NIR Spectroscopy in Animal Food Manufacturing

NIR Spectroscopy in Animal Food Manufacturing & Technological Advances

Introduction

NIR spectroscopy has a long history dating back to the discovery of the first non-visible part of the electromagnetic spectrum by Sir William Frederick Herschel in 1800. Herschel was testing different filters to pass sunlight through and noticed that filters of different colors generated varying amounts of heat. Further experiments entailed passing light through a prism to measure the different colors of light and in the process, Herschel took a measurement just beyond the red end of the visible spectrum. He detected a temperature one degree higher than that of red light and concluded that there was an invisible form of light beyond the visible spectrum. At the time, Herschel referred to this region as “calorific rays” and it was later renamed “infrared”.

Until the 1950s, the use of NIR spectroscopy as an analytical technique was dormant and it took a backseat to other analytical methods. Mid-IR spectroscopy gained acceptance in the 1900s after the work of William Coblentz, who was the first researcher to obtain absorbance spectra of pure substances and verify their usefulness for the identification of organic functional groups. NIR spectroscopy was neglected for many years because spectroscopists could not find any attractive information in the NIR region which was occupied by broad, superimposed, and weak absorption bands.

The work of Karl Norris at the USDA Agricultural Research Center pioneered the use of NIR spectroscopy as a modern analytical technique. Norris built his own spectrometer using tilting filters and figured out that by using a computer with the spectral data as the independent variable, predicting the composition of an agricultural material was possible by calibrating the instrument with the laboratory determined composition as the dependent variable.

A groundbreaking paper published in 1965 by Norris paved the way for the adoption of a two wavelength model which could abandon Beer’s Law as a prerequisite for quantitative analysis. The highly scattered transmission of NIR light prevents Beer’s Law from being an effective equation for analysis. The same paper also pioneered the use of diffuse reflectance as a non-destructive technique in the NIR region, allowing for direct measurement of samples with minimal preparation and without any pre-treatment. Ultimately in the 1970s, Norris developed the first commercial NIR analyzers for measuring protein in wheat and moisture in flour.

After the door was opened, the 1980s saw rapid advancement in the use of NIR spectroscopy as an analytical technique. Developments in instrument improvement, data treatment and processing, and the use of chemometrics for calibration models (largely dependent on advancements in computers) paved the way for widespread use of NIR spectroscopy in numerous industries.

Agriculture, Food & Beverage, Pharmaceuticals, Chemical, Petrochemical, and Polymers are just some of the industries where NIR spectroscopy emerged as an important analytical technique. The advantages of being fast, non-destructive, and non-invasive with strong light penetration and requiring minimal sample preparation have made NIR spectroscopy the analytical technique of choice for many applications. It is especially well suited for measuring small amounts of water and C-H bonds.

After the initial boom in the 1980s and 1990s in the development of NIR spectroscopy, the technique was prominent but primarily used for benchtop and laboratory measurements. Technological advances in hardware and software have helped moved the focus on NIR spectroscopy from use as an offline analytical tool to an inline method that can be used for process control.

The Process Analytical Technology (PAT) initiative by the FDA created the framework for designing systems that can be used analyzing and controlling an entire manufacturing process. While formal implementation of this initiative has been limited outside of the pharmaceutical industry, the principles of PAT and benefits of real-time analysis are recognized. They have long been used in the petrochemical and chemical industries and have been implemented at many leading human and animal food manufacturers.

Some of the advancements that have helped advance the use of NIR spectroscopy as a real-time analytical tool include improved fiber optics and probes, multiplexed instruments, pre-built calibrations, instrument portability, and web-based platforms that support database management, trend and quality control analysis, and real-time virtual instrument and software maintenance and support.

In the sections below, some real examples of the use of NIR spectroscopy in animal feed manufacturing are discussed as well as detailed analysis of recent technological advancements.

Animal Feed Manufacturing and NIR Spectroscopy

A number of prominent animal food manufacturing companies have implemented NIR spectroscopy as an analytical method for both off-line quality control and as a real-time process manufacturing and control tool. The second largest pet food manufacturer in the world uses NIR spectroscopy for analysis of incoming raw materials while still in the package. Their adaptation of the technology has proven so successful that they have shared it with other companies, including other pet food companies, in the interest of health and safety.

A leading global company in aquaculture nutrition based out of Norway uses NIR spectroscopy during the manufacturing process. The company has production facilities in 18 countries on five continents and produces more than 2.5 million tons of aquaculture feed annually for more than 60 species. Every manufacturing plant worldwide uses NIR spectroscopy to provide rapid analyses of feed ingredients, intermediate products, and finished feeds. Over 600,000 samples are analyzed annually generating over three million analytical results. The company maintains an Aquaculture Research Center (ARC) that is responsible for innovations, calibration development, performance monitoring, and support to the plants through a global network.

While such companies are often guarded in the details of their analysis, the company’s website does show a graphic of a standard aquaculture manufacturing process and where NIR spectroscopy is used in the process. NIR spectroscopy is used for analysis of incoming raw materials before being placed into storage. After storage, materials are formulated, placed in dosing silos, weighed, and run through a premixer, hammermill, and mixer.

After mixing, the second NIR analysis takes place to verify that the material composition is the desired formulation before further processing. The formulated material is again stored in silos, passed through a preconditioner, extruder, and dryer and is again analyzed using NIR spectroscopy before material coating. Proper moisture content is essential after drying and the NIR analysis provides a fast and accurate reading before the material is coated and cooled. After cooling and before packing, NIR analysis is used for a final quality control check to ensure the composition meets proper standards before shipping.

The NIR team leader at the company ARC has stated “NIR is a very practical method to use for the production and quality teams at feed plants. Samples need little or no preparation and results from multiple parameters are available almost instantly. The information they produce enables formulators to ensure finished feeds match the specifications precisely and that customers receive feeds with all the benefits the research team and product developers intended.”

While the benefits of implementing NIR spectroscopy as a replacement for traditional expensive and time-consuming analytical tests are pronounced even in an off-line method, the real potential benefits are from using it as a real-time process and formulation control tool.

The Canadian Feed Research Centre (CFRC) was founded in 2014 as part of the College of Agriculture and Bioresources at the University of Saskatchewan. The mission of the CFRC is to research, develop, and commercialize new and better high value animal feeds from low value crops and co-products from bioprocessing and biofuels industries.

One of the research projects at CFRC is using an FT-NIR spectroscopy system that supports input from six probes that operate from a single instrument. The ultimate goal of the project is to use the continuously collected data to provide feedback in both mill operations and formulation.

The manager of the CFRC discussed the implementation of the probes in an interview. The first three probes in the FT-NIR system are used to measure ingredients before and after the hammer mill and at the mixer. The first probe before the hammer mill analyzes ingredients for moisture, particle size, and proximate analysis (generally defined as protein, fat, moisture, ash, and carbohydrates). The second probe after the hammer mill measures particle size and moisture. Ultimately, such real-time data could be used to adjust feed formulations. The third probe at the mixer monitors the homogeneity of the feed, giving feedback to optimize mixing time.

While blend monitoring is used in the pharmaceutical industry by manufacturers who have implemented the PAT initiative, documented applications of blend monitoring in the food industry are minimal at the present time. However, the feed industry does use NIR spectroscopy for blend analysis.

Other sensor locations include after steam flaking and before cooling, before and after conditioning of the meal, after pelleting, and after cooling. For steam flaked products, moisture and degree of gelatinization are measured. For pellet products, meal is analyzed for moisture, proximate analysis, and protein and carbohydrate profiles both before and after conditioning. Similar analysis is done after the pellet mill and after cooling as a final quality control check to ensure the product values meet standards and are uniform in composition.

The “Holy Grail” for feed formulators and animal food manufacturers has been described as developing the technology using FT-NIR spectroscopy to continuously adjust feed formulations based on real-time feedback and monitoring of the nutrient composition. While the shifting of feed formulations “on the fly” based on the real-time measuring of feed ingredient composition is not implemented just yet, the technology does exist to do so and such monitoring is considered a long-term goal for the industry.

Numerous technological advances in the use of NIR spectroscopy have been developed in recent years, such as advancements in instrument portability, pre-built calibrations, database management, trend and quality control analysis, and web-based platforms that facilitate instrument ease of use and third party support. In the section below, some of these technological advances are examined in greater detail.

Technological Advances in NIR Spectroscopy

Instrument Portability

While there has been profound progress in implementing NIR spectroscopy as a process control tool, other advancements have occurred at the opposite end of the spectrum. Similar advancements have occurred in the design and technical implementation of portable and handheld NIR spectrometers. The benefits and shortcomings of portable NIR instruments have been examined in great detail in recent years. They are specifically designed to be compact, lightweight, and easy to use in a manner that facilitates the ability to make on-site measurements, especially in the agricultural industry.

Variability in the raw materials used to make products in the feed and forage industries is a big issue in product optimization, especially when implementing feed formulation. Portable NIR spectrometers can be used in the field on-site to determine the composition of raw materials.

The first portable NIR instruments that were developed in the late 1990s had a number of limitations. They had limited wavelength range, low resolution, poor spectra reproducibility, limited optical window size, the lack of a compartment cell to measure liquids, and were required to be connected to an external computer.

The strategies to reduce the size of portable spectrometers generally are done with light splitting and the detecting components. Advances in micro NIR instrumentation have been made due to rapid progress in the use of sensing technologies such as Linear Variable Filters (LVF) and Micro-Electro-Mechanical Systems (MEMS) along with the integration of these technologies with micro-optics.

While FT-NIR spectrometers are considered to have several advantages over traditional dispersive spectrometers such as greater robustness for industrial applications, better resolution, and higher spectra reproducibility, their design is more challenging for miniaturization.

MEMS technology has facilitated the design of some commercial products that have miniaturized FT-NIR sensors. These include using MEMS chips that consist of a traditional Michelson interferometer in one product and a Fabry-Perot interferometer in another.

Portable spectrometers can be classified into two types of categories related to detectors: array detector and single detector. Single detector instruments are preferred if reduction of cost is a priority but cooling may be required. A number of products have been launched that have reduced the list price of the instrument to less than $5000 and some are true portable instruments, as they can fit into the palm of a hand. However, consumers must be aware of the cost of accessories that may accompany an instrument that appears to be cost-effective.

Most handheld and portable NIR spectrometers come with software that can perform simple data treatment and plotting, PLS regression, and prediction of unknown samples on an often limited basis. They also allow for exporting to chemometric or multivariate data analysis software for model development, but such packages can be expensive and are often cost-prohibitive for vendors who need to keep costs low.

Ultimately, the practical end user solution for implementation of a portable NIR system can involve significant additional costs regarding both software and hardware. The consumer must also consider the fact that although significant improvements have been made in the performance of portable instruments, there is no disputing the natural laws regarding light and optics. Any smaller system will get less light through it and thus, smaller optics will always result in reduced performance compared to a larger system.

In conclusion, these instruments can vary greatly in cost, size, weight, power requirements, robustness, ease-of-use, durability, accuracy, and reliability. There is no “one size fits all” instrument for different needs and applications. Evaluations of instrument performance as well as technological advances will continue in the development of handheld and portable NIR spectrometers and these compact or miniaturized instruments potentially represent a new era in NIR technology.

Pre-Built Calibrations and Ease-Of-Use

With the advent and expansion of the use of NIR spectroscopy as both a real-time process control tool, as well as a means of portable field analysis, there is a need to simplify the use of NIR analytical instrumentation as much as possible. The work required to test samples and develop calibrations using chemometrics requires highly skilled labor and scientists. In the case of feed and forage analysis, many smaller farms and feed manufacturers simply do not have the skilled workers necessary to perform this work on a large scale. This has created a need for plug-and-play solutions with minimal work required on the front end beyond the scanning of samples.

One powerful tool used to facilitate ease-of-use of NIR spectrometers is fully packaged analyzer systems that contain pre-built calibrations. Many leading vendors in the NIR spectrometer industry offer packaged software with calibrations for various feed types, such as poultry, swine, cattle, aqua, grains, and forages. Calibrations for standard proximate analysis are offered as well as for more specialized parameters, such as minerals, fibers, amino acids, and energy parameters.

Some companies have developed proprietary NIR calibration models over decades encompassing hundreds of thousands of samples and millions of spectra. Despite the vast amount of sample data that is available in pre-built calibrations, support is necessary to update models with new samples and data for specific applications and sample types. Calibration updates are especially important for natural products where the amount of variation can be great in similar sample types based on differences in climate, soil, growing conditions, and many other factors.

Such services are not limited to instrument vendors. There are large companies that offer full third party support for farms and feed manufacturers to facilitate the use of NIR spectroscopy in their product analysis. Some of the services that are offered include database management, trend analysis, quality control development, and calibration updates.

The use of web-based platforms and cloud systems have greatly facilitated the development of such technology in animal food analysis using NIR spectroscopy. Such technology has emerged to the extent that there are now portable instruments that are run using smartphone apps and customer portals.

In the section below, support services for NIR spectroscopy are examined in further detail.

Third Party Support and Web-Based Platforms

Advances in technology, IT tools, and web platforms have greatly increased the viability of using NIR spectroscopy as a real-time management tool for feed and forage analysis, especially for use in managing variability.

One example of the use of NIR spectroscopy to assess the characteristics of feed ingredients is the Precise Nutrition Evaluation (PNE) program. The program is run by a large industry player in the animal nutrition sector that has over 3,900 customers and employs nearly 2,400 people on multiple continents.

The program began in 1997 with the purchase of an NIR spectrometer to complement their amino acid analytical program which used HPLC. At the time, there was extensive research being conducted on digestible amino acids in formulating rations. The program has evolved to entail prediction models and profiles for over 35 raw materials using a network of two hundred and fifty NIR spectrometers.

Some of the raw materials include barley, corn, oats, sorghum, wheat DDGS (Dried Distiller’s Grains with Solubles), soybean meal, rapeseed, canola, sunflower meal, and rice bran. Modeled and predicted parameters include standard proximate analysis, total and digestible amino acids, energy parameters such as metabolizable energy (ME), and total and phytic phosphorus for poultry. Calibrations are updated yearly using an in vivo evaluation program that includes designed experiments for adding data.

PNE runs on a web-based program and cloud system that is designed for minimal user input, requiring only the spectra of scanned samples and uploading of spectra to the platform. After data uploading, real-time predictions of the nutritional values of raw materials are provided. Analysis of the results is presented in a clear and concise way designed to facilitate decision making. Tools include quality control and traceability parameters, analysis trending, accurate sourcing of raw materials, benchmarking of results, and comparison between suppliers locally, other countries, and worldwide for raw material analysis. The structure of the platform and vast network allows for comparison on a local and worldwide level.

The program has been designed in extensive collaboration with multiple large vendors in the NIR spectroscopy industry. One documented application that has been openly published by the PNE program is for evaluating digestible lysine as a tool for identifying drift and variation in DDGS. The variation in the nutritional parameters of DDGS as a feed material has been well documented and with its increased use in animal feed products, precise analysis of fat and protein especially in DDGS is becoming more important.

The evaluation of digestible lysine content in DDGS using the PNE programs has reduced safety margins in nutrient specifications and inclusion rate, allowing for substantial feed cost savings. The company claims an average savings of two Euro per ton of feed using this analysis. Instrument vendors also offer specific packages with similar support on a more localized scale, especially with pre-built calibrations and updates.

An example of a true third party service that implements solutions for NIR analysis in not only the animal feed industry but also many other segments of the food and beverage industries is Eurofins Quality Trait Analysis (QTA). Industries that are served by the program include beer, biodiesel, chemicals, crop protection, food, food ingredients, hemp, hops, olives, and spices.

The program operates on a subscription basis for vendors and is easy to use by design. Experts work with clients to develop a testing package for specific needs. A spectrometer is chosen from a number of vendors, installed on-site either in the laboratory or on-line, and training is provided for instrument use, validation testing, data management, and calibration model updates.

The company has an extensive calibration library that contains accurate and robust models. Experts provide support for either the use of the library or the development of new calibrations if needed. The program operates through a cloud based system with full 24/7 support for both technical support and data analysis.

By design, the QTA system handles all the technical development, configuration, management, and monitoring needed to implement NIR spectroscopy for quality control testing that facilitates such testing by non-technical personnel.

The Future of NIR Spectroscopy in Animal Feed Analysis

NIR spectroscopy has emerged as a fast, non-invasive testing method for parameters of interest in animal food quality control. It offers the advantages of little to no sample preparation, the ability to be used for large-scale testing, and is able to determine multiple parameters with a single measurement.

The use of NIR spectroscopy has evolved from research and development studies in the animal food industry to being used extensively by many large companies in the animal feed, animal forage, and pet food industries. Technological advances have facilitated the emergence of NIR spectrometers from being primarily benchtop and laboratory instruments to becoming both a real-time process control tool in the manufacturing process and an on-site field analyzer.

NIR spectroscopy can be used for analysis from the beginning to the end of the animal feed manufacturing process. It is used for raw material ID and component analysis as well as a tool for checking variation and shipment within batches.

NIR spectroscopy is an aid for feed formulation by determining variation in materials. It is used as a real-time process control tool in the feed mill. When products are finished, NIR spectroscopy can be used for the final quality control check.

The principles of PAT that have been established in pharmaceutical manufacturing provide a basis for similar quality analysis in both human and animal food products. Field instruments can account for nutritional variation in raw materials before the manufacturing process begins. Moving manufacturing analysis from offline and at-line analysis that requires sampling from the process to real-time in-line analysis gives manufacturers the tools for real-time process optimization and control.

The technological advances from web-based platforms have helped move NIR spectroscopy as an analytical tool from a more localized scale to a world-wide network where data can be analyzed in a manner that facilitates product optimization. Research and development for new technologies and products is ongoing as the benefits from using NIR spectroscopy for animal feed analysis are vast and show great potential for improving product quality and reducing costs. As the market for animal feed continues to grow in coming years, NIR spectroscopy will play an essential role in many parts of the analytical process.