Characterization and enhancement of sample introduction and ion transmission in combined gas chromatography/tandem mass ...

MISSING IMAGE

Material Information

Title:
Characterization and enhancement of sample introduction and ion transmission in combined gas chromatography/tandem mass spectrometry
Physical Description:
viii, 173 leaves : ill. ; 28 cm.
Language:
English
Creator:
Hail, Mark Edward
Publication Date:

Subjects

Subjects / Keywords:
Gas chromatography   ( lcsh )
Mass spectrometry   ( lcsh )
Chemistry thesis Ph. D   ( lcsh )
Dissertations, Academic -- Chemistry -- UF
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1989.
Bibliography:
Includes bibliographical references (leaves 167-172).
Statement of Responsibility:
by Mark Edward Hail.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001479605
oclc - 21071961
notis - AGZ1563
sobekcm - AA00004805_00001
System ID:
AA00004805:00001

Full Text













CHARACTERIZATION AND ENHANCEMENT OF SAMPLE INTRODUCTION AND
ION TRANSMISSION IN COMBINED
GAS CHROMATOGRAPHY/TANDEM MASS SPECTROMETRY













By

MARK EDWARD HAIL


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY






UNIVERSITY OF FLORIDA


1989






































To my loving wife, best pal, and big toe, Amy.

















ACKNOWLEDGMENTS


I would like to express sincere gratitude to my graduate research

director, Dr. Richard A. Yost, for his guidance and friendship over the

past four years. Rick deserves some sort of medal, or perhaps more

appropriately, a new prescription for his glasses, for all of the reading

and reviewing of manuscripts and thesis chapters that he has done during

the past few months. I would like to thank the members of my graduate

committee, Drs. James D. Winefordner, Anna Brajter-Toth, Samuel 0.

Colgate, and Henri A. Van Rinsvelt. I would also like to thank Dr. John

G. Dorsey (a fellow Lexingtonian) for his helpful discussions.

I would like to acknowledge the U. S. Air Force Engineering and

Services Center at Tyndall Air Force Base and the U. S. Army Chemical

Research Development and Engineering Center at Aberdeen Proving Grounds

for providing the financial support for this work.

No graduate student would be very productive without the help of the

departmental staff. For this reason, I thank Chester Eastman, Vern Cook,

and Dailey Burch of the machine shop, and Mark Ross, Steve Miles, and Dean

Schoenfeld of the electronics shop. I extend special thanks to Russ

Pierce of the electronics shop, who was particularly helpful in answering

questions and troubleshooting problems in the electronics lab and in my

research.

I would like to thank my friends who made this period of my life so

enjoyable, including all of the Yosties, and post-Yostie Jodie Johnson.

Special thanks to my two office mates and soon-to-be coworkers, Ken










Matuszak and Dave "Cool One" Berberich. I would also like to thank those

that dwelled with me in that most humble abode known as the thunderdome,

where one man enters and two people leave. In order of appearance (and

in some cases disappearance), thanks to Todd "Spud" Gillespie, Stephen

"Judas" Brooks, Mark "Citizen" Barnes, and Jerry "Citrus Connection"

Grunewald.

I would like to thank my family, Mom, Dad, Carl Jr., Beccy, and Steve,

who have been so supportive.

Finally, I would like to thank my wife, Amy, the person who was always

there to pat me on the back, who did my laundry and fed my face when I

didn't have time, and who was always there for entertainment just by being

herself.

















TABLE OF CONTENTS



ACKNOWLEDGMENTS . . ... .. iii

ABSTRACT .... . . . vii

CHAPTERS

1 INTRODUCTION . . ... 1

Short GC Columns in GC/MS and GC/MS/MS . .. 1-
Short GC Columns at Sub-ambient Pressures . 2
Mixture Analysis by GC, GC/MS and GC/MS/MS . 4
Organization of Dissertation . .. 10

2 THEORETICAL AND PRACTICAL ASPECTS OF SHORT OPEN
TUBULAR COLUMNS AT SUB-AMBIENT PRESSURES IN
GAS CHROMATOGRAPHY/MASS SPECTROMETRY . .... 12

Introduction and Theory . . ... 12
Experimental Section . .... .. 15
Calculations . .... ..... 15
Mass Spectrometry . .... .. 15
Gas Chromatography . . 16
Average Velocity Measurement . ... 19
Results and Discussion . .... 20
Effect of the Pressure Drop . ... 20
Sub-ambient Inlet Pressures . ... 26
Experimental vs. Theoretical Performance ... 37
Choice of the Carrier Gas . ... 41
Conclusions . . . 43

3 CONCEPTS FOR PORTABLE GAS CHROMATOGRAPHIC
INSTRUMENTATION IN GC/MS: FLOW RATE
PROGRAMMING AND DIRECT COLUMN HEATING . ... .46

Introduction . . ... 46
Flow Rate Programming . .... .47
Direct Resistive Heating of Al-Clad Capillary Columns 48
GC Probe . . ... 51
Experimental Section . . .. 52
Experimental Conditions . .... 52
Flow Control System . .... 53
Direct Resistive Heating . .... 57










GC Probe Design . . ... .. 61
Results and Discussion . ... 64
Flow Rate Programming . .... 64
Direct Resistive Heating . .... 77
GC Probe Performance . .... 85
Conclusions . . .. .. .. 89

4 GAS CHROMATOGRAPHIC SAMPLE INTRODUCTION INTO THE COLLISION
CELL OF A TRIPLE QUADRUPOLE MASS SPECTROMETER
FOR MASS-SELECTION OF REACTANT IONS FOR
CHARGE EXCHANGE AND CHEMICAL IONIZATION . .. 90

Introduction and Theory . .... 90
Experimental Section . ... ....... 93
Mass Spectrometry . . ... 93
Gas Chromatography . .... 95
GC/MS Interface . .... ..... 96
Results and Discussion . . ... 98
Ionization Energy Measurements . ... 98
Simultaneous Structural and Molecular Weight Information .101
Selectivity of Ionization . ... .106
Detection Capabilities of GC/MSR/MS . ... 112
Chromatographic Integrity . ... .116
Conclusions . . .. ... ... 116

5 CHARACTERIZATION AND OPTIMIZATION OF ION TRANSMISSION
IN TRIPLE QUADRUPOLE MASS SPECTROMETRY . ... 118

Introduction . . ... ... 118
Experimental Section . .... .119
Instrumentation . . ... .119
Procedures . ... .. .. ... 125
Results and Discussion . .... 126
Mass-dependent Optimization . ... .126
Quadrupole Mass Filters . .... 135
Ion Transmission Characteristics of the Center Quadrupole 140
Ion Transmission at High Collision Energies ... 153
Conclusions . . ... .. .... 157

6 CONCLUSIONS AND FUTURE WORK . ... .159

Conclusions . . .. .. .. 159
Suggestions for Future Work . ... .161
Reduced Column Pressures with Conventional GC Detectors 161
Directly-heated Open Tubular Columns . .. 162
Resonance Excitation and Ion Storage in Q2 ... 164

LITERATURE CITED . . . 167

BIOGRAPHICAL SKETCH . ... ...... 173

















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy






CHARACTERIZATION AND ENHANCEMENT OF SAMPLE INTRODUCTION
AND ION TRANSMISSION IN COMBINED
GAS CHROMATOGRAPHY/TANDEM MASS SPECTROMETRY

By

Mark Edward Hail

May 1989


Chairman: Richard A. Yost
Major Department: Chemistry

Various theoretical and practical aspects of the use of short open

tubular columns in combined gas chromatography/mass spectrometry (GC/MS)

and tandem mass spectrometry (GC/MS/MS) have been investigated, including

the advantages of low-pressure operation with short and/or wide-bore open

tubular columns. It is shown both theoretically and experimentally that

short columns with vacuum outlet require sub-atmospheric inlet pressures

if optimum gas velocities are to be obtained. The use of sub-ambient

inlet pressures is also shown to improve the sensitivity when short

columns are used with electron ionization in GC/MS.

Several considerations are addressed for the design of portable GC

instrumentation. In particular, exponential flow rate programming as well

as resistively-heated aluminum-clad open tubular columns have been

evaluated for low-power, high-throughput analysis in GC/MS. Aluminum-clad


vii










open tubular columns are heated directly by applying a voltage across the

capillary, and the column temperature is sensed by measuring the column

resistance. These investigations have led to the development of a compact

gas chromatograph probe for use in GC/MS.

A gas chromatograph has been interfaced to the collision cell of a

triple quadrupole tandem mass spectrometer for performing mass-selected

ion-molecule reactions. Reactant ions are selected with the first

quadrupole and are allowed to react in the second quadrupole collision

cell with the effluent from a short open tubular GC column. The ion-

molecule reaction product ions are mass-analyzed by the third quadrupole.

The advantages of using mass-selected reactions for controlling the

selectivity of charge exchange and chemical ionization are demonstrated.

In addition, this configuration is shown to provide both structural

information and molecular weight information in the same chromatogram by

alternating between different reactant ions.

The ion transmission characteristics of a triple quadrupole mass

spectrometer (TQMS) have been studied. The unique high level of computer

control of the TQMS has allowed for rapid characterization and

understanding of the ion optical parameters. Strategies for obtaining

optimum ion transmission and minimization of mass dependencies in the ion

optics for single MS experiments as well as for MS/MS experiments are

discussed.


viii

















CHAPTER 1
INTRODUCTION


In this dissertation a variety of concepts are presented for the

enhancement of sample introduction and ion transmission in analytical mass

spectrometry (MS) and tandem mass spectrometry (MS/MS). Due to the

inherent speed and selectivity of MS and MS/MS, considerable attention was

devoted to the study of rapid gas chromatographic (GC) separation

techniques. A new tandem mass spectrometric method, in which the

selectivity of ion-molecule reactions is exploited for gas chromatographic

mixture analysis, is also described. Finally, the results of a detailed

characterization of the ion optical parameters of a triple quadrupole

tandem mass spectrometer are presented.


Short GC Columns in GC/MS and GC/MS/MS


In this research group, it has been found that short GC columns can be

used for the extremely rapid analysis of complex mixtures by taking

advantage of the inherent selectivity of MS, and particularly MS/MS [1-

3]. In the initial results that were obtained with very short (e.g., 50

cm) packed columns, it was found that the .sensitivity and selectivity

obtainable with short-column GC in conjunction with MS/MS in many cases

exceeded that of high-resolution GC/MS (HRGC/MS) [1]. In addition, it

was shown that 100 injections per hour could be performed with the short-

column method, while only 2 injections per hour could be made with the

HRGC/MS method employing an 18 m capillary column [1]. Later, short open








2

tubular capillary columns were evaluated (including an investigation of

the associated chromatographic theory) due to their inherent advantages

over packed columns [2,3]. These studies began to demonstrate the

advantages of vacuum outlet operation with short open tubular columns.

It was found that there were added benefits other than speed of analysis

that were important. For instance, because of the high gas velocities and

low operating pressures associated with short open tubular columns under

vacuum outlet conditions, it was found that solutes tended to elute at

temperatures well below their boiling points. Thus, it was discovered

that short-column open tubular GC in conjunction with MS or MS/MS was

ideal for the analysis of involatile, thermally labile, or polar solutes

[2,3]. More recently, McClennen et al. have taken advantage of this and

have demonstrated that large involatile and/or polar molecules can be

analyzed by examining the Curie-point pyrolysis products of these

molecules with short-column GC/MS [4,5].


Short GC Columns at Sub-ambient Pressures


Many of the chromatographic separation techniques currently in

widespread use require relatively high operating pressures, namely, high-

pressure liquid chromatography (HPLC), supercritical fluid chromatography

(SFC), and even high-resolution capillary open tubular gas chromatography

(HRGC). In contrast, research in the area of low-pressure separation

techniques has been limited. More than 25 years ago, Giddings

demonstrated that the minimum analysis times in gas chromatography could

be achieved when the column outlet was connected to a vacuum [6].

However, to date, a majority of the research done in high-speed GC has

dealt with the use of small diameter columns and the development of high-











speed injectors required for use with these columns [7-13]. Recently,

interest in vacuum outlet operation has been renewed, as Leclercq and

coworkers have published a number of articles that have addressed the

theory and advantages of vacuum outlet operation for reducing analysis

times in open tubular GC [14-17].

Vacuum outlet operation in gas chromatography results in lower column

pressures, which increases the diffusivity of the solute in the gas phase

and leads to increased optimum carrier gas velocities and shorter analysis

times [14]. The advantages of vacuum outlet operation are inherently

obtained when gas chromatography is combined with mass spectrometry

(GC/MS), provided the column is inserted directly into the vacuum of the

mass spectrometer. Consideration of the theory indicates that low outlet

pressure operation has a more pronounced effect if wide-bore columns [16]

and/or short columns [3] are used. As shorter column lengths are used to

minimize analysis times, lower inlet pressures are required if optimum

chromatographic performance is to be obtained. The use of very short

lengths or wide-bore columns is limited when conventional inlet pressures

are used because of the gas load imposed on the mass spectrometer.

Alternatively, the inlet can be operated at sub-ambient pressures to

obtain optimum performance and/or to reduce the flow rate of carrier gas

entering the mass spectrometer vacuum system [18]. Unfortunately, the

value of sub-atmospheric column inlet pressures has not been widely

appreciated, due in part to the difficulty of operating traditional GC

injection ports at vacuum.









4

Mixture Analysis by GC, GC/MS. and GC/MS/MS


Application of short columns with conventional GC detectors has been

limited to the analysis of relatively simple mixtures due to the non-

selective nature of the detectors and the reduced resolving power of short

columns. However, as stated previously, the inherent selectivity of MS

and MS/MS can be utilized to atone for the chromatographic resolution lost

when shorter columns are used. Perhaps a better comprehension of the

additional selectivity of the mass spectrometric detection methods over

conventional detection methods in gas chromatography can be obtained by

considering the possible cases which result in interference. These

coincidences are summarized in Table 1.1, which shows a comparison of GC

employing flame ionization detection (FID) as well as GC with various mass

spectrometric detection methods. A schematic comparison of the three mass

spectrometric techniques, all of which can be performed on the triple

quadrupole tandem mass spectrometer used in this work, is shown in Figure

1.1.

Mixture overlap will readily occur when non-selective detection methods

(e.g., FID) are used in complex mixture analysis, unless a column of

sufficient length is used to chromatographically separate all possible

interference from the analyte of interest. In many cases, no column,

regardless of length, would be sufficient to accomplish this goal; thus,

other extraction and sample cleanup procedures must be used in addition

to the already time-consuming chromatographic analysis. In addition,

since chromatographic resolving power increases only with the square root

of the column length, while analysis times increase proportionally with

column length, increasing the column length to gain the necessary











Table 1.1. Cases leading to mixture overlap interferencee) in gas
chromatographic mixture analysis with flame ionization detection and
various methods of mass spectrometric detection. The '?'s in the table
indicate the conditions under which targeted mixture components will be
obscured by interference.



Analytical Method


GC/FID


GC/MS


GC/MS/MS


GC/MSR/MSa


Coincidences Resulting in Interferences


(1) coelute

?


(1) coelute
4.
(2) produce
ions of
same m/z


(1) coelute

(2) produce
ions of
same m/z
4
(3) fragment to
daughter ions
of same m/z
4


(1) coelute
4.
(2) react to form
product ions
of same m/z

?


a GC/MSR/MS GC/mass-selected reaction/MS, see Figure 1.1.












w C e C
0 u 040 00
o n 0 04 0
0 0 i-l 0 41 *-4
,c 0 0 0 l-4
) 0 r-0 C
U 0 v-q U) 4)
o\ o1 .0


4j u C Cn
4 j rq (d W


4 0 CO
0 r o
>

u0 ..o c



aEn ^ 0
U 4



4a J





00
(6C*s




4-J U M ^



s **4
0 C







0902.
* M O 0*
Cd w 5 C4



S -> c o (
4 C -4
g 9 0 Go



4 0 0) 0






60 M .u
4.) 0 O p U








4 S:
4rl a 4 a3



S^4 0
.-4 0 ro




* I


0 4.) Cd U
:1 $ a W1 C





"4 04 1 -4
t0 P. i-l ) 4


-- 3 ) 4-l C .-4 0






















U)

\

Cl)



U
(3


z
0
co
--- <
-JU
-1U
0
U


0)








C)
U
(3


z^
)-
4 to




U
0011


-Q


U)




uJ


CV)
0




Cu
a








8

resolution tends to be very costly in terms of the time required for an

analysis.

As shown in Table 1.1, the possibilities for interference in mixture

analysis are greatly reduced when mass spectrometric detection is used.

In GC/MS (Figure 1.1 (a)), other components) of the mixture must coelute

with the analyte of interest and must yield ions of the same m/z as the

analyte of interest in order for interference to occur. Even more

selectivity is obtainable with GC/MS/MS (Figure 1.1 (b)). In order for

mixture overlap to occur in GC/MS/MS, undesirable interferents must not

only coelute and be ionized to produce the same nominal mass ions, but the

interfering parent ions must also fragment via collisionally-activated

dissociation (CAD) to produce all of the same daughter ions as the

analyte.

Described in this dissertation is a new GC/tandem mass spectrometric

technique, GC/mass-selected reaction/MS (GC/MSR/MS), which involves mass-

selection of reactant ions and reaction of these ions with the effluent

from a gas chromatographic column. It can be seen from Figure 1.1 (c)

that this method involves introduction of the GC column into the second

quadrupole collision cell (Q2) rather than into the ion source. This

method takes advantage of the intrinsic selectivity of ion-molecule

reactions. Hence, for overlap to occur in mixture analysis with

GC/MSR/MS, interference must react in Q2 to produce all of the same m/z

product ions as the analyte of interest.

It can be seen from the above comparisons that the added resolution

elements available with the mass spectrometric techniques in many cases

obviate the need of high-resolution chromatographic separation in mixture

analysis. Therefore, the use of short open tubular columns and the










concomitant short analysis times are readily justified. In addition, the

capability to perform any of the three GC/MS techniques (GC/MS, GC/MS/MS,

or GC/MSR/MS) certainly demonstrates the versatility of the triple

quadrupole mass spectrometer (TQMS).

Although instrumentation for MS/MS tends to be rather expensive (e.g.,

$350,000 for a TQMS) when compared to other analytical techniques (e.g.,

$15,000 for a GC), it has been pointed out that MS/MS is a cost-effective

technique because of the short analysis times that are possible [19].

This is most likely the reason for the increasing popularity as well as

the widespread availability of commercial instrumentation for MS/MS. The

MS/MS technique has certainly benefited from recent advances in computers

and electronics, as these progressions have led to the development of a

wealth of new, exciting, computer-controlled instrumentation. One such

instrument is the Finnigan MAT TSQ70, which was used throughout this work.

This instrument, with its unique level of computer control, has allowed

a detailed study of the ion optical parameters and their effects on ion

transmission in triple quadrupole mass spectrometry. In fact, many of the

undesirable ion optical effects (e.g., mass dependencies) can be avoided

due to the capability of this instrument to vary any the ion optical

parameters as a function of mass within each mass spectral scan. These

types of studies have provided a more thorough understanding of the ion

optical parameters that effect the performance of this instrument, which

will ultimately improve the analytical capabilities during its use.










Organization of Dissertation


This dissertation is divided into six individual chapters. This

chapter has served to present an overall perspective as well as the

purpose of the work. Global conclusions and suggestions for future

research are found in Chapter 6. Chapters 2-5 make up the body of the

dissertation. Due to the widely varying material presented in this

dissertation, there is no particular chapter devoted entirely to the

description of experimental conditions. Rather, each chapter (of Chapters

2-5) consists of a separate introduction, experimental section, discussion

of results, and conclusions.

In Chapter 2, the advantages of utilizing low column pressures with

short open tubular columns in GC/MS are examined by consideration of the

theoretical chromatographic relationships. In particular, the advantages

of operating the column inlet, as well as the column outlet, at sub-

atmospheric pressures are stressed. Experimental data are presented that

demonstrate the practicality of the combination of short GC columns with

mass spectrometry.

In Chapter 3, considerations for portable and/or high-throughput sample

introduction schemes for GC/MS are presented, including investigations of

exponential flow rate programming and direct resistive heating of

aluminum-clad open tubular columns. A portable direct insertion gas

chromatograph probe is described that incorporates these concepts.

The fourth chapter describes the GC/MSR/MS technique, including a

brief discussion of the utility of selective ion-molecule reactions for

mixture analysis. This work represents the first experiments ever

performed in which samples are introduced from a GC column into the center

quadrupole collision cell of a TQMS. Experimental data are presented that








11

demonstrate the potential selectivity, detection and screening

capabilities of this method. A method is also described which allows for

the extremely rapid measurement of ionization energies on the computer-

controlled TQMS.

In Chapter 5, the effects of various ion optical parameters of the

TQMS for MS and MS/MS are investigated. In particular, it is shown that

the mass-dependencies in the ion optics of the computer-controlled TQMS

can be minimized, since all of the ion optical parameters may be varied

as a function of mass during a mass spectral scan. The insights gained

in these experiments have yielded a better understanding of the complex

inter-relation of the parameters that affect the performance and

reproducibility of the results obtained with this instrument.
















CHAPTER 2
THEORETICAL AND PRACTICAL ASPECTS OF SHORT OPEN TUBULAR COLUMNS
AT SUB-AMBIENT PRESSURES IN GAS CHROMATOGRAPHY/MASS SPECTROMETRY


Introduction and Theory


In this chapter, various theoretical and practical aspects of the use

of short open tubular columns in GC/MS are discussed, including the

advantages of low-pressure operation with short and/or wide-bore open

tubular columns. It is shown that short columns with vacuum outlet

require sub-atmospheric inlet pressures if optimum gas velocities are to

be obtained. Since sub-atmospheric inlet pressures are not obtainable

with most gas chromatographic inlet systems, a computer-controlled flow

control system has been investigated that allows the injection port to be

operated at sub-ambient pressures with either split or splitless

injections. Plate heights have been experimentally determined at sub-

ambient pressures and have been compared with theoretical predictions.

The effects of extra column band broadening are evident in the

experimental data and demonstrate the need for high-speed injection

systems if maximum performance is to be obtained with short columns. The

use of sub-ambient inlet pressures is also shown to improve the

sensitivity when short columns are used with electron ionization in GC/MS.

Finally, it is shown that hydrogen carrier gas can be used to obtain the

fastest analyses without altering the relative abundances of ions obtained

with methane chemical ionization mass spectrometry.










The theoretical dependence of the plate height, H, on the average gas

velocity, v, in open tubular gas chromatography is given by the Golay

equation [20]



H B/v + (Cg+CL)V (2.1)



where B is a term relating to the longitudinal diffusion of the solute

zone, and Cg and CL are terms relating to mass transfer in the gas phase

and liquid stationary phase, respectively. For open tubular columns

employing a thin-film stationary phase, mass transfer in the liquid phase

is negligible and Cg dominates [14]. The Golay equation for thin-film

columns corrected for gas compressibility [21] can be rewritten in terms

of the column parameters yielding [14]



2D,0 f (1+6k+llk2)r2vf
H-- + (2.2)
vo 24(l+k)2Do


where Dg is the diffusion coefficient of the solute in the gas phase at

the outlet pressure, f is the Giddings correction factor, vo is the gas

velocity at the column outlet, k is the partition ratio, and r is the

column radius. The Giddings correction factor accounts for the

decompression of the carrier gas when a pressure gradient exists and is

given by



9(P4-1)(P2-1)
f -8(- ---- (2.3)
8(P3-1)2










where P is the inlet to outlet pressure ratio, Pi/Po. The gas

decompression factor approaches a value of 9/8 when large pressure

gradients are employed, as in GC/MS. Most commonly, H is expressed in

terms of the average velocity, v. If this is done, vo and Dgo of eq 2.2

are replaced by the average velocity, v, and an average diffusion

coefficient, D An average diffusion coefficient can be estimated from

[14]



P1D1
D (2.4)
P



where Dg1 is the diffusion coefficient at 1 atmosphere pressure, P1, and P

is the average column pressure. The length-average column pressure, P,

is dependent on the pressure gradient and is given by [14]



2Po(P3-1)
P- (2.5)
3(P2-1)



which reduces to 2P,/3 for vacuum outlet conditions.

As shown by eq 2.4, lowering the average column pressure increases the

diffusion coefficient. An increase in the diffusion coefficient has been

shown to result in an increase in the optimum gas velocity and decreased

analysis times [3,14-18]. In addition, a larger D results in a slower

rate of increase of H (from the Cg term) as the gas velocity is increased

beyond the optimum. This means that even higher gas velocities can be

used without severe losses of efficiency.








15

There are a number of ways of decreasing the average column pressure.

Lower average column pressures are readily obtained if the column outlet

is at vacuum as in GC/MS. At lower outlet pressures, the greatest gains

in terms of speed of analysis (as compared to normal atmospheric outlet

operation) can be obtained if short and/or wide-bore columns are used

[3,16]. Lower average column pressures can also be obtained if low-

viscosity carrier gases are used [22]; hence, the use of hydrogen as a

carrier gas will always result in the shortest analysis times.


Experimental Section


Calculations


All theoretical calculations were done on an IBM PC/XT or a PC/AT, with

programs written in TurboBASIC (Borland International). The results of

the calculations (e.g., H vs. v data) were stored on disk in ASCII format,

which allowed the data to be retrieved and plotted with a commercially

available plotting program (Grapher, Golden Software, Inc.).


Mass Spectrometry


A Finnigan MAT TSQ70 triple quadrupole mass spectrometer was used in

these studies. Electron energies of 70 and 100 eV were used for electron

ionization (El) and chemical ionization (CI), respectively. The

appropriate interchangeable ion volumes were used to obtain El and CI

spectra. In order to obtain true ion source pressures, indicated ion

source pressures were calibrated with a capacitance manometer. This was

done by connecting a capacitance manometer (MKS Instruments model 127AA)

to the CI ion source via the side flange (normally used for GC/MS transfer

line) of the TSQ70. The capacitance manometer was used to calibrate the








16

ion source thermocouple gauge at various pressures of methane reagent gas.

Ion source pressures were determined as a function of flow rate of helium

and hydrogen carrier gases by connecting a mass flow controller (MKS

1159A) to the CI ion source and recording the true pressures read from the

capacitance manometer. All ion source pressures reported are the absolute

pressures obtained from the calibrations. Ion source pressures for

chemical ionization were typically 0.2-0.5 torr. The ionizer temperature

was 150*C. The preamplifier gain was set to 108 V/A; the electron

multiplier was operated at 1000 V.


Gas Chromatographv


A Varian model 3400 GC equipped with a split/splitless injector with

a 2 mm i.d. glass liner was used. Bonded phase fused silica open tubular

columns were used with an inner diameter of 0.25 mm and a 0.25 pm film

thickness (DB-5, J&W Scientific). Column lengths were typically 3 m. All

columns were inserted directly into the ion source. Column efficiencies

were measured with n-pentadecane (C15H32) with the column and interface at

100*C (k 32) and an injection temperature of 200*C. For all other

applications temperature programming was used with the interface at 150*C.

Column temperatures were programmed at a rate of 25'C/min from 50C to

150*C, after an initial hold period of 12 seconds.

A schematic of the gas chromatographic system, modified to allow for

either normal pressure-regulated operation or flow-controlled operation,

is shown in Figure 2.1. With valve 1 (Vl) rotated to position A, the GC

can be operated in the normal configuration with inlet pressures above 1

atmosphere. With V1 rotated to position B, flow regulation is used,

allowing inlet pressures above or below one atmosphere to be used. In










17

0
.m








z 4




100
r-4


O
.0
41
a. Q

W )4



OU







OZ.
S0 0



So

o B
0
w ry


I" I,, Io -











0. 0
:I




w Wa
So
I-- P g


















Lz J 41



< 0
So S














4) o4O
I *------~ j







i-l )-
CL








18

addition, computer control of the flow controller allows the flow rate of

the carrier gas to be programmed with time if desired (see Chapter 3).

The mass flow controller (MFC) used was a MKS model 1159A with a type

246 power supply/digital readout. The output flow rate of the MFC, which

has a range of 0-100 mL/min for N2, is set by a 0-5 V control voltage from

the TSQ70. The TSQ70 has two spare digital-to-analog converters (DACs),

or user outputs, one of which was used to control the flow rate of the

MFC.

In order to achieve the sub-ambient inlet pressures, a 30 L/min

mechanical pump (Alcatel model 8A) was connected to the split line of the

injection port. A Bourdon tube pressure gauge (Omega Engineering model

30V/30) capable of indicating pressures above or below one atmosphere was

connected to the septum sweep of the GC injection port. Valve V2 is a

solenoid valve that can be opened or closed under computer control. The

inlet pressure in the vacuum inlet mode is determined by the flow rate of

carrier gas from the MFC and the adjustment of the needle valve (V3). The

occurrence of air leaks in the system was monitored by observing the

intensity of the N+ (m/z 28) and 02+ (m/z 32) ions. It was necessary to

replace the plastic fittings and viton ferrules of the Varian 3400 carrier

gas lines with stainless steel fittings to minimize leakage of air into

the system. Leakage of air through the septum was minimized by replacing

the septum after frequent use (e.g., after 30 injections). Once these

steps were taken, no additional increase in the intensity of the air peaks

was observed when vacuum inlet operation was used instead of normal

atmospheric operation.

During a split injection, samples are injected directly into the low-

pressure injection port with valve V2 open; thus, most of the sample is








19

swept away by the mechanical pump. One of the problems of vacuum inlet

GC that has limited its use in the past is that only split injections

could be performed. This is certainly a limitation, especially when

analyzing samples with analytes present at trace levels. Splitless

injection is commonly used for trace analysis [23]. With the flow control

system shown in Figure 2.1, the advantages of vacuum inlet GC, as well as

splitless operation, can be utilized. A splitless injection is made with

V2 initially closed, with the injection port slightly above atmospheric

pressure. After allowing ample time for the sample to enter the column,

the inlet pressure is then rapidly reduced by opening V2. Based on the

internal volume of the injection port (0.24 mL), at a typical flow rate

of 5 mL/min, the injection port is completely flushed with carrier gas in

approximately 3 s. Thus, leaving V2 closed for any period longer than 3

s should be sufficient time to insure that all the sample has entered the

column. Nevertheless, in all experiments utilizing splitless injections

reported here, V2 was left closed for 12 s or more.


Average Velocity Measurement


It was found that the average velocity of the carrier gas could not be

accurately measured by conventional means with the short columns that were

used. Traditionally, the average velocity is determined from the dead

time, which is the time required for the elution of unretained air or

methane. A 3 m x 0.25 mm i.d. column at 50*C and 1 atmosphere inlet

pressure of helium carrier gas has a gas velocity of 240 cm/s and a dead

time of 1.25 s. Dead times that are this short are nearly impossible to

measure with a reasonable degree of accuracy. An alternative method is

to measure the gas flow rate and calculate the gas velocity. This is not










entirely straightforward in GC/MS. One can obtain a measure of the column

flow rate by measuring the gas flow rate exiting the forepump of the

vacuum system, since all of the carrier gas exiting the column must exit

through this pump. However, the pulsations produced by the pump limit

the accuracy and precision of this measurement, especially at low flow

rates. A more accurate indication of the average gas velocity when short

columns are used can be obtained by measurement of the inlet pressure.

If the inlet pressure, Pi, is known, the average velocity under vacuum

outlet conditions can be calculated from the Poiseuille equation [14]



v 3P7r2/32iL (2.6)



where r is column radius, r is the viscosity of the carrier gas at the

column temperature, and L is the column length. With the pressure gauge

that was used, inlet pressures ranging from -14.7 psig to 30 psig could

be measured with a rated accuracy of 3% [24].


Results and Discussion


Effects of the Pressure Drop


As was stated previously, lower average column pressures result in

increased gas-phase diffusion coefficients and increased optimum gas

velocities. Lower column pressures are obtained in GC/MS, since the

outlet of the column is at vacuum. Figure 2.2 shows Golay plots

(calculated from eq 2.2) demonstrating the effect on the plate height of

using vacuum outlet or atmospheric outlet with 0.25 mm i.d. open tubular

columns with lengths of 3 m and 30 m. As shown in the figure, the most

significant increase in the optimum velocity is obtained with the short













0.10 i



0.08


0.06


S0.04
0c
4-,-

- 0.02



0.00


30 m


3 m


- -Vacuum Outlet
Atmospheric Outlet


- -


Average Velocity


cm/s)


Figure 2.2. Theoretical dependence of plate height on the average carrier
gas velocity for 3 m and 30 m 0.25 mm i.d open tubular columns under
vacuum outlet and atmospheric outlet conditions. (Conditions: thin-film
stationary phase; He carrier gas; k 10; 100'C; D 0.27 cm2/s; r -
2.28 x 10'4 poise)








22

column operated with vacuum outlet. The speed of analysis is ultimately

dependent on the gas velocity and the column length. Under vacuum outlet

conditions, analysis times are reduced by a factor of 30, if a 3 m column

is used instead of a 30 m column, and both are operated at their optimum

gas velocities. The reduced analysis times are due to a reduction of the

column length by a factor of 10, as well as the increase in the optimum

gas velocity by a factor of 3 for the shorter column. In addition, since

chromatographic resolution is proportional to the square root of the

number of theoretical plates, the resolution lost by using a column that

is 10 times shorter is only a factor of the square root of 10 (i.e.,

3.16). As shown in Figure 2.2, the theoretical minimum plate heights

obtained with vacuum outlet operation are slightly greater than those

which are obtained with atmospheric outlet pressure. This can be

attributed to the gas decompression term, f, in eq 2.2, which results in

an increase of the minimum plate height by a factor of 9/8 (in the worst

case) when vacuum outlet is employed instead of atmospheric outlet

operation [14].

A better understanding of the effects of the outlet pressure on the

plate height for a short column can be obtained from an investigation of

the pressure drop. The pressure, p(x), at each point, x, along an open

tubular column can be calculated from the inlet pressure, Pi, the outlet

pressure, Po, and the column length, L [25]:



p(x) [(Po2-Pi2)x/L + Pi2]1/2 (2.7)



This equation was used to calculate the pressure at each point along a 3

m x 0.25 mm i.d. open tubular column with various outlet pressures. As








23

shown in Figure 2.3 (a), the pressures calculated for outlet pressures of

100 torr and 1 torr are virtually the same over most of the length of the

column. Figure 2.3 (b) shows the theoretical dependence of the plate

height on the average gas velocity for the same outlet pressures used in

Figure 2.3 (a). Note that the effects of the outlet pressure on the plate

heights calculated are nearly the same with outlet pressures of 100 torr

and 1 torr. It is apparent from these data that any GC detector that

could be operated at the reduced pressures would offer significant

advantages in terms of speed of analysis. In addition, if operation at

1 torr were not possible, operation with outlet pressures of 100 torr

should yield nearly the same chromatographic results (at least with the

length and diameter column used in this example). Another advantage of

the low outlet pressure operation is that chromatograms obtained on a

stand-alone GC (e.g., with flame ionization or electron capture detection)

could be reproduced on a GC/MS system. If this could be done, routine

chromatographic optimization and methods development could be performed

on the low outlet pressure GC for analyses to be performed later on the

GC/MS system. This would obviate the use of valuable instrument time on

a GC/MS system for chromatographic methods development. Investigations

in these areas are currently underway in our laboratory.

The effect of the outlet pressure on the optimum velocity is even more

pronounced for short wide-bore columns. Figure 2.4 shows the effect of

using vacuum outlet or atmospheric outlet on 3 m columns with inner

diameters of 0.53 mm and 0.25 mm. As shown in the figure, the use of

wide-bore columns when compared to narrow-bore columns of the same length

results in a more significant increase of the optimum velocity with vacuum

outlet operation. However, it is evident from Figure 2.4 that a lower










1000


800


600


400


200


Po=760 torr

P,=400 torr

/Po=100 torr

/Po=1 torr


--r -I
I I II
IOptima
I I I
I I II


-. -I I -


. r T r .
100 200
Average Velocity (cm/s)


300


Figure 2.3. (a) Calculated pressures along a 3 m x 0.25 mm i.d. open
tubular column with an inlet pressure of 860 torr of helium and various
outlet pressures. (b) Theoretical dependence of the plate height on the
average gas velocity for the same outlet pressures used in (a). The
conditions were the same as those listed with Figure 2.2.


Po=760 torr





Po=400 torr




Po=100 torr
Po=1 torr


(b)


0.10


0.08


0.06


0.04


0.02


/* ff


C.)


I

a,


u.uu


I
















,. 0.53 mm i.d.


0.25 mm i.d.,,


//
//
//
/J 0.53 mm i.d.
//
//
//
//. 0.25 mm i.d.
//


/
'- /


0.12


E


0.08
C-


I


0 0.04





0.00


(cm/s)


Figure 2.4. Theoretical dependence of the plate height on the average gas
velocity for 3 m x 0.25 mm i.d. and 3 m x 0.53 mm i.d. open tubular
columns with atmospheric or vacuum outlet operation. The conditions were
the same as those listed with Figure 2.2.


I I


- -- Vacuum Outlet
- Atmospheric Outlet








26

minimum plate height can be obtained with the narrow-bore column. Since

the maximum number of theoretical plates is equal to column length divided

by the minimum plate height, more theoretical plates can be obtained with

the narrow-bore column. If a constant number of theoretical plates are

required, analysis times will be shorter for the narrow-bore column. For

example, approximately 5600 plates can be obtained with the 3 m x 0.53 mm

i.d. column at an optimum velocity of 182 cm/s. The same number of

theoretical plates and optimum velocity can be obtained with a 1.4 m x

0.25 mm i.d. column. Based on the lengths of the two different diameter

columns used to generate 5600 plates, the analysis times should be faster

for the narrow-bore column by a factor of 2.1 (i.e., 3 divided by 1.4)


Sub-ambient Inlet Pressures


The optimum gas velocity for thin-film open tubular columns operated

under vacuum outlet conditions, voptvac can be calculated from [14]


(P13(l+k) Dg,1Hmin)1/2
Uopt,vac (2.8)
(2Ln(llk2+6k+l))1/2



where Hmi is the minimum plate height obtained at vot. As can be seen

from eq 2.8, the optimum velocity is inversely proportional to the square

root of the column length; thus, a reduction of the column length results

in higher optimum velocities. In addition, the optimum inlet pressure

needed to obtain the optimum gas velocity is given by [14]



(72P1L1Dg,1)1/2
i,opt,vac -= (2.9)
(Hmin)1/2r










Thus, short columns and/or columns with larger inner diameters require

lower inlet pressures. In addition, optimum inlet pressures are predicted

to be lower for low-viscosity carrier gases. This is demonstrated in

Figure 2.5 (a) which shows a plot of optimum velocity vs. column length

and Figure 2.5 (b) which shows a plot of optimum inlet pressure vs. column

length for 0.25 mm i.d. open tubular columns with both He and H2 carrier

gases. As shown in the figures, short columns with vacuum outlet require

sub-atmospheric inlet pressures, if optimum gas velocities and hence,

minimum plate heights, are to be obtained. Most GCs designed to be used

with open tubular capillary columns utilize pressure regulators that are

referenced to one atmosphere, and are incapable of operating with the

injection port at reduced pressures. Because of this limitation, the

advantages of short-column vacuum outlet GC have never been fully

exploited.

Sub-ambient inlet pressures are readily obtained with the flow-

regulated system used in this work. The chromatograms shown in Figure 2.6

demonstrate the increased resolution obtained on a 3 m x 0.25 mm i.d. open

tubular column when the injection port was operated at a reduced pressure.

The chromatogram in Figure 2.6 (a) was obtained with an inlet pressure of

940 torr and a gas velocity of 297 cm/s. For the chromatogram in Figure

2.6 (b), the inlet pressure was kept at 940 torr for the initial 12 s,

after which the pressure was reduced to 380 torr and a gas velocity of 120

cm/s. Note that the two components unresolved chromatographically in

Figure 2.6 (a) exhibit baseline resolution in Figure 2.6 (b). Moreover,

only 35 s of additional time was required for the chromatogram in Figure

2.6 (b).










(a)


,500
mc
E
%400


o 300

a I
S200

E
S100

0
0
C




1400
L
01200

1000-
,n
800

600
C
400
E
E 200
0.
o O


Column



(b)


Co.... '
Column


15 20
Length


15 20Length
Length


25
(m)


30





H He


SH2


25
(m)


Figure 2.5. Calculated optimum velocity (a) and optimum inlet pressure
(b) as a function of column length for 0.25 mm i.d. open tubular columns
with helium and hydrogen carrier gases illustrating need for sub-
atmospheric inlet pressures with short columns.


'I
I'\


--- Sub-atmospheric Inlet



""--H2

He


//
//
1/
I/
I


.......................... -_ _, _


































Figure 2.6. GC/MS chromatograms obtained with a 3 m x 0.25 mm i.d. open
tubular column at two different inlet pressures: (a) Pi 940 torr, v 297
cm/s, (b) Pi 380 torr, v 120 cm/s. Individual traces are for the
molecular ions of acenaphthylene (m/z 152) and tetradecane (m/z 198), as
well as the reconstructed ion current (RIC) from m/z 35 to 300.










100%.


4.'
(I)



(a) ,

4..
CU
*13


100%l


100%I


1:20


>,
C
4.
e
(b) .


.i 100
It


m/z 152


tetradecane
m/z 198






RIC I


1:25


1:30 1:35


1:40


1:45


1:50


acenaphthylene
m/z 152


2:00 2:05 2:10 2:15 2:20 2:25 2:30


Retention Time (min:s)


1:55


m/z 198


i . .


""'"''








31

It has been suggested that chromatographic analyses be performed at the

optimum practical gas velocity (OPGV) rather than the optimum velocity

[26,27]. The OPGV has been defined as the velocity that yields the

highest number of theoretical plates per unit time [27]. The number of

theoretical plates per unit time can be determined by dividing the number

of theoretical plates by the retention time (i.e., N/tR). Thus, if N/tR

is plotted as a function of the average gas velocity, the OPGV will be

represented by the maximum on the curve. Figure 2.7 shows examples of

these calculated curves demonstrating the determination of OPGVs for 3 m

and 30 m 0.25 mm i.d. columns operated with atmospheric and vacuum outlet.

For this comparison, the value of the partition ratio (k) was chosen to

be 30. Although this is far from the optimum partition ratio of 1.76

suggested by Guiochon [22], a partition ratio of 30 is a typical value

with the inherent low operating temperatures employed with short columns.

Since chromatographic resolution is proportional to the square root of the

number of theoretical plates, the percent resolution lost (%Rost) by

operation at any gas velocity other than the optimum is given by



%RLost 100(1 (N/Npt)1/2) (2.10)



where Npt is the number of theoretical plates obtained at the optimum

velocity and N is the number of theoretical plates obtained at an average

velocity, v.

As shown by the results of this comparison summarized in Table 2.1,

vacuum outlet operation results in the shortest analysis times, regardless

of whether the columns are operated at their optimum velocities or optimum

























' t1


3m





50 m




= Vopt
- Vacuum Outlet
Atmospheric Outlet


100 200
Average Velocity (cm/s)


Figure 2.7. Examples of calculated plots to determine optimum practical
gas velocities (OPGVs) for 3 m and 30 m 0.25 mm i.d. open tubular columns
operated with vacuum or atmospheric outlet. For each plot, the number of
theoretical plates generated per second has been normalized to a maximum
of 100X. The maxima of the curves indicate the OPGVs, while the stars
indicate the optimum velocities. The conditions were the same as those
of Figure 2.2, except k 30.


100


-


I



















\o o\ o.0


o
o







-4
0





4-4
0








I

41















O



O
0




44




0











(.$
0






0
0








o










1.
4









C
a)














r-4W
0
0







0





















.0
4 0
00



e .






. 0








E-i o


Ln


4 CM4
-4


(4 c






CN c


202 U
a 0 e 0
S > 0 >





Sc 0
1n e


co .0 u 4 o 4-


o
0





4
NI







o
U

















0-U
o





m


n or- o


0
r-4

CM



0*
0




e
4-4




14r
Co



0
-4











CO
U
4 ->









P,







S .
0




0



0 0














41 41 0
-IU

0 00

Ul I) 4.)

> >N. 0
r-4 r-4 0?


o .,










-4
Z E




u r
a U






U -U



0 4
-)4









0


on
0

I






0
U









00
a CN




o o
0






o M
cu
0

o o0

a 0-
(0
0) 0)
4.

o >,





0 u



U) U











practical velocities. As stated previously, atmospheric outlet operation

results in a lower minimum plate height (typically 10% lower), and thus

more theoretical plates at the optimum velocity, than vacuum outlet

operation. For example, if operated at their optimum velocities, a 3 m

column with atmospheric outlet would yield 5.6% better resolution than a

3 m column with vacuum outlet. However, the analysis times with the 3 m

column would be 2.8 times shorter if vacuum outlet was used instead of

atmospheric outlet. Moreover, if operated at their optimum practical gas

velocities, columns with vacuum outlet would result in faster analyses and

more theoretical plates (and hence, better resolution) than columns with

atmospheric outlet. For example, analysis times are 0.3 min shorter on

the 3 m column with a 15% improvement in the resolution if vacuum outlet

at its OPGV is used instead of atmospheric outlet at its OPGV. It can be

seen from Table 2.1, that for any column length, vacuum outlet operation

yields the shortest analysis times and the highest number of theoretical

plates per unit time, regardless of whether optimum velocities or optimum

practical velocities are used.

There are other significant advantages of utilizing reduced inlet

pressures with short columns in GC/MS. It has previously been reported

that short columns result in narrower bands and higher sample

concentrations at the detector than long columns [3]; thus, at least from

a theoretical perspective, short columns should provide the best possible

detection limits. In practice, short columns operated at conventional

inlet pressures result in high flow rates of carrier gas into the ion

source. In fact, sub-ambient inlet pressures are a necessity if short

wide-bore columns are to be used at all in GC/MS, since conventional inlet

pressures result in analyzer pressures that are excessive for the normal










operation of the mass spectrometer (see Chapter 3). It is possible that

the advantage of low detection limits obtainable with short columns in

GC/MS might be lost if excessive flow rates (which can disrupt the

ionization process in the ion source) are used. Experience in this

laboratory has shown that the sensitivity obtainable with chemical

ionization (CI) is not greatly affected by the high flow rates of a

different carrier gas (e.g., He or H2) (see Chapter 3). However, when

electron ionization (El) is used the sensitivity decreases dramatically

with increasing flow rates of carrier gas into the ion source, especially

when flow rates above ca. 10 mL/min are used. Thus, lower inlet pressures

are preferred for use with El and short columns. The increased

sensitivity of the vacuum inlet operational mode is demonstrated in Table

2.2, which shows the integrated peak areas of the molecular ion of n-

hexylbenzene obtained with different injection methods and inlet

pressures. As shown in the table, splitless followed by low-pressure

operation at 560 torr yielded the best sensitivity. The vacuum split

injection is inherently less sensitive, since less sample enters the

column and is detected. For the splitless injections, the peak areas were

significantly lower when the higher inlet pressures (and flow rates) were

employed. For example, the peak area obtained by splitless operation with

an inlet pressure of 560 torr was approximately two orders of magnitude

greater than that obtained at 2155 torr.

The effects of splitless operation on retention times was also studied.

Traditionally, both split and splitless operation are performed at

constant inlet pressure. Thus, retention times are the same regardless

of which mode is used, provided that the same inlet pressures are used.

In the splitless method described here, the initial inlet pressure during











Table 2.2. Comparison of peak areas for four different injection of n-
hexylbenzene obtained with different injection methods and inlet pressures
of He carrier gas.


Injection
Method


split
splitlessd
splitless
splitless


Inlet
Pressure
(torr)


560
560
940
2155


Average
Velocity
(cm/s)a


176
176
297
679


Flow
Rate
(mL/min)b


2.6
2.6
5.0
30.0


Peak Area of
M+ (m/z 162)


2.56 x 106
2.09 x 107
7.95 x 106
2.11 x 105


a average velocities calculated with eq 2.6
b indicated flow rates from flow controller
c split ratio 10:1
d inlet pressure 940 torr for initial 12 s, final inlet pressure 560
torr








37

injection is much higher than that during the remainder of the elution.

Since a change in the inlet pressure results in a change in the gas

velocity, it was expected that the retention times for split and splitless

operation would not be the same. In order to determine the effects of the

large pressure change, the retention times for split injections and

splitless injections for a mixture of alkylbenzenes were compared. The

results are shown in Table 2.3. The split injection was performed at a

constant inlet pressure of 560 torr. The splitless injections were

performed with an initial pressure of 940 torr and a final pressure of 560

torr. For splitless operation, the split valve was left closed for 12 s

in one case and for 30 s in another. As shown in Table 2.3, the large

change in inlet pressure does not greatly affect the retention times, even

when a splitless time of 30 s is used. This is most likely due to

trapping of the solutes at the head of the column at the initial column

temperature of 50"C.


Experimental vs. Theoretical Performance


Figure 2.8 shows a comparison of experimental and calculated (from eq

2.2) plate heights obtained with a 3.1 m x 0.25 mm i.d. open tubular

column operated with H2 carrier gas and the column outlet at vacuum. As

can be seen in the figure, the experimental performance at higher

velocities is not as good as predicted by theory. The differences in the

experimental and theoretically calculated data can be attributed to extra

column variances. Extra column contributions are the result of band

broadening due to instrumental time constants and, thus are not accounted

for in the Golay equation (eq 2.1). The extra column variances result in

the addition of a constant band width to the theoretical band width. This










Table 2.3. Comparison of retention times for components of an
alkylbenzene mixture utilizing vacuum inlet split and splitless injection.



Injection Method

Splita Splitless, 12sb Splitless, 30 sb

Component Retention times (min:s)

n-hexylbenzene 1:17 1:14 1:10
n-octylbenzene 2:16 2:14 2:14
n-decylbenzene 3:13 3:12 3:13
n-dodecylbenzene 4:08 4:08 4:09

a inlet pressure 560 torr
b inlet pressure 940 torr for initial 12 s, final inlet pressure 560
torr








39
Inlet Pressure (torr)
400 800 1200
. .* I . I ,. .. I .


(a)


, 10.0

SC
-o 8.0 -

I
6.0


c 4.0


N 2.0
ao
CL


..........


200
Average


400
Velocity


. .
600
(cm/s)


800


/ Theoretical
with extra-column
H = B/v + Cv + D[2


200 400 600
Average Velocity, v (cm/s)


Figure 2.8. Comparison of theoretical and experimental data obtained for
a 3.1 m x 0.25 mm i.d. open tubular column at 100*C with hydrogen as a
carrier gas: (a) peak widths at 10% height and (b) plate heights. The
experimental data agree well with theory if an extra column variance (aec2
- 0.073s2) is included in the calculations.


Experimental
-0-o 1 0.4 s
Theoretical


u.u


0.30


,0.25
E
0.20
I

S0.15


0.10


0-0.05


0.00


(b)








40

is demonstrated in Figure 2.8 (a), which shows the expected (theoretical)

peak widths and the experimentally observed peak widths for n-pentadecane

on a 3.1 x 0.25 mm i.d. open tubular column. The extra column band

broadening results in an additional 0.4 s to the theoretical band width

at 10% peak height. Guiochon et al. have elaborately discussed the role

of extra column effects and have stressed their importance in high-speed

GC applications where short and/or small diameter columns are used [8].

They demonstrated that an additional term (Dv2) can be added to the Golay

equation which can be used to describe the apparent plate height, Ha.



Ha B/v + Cu + Dv2 (2.11)

with:

D ae/(l+k)2L (2.12)



where o 2 is the extra column variance. When eq 2.11 was used with an

extra column variance of 0.073 s2, the resulting curve was an excellent

description of the experimental data, as shown in Figure 2.8 (b). As

shown in the figure and by eqs 2.11 and 2.12, the extra column

contributions become more significant as shorter column lengths or higher

velocities are used. It is believed that the instrumental variances

observed here arise from sample injection rather than detection. The n-

pentadecane used for the plate height determinations was manually injected

as a liquid into the hot injection port. There is expected to be a finite

time required for vaporization of the sample and entrance of the solute

onto the column, which could result in the observed peak broadening. The

residence times of samples eluting from the GC column into the ion source

is not expected to cause a significant amount of band broadening. In








41

addition, the data were collected utilizing a scan time of 0.05 s over the

m/z of interest, which resulted in at least 40 data points per GC peak.

Therefore, no significant time constant was introduced during the data

collection. Because of the extra column effects, the use of short columns

dictates careful consideration of injection technique and injector design

if the maximum performance is to be obtained.


Choice of the Carrier Gas


The importance of the choice of the carrier gas for minimizing analysis

times has previously been discussed by Giddings [6]. Hydrogen, which has

the largest gas viscosity to diffusivity ratio, is predicted to yield the

fastest analyses. This can be seen from the experimentally determined

plate heights for H2 and He on a 3.1 m x 0.25 mm i.d. open tubular column,

as shown in Figure 2.9. Hydrogen is preferred, since the optimum velocity

is higher and the Golay curve is flatter than that of helium. The dashed

line parts of the curves of Figure 2.9 indicate where sub-ambient inlet

pressures were used. The use of H2 results in lower optimum inlet

pressures than He as predicted by eq 2.9 and shown in Figure 2.5 (b).

Even though H2 is predicted to yield the fastest chromatographic

analyses, the effects of its use as a carrier gas on resulting mass

spectra have not been previously studied. When H2 is used a carrier gas,

H3* ions are produced in the ion source. The presence of these ions does

not appear to affect El spectra and in an El source do not cause a

significant amount of chemical ionization (i.e., no increase in the MH+/M+

ratio was observed). This is most likely due to the fact that pressures

in the El ion source are not high enough to cause a significant amount of

reaction to occur. However, the reaction of H3+ ions with sample molecules

















He


0.80



0.60
E



._ 0.40
7-
I)


-0.20



0.00


200
Average


400
Velocity


600
(cm/s)


800


Figure 2.9. Comparison of experimentally determined dependence of the
plate height on average velocity for helium and hydrogen carrier gases
with a 3.1 m x 0.25 mm i.d. open tubular column at 100*C. The dashed
lines indicate where sub-atmospheric inlet pressures are required. The
curves were calculated with eq 2.11.


H2



sub-atmospheric
inlet pressures


.


O !


, 4
gs-"








43

in a chemical ionization (CI) source is possible due to the higher

operating pressures. It was also predicted that the H3+ ions would alter

the mass spectra normally obtained with methane CI. On the contrary, the

presence of H2 carrier gas does not appear to affect methane CI spectra.

As shown in Figure 2.10 (a) and (b), the CI spectrum of 2,4

dimethylaniline obtained with 0.5 torr H2 and 0.2 torr CH4 (b) is the same

as that obtained with 0.2 torr CH4 and no H2 (a). In fact, as soon as

methane is introduced in the ion source with H3+ ions present, the H3* ions

disappear. This is apparently due to the fact that the hydrogen reagent

ions (H3*) are used in the protonation of methane. This is not surprising,

since methane has a higher proton affinity than hydrogen [28]. It is

expected that similar behavior would be observed with other commonly used

proton transfer CI reagent gases (e.g., isobutane, ammonia, etc.) because

of the low proton affinity of hydrogen. In the absence of methane,

hydrogen can be used as both the carrier gas and the CI reagent gas, as

shown in Figure 2.10 (c). The spectra that are obtained with hydrogen CI

exhibit more fragmentation than those obtained with methane CI, which is

due to the greater exothermicity of the proton transfer reaction between

the H3+ reagent ions and the sample molecules. With hydrogen CI, no adduct

ions are observed as is normally the case with methane CI. The absence

of these adducts is often desirable, since their presence may complicate

interpretation of the mass spectra.


Conclusions


A better understanding of the influence of the pressure drop on the

chromatographic behavior of open tubular columns in GC/MS has allowed for

better utilization of the inherent advantages associated with short














Positive ion CI spectra of 2,4-dimethylaniline (MW 121)


1ool (a)


(.J *-- 1--1- '----.--- 1. .1


60 80 100


40 1 6 1 I
140 160 180


(b)


100 120


140 160 180 200


122
121"


107

106
^ -, ___ -JL -__,___[


140


160 180


m/z


Figure 2.10. Effect of using hydrogen carrier gas on the chemical
ionization mass spectra of 2,4 dimethylaniline: (a) 0.2 torr methane and
no hydrogen present, (b) 0.2 torr methane with 0.5 torr hydrogen present,
(c) 0.5 torr hydrogen and no methane present.


NH,



CH,


100-









0-


60 80 100










columns. The large pressure drop associated with vacuum outlet GC should

not be considered a disadvantage. On the contrary, the low-pressure

outlet results in increased optimum gas velocities and decreased analysis

times. It has been shown that vacuum outlet operation yields the shortest

analysis times, regardless of whether optimum velocities or optimum

practical velocities are used. In addition, the chromatographic

resolution obtained when optimum practical gas velocities are used is

predicted to be higher for vacuum outlet than for atmospheric outlet

operation. As shorter column lengths are used with vacuum outlet

operation, lower inlet pressures are needed to obtain optimum performance.

Sub-ambient inlet pressures can be obtained with a relatively simple flow

control system as described here. The capability of splitless operation

has greatly extended the utility of vacuum inlet operation. The lower

inlet pressures with short columns are also shown to provide the best

sensitivity when electron ionization is used, since the flow rates are

lower. The use of hydrogen carrier gas results in even lower optimum

inlet pressures and higher optimum velocities than helium. Hydrogen can

be used as the chemical ionization reagent gas or it can be used in the

presence of other commonly used proton transfer chemical ionization

reagent gases without affecting the mass spectra.
















CHAPTER 3
CONCEPTS FOR PORTABLE GAS CHROMATOGRAPHIC
INSTRUMENTATION IN GC/MS: FLOW RATE
PROGRAMMING AND DIRECT COLUMN HEATING


Introduction


The object of any chromatographic analysis is the separation and

identification of components in mixtures. However, chromatographic

mixture analysis is often complicated by the "general elution problem"

[29]. This predicament results both in poor resolution in the early part

of a chromatogram due to coeluting peaks, and in severe peak broadening

near the end of a chromatographic run. In gas chromatography, temperature

programming is often used to circumvent this problem. In temperature-

programmed GC (TPGC), the column temperature is increased (usually

linearly) during the course of the analysis. Since the temperature range

covered can be up to 300C (and higher depending on the stationary phase),

TPGC offers a wide dynamic range. In analyses where temperature

programming is used, the rate at which samples can be analyzed is not only

dependent on the time required by the separation, but also on the time

required to equilibrate (cool) the column for the next injection. Since

the column temperature is changed by changing the temperature of a large

gas chromatograph oven, the rate at which a GC column can be heated or

cooled in TPGC is usually slow (typically 50*C/min or less depending on

the manufacturer of the GC oven). Moreover, the need to heat a GC oven by

conventional convection heating, only to cool it back down, requires










significant amounts of electrical power, making temperature programming

unattractive for portable instrumentation.

In this chapter, two approaches to solving these problems are

presented. The first approach involved investigation of flow rate

programming as a robust alternative to temperature programming. The

second approach examined was direct resistive heating of metal-clad

capillary GC columns, as opposed to oven (convection) heating. Direct

resistive heating is especially exciting, and should be widely applicable

in portable GC instrumentation for environmental or process monitoring.

The concepts developed in these studies have led to the development of a

portable GC probe, which appears to be particularly attractive for future

applications involving portable mass spectrometers or as a convenient

analytical tool for that could be used in any mass spectrometry

laboratory.


Flow Rate Programming


Flow programming of the carrier gas has previously been examined as

an alternative to temperature programming [25,29-35]. Use of this

technique was limited in the early years of application due to the high

inlet pressures required with packed columns. The high permeability of

short and/or wide-bore open tubular columns allows for high flow rates to

be achieved at relatively low inlet pressures [34]. In addition, it has

more recently been shown that exponential flow rate programming under

isothermal conditions results in peak distributions that are very similar

to those obtained with linear temperature programming [29,33-35]. One

advantage of programmed flow GC is that the sample throughput is high,

since it is not necessary to wait for the column to cool for the next








48

injection. Although resetting of the column flow rate is required, this

step is essentially instantaneous due to the high permeability of open

tubular columns [34]. This technique might prove attractive for a field

instrument that could not be rapidly heated and cooled, or in the

laboratory where high-throughput analyses are desired. Flow rate

programming is well suited for the analysis of thermally labile compounds,

since the column can be operated isothermally at relatively low

temperatures. It has also been pointed out that bleed of the liquid

stationary phase from the column is reduced since lower temperatures are

required [34]. This could prove to be an important factor in the

reduction of chemical noise at the detector, or could help to extend the

life of GC columns. One disadvantage of the method that is evident from

GC theory, is that chromatographic efficiency is reduced if the flow rate

is increased beyond the optimum. However, as was shown in Chapter 2,

optimum flow rates are higher when the column outlet is at vacuum as in

GC/MS. In addition, the rate of increase of the plate height is much less

for short columns than for long columns, when higher than optimum

velocities are used. Nevertheless, as will be shown in this chapter, the

maximum allowable flow rate is ultimately dependent on the pumping speed

and associated performance of the mass spectrometer.


Direct Resistive Heating of Al-clad Capillary Columns


Another alternative that avoids many of the limitations of oven

heating is to heat the column directly instead of by convection.

Recently, aluminum-clad fused silica open tubular (Al-clad FSOT) capillary

columns have become commercially available (SGE, Quadrex). The potential

advantages of directly-heated capillary GC columns have been suggested








49

[36], but never demonstrated. In Chapter 4 it will be shown that these

directly-heated columns can be used as simple transfer lines for

introducing samples via GC into the collision cell of a triple quadrupole

mass spectrometer. In this chapter, a method is described for heating

these columns directly by passing an electrical current through the thin

aluminum cladding coated on the outer surface of the column. A method of

sensing the temperature of the Al-clad capillary directly, by measuring

the column resistance instead of using external thermocouples or other

temperature sensors, is also presented.

The principles of resistive heating are relatively simple. When a

voltage is dropped across an Al-clad capillary, the column temperature

increases depending on the amount of power dissipated. The power

dissipated, P, is dependent on the current through the column, i, and the

voltage across the column, V.



P iV (3.1)



The current is dependent on the voltage drop and the electrical

resistance, R.



i V/R (3.2)



The electrical resistance is dependent on the length, diameter, and

thickness of the Al coating on the outside of the capillary. The amount

of heat (or energy) required to increase the temperature of any substance

by an increment AT is given by [37]








50

Q mCAT (3.3)



where Q is the amount of heat (or energy, J or W-s), m is the mass of the

material (g), C is the specific heat of the material (J/g*C), and AT is

the change in temperature of the material (*C). In addition, since the

heat is produced resistively, the amount of energy consumed over a given

time period At is given by



Q PAt (3.4)



From eqs 3.3 and 3.4 it is apparent that objects of large mass require

more heat or more power over a given time period to reach a given

temperature. Al-clad FSOT GC columns are coated with a very thin aluminum

film, and have a very small thermal mass. The low thermal mass of these

columns allows them to be rapidly heated and cooled with much less

electrical power than is normally required with chromatograph ovens.

Since no large oven is needed, direct resistive heating is extremely

attractive for use in portable GC instrumentation.

The low thermal mass of the Al-clad columns does present some problems

if conventional methods are used to measure the column temperature.

Currently, there are no thermocouples or resistance thermometers available

that are small enough to accurately measure the temperature of these

columns. Even if small sensors are used, placing the sensor against the

column causes a local cold spot and leads to inaccurate temperature

measurement. An alternative method of measuring the column temperature

that avoids this problem is to use the column itself as the temperature

sensor. The resistance of any metallic conductor is linearly related to








51

its temperature over a wide temperature range and is given by [38]



RT Ro(1 + aT) (3.5)



where RT is the resistance (0) at temperature T ("C), Ro is the resistance

at 0"C, and a is the temperature coefficient of resistivity of the metal

(*C-1). In fact, eq 3.5 is the basis of operation of Pt resistance

thermometers, which are widely used in scientific applications. The

resistance of a column is readily calculated during resistive heating if

the current through the column and the voltage across the column are

measured. Once a calibration of temperature versus resistance has been

performed, a determination of the column resistance, can be used as a

direct measure of the column temperature.


GC Probe


The concepts discussed above have been applied in the development of

a compact GC probe, which resembles a conventional direct insertion probe,

and interfaces with a commercially available mass spectrometer. A GC

probe utilizing packed columns operated isothermally has previously been

described [39]. The GC probe described here allows the temperature to be

rapidly cycled for high-throughput GC/MS analyses requiring very little

electrical power. Removal or insertion of the probe from the mass

spectrometer is through the normal probe inlet assembly. This facilitates

changing of the column or allows for rapid changing to other techniques

(e.g., solids probe, FAB, etc.) with minimal effort. Once the probe is

removed from the mass spectrometer, the column can be rinsed with solvent

without removing the column from the probe. This is attractive in cases










when particularly "dirty" samples are being analyzed, as may be the case

for environmental monitoring in the field.


Experimental Section


Experimental Conditions


Mass spectrometry. A Finnigan MAT TSQ70 triple quadrupole mass

spectrometer was used for this work. The instrument was tuned with FC43

(perfluorotributylamine) with the GC probe in place. The electron

multiplier was maintained at 800-1200 V with the preamplifier gain set at

108 V/A. Electron energies of 70 eV and 100 eV were used for electron

ionization (EI) and chemical ionization (CI), respectively. Normal Ql

mass spectra were acquired for all MS experiments. Methane reagent gas

was used in all CI experiments with ion source pressures of 0.2-0.5 torr.

The same removable ion volumes commonly used with the direct insertion

probe were used with the GC probe.

Flow rate programming. All experiments with flow rate programming

were performed on a Varian model 3400 GC equipped with a split/splitless

injector. Open tubular GC columns, approximately 3 m in length with an

internal diameter of 0.25 mm and a 0.25 pm film thickness (DB5, J&W

Scientific) were used. Injection temperatures were typically 220*C;

column temperatures are reported with each analysis.

GC probe. Wide-bore (0.53 mm i.d.) Al-clad FSOT columns with a non-

polar 1.0 pm film BP-1 stationary phase (SGE) were used with the GC probe.

Narrow-bore columns (0.33 mm i.d., 0.5 pm film, BP-5 stationary phase)

were evaluated in terms of heating requirements; however, for all

chromatograms obtained with the GC probe, wide-bore columns were used,

which allowed for a mechanically simple injection port.










Flow Control System


Flow rate programming. For many of the experiments performed (e.g.,

exponential flow rate programming and sub-ambient inlet pressure GC) it

was necessary to be able to control the column flow rate instead of the

column inlet pressure. Several methods for programming the column flow

rate have appeared in the literature [25,29-33]. For example, a very

simple method of flow programming was achieved by manually manipulating

a needle valve in the splitter line [33]. However, due to the imprecision

and inconvenience of this method more reliable methods were developed.

Others have achieved flow control via feedback from a pressure transducer

with a motor-controlled regulator valve [29,34,35]. This type of system

first requires a calibration step to determine the dependence of flow rate

on inlet pressure. After the flow function is determined and programmed

into the computer, the pressure transducer detects any difference between

the inlet pressure and the setpoint flow. If a difference is detected,

a DC motor adjusts the regulator until the correct flow is achieved. The

main disadvantage of this method is the slow response time, since several

seconds are required for the motor to reposition the valve. The response

time of the flow controller becomes even more important for short columns

since the analysis times are typically only a few minutes long. For this

reason it was decided that commercially available mass flow controllers

would be most suitable for these applications. These types of transducers

incorporate thermal mass flow sensors to determine mass flow rates of

gases. Sensors placed at both ends of a laminar flow tube detect

differences in heat transferred along the tube, and directly relate this

difference to the flow rate of the input gas. The measured flow rate is

compared to the setpoint value and is adjusted by an electromagnetically








54

controlled automatic valve. The transducer and controller valve are

contained in a single package measuring approximately 1/2 in. x 5 in. x

5 in. An external power supply/readout unit provides the appropriate

power, real-time readout of the measured flow, and capability for computer

interfacing. The advantages of this system are fast response (500 ms for

the transducer), excellent precision (0.2% of full scale), and high

accuracy (0.5% of full scale). The flow controller is interfaced to the

mass spectrometer electronics via one of two digital-to-analog converters

(DACs, known as user outputs) that are variable over the range 5 V DC

via the TSQ70 trackball or Instrument Control Language (ICL) procedure.

A control voltage range of 0-5 V is used to adjust the flow controller

flow rate over its full range.

One concern about the use of a mass flow controller was its ability

to regulate flow when the output was connected to the vacuum of the mass

spectrometer, as it might be in short-column or vacuum inlet GC. In order

to test the performance of the transducer in this configuration, the flow

controller was utilized to control the input of collision gas into the

collision cell of the TSQ70. In fact, this configuration worked so well

that a flow controller dedicated for control of collision gas pressure has

since been installed.

For exponential flow rate programming experiments, the flow control

system was essentially the same as that described in Chapter 2. The only

difference was that, instead of a MKS 1159A mass flow controller, a MKS

1259B mass flow controller was used, which has a flow control range of 0-

20 mL/min for nitrogen and 0-29.6 mL/min for helium. It was later found

that the flow rate range of this flow controller limited the experiments

that could be performed. For example, during flow programming








55

experiments, the split and sweep valves had to be left closed during the

entire GC run to obtain the widest range of useable flow rates, which

limited analyses to splitless operation. A flow controller with a wider

flow rate range (0-100 mL/min for nitrogen) was eventually installed to

alleviate this problem. The flow function for an exponential flow program

is controlled automatically with the DAC user output of the TSQ70 with an

Instrument Control Language (ICL) program. The flow rate at time t, Ft,

is given by


kt (3.6)
Ft -F +et


where Fo is the initial flow rate and k is a constant. In a flow rate

programming experiment, the initial and final flow rates, as well as the

time for the delay and end of the programming period are selected and

entered into the computer program. The constant, k, is calculated in the

computer program such that the flow rate at any time during the analysis

is determined by eq 3.6.

GC probe. Since short wide-bore columns (typically 3 m in length)

were used in the GC probe, the injection port could not be operated at

conventional injection port pressures (i.e., atmospheric pressure and

above) without a carrier gas flow rate exceeding the normal operating

pressure of the mass spectrometer. Instead, the injection port was

maintained at lower pressures (typically 100-200 torr) with the flow

control system described in Chapter 2. This system utilized a MKS model

1159A flow controller with a range of 0-100 mL/min for nitrogen and 0-145

mL/min for helium. The use of sub-atmospheric inlet pressures did require

some consideration of injection technique, since it was possible for the

sample to be prematurely aspirated from the syringe by the vacuum in the








56

injection port. It was found that the best way to avoid this problem was

to use a gas-tight syringe and to draw the sample as well as a small

volume (e.g., 1 pL) of air into the syringe barrel prior to injection.

The column flow rate of He carrier gas used with the sub-ambient

inlet pressures was typically 3.5 mL/min. Flow rates were determined by

measuring the flow output of the flow controller with a soap bubble meter

prior to connecting the carrier gas line to the injection port. Davies

has shown that, under the conditions of vacuum outlet, flow rates can be

calculated by [40,41]



Q 64itL3/9t02 (3.7)



where Q is the flow rate (mL/s), to is the retention time of the unretained

air peak (s), q is the gas viscosity (poise), and L is the column length

(m). However, when short columns are used, the air peak times are so

short that they are not easily measured. For example, a 3 m column

operated at a typical gas velocity of 300 cm/s has an air peak time of 1

s. As discussed in Chapter 2, the average gas velocity can be calculated

if the inlet pressure is known. However, due to the very simple design

of the injection port of the GC probe, a conventional pressure gauge could

not be used to measure the inlet pressure. Therefore, it was more

convenient to measure flow rates instead. Once the flow rates were

determined, the inlet pressures and gas velocities were readily

calculated. Given that the average gas velocity is


v L/tO


(3.8)








57

Average velocities can be calculated if eq 3.8 is substituted into eq 3.7

yielding



u -(375Q/ L) (375Q/16L) 9)



where Q is in mL/min, L is in m, q is in poise, and v is in cm/s. The

inlet pressure (Pi, dynes/cm2) can be calculated from the Poisuelle

equation for vacuum outlet conditions [14]



Pi 32vqL/3r2 (3.10)



where r is the column radius in cm, and L is in cm. A 3 m wide-bore

column, operated at a typical flow rate of 3.5 mL/min of He at 25*C, has

a calculated gas velocity of 204 cm/s and an inlet pressure of 146 torr.


Direct Resistive Heating


Electronics. A programmable DC power supply was constructed in-house

and was used to control the voltage across the Al-clad GC columns. The

electronic components for the power supply are contained in a box which

is approximately 10 in. x 6 in. x 3.5 in. A schematic of the overall

system electronics is shown in Figure 3.1, with details of the DC power

supply shown in Figure 3.2. The output of the power supply (up to 30 V)

is controlled by an input control voltage of 0-5 V DC. A DAC from the

TSQ70 was used to generate a control voltage for the power supply. The

control voltage (from the potentiometer or DAC) is fed into a voltage

follower with variable gain. The output of the op-amp provides a bias to

the base of power transistor that determines the output of the supply.







































































































-J


(0


o




c0

44



0
44


















0
C(










o

4-1
0

























0
(U


















*4









4r.
0
0





















o
1*1







cn
0




































*C
o
O
0


C
U4
U)
0)

0


0

(4
0
bO
*-I


U



*U








(r
CK




























C>
U-



>e
v,,

o


*o >
S- 10
0 4.


40


0


U n u


"4
41
U






O
0
u




0











3




0






0













54
0
*M



















0

0

*4
"4-


41


$'4







bO
bo














































1-4
04
*

















I-



^1
00











Ul
1-









>-

r
*-I








60

The output of the power transistor is rated at 30 V and 30 A. However,

in the present configuration the output of the supply is limited by the

transformer, which is rated at approximately 6 A. Typically, the gain of

the voltage follower was set such that the power supply yielded the

maximum desired temperature at the maximum control voltage. This

versatility is important, since a change in the length or diameter of the

column changed the amount of power required for the column to reach a

particular temperature. No additional circuitry had to be designed to

measure the column voltages and currents. The voltage drop across the

column, as well as the voltage drop across a 0.261 0 current sensing

resistor were measured with an external devices sensor circuit already

present in the TSQ70 electronics. This circuit consists of five

differential amplifiers, an analog multiplexer, and an analog-to-digital

converter (ADC). The voltages to be determined were fed into two separate

differential amplifiers of the sensor circuit. This allows the voltages

to be displayed and/or plotted in real time with the computer-controlled

TSQ70 and the ICL. Programs were written in ICL that allowed the voltage

across the column to be increased linearly with time during the

acquisition of data. The linear voltage ramp was used to simulate the

linear temperature ramp commonly used in temperature programming.

However, it should be pointed out that any desired voltage ramp could be

created with this system (e.g., exponential, logarithmic, etc.), by a

simple modification of the ICL program.

Temperature sensing. The resistance of the column, obtained by

measurement of the voltage and current, was used as an indication of the

column temperature. Al-clad GC columns were "calibrated" by placing them

in a GC oven (Varian model 3400) and recording the resistances of the








61

columns as a function of GC oven temperature. A wide-bore Al-clad FSOT

column (0.53 mm i.d. x 3.14 m), and two different lengths (4.77 m and 3.05

m) cut from the same narrow-bore (0.33 mm i.d.) Al-clad FSOT column were

calibrated using this method. The nominal resistances of these columns

were very low (ca. 0.9 D/m for the narrow-bore column and 0.5 G/m for the

wide-bore columns), and thus were too small to be accurately and precisely

measured with a typical digital multimeter (DMM). In order to obtain

accurate resistance measurements for the calibration, the DC power supply

was used to apply a small voltage drop (e.g., 50 mV) across the column.

The current through the column and the voltage drop across the column were

measured with two DMMs (Fluke model 75). The resistances were then

calculated from the voltages and currents measured. The voltages applied

during calibration were not high enough to cause a significant amount of

column heating.


GC Probe Design


The GC probe is schematically illustrated in Figure 3.3. The 1.27 cm

x 28 cm stainless steel probe shaft was adapted from a Finnigan 4500

series ion volume insertion/removal tool. The probe shaft is threaded and

screws into one of the two 4 in. square end-blocks (machined from phenolic

material). The injection port is positioned in a milled slot on the other

end-block and is held firmly in place by a set screw. The column is

wrapped around a 2.5 in. diameter Teflon spool. The spool is held between

the two end-blocks and contains a slot that allows the column to be

inserted through the spool and down the length of the probe shaft. The

column is covered with Nextel glass braid insulation (Omega Engineering)

and is coiled around the Teflon spool. The insulation serves to
















o .Z
0 _0
u o s-i-
.J I-W
LI u z
U UZ
I oUJ
-Z-JO









d I-' -
Z <



-CO




m;<_1 ------ -

UZ
2 0


U 0









0Wi
w 0,
xz
CLw











Ct-


CLW





U










z
0









u


z

Iz


a


Ia
41)
-S










electrically isolate each strand of the coiled column. A 1/16 in.

stainless steel union (Swagelok) with a graphitised vespel ferrule and a

Viton o-ring are used to form a vacuum-tight seal at the end of the probe

shaft. Approximately 2 cm of the column extends from the end of Swagelok

union. A copper connector slides over this last few cm of the column and

electrically connects the end of the column to the probe shaft.

Electrical leads from the DC power supply plug into the banana jacks

mounted on the probe assembly. It is very important to minimize the

contact resistance of the electrical connections to the column, since poor

connections result in power consumption and local heating at the point of

contact. In addition, large connectors have a large thermal mass and

dissipate heat near the contact. For these reasons, the electrical

connection to the head of the column was made with a small removable clip

lead. The probe shaft, which is used as a ground return, remains cool

during column heating, since the resistance of the probe is much less than

that of the columns used. The nominal resistance of the column in the

probe was always checked prior to applying power to insure good electrical

connections. The GC probe shaft is inserted through the probe inlet

assembly and the end of the probe fits firmly against the ion volume

contained in the ion source block.

A compact, low thermal mass injection port was fabricated from a 1/8

in. x 1/16 in. stainless steel tube reducing union (Swagelok). For the

introduction of carrier gas into the injection port, a hole was drilled

into the side of this fitting and a 1/16 in. o.d. stainless steel tube was

silver-soldered in place. The wide-bore column is inserted into the

smaller end of the union and is sealed with a nut and a graphitised vespel

ferrule. The top of the injection port contains an 1/8 in. nut and a








64

septum. A standard GC syringe (SGE), equipped with a 26-gauge needle, was

used to inject samples directly onto the wide-bore columns. The inside

of the injection port contains a stainless steel guide, which directs the

syringe needle to the column entrance. Upon injection, the syringe needle

enters the column and is inserted far enough that the sample is injected

past the point where the electrical connection is made. This minimizes

sample loss due to unvolatilized components. The injection port remains

cold when the column is heated, and thus does not have to be cooled

between injections.


Results and Discussion


Flow Rate Programming


Effect of carrier gas flow rates on mass spectrometer performance.

Exponential flow rate programming has been used in many instances with

conventional GC detectors [25,29-35]; however, the effects of the

resultant high (and varying) flow rates on the operation of the mass

spectrometer have not been studied. One problem that became apparent from

some initial experiments employing electron ionization (El) was the

reduction in the analyte signal (sensitivity) associated with high flow

rates of carrier gas into the mass spectrometer ion source. This was

briefly addressed in Chapter 2. Initially, it was unclear whether this

problem was linked to mass analysis, ion formation, or a combination of

both. It is possible that the increased pressure in the mass analyzer

(which causes a decrease in the mean free path of the ions) results in

scattering losses and a reduction of the number of ions reaching the

detector. Analyte signal may also be decreased due to inefficient ion

formation with El. This associated decrease in sensitivity with El has








65

also been observed in supercritical fluid chromatography/mass spectrometry

(SFC/MS), where the pressure in the ion source is elevated due to the

presence of the constantly eluting mobile phase [42].

It was discovered that the problem of reduced analyte sensitivity with

high carrier gas flow rates could be overcome by employing chemical

ionization (CI) instead of El. Due to the presence of a reagent gas

(e.g., methane) at much higher operating pressures (typically 0.1-1 torr

for CI vs. 10-4 torr for El), ion formation with CI appears unaffected by

the increase in carrier gas flow rates. Figures 3.4 and 3.5 show a

comparison of the performance of El and methane CI respectively under

identical changes in column flow rates. For these experiments, the

chromatograms were obtained in the splitless mode with a 3.2 m, 0.25 mm

i.d., DB-5 capillary column operated isothermally at 80*C. The sample

injected was 30 ng each of C14, C15, and C16 n-alkanes. Figure 3.4 shows

the chromatograms of the major fragment ion (m/z 85) obtained with El.

For Figure 3.4 (a) the column flow rate was 8 mL/min and that for Figure

3.4 (b) was 20 mL/min. Both traces are plotted on the same time and

intensity scale for ease of comparison. The integrated peak areas for the

components at the two flow rates are given in Table 3.1. As shown in the

table, the m/z 85 peak areas for the three hydrocarbons in a particular

chromatogram are reproducible within the expected limits of error.

However, the areas of the peaks at the higher flow rate are approximately

a factor of four less than the peak areas obtained at the lower flow rate.

Although the resulting sensitivity for only one particular ion is

considered in this comparison, all of the sample ions were observed to

decrease by approximately the same amount. The major sample ions obtained

with methane CI of n-alkanes are formed by hydride abstraction to yield










66





Aiisueiul. enlosqv y 5
















(N r
10 0 T> 2
S'0


C, u















0- E
00
0 0

























N 0
oc

0
























04
0
$4

























I0
( 4-4
o 0 z

1c. 4 .34.











C.)
CU 0S






- 04)






CU S
E L~ -40










67
*4 0


co






t o
0 O 0


N0



Nr04
(0d


Qb.4

0 -
SCM







too
C to e
} ..4 ) >


CMe ,o
N 0 4)


.00
C r
0 0^ >




C C
N N 0



C 0)) 5 *4





E 3 0




N 3 _
0
4) ,









4 %.3 -4

C(4 co -
U~E e










Table 3.1. Integrated peak
He carrier gas flow rates.


areas for comparison of El and CI at different


Flow Rate
(mL/min)


Hydrocarbon


85
85
85

85
85
85

197
211
225

197
211
225


Ionization
Method

El
El
El

El
El
El

CI
CI
CI

CI
CI
CI


Peak
Area


220694
213729
227111

53244
53148
52292

1893613
1827399
1617471

1758975
1804526
1679119










(M-H)* ions. Figure 3.5 shows the mass chromatograms of these ions for

the two flow rates. The resultant integrated peak areas for these ions

are shown in Table 3.1. As shown in the table, the variation in the peak

areas when the flow was increased from 8 to 20 mL/min is insignificant and

within the limits of experimental error. These results indicate that the

chemical ionization process is less affected by the increased He carrier

gas flow rates. This may be due to the fact that sample molecules in the

CI ion source exist in a large excess of reagent molecules, thus the

variation of He flow rates does not significantly affect sample

ionization. The increased sensitivity (larger signal intensities and peak

areas) obtained with CI for this alkane mixture can be attributed to the

differences in the amount of fragmentation obtained with El and CI. The

El spectra exhibited a large amount of fragmentation, and thus resulted

in an abundance of fragment ions. In contrast, abundant pseudo-molecular

(M-H*) ions were obtained with CI with less fragmentation than EI. Thus,

the integrated peak areas for one selected ion were larger for CI than for

EI.

Once it was determined that EI, but not CI, sensitivity was degraded

by high flow rates of He carrier gas, it was postulated that increasing

the electron energy might help to improve the sensitivity of El with the

high flow rates of carrier gas. The electron energy, which is the

potential difference between the filament and the source block, represents

the kinetic energy of the electrons impinging upon the sample molecules.

The electron energy is usually set higher in CI than in El (100 eV instead

of 70 eV) to insure that the electrons have sufficient energy to penetrate

the source region containing ca. 1 torr of reagent gas. However,

experiments have shown that an increase or decrease in electron energy








70

and/or optimization of any of the other ion optical parameters does not

improve the performance of El when relatively high flow rates are used.

These data indicate that CI is the ionization method that should be used

if high flow rates (>10 mL/min) of carrier gas are to be used. This is

not a severe limitation since CI is often more sensitive than El and is,

thus, usually preferred for trace analysis. In addition, since CI is a

"softer" ionization method (less fragmentation than EI), molecular weight

information is often more readily obtained. If more fragmentation were

desired, it might be possible to employ charge exchange ionization instead

of El or CI, as shown in Chapter 4, and as has been done in SFC/MS [42].

Comparison of temperature programming and flow rate Drogramming. In

order to illustrate the similarities and differences of linear temperature

programming and exponential flow rate programming, a series of

chromatograms was obtained under each set of conditions. In these

experiments, a 3 m x 0.25 mm i.d. DB-5 capillary column was used. The

sample was a GC test mixture obtained from J&W Scientific that is often

used for evaluating the chromatographic performance of non-polar bonded-

phase columns. The sample contained 23 ng/pL of each of seven components,

which are listed along with their boiling points in the caption of Figure

3.6.

Figures 3.6 and 3.7 show the separation of this mixture with

exponential flow rate programming and temperature programming,

respectively. Figure 3.6 (a) shows the chromatogram obtained with

isorheic (constant flow rate) operation at 5 mL/min of He carrier gas and

an isothermal column temperature of 65C. Figures 3.6 (b) and (c) show

the mixture separated isothermally at 65*C using exponential flow rate

programming. The rate of the exponential was adjusted such that the final

































Figure 3.6. Reconstructed ion chromatograms (methane CI) of GC test
mixture obtained on a 3 m column with isothermal operation at 650C and
various flow conditions of He carrier gas: (a) isorheic at 5 mL/min; (b)
exponential flow program from 5 mL/min to 30 mL/min in 4 minutes; (c)
exponential flow program from 5 mL/min to 30 mL/min in 3 minutes. The
components of the mixture and their boiling points are: 1, 2-chlorophenol
(175*C); 2, n-undecane (196*C); 3, 2,4-dimethylaniline (214*C); 4, 1-
undecanol (243*C); 5, acenaphthylene (270*C); 6, tetradecane (232*C); 7,
pentadecane (271*C).














3








(a) I
RIC






5
2


4
7

I I I I
0:21 1:55 2:20 5:11
Retention Time (min:s)



30 mLmin
-RIC
3
-flow program







c5c
(b) i



2

5
mLImin 4 7



0:21 1:55 2:18 4:01
Retention Time (min:s)


30 mL/min
-RIC

3

-flow program
low program


0
*C

(c) )

56

2
I /

5 2
mLlmin 4A


Retention Time (min:s)





































Figure 3.7. Reconstructed ion chromatograms
with isorheic operation at 5 mL/min He
temperature program rates: (a) 2*C/min; (b)
components of the mixture are identified in


of GC test mixture performed
carrier gas and different
5*C/min; (c) 15C/min. The
Figure 3.6.


















(a) 6








C














a5 C


(c)


Retention Time (min:s)








75

flow rate of 30 mL/min was reached after 4 minutes and 3 minutes for

Figures 3.6 (b) and (c), respectively. Figure 3.7 illustrates isorheic

separations of the mixture with temperature programming at rates of

2*C/min, 5*C/min, and 15"C/min. It can be seen from Figures 3.6 and 3.7

that exponential flow rate programming yields similar results to linear

temperature programming.

It is evident from this comparison that a change in flow rate does not

influence retention as much as a change in temperature. For example, a

temperature program rate of 2*C/min and a resulting temperature increase

of 8C over a period of 4 minutes (Figure 3.7 (a)) yielded nearly the same

results as an exponential flow program where the flow rate was increased

sixfold from 5 mL/min to 30 mL/min in 4 minutes (Figure 3.6 (b)). The

differences of temperature and flow rate programming are readily

understood if the thermodynamics of the chromatographic process are

considered. The retention time, tR, of a solute is given by [43]



t, L(l+k)/u (3.11)



where u is the average carrier gas velocity, L is the column length, and

k is the partition ratio; thus, the retention time is inversely

proportional to the average velocity. It is also evident from eq 3.9 that

under vacuum outlet conditions the average velocity is proportional to the

square root of the column flow rate; therefore, an increase in flow rate

by a factor of 6 (e.g., from 5 to 30 mL/min) would decrease the retention

time by only the square root of 6 (or 2.4). In contrast, a change in

column temperature affects the equilibrium of the solute between the

mobile phase and stationary phase, and thus results in a change in the








76

partition ratio, k. The effect of the column temperature on the partition

ratio is given by [43]



In k -AG/RT In B (3.12)



where AG is the change in free energy for the evaporation of the solute

from the liquid stationary phase, T is the absolute temperature, R is the

gas constant, and B is the volume ratio of the mobile phase to the liquid

stationary phase. Consider n-pentadecane, which has a retention time of

311 s on the 3 m x 0.25 mm i.d. column at 65*C. The associated values of

k, AG, and B for these conditions are 259, -7.9 kcal/mol, and 500,

respectively. In order to decrease the retention time of this compound

by a factor of 2.4 (the same factor afforded by a factor of 6 increase in

flow rate), the partition ratio factor (l+k) of eq 3.11 must also be

reduced by this amount (i.e., k must equal 108). From eq 3.12, the

temperature must be increased by only 27*C (i.e., to 92C) to decrease the

partition ratio and the retention time by a factor of 2.4.

The calculations discussed above as well as the experimental data

demonstrate that a change in temperature has a greater effect on retention

than does a change in flow rate; therefore, the range of boiling points

spanned in a flow program will not be as large as that obtainable with

temperature programming. Nevertheless, the chromatograms obtained with

flow programming are certainly an improvement over isothermal operation

(Figure 3.6). The range of boiling points covered is still relatively

wide, as the components in the test mixture have boiling points ranging

from 175C (2-chlorophenol) to 2710C (pentadecane). Perhaps the real

advantage of flow rate programming is the low elution temperatures. The








77

components of Figure 3.6 eluted as much as 200*C below their boiling

points. This benefit might be of particular value in the separation of

thermally labile components or if limited power were available for the

heating of a portable GC.


Direct Resistive Heating


Calibration of columns for temperature sensing. Data obtained for the

calibration of the column resistance vs. temperature are shown in Figure

3.8. The column resistances were normalized by dividing by the length of

the column used in the calibration. Length normalization was necessary,

since the same length of column was not always used in the GC probe. Once

the calibrations were performed, the temperature of a resistively-heated

column could be determined from the relationship



T aR/L + b (3.13)



where T is temperature (*C), R is the resistance (0), L is the column

length (m), and a and b are the slope and y-intercept of the linear

calibration curve, respectively. The slopes, y-intercepts, and standard

deviations from the linear least squares fits of the calibration data are

shown in Table 3.2. As shown in Figure 3.8 and Table 3.2, essentially the

same calibrations were obtained for the two different lengths of the

narrow-bore column. This is an indication of the uniform thickness of the

Al coating. Since these columns were cut from the same original column,

no generalizations can be made about the column-to-column uniformity. As

shown in Figure 3.8, the resistance of the wide-bore column is less than

that of the narrow-bore column. This is due to the fact that there is a
















0.53 mm i.d.
ooao L = 3.14 m


( 0.33 mm i.d.
ooooo L = 3.05 m
SAAA L = 4.77 m


Figure 3.8.
(0/m) for
temperature
calibration


Calibrations of column temperature vs. normalized resistance
two different diameter Al-clad capillary columns. The
of a resistively-heated column could be determined from the
line with knowledge of the column resistance and the length.


270


220




0
L 170


.120
E








79

Table 3.2. Results of linear least squares fits of temperature
calibrations shown in Figure 3.8.


slope s.d.8 v-intercept s.d.b


mm i.d. x 3.05 m
mm i.d. x 4.77 m
mm i.d. x 3.14 m


276.7 0.7
280.4 0.9
522.2 1.7


-217.9 0.4
-220.8 0.5
-221.7 0.5


slope standard deviation, in units of *C/Om-1
by-intercept standard deviation, in units of C
Correlation coefficient for the linear least squares fit


0.33
0.33
0.53


0.99991
0.99987
0.99986










greater mass of Al per unit length on the wide-bore column, which can be

attributed to the greater surface area and/or thicker Al cladding on the

wide-bore column. Note that the y-intercepts obtained were practically

the same regardless of the column diameter. This can be better understood

by a rearrangement of eq 3.5



T RT/aRo 1/a (3.14)



which demonstrates that the y-intercept (1/a) is dependent only on the

temperature coefficient of resistivity (a), a fundamental property of the

metal. Due to the limited availability of these columns, it has not yet

been possible to compare calibrations from different columns of the same

diameter. If the column-to-column Al coating thicknesses were consistent,

then calibrations would not have to be repeated for each new column.

Rather, the calibrations would only need to be done for each new diameter

column that was used.

One advantage of this method of direct column temperature sensing,

since thermocouples or other temperature sensors are not needed, is the

simplicity of the electronic circuitry required. This could be an

important advantage in the design of portable instrumentation for

environmental or process monitoring. Another advantage that is

particularly important in this application is that measurement of the

column resistance is effectively a length-average temperature measurement.

The temperature measured at each point along the column may vary,

depending on the insulation and thickness of the Al coating at a

particular spot; thus, measurement of the column temperature at one of

these spots may not be an accurate representation of the average column








81

temperature. In addition, we have observed that temperature sensing by

conventional means (e.g., with a thermocouple placed in contact with the

column) causes the temperature of the column to drop near the point of

measurement and leads to inaccurate determinations. This direct

measurement of column temperature might be attractive for other

applications as well. For example, it is normally assumed that the

temperature of a column inside an oven directly follows the oven

temperature; this method could be used to directly measure the actual

temperature of an Al-clad GC column inside a chromatograph oven.

Temperature response characteristics. As mentioned previously, direct

resistive heating of Al-clad columns allows for extremely rapid cycling

of the column temperature due to their extremely low thermal mass. Figure

3.9 shows the temperature response of a 2.3 m x 0.53 mm i.d. Al-clad FSOT

column installed in the GC probe, after the voltage output of the

transistor of the power supply had been stepped from 0.5 V to 8.2 V and

then returned to 0.5 V after a period of 25 s. Note that the voltage

across the column (Figure 3.9 (a)) did not remain constant during the

heating period, but increased from 6.8 V to 7.5 V. Similarly, the column

voltage did not instantaneously return to its initial value of 0.45 V at

the end of the heating cycle. The slow response of the column voltage can

be attributed to an increase in the column resistance as the temperature

is increased. The output voltage of the transistor and the resistance of

the 0.261 0 resistor remain constant during the heating period; thus, as

the column resistance increases, a greater fraction of the transistor

output voltage is dropped across the column. This causes the column

voltage to slowly increase during heating and slowly decrease during

cooling.















































































uuunlo3 ssoaoV
a6olOA pailddy


*on


>> I
U') C.)
0 ta


(o|9Ais!saJ painsoau)
(o) aJnioajaduwaj ulUnlo3


0 0
44

1

1o

E-4
r ,





o*
0








o

4,








4 a4
o













O-44




c 0
41

41




"'4

























a 0




,4 0
o -



W -4
1.4












co
41 B

u -

0o


4)





















4L
Q- 0


r (d
1- Q

s ^ nv
2^1



C) b0








83

During the 25 s heating cycle, the column temperature increased from

43*C to 150*C. During the first second of the 25 s heating period, the

column temperature increased from 43*C to 83C, which represents an

initial heating rate of 40*C/s or 2400'C/min. The heating and cooling

response of the column exhibits an exponential behavior, much like that

caused by impedance in an electrical circuit. However, the response times

for heating and cooling are not the same. In this example, the heating

response time of the column is defined as the time required to raise the

column temperature 90% of the way from 43C to 150*C (i.e., to 139*C).

As shown in Figure 3.9, this response time was found to be approximately

11 s. The temperature increase from 43*C to 139*C in 11 s represents an

average rate of increase of 8.7*C/s or 524*C/min. For comparison, the

ballistic heating rate of a typical gas chromatograph oven from 50C to

150*C is only about 50*C/min [44]. In addition, the maximum heating rate

for a typical on-column injector is only 150*C/min [44]. Obviously, the

method of direct resistive heating offers considerable improvement in

terms of rapidity of response when compared to oven heating.

In applications where extremely rapid temperature ramps are used (as

in the example shown above), the speed of repetitive analyses will most

likely be limited by the time required to re-equilibrate (cool) the column

temperature between injections. During the first second of cooling, the

column temperature dropped from 152*C to 112C, which is an initial

cooling rate of 40*C/s or 2400*C/min. Note that the initial heating and

cooling rates are the same. The response time for cooling the column,

defined as the time required to cool the column 90% of the way back down

to 43C from 150C (i.e., to 54*C), was determined to be approximately 35

s. This represents an average rate of cooling of 2.7*C/s or 165*C/min.










For comparison, the temperature cool-down time for a typical gas

chromatograph oven over the same temperature range would be approximately

4 minutes (about 7 times longer) [44].

As shown in Figure 3.9, there is some thermal lag associated with the

heating and cooling these of columns. A small amount of thermal lag can

be attributed to the voltage regulation of the power supply. The power

supply regulates the voltage output of the transistor; thus, the voltage

across the column changes as the temperature (and hence the resistance)

of the column changes. Much of this thermal lag can be attributed to the

slow thermal response of the insulation that was used to electrically

isolate the column; thus, it is expected that these columns would exhibit

even faster thermal response if the glass braid insulation were

eliminated. This would be possible if the column could be wrapped around

a grooved spool, which would separate each strand of the column and

prevent electrical shorting. Nevertheless, even without any further

improvements, the temperature response characteristics are impressive when

compared to conventional GC ovens.

Another advantage of direct column heating is that very little

electrical power (or energy) is needed. For example, approximately 35 W

of electrical power applied for 25 s (0.875 kJ of energy) was required to

resistively heat the 2.3 m column to 150C. For comparison, a typical GC

oven utilizes a 2400 W heater (240 V AC, 10 A) [44]. Assuming an average

ballistic heating rate of 50C/min, 3 minutes would be required to heat

the oven from 50C to 150*C. Approximately 432 kJ of energy would be

consumed, which is nearly 500 times more energy than that required by

direct resistive heating. In addition, a typical injector is heated

separately from the oven and requires an additional 100-300 W of power








85

[44]. When the entire column is heated resistively and on-column

injection is used, as is the case with the GC probe, no additional

injector heaters are required. The power required to resistively heat a

column to a particular temperature is dependent on the thermal mass of the

column (eq 3.3). For columns of the same diameter and Al coating

thickness, the power needed to reach a particular temperature is directly

proportional to the column length. For example, a 23 m column would

require ten times more power than the 2.3 m column, or 350 W, to reach

1500C.

One limitation of the present system is that the voltages sensed at

the lower column temperatures are too small to yield a high degree of

precision for the column resistance (and hence temperature) measurement.

This is the reason for the quantization noise (at low voltages and

temperatures) in the temperature profile of Figure 3.9. This limitation

could be remedied if an autoranging circuit was used to automatically

adjust the sensitivity of the resistance-sensing electronics.


GC Probe Performance


Figures 3.10 (a) and 3.10 (b) illustrate two different reconstructed

ion chromatograms that were obtained with the GC probe and two different

temperature ramps using a 3.19 m x 0.53 mm i.d. Al-clad FSOT column. A

0.5 pL on-column injection was made in both cases. The concentration of

each component (identified in the figure) was 250 ng/pL in hexane. In

Figure 3.10 (a), the voltage across the column was programmed, after a

delay of 15 s, from 0 V (25*C) to 8.6 V in 2 minutes. The voltage was

held constant for another 1.5 minutes, resulting in a final temperature

of 143*C. In Figure 3.10 (b), the same initial voltage (temperature) was



































Figure 3.10. Reconstructed ion chromatograms obtained with the GC probe
using a 3.19 m x 0.53 mm i.d. column and two different linear column
voltage programs: (a) final voltage of 8.6 V after 2 minutes, (b) final
voltage of 10.8 V after 1 minute. The components of the mixture are: 1,
2-chlorophenol; 2, 2,4-dimethylaniline; 3, n-undecane; 4, 1-undecanol; 5,
acenaphthylene; 6, tetradecane; 7, pentadecane.























1.0 1.5
Retention


2.0
Time


I


1 1 1 .1 ...... 11 I I III
2.5 3.0
(min)


0.8 1.2
Retention Time (min)


8.6 V
143*C
applied voltage n dce
across column indicated column
S-- temperature
3 5
0 v^ RRIC

25C 4


\ 1


RU,


Q)
C
(a)
a)
[15


0.0


i)
Cn
(b) |
CO
Q)
cj


0.5


3.5


0.0


j


`---'








88

used as in Figure 3.10 (a). The delay was 10 s and the final voltage was

10.8 V after 1 minute. The voltage was held constant at 10.8 V for an

additional 30 s, resulting in a final temperature of 168C.

On average, the peak widths at 50% height are approximately 3 s in

Figure 3.10 (a) and 2 s in Figure 3.10 (b). The earliest eluting peak (2-

chlorophenol) has a larger peak width in both cases. This can be

attributed to the deleterious effects of the solvent. During cold on-

column injection, components that are much less volatile than the solvent

are focused at the head of the column and exhibit very narrow band widths.

In contrast, the volatile, early eluting components (if not well separated

from the solvent front) can partition in the liquid solvent as well,

resulting in increased solute peak widths [45]. This problem can often

be avoided if smaller volumes of solvent are injected, or more reliably,

if cryofocusing is used to trap the solutes in a narrow band at the head

of the column [46]. The peak widths obtained here are extremely narrow

considering that the column that was used had a stationary phase film that

is normally considered thick (1.0 pm). In addition to increasing the

chromatographic resolution, the capability of obtaining narrow band widths

increases the sensitivity and lowers the limit of detection [3,14].

In another study, the reproducibility of retention times and peak

areas obtainable with the GC probe were measured. The retention times and

resulting integrated peak areas were measured for four on-column,

temperature-programmed injections of 150 ng tetradecane on a 3.3 m x 0.53

mm i.d. column. Tetradecane, which has a boiling point of 232*C, eluted

at 125*C with a retention time of 90.00 0.81 s (relative standard

deviation (RSD) of 0.9%). The integrated peak area of the m/z 198

molecular ion of tetradecane was determined with a precision of 3.5% RSD,








89

which is certainly within the limits of error for injections performed

with a standard syringe.


Conclusions


Several important concepts for low-power and/or high-throughput

analysis in GC/MS have been addressed. It has been shown that exponential

flow rate programming can be used to approximate linear temperature

programming. When chemical ionization is used, flow rates up to 30 mL/min

can be used without detrimental effects on instrument sensitivity.

Although the range of boiling points separated with flow programming is

less than that which can be obtained with temperature programming, flow

programming does represent a significant improvement over isothermal

operation. The inherent lower elution temperatures of flow rate

programming might also be attractive for the analysis of thermally labile

solutes.

The direct resistive heating of Al-clad FSOT columns has also been

demonstrated. This direct method of column heating obviates the need for

a large chromatograph oven, and should be widely applicable in portable,

low-power gas chromatographic instrumentation. This was demonstrated with

the design of a compact GC probe for high-throughput analyses in GC/MS and

GC/MS/MS. Measurement of the column resistance was shown to be a simple

yet accurate method of directly sensing the temperature of low thermal

mass Al-clad columns. It was demonstrated that linear voltage programming

of the Al-clad columns can be used to approximate linear temperature

programming. Future application of proportional-integral-differential

(PID) feedback control of the column temperature and elimination of the

column insulation is expected to further improve the thermal response

characteristics of direct resistive heating of these columns.
















CHAPTER 4
GAS CHROMATOGRAPHIC SAMPLE INTRODUCTION INTO
THE COLLISION CELL OF A TRIPLE QUADRUPOLE MASS
SPECTROMETER FOR MASS-SELECTION OF REACTANT
IONS FOR CHARGE EXCHANGE AND CHEMICAL IONIZATION


Introduction and Theory


The maximum selectivity of any analytical technique can be realized

if all of the variable parameters of the method are considered. These

variable parameters can be considered as resolution elements that affect

the informing power (i.e., the amount of information available) in the

analytical method [47]. In mass spectrometry, two resolution elements

which may be varied are mass analysis and ionization. The informing power

of the mass analysis can be augmented by increasing the mass range,

increasing the mass resolution, or increasing the number of stages of mass

analysis (i.e., tandem mass spectrometry). Since the ionization method

is a resolution element, it too will affect the informing power. A

certain degree of "resolution" (or selectivity) can be obtained with

ionization by employing techniques that allow ionization of only the

desired components of a mixture, thereby excluding the ionization (and

hence detection) of undesirable interferents. This selectivity of

ionization is not frequently exploited, but can be controlled if careful

consideration is made for the selection of reactant ions for charge

exchange (CE) ionization or chemical ionization (CI) [48]. Traditionally,

ion-molecule reactions for CE or CI are performed in a high-pressure

(e.g., 1 torr) ion source [48-50]. However, there are limitations to








91

performing these reactions in the ion source. When a reactant gas is

introduced into the ion source, it is rare that only a single m/z reactant

ion is formed; due to the presence of other undesirable reactant ions,

the ionization process may not be well-controlled. In fact, many

ionization processes including CE, CI, (each with various reactant ions)

and even El may compete with the ionization technique of interest. This

mixed-mode ionization certainly limits the selectivity of the ionization

process. In addition, instrumental parameters such as ion source pressure

affect the relative abundances of the various reactant ions; hence,

reproducible CE or CI spectra are often not easily obtained.

A novel approach that eliminates the mixed-mode ionization discussed

above is to mass-select the desired reactant ion before allowing it to

react with the sample. Mass-selection of reactant ions has previously

been used for studying ion-molecule reactions and reactive collisions.

A variety of tandem mass spectrometers have been used, including sector

instruments [51], a double quadrupole instrument [52], quadrupole ion

traps [53-55] and triple quadrupole mass spectrometers [56-62]. Crawford

and co-workers have used Kr as a charge exchange collision gas in a triple

quadrupole mass spectrometer (TQMS) for monitoring carbon monoxide in the

presence of hydrocarbons [63]. However, to date there has been no

demonstration of the analytical utility of mass-selecting the reactant ion

for mixture analysis, or of the combination of mass-selected reactions

(MSRs) with GC. For this reason, a gas chromatograph has been interfaced

to the collision cell of a TQMS. In gas chromatography/mass-selected

reaction/mass spectrometry (GC/MSR/MS), reactant ions are selected with

the first quadrupole mass filter (Ql) and are allowed to react with

neutral sample molecules which elute from a GC column into the second








92

quadrupole collision cell (Q2). The third quadrupole is scanned for the

products of the ion-molecule reactions. In this chapter, the selectivity

and sensitivity of a TQMS system combining short-column GC with mass-

selected reactions are discussed. Also, the capability of obtaining both

molecular weight and structural information in the same chromatogram is

demonstrated. A highly useful method of rapidly determining ionization

energies with the TQMS is also demonstrated.

The selectivity of charge exchange (CE) and chemical ionization (CI)

techniques arises from control of the energetic of the ion-molecule

reactions between reactant ions and sample molecules. A CE reaction may

occur if the ionization energy of the compound of interest, IE[M], is less

than the recombination energy of the reactant ion, RE[R'].



R+' + M M+' + R (4.1)

AH IE[M] RE[R+*] (4.2)



For monoatomic species, the recombination energy of an ion is the same as

the ionization energy of the neutral. However, polyatomic ions may

possess excess vibrational/rotational energy; therefore, this equality

cannot always be assumed. The amount of internal energy deposited into

M" is determined by the exothermicity of the CE reaction.

Similarly, proton transfer CI reactions may occur if the proton

affinity of the sample molecule, PA[M], is greater than the proton

affinity of the conjugate base, R, of the reactant ion [R+H+], PA[R].



[R+H]* + M [M+H]* + R (4.3)

AH PA[R] PA[M] (4.4)