EPA/600/4-90/022
September 1990
PERFORMANCE EVALUATION OF PARTICLE BEAM
LIQUID CHROMATOGRAPHY/MASS SPECTROMETRY
FOR THE MEASUREMENT OF ACID HERBICIDES
by
Chris M. Pace
Dennis A. Miller
Mark R. Roby
Environmental Programs
Lockheed Engineering & Sciences Company
Las Vegas, Nevada 89114
EPA Contract No. 68-03-3249
ENVIRONMENTAL MONITORING SYSTEMS LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
LAS VEGAS, NEVADA 89114
-------
TECHNICAL REPORT DATA
(Pleate read Instructions on the reverse before complete
1. REPORT NO. 2.
EPA/600/4-90/022
3 f PB90-270547
4 TITLE AND SUBTITLE
PERFORMANCE EVALUATION OF PARTICLE BEAM LIQUID
CHROMATOGRAPHY/MASS SPECTROMETRY FOR THE MEASUREMENT
OF ACID HERBICIDES
6. = = onBT DATE
September 199U
6 PERFORMING ORGANIZATION CODE
7 AUTHORIS)
Chris M. Pace, Dennis A. Miller, and Mark R. Roby
8 PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Lockheed Engineering and Sciences Company
1050 E. Flamingo Rd., Suite 120
Las Vegas, NV 89119
10 PROGRAM ELEMENT NO.
11 CONTRACT/GRANT NO
68-03-3249
12. SPONSORING AGENCY NAME AND ADDRESS
U.S. Environmental Protection Agency
Environmental Monitoring Systems Laboratory
P.O. Box 93478
Las Vegas, NV 89193-3478
13 TYPE OF REPORT AND PERIOD COVERED
Project Report 5/1/89-3/30/90
14 SPONSORING AGENCY CODE
EPA/600/07
15 SUPPLEMENTARY NOTES
16 ABSTRACT
Particle beam liquid chromatography/mass spectrometry (LC/MS) was
evaluated for the measurement of acid herbicides. An acetic
acid/ammonium acetate/methanol solvent system with a C-8 reversed phase
column gave baseline resolution of all target analytes. Detection limits
in the full scan mode were 100 ng to 500 ng for most of the target
analytes. Dalapon and dinoseb were not detected. Response curves over
the range 200 ng to 2000 ng were non-linear for most of the analytes.
Response factors tended to increase with increasing analyte
concentration. Mass spectra were variable and exhibited abundant ions
corresponding to "thermal" decomposition mechanisms. Spectral appearance
was dependent on analyte concentration, source conditions, and source
temperature. Only spectra acquired at high concentration were library
matchable. Therefore, a rugged and reliable method to identify and
quantify acid herbicides in environmental samples based on particle beam
LC/MS technology does not appear feasible at this time.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19 SECURITY CLASS (Tins Report)
UNCLASSIFIED
21.NOL.OF PAGES
65
20 SECURITY CLASS (Tills page)
UNCLASSIFIED
22 PRICE
EPA Form 2220-1 (R»». 4-77) previous edition is obsolete
i
-------
NOTICE
The information in this document has been funded wholly or inpart by the United States
Environmental Protection Agency under contract number 68-03-3249 to Lockheed
Engineering & Sciences Company. It has been subject to the agency's peer and
administrative review, and it has been approved for publication as an EPA document.
Mention of trade name or commercial products does not constitute endorsement or
recommendation for use.
n
-------
ABSTRACT
Particle beam LC/MS was evaluated for the measurement of phenoxyacid herbicides. An acetic acid/
ammonium acetate/methanol solvent system with a C-8 reversed phase column gave baseline resolution
of all target analytes. Detection limits in the full scan mode were 100 ng to 500 ng for most of the target
analytes. Dalapon and Dinoseb were not detected. Response curves over the range 200 ng to 2000
ng were non-linear for most of the analytes. Response factors tended to increase with increasing analyte
concentration. Mass spectra were variable and exhibited abundant ions corresponding to thermal'
decomposition mechanisms. Spectral appearance was dependent on analyte concentration, source
conditions, and source temperature. Only spectra acquired at high concentration were library matchable.
This report was submitted in fulfillment of contract #68-03-3249 by Lockheed Engineenng & Sciences
Company under the sponsorship of the United States Environmental Protection Agency. This report
covers the period from May 1,1989 to March 30, 1990 and was completed June 15,1990.
iii
-------
CONTENTS
Abstract iii
Tables V1-
Figures vii
Abbreviations and Symbols vi i i
Introduction 1
Conclusions 2
Recommendations 5
B?)erimental 6
Results and Discussion 7
Chromatography 7
Response Characteristics 13
Detection Limits 13
Response Curves 14
Reproducibility 18
Matrix Effects 19
Spectral Quality 24
References 38
Appendix A: Particle Beam Mass Spectra of Phenoxyacid Herbicides A-1
Appendix B: Comparison of Two Particle Beam Type Interfaces for the
Measurement of Phenoxyacid Herbicides by LC/MS B-1
v
Preceding page blank
-------
TABLES
Number Page
1 HPLC Parameters 8
2 Estimated Detection Limits 13
3 Effect of Ammonium Acetate on Detection Limit 14
4 On Column Response Factors 15
5 Effect of Ammonium Acetate on Response 16
6 Day to Day Response Variation 18
7 Single Day Variation in Instrument Response 19
8 Phenolic Compounds in Soil Extract 20
9 Matrix Effects on Response Factors 21
10 Wiley Library Search Results 24
11 Effects of Concentration on Mass Spectral Quality 29
12 Effects of Source Temperature on Mass Spectral Quality 30
13 Matrix Effects on 2, 4-D Spectra 31
B1 LC/MS Operating Parameters B-3
B2 Estimated Instrument Detection Limits B-4
B3 Comparative Response Curves B-5
B4 Comparative Response Precision at 1000 ng B-6
B5 Comparative mass Spectra B-7
vi
-------
FIGURES
Number Page
1 Acid Herbicide Target Analytes 2
2 Method I Acid Herbicide Chromatogram (UV detection at 280 nm) 11
3 Method IV Acid Herbicide Chromatogram (UV detection at 280 nm) 12
4 2,4-D Response Curve 17
5 LC of Spiked Soil Extract (UV at 280 nm) 22
6 Particle Beam TIC of Spiked Soil Extract 23
7 Particle Beam TIC of Phenoxyacid Herbicides (2 jig each) 25
8 Comparative Mass Spectra of 2,4-D 26
9 Comparative Mass Spectra of MCPA 27
10 NRSICs of M/Z 141 and 142 from MCPA 33
11 NRSICs from the Particle Beam Mass Spectrum of 2,4-D 35
12 Electron Impact (A) and Methane PCI (B) Particle Beam Spectra of Silvex 35
13 Mechanism of Ion Formation in Particle Beam PCI of Silvex 36
14 NRSICs from the Particle Beam Methane PCI Spectrum of Silvex 37
A1 Particle Beam mass Spectrum of Dicamba A-2
A2 Particle Beam mass Spectrum of 2,4-D A-3
A3 Particle Beam mass Spectrum of MCPA A-4
A4 Particle Beam mass Spectrum of 2,4,5-T A-5
A5 Particle Beam mass Spectrum of Dichlorprop A-6
A6 Particle Beam mass Spectrum of MCPP A-7
A7 Particle Beam mass Spectrum of Silvex A-8
A8 Particle Beam mass Spectrum of 2,4-DB A-9
vii
-------
ABBREVIATIONS AND SYMBOLS
amu atomic mass units
avg average
°C degrees centigrade
2,4 D 2,4-dichlorophenoxyacetic acid
2,4 DB 4-(2,4-dichlorphenoxy)-butanoic acid
DLI Direct Liquid Introduction
ECD Electron Capture Detector
EDL Estimated Detection Limit
EI Electron Impact
eV electron volts
GC Gas Chromatography
HOAc Acetic Acid
HPLC High Performance Liquid Chromatography
id internal diameter
LC Liquid Chromatography
LC/MS Liquid Chromatography/Mass Spectrometry
M Molar
p A Microampere
MAGIC-LC/MS Monodisperse Aerosol Generation Interface for Combined Liquid
Chromatography/Mass Spectrometry
MB Moving Belt
MCPA 4-chloro-2-methylphenoxy-acetic acid
MCPP 2-(4-chloro-2-methylphenoxy)-propanoic acid
/ig microgram
HqIL micrograms per liter
min minute
mL milliliter
/iL microliter
mm millimeter
MS Mass Spectrometry
MW Molecular Weight
m/z mass to charge ratio
ng nanogram
NH4OAc Ammonium Acetate
nm nanometer
NRSIC Normalized Reconstructed Single Ion Chromatogram
PB Particle Beam
PCI Positive Chemical Ionization
psi pounds per square inch
RF Response Factor
RSD Relative Standard Deviation
RT Retention Time
SARM Standard Army Reference Material
sec second
vi ii
-------
ABBREVIATIONS AND SYMBOLS (continued)
S/N
signal to noise ratio
SW-846
Test Methods for Evaluating Solid Wastes
2,4,5-T
2.4,5-trichlorophenoxyacetic acid
TIC
Total Ion Chromatogram
THAMA
Toxic and Hazardous Materials Agency
THF
tetrahydrofuran
torr
torricelli
2,4,5-TP
2-(2,4,5-trichloropenoxy)-propanoic acid
TS
Thermospray
UV
ultraviolet
USEPA
United States Environmental Protection Agency
V
volts
ix
-------
INTRODUCTION
Particle beam (PB) high pressure liquid chromatography (HPLC) mass spectrometry (MS) is a term used
to describe the analytical technique of coupling HPLC to MS by means of the PB interface. This
technique is capable of producing liquid chromatographic separation and Electron impact (El) mass
spectra for polar non-volatile and/or thermally labile organic compounds. PB HPLC/MS is a relatively
new technique using the MAGIC (Monodisperse Aerosol Generation Interface for Combined) LC/MS
technology developed by Willoughby and Browner (1) in 1984. The PB interface is only one of several
LC/MS interfaces that have been developed. A few of the others include direct liquid introduction (DLI),
moving belt (MB), and thermospray (TS) LC/MS interfaces. Both DU and MB type interfaces have
limitations that have caused these methods not to be widely used (2,3). The TS LC/MS interface is
probably the most widely utilized today. The lack of spectral information produced by this soft ionization
technique (4) has limited the use of TS LC/MS to the identification of known organic compounds. The
development of MAGIC LC/MS (and thus PB LC/MS) technology has led to an interface capable of
removing a large portion of the mobile phase from the LC effluent. Once most of the mobile phase has
been removed the analyte particles enter the mass spectrometer ion source where they are vaporized
and subsequently ionized. The ionization of the analyte leads to electron impact (El) type spectra which
contain substantial structural information (5). It is the El mass spectral capability of the PB interface that
has captured the interest of many scientists in the environmental chemistry field.
The development of LC/MS interfaces arose from the need to analyze an enormous number of polar non-
volatile and/or thermally labile compounds which cannot pass through or have difficulty in passing
through a gas chromatograph (GC). Although LC systems alone are capable of analyzing these types
of compounds, conventional LC detectors do not provide the analyte identification capabilities of a mass
spectrometer.
The phenoxyacid herbicides are one class of compounds that a LC/MS method for identification and
quantification would be very useful. Presently these compounds are analyzed for by method 8150 and
8151 of the SW-846. These methods involve hydrolysis and derivatization with diazomethane before
analysis by GC with an electron capture detector (ECD). The derivitization step is both time consuming
and dangerous; thus a method eliminating its use would be advantageous.
This report describes the application of PB LC/MS to the analysis of phenoxyacid herbicides and other
chlorinated acid herbicides. The target analytes examined were: Dalapon; Dicamba; Dichlorprop;
Dinoseb; MCPA; MCPP; 2,4-D; 2,4-DB; 2,4,5-T; 2,4,5-TP (Silvex); and the butoxy ethyl esters of 2,4-D and
2,4,5-T. Factors affecting both the chromatography and the response characteristics (including spectral
quality) of the acid herbicides were investigated. The acid herbicide target analytes are displayed in
figure 1.
1
-------
h3c cci2cooh
Dalapon
COOH
CIv^N^OCH.
Wc, '
Dicamba
CH,
O-CH-COOH
.CI
CI
Dichlorprop
OH
CH.
°2NY^VCH_CH2~CH3
V
no2
Dinoseb
h3c
o-ch2cooh
ci
MCPA
fH»
O-CH-COOH
h3c ^
CI
MCPP
0-CH2COOH 0-(CH2)3-C00H 0—CHjCOOH
1 Cl c,nA
yLcl
CI
2. 4.5 -T
2, 4 - DB
i ¦*
O-CH-COOH
CI
2, 4, 5 - TP (Silvex)
Figure 1. Acid Herbicide Target Anafytes
2
-------
CONCLUSION
A rugged and reliable method to identify and quantitate acid herbicides in environmental samples based
on particle beam LC/MS technology does not appear feasible at this time. A number of fundamental
issues need to be addressed before a reliable method can be developed. Among these issues are poor
sensitivity, non-linear response, response drift, and variation in spectral quality.
With the exception of Dalapon and Dinoseb, the instrument detection limits for the acid herbicides are
in the 100 ng to 500 ng range as determined under full-scan operation of the mass spectrometer.
Dalapon and Dinoseb were not detected. Based on SW-846 extraction procedures and assuming 10
li L injection volumes, the observed instrument detection limits would result in method detection limits of
50 to 250 iig/L These values are comparable to 8150 detection limits for MCPA and MCPP, but are
considerably higher than the method detection limits for the remaining acid herbicides. Selected ion
monitoring of the acid herbicides improved detection limits by 10 to 50 times, but identification based
on full scan mass spectra is lost.
Response curves for most of the acid herbicides exhibited non-linear behavior over the range 200 ng
to 2000 ng. Response tended to increase with increasing concentration. This effect was more
pronounced at higher concentrations and was observed in both total ion chromatograms and extracted
ion chromatograms. Accordingly, accurate results cannot be expected over a range greater than 10
when using linear calibration models. The addition of the volatile buffer, ammonium acetate, to the
mobile phase enhanced signal intensity but did not improve the linear response range.
The response factors for the acid herbicides exhibited relative standard deviations (RSDs) of roughly 50%
over the course of a six day period in which 11 single level standards were analyzed. This data indicates
measurement of the herbicides requires daily instrument calibration or recalibration whenever the
instrument is placed in the standby mode. Response factors for herbicide standards analyzed
consecutively (n=3) at a single level (1000 ng) exhibited RSD values of 10% or less. This data indicates
the best precision observed. Data from an interlab study revealed mixed results in terms of precision
(6). The RSD values at high concentration (2000 ng) averaged 6%, but averaged 22% at lower
concentrations (200 ng to 1000 ng) with some response factors varying by more than 50%. Although
precision data from the interlab study was acquired under different mobile phase conditions than the
single level precision data, it does suggest that precision at lower concentrations may not be acceptable.
Spectral quality was found to be dependant on the amount of analyte reaching the ion source of the
mass spectrometer. At high levels of sample (2 |ig), the El mass spectra are of library matchable quality
for Tive of the phenoxyacid herbicides examined. However, at lower sample amounts the mass spectra
show an increase in the relative abundance of thermal decomposition ions. These decomposition ions
represent non-classical gas phase El mass spectra and cause misidentification by the data system. At
lower concentrations, the mass spectra are not suitable to permit unambiguous acid herbicide
identification. The addition of a volatile buffer,ammonium acetate, suppressed the relative abundances
of the thermal decomposition ions to some extent. Varying the ion source temperature showed that the
best spectra are produced at low (200° C) source temperatures. A serious draw back to operating the
ion source at low temperatures was strong surface adsorption and extensive peak tailing. The higher
the ion source temperature (to 300° C) the sharper the peaks become but the amount of thermal
decomposition also increases. A source temperature of 250° C was found to be a compromise between
3
-------
tailing and decomposition. Investigation into the cause of thermal decomposition ions leads us to the
hypothesis that these ions are formed through interactions of analyte particles with the ion source surface
resulting in decomposition to a more thermally stable and volatile species followed by desorption and
subsequent ionization. The extent to which these events occur are variable; depending on analyte
concentration, source temperature, and source cleanliness, and may be the pnncipal cause of non-linear
response over an extended range and response vanation at low concentrations.
However, results obtained from an interlab study on the measurement of acid herbicides by LC/MS (6)
do not entirely support these conclusions. Of the two other tabs reporting particle beam results, one,
using an Extrel ThermBeam, submitted data exhibiting problems similar to those described in this report
and the other, using Hewlett Packard particle beam equipment, reported linear response over the range
200 ng to 2000 ng and did not observe thermal degradation' spectra at any concentration level. These
results suggest that while a particle beam based method has potential for the measurement of acid
herbicides, it is not yet rugged and requires further work to understand the disparate results obtained
by different laboratories.
4
-------
RECOMMENDATIONS
Several problem areas have been identified with the use of particle beam LC/MS to measure acid
herbicides. Mass spectral variation may be the most difficult problem to overcome. However, other
areas of difficulty may be improved on more readily. The following section discusses several approaches
that may prove useful.
The chromatographic behavior of the acid herbicides has been examined on several reversed phase
columns with a variety of solvent combinations. An acetic acid/ammonium acetate/methanol based
solvent system on a Spherisorb C-8 column gave the best performance in terms of both analyte
separation and overall instrument behavior. The chromatographic performance is sensitive to injection
volume and sample solvent, particularly for early eluting analytes. To gain flexibility in the size of the
injection volume, the sample solvent must be compatible with the elution solvent with respect to elution
strength. Otherwise, injection volumes will be severely restricted.
Some improvement in detection limits may be achieved by summing over all isotopic ions associated with
the base peak rather than quantitating on the base peak alone. Further enhancements may be achieved
by incorporating specialized injection techniques. One of these techniques, well suited to liquid
chromatography, is on column preconcentration used in conjunction with a switching valve. In addition,
larger sample size or more extensive concentration may be incorporated into the sample preparation to
improve overall method detection limits.
Inaccuracy resulting from inappropriate calibration may be minimized by operating over a narrower
calibration range. For the acid herbicides, linear calibration models may be used provided the calibration
range is less than a factor of 10. For an extended range, a non-linear calibration, must be employed
to maintain acceptable accuracy. In addition, the use of an internal standard should be investigated as
a measure to compensate for possible drift in instrument response.
Sample preparation techniques have not been directly examined as part of an overall analytical method
during these studies. However, a feasibility study using C-18 solid phase extraction cartridges to extract
acid herbicides from solid samples was conducted (7). This study determined 90% to 100% recoveries
from sandy soils and 50% to 60% recoveries from fly ash as measured by HPLC with UV detection.
These results suggest that solid phase extraction is a suitable alternative to the ether based extraction
procedures of Method 8150. Literature reports describe excellent herbicide recoveries from aqueous
samples using solid phase extraction (8).
The use of HPLC to separate and quantitate acid herbicides as the free acids is a sound concept. HPLC
avoids the complicating and time consuming features of diazomethane derivatization now required by
Method 8150. Although detection by particle beam mass spectrometry may not prove suitable,
alternative means of detection and quantitation are available. These include thermospray mass
spectrometry and HPLC with UV/VIS or electrochemical detection. A rugged and reliable method, with
detection limits comparable to Method 8150, based on solid phase extraction with HPLC/UV analysis
could be developed in a relatively short period of time. UV detection by diode array would, in addition,
permit an element of identification based on UV spectral features.
5
-------
EXPERIMENTAL
The instruments used in these experiments were all manufactured by Hewlett-Packard. The initial
chromatographic separations were developed on a 1090L liquid chromatograph equipped with an
autosampler and a diode array detector. The LC system was controlled by a Hewlett-Packard HPLC
ChemStation. The PB LC/MS consisted of a 1090L Liquid chromatograph equipped with an autoinjector
and filter photometric detector. The LC was connected to a 5988A mass spectrometer by a 59980A
particle beam interface. The LC was controlled by a local user-interface while the MS was controlled by
a HP 59970 MS ChemStation.
LC flow rates of either 0.25 mL/min or 0.4mL/min were used. Several mobile phases and LC columns
were considered and are discussed in the chromatography portion of the results and discussion section.
Routine PB parameters were: nebulizer setting of 12, nebulization helium pressure of 30-50 psi,
desolvation chamber temperature of 45° to 55° C, and PB probe distance to the source of 0.5 mm. The
PB desolvation chamber vacuum pressure was estimated at 200 torr by the instrument manufacturer.
The pressure in the first stage of the momentum separator was typically 10 torr and that of the second
stage was typically 0.5 torr as measured by Hastings-Raydist gauges. The mass spectrometer ion
source for these studies was slightly modified. First, a stainless steel plug was inserted into the GC inlet
of the source. Second, the PB inlet to the source was drilled out to a larger diameter by the instrument
manufacturer. Except for one set of experiments the ion source was operated at 250° C. A typical MS
operating pressure of 1.2 x 10 s torr was measured by a Bayard-Alpert ion gauge tube. The mass
spectrometer was run in El mode (except for one experiment) with a filament emission current of 300 jiA
and an electron energy of 70 eV. The MS electron multiplier was a Galileo channeltron and was typically
operated at 2200 V. The MS system was tuned to maximize the m/z 219 ion of PFTBA introduced
through a reservoir on the PB transfer tube.
The acid herbicide standards were pure compounds ( > 97%) obtained from the U.S. EPA Repository
(Research Triangle Park, North Carolina). The compounds were subsequently diluted with acetonitrile.
6
-------
RESULTS AND DISCUSSION
CHROMATOGRAPHY
The liquid chromatographic behavior of the acid herbicides was investigated with a number of C-8 and
C-18 reversed phase columns and several acidic mobile phases. In most instances, the mobile phase
was made acidic with acetic acid to suppress dissociation of the herbicide acid functionality. Without
an acidic component in the mobile phase, the acid herbicides were poorly retained with considerable
coelution and peak splitting. A 1% acetic acid solution was found to be sufficient to suppress
dissociation. Lesser amounts of acetic acid may also be sufficient but these were not investigated.
The general results of the chromatographic studies are summarized in table 1. Nominal retention times
for each acid herbicide are listed for each of the chromatographic conditions examined.
Chromatographic conditions are identified with a roman numeral as described in more detail further in
the table.
Methanol, acetonitrile, and tetrahydrofuran (THF) were examined as mobile phase organic modifiers.
Each of these solvents exhibited differences in selectivity and in some instances resulted in reversal of
elution order. The methanol based mobile phase (Method I) exhibited superior selectivity for the acid
herbicides in comparison to the acetonitrile (Method II) and THF (Method III) based solvent systems. The
methanol based system was capable of separating 2,4-D and MCPA whereas acetonitrile and THF were
not. Methanol also gave better resolution between Dichlorprop, MCPP, and 2,4,5-T although these were
not baseline separated.
The methanol/acetic acid solvent system was examined on five different C-18 columns: Supelcosil LC-18,
Vydac 201 TP, Supelcosil LC-18-DB, Adsorbosphere HS, and a Spherisorb S5 ODS1. Chromatographic
behavior was also examined on two C-8 columns: Spherisorb S3 C-8 and a Microsorb C-8. Although
some differences in selectivity were observed, the Supelcosil LC-18 gave the best overall separation of
Dichlorprop, MCPP, and 2,4,5-T. The early LC/MS performance studies and an interlab study (6) were
conducted with the 1% acetic acid/methanol mobile phase on the Supelcosil LC-18 (2.1 x 150 mm)
column (Method I). A representative chromatogram using method I conditions is shown in figure 2.
Subsequent chromatographic studies with ammonium acetate added to the 1% acetic acid/methanol
solvent system gave baseline separation of the acid herbicides on the Spherisorb C-8 column (Method
IV). A representative chromatogram using Method IV conditions is shown in figure 3. Baseline
separation was not observed on the Supelcosil LC-18 column but the chromatography was considerably
improved with the ammonium acetate buffer (Method V). A buffer system composed of ammonium
formate and 1% formic acid was also examined but was not capable of separating Dichlorprop from
2,4,5-T (Method VI).
The ammonium acetate/methanol solvent system was further modified to accommodate the particle beam
interface. Because of the high initial aqueous content of the mobile phase a substantial loss of sensitivity
for the acid herbicides was observed. To compensate for loss of sensitivity, the flow rate was reduced
to 0.25 mL/min and the nebulizer helium pressure increased to 50 psi (Method VII). The decreased flow
rate and higher helium pressure resulted in sensitivity equivalent to previous observation. These
chromatographic conditions were used for all subsequent studies with the interface. The herbicide esters
7
-------
could be chromatographed along with the free acids by extending the gradient time to deliver
proportionately higher amounts of methanol.
We have observed peak distortion of early eluting components (< 10 mm) when sample solvent strength
is considerably different from the eluting solvent strength. For the 4.6 mm id columns, distortion was
observed with injection volumes greater than 25 |il_ This effect was more pronounced on the narrow
bore columns (ca. 2 mm id). Injection volumes greater than 4 |iL resulted in peak distortion on the
narrow bore columns. To give more latitude in injection volume size, the sample solvent must closely
match the strength of the eluting solvent.
TABLE 1. HPLC PARAMETERS
t
A. Retention Times (minutes)
I
II
III
IV
V
VI
VII
Dalapon
ND
ND
ND
ND
ND
ND
ND
Dicamba
2.76
3.85
7.02
1.94
4.42
7.17
2.63
2,4-D
5.92
6.92®
10.242
6.85
11.17
10.75
7.98
MCPA
6.85
6.92®
10.242
8.50
12.56
11.67
10.19
Dichlorprop
9.221
10.391,2
12.84s
10.32
14.371
13.602
11.82
2,4,5-T
9.66'
11.05'
14.21
9.59
14.001
13.602
13.12
MCPP
9.98
10.3912
12.842
11.75
15.48
14.18
15.37
2,4-DB
12.70
13.46
13.33
15.02
18.70
15.22
21.16
Silvex
13.77
14.91
16.64
12.59
16.60
15.88
16.91
Oinoseb
17.56
18.36
'partially resolved
2co-elute
NDnot detected
8
-------
Conditions
Column: Supelcosil LC-18 5 jim 2.1 x 150 mm
Flow:
Gradient:
0.4 mL/min Temperature: ambient
Time (min)
initial
2
12
18
1 % Acetic Acid
50%
50%
40%
0
1 % HOAc in methanol
50%
50%
60%
100%
Column:
Flow:
Gradient:
Same as I
0.4 mL/min
Time (min)
initial
10
18
Temperature: ambient
1 % Acetic Acid 1% HOAc in acetonitrile
60% 40%
60% 40%
20% 80%
Column:
Flow:
Gradient:
Same as I
0.4 mL/min
Time (min)
initial
4
24
Temperature: 37.5° C
1 % Acetic Acid
70%
70%
50%
tetrahvdrofuran
30%
30%
50%
Column: Spherisorb S3 C-8 5 |im 2x100 mm
Flow: 0.4 mL/min Temperature: 37.5° C
Gradient: Time (min) 1% Acetic Acid +0.01 M NH.OAc methanol
initial 75% 25%
2 75% 25%
15 25% 75%
Column: Supelcosil LC-18 C-8 5 jim 2.1 x 250 mm
Flow: 0.5 mL/min Temperature: 40°C
Gradient: Time (mini 1% Acetic Acid +0.01 M NH.OAc
initial 75%
2 75%
15 25%
methanol
25%
25%
75%
Column:
Same as Method IV but with 1% formic acid +0.01 M ammonium formate buffer
-------
Column: Same as IV
Flow: 0.25 mL/min Temperature: 50° C
Gradient: Time (min) 1% Acetic Acid +0,01 M NH.OAc 1% HOAc in methanol
initial 75% 25%
2 75% 25%
25 40% 60%
10
-------
2-
0H
2
4
6
8
T i me C m i n . )
Figure 2. Method I Acid Herbicide Chromatogram (UV detection at 280 rim).
11
-------
D
(E
E
120-
100-
80-
G0-
40-
20-
oc
m
r
a;
o
a
i
(VI
a:
Q.
u
r
0-'-J ¦ ¦
j
»-
i
in
m
(VI
~QT
O
a:
a.
a
o
_i
X
u
JU
a.
a.
o
E
X
u
>
_l
n
in
m
a
i
OJ
G 8 10
T i me C m i n . )
1 4
Figure 3. Method IV Acid Herbicide Chromatogram (UV detection at 2B0 nm).
12
-------
RESPONSE CHARACTERISTICS
Detection Limits
The detection limits for an on-column injection of the phenoxyacid herbicides are listed in table 2. Most
detection limcts were estimated using LC Method Vil. The limits of detection were estimated from a full
scan total ion chromatogram (TIC) at a concentration near these limits. The individual peaks were
extrapolated to give a signal to noise ratio of three. Slightly better limits of detection can be expected
by using the extracted ion chromatograms associated with the quantitation ion of each of the
phenoxyacid herbicides. Selected ion monitoring of the herbicides improved detection limits by 10-50
fold.
Table 2. Estimated Detection Limits (3 x S/N)
Compound
Detection Limit (na)
Dalapon
ND
Dicamba
520
2,4-D
130
MCPA
260
2,4,5-T
320
Dichloroprop
130
MCPP
400
Silvex
130
2,4-DB
120
Dinoseb'
> 2500
2,4-D butoxy"
ethyl ester
50
2,4, S-T butoxy'
ethyl ester
50
ND Not detected
' LC Method I
13
-------
Table 3. Effect of Ammonium Acetate on Detection Limit
Det. Limit (ng)
without NH.OAc
Det. Limit (ng)
with NH.OAc
2,4-D 1. 139
2. 143
3. 155
153
142
112
avg. 146
136
MCPA 1. 274
2. 261
3. 227
208
139
157
avg. 254
167
Table 3 displays the limits of detection for two phenoxyacid herbicides with and without the presence
of a volatile buffer added to the mobile phase. The mobile phase without a volatile buffer consisted of
50% methanol and 50% water with 1% acetic acid (HOAc). The mobile phase with a volatile buffer was
the same as above with the addition of 0.01 M ammonium acetate (NH40Ac) to the water/acetic acid
solvent. The herbicides 2,4-D and MCPA were chosen as representative compounds based on structural
similarities to the other phenoxyacid herbicides examined in this study. The two herbicides listed in table
3 were analyzed by flow injection (bypass of analytical column). Again, the limits of detection were
calculated from a TIC extrapolated to give a signal to noise ratio of three. The limits of detection for each
mobile phase system (with and without buffer) are the results of averaging three separate injections.
These results indicate a slight improvement (7%) in detection limit for 2,4-D and a substantial
improvement (34%) for MCPA with addition of ammonium acetate to the mobile phase. Although the
improvement for 2,4-D is within the experimental error observed for the PB LC/MS system, the
improvement in detection limit for MCPA appears to be the result of addition of the buffer to the mobile
phase. Differential signal enhancement in the presence of ammonium acetate has been reported by
another laboratory (9).
Response Curves
A four point calibration curve for eight of the target analytes was run using LC method VII (ammonium
acetate). Response factors at each point were determined by dividing quantitation ion area counts by
the amount injected. Response factors were determined over the range 200 ng to 2000 ng. Results are
summarized in table 4. Two of the target analytes, Dicamba and MCPP, exhibited near linear response
over the range examined. The response to Silvex and 2,4-DB was distinctly non-linear. Response to
the remaining target analytes was intermediate with relative standard deviations between 15% and 25%.
14
-------
Table 4. On-Column Response Factors
Response Factors
Quant.
Avg.
ion
200 na
500 na
1000 na
2000 na
RF
RSD ft
Dicamba
173
111
99
112
112
108
6
2,4-D
162
936
966
1325
1509
1184
24
MCPA
141+142
1207
1098
1350
1596
1313
16
2,4,5-T
196
429
373
454
625
470
23
Dichlorprop
162
1417
1347
1590
1965
1580
17
MCPP
141+142
1349
1339
1282
1308
1320
2
Silvex
196
1108
1409
2070
2424
1753
34
2,4-DB
162
1763
2413
3617
4696
3122
42
The data summarized in table 4 represents the best observed in terms of response factor RSDs. Other
calibration curves over a similar range of concentrations but with different chromatographic conditions
exhibited considerably higher RSDs for most of the target analytes. Examination of accumulated
response curve data reveals two trends. First, response curves for most of the target analytes were non-
linear over the range 200 ng to 2000 ng. Secondly, response factors tend to increase with increasing
analyte amounts.
To investigate the effect of ammonium acetate on response, two analytes were selected for further study
The response curves for 2,4-0 and MCPA were determined by flow injection analysis. Levels of 200,500,
1000, and 2000 ng were obtained by injecting 1, 2.5, 5 and 10 |iL respectively of a 200 ng/|iL solution
of each compound. Response factors (RFs) were calculated by extracting the quantitation ions from the
full-scan (80-400 amu) TIC and summing the area under the peak. For 2,4-D the quantitation ion was
m/z 162. For MCPA, mJz 141 and m/z 142 were summed together for quantitation. Each point of the
curve was determined by averaging three separate injections. To investigate the effects that the volatile
buffer, ammonium acetate, has on the response characteristics of these compounds, the response
curves were generated with and without the buffer. Table 5 summarizes the results for these two
compounds. Listed in this table are the average of three injections at each level, average RF, and the
%RSD of those injections. Also the average response factors, RF(T), for the entire curve and associated
%RSD(T) are given in this table. An increase in the average response factor for the entire curve is seen
for both 2,4-D and MCPA upon the addition of ammonium acetate. This substantiates previous evidence
of signal enhancement upon addition of a volatile buffer to the LC mobile phase. Examination of the
%RSD(T)s for these compounds supports previous observation. First, the response curves were not
linear over the range examined and, secondly, a definite trend was observed wherein the response
factors increase with increasing amounts of analyte. In addition, the %RSD(T) are not effected by the
addition of the volatile buffer indicating that linearity is not improved.
15
-------
Figure 4 illustrates the non-linear response to 2,4-D. This figure shows the plot of RF verses the absolute
amount of 2,4-D injected. It clearly indicates that the RFs for 2,4-D are not linear (a horizontal line would
indicate linearity). Most of the phenoxyacids examined exhibit this type of response behavior. The
apparent linear response range for the phenoxyacids is somewhat less than a factor of 10. A polynomial
type curve fit might be a more appropriate calibration model for these compounds over an extended
range.
Table 5. Effects of Ammonium Acetate on Response
50% methanol
50% water with 1% HOAc
ComDound
Amount (no)
RF
%RSD
* RE
%RSP
2,4-D
200
219
4.6
321
7.1
m/z 162
500
274
2.2
437
1.6
1000
362
2.3
574
6.0
2000
491
6.3
725
2.2
RF(T)
- 337
RF(T)
- 514
%RSD(T)
- 35.3
%RSD(T)
- 33.9
MCPA
200
460
7.9
768
12.6
m/z 141
500
593
5.8
965
8.7
and 142
1000
596
7.5
1206
7.9
2000
779
11.9
1399
3.8
RF(T)
- 607
RF(T) -
1085
%RSD(T)
- 21.6
%RSD(T) -
25.4
50% methanol
50 % water with 1% HOAc
and 0.01 M NH40Ac
16
-------
520
500
480
460
440
420
400
380
360
340
320
300
280
260
240
220
200
2,4 —D 50:50 ME0H/1%H0AC
02 04 06 08 1 12 14 16 18 2
(Thousands)
amount (ng)
-------
REPRODUCIBILITY
The data in table 5 indicates that single level RFs were reproducible with a maximum RSD value of 12.6%
and most RSD values below 10%. Each of these RSD values were calculated from three consecutive
flow injections at the levels indicated. Table 6 lists the RSD values at the 1000 ng level on four separate
days over a six day period during which the phenoxyacid herbicide LC/MS interlaboratory study was
conducted (6). Eleven on-column injections were made using LC method I. The quantitation ions are
listed in the table. Every attempt was made to return the instrument to the same operating condition
each day. This table indicates substantial vanation in single level reproducibility over the time frame
examined.
Table 6. Day to Day Response Variation (Single Level at 1000 ng)
Quantitation
ComDOund
ion fm/z)
%RSD
Dicamba
203
43.5
2,4-D
162
43.2
MCPA
142
44.0
Dichloroprop
162
51.2
2,45-T
196
46.0
MCPP
142
43.0
2,4-DB
162
43.6
Silvex
196
51.4
(n = 11)
18
-------
Table 7 displays the results of 4 single level injections of the phenoxyacid herbicides (excluding dicamba)
over the course of one day. The four single level 1000 ng standards demonstrated relatively stable
response over the course of about 2.5 hours that it took to analyze them.
Table 7. Single Day Variation in Instrument Response
Quantitation
Compound ion (m/z) %RSD
2.4-D 162 1Z5
MCPA 142 8.4
Dichlorprop 162 7.9
2,45-T 196 3.2
MCPP 142 11.1
2,4-DB 162 3.1
Silvex 196 1.7
(n = 4)
The LC/MS conditions were the same as those for LC Method I. The RSO values in tables 6 and 7 show
that the RFs for the herbicides run consecutively on one day are substantially more reproducible than
the RFs generated over several days. Results taken from the interlab study (6) exhibited low RSD values
at high concentration but considerably higher RSO values at lower concentrations. Data from this study
was acquired in triplicate with LC Method I. At 2000 ng injected, the RSD values averaged 6% with a
low of 1.4% for 2,4-DB and a high of 14% for MCPP. At 1000 ng, RSD values averaged 17% with a low
of 1.5% for 2,4-DB and a high of 49% for MCPA. At 200 ng, RSD values averaged 29% with a low of
3.5% for 2,4-DB and a high of 105% for Silvex. These results suggest that instrument response is stable
at higher concentrations but at concentrations approaching the detection limit instability can be expected.
MATRIX EFFECTS
An expenment was designed to examine matrix effects on phenoxyacid herbicide response. Instrument
response to a mixed herbicide standard solution (200 ng/iiL) was compared to the response observed
for acid herbicides in a soil extract and a soil extract containing phenolic compounds. The soil extracts
were prepared from a U.S. Army standard soil obtained from the THAMA SARM repository (DAA05-81-
A284).
19
-------
Briefly, a 10 g soil sample was sonicated with 10 mL acetonitrile and then centrifuged to settle the solids
A 5 mL aliquot of the supernatant was passed through a 0.5 g C-18 solid phase extraction cartridge and
the eluent concentrated to a final volume of 2 mL Two samples were prepared from the resulting soil
extract. Sample 1 was prepared by spiking the acid herbicides into 500 |iL of soil extract and diluting
to a final volume of 1 mL with an aqueous solution containing 1% acetic acid and 0.01 M ammonium
acetate. Sample 2 was prepared in the same way except that 14 phenolic compounds and benzoic acid
were spiked in addition to the acid herbicides. Components in the phenol spike are listed in table 8.
Table 8. Phenolic Compounds Spiked in Soil Extract
Compounds
1.
benzoic acid
9.
2,4-dinitrophenol
2.
p-chloro-m-cresol
10.
2-nitrophenol
3.
2-chlorophenol
11.
4-nitrophenol
4.
o-cresol
12.
pentachlorophenol
5.
p-cresol
13.
phenol
6.
2,4-dichlorophenol
14.
2,4,5-trichlorophenol
7.
2.4-dimethylphenol
15.
2,4,6-trichlorophenol
8.
4,6-dinitro-o-cresol
The concentration of the acid herbicides and phenols in the final sample volumes were the same as the
acid herbicide concentration in the standard solution used for comparative purposes (200 ng/pL). The
two extracts and the standard solution were analyzed using LC Method VII with a 5 pL injection volume
(1000 ng each analyte on column), The standard solution was measured in triplicate; two prior to the
extract samples and one after. The measured response factors from each of the standards and the soil
extracts are listed in table 9 in the order in which they were analyzed.
20
-------
Table 9. Matrix Effects on Response Factors
Response Factors
soil +
quant.
STD-1
STD-2
soil
phenols
STD-3
ComDOund
ion (m/z)
1000 no
1000 na
1000 na
1000 na
1000 na
Dicamba
203
85
112
116
137
80
2,4-D
162
1111
1325
1043
1036
1035
MCPA
142
694
845
671
666
698
2,4,5-T
196
547
454
429
462
437
Dichloroprop
162
1422
1590
1473
1414
1423
MCPP
142
901
979
864
868
893
Silvex
196
1749
2070
1787
1730
1806
2,4-DB
162
3126
3617
3245
3079
3139
Comparing response factors from the soil extracts against those from the standards indicates the soil
matrix has no significant effect on instrument response for most of the target analytes. With the
exception of Dicamba and MCPP, response to the acid herbicides in the soil extracts were within one
standard deviation of the measured responses from the standards. Soil extract response factors for
MCPP were just slightly greater than one standard deviation. The response to Dicamba did appear to
be effected. However, this observation may be due to inaccurate integration and poor peak shape
resulting from sample solvent mismatch with the elution solvent
In conducting the matrix effect experiments, none of the 14 phenols or benzoic acid exhibited
appreciable response on the PB LC/MS system. Figure 5 displays the UV chromatogram of sample 2
(soil extract, phenols and herbicides). Figure 6 is the corresponding PB LC/MS TIC showing that the
phenols were not detected to any appreciable extent
In conclusion, it appears that matrix components from the extract and the soil extract plus phenols do
not effect the response characteristics by a significant amount Also it would appear that the phenols
used in this study are not amenable to analysis by PB LC/MS.
21
-------
Figure 5. LC of Spiked Soil Extract (UV at 280 nm).
22
-------
IS
T t ma (mln.)
Figure 6. Particle Beam TIC of Spiked Soil Extract.
23
-------
SPECTRAL QUALITY
There has been evidence of variations in spectral quality from this laboratory and others (10). Efforts
were made to characterize the variations in spectral quality and will be described in the following section.
Figure 7 displays a TIC of eight phenoxyacid herbicides from an on-column injection of 2 |ig per
component. LC Method VII was used to obtain this chromatogram. The mass spectrum for each peak
is included in appendix A (figures A1 through A8). These spectra were obtained with a recently cleaned
ion source volume and at the highest absolute amount, 2|ig, used in the calibration curves. Table 10
summarizes the results of a library search routine using a probability-based matching (PBM) scheme from
the Wiley database as provided with the HP MS ChemStation. All eight phenoxyacid herbicides
investigated had entries in the Wiley library. Presumably the reason 2,4,5-T, dichlorprop, and 2,4-DB
were not matchable was because the molecular ions in the PB spectra were not as intense as those in
the library spectra It is not surprising that some differences exist between PB generated spectra and
library spectra produced by solid probe analysis.
Table 10. Wiley Library Search Results
Compound Library Match PBM%
Dicamba
yes
83
2,4-D
yes
87
MCPA
yes
96
2,4,5-T
no
-
Dichloroprop
no
-
MCPP
yes
98
Silvex
yes
90
2,4-DB
no
-
The spectra obtained from the TIC in figure 7 were obtained under the "ideal* conditions of a recently
cleaned ion source and high analyte concentration. Figures 8 and 9 give examples of the spectral
variations observed with these types of compounds. These figures are of two representative herbicides,
2,4-D and MCPA. In both figures, spectra A are the library matchable spectra obtained from the TIC in
figure 7 and spectra B represent some earlier work in which the spectral quality was considered very
poor. In figure 8 spectrum B, the ions at m/z 128, 139 and 168 have higher relative abundances
compared to the base peak, m/z 162, than those ions in spectrum A. Also note the absence of a
molecular ion at m/z 220 in spectrum B. In figure 9, spectra A and B also show striking differences.
First, the molecular ion at m/z 200 is absent in spectrum B. Second, m/z 107 has become the base
peak in spectrum B. Finally, the ratio of m/z 141 to m/z 142 is about 2:1 in spectrum A and 1:4 in
spectrum B. When spectra of the quality shown in B of figures 8 and 9 were obtained, we proposed that
the ions m/z 128, 139 and 168 for 2,4-D were generated by thermal" decomposition of the analyte.
Some evidence for decomposition is the non-classical electron impact loss of 52 mass units (thought to
be loss of HOC I) from 2,4-D shown in figure 8 spectrum B producing m/z 168.
24
-------
3.0E+5-
m
a
i
2.SE+5-
2.0E+5-
•
o
c
«
u
c
3
ji
ac
1.5E+S
1.0E+5-
cr
m
n
tE
O
5.0E+4-
0.0E+0
O
I
cvi
Q.
o
az
a.
o
at
o
_i
zc
o
CE
a.
o
VJ
I
in
cvi
VJ
x
LiJ
>
»—I
cn
OJ
a_
o_
o
—r-
4
—r-
6
10 12 14
T1 me (mln. )
¦—i—
16
1—i—
IS
•—t—
20
22
24
Figure 7. Particle Beam TIC of Phenoxyacid Herbicides (2 |ig each).
25
-------
2.4-D
MW 220
1B0 180
Ma»«/Ch»rflB
1200
1000
160 180
Mtaa/Charga
Figure 8. Comparative Mass Spectra of 2,4-D.
26
-------
•
u
c
a
-a
c
3
jO
a:
18000:
16000
14000
I2000d
10000
0000
6000
4000
2000
0
91
'"¦c"
MCPR
MW 200
107
/
|l-—I1 I¦ 111' 11
/
14 1
125
/
'I"'*!"'
159
X
100
120
140
157
I. 11111, ii t
200
/
182
\ /
160 180
Hw/Ch»rg»
220
/ .
4^
224
200
220
245
/
—i—•
240
0
a
c
0
u
c
3
-Q
It
1800:
1600
1400:
1200
1000
800
600
400*1
200
0
07
142
\
144
120
140
160 180
Mas«/Charge
200
220
B
240
Figure 9. Comparative Mass Spectra of MCPA.
27
-------
The other phenoxyacid herbicides investigated exhibited evidence of thermal* decomposition similar to
those described above for 2,4-D and MCPA. We also observed that other compounds such as
polynuclear aromatics (PNAs), and the esters of 2,4-D and 2,4,5-T did not exhibit the poor quality spectra
observed for the acid herbicides. We attributed the tendency of the acid herbicides to decompose in
the ion source to the acid functionality.
Due to the difficulties experienced with spectral quality, technical representatives from Hewlett Packard
were asked to verify instrument performance. Hewlett-Packard found an air leak in the desolvation
chamber and replaced the chamber. In addition, a modified source block was installed (see experimental
section). HP representatives concluded that the problem of poor spectral quality was due to LC column
bleed. However, subsequent flow injection analysis found that the problem of thermal decomposition
with the acid herbicides persisted without an analytical column. In subsequent work, we observed that
on changing from commercially available HPLC grade water to water produced with an in-house water
purification system the intensity of the thermal decomposition ions was reduced but not eliminated.
Experiments were performed to determine what additional factors affected the spectral quality of the
phenoxyacid herbicides. Two representative acids, 2,4-D and MCPA, were analyzed with and without
NH40Ac at two concentration levels, 500 ng and 2000 ng. Analytes were introduced by flow injection.
Ions indicative of thermal decomposition for 2,4-D and MCPA are listed in table 11 as a function of
concentration.
All values in this table are expressed as the abundance of a given ion relative to the base peak (base
peak abundance is listed as 100 in table 11). Two general trends were observed in this data First, the
relative abundances of certain ions appear to be dependant on the amount of the compound injected.
For example, the mJz 128 ion of 2,4-D decreased in abundance from 98% at the 500 ng level to 62% at
the 2000 ng level for mobile phase condition #1. For MCPA, the ion at m/z 141, which is the base peak
under ideal conditions (see figure 9A), increased in abundance from 38% at 500 ng to 54% at the 2000
ng level. Second, the addition of the volatile buffer, ammonium acetate, appeared to suppress the
formation of thermal decomposition ions for both compounds and at both levels.
28
-------
Table 11. Effects of Concentration on Mass Spectral Quality
Relative Abundance of Thermal Degradation Ions
Mobile
Ions (m/z)
Compound
Amount (na)
Phase
128
139
162
168
220
2,4-D
500
1
98
79
100
67
12.2
500
2
74
66
100
55
9.2
2000
1
62
63
100
48
8.9
2000
2
47
49
100
35
9.5
107
141
142
200
MCPA
500
1
100
38
82
14.4
500
2
100
43
84
16.8
2000
1
100
54
92
21.4
2000
2
100
76
92
33.6
mobile phase #1: 50% methanol
50% water with 1% HOAc
mobile phase #2: 50% methanol
50% water with 1% HOAc with .01M NH40Ac
29
-------
Table 12. Effects of Source Temperature on Mass Spectral Quality
Compound
TemDerature (C)
128
139
162
168
220
2,4-D
200
22
28
100
18
18.8
225
22
21
100
14
14.0
250
37
28
100
17
14.5
300
43
32
100
18
15.3
107
141
142
200
MCPA
200
47
100
54
47.8
225
90
100
85
39.3
250
90
100
86
40.1
300
100
77
87
31.6
Table 12 lists the relative abundances of diagnostic ions from 2,4-D and MCPA as a function of ion
source temperature. The mobile phase was 100% acetonitrile and both compounds were introduced into
the PB LC/MS system by flow injection (2000 ng). For MCPA at 200° C we observed ion abundances
that were library matchable. Upon heating the ion source towards 300° C, the ion at m/z 107 became
the base peak and m/z 142 became larger than m/z 141. Similar behavior was observed for 2,4-D. For
this compound, the thermal decomposition ion at m/z 128 increased in relative abundance with
increasing temperature. Although the spectra at 200° C are of better quality for library matching, the
corresponding TIC peaks exhibited extensive tailing at the lower temperatures. The peak shape of the
acids became sharper upon heating towards 300° C but spectral quality suffered as seen in table 12.
Table 13 lists the abundances for several diagnostic ions of 2,4-0 at a single level under the influence
of different matrices. These results were obtained from on-column injections with PB LC/MS conditions
the same at those described previously in the matrix effect section. For each matrix, an absolute amount
of 1000 ng was injected. From these results it does not appear that the two matrices influenced the
spectra of 2,4-D at the 1000 ng level.
30
-------
Table 13. Matrix Effects on 2,4-D Mass Spectra
% relative abundance
Ion (m/z)
Standard Soil
Soil +
Phenols
220
168
162
139
128
14.7 16.2 14.4
39.9 37.7 37.0
100 100 100
46.8 47.1 47.3
36.7 32.7 36.6
The results described thus far indicate that particle beam spectra of the phenoxyacid herbicides are
variable and depend on source cleanliness, quality of the water in the mobile phase, source temperature
and analyte concentration. Additional characterization of the spectral variation is presented below.
The changes in the ion ratio between nVz 141 and m/z 142 for MCPA described earlier were further
investigated. The ion at m/z 141 is formed by cleavage of the bond connecting the acid side chain to
the phenoxy oxygen. The Wiley reference spectra of MCPA shows the ratio between m/z 141 and m/z
142 to be about 2:1. Our PB LC/MS will only produce a ratio of 2:1 with a scrupulously clean ion source.
After a short period erf operation under 'normal1 conditions, the ion at m/z 142 increases in abundance
and overtakes m/z 141. Figure 10 displays the normalized reconstructed single ion chromatograms
(NRSICs) of m/z 141 and m/z 142 from MCPA at an ion source temperature of 225° C. This figure
indicates that the species responsible for the ion at m/z 142 is residing in the ion source for a longer
period of time than the species giving rise to the ion at m/z 141. NRSICs have been used to explain
surface assisted ionization mechanisms (11). The NRSICs suggest that the ion at m/z 142, with the help
of a 'dirty* ion source, might be formed after the molecular species has been altered in some way by
contact with the source surface. This altered species could then desorb from the source surface and
ionize with the formation of m/z 142.
Speculating in this instance, MCPA may, on prolonged contact with the hot source, thermally decompose
to the corresponding phenol which then desorbs and is ionized giving rise to the species m/z 142.
NRSICs were generated for several of the ions in the PB mass spectra of 2,4-D and are displayed in
figure 11. This figure clearty shows that the species responsible for m/z 128 and m/z 139, thought to
be formed by thermal decomposition, are remaining in contact with the source surface for a longer
period of time than the molecular species.
31
-------
100-1
NRSICs
MCPfl
EI PB LC/MS
SOURCE 225 C
90-
80-
70-
60-
142
30-
40
30
20
10-
2.3
i .9
2.0
(mln.>
2. i
2.2
1.7
1 .8
T1
Figure 10. NRSICs of m/z 141 from Particle Beam Spectrum of MCPA.
32
-------
100-1
NRSICs
2,4-D
EI PB LC/MS
SOURCE 225 C
— 128
90
— 13
80
70
60
— 162
90
30
--220
20-
10-
e.80
a. as
8.43
8.60
8.65
8.75
Figure 11. NRSICs from the Particle Beam Mass Spectrum of 2,4-0.
33
-------
Some further evidence that the molecular species of the phenoxyacid herbicides are undergoing
alteration on the ion source surface and resulting in what we have called thermal decomposition ions'
has been obtained. Figure 12 spectrum B is the methane positive chemical ionization (PCI) mass spectra
of Silvex obtained at an ion source temperature of 250° C. The ion source pressure was not readily
measured due to the plug inserted into the GC entrance port of the ion source. The methane flow rate
was set at the level necessary for maximizing the ion abundance of m/z 219 and m/z 414 for PFTBA.
Also entering the ion source was the LC mobile phase which in this case was 100% methanol.
Therefore, true methane PCI may not be possible on this system. Nevertheless, the spectra obtained
suggest possible origins of the thermal decomposition ions from Silvex. Spectrum A in figure 12 is a El
spectrum of Silvex that is easily library matchable. In spectrum B, a very small abundance of the
protonated molecular ion, m/z 269, is observed. The fragment at m/z 223 is an expected loss of the acid
functionality from the (M+H)+ ion due to excess internal energy. What is not expected are the ions at
m/z 163 and m/z 197 in figure 12 spectrum B. These ions appear to be the protonated anologs of m/z
162 and m/z 196 seen in figure 12 spectrum A.
One would not expect these ions under normal gas phase PCI mass spectrometry. These ions would
be possible if the analyte molecule contacted the ion source surface and thermally decomposed into
a species which then desorbs and is subsequently ionized. The presence of the m/z 197 ion in the PCI
spectrum of Silvex implies that the base peak, m/z 196, in the El spectrum does not arise from
fragmentation of the ionized parent molecule, but rather arises from ionization of the corresponding
phenol which results from thermal decomposition of Silvex in the ion source. This interpretation further
suggests that so called reference spectra, direct insertion probe or otherwise, are not true El type
spectra but are rather a mix of El and thermal decomposition mechanisms. Further work is required to
clarify this issue.
Figure 13 shows a simple mechanism of how some of the ions of the methane PCI mass spectra (figure
12 spectrum B) could have been produced. In this figure the analyte molecule, Silvex, is represented
by M. The analyte molecule is shown in this figure reacting in one of several pathways. First, M is
shown reacting with the reagent gas ions, PH+, forming the protonated molecular species (M+H)+.
This species then looses 46 mass units forming the ion at m/z 223. Second, M could absorb onto the
ion source surface. Once absorbed onto the ion source surface M could desorb unaltered followed by
ionization as described above. Another alternative for the adsorbed molecular species would be a
surface assisted reaction which converts the original molecule into unknown species X and Y. Figure
13 shows these altered species, X and Y, desorbing from the ion source surface and being ionized by
the reagent gas ions. Figure 14 displays the NRSICs for m/z 223,197 and 163 in the PCI mass spectra
of Silvex. These traces show that the species responsible for m/z 197 and 163 are residing in the ion
source for a longer period of time than the species responsible for m/z 223.
In summary, there appears to be substantial interaction of the phenoxyacid herbicides with the ion
source surface. This interaction leads to thermal decomposition of the analyte, and gives rise to spectral
variation. Variation in mass spectral quality depends on ion source cleanliness, source temperature,
solvent quality in the mobile phase, and analyte concentration.
34
-------
35000
30000
85000
20000
15000
10000
5000
0
SILVEX
MH 268
196
9?
132
162
\
164
125
/
4
198
R
EI
ll
268
223
/
234
lL
252
2B5
100 120 140 160 160 200
Mass/Charge
220
240
260
280
*300
4000"
3500-
3000-
0
u
2500-
c
0
"a
2000-
c
3
a:
1500-
1000-
500-
0-
163
19?
129
103
/
131
I ¦¦i|nli| I <|
¦i.lil 11il,i¦ 11|.1 I I
165
It,I ijilil.l.l |
k
199
225
/
ll. I.I.I, I.I.Jil,
METHRNE PCI
253
23? /
/
¦|H- , -
269
¦ | .H|i- . I"
29?
\
100 120 140 160 180 200
Mass/Charga
220
240
260
280
300
Figure 12. Electron Impact (A) and Methane PCI (B) Particle Beam Spectra of Silvex.
35
-------
Ion source
surface
\\
PH+ \
(M+H) + < M > \
m/z 269 | l\
1 U
I 1 X
< „K
—C00H2+ \
PH+ \
I (X+H)+ < X<— \
V m/z 163 \
(M-46)+ PH+ \
m/z 223 (Y+H) + < Y<—| \
m/z 197 \
!\
Figure 13. Mechanism of Ion Formation in Particle Beam PCI of Silvex.
36
-------
NRSICs
SILVEX
PCI PB LC/MS
SOURCE 250 C
c-163
— 197
— 223
10. 35
10. 40
10.45
10.50
10.55
10. 60
10. 65
10. 75
10.80
TI mo (mln. )
Figure 14. NRSICs from the Particle Beam Methane PCI Spectrum of Silvex.
37
-------
REFERENCES
1. Willoughby, R.C., and R.F. Browner, Anal. Chem. 6, 2626 (1984).
2. Arpino, R.J., J Chromatoqr. 323, 3, (1985).
3. Krost. K.J.. Anal. Chem. 57. 763 (1985V
4. Vestal, M.L, Int. J. Mass Spectrom, Ion Phvs. 46, 193 (1983).
5. McLafferty, F.W., Interpretation of Mass Spectra Third edition, ed. by N.J. Turro, p. 10. University
Science Books, Mill Valley, California (1980).
6. Interlab study on the performance of LC/MS to measure phenoxyacid herbicides. Unpublished.
USEPA, EMSL-Las Vegas.
7. Roby, M.R., Pace, C.M., and D.A. Miller. October/November Monthly Report. EPA Technical
Directive 70.37, EMSL-Las Vegas. Dec. 11, 1989
8. DiCorcia, A., Marchetti, M., and R. Samperi. Analytical Chemists. 61, 1363 (1989).
9. Bellar, T.A., T.D. Behymer, and W.L Budde, J. Am. Soc. Mass Spectrum 1, 92 (1990).
10. Experimental results from subcontract with West Coast Analytical Services, Inc., Santa Fe Spring,
California (1990).
11. Sears, LJ., J.A. Campbell, and E.P. Grimsrud, Biomedical and Environmental Mass Spectrometry
14, 401 (1987).
38
-------
Appendix A
Particle Beam Mass Spectra
of
Phenoxyacid Herbicides
A-1
-------
o
o
c
a
•a
c
3
.o
tc
1000"
1600-
1400-
1200-
1000-
800-
600-
400-
200-
DICRMBfl
MW 220
139
/
111
87
/
0-
i
—r"
80
illL
1 13
1S7
I
73
203
W
u
220
222
267
/
293
100 120 140 160 180 200
M>ss/Ch»rga
220 240
260
280
Figure A1. Particle Beam Mass Spectrum of Dicamba.
A-2
-------
62
25000-
2,4-D
MW 220
20000
e
o
c
0
"O
C
3
il
a:
15000-
10000-
111
139
5000-
98
/
141
85
/ -
lill[ll IJmJ
Li
m
llUt
1
220
175
222
V-4
200
/
239
265
/
80 100 120 140 160 180
M«ts/Ch»ra«
200
220
240
260
Figure A2. Particle Beam Mass Spectrum of 2,4-D.
A-3
-------
e
u
c
d
U
c
3
cr
18000-
16000-
14000-
12000-
10000-
8000-
6000-
4000-
2000-
MCPfl
MN 200
107
/
12S
/
89
91
I",'H
80
iU
100
JjiiU
120
II"!1!-"!
41
200
\
155
/
k,A
162
u
k
186
\
140 160
Mas s/Chargi
203
220
/
Ll
245
180
200
220
Figure A3. Particle Beam Mass Spectrum of MCPA.
A-4
-------
6000-
2,4,5-T
MW 254
5000-
96
4000
o
u
c
at
U
c
3
ja
-------
30000-
162
23000-
DICHLOROPROP
MW 234
164
20000-
o
0
c
a
¦a
c
3
j>
1
15000-
10000
90
/
128
5000-
91
126
\
ill! .1
llkJ
ilai
iiIii,-iNIi..
234
189
/
200
liiji li|ii .1I1II4I1I
[In >1.1.1.1
216
256
/
279
/
.
80 100 120 140 160 180 200
Mttt/Charga
220
240
260
280
Figure A5. Particle Beam Mass Spectrum of Dichlorprop.
A-6
-------
14000
MW 214
10000
4000-
2000-
Uu
160 100
M>*«^Ch«rge
Figure A6. Particle Beam Mass Spectrum of MCPP.
A-7
-------
e
o
c
«
"O
c
3
J)
(E
35000-
30000-
25000-
20000-
15000-
10000-
5000-
SILVEX
MM 268
9 7
132
162
/
-Jill"
69
/
iM
lu^ii
i.|nl,i
142
/
168
|ll|li |IiIi|„J[I».,.i1||l
196
223
/
268
/
80 100 120 140 160 180 200
Mas8/"CharHP
252
v llllL|.lll|il...
285
/
313
220
240
260
280
300
Figure A7. Particle Beam Mass Spectrum of Sitvex.
A-8
-------
2,4-DB
MH 24B
leaea
160 100 200
MassxCharga
Figure A8. Particle Beam Mass Spectrum of 2,4-DB.
A-9
-------
Appendix B
Comparison of Two Particle Beam Type Interfaces
for the Measurement of Phenoxyacid Herbicides by LC/MS
B-1
-------
Comparison of Two Particle Beam Type Interfaces for the
Measurement of Phenoxyacid Herbicides by LC/MS
The performance of two particle beam type LC/MS systems were compared for the ability to measure
phenoxyacid and other acidic herbicides. Comparisons are based on results obtained by West Coast
Analytical Services, Inc. (WCAS) on an EXTREL ThermaBeam Interface and by Lockheed Engineering
& Sciences Company (LESC) on a Hewlett Packard 599B0A Particle Beam interface. The principle
difference between the two interfaces is that the EXTREL ThermaBeam provides a heated nebulizer with
temperature control; the HP Particle Beam does not. The thermal aerosol" resulting from the heated
nebulizer is reported to give smaller particles with more uniform size than a pneumatic aerosol.
The results presented here were not obtained under strict experimental conditions and should not be
considered as a formal comparative study. Results from WCAS were derived from data acquired during
the course of an mterlab study evaluating LC/MS measurement of acid herbicides (6). Some of the data
for the HP Particle Beam system was acquired during the same interlab study, but much of the data was
acquired during the course of instrument characterization studies. Results from the two instruments
should be interpreted with the awareness that differences may reflect variation in experimental conditions
rather than true differences in instrument performance. Nominal operating parameters for each
instrument are listed in table B1.
B-2
-------
Table B1. LC/MS Operating Parameters
WCAS/EXTREL
LESC/Hewlett Packard
I. HPLC:
column
injection
volume
flow rate
solvents
gradient
program
Waters 600 MS
Ultracarb 7ji ODS (30)
250 x 2 mm (7 jim)
20 |iL
0.5 mL/min.
A = 1% formic acid
B = acetonitrile
Hewlett Packard 1090L
Spherisorb S3 C-8
100 x 2 mm (3 jim)
4 (iL
0.25 mL/min.
A = 1% acetic acid + 0.01 M
ammonium acetate
B = methanol
Time (min.)
A(%)
B(%)
Time (min.)
A(%)
B(%)
initial
85
15
initial
75
25
5
75
25
2
75
25
10
45
55
25
40
60
15
45
55
20
4
96
27
4
96
II. Interface: EXTRELVThermaBeam Hewlett Packard 59980A
helium pressure 35 psi 50 psi
desolvation
chamber 85° C 55° C
nebulizer position NA 12
probe position NA 0.5 mm
nebulizer temp. 235° C ambient
pressure stage 1 5 torr 10 torr
stage 2 NA 0.5 torr
III. Mass Spectrometer:
Source temp.
Scan
ionization
multiplier
MS-EXTREL 400 ELQ
250° C
45 to 650, 3 sec.
El, 70 eV
1850 volts
Hewlett Packard
250° C
80 to 400, 1.23 sec.
El, 70 eV
2200 volts
B-3
-------
I. Instrument Detection Limits
Estimated detection limits (EDL) for each of the instruments are listed in table B2 along with nominal
retention time (RT) and quantification ions. WCAS determined detection limits by calculating a value 3
times the standard deviation of 7 replicate measurements. Measurements were made on a mixed low
level calibration solution (200 ng each analyte) using quantification ions. Detection limits on the HP
Particle Beam were estimated by calculating a value 3 times the signal to noise ratio for each target
analyte in the total ion chromatogram (TIC) of a low level mixed standard (200 ng each analyte).
Neither instrument was capable of detecting Dalapon. Dinoseb was only detected at high levels ( >
2500 ng). Dalapon and Dinoseb may be sufficiently volatile that they are lost to the vacuum in the
interface. Instrument detection limits for the remaining target analytes were on the order of 100 to 500
ng for both instruments. Although some compound specific differences in detection limits were observed
between instruments, these probably reflect operational differences rather than true differences in
instrument performance.
Table B2. Estimated Instrument Detection Limits
ThermaBeam Particle Beam
quant. quant.
RT
EDL
ion
RT
EDL
ion
Dalapon
ND
—
—
ND
—
Dicamba
21.2
470 ng
173
2.6
520 ng
173
2,4-D
22.9
60 ng
63
8.1
130 ng
162
MCPA
22.9
70 ng
107
10.2
260 ng
141 + 142
Dichlorprop
25.0
150 ng
128
13.2
130 ng
162
2,4,5-T
25.2
660 ng
198
11.8
320 ng
196
MCPP
25.0
100 ng
107
15.4
400 ng
141 + 142
Silvex
27.0
410 ng
198
17.0
130 ng
196
2,4-DB
25.9
200 ng
128
21.1
120 ng
162
Dinoseb
30.0
2700 ng
211
-
> 2500 ng
--
2,4-D butoxy
-
-
-
-
50 ng
220
ethyl ester
2,4,5-t butoxy
32.5
100 ng
57
-
30 ng
254
ethyl ester
ND = not detected
RT = retention time
EDL = estimated detection limit
B-4
-------
II. Response Characteristics
Both the EXTREL and Hewlett Packard instruments exhibit similar response tendencies. Table B3 lists
response factors from representative four point calibration curves covering the range 200 ng to 2000 ng.
Response factors were calculated as the ratio of quantification ion area counts to the amount injected.
For most of the target analytes, response factors tended to increase with increasing concentration on
both instruments. A 5000 ng standard on the ThermaBeam instrument exhibited a disproportionately
larger increase in response. This data was omitted from table B3 for comparative purposes. A
comparable high level standard was not run on the HP Particle Beam.
Table B3. Comparative Response Curves
A. ThermaBeam
Avg.
200 na
500 na
1000 na
2000 na
EE
RSD%
Dicamba
..
0.2
3.1
1.6
89
2,4-D
38.5
61.5
68.9
69.6
59.6
24
MCPA
50.1
49.9
47.6
70.0
54.4
19
Dichlorprop
35.3
42.8
39.7
49.1
41.7
14
MCPP
38.4
48.0
43.3
56.7
46.6
17
2,4,5-T
5.5
20.0
27.8
38.9
23.0
61
2,4-DB
32.1
48.0
62.8
69.0
53.0
31
Silvex
8.8
12.6
17.2
22.1
15.2
38
B. Particle Beam
Avg.
200 na
500 na
1000 na
2000 na
BE
RSD%
Dicamba
111
99
112
112
108
6
2,4-D
936
966
1325
1509
1184
24
MCPA
1207
1098
1350
1596
1313
16
Dichlorprop
1417
1347
1590
1965
1580
17
MCPP
1349
1339
1282
1308
1320
2.3
2.4,5-T
429
373
454
625
470
23
2,4-DB
1763
2413
3617
4696
3122
42
Silvex
1108
1409
2070
2424
1753
34
The relative standard deviation in response factors on the ThermaBeam instrument ranged from a low
of 14% for Dichlorprop to a high of 89% for Dicamba Only MCPA, Dichlorprop, and MCPP gave
responses approaching linearity (i.e. RSD < 20%). Dicamba; 2,4,5-T; 2,4-DB; and Silvex exhibited
extreme deviation from linear response behavior over the range examined.
B-5
-------
Response factor RSD values from 200 ng to 2000 ng ranged from a low of 2% from MCPP to a high of
42% for 2.4-DB on the Particle Beam instrument. Instrument response to Dicamba and MCPP was
essentially linear (i.e. RSD < 10%). Response behavior approaching linearity was observed for MCPA
and Dichlorprop (RSD < 20%). 2,4-DB and Silvex exhibited extreme deviation from linear response over
the range examined. The apparent linear response range for both instruments is somewhat less than
one order of magnitude for most of the target analytes.
III. Reproducibility
Both instruments exhibit poor response precision on a day to day basis. Relative standard deviations
(RSD) in response factors at a single level (1000 ng) are summarized in table B4. Day to day RSD
values were calculated from the 1000 ng standard run on four separate days. In each case, the interface
was put in the standby mode at the end of each days run sequence. ThermaBeam day to day values
ranged from 23% for 2,4-DB to 99% for Dicamba The HP Particle Beam exhibited a narrower range of
variation, 43% for 2,4-D and MCPP to 53% for 2,4,5-T butoxy ethyl ester, but nevertheless, exhibited large
RSD values (> 40%). The high day to day RSD values indicate that both instruments require
recalibration on a daily basis or whenever the interface is placed in standby following initial calibration
Table B4. Comparative Response Precision at 1000 ng
ThermaBeam Particle Beam
Day to Day Within Day Day to Day Within Day
Dicamba
99%
37%
44%
18%
2,4-D
25%
11%
43%
13%
MCPA
63%
22%
44%
12%
Dichlorprop
38%
18%
51%
6%
2,4,5-T
30%
13%
46%
12%
MCPA
68%
19%
43%
5%
2,4-DB
23%
10%
44%
8%
Silvex
31%
8%
51%
9%
2,4,5-T butoxy
28%
15%
53%
-
ethyl ester
Within day variations in instrument response to all target analytes were considerably better than the day
to day response for both instruments. The within day RSD values were calculated at the 1000 ng level
with response factors taken from an initial calibration and all subsequent calibration check samples
acquired during the course of a single day (n=3). ThermaBeam single day RSD values ranged from a
low of 8% for Silvex to a high of 37% for Dicamba Although RSD values for Dicamba and MCPA were
somewhat high, the remaining target analytes exhibited single day RSD values less than 20%. The HP
Particle Beam single day RSD values ranged from a low of 6% for Dichlorprop to high of 18% for
Dicamba Single day response variations for all target analytes at the 1000 ng level were less than 20%
on the Particle Beam instrument. These results indicate single day response precision was acceptable
for both instruments.
B-6
-------
IV. Spectral Quality
In previous work with the HP Particle Beam, mass spectral quality for most of the target analytes was
variable. Spectral quality appeared to be a function of ion source cleanliness, water purity, and analyte
concentration. We speculate that changes in spectral appearance are the result of thermal'
decomposition in the ion source. Indicators of thermal' decomposition are loss in molecular ion intensity
and the appearance of ions not present or present only at low levels in library reference spectra. Table
B5 compares relative ion intensities for ions characteristic of decomposition. The four analytes selected
are representative of the structural types within the target analyte group. Reference spectra are from the
Wiley library which were acquired by direct insertion probe.
Table B5. Comparative Mass Spectra
2,4-D
MCPP
base peak
M+
139
128
base peak M+
107
Reference
162
60%
1%
3%
142
32%
81%
ThermaBeam
162
15%
39%
85%
107
12%
100%
Particle Beam
162
26%
22%
19%
142
33%
63%
2,4,5-T
Dicamba
base peak
M+
173
162
base peak
M+
203
139
Reference
196
50%
2%
3%
173
92%
20%
4%
ThermaBeam
198
15%
18%
38%
173
61%
31%
71%
Particle Beam
196
12%
26%
30%
173
59%
68%
69%
Examination of table B5 reveals that both the ThermaBeam and the Particle Beam produced spectra
indicative of thermal decomposition. In most instances, the base peak is the same for reference spectra
as it is in the particle beam type spectra. However, relative intensities of the diagnostic ions are much
larger in the particle beam type spectra than the reference spectra In only one case, MCPP by the
Particle Beam, does the experimental spectrum match the reference spectrum.
Comparing spectra from the two instruments reveals similar diagnostic ion intensities in the spectra of
2,4,5-T and the spectra of Dicamba. Considerable differences in diagnostic ion intensities cire apparent
in the spectra of 2,4-D and MCPP. The ThermaBeam spectra exhibit more extensive thermal
decomposition. In the case of the ThermaBeam spectrum of MCPP, thermal decomposition was
sufficiently extensive to shift the base peak from m/z 142 to m/z 107.
B-7
-------
Based on this single set of spectral data from the ThermaBeam instrument, it cannot be concluded that
ThermaBeam spectra exhibit more extensive thermal decomposition of phenoxyacids than the Particle
Beam instrument. We have observed, on numerous occasions, more extensive thermal decomposition
spectra on the Particle Beam instrument than those reported here for the ThermaBeam. Because only
one set of spectral data acquired at a single concentration level was reported for the ThermaBeam, no
conclusions regarding instrument differences in spectral variation as a function of analyte concentration
can be drawn.
V. Summary
Based on the available information and data, the EXTREL ThermaBeam and the Hewlett Packard Particle
Beam interfaces gave equivalent performance for the measurement of phenoxyacid herbicides. Detection
limits for both instruments were on the order of 100 ng to 500 ng. Neither instrument was capable . of
measuring Dalapon or Dinoseb. Response curves were non-linear over a concentration range greater
than 10 with response factors tending to increase with increasing concentration. Day to day precision
was poor on both instruments. Within day precision was considerably better with RSD values less than
20% for most target analytes. The phenoxyacid mass spectra generated by both instruments gave
indications of thermal decomposition rather than true electron impact spectra and generally did not
match library reference spectra Vanation in spectral quality as a function of concentration were not
compared.
B-8
------- |