Document Sample
spectrometer Powered By Docstoc
					ANALYSIS/Spectrophotometric Techniques ± Infrared Spectroscopy


Further Reading
Andrews DL (ed.) (1990) Perspectives in Modern Chemical Spectroscopy. Berlin: Springer-Verlag. Banwell CN and McCash H (1994) Fundamentals of Molecular Spectroscopy, 4th edn. London: McGraw-Hill. Horman I (1984) NMR spectroscopy. In: Charalambous G (ed.) NMR Spectroscopy in Analysis of Food and Beverages, pp. 205±264. London: Academic Press. Wilson RH (ed.) (1994) Spectroscopic Techniques for Food Analysis. New York: VCH.

The theoretical basis of infrared spectroscopy followed by many applications within the laboratory as well as in the process line will be described brie¯y in this article.

Theory of Infrared Spectroscopy
Electromagnetic waves can interact with materials in different ways. The wave can pass through the material without interaction (transmission), the radiation can be re¯ected at the surface (re¯ection) or some part of the energy of the wave can be absorbed by the material. The absorption of energy by a molecule can cause the molecule to move translationally or to rotate, or it can cause speci®c groups within the molecule to vibrate or some electrons of the molecule to get excited. In the range from 2500 to 25 000 nm (mid infrared region), transitions from the ground state to the ®rst excited state of different speci®c groups occur. In the near infrared range, 800 to 2500 nm, transitions from the ground state to higher excited states (mainly second excited vibrational state) and combinations of two vibrations (each one from the ground state to the ®rst excited state) can be found. In the water molecule, a symmetric and asymmetric excitation of the bonds between the oxygen and hydrogen atoms occurs at approximately 2700 nm (Figure 1). Around 6250 nm, the bond angle between the oxygen and hydrogen atoms will change with time; this is called the deformation vibration. To induce vibrations, the energy of the radiation must exactly match the energy difference between the states. Furthermore, the dipole moment of the molecule must change. Otherwise, this vibrational state cannot be induced by infrared radiation. These two conditions are called resonance conditions for unpolarized light. To measure the excitations, one needs an instrument that allows continuous change in the wavelength (spectrometer). The ratio of the transmitted radiation intensity in the presence of a sample to the radiation intensity without sample is called transmittance. Measuring this ratio at all wavelengths as a function of the wavelength, one obtains a so-called

Spectrophotometric Techniques ± Infrared Spectroscopy
L Rudzik, Ahlemer Institut der Landwirtschaftskammer, Hannover, Germany È E Wust, University of Applied Sciences, Hannover, Germany
Copyright 2002, Elsevier Science Ltd. All Rights Reserved

Nowadays, different spectroscopic techniques are used in dairy laboratories. UV±VIS-spectroscopy (spectroscopy in the ultraviolet and visible region of the electromagnetic spectrum) plays an important role in enzymatic analyses and the analysis of enzyme-linked immunosorbent assay (ELISA) plates. Microwave spectroscopy is mainly used for the determination of the water content of powders, which can also be used within the process line (in-line measurement). Lowresolution nuclear magnetic resonance spectroscopy can be applied to determine the fat and water content of low-moisture products. The most important spectroscopic method in the dairy industry is infrared spectroscopy in the mid and near infrared region of the electromagnetic radiation. The reasons for its importance include:

 the method is fast and reliable and gives accurate  the analysis can be done with nearly no sample    
(A) y O H H x H (B) y O H x H (C) y O H x

preparation it is possible to determine the concentrations of various constituents simultaneously the method is inexpensive the analysis can be performed in the process line the method is environmentally friendly.

Figure 1 Vibrational modes of the water molecule: (A) symmetric valence vibration, (B) asymmetric valence vibration and (C) deformation vibration.


ANALYSIS/Spectrophotometric Techniques ± Infrared Spectroscopy

`spectrum'. The `valleys' in the spectrum represent wavelengths where transitions occur. The spectrum pro®le depends on functional groups that have been excited. Therefore, the spectrum has been used to explain molecular structure. Two kinds of analysis are possible with respect to the spectra: (1) qualitative analysis, and (2) quantitative analysis.
Qualitative Analysis

the low concentration range is c ˆ F0 ‡ FA ‰3Š

A qualitative analysis compares spectra and looks for similarities or differences. First, one creates a library of spectra of known substances. Having the spectrum of an unknown substance, one computes spectral distances to ®gure out if the spectrum of the unknown substance is the same as a spectrum in a library or not. If the distance is close to zero, both substances are identical. If the distance is large, it is not possible to identify the substance. Computing the distances to all library spectra, one obtains a so-called `hit-list'. This list gives the best matching library spectra, so that the user can get information useful for identi®cation. Another possibility is to perform a cluster analysis and to show the result as a dendrogram, which shows spectral similarities in a visual form. Instead of using the whole spectral information, one can perform similar kind of identi®cation based on reduced, compressed spectral information (principal component analysis, Fourier analysis, wavelets, etc.).
Quantitative Analysis

where the factor F represents the slope of the regression line and F0 the intercept. In addition, eqn [2] can be inverted so that one can directly predict the concentration of the molecule(s) of interest. To determine the values F0 and F, one must use standards, i.e. samples with known concentration (for example determined by reference analysis) and absorbance values. These samples are called calibration samples or calibration standards. With this information, linear regression determines F0 and F. To predict the concentration of an `unknown' sample, one has to measure the absorbance value, A, and then compute the concentration c according to eqn [3]. Having a mixture of different compounds that absorb radiation, the absorbance value, A, at one wavelength is not suf®cient to determine the concentration. One must use the absorption information at more wavelengths: c ˆ F0 ‡ F1 A1 ‡ Á Á Á ‡ Fn An ‰4Š

The intensity of radiation which is absorbed by the molecules is directly related to the number of the appropriate molecules and is described by the law of Lambert and Beer:
1 I…x† ˆ I0 0Àxc


where I0 is the intensity of the source, x is the optical path length (i.e. x describes the distance which the wave has travelled in the absorbing medium with absorption coef®cient ), and c is the concentration of the absorbing medium. Therefore, I(x) is the intensity of the wave after the distance x. Usually, the relation is transformed to the equation ! I0 ˆ xc ‰2Š A ˆ log10 I…x† The term A is called the absorbance. Having one kind of absorbing molecules only, the absorbance is directly related to the concentration of the molecule, A. Due to the fact that this relationship is only valid for highly diluted samples, the general equation for

The index (1, F F F , n) re¯ects the number of the used wavelengths. Ai is the absorbance value at the ith wavelength. With calibration samples, partial least square regression is the most common way of determining of F-values. Other strategies for the relation between absorbance values and concentration, like neural networks, fuzzy regression and so on, can be found in the literature. Such an approach (prediction of concentrations) will not lead to as accurate results as those obtained with calibration samples. However, the advantage is that the prediction is nearly independent of the person performing the analysis. Qualitative and quantitative analysis both require a high quality of spectral information. Standardization of sample preparation as well as spectral data pretreatment are necessary in some cases.
Standardization and Spectral Data Processing

The spectrum of any sample must have a high reproducibility. This can be reached by a standardization of the sample preparation. In some cases, the temperature must be constant, because the infrared spectra are very sensitive with respect to temperature variations (since heat is a form of infrared radiation). In other cases, grinding the sample is necessary due to variations in the particle size distribution. Pressure can also be of importance. Furthermore, the measuring device must allow a standardized intake of the sample into the infrared instrument.

ANALYSIS/Spectrophotometric Techniques ± Infrared Spectroscopy


Having optimized the sample preparation, some minor deviations still occur. These can partly be removed by mathematical processing of spectral data.

Applications in the Dairy Industry
Incoming Product Control

offers the possibility of ful®lling the task nearly simultaneously with the production process (on-line). To go a step further, infrared spectroscopy allows measurement of the compound of interest directly in the process line (in-line measurement). The following applications are used in the dairy industry: 1. The water content of milk powder may be determined directly after the drying chamber with a near infrared spectrometer. Having this information, one can regulate the concentrate feed to the chamber. 2. The water content of butter may be measured at the end of the buttermaking machine with a near infrared spectrometer. This can be used to control the separation by pressing which in¯uences the water content. 3. To standardize fat and protein for cheesemilk or fat for market milk, a mid infrared spectrometer may be used to measure these compounds for process control. 4. The formation of the coagulum during the cheesemaking process may be monitored with near infrared diffuse re¯ectance spectroscopy with ®bre optics. 5. The dry matter content of quark may be determined by a near infrared spectrometer in transmission mode with ®bre optics. Figure 2 shows the technical set-up of this last example for the production line of quark and Figure 3 the in-line device in the process line. Figure 4 shows

Considering a medium-size dairy supplied by 10 000 farmers, 10 000 samples have to be analysed nearly every or every second day depending on the collection of the raw milk. The fat, protein and lactose contents are determined for each sample with mid infrared instruments. These parameters are used for the payment of the farmers. The average fat content is around 4.3%, the protein around 3.3% and the lactose around 4.7%. The accuracy of determination is approximately Æ0.03% for these constituents. Nowadays, it is also possible to separate the total protein content into casein and whey protein using infrared spectroscopy. At the moment, suitability of this technique to measure urea and citrate concentrations is being discussed. Some dairy products need additives (e.g. stabilizers in yoghurt or dessert products). The routine check of a truck load is often performed by inspection of the delivery document and a sensory test, because everything else is very time consuming. In some cases the additives are brought into the store and used in production after further investigation. Both situations have drawbacks: in the ®rst case, the sensory test is not suf®cient to detect speci®c problems with the additive. In the second situation, if the load does not meet the requirements, it must be returned. A robust analytical method is necessary which is fast enough so that it can be used while the truck is waiting at the manufacturing site, and gives more information than a sensory assessment. A near infrared instrument with ®bre optics, for instance, can be used for this purpose at the entrance to the plant. Having a spectral library of correct additives, one can compare the spectrum of the delivered additive with the spectra in the library. A hit-list or dendrogram will show the classi®cation result. However, one should be aware that infrared spectroscopy is not a method which can solve everything. One has to know the limitations of the method.
Process Control


2 7 4 1 6 5
Figure 2 Schematic representation of a quark production line. The milk is coagulated in the tank (1), pumped through a heater (2) and feeding pipe (3) to the separator (4), where the milk is separated into whey and quark. A pump (5) brings the quark through the in-line device (6) and a cooler (8) to the packaging unit. The near infrared spectrometer (7) is equipped with a pair of ®bres and works in transmission mode.


To control a process, it is necessary to have the essential information just in time. Traditional chemical analysis of constituent concentrations is too time consuming, because the process could have changed in the meantime. Infrared spectroscopy


ANALYSIS/Spectrophotometric Techniques ± Infrared Spectroscopy

the dry matter distribution in quark in response to different means of process control: the solid curve shows the dry matter distribution after 4 weeks of controlling the separation process in the traditional manner. Using the infrared prediction of the dry matter content to control the separator, the distribution is reduced by a factor of 2 (long dashed curve). To reconcile economic considerations with the legislative requirements with respect to the dry matter content, one can reduce the internal value of the desired dry matter content (short dashed curve). Economical considerations can be found in the literature. Apart from these quantitative measurements, qualitative in-line analysis is possible to control or improve the process.
End Product Control

Nearly all major constituents in all dairy products can be analysed by infrared spectroscopy. Some important parameters measured by infrared spectroscopy are listed in Table 1. The accuracy of the near infrared prediction, i.e. the difference between

the results obtained by the reference method and the infrared prediction, is close to the repeatability of the reference method. The absolute difference of two analytical results (same person, same instruments and chemicals, short time between the analyses) with the reference method, on identical material, should be within the repeatability value of the reference method, at 95% probability. In principle, infrared spectroscopy cannot do better, because results obtained with the reference method are used for calibration. There is a tendency to move this type of analysis out of the laboratory towards the production line, as well as to adapt infrared spectroscopy to the process line. Another use for infrared spectroscopy is in the routine identi®cation of microorganisms, based on mid infrared spectroscopy. After a cultivation (multiplication), the spectrum of the unknown species is compared with the spectra of a library. A hitlist or a dendrogram helps to identify the unknown species. The advantage of the method is that 2 to 3 days can be saved. This technique has also been developed as a routine method within larger
Table 1 Constituents of dairy products determined by infrared spectroscopy Product Liquid Raw milk Skim milk Market milk Coffee cream Evaporated milk Whipped cream UHT cream Cocoa concentrate Constituents Fat, protein, casein, whey protein, lactose Dry matter, protein, casein Fat, protein, dry matter Dry matter, fat Dry matter, fat Dry matter, fat Dry matter, fat Dry matter Dry Dry Dry Dry matter, protein matter matter, fat, protein matter, fat fat fat, protein, lactose fat, protein fat salt fat

Figure 3 The in-line measuring device in a quark production line.

Part of the batch

0.25 0.20 0.15 0.10 0.05 0.00 17 17.5 18 18.5 19 Dry matter content (%) Using NIR 19.5

Viscous Low-fat quark Modi®ed quark Quark Fruit quark Powder Skim milk powder Milk powder Coffee creamer Cappuccino Yeast autolysate Creamer Others Feeding stuffs Ice cream mix Lecithin

Water, Water, Water, Water, Water, Water,



Figure 4 Dry matter distribution by using different ways of process control.

Various constituents Dry matter, fat Various constituents

ANALYSIS/Spectrophotometric Techniques ± Infrared Spectroscopy


hospitals and in other industrial areas (e.g. the pharmaceutical industry and breweries). To ensure proper calibrations, one must arrange a monitoring system, which is called good laboratory practice (GLP) for infrared calibrations.
Good Laboratory Practice for Infrared Calibrations

As with all methods of chemical analysis, one must check the performance of the method regularly. Due to the fact that the infrared method is an indirect method (requiring a calibration step) for the determination of the constituent concentrations, one must set up a more complex checking routine testing three kinds of potential problems. 1. One must ensure that the instrument operates within an acceptable error. This can be done by taking the infrared spectra of inert standards over time (a certi®ed material of known concentration is used to check whether the method is reliable). If the difference between the standard spectrum and the spectrum obtained with the instrument is unacceptable, the instrument must be readjusted. 2. If sample preparation is a necessary step for the technique, it must be tested to determine if the operators satisfy the demands of the standard operating procedure. This can be evaluated by preparation of the same material several times and predicting the constituent concentration. Limits will help to clarify this step. This is also done with other chemical analysis methods. 3. The performance of the calibration must be monitored. This is not always possible with certi®ed material in the case of a calibration because certi®ed materials sometimes do not exist (e.g. for a calibration for fat in yoghurt; no yoghurt exists which can be used as certi®ed material). The only way is to analyse the corresponding sample by the reference method and compare the difference between reference value and infrared prediction over time. Preset limits of the difference can help to de®ne warning and action levels. The last point is very important, because changes in the recipe can in¯uence the infrared spectrum. The difference between the results obtained by the reference method and the infrared predictions needs to be constantly monitored; sometimes a new calibration is necessary. Usually, the difference between the results obtained by the reference method and the infrared prediction is plotted versus time. This task can be accomplished by a control chart that also shows the warning and action levels.

Having arranged such a system, one can show that the infrared method gives more accurate values than the reference method when performing a multiple analyses of the same, inert material. The infrared predictions are more constant and have smaller variations than the reference values. To set up such a GLP system, much experience is necessary. The fast way is to operate a network so that the performance of many instruments can be monitored simultaneously.

There are three kinds of networks, illustrated by an example from Lower Saxony.
Service Network

Since 1988 the Ahlemer Institut of the Landwirtschaftskammer Hannover has operated a service network, which has been accredited by the German Accreditation Council (DAP). There are nine dairies with 12 near infrared instruments connected by telephone and modem to the Ahlemer Institut (Figure 5). The Institute conducts feasibility studies, validates new applications and performs the GLP procedure. The advantage is that the individual dairies do not need to employ trained and expensive personnel to

Sittensen Zeven Uelzen

Moers Rehburg Hannover

Georgsmarienhütte Weissenfels



Figure 5 Map of the Milchwirtschaftliches Infrarot Netzwerk (MIRN: Infrared Instrument Network for Dairies) including Lower Saxony (a state in Germany) and the dairies connected.


ANALYSIS/Spectrophotometric Techniques ± Infrared Spectroscopy

perform the infrared analysis. Furthermore, persons with different scienti®c backgrounds work together in the Institute, so that any problems can be solved in less time. Nowadays, businesses generally focus on their key activity and therefore are outsourcing their other activities.
Surveillance Network

The Ahlemer Institut has been given responsibility by the Lower Saxony government to check the mid infrared instruments which are used for computing the payment to the farmers according to the constituent concentrations in raw milk. This is done in the following manner: 1. On a weekly basis calibration standards are sent to the laboratories to test the performance of instrument and calibration. After approximately 200 raw milk samples, one of the standards has to be reanalysed. Analysis is repeated after each 200 raw milk samples. The results of tests on standards can be transferred by modem to the Institute for further evaluation. 2. Each month, a series of standards is prepared to check the calibration over a wider concentration range. These results are transferred to the Institute. The advantages of this network are that the calibration is monitored more often, the checking is done without the necessity for travelling and is thus less time consuming and the Institute staff can help with their expertise. The network serves as a con®dence-building measure between farmers and laboratories.
Harmonization Network

funded by the European Union, it has been demonstrated how one can achieve this harmonization goal for infrared spectroscopy. The method is based on the concept of `matching instruments', where one instrument is used as the `master' (reference) instrument. Having compared the characteristics of the master and the other instruments, the spectra of the other instruments are transformed so that they match those of the master instrument. Spectra obtained in this way look as if they were obtained with the master instrument. Using the calibration of the master instrument, the correct sample composition can be predicted. This ensures that all predictions include the same information, and therefore all instruments behave in the same manner.

Summary and Future Trends
Infrared spectroscopy is a powerful tool for determining constituent concentrations and qualitative characteristics of dairy products. Many examples and applications show that the technique is accurate and fast and is therefore used for process control. To ensure the proper performance of instruments, it is necessary to establish GLP guidelines. Within these guidelines, the main point is the monitoring of calibration. The results suggest what one must do: to adjust the existing calibration or to conduct a new calibration. To obtain optimal results, much experience is necessary. Some dairies use the service of a network, thus outsourcing the calibration and application work. Infrared spectra are dominated by the water content of the product and nearly all dairy products have high moisture levels (except powder products). Spectra of water and milk, for example, look very similar. Reliable information can only be obtained by applying statistical methods. In another excitation technique, Raman spectroscopy, water does not disturb the spectra in such an extreme manner. This method has the potential for further applications.
See also: Analysis: Atomic Spectrometric Techniques; Biosensors; Spectroscopy, Overview.

Harmonization of analytical results is a big issue in two ways: 1. Large dairies with more production sites, which transport milk or products from one site to another, would like to ensure that all measurements, performed on the same product at different locations, are the same or at least in good agreement. Otherwise the mass balance creates problems. 2. Results of tests on exported products, obtained at different laboratories, should be in good agreement. With respect to the chemical methods, standard operating procedures are de®ned as well as precision parameters (used for checking the results). However, the results are strongly dependent upon individual operators. Within a research and development project

Further Reading
Barnes RJ, Dhanoa MS and Lister SJ (1989) Standard Normal Variate Transformation and de-trending of near-infrared diffuse re¯ectance spectra. Applied Spectroscopy 43: 772±778. Helm D, Labischinski H, Schallehn G and Naumann D (1991) Classi®cation and identi®cation of bacteria by Fourier transform infrared spectroscopy. Journal of General Microbiology 137: 69±94.

ANALYSIS/Atomic Spectrometric Techniques Martens H and Naes T (1989) Multivariate Calibration. New York: John Wiley. Naumann D, Fijala V and Labischinski H (1988) The differentiation and identi®cation of pathogenic bacteria using FT-IR and multivariate statistical analysis. Mikrochimica Acta 1: 373±397. Osborne BG and Fearn T (1986) Near Infrared Spectroscopy in Food Analysis. Harlow: Longman. Savitzky A and Golay MJE (1964) Smoothing and differentiation of data by simpli®ed least squares procedures. Analytical Chemistry 36: 1627±1633. Wang Y, Veltkamp DJ and Kowalski BR (1991) Multivariate instrument standardization. Analytical Chemistry 63: 2750±2758. Wietbrauk H, Hulsen U and Wust E (1998) Ressourcen bei È È der Frischkaseproduktion. Deutsche Milchwirtschaft È 23: 993±996. Wuest E, Neemann H and Rudzik L (1992) NIR-calibration methods and their tolerance with respect to random errors of the reference values. In: Hildrum KI, Issakson T, Naes T and Tendberg A (eds.) Proceedings of the 5th International Conference of Near Infrared Spectroscopy, pp. 67±72. London: Ellis Horwood.


Atomic Spectrometric Analysis
In atomic spectroscopy, the sample is placed in an environment that is hot enough to break molecular bonds and produce atoms. The atoms can be identi®ed and their concentration measured by the emission or absorption of characteristic radiation. In atomic absorption spectrometry (AAS), a light source emitting radiation characteristic of a speci®c element (usually a hollow cathode lamp) is passed through the atomized sample and the transmitted radiation is measured. In atomic emission spectrometry (AES), the sample is heated to suf®ciently high temperatures that the electrons of the atoms are excited from their ground electronic state to an excited state. As the electrons return to ground state, they emit radiation characteristic of the element present. In conventional AAS and AES, photon detectors are used to detect the radiation transmitted through the atomizer. Increasingly, however, mass spectrometry is used as a detector and very high sensitivity is achieved.

Minerals Analysed in Dairy Products Using Atomic Spectrometry

Atomic Spectrometric Techniques
C M M Smith, University College, Cork, Ireland
Copyright 2002, Elsevier Science Ltd. All Rights Reserved

Atomic spectrometric techniques are used for the qualitative and quantitative determination of approximately 70 elements. Element concentrations down to the part per billion level (ng mlÀ1) are detectable using these techniques. Atomic spectroscopy is used routinely in many laboratories because of this high sensitivity. Other advantages of the techniques include high selectivity, reasonable cost, speed and ease of use. In fully automated mode, hundreds of analyses per day can be carried out with little operator input. The commercially available systems for atomic spectrometry vary signi®cantly in terms of cost, ease of operation and analytical performance. In this overview, the techniques will be discussed with relation to their relevance to the analysis of the minerals found in milk and dairy products. Sampling and sample preparation (often the most critical part of an analysis) will also be considered.

All the 15 elements for which United States Recommended Dietary Allowance (RDA) and Adequate Daily Dietary Intake values exist (Ca, P, I, Fe, Mg, Cu, Zn, Se, Cr, Mo, Mn, F, Na, Cl and K) can be determined using some type of atomic spectrometric technique. In addition, many of the elements that are considered essential, but have no de®ned requirement levels for humans, can also be determined using atomic spectrometry. This latter group includes As, Ni, Si, B, Cd, Pb, Li, Sn, V and Co. A number of elements in both of these categories can be toxic at high levels and so their analysis in food products is also important. The atomic spectrometric method of choice for the determination of minerals in dairy products depends, of course, on the element to be determined, but also on the type of product to be analysed.

Sample Preparation
Atomic spectrometric techniques are conventionally optimized for handling liquid samples. Some notable exceptions to this rule (arc and spark emission and laser ablation techniques, for example) are not utilized traditionally for the analysis of foods. As a result, the ®rst step in most analyses is the digestion or decomposition of the sample to facilitate liquid sample handling. It is important that the entire sample is digested to ensure that the elements(s) of

Shared By: