oil extraction and analysis phần 9 pps

This decrease also will result in a lower estimate of total oil content because hydrogen is used as a reporter in the NMR measurement.. Hahn Echo pulse sequence: RF pulse 1{---delay ---}

Chapter 9

Simple Methods for Measuring Total Oil Content

by Benchtop NMR

P.H Krygsman and A.E Barrett

Bruker Optics Ltd., Minispec Division, 555 Steeles Ave., East Milton, Ontario L9T 1Y6, Canada

Abstract

Low-resolution time domain nuclear magnetic resonance (TD-NMR) is an interna-tionally recognized and commercially important analytical tool for measuring the oil content in oilseeds and oilseed residues It is commonly employed as a simulta-neous determination of oil and moisture content, as in ISO 10565 or AOCS Ak

4-95 Because the method is very fast, compared with oven drying to determine moisture and extraction to determine oil, rapid and more frequent testing is possi-ble, and dry-weight oil is easily reported The NMR method is widely used in European countries, especially France and Germany, for oil in sunflower, rape, and soybean A related method used in Spain is oil in olives for assigning commercial value In North America and South America, official oil content in canola and sun-flower is determined by the NMR method All other oilseeds are amenable to the method, including soy, flax, corn, cotton, and peanut A large part of the reason for the success of the method is the simplicity of calibration In cases in which rapid and accurate screening of oil content is the primary concern, simple calibration maintenance is desired Calibration for oil can involve as few as two or three refer-ence oilseed samples Often 5–20 calibration samples are used to satisfy statistical variation in the natural product Repeatability of measurement is usually limited by subsampling error If the same sample is measured repeatedly, the SD is typically

≤0.02 Agreement between duplicate samples taken from the same lot depends on the oilseed and size of sample used It was found to be ~0.1–0.2 for soybean mea-sured in ~22-g samples TD-NMR measures oil by detecting the hydrogen in the liquid phase of the sample The NMR signal for oil is well isolated and normally free of interference from other components of the oilseed; under normal condi-tions, the signal per gram increases in a simply linear fashion with increasing oil content Moisture present in the seed <10% by weight (14% in soy) is strongly associated with solid components, and relaxes at least an order of magnitude faster than oil If excess water is present in the sample, it must be removed by a predry-ing step Water in various degrees of “free” states may relax with a time constant similar to that of oil, or even longer Vegetable oils contain a mixture of fatty acids

in triacylglycerides, and the fatty acid profile can vary considerably If unsaturation

is introduced, the hydrogen content (%) decreases This decrease also will result in

a lower estimate of total oil content because hydrogen is used as a reporter in the NMR measurement The hydrogen content (%) can easily be determined in isolated oil (extracted or expressed) to ensure that it fits within an acceptable range

Introduction

Oilseed characterization naturally requires measurement of total oil content Traditionally, this involved solvent extraction, but many have been motivated by concern over safety and the need for better analysis speed to seek alternatives

Spectroscopy is a natural place to look, especially if oil can be measured in situ.

One of the most successful spectroscopic methods for oil in oilseeds is low-resolu-tion benchtop nuclear magnetic resonance (NMR) Today, benchtop NMR analyz-ers are commercially important analytical tools for measuring the oil content in oilseeds and oilseed residues (1–3) Low-resolution NMR provides rapid and rou-tine analyses, with simple calibration based on a primary method Because it is nondestructive, samples can be retained to be measured again, used for other

analyses, or for planting in field studies Analysis is via direct spectroscopic

detec-tion of oil without reagents or solvents, making it environmentally friendly, safe for the operator, and low-cost in terms of consumables and disposal fees NMR detects oil in the whole sample at once and there is no need to adjust the calibration for surface effects such as color or texture This makes it an ideal tool for small samples encountered in some seed breeding programs

Official methods exist for two forms of low-resolution NMR Continuous wave NMR (CW-NMR) is used for measurement of oil content alone in dried seeds (4,5) Pulsed time domain NMR (TD-NMR) is used for simultaneous deter-mination of oil and moisture in seeds with as-is moisture content (6–8) Of the two NMR technologies, TD-NMR is the most recent In fact, CW-NMR equipment is

no longer available, having been replaced by pulsed NMR technology

These methods have been validated and organized into official methods (see

Table 9.1) The ISO methods are widely used in European countries especially France and Germany for oil in sunflower, rape, and soybean A related method is used in Spain for assigning commercial value to olives In North America and South America, the official oil content in canola and sunflower is determined by the NMR method All other oilseeds are amenable to the method, including soy, flax, and corn, cotton, and peanut

Oil is stored in the seed in structures called oleosomes or sometimes deposited

as droplets in the cytoplasm Vegetable oil is present largely in the form of triacyl-glycerides (TAG), whose chemical structure is made up of carbon, oxygen, and hydrogen (Fig 9.1)

NMR analyzers detect the signal from receptive atomic nuclei when the sample

is placed in a magnetic field Only those atoms with appropriate nuclear

character-istics are detected under the particular conditions used, i.e., depending specifically

on the magnetic field and radio frequency (RF) used In the case of oilseed mea-surements, hydrogen is used as the reporter of oil content Detection of carbon or oxygen is impractical for a low magnetic field application because of poor sensitiv-ity

Instrumentation

The choice of appropriate NMR hardware is important when initiating an oilseed measurement program Issues to consider include seed size, oil content consistency from seed to seed, and sample availability For bulk screening of oilseeds, the instrument of choice is one that can accommodate a sufficiently large sample to allow a statistically accurate result to be obtained, when combined with proper sampling and reduction For ISO and AOCS methods, sample size is typically 50

mL in 40-mm diameter tubes (Fig 9.2) (7,8,10,11) Bulk screening of small oil-seeds such as flax, canola, and mustard can be done with 18-mm sample tubes, which accommodate ~10 mL, or in 25-mm sample tubes, which accommodate ~20

mL USDA GIPSA mandates that sunflower samples be measured in 50-mm sample tubes so that sample volumes of ~110 mL can be accommodated (9)

Single seeds are measured in some breeding programs If the task is to find the highest oil-producing plant or seed, the best results are obtained with a small diameter sample compartment (e.g., 13 or 18 mm) For example, single peanuts can be

TABLE 9.1

Official Methods for Oil Content Determination by Low Resolution NMR

Oilseeds Oil: AOCS Recommended CW-NMR 5

Practice Ak 3-94: 1999, 2000 Oilseeds Oil and Moisture; ISO 10565 Pulsed NMR 7 Oilseeds Oil and Moisture: AOCS Recommended Pulsed NMR 8

Practice Ak 4-95 Oilseeds Oil: USDA GIPSA Certificate Pulsed NMR 9

No FGIS00-101 Oilseed residues Oil and Moisture: ISO 10632 Pulsed NMR 10 Oilseed residues Oil and Moisture: AOCS Ak 5-01 Pulsed NMR 11

Fig 9.1 The chemical structure of a

triglyceride.

accommodated in 18- or 25-mm tubes and measured at 10 or 20 MHz, but single canola or flax seeds should be measured in a 60-MHz system to obtain an adequate signal-to-noise ratio Standard tube sizes at 60 MHz are 5 or 7.5 mm in diameter

Experimental

In the absence of free water, the NMR experiment used to determine total oil is the Hahn Echo (12), or spin echo Common terminology for a pulsed NMR experiment

is a pulse sequence, because the experiment is actually a series of events including

RF pulses, evolution periods, and detection of RF signals over time.

Hahn Echo pulse sequence:

RF pulse 1{ -delay -}RF pulse 2 { -delay -}Digitize{ -delay -}

where

RF pulse 1 causes a 90° shift in the sample magnetization into the detector plane

RF pulse 2 causes a 180° shift in the sample magnetization within the detector

plane

Requirements of the method include:

• A liquid state of the oil in the presence of other nonliquid seed components

• Hydrogen as a reporter of oil content, in proportion to the amount of oil

• Moisture content at or below the normal preservation levels for the seed Measurements are made on whole seed and are totally nondestructive The sample does not need to be ground prior to measurement A typical procedure is as follows:

1 Weigh sample into sample tube

Fig 9.2 Sample tubes used

in the bulk screening of oilseeds.

2 Insert into the instrument; measurement starts automatically

3 After 30 s, result is reported, and the sample can be removed

In some cases, it is necessary to preheat samples to liquefy fat, e.g., cocoa beans, which contain a high percentage of saturated fat that is semisolid at room temperature, or when laboratory temperatures vary >1–2°

If the moisture content is above the recommended upper limit (6–8), the oilseed can be weighed as is, then dried for 3 h and reweighed Weight loss in a predrying step is used to quantify the partial moisture content; the remainder is determined by NMR

Total moisture = W b + [(m – m 0 )/m] × 100

where

m is the original mass (all moisture included)

m 0is the mass of partially dried sample tested by NMR

W bis the moisture determined by NMR in the partially dried sample

Preparation of the oil calibration is quite simple compared with other sec-ondary spectroscopic techniques, e.g., near infrared It is recommended that at least 3–5 well-characterized reference samples be collected Often 5–20 calibration sam-ples are used to satisfy statistical variation in the natural product The calibration for oil in sunflower uses 2 well-characterized samples (9) A linear calibration line

is constructed for oil content by measuring the NMR signal at the echo, normalizing

by sample mass, and plotting vs the oil reference value The calibration equation for oil is as follows:

A21/m1= sx1+ i

where

A21is the NMR signal at the echo for sample 1

m1is the mass of sample 1

x1is the reference oil value for sample 1

s is the slope of the calibration line

i is the intercept of the calibration line

Moisture content by NMR is proportional to the difference between the free induction decay (FID) intensity at time ~60 ms and the echo amplitude due to oil The calibration equation for moisture is as follows:

(A11– A21)/m1= sx1+ i

where for moisture

A11 is the NMR signal at the beginning of the decay for sample 1 (past the solids decay)

A21 is the NMR signal at the echo for sample 1

m1 is the mass of sample 1

x1 is the reference oil value for sample 1

s is the slope of the calibration line

i is the intercept of the calibration line

Oil calibration reference values can be determined by an extraction reference method (13,14, or for oilseed residues, 15,16) Reference samples should be of the same species as the test samples, and have a similar fatty acid profile (especially for canola and sunflower)

Moisture calibration reference values are more difficult to maintain than oil because the moisture content changes with exposure to atmospheric humidity It is necessary to maintain samples in a sealed container or determine the moisture con-tent in the reference samples just before use as a calibration standard This is nor-mally done by a reference drying method, e.g., drying oven maintained at 103°C It has been noted for canola that as moisture falls below 4 or 5%, the moisture cali-bration is no longer linear (personal communication, Ken Howard, Canadian Grain Commission, 2003) Therefore, a separate calibration line constructed using seed that contains <5% moisture is necessary for canola The most likely reason for nonlinearity at the low end is the tighter association of moisture with the solid matrix of the remaining moisture, resulting in faster NMR signal decay It is possi-ble that a saturation fit can produce a single calibration or the calibration might be linearized by setting the first sample window earlier

The moisture limit may vary considerably by oilseed, with some having much higher moisture levels than indicated in the ISO method According to checks by the reference methods, the NMR method still yields the correct oil and moisture results for canola containing 18% moisture, and corn or sunflower containing 35% moisture (personal communication, Ken Howard, Canadian Grain Commission, 2003)

Results and Discussion

TD-NMR spectroscopy discriminates oil from other components of the seed by relax-ation time differences Hydrogen present in the oil has a significantly slower decay than hydrogen in other constituents of the mature oilseed NMR relaxation time for oil protons in a liquid environment is on the order of hundreds of milliseconds In particu-lar, vegetable oil relaxation time is on the order of 100–200 ms In an oilseed, oil relax-ation time is much the same as that of the extracted oil (Fig 9.3)

Moisture present in mature seeds at preservation levels of dryness is essentially bound (which is why seeds do not spoil) It is associated with polar surfaces of

large molecular weight, solidified carbohydrate, and protein In this state, water protons have a T2 relaxation on the order of 1 ms All other solids present in the seed have relaxation times shorter than bound moisture

For CW-NMR, the T2 relaxation characteristics translate into a relatively sharp peak in the NMR spectrum due to oil, superimposed (due to lack of resolution) on a broad feature due to solids and bound moisture Therefore, the CW-NMR signal inher-ently has the amplitude due to bound moisture and solids included In the case of TD-NMR, the signal is analyzed in the time domain, and the characteristic results in a slower decay component, which can be evaluated at a point in time when only the liq-uid signal remains The TD-NMR signal from oilseeds, measured by the Hahn Echo pulse sequence, typically appears as in Figure 9.4

Liquid (free) water T2 relaxation is on the order of 1000–3000 ms, approxi-mately an order of magnitude longer than oil If a seed is physically wet, free water can be detected in the signal in the form of a long relaxation time component, longer than 200 ms One complication of resolving water from other components

by relaxation time is that water can have several physical states with very different relaxation times As the hydration level increases from dry seed (<10 or 14% mois-ture) through intermediate levels (15–20%) and then to wet, some of the moisture can take on relaxation properties intermediate between bound and free In this case,

it is difficult to distinguish water from oil in the same relaxation range As a result, the international methods have been written for low-moisture or dry seeds

Repeatability of the instrument itself can be determined readily because the sam-ple is not consumed by the test Repeat measurement of the percentage of hydrogen

Fig 9.3 Comparison of NMR relaxation times for canola oil Refined commercial

brand canola oil (biexponential: 59% has T2= 105 ms, 41% at 326 ms); Canola oil in seed (biexponential: 56% has T2= 86 ms, 44% has T2= 276 ms); and reference

mate-rial n-dodecane (T2= 1340 ms).

Oil NMR Signal Decay by T2-Relaxation

Dodecane Canola Oil Canola Seed

Time (ms)

using the same dodecane reference sample (theoretical %H = 15.385) gave an average of 15.389% and SD of 0.0021 for 7 measurements (see Table 9.2) Therefore, the NMR method is capable of excellent precision

Similar repeatability tests for oil content in soybean samples are presented in

Table 9.3 Three samples with known oil content were measured three times The average of the results agreed well with the extraction values, and the SD for all sam-ples was ≤0.02 Moisture results for soybeans had similar precision (see Table 9.4)

Fig 9.4 TD-NMR signal from typical oilseed: 41% oil, 25% protein, 4.3% moisture.

Upper panel: Hahn Echo experiment showing regions where signals are sampled for moisture and oil Lower panel: Expansion of the first 0.5 ms, showing fast decay

peri-od due to solids followed by slower decay due to moisture and oil.

TABLE 9.2

Repeatability and Accuracy of TD-NMR for the Percentage of Hydrogen in Dodecane (15.385 %H)a

Mean amplitude Mean amplitude Measured Sample sample reference reference (%H) (%H)

aResults using the Minispec mq10 (Bruker, Rheinstetten, Germany).

TABLE 9.3

Repeatability and Accuracy of TD-NMR for the Percentage of Oil in Soybeans

Analyzed weight (g) 21.4200 21.2784 21.6552 NMR % oil 1 20.01 19.66 20.07 NMR % oil 2 19.99 19.68 20.08 NMR % oil 3 20.03 19.69 20.09 NMR average % 20.01 19.68 20.08 Given oil 19.86 ± 0.07 19.04 ± 0.11 19.82 ± 0.14

TABLE 9.4

Repeatability and Accuracy of TD-NMR for the Percentage of Moisture in Soybeans

Analyzed weight (g) 21.4200 21.2784 21.6552 NMR % moisture 1 10.64 12.33 10.87 NMR % moisture 2 10.67 12.38 10.87 NMR % moisture 3 10.67 12.34 10.86 NMR average % 10.66 12.35 10.87 Given moisture 10.45 ± 0.01 12.294 ± 0.03 11.16 ± 0.14

Agreement with the reference method was excellent, and the SD was ≤0.03 at 10–12% moisture

Repeatability considering intersample variation can be tested by taking sepa-rate samples (A and B) from a batch of seeds Results for three seed types are shown in Table 9.5 Agreement using 20-g sample size (40-mm diameter sample tube) was very good among separate subsamples (largest difference was 0.18 for oil, and 0.13 for moisture), indicating that this sample size is adequate for deter-mining accurate oil and moisture values for soybeans

The NMR experiment permits very stable results, even when the seed is dam-aged or discolored The signal from the whole seed is measured equally; therefore, the distribution of oil in the seed cross section, or presentation of the seed, matters little

The fatty acid profile is a factor to be considered in the NMR calibration (17) Fatty acids present in oilseeds differ significantly in the percentage of hydrogen and in iodine value (refer to Table 9.6) As a result, oil in oilseeds can vary some-what in the percentage of hydrogen, depending on the fatty acid profile How then can the NMR method work as well as it does? The answer is that many oilseed varieties end up having a very similar total percentage of hydrogen even when the fatty acid profile and iodine value differ Refer to Tables 9.6 and 9.7 for the case of canola and mustard varieties from the Canadian Grain Commission 2002 crop sur-veys (18,19) The canola varieties listed have a percentage of hydrogen close to 11.6 or 11.7% Rapeseed oil differs the most from the canola varieties If measured

against a calibration generated for the canola B juncea, the result would read high

by ~2% of the oil reading

Some seed varieties clearly require a separate calibration for oil content, based

on extraction results from the seed variety to be measured The percentage of hydrogen is one important factor to consider when choosing a calibration grouping Ideally, the percentage of hydrogen would be determined simultaneously with total oil determination by NMR to account for the lower percentage of hydrogen However, this is not feasible with the speed desired for screening a large volume of seed The usual solution is to spot check the NMR results by the reference method

TABLE 9.5

Sampling Repeatability of TD-NMR for Oil and Moisture in Soybeans

Average % oil (3×) Average % moisture (3×)