AALTO YLIOPISTON TEKNILLINEN KORKEAKOULU
Signaalinkäsittelyn ja akustiikan laitos
MIKES Aalto Mittaustekniikka
Optical rotary dispersion of sugars
16.9.2012
TY Teemu Kääriäinen, 69724R
teemu.j.kaariainen@aalto.
Mittaustekniikan erikoistyö kurssiin S-108.3120 Erikoistyö.
Opintopisteet (ECTS):
Arvosana (1 5):
Ohjaajan allekirjoitus:
TkT Toni Laurila
Contents
1 Introduction
3
2 Method
4
3 Setup
8
4 Results
11
2
1 Introduction
1
Introduction
This feasibility study on optical rotary dispersion of sugars was accomplished
in 2011-2012 as a part of the LUMO (Liquid Monitoring Using Novel Optical Sensors) research project funded by TEKES and six industrial partners.
The motivation for the work is the need for new and alternative aordable
measurement devices required for the analysis and monitoring of biofuel processes. Cellulostic ethanol is produced from materials such as wood, plants
and grasses. The process involves breaking the cellulose down to sugars [1]
and thus there is a need for a device capable of measuring the concentrations
of sugars in a solution that contains dierent sugar types. At the time this
is done by chemical analysis which is time consuming and expensive. The
suggested measurement technique could be used in real time, without the
need of taking samples and the parts needed for the device are low cost.
Sugars are well known for their natural property to rotate the polarization
of light upon transmission. The amount of rotation is wavelength dependent and dierent for specic sugar types. This phenomenon is utilized in
optical measurement devices in food and beverage industry. By measuring
the amount of rotation, sugars can be identied and their purity examined
[2]. The scope of this work is to research if the wavelength dependence of
the phenomenon, optical rotary dispersion (ORD) can be used to measure
concentrations of sugars in a solution that contains at least two dierent
sugar types.
The rst chapter describes the background information and the theory of
optical activity. The method of using ORD to identify dierent sugar components is explained in chapter 2. In the third chapter the experimental
set-up for measuring ORD is described. Finally the results are analyzed and
conclusions drawn in chapter 4.
3
2 Method
2
Method
Certain materials, named as optically active, have natural ability to change
the polarization of transmitted light. Sugars are well known for being optically active. Due to the chiral structure of sugar molecules, the refractive
index is dierent for right and left handed circularly polarized light. For
linearly polarized light this means that the plane of polarization rotates as
it travels through the material as illustrated in Figure 1.
Figure 1: Polarization rotation
In the case of sugar-water solutions, the amount of polarization rotation
depends on the optical path length, temperature, concentration and the
specic rotation of the sugar. The specic rotation is unique for dierent
species and is dened as [3]
[α]Tλ =
α
lp
(1)
where α is the observed rotation in degrees, l is the path length in dm and p
is the density of the liquid in g/mL. The specic rotation is wavelength and
temperature dependent and is usually given at the standard wavelength of
589 nm (Sodium D line) at 20 o C temperature, identied as [α]20
D . Values of
specic rotations for sugars used in this work are listed in Table 1.
The sign of the specic rotation tells the direction of the rotation when
viewed towards the light source. Rotation of the polarisation is clockwise
for + and counterclockwise for -. Temperature dependence of the specic
rotation is relatively small and can be modelled with equation [3]
4
2 Method
Table 1: Specic rotation of sugars
Sugar
[α]20
D
Sucrose
66.3
Fructose
91.7
Glucose
54.3
α(t) = α(20o C)[1 − 0.000471(t − 20)]
(2)
For example 1 o C decrease in temperature increases the specic rotation of
fructose by 0.04 degrees.
Measurement instruments that use optical rotation are called polarimeters.
They are widely used in food and beverage industry, for example, in identifying and quantifying purity of sugars. Polarimeters that are used specifically for measuring sugar-water solutions are called Saccharimeters. There
are commercial devices available that can have measurement accuracy up
to parts-per billion (ppb) level. Such devices work at a single wavelength
and the net rotation observed is always the sum of the dierent optically
active components in the sample, and thus, these devices fail to determine
individual concentrations in samples containing multiple arts of sugars.
Optical rotary dispersion (ORD) spectrum is the specic rotation of the
material as the function of the wavelength of the light. The ORD spectrum
is called "plain" for molecules lacking a chromophore (absorption in the
measurement bandwidth). The specic rotation is inversely dependent on
the square of wavelength and can be modeled with Drude's expression [4]
[α] =
λ2
A
− λ20
(3)
where A is the rotation- and λ0 is the diusion constant. For optically
5
2 Method
active materials with a chromophore the ORD spectrum is more "anomalous"
due to absorption within the measurement range. In that case there is a
rapid change in the direction of the polarization rotation at the absorption
wavelength. This behavior of the ORD spectrum is called Cotton eect. The
Cotton eect is said to be positive if the change in rotation is from negative to
positive with decreasing wavelength and negative otherwise. The absorption
peak is shown in Figure 2A and the cotton eect is shown in 2B (positive)
and 2C (negative). The direction of the change depends on the type of the
material.
Figure 2: Cotton eect
In an attempt to identify multiple sugars and measure their concentrations
in sugar-water mixtures, ORD spectra for sucrose, glucose and fructose were
measured in the visible wavelength region 500-750 nm. ORD spectra were
also measured for solutions containing two dierent sugar types at varying
6
2 Method
concentrations. Using Equation (1) the rotation of the mixed sample can be
expressed as
α=
c22
c21
l[α]1 +
l[α]2
c1 + c2
c1 + c2
(4)
where l is the optical path length, [α]1 and [α]2 are the specic rotations and
c1 and c2 are the concentrations of the sugar types. Levenber-MaquardtFletcher [5] algorithm is used to calculate values c1 and c2 from the sample
data.
7
3 Setup
3
Setup
The measurement setup was built at MIKES in Espoo, Finland. The shcematic and a photo of the measurement set-up are shown Figure 3. Light from
a halogen source 1 is polarized and collimated 2 through the sample tube
3. The second linear polarizer 4 is crossed in resepect to the rst polarizer.
Transmitted light is focused to ber connected to a spectrometer 5.
Figure 3: Experimental setup, components are described in the text.
The initial idea was to employ low-cost parts, white LED as the light source
and a CCD (Charge Coupled Device) camera for detection. The collimated
light beam is dispersed in to a spectrum using a prism and the spectrum is
recorded with the CCD camera. For research purposes, more suitable parts
were chosen. Broadband Hamamatsu L10290 was used as a light source
and an ocean optics HR4000 grating spectrometer for detection. In grating
spectrometer an optical grating is used to disperse the broadband light at
a CCD linear detector array1 . CCD consist of an array of MOS (Metal
Oxide Semiconductor) capacitors that store a charge according to the light
intensity that hits the photoresistive surface of the capacitor. The charges
1
More on the spectrometer see: http://www.oceanoptics.com/Products/hr4000.asp
[cited 1/2013]
8
3 Setup
are converted into voltage and read row by row to processing2 . A plastic
20 cm long sample tube was used with 0.1 mm thick glass windows as a
cuvette containing the sample. The sample is injected and removed from the
tube with a syringe. The sample tube is washed between measurements to
prevent contamination of samples.
The optically active sugar solution rotates the plane of the polarized light
according to equation (1). The maximum contrast is obtained when the
sugar solution has rotated the polarized light by 90 degree due to the crossed
polarizers and in contrast, no intensity is seen by the spectrometer when no
optical activity of the sample is involved. The angle of rotation depends on
the transmitted intensity according to Malus' law
I(θ) = I0 cos2 (θ)
(5)
where θ is the angle from the axis that gives the maximum intensity I0 .
The intensity transmission of the setup was calibrated with pure water at
room temperature using an adjustable polarizer as the second polarizer. The
relative transmitted intensity was measured every 10 o . The results are shown
in Figure 4 with polynomial t and the theoretical values from equation (5).
The polynomial t was used to convert the transmitted intensity in to angle
of rotation.
The procedure for measuring and calculating the ORD specta is as follows:
1. Sugar solution is injected in the sample tube
2. Intensity thourgh the crossed polarizers is measured
3. Second linear polarizer is removed to measure the reference
2
More on CCD:s see for example: http://spi.rit.edu/classes/phys445/lectures/ccd1/ccd1.html
[cited 1/2013]
9
3 Setup
4. The relative intensity is converted to rotation using the polynomial t
obtained from calibration
The attenuation of the linear polarizer is non zero and the transmission of
the reference spectrum will be higher than expected due to absence of the
second polarizer. The measured spectrum must be compansated with the
attenuation spectrum of one linear polarizer.
Figure 4: Calibration of the setup
10
4 Results
4
Results
All measurements were done at room temperature. Measured specic rotations with Drudes t (3) are shown in Figure 5. The circles represents
values, for Sucrose and Fructose from the supplier of the sugar samples [8]
and for Glucose [7]. The ORD spectra for all three sugars species are so
called "plain" curves explained earlier in Methods.
Figure 5: Measured specic rotations
Measured values for glucose were higher than expected due to mutarotation.
20 of +112o and
Glucose has two stereoisomeric forms, a-D-glucose with αD
20 of +19o . When dissolved in water the equilibrium
β -D-glucose with αD
mixture of 36% of a-glucose and 64% of β -glucose starts to form. Conversion
occurs through s small amount of open chain form. The dierent structures
of Glucose are shown in gure 6
Observer rotation of 1:4 and 4:1 ratio of glucose-fructose mixture are shown in
Figures 7 and 8. Blue lines represent the measured rotation of the samples,
the red and green curves represent the eect of the glucose and fructose
11
4 Results
respectively.
Figure 6: Structure of Glucose
Figure 7: Observed rotation of glucose fructose mixture 1
Figure 8: Observed rotation of glucose fructose mixture 2
12
4 Results
Although due to featureless shape of the ORD spectra it was undoubtedtly
clear that accurate solution analysis can not be achieved, an attempt was
made with the Levenberg-Maquardt-Fletcher algotithm [5].
The algorithm was run 10 000 times with random initial guess of the solution in range of ±50% percent of the actual concentrations. The maxmimum
error tolerance for equation (4) was set to 0.01. The variation of the solutions using the algorithm was highly dependent on the initial guesses of the
concentrations. The solutions outside the 50% percent range were discarded.
The amount of discarded solutions was 9636 for mixture 1 and 9805 for mixture 2, so most of the solutions were out of bounds. The inaccuray of the
accepted solutions are shown in Figures 9 and 10. The x-axis represents the
number of the accepted solutions and y-axis inaccuracy for that solution.
The average inaccuracy was around 29% for mixture 1 and 25% for mixture 2. The deviation of the accepted results is very high and no reliable
conclusions of the concentrations can be made with the results.
As expected, no accurate results could be achieved and the idea of using
ORD in visible region for measuring concentrations in a solution containing
at least two dierent sugar species is not likely to be achieved in a low cost
measurement devices, if at all. One possible way to continue the study is to
use UV light instead of visible, but this brings up challenges and expenses
in the detection scheme, sample tube and polarizers.
For samples containing only one sugar specie or other optically active substance, ORD can possibly enchant the accuracy of the measurement compared to single wavelength measurement. To be used in the biofuel production, the sugar species must rst be separated using some alternative
method.
13
4 Results
Figure 9: Inaccuracy mixture 1
Figure 10: Inaccuracy mixture 2
14
References
References
[1] Carlos A. Cardona et al. Fuel ethanol production: Process design trend
and integration opportunities
[2] Yeshajahu Poeranz, Food Analysis: Theory and Practice
[3] Standardbase,
Polarimetry.
Available
at
http://www.standardbase.hu/tech/SITechPolar.pdf ,[cited 1/2013]
[4] Paul Drude, The theory of optics. 1902
[5] LMFnlsq
- Solution of nonlinear least squares. Available at
http://www.mathworks.com/matlabcentral/leexchange/17534lmfnlsq-solution-of-nonlinea ,[cited 1/2013]
[6] Laurence Barron, Molecular Light Scattering and Optical Activity. Cambridge university press
[7] Steven Nanis, Measuring the specic rotation of optically active solutions.
Bennington College Spring 2011
[8] Merck-chemicals http://www.merck. ,[cited 1/2013]
15