Todd Frank Director of Manufacturing MicroTracers, Inc' San Francisco, Ca' USA

1
Todd Frank Director of Manufacturing
Micro-Tracers, Inc. San Francisco, Ca. USA
ELANCO SYMPOSIUM
Medellin, Columbia
23 July 2009
2
ELANCO SYMPOSIUM

• Microtracer FS-Red/Natural Yellow

Used by ELANCO to Code
RUMENSIN
3
Micro-Tracers,Inc. a Qualified Supplier

• Founded 1961, operated under same management.
• Financially stable, committed to research.
• Associated with Anresco,Inc. (1943) commercial
  analytical laboratory.

4
What Are Microtracers ?

• Many Microtracer variations. First came
  Microtracers S colored salt then G colored
  graphite.
• Microtracers F, FS and RF are colored iron grits,
  stainless steel grits and reduced elemental iron
  powder.

5
History of Micro-Tracers Inc

• Founder (1961)Dr. Sylvan Eisenberg

• President, Mr. David Eisenberg, MBA

6
MICROTRACERS
7
What Is FS-Red/Natural Yellow?

• Stainless steel particles
• Coated with food grade dye.
• Same size.
• Guaranteed particle number per gram.

8
Quality of Microtracers

• Sieve Analysis.
• Particle Counts.
• Color Analysis.
• Periodic testing for dioxins (none found) and for
  arsenic.

9
The Use of Microtracers in Rumensin
Coding Rumensin and Feeds Containing it as
Proprietary
Providing a Value Added Quick Test for
Rumensin In Feeds to be used by Feed
Manufacturers and Elanco Technical Support
Personnel.
10
Coding Rumensin as Proprietary
Qualitative Test Mash feed one minute
Pelleted feed two minutes
Quantitative Test Counting colored spots
Some skill required 5 minutes
Test Rumensin and Third party pre-mixes first.
11
Coding Rumensin as Proprietary As Value Added
Benefit to Feed Manufacturers
Allow immediate confirmation Rumensin has been
added to feed.
Permit validation of mixer performance by
taking/analyzing a series of samples from one
batch-or one sample from each on many batches
Permit validation of cross contamination control
procedures
12
Coding Rumensin as Proprietary
Microtracer FS-Red Natural Yellow is Exclusive to
Elanco
Stable in Rumensin premix
Not with Confused by other Microtracers
13
Coding Rumensin as Proprietary Quantitative Use
Preciseness of Test Requires Counting More
Particles
Poisson statistics Count has inherent standard
Deviation equal to its square root
100 particles 10 Standard
Deviation 10
Coefficient of Variation
25 particles 5 Standard Deviation
20 Coefficient Variation
14
Analytical Procedure

• Mason Jar Technique.
• Qualitative.
• Rotary Detector Technique.
• Qualitative and Quantitative.

15
C-1A/ Mason Jar TechniqueMason Jar Technique.

• Materials
• 500 ml Mason Jar
• Scale
• Developing solution
• Filter paper (70 mm)
• Grinder for pelleted feed.

• Test Method
• Weigh 100 g sample
• Insert filter paper into the magnetic cap and
  close jar.
• Shake jar so sample contacts filter paper
• Spray developing Solution and identify the color
  of the Microtracer Spots.

16
Mason Jar Technique
17
Mason Jar Technique
18
Mason Jar Technique
19
Mason Jar Technique
20
Mason Jar Technique
21
Rotary Detector Technique.

• Materials
• Rotary Detector.
• Scale.
• Demagnetizer.
• Heating plate.
• Developing Solution.
• Filter paper 7.5 mm
• Grinder for pelleted feed

• Test Method
• Weigh 100 g sample
• Place filter paper on the spindle of the rotary
  magnet.
• Transfer the sample of feed to the top hopper of
  the Rotary
• Transfer the Microtracer to a scoop,
  demagnetize, then disperse over a large wetted
  filter paper on an aluminium plate. Then dry on
  hot plate to develop spots. Count the spots.

22
Rotary Detector Technique
23
Rotary Detector Technique
24
Rotary Detector Technique
25
Rotary Detector Technique
26
Rotary Detector Technique
27
FS-Red/ Natural Yellow Spots Developed with 50
Water and Alcohol
28
Coding Rumensin as Proprietary Quantitative Use
Errors
Tracer count not precise per lot 10 CV
Tracer recovery from mash feed in mixer 100 Mash
feed at truck 85 Pelleted feeds 70
29
Coding Rumensin as ProprietaryQuantitative
UseLosses in Feed Manufacturing
5 to each feedmill magnets
5 to abrasion in mixing
5 to dissolution during steam pelleting
5 abrasion in grinding pellets to mash for
analysis
30
ELANCO SYMPOSIUM

• Use of Microtracers in Validating Feed Mixing
• Control of Cross Contamination

31
Estimating the Level of Rumensin in Finished
Feeds from Microtracer Particle Count

• 100 grams
  500 grams

Expected Tracer Count
Feed from Mixer- 100 Tracer recovery
12 3.5 or 30 62 8 or 13
Feed from truck-mash- 85 recovery
103.2 or 32 53 7 or 14
Feed from truck- pellets- 70 recovery
82.7 or 34 43 6 or 15
Coefficient of Variation (CV) based on one
standard deviation range (67 likelihood of
actual count falling in range) For two SD range
(95 likelihood) double CV.
Qualitatively, the likelihood of obtaining no
tracer if the actual expected count is 8
particles is less than 1 in 400 test.
32
Critical Issues to Consider in Relying Upon
Microtracer Test

• The test must be run properly. You must use the
  correct developer.(50 ethanol is the correct
  solvent to develop Rumensin tracer spots.
• Run a control positive each day, a feed known
  to contain Rumensin-to confirm you find
  Microtracer from the feed you know contains it.
• The test is only as good as the sample analyzed.

33
The Use of Microtracers as a Value Added
Benefit to Feed Manufacturers

• The ability to confirm Rumensin is in a feed at a
  formulated level.
• The ability to test every truckload of
  sensitive feed (horse) before it leaves the
  feedmill to be as certain as possible Rumensin is
  not present.
• If the feed manufacturers keeps a retained sample
  of each truckload of feed, these can be tested if
  a feed manufacturing error is suspected of having
  occurred.
• The ability to validate the complete mixing of
  Rumensin into final feed and to validate
  cross-contamination control procedures.

34
Basic Premises of Formula Feed Manufacturing-

• Complete mixing of feed is good.
• Dangerous cross-contamination of feeds is bad.

35
Official Status for validating mixer performance
and cross-contamination validation

• 1. Colored iron particles are included in GPM
  test method (Holland, 2006)
• Colored iron powder is included in Standard of
  American Society of Agricultural Engineers (USA,
  2007)

36
Basic Issues with Mixer Evaluation

• 1. How often to test?
• How many samples to take?

37
How does one know if mixing is complete and
equipment cleanout adequate?

• 1. Initial feedmill design .
• 2. Testing at startup.
• 3. Developing manufacturing procedures.
• 4. Periodic testing.

38
Optimizing Mixing.

• 1. Mixing time.
• 2. Batch size.
• 3. Mixer speed.
• 4. Particle size of ingredients.

39
Validating Mixing.

• 1. Selection of tracer.
• 2. Addition of tracer to the batch.
• 3. Sampling.
• 4. Analyzing the samples.
• 5. Interpreting the test results.

40
Selection of tracer.

• 1. Tracer from one source.
• 2. Tracer should be microingredient.
• 3. Analytical procedure should be accurate and
  reliable.
• 4. Results quickly available.
• 5. Results must be interpretable.

41
Commonly used tracers.

• 1. Salt.
• 2. Minerals.
• 3. Vitamins, drugs and amino acids.
• 4. Methyl violet.
• 5. Colored iron particulates and colored iron
  powder.

42
Tracer Addition to Feed

• 1. Tracer Should be Premixed.
• 2. Location of Tracer Addition Will Depend Upon
  Purpose of Test and Practicalities.

43
Sampling.

• 1. Best- from the mixer.
• 2. Often from screw conveyer exiting surge bin.
• 3. Samples must be grab.
• 4. How many samples?

44
Analyzing the samples.

• 1. Depends upon tracer.
• 2. Vitamins, minerals, drugs, amino acids or salt
  require laboratory analysis.
• 3. Salt analysis by Quantab test strips and
  colored iron particles -analysis at the feedmill.
  

45
Interpreting Test Data.

• Complete Mixing- When CV for series of samples
  equals CV for repeat analysis of one sample.

46
Interpreting Test Data- contd.

• 1. Typical CVs -drug assays 20, salt- 3,
  minerals- 7, methyl violet or colored iron
  powder- 5.
• 2. CV for particle counts defined by Poisson and
  Chi-Squared statistics.

47
Poisson Distribution

• The poisson Distribution describes the
  probability that a specified count will occur as
  a function of the average count and the number of
  such counts. The distribution assumes that counts
  are integral and independent of each other
• At one time there is only one event.
• Each event is independent of others (no influence
  between each other).
• The events are independent (not influenced by a
  number of events happening in the past)
• The distribution is stationary (the mean event
  rate do not change with time)

48
E. Interpretation of Results.

• Mathematical sizes
• Micro tracer count x1, x2, x3
• Number of repeated experiments n
• Mean X
• Difference between single event and mean d1, d2,
  d3
• Sum of differences squared d1²d2²d3²S
• Chi-Square ?2 S X
• (see ?2 table)

• Determination of number of independent elements
  x-2n
• Finding of probability p for n and ?2
• Homogeneous mixture P gt 0.05
• Mixing is marginal 0.01lt P lt 0.05
• Non-homogeneous mixture P lt 0.01

49
Interpreting Particle Counts as evidence of
mixing.

• 1.Poisson Complete Mix- series of analyses
  -standard deviation (sd) square root of average
  count. Average of 100 then sd 10 and CV
  10/100 or 10.
• 2. Average 25 then sd 5 and CV 5/25 or 20.
• 3. Average 1 then sd 1 and CV 1/1 or 100.

50
Experimental Design for Mixer Tests.

• Guidline of Authorities Mixing homogeneity
  1 100 000
• Example
• 10 g / t of Feed
• 25 000 particles / gram of tracer F
• 250 000 particles / ton of Feed
• 250 particles / kg of Feed
• 50 particles / 200 grams of Feed

51
Examples of Mixer Study Homogeneous Mixture

• Samples Value Mean x n- d n ( x n-
  d n) ²
• 1 47 50 3
  9
• 2 57 50 7
  49
• 3 45 50 5
  25
• 4 55 50 5
  25
• 5 50 50 0
  0
• Mean x 50 Sum d n² S 108
• Number of repeated experiments n5
• Chi-Square ?2 S X 2 (108 50 2)
• Table
  Horizontal n 23
• Vertical ?2 2
• Probability P
  0.572

52
Table for determination of probability,
horizontal number of degrees of freedom,
vertical chi squared values
53
Interpreting Particle Counts as Evidence of
Mixing- contd.

• 1. When study CV exceeds Poisson- Chi-Squared
  Statistics employed.
• 2. If CV 13 when 10 predicted and 50 samples
  analyzed- Chi-Squared Likelihood less than 1.
  Mixing is Incomplete.

54
Non-homogeneous Mixture

• Samples Value Mean x n- d n ( x
  n- d n) ²
• 1 43 53 10 100
• 2 57 53 4 16
• 3 70 53 17 289
• 4 35 53 18 324
• 5 61 53 8
  64
• Mean x 53 Sum d n²S 793
• Number of repeated experiments n5
• Chi-Square ?2 S X 15 (793 53 15)
• Table
  Horizontal n 2 3
• Vertical ?2 15
  
• Probability P
  0.002

55
Study Where 15 Tracers Employed- two mills
expected to have complete mixing and one mill
expected to have incomplete mixing.

• 1. 12 of 14 tracers correctly identified problem
  at mill 3.
• 2. 3 external tracers all correct.
• 3. Fat and magnesium results in error.
• 4. Analytical variances greater than expected.

56
Mixing Data from Study Where Many Tracers
Employed- 3 feedmills
57
Mixing Data from Study Where Many Tracers
Employed- 3 feedmills (cont.)
58
Mixing Data from Study Where Many Tracers
Employed- 3 feedmills (cont.)
59
Problems in Determining Cross-Contamination of
Feeds.

• 1. Analytical methods for ruminant by-products in
  feeds not adequate.
• 2. Analytical methods for drugs in feeds at 1
  formulated levels not adequate.
• 3. Analytical methods for drugs in meat, poultry
  and fish are more sensitive.

60
Study of Cross-Contamination of Amprolium Using
Colored Iron Particles and Powder.

• 1. Colored Iron Particles and Colored Iron Powder
  Formulated at 1-kilo per tonne of Amprolium 2.5
  Premix.
• 2. Samples Taken from Premix, 3 Succeeding
  Batches and Pellets from Combined Batches.
• 3. Amprolium Determined Chemically and Colored
  Iron Particles and Colored Iron Powder Determined
  by tracer tests.

61
Premix Plant Cross-Contamination Study
62
Premix Plant Cross-Contamination Study (cont.)
63
Premix Plant Cross-Contamination Study-
Conclusions

• 1. The level of detection (LOD) for the amprolium
  assay was 50 ppm or 0.2 the formulated level.
• 2. The LOD for the colored iron particles and
  colored iron powder was 0.02.

64
Premix Plant Cross-Contamination Study-
Conclusions- continued.

• 3. Below 500ppm Amprolium chemical assays
  yielded higher results than tracer results.
• 4. The amprolium was powdered and the iron
  particles granulated.
• 5. Cross-contamination higher for powders.

65
QUESTIONS?

• Please e-mail at
• tfrankmt_at_sbcglobal.net
• or
• MICROTRACE_at_aol.com
• THANK YOU