of modern statistical methcds (standard
deviation, variance, spread).
It is essential to take into account the form or the frequency distribution of the
errors. These are certainly not, in all cases, of the normal Gaussian type. For instance, a
rectangular distribution may be encountered, i.e. distribution by which inside a certain interval
each error has the same a priori probability, and outside this interval a probability 0, e.g.
errors in reading divisions of the scale.
For this reason, modern methods which waive any supposition of the form or the frequency
distribution, the so-called rank correlation methods (2), are very important for this type of
investigation. A few applications to ape ctrophotome tric problems have been published (3).
5. DISCUSSION OF ERRORS
A discussion of errors, based on the Gaussian propagation law for fluctuatior..s, is
essential for the Interpretation of the fluctuations found experimentally.
A starting point must be the expression for the extinction, as a function of the
magnitudes actually measured (for instance galvanometer readings with spectrophotometers of the
deflection type).
Assuming that the extinction E is a function of the experimental magnitudes At ,/ and Z
9 = f(X,y,Z),
the variance of E (square of the standard deviation) is then given by
s2(&') _ (s X' SP-(X) + [_fj2 S'-N) + {rj 2 S2(z)
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
To this must be added the contributions of factors, which fall outside the functional
relation such as errors in scale reading.
These observations can lead to an evaluation of the relative significance of the
various sources of fluctuations and thereby point the way to improvements in the construction
and use of the instrument. As a general rule, the important point Is that a partial
variance, which is four times or more as great as the other put together, is the dominating
factor.
REFERENCES
(1) R. SCHMIDT. Metals 1 (1946/47) 37.
(2) M.G. KENDALL. Rank Correlation Methods. London 1948.
(3) R. SCHMIDT. Bulletin of the Photoelectric Spectrometry Group No. 5,
October 1952, p.liS.
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
-' 17 -
III. THE BEHAVIOUR OF THE SPEKKER ABSO'RPTIOMETER
R. Schmidt
Chemisch Weekbiad _42 ( 1953) 494-496
( From Dutch)
The Spekker absorptiometer has been critically evaluated in its two models H.560
and H.760.
It has been conclusively shown that neither of the two models exhibits a linear
characteristic of the drum diaphragm when equiped with a tungsten lamp. This causes
systematic differences between the results obtained on the same solutions with the two
principles of measurements, proposed by the manufacturer and by Vaughan respectively.
Since this is not the case with a H.560 instrument equiped With a mercury lamp (except
when the diaphragm is wide open) the cam which actuates the diaphragm cannot be held
responsible. The difficulties are linked up with the tungsten lamp, which is also
responsible for significant calibration curve drifts. Within equivalent calibration
curves the standard deviation of one measurement is about 2.5 x 10-3 irrespective of
absorbance (up to 0.65), wavelength or principle of measurement, provided the
galvanometer is sufficiently sensitive.
1. DESIGN OF THE INSTRUMENT AND PRINCIPLES OF MEASUREMENT
The Spekker absorptiometer is an Instrument of the substitution type for which a
schematic diagram is given In Figure 1.
The substitution is obtained by means of the combination cell and the measuring
diaphragm, which can be adjusted. The photocurrent, which at the measuring side of the
instrument, is generated by the photocell, is compensated by the photocurrent, which is
generated by a second photoelement with variable diaphragm. The method of measurement
must be independent of variations. In-the light current. This requirement Is met provided
the ratio between the transmissions of both halves of the Instrument-and that of the response
of both photocells are independent of the light current and the spectral distribution of
energy (1)*, Two methods of measurement have been given by the firm Hilger (2) and by
Vaughan (3) respectively.
Hilger: Zero setting with measuring diaphragm fully open and test solution
to the cell: blank measurement with measuring diaphragm adjusted
until the galvanometer again reads zero.
Vaughan: Zero setting with blank cell and measuring diaphragm opened by
1/10: measurement with test solution in the cell and the
measuring diaphragm adjusted until the galvanometer again reads
zero.
The manufacturers maintain that using their method a greater accuracy is obtained due
to the fact that the reading is more accurate. is will appear from what follows, it is
not possible to uphold this claim. Relatively few objective data are known regarding the
capabilities of the H.760'instrument. Isbell (4) has compared the improvOtehts in comparison
with the older types, without furnishing quantitative data.
Pollak and Nicholas (5) have reconstructed a Spekker absorption meter using a.
mercury lamp for the purpose of obtaining greater accuracy. This paper, however, gives
little positive information.
* For references, see page D.
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
2. INSTRUMENTS USED AND EXPERIMENTAL TECHNIQUE
7
Observations were made with two types:
(a) an older type (H. 560)"equiped with a tungsten lamp, as well as a mercury lamp, and
(b) the latest type(H.780);with a tungsten lamp.
The older instrument as equiped by us with a Kipp A75 galvanometer with adjustable
sensitivity, so that under any condition it can be operated with the same sensitivity for zero
Indication. In addition, a 500 watt resistance of 5 ohms was connected, in series, with the
tungsten lamp, in order to prolong the life of the lamp. All measurements were made with suitable
concentrations of a grey solution prepared from
16.7 gr.
potassium chromium (III)-alum
33.3 gr.
copper sulphate
3995 gr.
ammonium cobalt'
(II)-sulphate (6 H2O)
0.12 gr.
potassium dichromate
3. EFFECT OF VOLTAGE FLUCTUATIONS ON THE LAMP
From many series of observations in which the tungsten lamp was fed via a varlac, by which
the voltage was varied between 170 and 230 V it appeared that the voltage had no influence on the
measurement. The standard deviation of the extinctions obtained "between voltages" was, for
none of the filters, greater than values obtained normally over a short period with one voltage
(approximately 3 x 1073 extinction). Even alteration of the voltage between zero adjustment
and the measurement had no effect. The construction of an instrument, which is inde;endent
of the fluctuations in the light source has therefore certainly been achieved.
4. CHARACTERISTICS OF THE MEASURING DIAPHRAGM
The heart of the instrument is the measuring diaphragm, which is characterised, by the
following expression
trn (& = A a = A 10'R
tm, (00 = photocurrent in the galvanometer supplied by the photocell behind the
diaphragm for a relative opening Ct of the diaphragm.
R = the reading on the diaphragm scale.
This characteristic can be simply determined by setting the measuring diaphragm at a value
desired, setting the galvanometer with the left-hand diaphragm at zero and finally by covering up
the left-hand photocell. The galvanometer will then indicate the photocurrent, which is generated
in the measurement side (tm).
The characteris tics defined in this way for the tungsten lamp and the mercury lamp are
different (Figure 2). The curve for the tungsten lamp is non-linear, whereas that for the
mercury lamp, is practically linear, except at openings in the neighbourhood of 1.0. On repeating
the measurement, the characteristic of the tungsten lamp shows a very pronounced systematic
variation between 0,4 and 0.8 (with the H.560 model). On the contrary, the characteristic of
the mercury lamps shows a random variation neglecting the non-linearity at large apertures.
All this is reflected in the calibration curves determined for the instrument. When using the
tungsten lamp, the calibration curves are non-linear. In addition the calibration curves differ
because of the curvature of the diaphragn. (According to measurements made by Hilger and Vaughan).
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
Approved For Release 2007/10/23: CIA-RDP78-04861AO00400030023-1
When using the H. 760 model, tested by us It has been found that the calibration
curves differ significantly for the two methods of measurement. From this fact it may
be concluded by analogy that here also the characteristic of the tungsten lamp is not.
linear. While, however, the curve for the model H.580 is concave in relation to the
aperture axes, that for the H.760 model must be convex in relation to this axis (Figure 3)
The difference in behaviour between the tungsten and mercury lamp Is a proof, that the cam,
which actuates the diaphragm is correctly constructed.
FLUCTUATING SYSTEMATIC DIFFERENCES
BY means of a technique described elsewhere, and a parameter free statistical
investigation of the results obtained with the H.760 model, it can be demonstrated that
this model shows fluctuating systematic differences (calibration curve drift) If the
tungsten lamp is used. That means that the calibration curves are not all of equal value,
and cannot be used indiscriminately. In a somewhat different way, the same thing is
demonstrated for the H.560 model.
on the contrary, with the H.560 model using a mercury lamp, there are no signs of
any significant drift of the calibration curves.
It is quite justifiable to relate this association between the drift of the
calibration curve and the use of the tungsten lamp with the systematic deviations in the
characteristic of the measuring diaphragm. This claim is supported by the fact, that it
has not been found possible, using random observations, to obtain consistent estimates of
the standard deviation of the extinction measurement and especially in the region where
systematic deviations of the diaphragm characteristic were noticeable.
No absolutely valid explanation of these systematic deviations has yet been found.
A possible cause is the existence of extraneous light In the measuring diaphragm, due to
reflections from the filter, lens and the lamp holder.
6. RANDOM DEVIATIONS (FLUCTUATIONS)
From the data relating to four series of nine calibration curves determined with
varying concentrations of the grey solution for different methods of measurement and
different filters, the standard deviation at one observation was calculated as a function
of the extinction for each method of measurement using two filters (Ilford 602 and 607).
It appeared from this calculation, that the,standard deviation is independent of the filter
and of the method of measurement employed and depends only on the extinction. The standard
deviation increases from 3 x 10 3 at extinction 0 to 6 x 10 3 at extinction 0.8.
In this calculation, the fact that not all the calibration curves of gU series are
interchangeable was neglected. If we calculate the standard deviation for equivalent
calibration curves, then it appears that the dependence on the extinction disappears, so
that up to the highest extinction measured, viz. 0.85, a uniform standard deviation of
2,5 x 107,3 is obtained provided the galvanometer is sufficiently sensitive*, (Figure 4).
This fact is very noticeable because, with all conceivable sources of fluctuations
(reading zero adjustment, manipulation of the diaphragm, galvanometer reading) we expect
an exponential relationship between the reading of the diaphragm scale and the standard
deviation. On the other hand, the standard deviation thus found, Is approximately equal
to that mentioned earlier in connection with the effect of varying the voltage of the
tungsten lamp (section 3).
* This is the case when the displacement of the measuring diaphragm from the
position corresponding to zero current through the galvanometer required
to give a higher reading of 0.1 to Q 3 extinction units on the extinction
scale will at the same time produce a galvanometer deflection of half the
full scale length of a Cambridge or de Kipp galvanometer.
Approved For Release 2007/10/23: CIA-RDP78-04861AO00400030023-1
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
REFERENCES
See. also R.H. ALLER in Mr. MELLON Analytical Absorption Spectroscopy, New York,
London 1950., p.173=174.
(2) See Ref. 4,
O See,,, amongst others, F.W. HAYWOOD, A.A. R. WOOD. Metallurgical analysis by means of
theSpekker Photoelectric Absorptiometer. London (Hilger).
(4) R.A.C. ISBELL. Analyst. 74 (1949) M.
(5) F.F. POLLAK, J.W. NICHOLAS. Metallurgia 44 (1951) 319.
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
K, D? F? C? L C. F. D.
Fig.1: Schematic diagram of the Spekker Absorptiometer
L - lamp; C - cell; F = filter; D = diaphragm;
K = photocell. The indices v and m correspond to
the comparison and measuring sides respectively.
0 0.5 1,0a O,5 I.0-
Fig.2: Reproducibility of the measuring diaphragm
characteristic of the Spekker.
I H560 Tundaten
H760: Tunga te4 H56o Mercury
ywGNA
CONC
Fig.3: Calibration curves according to the methods
of Hilger and Vaughan for different instruments.
S[E].1Oa
Equivalent calibration curves
O,5 1,0 E
Fig.4: Reproducibility over long period of the extinction
measurements as a function of the extinction expressed as
a standard deviation of the extinction within equivalent
calibration curves.
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
Approved For Release 2007/10/23: CIA-RDP78-04861A000400030023-1
IV. PRECISION EXTINCTION MEASUREMENTS WITH
SIMPLE SPECTROPHOTOMETERS OF THE
DEFLECTION TYPE
Chemisch Weekblad _49 (1953) 496-499
(From _ Dutch )
This paper discusses the derivation and the significance of a formula for the
variance of extinctions measured with spectrophotometers of the deflection (Coleman Jr.
Model 6A,Engel colorimeter and the like). It Is shown inter alla, that the minimum of
the variation fraction for the extinction is not generally situated at the value 0.434,
but at a value which is determined by the ratio of two of the instrument parameters.
This ratio can be simply determined by experiment. The precision of extinction measurements
made with a Coleman Jr. Model'6A spectrophotometer is discussed on the basis of the
foregoing remarks.
E = log tref./tX'
In order to. measure the extinction of an unknown solution with a spectrophotometer
of the deflection type, two steps are necessary.
(a) The determination of the galvanometer deflection uref, for a comparison
solution - in practice this means adjusting the spectrophotometer
sensitivity so that Ure f. amounts to a full scale deflection L
The extinction is then equal to
E = log uref./"x ...... (1)
Since the factor U Is proportional to the photocurrent t of the spectrophotometer detector
we get
...... (2)
From (i) by differentiation squaring of the differentials and summation of the
squaras of dE over an infinitely large number of observations an expression can be derived
for the variance of the extinction vari(E) with respect to the reading errors,
vari (E) 0.4342 \2/12 L2 (OC2 + 102E) ...... (3)
1. A FORMULA FOR THE VARIANCE OF THE EXTINCTION
(b) The determination of the galvanometer deflection lAx for the.unknown
solution Itself.
= the smallest perceptible distance on the galvanometer scale and bears
the following relationship to var (U)*
var (u) = A.2/12 ?
Oc = var (uref)/var (ux); O is a constant of,