the determination of extractables and leachables in...
TRANSCRIPT
The determination of extractables and leachables in pharmaceuticalpackaging materials using headspaceGC/MS
Application Note Albert E. GudatRoger L. Firor
Abstract
Using the Agilent G1888 Network Headspace/6890N GC/5975 inert MSD
system several primary packaging materials for pharmaceuticals were
evaluated in a study. In this Application Note results for two of these
materials, a labeled HDPE bottle and a soft elastomer liner from the
inside of the screw cap are presented. A multiple headspace extraction
technique (MHE) was used and the highest attainable amount of extracta-
bles that could ever be concentrated in the drug product was calculated.
Further, the extraction efficiency of the MHE procedure was studied
using benzaldehyde and phenol as target compounds. The analytical
results obtained from the inert 5975 headspace-GC/MS provide excellent
sensitivity when utilizing SIM/Scan data acquisition with multiple head-
space extraction (MHE). The synchronous SIM/Scan feature of the 5975
MSD allows for very low level SIM detection while also searching for
unkowns using scan data.
release low molecular weightchemicals into the product andpossibly compromise the biologi-cal safety and efficacy of the drugproduct. These chemicals includeboth extractable and leachablecomponents. Extractables arechemical substances that areobtained by exposing the packag-ing to a variety of solvents underexaggerated incubation conditionsof time and temperature. Leachablesdiffer from extractables in thatthey are chemical substances thatmigrate under normal conditionsof use from the CCS into a drugproduct. Leachables are thereforea subset of extractables; allextractables are potential leach-ables of toxicological concern.
Pharmaceutical container closuresystems are usually fabricatedfrom glass, plastic, or a variety ofmetals including aluminum andstainless steel. Extractables maycome from these primary packagingmaterials or from secondary pack-aging components. For example,plastics consist of the polymer,residual monomers or low molec-ular weight oligomers, additivessuch as plasticizers, phenolicantioxidants, UV stabilizers, col-orants, catalysts, raw materialimpurities and many other produc-tion aids. Secondary packagingmay include laminates such asadhesives and release agents, inks,epoxides, urethanes, acrylates and polyesters. Residual solventsthat originate from the packagingmaterials would have to be classi-fied as extractables
ExperimentalMass Spectrometry (MS) is per-haps the most widely applied technique to detect, identify and
IntroductionThe containers used for pharma-ceutical packaging are regulatedby many different government andnon-government bodies. Forexamples, FDA guidances andUSP procedures provide some ofthe requirements to help drugmanufacturers in the USA developthe necessary procedures. Forinternational distribution, the situ-ation becomes complicated byinconsistencies in requirements indifferent countries. However, theFDA and USP work with othercountries and the InternationalConference on Harmonization(ICH) to harmonize monographsand procedures.
Evaluation of packaging materialsfor potential extractables andleachables is critical to guaranteethe integrity of the drug productand assure compliance with CFRTitle 21, Part 211.65 which states:Equipment shall be constructed sothat surfaces that contact compo-nents, in-process materials, ordrug products shall not be reac-tive, additive, or absorptive so asto alter the safety, identity,strength, quality, or purity of thedrug product beyond official orother established requirements1.The US FD&C Act also states thata drug or device shall be deemedto be adulterated if its container iscomposed in whole or part of anypoisonous or deleterious sub-stance which may render the con-tents injurious to health2. A goodsummary of regulatory and scien-tific considerations on testing forextractables and leachables canbe found in reference 3.
A container or drug deliverydevice, often referred to as a con-tainer closure system (CCS), forpharmaceutical products may
quantify organic extractables. For this application an Agilent G1888/6890/5975Headspace/GC/MSD system isused to characterize extractables.Headspace vials of 10 mL volumewere used and the headspace sam-pler was equipped with a 1 mLsample loop. Sufficient flow mustbe maintained through the systemto avoid excessive peak broaden-ing, therefore split injection isused. At the chosen split ratio of1:1, the headspace sample loop isswept fast enough to producegood peak shapes. Headspacevials containing 10 micro liters ofa standard in DMSO or methanolwere equilibrated for 60 minuteswhereas the milled samples ofhigh density polyethylene (HDPE)and liner polymer from the clo-sure system were equilibrated for300 and 120 minutes, respectively.The headspace temperature wasselected to be just below the melt-ing point but above the glass tran-sition temperature of the polymer:125°C for the HDPE and 100 °Cfor the liner material.
In static headspace extraction ananalyte is partitioned between thecondensed phase and the gaseousphase until equilibrium is reached.However, since a single extractionwill never force all of the analytecontained in the solid sample tomigrate into the gaseous head-space, a multiple headspaceextraction technique (MHE) isused and the highest attainableamount of extractables that couldever be concentrated in the drugproduct is calculated. This is theequivalent of doing an exhaustiveextraction on the sample thatgives the highest concentration forthe analyte. This is the worst-casescenario with the highest associat-ed risk attributed to the analyte.
2
Reference 4 gives an extensivedescription of the MHE technique.Appendix 1 from that ApplicationNote has been duplicated at theend of this note and gives thesalient features of the MHE tech-nique and an example for calculat-ing the concentration of an ana-lyte. Since standard chromatogra-phy software does not usuallyinclude a quantitation procedurefor MHE, an Excel template (table 2and figure 8) was designed to calculate properties derived fromthe semi-logarithmic plot: the cor-relation coefficients, slopes, totalpeak areas and the concentrationof the analyte in the sample. Thesample weight and the amount ofstandard added to the headspacevials is also entered in the spread-sheet. Note that the standard ofthe analyte does not have to beprepared in the sample matrix andis usually prepared in solutionusing a high boiling point solventsuch as DMSO. Matrix effects donot influence the concentrationcalculated with the MHE techniques.This is the beauty of a matrix-inde-pendent MHE calculation.
In order for the MHE technique towork efficiently with solid sam-ples, the sample must be reducedto a fine powder to maximize thesurface area to volume ratio. Thiswas accomplished with a 6750SPEX SamplePrep Freezer/Millthat cools the sample to cryogenictemperatures and pulverizes it bymagnetically shuttling a steelimpactor back and forth againsttwo stationary end plugs. Sincethe vial is closed, the integrity ofits contents is maintained: haz-ardous or critical samples are easi-ly controlled, cleanup is simpli-fied, and cross-sample contamina-tion is eliminated. Because thecontainer holding the sample is
immersed in liquid nitrogenthroughout the milling cycle, thesample is kept at cryogenic tem-peratures and its key aspects arepreserved. In contrast, conven-tional room-temperature grindingmay alter the sample compositionby the heat generated duringgrinding and some of the volatilecomponents may escape.
The packaging materials exten-sively evaluated in this studyinclude a HDPE (high-densitypolyethylene) bottle with theprinted label still attached and a
soft elastomer liner from theinside of the screw cap. A blisterpack, an insulin syringe, and apolypropylene (PP) containerwere also tested but the resultsare not reported here. All thesecontainers except the syringe hadcontained pharmaceutical drugproducts. The insides of thesecontainers were carefully cleanedto ensure that no drug residueremained. All the samples pro-duced good chromatograms withan abundance of peaks.
6890N GC
Injection port Volatiles interfaceTemperature 225 °CSplit ratio 1:1Carrier gas HeliumConstant flow mode 1.2 mL/min
GC oven program
Initial temperature 40 °C Initial time 4 minRate 20 °C/min.Final temperature 300 °CFinal time 5 min Columns: 30 m x 0.25 mm x 0.5 µm DB-5MS, part# 122-5536
G1888A Headspace sampler
Loop size 1 mLVial pressure 25 psigHeadspace oven 125 °C for HDPE, 100 °C for liner Loop temperature 150 °CTransfer line temperature 150 °CAdvanced function 8 MHE on, 8
Equilibration time 300 min., no shakeGC cycle time 30 minPressurization 0.3 minVent (loop fill) 0.3 minInject 3 min
5975 Inert MSD
Synchronous SIM/Scan mode onSIM 1 group, 3 ionsScan 35 to 350 amu, samples 2 ^2
Threshold 100 EM offset +300
Source temperature 230 °CQuad temperature 150 °CTune atune.uChemStation software G1701DA D.02.00
Table 1Instrument conditions.
3
Experimental results and discussionFigure 1 shows the chromato-grams acquired in synchronousSIM/Scan mode from a 0.208 gsample of cryo-milled HDPE. Anenlargement of the scan data isshown in figure 2. The chro-matogram shows many peaksmost of which are from siloxanes(probably from the column) andalkenes. The chromatogram isvery complex and many peaks arecomposed of multiple overlappingcomponents and the apex spec-trum is actually a composite ofthese constituents. A mass spec-tral library search would give apoor match, at best, and certainlywould not identify all of the indi-vidual components that make upthe composite spectrum.Deconvolution produces “clean”spectra for each overlapping com-ponent. These individual spectracan then be library searched witha high expectation for goodresults. A simplified illustration ofthis deconvolution process isshown in figure 3.
When the data for HDPE wasprocessed with the AutomatedSpectral Deconvolution andIdentification System (AMDIS)from NIST, the program found 229component spectra and identified15 targets from a database of 731hazardous chemicals, the AgilentHazardous Chemical DatabaseLibrary (HCD)5,6. A typical outputfrom AMDIS is shown in figure 4.Among the compounds identifiedwere phenol, di-n-butyl phthalateand bis-2-ethylhexyl phthalate.Actually, a number of phthalateswere found between 12 and 19minutes and have been indicatedon figure 2 as the phthalate region.
2.00 6.00 10.00 14.00 18.00
5000000
1e+07 1.5e+07 2e+07
2.5e+07 3e+07 3.5e+07
4e+07 4.5e+07 5e+07 5.5e+07
6e+07 6.5e+07 7e+07
Time-->2.00 6.00 10.00 14.00 18.00
Time-->
Abundance AbundanceTIC: HDPE cryomilled inj#2.D
Scan Data of HDPE
10000
20000
30000
40000
50000
60000
70000
80000
90000
100000
110000
120000
TIC: HDPE cryomilled inj#2.D
3 ions: 77, 105, 106
SIM data of HDPE
Figure 1Synchronous SIM/Scan data from a 0.208 gram sample of cryo-milled HDPE polymer.
2.00 6.00 10.00 14.00 18.00
100000150000200000250000300000350000400000450000500000550000600000650000700000750000
Time-->
Abundance TIC: HDPE cryomilled inj#2.D\DATA.MS
Benzaldehyde
�
Scan of HDPE
Phthalates region
Figure 2Enlargement of the Scan data from figure 1.
TIC & Spectrum Deconvoluted peaks and spectra
Deconvolution
Matrix
Target
Interference
Figure 3AMDIS deconvolution pulls out individual components and their spectra.
4
Since most of the electron ioniza-tion spectra of phthalates look thesame and unequivocal identifica-tion of the phthalates is difficult,positive identification of thephthalates may only be accom-plished by looking at chemicalionization data which gives molec-ular weight information. Also iden-tified in figure 2 is the location ofthe benzaldehyde peak that isused for the MHE study of HDPE.
SIM data was collected for ben-zaldehyde using the 77,105 and106 ions that are characteristic ofbenzaldehyde. The area data givenin fables 2 – 4 are for the extract-ed 106-ion chromatograms.
Figures 5 and 6 illustrate howmatrix interference can compli-cate the mass-spectral identifica-tion of peaks. A very large silox-ane peak with an intense 281 ioncoeluted with the benzaldehyde.To eliminate the siloxane interfer-ence from the spectrum, extractedion chromatograms of 281, 105and 106 were generated. The threechromatograms were normalizedso that the peaks maxed out at10000 counts. A good backgroundspectrum with the proper ionabundance was now easily select-ed from the left of the benzalde-hye peak maximum. Figure 6shows the spectrum before andafter background correction. Thecorrected spectrum when searchagainst the Wiley library of spectraproduced an unequivocal hit forbenzaldehyde. However, theAMDIS did all this work for everypeak in the chromatogram in lessthan a minute.
9.70 9.80 10.00 10.20 10.400
1000
2000
3000
4000
5000
6000
7000
8000
9000
Time-->Time-->
Abundance��� ������ ����� �� ������ � ���� �������� ������������ �
105 and 106
281
Normalized chromatogramsused to guide background subtraction
��� ����� ������ �� ����� � ���� �������� ������������ � ��� !���� !���� �� !���� � ���� �������� ������������ �
9.70 9.80 10.00 10.20 10.400
50000100000150000200000250000300000350000400000450000500000550000600000650000
Abundance��� ������ ����� �� ������ � ���� �������� ������������
Original chromatograms
105 and 106
281
��� ����� ������ �� ����� � ���� �������� ��������������� !���� !���� �� !���� � ���� �������� ������������
Figure 5Manual MS deconvolution attempted for the peak at 9.8 min.
80 120 160 200 240 2800
50000100000150000200000250000300000350000400000450000500000550000600000
m/z-->
Abundance281
10577 267
249193133205
163179 23511989 149 293221
Before background correction
80 120 160 200 240 2800
2000
6000
10000
14000
18000
22000
26000
30000
34000
38000
42000
46000
m/z-->
Abundance105
77
269125 285
189 251162
221178149
88 234204136 294
After background subtraction
Figure 6The spectrum for the peak at 9.8 min before and after doing a background subtraction.
5
Figure 4AMDIS results for the scan data of HDPE with the target at 16.71 min identified as di-n-butylphthalate.
Figure 7 shows the changes in thebenzaldehyde peak as a functionof the number of extractions dur-ing the MHE. The peak areashould decreases exponentiallywith the extraction number.However, sometimes during MHEssome analytes may exhibit asmaller peak area from the firstextraction than what is observedfrom the second extraction, whilethe remaining extractions showthe expected exponential decreasein the peak areas. Such behaviorwas also observed for benzalde-hyde in HDPE (figure 7, 8 andtable 2).
Since headspace analysis involvesa gas-phase extraction, we shouldbe aware that the first extractionis always done with air for theextraction fluid because that is inthe vial when the sample is placedin the vial and capped. Subsequentextractions are done primarilywith helium because that is thegas used to pressurize the vialafter equilibration. The extractionefficiency with air is probably lessthan with helium which causes thepeak area of the first extraction tobe lower than what the graphicalextrapolation would indicate (figure 8). In such cases a modi-fied form of the equation may beused to calculate the total area forthe analyte (see Appendix for theoriginal equation). Now the TotalPeak Area is
where the second half of the equa-tion (the ratio) gives the total peakarea as derived from the regressionanalysis using the results from thesecond and subsequent extrac-tions.
9.75 9.80 9.85 9.90 9.95 10.000
5000
10000
15000
20000
25000
30000
35000
40000
45000
Time-->
AbundanceIon 106.00 (105.70 to 106.70): HDPE cryomilled inj#5.D\DATASIM.MS
benzaldehyde
5------
4------
3------1------2------
Ion 106.00 (105.70 to 106.70): HDPE cryomilled inj#4.D\DATASIM.MSIon 106.00 (105.70 to 106.70): HDPE cryomilled inj#3.D\DATASIM.MSIon 106.00 (105.70 to 106.70): HDPE cryomilled inj#2.D\DATASIM.MSIon 106.00 (105.70 to 106.70): HDPE cryomilled inj#1.D\DATASIM.MS
Figure 8Semi-logarithmic plot of the MHE raw data for benzaldehyde in HDPE.
Benzaldehyde in 0.208 grams of HDPE polymer (cryomilled)
extraction # sample standard standard stats1 126081 90157 0.821008 108302.52 129433 72611 0.004898 0.0162433 118095 58834 0.998153 0.0154874 98261 49099 1621.638 35 84345 40898 0.388965 0.000726 68811 329927 56676 27095 sample stats8 45872 22746 0.897668 151819.2
0.023684 0.078551regression correlation (E4 or E11) 0.873825502 0.998153435 0.873826 0.074896slope (k) = ln(E2 or E9) -0.107955422 -0.197222062 20.7766 3
0.116544 0.016828total area = (A(1)/(1-e(-k))) 1232073 503694
analyte in vial (mg) 0.026906838 0.011sample amt (mg) in vial 208
concentration (ppm) in wt/wt 129.36concentration (wt-%)=ppm * (10 ^ -4) 0.0856
6
126081 11809598261
6881156676
45872
90157
5883449099
4089832992
2709522746
84345
129433
72611
y = 171387e -0.1545x
R2 = 0.9545
y = 107935e -0.1963x
R2 = 0.999410000
100000
1000000
0 1 2 3 4 5 6 7 8 9Extraction number
Peak
are
a
Sample
Standard
Expon. (sample )
Expon. (standard )
Benzaldehyde in HDPE
)1(2
1 Ke
AA −−
+=Total Peak Area
Figure 7Multiple headspace extraction data for the benzaldehyde peak from HDPE. The numbers next tothe peak indicate the extraction number.
Table 2Excel template used for entering area data, sample weight and standard amounts used for thecalculations needed for the MHE procedure. Table contains raw data for benzaldehyde in HDPE.
7
Should the above argument forextraction efficiency be indeedvalidated by purging the head-space vial with helium before sam-ple introduction, another optionwould be to select the point asso-ciated with the first extraction inthe Excel semilogarithm plot anddrag that point onto the regressionline generated by the subsequentextraction points. The results forsuch a data treatment is shown infigure 9 and table 3. When the datapoint is dragged onto the regres-sion line, the peak area linked tothe table is no longer an integervalue but now has decimal points.Clearly the R2 of the regressionline, the correlation coefficient,has improved from 0.9546 (figure 8)to 0.9931 (figure 9). Note that thedata point for the 2nd extractionalso appears a little below the regression line.
Benzaldehyde in HDPE
172583.78
11809598261
6881156676
45872
90157
5883449099
4089832992
2709522746
129433
8434572611
y = 200518e -0.1807x
R2 = 0.9931
y = 107935e -0.1963x
R2 = 0.999410000
100000
1000000
0 1 2 3 4 5 6 7 8 9Extraction number
Peak
are
a
Sample
Standard
Expon. (sample)
Expon. (standard)
Benzaldehyde in 0.208 grams of HDPE polymer (cryomilled)samp before fix
extraction # sample standard standard stats1 172583.78 90157 126081 0.821008 108302.52 129433 72611 129433 0.004898 0.0162433 118095 58834 0.998153 0.0154874 98261 49099 1621.638 35 84345 40898 0.388965 0.000726 68811 329927 56676 27095 sample stats8 45872 22746 0.843035 195167.2
0.015703 0.052081regression correlation (E4 or E11) 0.975253934 0.998153435 0.975254 0.049658slope (k) = ln(E2 or E9) -0.170747071 -0.197222062 118.2314 3
0.291546 0.007398total area = (A(1)/(1-e(-k))) 1099503 503694
analyte in vial (mg) 0.024011686 0.011sample amt (mg) in vial 208
concentration (ppm) in wt/wt 115.44concentration (wt-%)=ppm * (10 ^ -4) 0.0856
Table 3Data for benzaldehye in HDPE but with a correction made to the area associated with the 1st extraction.
Figure 9Semi-logarithmic plot of the MHE data for benzaldehyde after forcing the area associated withthe 1st extraction to fall on the regression line.
Probably not all the air in the head-space vial had been replaced withhelium after the first extractionand the 2nd extraction was donewith a mixture of air and helium. Ifthat data point is also moved ontothe regression line (figure 10 andtable 4), the R2 value for the graphbecomes 0.9989 and the calculatedamount of benzaldehyde in thesample goes from 0.0261 grams for the original virgin data to0.0229 grams after manipulation,i.e., the calculation gives 14 % moreanalyte without the correction.
8
Benzaldehyde in HDPE
172583.78
11809598261
6881156676
45872
90157
5883449099
4089832992
2709522746
141579.37
8434572611
y = 207710e-0.186x
R2 = 0.9989
y = 107935e-0.1963x
R2 = 0.999410000
100000
1000000
0 1 2 3 4 5 6 7 8 9Extraction number
Peak
are
a
Sample
Standard
Expon. (sample)
Expon. (standard)
Benzaldehyde in 0.208 grams of HDPE polymer (cryomilled)samp before fix
extraction # sample standard standard stats1 172583.78 90157 126081 0.821008 108302.52 141579.37 72611 129433 0.004898 0.0162433 118095 58834 0.998153 0.0154874 98261 49099 1621.638 35 84345 40898 0.388965 0.000726 68811 329927 56676 27095 sample stats8 45872 22746 0.835507 204119.5
0.00453 0.015024regression correlation (E4 or E11) 0.998097615 0.998153435 0.998098 0.014325slope (k) = ln(E2 or E9) -0.179716781 -0.197222062 1573.967 3
0.322981 0.000616total area = (A(1)/(1-e(-k))) 1049185 503694
analyte in vial (mg) 0.022912801 0.011sample amt (mg) in vial 208
concentration (ppm) in wt/wt 110.16concentration (wt-%)=ppm * (10 ^ -4) 0.0856
Table 4Data for benzaldehyde in HDPE but with corrections made to the areas associated with the 1st and 2nd extractions.
Figure 10Semi-logarithmic plot of the MHE data for benzaldehyde after forcing the areas associated withthe 1st and 2nd extraction to fall on the regression line.
The synchronous SIM/Scan datafor the liner material and the spec-trum of phenol at 9.83 min areshown in figure 11. AMDIS resultsare shown in figure 12. 213 compo-nents and 15 targets are identified.The graphics show the phenol tar-get at 9.83 min and the deconvolut-ed spectrum.
Table 5 lists the MHE-area data forphenol extracted from the linermaterial and the standard preparedin methanol. DMSO could not beused in the preparation of this stan-dard because a solvent impuritywith a strong 94 ion co-eluted withthe phenol peak. The concentra-tion of phenol was calculated at0.63 ppm in wt/wt units. Figure 13gives the Excel MHE graphics forthe sample and standard.
9
4.00 8.00 12.00 16.00
100000
200000
300000
400000
500000
600000
700000
800000
900000
1000000
Time-->
Abundance
TIC: liner cryomilled inj#2.D
Cap Liner, Scan data
4.00 8.00 12.00 16.00
100002000030000400005000060000700008000090000
100000110000120000130000
Time-->
AbundanceTIC: liner cryomilled inj#2.D
Cap Liner, SIM data
40 45 50 55 60 65 70 75 80 85 90 950
5000
10000
15000
20000
25000
30000
35000
40000
45000
50000
55000
60000
65000
70000
75000
80000
m/z-->
Abundance
Scan 1733 (9.831 min): liner cryomilled inj#2.D
94
66
7340 6350534744 79 8860 847670
phenol in 0.084 grams of liner polymer (cryomilled)
extraction # sample standard standard stats1 25882.12 33028 0.630118 54417.162 19906.73 22052 0.013536 0.0448953 15631.47 14377 0.99743 0.0428064 11977 8406 1164.101 35 9017 5314 2.133035 0.0054976 7093 33897 5729 1881 sample stats8 4643 1223 0.769746 33794.04
0.004119 0.013662regression correlation (E4 or E11) 0.999257247 0.997429527 0.999257 0.013026slope (k) = ln(E2 or E9) -0.261695112 -0.461847924 4036.026 3
0.684843 0.000509total area = (A(1)/(1-e(-k))) 112407 89293
analyte in vial (mg) 5.31233E-05 0.0000422sample amt (mg) in vial 84
concentration (ppm) in wt/wt 0.63concentration (wt-%)=ppm * (10 ^ -4) 6.32421E-05
Table 5Data for phenol in liner polymer.
Figure 11Synchronous SIM/Scan data for 0.084 grams of cryo-milled liner polymer and the background-subtracted spectrum at 9.83 min.
ConclusionAnalytical results obtained fromthe inert 5975 Headspace-GC/MSprovide excellent sensitivity whenutilizing SIM/Scan data acquisitionwith multiple headspace extrac-tion (MHE). The synchronousSIM/Scan feature of the 5975 MSDallows for very low level SIMdetection while also searching forunkowns using scan data. AMDISin conjunction with the AgilentHazardous Chemicals DatabaseLibrary made identification ofcomponents and targets in com-plex matrices very easy and fast.The matrix-independent MHEtechnique can generate quantita-tive values for target extractablesfor risk evaluation in a worst-casescenario. A very good discussionof how the numbers generated forextractables and leachables maybe evaluated
relative to ICH concentration lim-its is given in reference 7. HereGoldstein et. al. describe a hypo-thetical extractable study for ace-tonitrile and acetic acid and com-pares the results to ICH guidelinesfor Class I, II, and III residual sol-vents. In a similar approach, thevalues for benzaldehyde and phe-nol as determined by MHE couldbe compared to Permitted DailyExposure (PDE) limits.
10
Figure 12AMDIS results for the scan data of the liner with the target at 9.83 min identified as phenol.
Phenol in liner
15631.47
11977
7093
57294643
33028
14377
8406
5314
3389
1881
1223
25882.12
9017
19906.7322052
y = 32611e -0.2489x
R2 = 0.9981
y = 56875e -0.4783x
R2 = 0.99851000
10000
100000
0 1 2 3 4 5 6 7 8 9Extraction number
Peak
are
a
Sample
Standard
Expon. (sample)
Expon. (standard)
Figure 13Semi-logarithmic plot of the MHE data for phenol in liner polymer.
References1. Code of Federal Regulations, Foodand Drugs Title 21, Part 211.65;U.S. Government Printing Office,
Washington, DC, rev. 1 April 2004.
2.Chapter V Drugs and Devices,Section 501(a) (3) Federal Food,
Drug, and Cosmetic Act as
Amended through December 21,
2000, U. S. Government Printing
Office, Washington D.C., 2001.
3.C. Jeanne Taborsky and Eric B.Sheinin, American Pharmaceutical
Review, Volume 9 Issue 4, pp.
10 – 15, May/June, 2006.
4. Albert E. Gudat and Susan M.Brillante, “Multiple HeadspaceExtraction – Capillary Gas Chro-matography for the QuantitativeDetermination of Volatiles in SolidMatrices.” Agilent Application
Note, Publication number 5965-
0978E, 2000.
5. Bruce Quimby and Mike Szelewski,“Screening for HazardousChemicals in Homeland Securityand Environmental Samples Usinga GC/MS/ECD/FPD with a 731Compound DRS Database”, Agilent
Application Note, Publication
number 5989-4834EN, 2006.
6. Philip L. Wylie, Michael J.Szelewski, Chin-Kai Meng andChristopher P. Sandy, ”Compre-hensive Pesticide Screening byGC/MS using DeconvolutionReporting Software, Agilent
Application Note, Publication
number 5989-1157EN, 2004.
7. Adam Goldstein, David Schieche,Jim Harter, Ravi Samavedam,Lindsay Wilkinson and AmyManocchi, American
Pharmaceutical Review, Volume
8 Issue 5, pp. 60 – 70, Sept/Oct,
2005.
AppendixIn the Multiple Headspace Extraction technique the sample is equilibrated at some temperature for a givenamount of time and the headspace above the sample is analyzed. This equilibration and measurementprocess is repeated multiple times and an exponential decrease in the peak areas is observed. If we performan infinite number of extractions, all the volatile impurities will be converted into the gas phase and the totalpeak area (eqn. 1) will correspond to the total content of the analyte in the sample:
However, such a large number of extractions per sample becomes impractical and we are forced to use arguments from kinetics to get the total peak area.
The rate of conversion of the the analyte from the solid matrix into the gas phase is assumed to follow 1storder kinetics
which upon integration becomes
If the gas extraction is carried out carefully and for equal times, and equal portions of the headspace gas areintroduced into the chromatograph, then the peak areas of a given analyte in the chromatogram will followthe same exponential law since at equlibrium the distribution coefficient K
dis a constant, K
d= c
c/c
gwhere c
c
and cg
are the concentrations of the analyte in the condensed and gas phase, respectively. For a discontinu-ous or stepwise gas extraction performed at equal time intervals, eqn. 3 now becomes
Note: n = 1 at t = 0 since t = n - 1A
n= the peak area of the nth injection A
1= the peak area of the 1st injection
= Ann=
∞
∑1
= A1 + A2 + A3 + ... + An (1)Total Peak Area
- dc
dt = k c (2)
c = c0 ek t- (3)
An = A en K
1
1( )− (4)
11
For an inifinitely large number of extractions, the total peak area for an analyte thus becomes
This decreasing geometric progression in eqn. 5 converges to
We therefore do not need to do a complete gas extraction to obtain the total peak area but we must obtainvalues for A
1and K. The A
1value is the measured peak area of the analyte after the 1st gas extraction and K
is the slope obtained from a regression analysis of the semilogrithmic plot of eqn. 4:
For two measurements eqn. 6 simplifies to:
Example Calculation:
Solid Sample ( 337 mg in vial) Analyte Standard ( 0.00866 mg in vial)
© Agilent Technologies, 2006
Published August 1, 2006Publication Number 5989-5494EN
www.agilent.com/chem/pharmaqaqc
Ann=
∞
∑1
= A1 ( 1 + e K- + e K- 2 + e K- 3 + ... ) (5)
Ann=
∞
∑1
= A
eK
1
1( )− - (6)
In
In
An = In A1 + ( 1 - n ) K or
An = - K ( n - 1 ) + In A1
Ann=
∞
∑1
= ( )A
eK
1
1− − = ( )
6072
10 4615− e
. = 16423 An
n=
∞
∑1
= ( )15609
10 6654− e
. = 32120
TotalArea
Amount
analyte
analyte
= TotalArea
Amount
s dard
s dard
tan
tan
Amountanalyte =
1642332120
* 0.00 866 = 0.00443 mg
Amount
Weight
analyte
sample
* 1 e 6 = =
0 00443337
. * 1 e 6
= 13 .1 ppmAnalyte conc. (ppm by wt/wt)
Ann=
∞
∑1
= ( )A
A A
1
2
1 2−
(9)
(7)
(8)
Slope = - K
-----n----->
ln An
Roger L. Firor and Albert E. Gudat
are Application Chemists at
Agilent Technologies, Inc., USA.