Caballero B. (ed.) Encyclopaedia of Food Science, Food Technology and Nutrition. Ten-Volume Set

Подождите немного. Документ загружается.

commodities were introduced in the early 1970s.

These instruments were dedicated primarily to the

determination of moisture, protein, and oil in grains

such as wheat, barley, and soybeans. In the interven-

ing years, rapid advances in optics, digital electronics,

and computer technology have greatly improved the

reliability of the technique, as well as making it pos-

sible to analyze a much larger range of constituents in

a wide array of products.

Origins of Spectra

0005 A molecule can absorb infrared radiation if it vibrates

in such a way that its charge distribution, and there-

fore its electric dipole moment, changes during the

vibration. A vibrating molecule, or molecular func-

tional group, can absorb radiant energy to move

from the lowest (ground) state to its first excited

state. The frequency of radiation that will make this

occur will be identical to the initial frequency of vibra-

tion, and is referred to as the fundamental absorption.

Typically, fundamental absorptions have frequencies

that fall in the midinfrared region of the electromag-

netic spectrum (4000–650 cm

1

, wavelength ¼2.5–

15 mm). Molecules can also absorb radiation to move

to a higher (second or third) excited state, such that the

frequency of the radiation absorbed is two or three

times that of the fundamental frequency. Such absorp-

tions are referred to as ‘overtones,’ and the intensities

of these absorptions are much lower than the corres-

ponding fundamental, since these transitions are less

favored. Combination bands can also occur if two or

more different vibrations interact to give bands with

frequencies that are sums of their fundamental fre-

quencies. Combination and overtone bands typically

have frequencies that fall in the NIR spectral region.

0006 Both the overtone and combination bands ob-

served in the NIR tend to be weak in intensity. How-

ever, this is actually an advantage, since absorption

bands that have sufficient intensity to be observed in

the NIR region arise primarily from functional groups

that have a hydrogen attached to a carbon, nitrogen,

or oxygen atom. These are common functional

groups in the major constituents of food, including

water, proteins, lipids, and carbohydrates. The NIR

absorption bands associated with a number of im-

portant food constituents are listed in Table 1. The

NIR absorption bands of many foods frequently over-

lap, yielding spectra that are quite complex. This can

be illustrated by observing the spectra of the starch

and zein (protein) components isolated from corn

(maize) shown in Figure 1. The protein fraction

exhibits absorption bands centered at about 2050

and 2180 nm, whereas starch has a strong absorption

band centered at 2100 nm. Note, however, that there

is significant overlap of the starch band into the

regions where protein absorbs. Because of such over-

laps, quantitative analysis of most foods requires

measurements to be made at multiple wavelengths

in order to relate the amount of a constituent present

to the spectral measurement.

Principles of Diffuse Reflectance

Measurements

0007One of the advantages of NIR spectroscopy is its

ability to take measurements directly on solid food

Wavelength (nm)

1900

1100 1300 1500 1700

2100 2300 2500

0.467

0.365

0.264

0.163

0.061

−0.040

Log 1/R

Starch

Zein

Corn fractions

fig0001Figure 1 Near infrared reflectance spectra of the starch and

zein (protein) components of corn (maize).

tbl0001 Table 1 Absorption bands for important food constituents

Constituent Absorber Wavelength (nm)

Protein, peptides Amide I, II, and III combinations 2050–2180

—N—H stretch first overtone 1520–1570

Carbohydrate C—O, O—H stretching combination 2060–2150

Lipid Methylene —CH stretch þ deformation combination 2300–2350

—CH

and —CH

C—H stretch first overtone 1690–1765

Water —O—H stretch first overtone 1400–1450

O—H stretch þ O—H deformation combination 1920–1950

SPECTROSCOPY/Near-infrared 5427

samples with little or no sample preparation. This is

typically accomplished by measuring the diffusely

reflected radiation as a beam of NIR energy is directed

on to the sample. When radiation strikes a solid or

granular material, part of the radiation is reflected

from the sample surface. This mirrorlike reflectance is

referred to as specular reflectance, and gives little

useful information about the sample. Most of this

specularly reflected radiation is directed back toward

the energy source. Another portion of the radiation

will penetrate through the sample surface and be

reflected off several sample particles before it exits

the sample. This is referred to as ‘diffuse reflectance,’

and the diffusely reflected radiation will emerge from

the surface of the sample at various angles through

180



. Each time the radiation interacts with a sample

particle, the chemical constituents in the sample can

absorb a portion of the radiation. Therefore, the dif-

fusely reflected radiation contains information about

the composition of the sample, indicated by the

amount of energy absorbed at specific wavelengths.

It should be noted that the size and shape of the

sample particles can affect the amount of radiation

that penetrates and escapes from the sample surface.

0008 In NIR spectroscopy, the intensity of the diffuse

reflectance from the sample is quantitatively meas-

ured at each wavelength of interest. The intensity of

radiation reflected from the sample is compared with

the intensity reflected from a nonabsorbing reference,

such as a ceramic or fluorocarbon material. Reflect-

ance (R) is calculated by the following formula:

R ¼ I=I

;

where: I ¼ the intensity of radiation reflected from

the sample at a given wavelength and I

¼ the inten-

sity of radiation reflected from the reference at the

same wavelength. Reflectance data are expressed

most commonly as log(1/R), an expression analogous

to absorbance in transmission spectroscopy. Various

mathematical transformations, such as derivatization

and scatter correction, can also be applied to the

spectra. Such treatments may help to remove spectral

differences that can occur due to variations in sample

particle size.

0009 Transmission measurements can also be made in

the NIR region, and this is often the method of choice

for liquid samples. A liquid can be placed in a quartz

cuvette, and the absorbance measured at the wave-

lengths of interest. Transmission measurements can

also be taken from solid samples, but generally only

in the 800–1100 nm range. In this wavelength region,

the absorption bands are higher overtones that are

very weak, allowing the radiation to penetrate through

several millimeters of a solid sample. The use of

transmission measurements can minimize errors that

could occur with reflectance measurements taken

from samples having heterogeneous surfaces.

Instrumentation

0010The radiation source in most NIR spectrometers is

a tungsten–halogen lamp with a quartz envelope,

similar to a projector lamp. Such lamps emit large

amounts of energy in both the NIR and visible regions.

Wavelength selection in many NIR instruments is

accomplished with a grating monochromator. These

may be scanning monochromator instruments where

a grating is used to disperse the radiation by wave-

length, and is then rotated to impinge a single wave-

length, or, more appropriately, a narrow band of

wavelengths, on to a sample at any given time. With

this arrangement, it may take several seconds to col-

lect a spectrum from a sample over the entire NIR

region. Some rapid scanning instruments impinge

light covering the entire NIR region on to the sample,

with the reflected or transmitted light then directed

on to a fixed grating that disperses the light by wave-

length, and also focuses it on to a multichannel array

detector that measures all wavelengths at once. These

instruments allow a spectrum to be collected from a

sample in less than 1 s.

0011Instruments dedicated to specific applications often

use optical filters to select from 6–20 discrete wave-

lengths that can be impinged on a sample. The filters

are selected to obtain wavelengths that are known to

be absorbed by the sample constituents. The instru-

ment inserts filters one at a time into the light beam to

direct individual wavelengths (or narrow bands of

wavelengths) of radiation on to the sample.

0012Recently, commercial NIR instruments based on

interferometry and using Fourier transform (FT)

mathematics have become available. In these FT in-

struments, the radiation is not dispersed as with a

grating monochromator, but rather all wavelengths

arrive at the detector simultaneously, and a Fourier

transform mathematical treatment is used to convert

the results into a typical NIR spectrum. Because all

wavelengths are measured simultaneously with an

FT-NIR instrument, these instruments can acquire

spectra very rapidly with a very high signal-to-noise

ratio. Also, such instruments have very high reso-

lution, indicating that they can discriminate differ-

ences in reflectance or absorbance occurring between

very small wavelength (or wavenumber) intervals.

While this may not be much of an advantage for

food and agricultural materials whose absorption

bands are very broad, it can be important when

working with mixtures of pure chemicals that have

narrow absorption bands, such as in the pharma-

ceutical industry. More complete descriptions of

5428 SPECTROSCOPY/Near-infrared

interferometric principles and Fourier transform

mathematics can be found in the Further Reading

section.

0013 Semiconductor detectors are most commonly used

to measure the NIR radiation reflected from or trans-

mitted through a sample. Silicon detectors are used

primarily in the 800–1100 nm range, whereas lead

sulfide (PbS) detectors are used for wavelengths

from 1100 to 2500 nm. Solid-state detectors with

a more rapid response, such as indium–gallium

arsenide (InGaAs), are being used more frequently

in FT and rapid scan dispersive instruments because

of their rapid response to changes in light intensity,

but they are limited to a maximum wavelength of

around 1700 nm.

0014 When taking diffuse reflectance measurements, ar-

rangement of the sample and detectors is extremely

important. Most commonly, solid or granular food

samples are packed tightly into a cell against a quartz

window. Quartz does not absorb in the NIR region,

and packing the samples against a window helps to

provide a smooth, uniform surface from which reflec-

tion can occur. For diffuse reflectance measurements,

it is important that the specular reflectance not be

measured. In some instruments, this is accomplished

by positioning the detectors at a 45



angle with re-

spect to the incoming infrared beam, such that the

mirrorlike radiation reflected back toward the infra-

red source is not measured. Other instruments use an

integrating sphere, which is a gold-coated metallic

sphere with two holes cut on opposite sides of the

sphere, one positioned directly over the sample, and

the other through which the radiation beam is intro-

duced. The detectors are mounted inside the sphere.

The sphere collects the diffusely reflected radiation

coming at various angles from the sample and focuses

it on to the detectors. The specular component exits

from the sphere through the same port by which the

incident radiation beam enters and strikes the sample.

0015 Recent advances in fiber optic technology allow

NIR energy to be transmitted through a fiber optic

cable some distance from the monochromator or

interferometer. This allows either reflectance or trans-

mission measurements to be made remotely from the

instrument. Various commercial sampling probes are

available that can be inserted directly into bulk granu-

lar materials, or can be inserted in a pipe carrying a

flowing liquid.

Quantitative and Qualitative Analysis

Using NIR

0016 Because of the overlapping nature of NIR absorption

bands arising from various food constituents, it is

usually necessary to take measurements at two or

more wavelengths to quantitate a food component

reliably. Most commonly, some type of multivariate

regression technique is used to develop mathematical

models, or calibrations, that relate NIR spectral

measurements to the amount of a particular constitu-

ent in a food. The most widely used regression tech-

niques include multiple linear regression (MLR),

partial least squares (PLS) regression, and principal-

components regression (PCR).

0017The first step in calibrating an NIR instrument for

quantitative analysis is to select a set of calibration, or

training, samples. The samples should be representa-

tive of the products that will be analyzed, contain the

constituent of interest at levels covering the range that

is expected to be encountered, and have a relatively

uniform distribution of concentrations across that

range. The calibration samples are analyzed by the

classical analytical method normally used for that

constituent, and spectral data are also obtained on

each sample using the NIR instrument at all available

wavelengths. All data are stored in computer memory.

0018When multiple linear regression is applied for cali-

bration development, it is used to select the optimum

wavelengths for measurement, along with an inter-

cept and associated coefficients for each wavelength,

such that an equation with the following form is

developed:

% constituent ¼ z þ a logð1=R

Þþb logð1=R

þ c logð1=R

Þþ...

Derivatized or other transformed spectral measure-

ments can be used in lieu of the raw log (1/R) values.

Wavelengths are selected based on statistical signifi-

cance by using a step forward or reverse stepwise

regression procedure, or by using a computer algo-

rithm that tests regressions using all possible combin-

ations of two, three, or four wavelengths to determine

the combination that provides the best results. Most

calibrations will use between two and six wave-

lengths, and one should always check to make sure

that the wavelengths chosen on the basis of statistical

significance also make sense from a spectroscopic

standpoint. The use of too many wavelengths can

result in ‘overfitting’ of the data, decreasing the pre-

dictive capability of the technique. Calibration results

are evaluated by comparing the multiple correlation

coefficients, Fs of regression, and standard errors for

the various equations developed. It is desirable to

maximize the correlation coefficient (generally, R

should be > 0.9) and minimize the standard error.

0019PLS and PCR regression techniques can make use

of the entire NIR spectrum, rather than a few selected

wavelengths, to predict sample composition. PLS and

PCR both use data reduction techniques to extract

SPECTROSCOPY/Near-infrared 5429

from a large number of variables (i.e., reflectance or

absorbance measurements at many wavelengths) a

much smaller number of new variables that account

for most of the variability in the samples. These new

variables then can be used to develop a regression

equation to predict the amount of a constituent in

samples of a food. When using PLS and PCR

methods, it is not necessary to eliminate spectral in-

formation, as it is when measurements at only a

limited number of wavelengths are used. PLS and

PCR methods have been found to yield improved

results for some analyses when compared with the

MLR technique.

0020 Artificial neural networks (ANN) have also re-

cently been used to predict composition from NIR

spectra. Neural networks may have some advantages

over the linear regression techniques for dealing with

highly complex samples, samples from diverse geo-

graphic regions, or samples where the relationships

between composition and spectral properties do not

change in a linear manner.

0021 No matter which calibration technique is used, the

models developed should always be tested by using

the instrument to predict the composition of a set of

test samples that are independent of the calibration

set, and comparing the results obtained with those

from the classical analytical method.

0022 NIR spectroscopy can also be used for qualitative

analysis, classifying a sample into one of two or more

groups, rather than providing quantitative measure-

ments. Discriminant analysis techniques can be used

to compare the NIR spectrum of an unknown sample

with the spectra of samples from different groups.

The unknown sample is then classified into the

group to which its spectrum is most similar. This

technique has been used most widely in the chemical

and pharmaceutical industries for raw material iden-

tification, but it is beginning to be used more widely

for food applications.

Advantages and Limitations of NIR

Spectroscopy

0023 NIR spectroscopy provides a number of advantages

over traditional wet chemical methods of analysis.

Foremost, once an instrument has been calibrated,

samples can be analyzed very rapidly, usually in less

than 1 min. Also, sample preparation is minimal, and

in some cases unnecessary. The technique requires no

chemical reagents, so no hazardous wastes are gener-

ated. The need for weighing of samples is eliminated

as well, thereby eliminating a traditional laboratory

bottleneck.

0024 NIR instruments are available that can be used in

a production plant environment either in an at-line

mode or as on-line monitors. Again, once an instru-

ment has been calibrated, it can be used routinely by

production personnel without extensive training. The

ability of this technology to provide real-time process

control has tremendous potential.

0025Limitations of the technology include the need for a

specific calibration for each product to be analyzed.

This is generally not a problem in a production envir-

onment where only a few products are likely to be

manufactured. However, this limits the usefulness of

the technique for situations such as contract labora-

tories, where a few of many different types of samples

may need to be tested each day. Also, early instru-

ments required frequent recalibration. For example,

when measuring protein in wheat, it was usually ne-

cessary to recalibrate an instrument when grain from

a new crop year arrived. However, the use of large

calibration sets containing samples from wide geo-

graphic areas and time frames, more stable instru-

ments, and more robust calibration techniques such

as PLS has made the need for frequent recalibration

much less. In fact, recent results have shown that

surprisingly diverse sample types can be analyzed

with a single calibration model. Research conducted

by Kays et al. at the USDA Russell Research Center in

Athens, GA found that dietary fiber could be meas-

ured successfully in products as diverse as snack

crackers, granola bars, sugar-coated cornflakes and

cookies, all with a single calibration.

0026Another limitation for users with low sample

throughput is the capital cost of commercial NIR

instruments. Instrument cost can vary widely,

depending on the level of sophistication and automa-

tion needed. However, capital expenditures can often

be recovered rapidly through labor and reagent

tbl0002Table 2 Applications of NIR spectroscopy in food analysis

Product Constituents measured

Cereal grains and flours Moisture, protein, starch, oil,

dietary fiber

Oilseeds, flours, meals Moisture, oil, protein, fiber

Bread Moisture, protein, fat, fiber

Cereal breakfast foods Moisture, dietary fiber, sugar

Pasta Moisture

Cheese Moisture, fat, protein

Butter Moisture, fat

Milk and whey powders Moisture, fat, protein, lactose

Red meat, poultry, fish Moisture, fat, protein

Processed meats Moisture, fat, protein

Dehydrated eggs Moisture, fat, protein

Fresh fruits and vegetables Total sugar, soluble solids

Dehydrated fruits Moisture, total sugar, fiber

Potato chips Moisture, fat

Beer Alcohol

Corn sweeteners Sugar

5430 SPECTROSCOPY/Near-infrared

savings, and by minimizing the production of out-of-

specification product.

Applications

0027 Applications of NIR spectroscopy to food analysis

have greatly expanded since its first use for measuring

protein and moisture in wheat and barley. An in-

depth discussion of specific applications is beyond

the scope of this chapter; however, a number of rep-

resentative applications are listed in Table 2.A

number of NIR procedures have been adopted as

approved methods of analysis by the American Asso-

ciation of Cereal Chemists. Typically, if a substance

or constituent absorbs in the NIR region, and is pre-

sent at a level of a few tenths of a per cent or greater, it

has potential for being measured by this technique.

0028 In addition to performing compositional analyses

of foods and ingredients, NIR has also been used to

measure specific chemical constituents in foods that

affect their end-use qualities, for directly predicting

processing characteristics of commodities that are

related to chemical composition, and for monitoring

changes that occur during processing. Examples of

such applications include determining the amylose

content in rice, an important determinant of rice

starch quality; predicting corn-processing quality for

dry and wet milling; estimating the baking quality of

hard wheat; monitoring the degree of cook during

extrusion processing; and monitoring the coagulation

of milk during cheese-making. In some of these cases,

correlations between NIR measurements and meas-

urements from the conventional technique may be

< 0.9, but the NIR technique is still good enough for

purposes of screening out very good and very poor

materials.

0029 Examples of qualitative applications of NIR spec-

troscopy include the classification of milk powders

based on the level of heat treatment, the classification

of wheat as hard red spring or hard red winter, the

detection of internally infesting insects inside kernels

of stored wheat, and the identification of orange juice

samples from different sources.

Further Reading

AACC (2000) Approved Methods of Analysis, 10th edn.

St. Paul, MN: American Association of Cereal Chemists.

Burns DA and Ciurczak EW (eds) (2001) Handbook of

Near-Infrared Analysis. New York: Marcel Dekker.

Griffiths PR and De Haseth JA (1986) Fourier Transform

Infrared Spectrometry. New York: John Wiley.

Kays SE, Windham WR and Barton FE (1998) Prediction

of total dietary fiber by near-infrared reflectance

spectroscopy in high-fat- and high-sugar-containing

cereal products. Journal of Agricultural and Food Chem-

istry 46: 854–861.

Kramer R (1998) Chemometric Techniques for Quantita-

tive Analysis. New York: Marcel Dekker.

Mark H (1991) Principles and Practice of Spectroscopic

Calibration. New York: John Wiley.

Martens H and Naes T (1989) Multivariate Calibration.

New York: John Wiley.

Osborne BG, Fearn T and Hindle PH (1993) Practical NIR

Spectroscopy with Applications in Food and Beverage

Analysis. Harlow, UK: Longman.

Wehling RL (1998) Infrared spectroscopy. In: Nielsen SS

(ed.) Food Analysis, 2nd edn. Gaithersburg, MD: Aspen.

Williams PC and Norris KH (2001) Near Infrared Tech-

nologies in the Agricultural and Food Industries, 2nd

edn. St. Paul, MN: American Association of Cereal

Chemists.

Fluorescence

L Munck, Royal Veterinary and Agricultural University,

Frederiksberg, Denmark

Introduction

0001Fluorescence (the name originates from the fluores-

cent mineral fluorspar) refers to cold light emission

(luminescence) by electron transfer in the singlet state

when molecules are excited by photons. The after-

glow lasts less than 10

9

s and it is independent of

temperature, in contrast to phosphorescence, which is

light from the triplet state defined as an afterglow

lasting longer than 10

6

s, and which is tempera-

ture-dependent.

0002Thus, the fluorescence phenomenon is a form of

energy transfer mediated in each case by a specific

electronic arrangement, which upon absorbing exci-

tation light of a specific wavelength transforms this

energy to another rather specific emission peak,

always at a longer wavelength. With suitable filters

and mirrors in the fluorescence microscope, fluores-

cent components are visualized against a black back-

ground after filtering out the shorter-wavelength light

used for excitation. Similarly, the grating monochro-

mator of a spectrofluorometer enables a sample to be

illuminated with light of a specific excitation wave-

length. The light emitted at 90



to the incident light

passes through another grating that allows only the

specific emitted light of interest to pass through to

reach the photodetector. Just as in the microscope,

absence of fluorescence gives a black background, so

in the spectrofluorometer the absence of fluorescence

SPECTROSCOPY/Fluorescence 5431

gives total darkness, i.e., no emitted light. This leads

to one of the main problems in fluorescence analysis –

how to calibrate against zero light and a zero-signal

baseline. This, together with differences in equipment

design, confers individual characteristics on spectro-

fluorometers to a much greater extent than with spec-

trophotometers. Spectrofluorometers are up to 1000

times more sensitive than most spectrophotometers,

but while this may confer considerable advantages, it

may also cause additional problems, e.g., with fluor-

escent impurities. Moreover, the fluorescence phe-

nomenon is sensitive to environmental conditions,

such as temperature, viscosity, pH, and interacting

substances, which can diminish the intensity of fluor-

escence by quenching.

0003 Due to the unique structure of the fluorescence

emission landscape from a sample, new data tech-

niques may deconvolute the excitation and emission

spectra of the pure underlying fluorophores as a form

of mathematical chromatography. With chemometric

methods, whole fluorescence spectra may be cali-

brated, e.g., to predict oxidation in vegetable oils,

thus complementing current predictive near infrared

and infrared spectroscopic screening methods with

methods which are both more sensitive and specific.

0004 Fluorescence analysis can be used to follow separ-

ation processes of plant and animal chemical com-

ponents and tissues in food production because of

naturally present fluorescence markers (autofluores-

cence or primary fluorescence). Thus, through pri-

mary fluorescence, and without using chemical

reagents, we can study the refinement process of

sugar, the health of chloroplasts in plants, respiration

enzymes in microorganisms and fish meat, and one

state of lipid oxidation in meat. The value of fluores-

cence techniques resides in the fact that specific

changes in food components during processing can

be followed in the fluorescence microscope and then

these changes can be quantified using homogenized

samples, and even with a turbid suspension, in a

spectrofluorometer. Specific fluorochromes, prepared

by organic synthesis or tagged on to specific immuno-

globulins, can also be added with different affinities

for chemical structures and enzymes (Table 1) and

thus biochemical changes in foods can be followed.

These are termed secondary fluorescence techni-

ques. Here again we can combine the structural visu-

alization in the fluorescence microscope with direct

spectroscopic measurements of the fluorescence in-

tensity.

Physics and Chemistry

0005In general, the vast majority of intensely fluorescent

organic compounds (fluorophors) are aromatic in

nature, such as benzene and naphthalene. Some non-

aromatic but highly conjugated compounds, such as

b-carotene and vitamin A, are also fluorescent. In

exploiting secondary fluorescence, nonfluorescent

derivatives such as fluorescein dibutyrate (FDB:

Figure 1) and calcofluor white Mr2 new (Table 1)

are used, which can be rendered strongly fluorescent

by enzymic modification or by binding, respectively.

(See Carotenoids: Occurrence, Properties, and Deter-

mination; Polycyclic Aromatic Hydrocarbons;

Retinol: Properties and Determination.)

0006Any fluorescent molecule has two characteristic

spectra (Figure 2): the excitation spectrum (the

tbl0001 Table 1 Examples of the use of fluorochromes

Compound Excitation

wavelength

(nm)

Emission

wavelength

(nm)

Application

Acridine orange 470 530–650 Starch, nuclei, pectin, microorganisms, DNA, RNA, acidic nuclear

proteins

Aniline blue 420–490 520 Cell wall, (1,3)-b-

D glucans, callose

Auramine 400–440 470 Bacteria, cutin

Benzoylarginyl-b-naphthylamide 338 410 Protease activity (trypsin)

Calcofluor white CFW 350 435 Cellulose, chitin, collagen, elastine creatine, bacteria, fungal

mycelium/spores

Calcofluor white Mr2 new 360 415 b-

D-Glucans in cell walls from, e.g., barley, oats, and wheat

Dansyl fluoride 495 525 Primary amines, A-T bases of DNA

Fluorescein dibutyrate 495 525 Esterase/lipase activity

Fluorescein isothiocyanate (FITC) 490 520 Used as a marker of immunoproteins; in yeast viability test and in

cell culture

Phenylalanyl-b-naphthylamide peptides 335 410 Amino peptidase assay

Resorufin butyrate 540 560 Cellulase

The conditions under which the fluorochromes are used largely determine their specificity and fluorescence properties. See Guilbault GG (ed.) (1990)

Practical Fluorescence ^ Theory, Methods and Techniques. New York: Marcel Dekker and Francisco A (1989). In Munck L (ed.) Fluorescence Analysis in Foods,

pp. 266–276. Harlow: Longman.

5432 SPECTROSCOPY/Fluorescence

relative efficiency of different wavelengths in causing

fluorescence) and the emission spectrum (the relative

intensity of radiation which is emitted at various

wavelengths at a given, e.g., maximal, excitation

wavelength). Theoretically, the shape of the excita-

tion spectrum of a fluorescent substance, discounting

instrumental artifacts, is identical with that of the

absorption spectrum as measured by spectrophotom-

etry. The longest-wavelength peak should be chosen

for excitation of the sample, thereby minimizing

photodecomposition of the fluorophore by ultraviolet

(UV) light. Rayleigh and Tyndal scattering can be

observed in the emission spectrum at the same wave-

length as the excitation wavelength, and also at twice

this value (second-order grating effect) in a grating

spectrofluorometer.

0007The basic equation defining the relationship of

fluorescence of a substance to its concentration is:

F ¼ FI

ð1  e

ebc

For very dilute solutions, this equation is reduced to

Beer’s law in spectrophotometry:

F ¼ KFI

ebc

where F is the quantum efficiency of yield (number of

quanta emitted divided by number or quanta

absorbed), I

is the incident radiant power, e is the

molar absorptivity, b is the path length of the meas-

urement cell, and c is the molar concentration of the

fluorophore. K designates an arbitrary constant with

no unit.

0008The sensitivity of fluorescence is large because a

signal over a zero background is measured. Its speci-

ficity is high because, although all compounds absorb

radiation, only a few emit, and those that do can be

differentiated on the basis that they very seldom have

the same excitation and emission maxima. The main

disadvantage of fluorescence is that the output is

strongly dependent on temperature although the

afterglow time is independent. It is also dependent

on pH, ionic strength, and viscosity, making it neces-

sary to standardize analysis conditions rigorously.

Fluorescence is affected by quenching, which reduces

OOCCH

Lipase

2CH

COOH

Fluorescent

Fluorescein

FDB

Nonfluorescent

Fluorescein dibutyrate

fig0001 Figure 1 Use of secondary fluorescence. Fluorescein dibutyrate used to detect lipase (esterase) activity by enzymic cleavage to the

yellow fluorescent substance fluorescein (excitation 495 nm, emission 525 nm). Reproduced from Spectroscopy: Fluorescence, En-

cyclopaedia of Food Science, Food Technology and Nutrition, Macrae R, Robinson RK and Sadler MJ (eds), 1993, Academic Press.

(a)

(b)

(c)

450

520

350

275

335

365

420

300 400 500 600

Wavelength (nm)

Fluorescence intensity (arbitrary units)

fig0002 Figure 2 Excitation and emission spectra in the detection of the

botanical components of wheat by their primary fluorescence.

Excitation (- - -) for (a) pericarp; (b) aleurone; (c) endosperm.

Emission spectra of manually dissected pericarp (——), aleur-

one (––), and endosperm (- –) obtained at maximum excitation

wavelength for these components: (a) 450 nm; (b) 350 nm; (c)

275 nm. The primary fluorescence probably originates from car-

otenoids (a), ferulic acid (b), and tryptophan (c). Measurements

were made on suspensions of ball-milled flour in glycerin. Re-

produced from Jensen SAA, Munck L and Martens H (1982) The

botanical constituents of wheat and wheat milling fractions. I.

Qualification by fluorescence. Cereal Chemistry 59: 477–484, with

permission.

SPECTROSCOPY/Fluorescence 5433

fluorescence by a competing deactivation process

that might be enhanced by oxygen, impurities (e.g.,

dichromate and iodide), high temperature and high

concentrations of the fluorophore.

0009 Concentration quenching may be substantial in

transmission mode (inner-cell effect), whereas it is

negligible in front-face (reflectance) fluorescence

with solid samples.

Instrumentation and Sample Preparation

Light Sources

0010 Spectra for two conventional light sources for micro-

scopy (HBO, mercury lamp; XBO, xenon lamp) and

spectrofluorometry (XBO) are presented in Figure 3.

Xenon lamps for spectrofluorometry need a certain

burning time in order to obtain a constant spectral

output, and they have a limited lifetime of 200–

1000 h. When working with the information from

complete spectra, lamp spectral stability is crucial,

and spectral instability cannot be totally compensated

for across complete spectra by using the rhodamine

standard in a ratio spectrofluorometer, as explained

below. Recently, monochromatic laser techniques

have increased light intensity and wavelength preci-

sion, but at high cost. The aim of increased light

intensity is often counteracted by photodecom-

position of the fluorophore which necessitates short

illumination periods.

Fluorescence Microscopy

0011Ideally a minimum of preparation should be used so

that the material can be examined in as near a native

state as possible. A suitable technique is cryosection-

ing at freezing temperatures (20



C) with and with-

out additives to prevent ice crystal formation.

Simpler, but also effective, is hand-sectioning with a

razor blade at room temperature. The sections or

macerated material are transferred to a microscope

slide, stained, and covered with a cover slip. Perman-

ent preparations need fixation of samples, e.g., with

glutaraldehyde, followed by embedding in a resin

such as glycol methacrylate (GMA), which is espe-

cially suitable for fluorescence, including immuno-

fluorescence; sections are then cut with a glass-knife

microtome and stained appropriately. GMA is easy

to handle at room temperature and need not (and

cannot) be removed from tissues prior to staining.

Tissue sections do not require hydration or dehydra-

tion prior to staining. GMA fluoresces minimally; the

polymerized block is much harder than paraffin and it

can thus be sectioned more thinly.

0012The most frequently used fluorescence microscopy

technique is epiillumination (reflectance), which is

most light-efficient owing to its specific filter/optics

arrangement (Figure 4), but the older transmission

mode (diaillumination) can also be used. There is an

inverse relationship between the number and volume

of glass pieces in an objective lens and its usefulness

in fluorescence techniques because a more sophisti-

cated optical construction compromises effective

light transmission. It is difficult to find light-efficient,

low-magnification objectives for fluorescence work

because of their low numerical aperture. Modern

sensitive films and video cameras can compensate

for the limiting factor of low light intensity in fluores-

cence microscopy.

0013Successful fluorescence microscopy is critically

dependent on the use of filters that have transmit-

tance characteristics such that they can excite max-

imum fluorescence in a specimen (excitation filter)

100

1000

10 000

400 500

Wavelength (nm)

600 700 800

HBO 100

100

(b)

(a)

1000

10 000

400 500 600 700 800

XBO 75

fig0003 Figure 3 Light sources. Examples of spectral outputs typical of

common illuminators in fluorescence microscopy and spectro-

fluorometry. (a) Mercury lamps (e.g., HBO series), mostly used in

microscopy, generate very intense peaks at certain wavelengths

(e.g., 365 nm (not shown), 450, 435, and 546 nm), while (b) xenon

(e.g., XBO series) lamps produce a more uniform spectrum at

lower intensity, and these are used in both fluorescence micro-

scopy and spectrofluorometry. Reproduced from Fulcher RG,

Irwing DW and Francisco A de (1989) Fluorescence microscopy:

Applications in food analysis. In: Munck L (ed.) Fluorescence An-

alysis in Foods, pp. 59–109. Harlow: Longman.

5434 SPECTROSCOPY/Fluorescence

and subsequently filter out all but the fluorescent

wavelengths (barrier filter) from the final image

(Figure 5). Fluorescence microscopes can be com-

bined with sophisticated monochromators and de-

tectors which can extend into the short-wave UV

region (220nm) with the use of quartz objectives.

001 4 An attractive refinement of the fluorescence micro-

scopy is the laser confocal microscope (such as the

Leica Lasertechnik TCS NT confocal microscope). A

laser beam is scanned over a fixed specimen, restrict-

ing illumination and detection to the same small

volume. Detection, via a computer-controlled video

system that controls the scanning function, is limited

to what is in focus, allowing the study of crude

samples in three dimensions and eliminating glare.

Up to four lasers can be attached to the microscope

from UV to near infrared. The specimens studied

are usually stained by fluorochromes responsive to

the specific wavelength of the laser.

Spectrofluorometry

0015 Recent research has shown that spectrofluorometers

may be used in combination with computers and

suitable software to predict important quality factors

in foods analogous with the established near infrared

spectroscopic techniques. (See Spectroscopy: Near-

infared.)

0016Applications are, for example, to monitor forma-

tion of colored substances (Maillard compounds) in

sugar processes, oxidation products in frying oils, and

purity of pectins. However, the application of these

techniques in industry has been delayed due to lack of

suitable instruments which are able to measure a

sample area large enough to obtain correct sampling

of the often nonhomogeneous food products.

0017Sample holders delivered with the instrument are

for front-surface (reflected) or transmitted light and

employ quartz cuvettes. Fluorescence of insoluble

particles such as wheat flour can be studied after

fine ball-milling and suspension (about 1 mg ml

1

)

in fluorescence-free glycerin/water which slows

down sedimentation.

0018A typical laboratory spectrofluorometer with a

measurement area of only 1  9mm is shown in

Figure 6. It is a convenient tool for obtaining com-

plete excitation and emission spectra of homogeneous

fluorescent samples, for identification and quantifica-

tion through data processing, and for measuring

fluorescence at a defined pair of excitation and emis-

sion wavelengths. For some applications in which

maximum excitation and emission wavelengths are

known, a simple and inexpensive filter fluorometer

is superior to a spectrofluorometer because the super-

ior accuracy in wavelength definition obtainable with

Exciter filter

Chromatic beam splitter

FT 420

Fluorescence emission

Chromatic

beam

splitter

FT 420

Radiation above 420 nm

below 420 nmRadiation

Radiation

below 420 nm

above 420 nmRadiation

Objective

− condenser

Exciting radiation

Specimen

(a) (b)

Specimen

fig0004 Figure 4 Diagrammatic representation of a typical fluorescence microscope epiillumination system equipped with a filter combin-

ation,asshownin Figure 5(a).(a)Broad-spectrumilluminationfirstpassesthroughacoloredglassfilterthattransmitsmaximallyat

365 nm. Upon striking the chromatic beam splitter in this example all wavelengths below 420 nm are reflected to the top surface of the

specimen through the objective, causing it to fluoresce at a longer wavelength. Both the excitation (by reflection) and fluorescent

wavelength are generated at the specimen surface and are again collected by the objective (b). However, only fluorescent wave-

lengths longer than 420 nm are now transmitted to the detector (e.g., the eye) by the beam splitter, which acts as a barrier filter. In most

instruments, an additional barrier filter is placed between the beam splitter and the detector to define further the range of fluorescence

wavelengths to be examined. Reproduced from Fulcher RG, Irwing DW and Francisco A de (1989) Fluorescence microscopy:

Applications in food analysis. In: Munck L (ed.) Fluorescence Analysis in Foods. Harlow: Longman, with permission.

SPECTROSCOPY/Fluorescence 5435

the latter instrument may be compromised by the

greater amount of light that can pass through the

filters of a filter fluorometer. Fluorescence thin-layer

chromatography (TLC) scanners, which are special-

ized (epi)spectrofluorometers, are often convenient

for studying, for example, bones in fish fillets and

connective tissue in minced meat. Filter and spectro-

fluorometry may be exploited to provide highly sen-

sitive detection of specific substances in high-pressure

liquid chromatography (HPLC) and flow-injection

analysis (FIA). (See Chromatography: Thin-layer

Chromatography; High-performance Liquid Chro-

matography.)

Applications

Monitoring Structure and Function in Food Systems

0019 In many applications, the evaluation of structural

(physical) properties of a food material accompanies

determination of its (bio)chemical characteristics. In

the case of wheat grain, it becomes readily apparent

that the bran tissues (pericarp and aleurone;

Figure 7:2:1 and 3) have yellow and blue fluores-

cence, respectively, at specific excitation wavelengths.

A method can then be developed to quantify fluores-

cence on the basis of excitation and emission spectra

obtained with suspensions of finely milled hand-

dissected samples of the pure botanical tissue com-

ponents (Figure 2). An additional finding will be that

the floury endosperm part of the seed autofluoresces

in the UV region, which cannot be detected by the eye

in the fluorescence microscope. This information

could be used tentatively to monitor the fractionation

process in a wheat mill by spectrofluorometry much

more accurately than with the present ash and color

analyses. It is also possible, for example, to optimize

the milling process to obtain an increased content of

the vitamin- and mineral-rich bran component, aleur-

one, in the flour, while still obtaining a relatively

white bread with good baking volume despite a high

ash content.

Wavelenght (nm)

(a)

Transmittance

400

0.2

0.4

0.6

0.8

1.0

500 600

Wavelenght (nm)

400

0.2

0.4

0.6

0.8

1.0

500 600

Wavelength (nm)

(c)

Transmittance

400

KP 425 KP 490 KP 560

FT 395

FT 425

FT 460

FT 510

FT 580

KP 500

0.2

0.4

0.6

0.8

1.0

500 600

Wavelength (nm)

(d)

400

0.2

0.4

0.6

0.8

1.0

500 600

(b)

G 365

BP 400−440

LP 420

LP 470

fig0005 Figure 5 (a, b) Filters. Typical spectral transmission curves of paired exciter (e) and barrier (b) filters designed to optimize

fluorescence by an HBO illumination. G 365 is a colored glass filter while the other filters displayed in the figure are interference

filters coated by vacuum evaporation of metallic oxides. The latter can be specified more flexibly, but are also more expensive. A

bandpass filter (BP 400–440) shows sharp cut-offs on each side of the curve (b). Short-pass (KP) interference filters (c) are used as

exciting filters. They suppress long-wavelength light, and show a steep cut-off at the high limit of transmittance. Barrier filters may also

be of the BP type but are mostly long-pass (LP) filters (d) showing steep cut-offs at the lower limit of transmittance and uniform

transmittance throughout the rest of the visible range, making it possible to detect a range of emission wavelengths with the same

barrier filter. Filters are examined in a high-quality spectrophotometer. Reproduced from Fulcher RG, Irwing DW and Francisco A de

(1989) Fluorescence microscopy: Applications in food analysis. In: Munck L (ed.) Fluorescence Analysis in Foods. Harlow: Longman, with

permission.

5436 SPECTROSCOPY/Fluorescence