Beruflich Dokumente
Kultur Dokumente
Contents
1
3.2
Conclusion
References
White Paper
White Paper
METTLERTOLEDO
Contamination (4%)
Chromatography (7%)
Columns (11%)
Integration (6%)
Collection (6%)
Instrument (8%)
Analysis (6%)
Calibration (9%)
Operator (19%)
The results from this survey indicated that the two largest sources of OOS results are sample processing followed
by human error and the most time-consuming (manual) task is sample processing (Figure 1). Even though no
follow-up survey has been published since, it appears that these results are still valid today. In fact, the instrumentation, data systems, and columns have improved significantly during the last 10 years, whilst the sample
processing has remained essentially the same. Since these other developments have saved a significant amount
of time in the workflow, it is fair to assume that the sample processing aspect of the laboratory work is likely to
account for even more than 61% of the FTE.
Figure 2 illustrates the accepted formal process for how to investigate OOS results:
Out-of-Specification Result
Laboratory Investigation
Checklist Approach: Investigation of standards used,
analytical techniques, etc.
No
Yes
Lab Error Identified ?
Retesting Error in Sampling
Procedure and Resample
No
Additional OOS
Result occur
Stop Production
White Paper
METTLERTOLEDO
White Paper
A great deal of effort is expended when a sample generates an OOS result and there have been numerous articles published on how this process should work. Obvious OOS results can take three days of work but serious
ones can take months of work to resolve. The cost can easily run into many thousands or tens of thousands of
dollars. Given the large impact that an OOS result has on a pharmaceutical company, the best course of action
should be to put every effort into avoiding them in the first place.
Besides trying to determine the root cause, the other significant issue seems to be the mounting Corrective and
Preventative Actions or CAPAs that a company generates over a number of years as a result of these laboratory
investigations. These CAPAs typically cause procedural changes to SOPs and other documents and over time
they become unmanageable and difficult to follow which has the potential to cause even more issues.
The overriding problem with CAPAs is that, in the vast majority of cases, the assumption tends to be made that
it is an isolated incident and so only a specific item in a workflow or process is addressed. In other cases, there
is a tendency to blame a single employee or a simple laboratory error. In some cases, this simple error may be
the only thing that that needs to be addressed but in many, if not most cases, it is the whole process or workflow
that needs to addressed instead of one step. This is especially true of sample preparation in the laboratory.
White Paper
METTLERTOLEDO
Figure 3: Measurement Uncertainty: Absolute (green line) and relative (blue line) measurement uncertainty of a weighing instrument. The
accuracy limit of the balance, the so-called minimum weight, is the intersection point between relative measurement uncertainty and the
required weighing accuracy.
Looking at the relative measurement uncertainty, which is the absolute measurement uncertainty divided by
the load, and usually indicated in per cent, we clearly see that the smaller the load is, the larger the relative
measurement uncertainty becomes. If you weigh at the very low end of the balances measurement range, the
relative uncertainty can become so high that the weighing result cannot be trusted anymore. It is good practice
to define accuracy (tolerance) requirements for every weighing process. For quantitative analysis this is even
stipulated by USP General Chapter <41>. Weighing in the red area as indicated in the figure will result in inaccurate measurements, as here the measurement uncertainty of the instrument is larger than the required accuracy
of the weighing process. Consequently, there is a specific accuracy limit for every weighing instrument the
so-called minimum sample weight, or short, minimum weight, and you have to weigh at least this amount of
sample in order to have a sufficiently small uncertainty that satisfies the specific weighing accuracy requirement.
While measurement uncertainty is described in much detail in the respective literature (8, 9), we want to
emphasize that for weighing small loads on analytical and microbalances and samples and standards usually are small loads as compared to the capacity of the balance the dominant contribution factor to weighing
uncertainty stems from repeatability (expressed as the standard deviation of a series of weighings). This is also
reflected in USP General Chapter <41> as discussed above.
Even though adherence to this USP requirement seems to be straightforward, many companies still have issues
with the correct interpretation. While environmental influences and operator variability which contribute to indeterminate errors and consequently to possible changes or fluctuations of the reading of a weighing device are
discussed later, another misconception which is prevalent in the industry is briefly discussed now. Many companies wrongly believe that the weight of the tare also accounts for the adherence to the minimum weight. In other
words, if the tare weighs more than the minimum weight, any sample quantity can be added and USP<41> is
automatically fulfilled. This would mean that with a large enough tare container you might believe that you could
even weigh a microgram on a 5-place balance and still comply with the uncertainty requirement of 0.1%. Such
an extreme example clearly shows us that this widespread misinterpretation indeed does not make any sense.
For this reason USP has attempted to clarify this issue in the latest revision of General Chapter <1251> (10):
The minimum weight applies to the sample weight, not to the tare or gross weight.
White Paper
METTLERTOLEDO
White Paper
Figure 4: Safety Factor: Variability of the relative measurement uncertainty due to changing environmental conditions and influences introduced by the operator. Weighing in the green area guarantees adherence to the weighing accuracy requirements (application of a safety
factor).
The safety factor describes that you would only weigh sufficiently above the minimum weight as determined at
calibration. For manual weighing, a safety factor of 2 is commonly used, provided you have reasonably stable
environmental conditions and trained operators. For very critical applications or a very unstable environment, an
even higher safety factor is recommended.
Later in this paper we will elaborate more in detail about typical minimum weight values and recommended
safety factors for automated gravimetric sample preparation as compared to manual weighing (see 5.5 How can
a gravimetric approach reduce the sample volume?).
White Paper
METTLERTOLEDO
For a typical simple sample preparation workflow, Figure 5 demonstrates the number of steps involved.
The steps are grouped into four distinct activities. The first concerns gathering materials and ensuring the equipment is clean and calibrated. There are a number of steps at the beginning that would not necessarily cause an
OOS result but would certainly feature in both GxP and safety audits. Resolving these audit issues can consume
a significant amount of time and effort and should also be avoided as they may lead to future OOS results.
The second activity involves weighing and labeling steps. Typically, these are manual time consuming operations and the weighing steps can contribute to OOS results. An additional difficulty is that these steps are manually intensive, so it can potentially be very difficult to determine the root cause of this operation.
White Paper
METTLERTOLEDO
White Paper
Following weighing, the diluent must be added. This is followed by dissolution and any successive dilutions that
may be required. The weighing, sonication and QSing steps are repeated for each standard and sample. Dilution
steps are a big potential source of errors, and consequently this activity will be discussed in more detail in the
section below.
Finally, the samples are analysed and the materials and equipment are tidied up. This requires disposal of
unused solutions, rinsing of flasks and pipettes, and other resupply steps.
Therefore, a simple sample preparation takes ten or more steps with an additional ten miscellaneous steps. If
two standards and a sample were prepared, approximately 40 steps would need to be carried out. A 40 step
process has a significant number of areas where problems could potentially occur. Furthermore, some of these
steps can be expanded and a detailed analysis with more complex steps, including operations such as extraction and filtering, might result in a process with one hundred or more steps. Given this number of manual steps
where indeterminate errors occur, it is surprising that OOS results are not even more frequent. Fortunately, many
but not all of these errors are found before the final results are obtained but they do significantly impact the productivity of the laboratory operation and the overall quality of the data.
White Paper
METTLERTOLEDO
There have been a number of OOS investigations where the analyst has forgotten to pre-rinse his volumetric
flasks. In fact, this implies that another analyst must have forgotten to perform a post rinsing operation too. The
problem created by this repeating issue is how to justify a CAPA that says retraining is addressing the problem
when in fact is doesnt. How do you know which analyst didnt do the post-rinsing, do you retrain all analysts? If
people continue to forget, what are the next steps? Do you spend hundreds of thousands of dollars on a system
to try and remove the contaminants from the glassware? Some companies have. But does it really make sense
to spend that much money on 100 year old technology?
4.2.4 Revalidation
Coleman and Harris also suggested in their paper (11) that the calibration of the glassware should be verified at
least every 10 years. This could potentially be a very expensive process considering that the number of volumetric flasks in a typical pharmaceutical analytical or QA/QC department can be very large. It could potentially be
cheaper to throw them all away and start again.
4.2.5 Tolerances
In Table 1, the published NIST relative percent errors associated with each size of volumetric glassware are listed.
In each case as the size (volume) of the glassware decreases, the error risk (tolerance) increases significantly.
Pipettes
Volume (mL)
Flasks
Relative % Error
Volume (mL)
Relative % Error
0.60
0.40
0.30
10
0.20
0.33
25
0.12
0.25
50
0.10
0.20
100
0.08
10
0.20
200
0.05
Any errors associated with the Class A glassware that does not meet the specification would be even larger, as
mentioned previously in 4.2.1 concerning the high failure rates.
Aside from the significant increase in the relative percent error, the smaller glassware is also very technique
dependent when it comes to manual manipulations. For example, in one company, it was found that many of
the analysts could not properly use a pipette below 2 mL and errors as high as 5% were noted in some cases
(13).
4.2.6 Cost
The cost of using volumetric glassware is an issue that is often taken for granted. A great deal of effort goes
into keeping glassware organized and stored in a laboratory. Everyone who has worked in the lab has probably
been charged with ordering and putting away the clean glassware at some point in their career. This is costing
a proportion of a FTE. The pre and post rinsing on a company wide basis, assuming a very conservative 25 mL
use per flask and 10,000 sample preparations, might be costing the company $10,000 or more each year at a
$40/liter average solvent cost. These annually recurring costs can add up to a substantial amount, especially
when you include the rising cost of waste disposal. Additional costs can be incurred by lab services groups that
transport the flasks to and from the washing facility; and attrition due to breakage and damage, which results in
approx. 10% loss each year at a cost of about $20 per flask.
4.3 Mixing
Most samples are sonicated to expedite the breakup of tablets, capsules, or powders. Sonication can cause OOS
results when there is a lack of robustness in the method. The lack of robustness arises from the improper use
of the sonicator and whether or not the instrument is tuned properly. Most sonicators have the following instructions on them:
White Paper
METTLERTOLEDO
White Paper
Do not place parts or containers directly on the bottom of the cleaning tank: use a try or wire to suspend
items.
Do not allow the solution to drop more that 3/8 inch below the operating level line with the cleaner on.
However, from industry feedback it seems that few people follow those instructions. The pictures in Figure 6
illustrate the difference between a tuned sonicator on the left and an untuned sonicator on the right.
The untuned system has most of its energy focused in the middle of the bath, where you can see the large hole
in the foil (on the right). Therefore, the energy of the system can vary significantly depending on the placement of
the sample into the bath.
Many methods need to have better instructions for the final mixing step. Most methods only state to mix well
without realizing that a volumetric flask is an extremely poor mixing vessel that requires it to be inverted a number of times to ensure proper mixing.
4.4 Labeling
Labeling can potentially cause OOS results due to label mix ups but the most significant issues with labeling
are usually identified at safety and GxP audit times. Regardless of what a labeling SOP in the company states,
when flasks in laboratories are examined, the labeling content usually ranges from the absolute minimum to
the very detailed. Of course, all of these permanent marker labels must be removed before sending them out to
be washed and that necessitates the use of methanol or acetone to wipe down the flasks which takes time and
wastes more solvent.
White Paper
METTLERTOLEDO
Figure 7: Gravimetric dispensing systems: QB5 on the left; QX1 on the right
The user has to define the concentration required and the target amount of solution to be prepared. The software
calculates the target amount of solid to dispense and then, according to the actual amount of solid dispensed,
delivers the appropriate amount of diluent to prepare an extremely accurate and precise concentration. Individual
dosing heads are used for each substance. The powder dosing heads are disposable and designed for use with
a single substance only. The solvent dispensing heads are also exclusively used for a single solvent, and no
valves or switching or washing of lines is employed. This means that all potential sources of cross contamination risk are eliminated.
The QB3 and QB5 systems automate the powder and liquid dispensing process. The difference between the QB3
and QB5 system is the type of balance integrated, and hence the weighing specifications. These differences are
described in Table 2. For both the QB3 and QB5 systems, the appropriate powder dispensing head has to be
selected and manually inserted into the system. Each dosing head is identified by RFID chip for process security.
This head has to be manually removed and replaced by the appropriate liquid dosing head at the Add diluent
step. All data is recorded electronically and can be automatically printed onto labels.
For fully automated sample preparation approach the QX7 is available. The QX7 can accommodate 10 individual powder dosing heads and 5 different solvent bottles on the system at any one time. Dispensing of solid and
liquid can occur unattended, with the system automatically selecting and replacing the appropriate heads and
placing target vials onto the balance position for each programmed sample preparation. Again heads are tracked
by RFID chip, target vials are identified by barcode and the data is stored electronically.
White Paper
METTLERTOLEDO
11
White Paper
Minimum weight
Readability
Number of sample preparations in
a batch
QB3
QB5
QX7
10 mg
6 mg
1.4 mg
0.01 mg
0.005 mg
0.001 mg [1 g]
30 1)
1 48
(4 racks of 12)
Minimum concentration
Maximum concentration
1 g/g
1 g/g
1 g/g
100 g
100 g
~10 g
10
1) If
Table 2: Specifications of Quantos QB3, QB5 and QX7 Gravimetric Dispensing Systems
These automated gravimetric systems are being adopted by analytical and QA/QC laboratories in the pharmaceutical industry.
12
White Paper
METTLERTOLEDO
White Paper
METTLERTOLEDO
13
White Paper
safety factor for automated weighing systems as Quantos, typically 1.5 instead of 2 within a controlled laboratory environment with trained operators. Consequently, sample sizes can be chosen much smaller for Quantos,
typically smaller by a factor of 3 as compared to manual weighing. While usually 50 mg are weighed manually
on a XP205 semi-micro balance, a Quantos QB5 using the same technology typically allows for weighing only
13.5 mg.
Besides of the smaller minimum weight and the smaller safety factor, an additional benefit of gravimetric sample
preparation is that the analyst is not constrained to make a volume based on the size of volumetric flask available. These two factors combined mean that smaller amounts of sample can be used, smaller volumes of solutions can be prepared, less solvent is consumed and there is less waste to dispose of. The automated nature of
the process also makes it safer for the analyst.
Figure 9: Comparison of volumetric and gravimetric sample preparation on the amount of solvent used.
Being constrained to Class A volumetric glassware forces an analyst to use much more substance than necessary because they are limited to the capacity of the flasks available. The amount of solvent used when preparing
samples using a manual volumetric approach is indicated by the red line in Figure 8. The sharp vertical drops in
the red line on the graph indicate the four discrete points at which the minimum weight of the substance matches
an available volumetric flask size (10ml, 25ml, 50ml, 100ml) and thus the only concentrations at which the
actual minimum weight can be used to achieve the desired concentration. In all other cases, the amount of
sample must be rounded up to match the next size of volumetric flask available.
Automated gravimetric sample preparation is not limited to these discrete intervals, as the smooth green curve
indicates. With this method, the minimum amount of sample (9 mg) can be weighed at every point on the curve,
and the corresponding amount of solvent can be added, to achieve the required concentration.
At the point indicated by the blue arrow (which corresponds to a concentration of 1.1 mg/ml), six times less
substance and solvent is used to achieve the same concentration gravimetrically. This is due to an equal contribution from both of the factors mentioned previously: the minimum amount of sample that can be weighed on
the Quantos QB5 is just over three times smaller; also, the total sample volume has been rounded up to the next
available size of volumetric flask (50ml).
For example, Table 3 illustrates the preparation of three concentrations of solution manually (volumetrically)
compared to automatically (gravimetrically) using either (a) an XP205 balance and an 50mL or 100mL volumetric flask; (b) a QB3 automated sample preparation system; (c) a QB5 automated sample preparation system. The amounts of substance and solvent consumed for each method are listing along with the savings.
14
White Paper
METTLERTOLEDO
Manual (XP205)
Typical minimum weight 1)
Recommended safety
14 mg
10 mg
6 mg
1.4 mg
1.5
1.5
1.5
factor 2)
0.5
1.0
1.5
0.5
1.0
1.5
0.5
1.0
1.5
0.5
1.0
1.5
50
50
37.5
15
15
15
2.1
2.1
2.1
100
50
25
30
15
7.5
18
4.5
4.2
2.1
1.05
70%
70%
60%
82%
82%
76%
96%
96%
94%
The values reported in this table are TYPICAL values (typical specifications). Determination of the minimum weight on site will
provide the user with the ultimate capability on the instrument in its specific location.
2) Note: The safety factor quoted is that recommended for stable environments and trained operators. For unstable environmental
conditions or insufficiently trained operators higher safety factors should be used.
Table 3: Substance saving based on automated sample preparation methods
So, if a 1.0 mg/mL solution is prepared manually in a volumetric flask, 50 mL of solvent and 50 mg of
substance are consumed. When the solution is prepared using an automated gravimetric method on the QB5
system, 9 mL of solvent and 9 mg of substance are sufficient (assuming a solvent density of 1 g/mL).
A saving of 82% in both substance and solvent can be realized whilst remaining compliant with USP General
Chapter <41>. Using the QX7 system, the gravimetric sample preparation method delivers solvent and substance
savings of 96%.
Step
200 mL container
Weigh 50 mg sample
Sample transfer
Fill to mark
Volumetric
Gravimetric
Determinate Errors
Indeterminate Errors
0.05%
Uncalibrated
Determinate Errors
Indeterminate Errors
0.1% balance
Others are accounted
for using a safety factor
of 2 or higher
Re-weighing container
Powder transfer
0.1% balance
Others are accounted
for using a safety factor
of 1.5, if automated
(in a controlled lab
environment)
As you can see in Table 4, with gravimetric sample preparation the number of determinate errors is reduced and
the indeterminate errors which tend to be much larger than the determinate ones are essentially eliminated or
accounted for.
White Paper
METTLERTOLEDO
15
White Paper
Dose API
(mg)
Solvent
added (g)
Solution
Concentration
(mg/g)
Area
Area
correlated to
0.6 mg/g
Injector
Accuracy
Solution 1
Area
10.105
16.7481
0.60299
2596.8833
2584.00634
Inj. 1
2593.06665
10.320
17.1048
0.60298
2595.35474
2582.52818
Inj. 2
2604.43384
10.140
16.8063
0.60298
2597.53027
2584.69296
Inj. 3
2604.12378
10.125
16.7815
0.60298
2595.5564
2582.72885
Inj. 4
2604.89429
10.250
16.9885
0.60299
2597.12427
2584.24611
Inj. 5
2607.99683
10.200
16.9058
0.60298
2596.04517
2583.2152
Inj. 6
2605.37231
10.130
16.7895
0.60299
2602.42212
2589.51769
Inj. 7
2602.29541
10.040
16.6408
0.60297
2609.64868
2596.79455
Inj. 8
2608.33862
10.275
17.0297
0.60299
2604.74341
2591.82747
Inj. 9
2593.23218
Inj. 10
2601.73145
Mean
10.176
16.866
0.603
2599.479
2586.617
Mean
2602.549
5.37642343
RSD (%)
0.21
RSD (%)
0.09092503 0.150659114
0.894
0.893
0.19
0.19
For the nine individual samples, the weight of solid dispensed is shown in column 2. The weight of solvent dispensed is shown in column 3 and the concentrations achieved are shown in column 4. Accurate solvent weighing compensates for any variation in the amount of solid dispensed (which is already within a narrow range
10.040 and 10.320 mg), to achieve an accurate concentration of 0.603 mg/g in every case. The repeatability of
the concentrations is excellent, as indicated by the low standard deviation, RSD = 0.001% (pale blue highlighted
cell). When the peak area is correlated after analysis, the RSD across these nine solutions is 0.19% (dark blue
highlighted cell).
The two columns at the right hand side of Table 5 show ten injections of the same solution, with the aim of
eliminating the contribution of the analytical system to the overall variability. The RSD of these samples is 0.21%
(green highlighted cell). So, in conclusion, automated standard preparation with solvent weighing shows no significant variability in the solution concentrations, that cant be attributed to the analytical instrumentation itself.
When the same experiment was repeated by preparing the individual standards manually, the results in table 6
were obtained. The RSD on the correlated areas was 0.57% (compared to 0.19% for automated). This IS significant compared to the accuracy of the analytical instrumentation, and means that the variability can be almost
only attributed to the manual preparation process. As the weighing process and the HPLC analysis are independent from each other, the associated %RSD are not added arithmetically but quadratically in order to determine
the %RSD of the peak area. As the %RSD of the peak area is 0.57% (sample weighed manually), and the %RSD
for the overall variability of the analytical instrumentation is 0.21%, the calculated %RSD for the manual sample
preparation is 0.54%, which fits nicely to the experimentally determined %RSD of 0.60% of the manual weighing process. In other words, for this experiment the limiting factor for the overall accuracy of the HPLC analysis
was the manual sample preparation, whereas for automated sample preparation the limiting factor for the overall
accuracy is the HPLC itself.
16
White Paper
METTLERTOLEDO
Manual Dose
#
Blend dosed
mg
Flask Size
ml
Conc. achieved
mg/ml
12.920
20
0.646
12.960
20
0.648
12.860
20
0.643
12.840
20
0.642
13.060
20
0.653
12.980
20
0.649
12.860
20
0.643
12.800
20
0.640
12.860
20
0.643
10
12.920
20
0.646
Mean
12.906
20
0.645
0.078
0.0
0.004
% RSD
0.60%
0.0
0.57%
Manual Volumetric
Automated Gravimetric
Substance saving
White Paper
METTLERTOLEDO
17
White Paper
Manual Volumetric
Automated Gravimetric
User Safety
6 Conclusion
The most important measure to guarantee accurate weighing and consequently to avoid the possibility of
OOS due to weighing is the determination of the minimum weight of the balances. Consequently, it is important to always weigh above the minimum weight in order to comply with the respective accuracy requirements.
For automated dosing systems such as Quantos, the minimum weight is significantly smaller as compared to
manual weighing. It is good practice to apply a safety factor in order to compensate the variability of the minimum weight due to different operators and changing environmental conditions, however, the safety factor can be
chosen significantly smaller for automated weighing systems as environmental effects are reduced and the variability introduced by the operator is completely removed.
Reducing the occurrence of OOS results in the laboratory requires close attention to the details of where errors
can occur, a critical evaluation of the overall process workflow, and a concerted effort to change those practices
that lead to OOS results or errors in the data. New technologies must be brought into the laboratory to finally
improve the data quality that is being generated by laboratories around the world. In addition, most companies
want and need to achieve higher productivity with the same or less resources. This efficiency cannot occur without a fundamental change in the way sample preparation is currently performed which has had little improvement for the best part of a century and still accounts for more than 60% of our time spent in the laboratory.
18
White Paper
METTLERTOLEDO
Gravimetric sample preparation, which is defined as automated weighing and dispensing of both the solid and
the solvent, is based on the weighing standard GWP. Furthermore, it has several important benefits in a modern laboratory: it improves user safety due to reduced exposure risk; it improves process safety via electronic
recording and tracking of data; and it can save substance and solvent due to reduced minimum weight and
elimination of volumetric flasks.
However, the most important benefit of gravimetric sample preparation is that it is an innovative way to enhance
the accuracy and drastically reduce the variability in sample processing steps which has been shown to be the
major source of Out-of-specification results. Some areas of high error risk in the process, such as those involving use of volumetric flasks, are eliminated completely. Gravimetric sample preparation has the effect of reducing
laboratory errors and increasing laboratory efficiency.
7 References
1. United States of America v. Barr Laboratories, Inc., 812 F Supp 458 (DNJ 1993).
2. U.S. Food and Drug Administration. Guidance for Industry: Investigating out-of-specification (OOS) test
results for pharmaceutical production. FDA. Available at: www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM070287.pdf.
3. Majors, R.E. LC/GC Magazine, 1991, 1997, 2002
4. General Chapter <41> Weights and Balances, US Pharmacopeia USP34 NF29, Rockville, Maryland,
2013, Online-Edition
5. GWP - Good Weighing Practice A Risk-Based Approach to Select and Test Weighing Instruments,
White Paper, Mettler-Toledo AG, Greifensee, Switzerland, July 2009.
6. Reichmuth A., Fritsch K., Good Weighing Practices in the Pharmaceutical Industry Risk-Based Qualification and Life Cycle Management of Weighing Systems, Pharmaceutical Engineering, Volume 29, Number 6,
ISPE, Tampa FL, USA, 2009.
7. Fritsch K., Quenot J.-L., Good Weighing Practices Avoid OOS Results with Proper Weighing, Pharmaceutical Formulation & Quality, Volume 14, Number 1, Wiley-Blackwell, Hoboken, NJ, USA, 2012.
8. Evaluation of Measurement Data Guide to the Expression of Uncertainty in Measurement (GUM), JCGM
100:2008, Bureau International des Poids et Mesures, Svres, France, 2008. Available at www.bipm.org.
9. Guidelines on the Calibration of Non-Automatic Weighing Systems, EURAMET cg-18, Version 3.0, European
Association of National Metrology Institutes, Braunschweig, Germany, 2011. Available at www.euramet.org.
10. General Chapter <1251> "Weighing on an Analytical Balance", Second Supplement to USP36-NF31,
June 2013, Rockville, Maryland, Online-Edition.
11. Important Technical Guidance on Glassware, Tom Coleman and Georgia Harris, NIST, Aug. 2005.
12. "Traceability: Volumetric apparatus", LAB15 Guidance, Edition 2, UKAS, 2009. Available at:
http://www.ukas.com/library/Technical-Information/Pubs-Technical-Articles/Pubs-List/LAB15.pdf
13. Dr Charles Ray, former Associate Director of Analytical R&D at Bristol-Myers Squibb Co., personal
experience from managing Analytical R&D teams in leading pharma companies.
14. Fritsch, K., Ratcliff, J., Ray Ch., Reducing Variability and Out-of-Specification Results by Implementing High
Quality Gravimetric Sample Preparation (GSP), Pharmaceutical Engineering, Volume 32, Number 1, ISPE,
Tampa FL, USA, 2012.
White Paper
METTLERTOLEDO
19
www.mt.com
Mettler Toledo AG
Laboratory & Weighing Technologies
CH-8606 Greifensee, Switzerland
Tel. +41 (0)44 944 22 11
Fax +41 (0)44 944 30 60
Subject to technical changes
07/2013 Mettler-Toledo AG
Printed in Switzerland 30086085
Global MarCom Switzerland