1%..
r

-

...

.¢.:...fi.I.¢s

rfiuw~nv~wwflh
3th.

tflnf

1 ,

1. .
y 1r;
a 6 1.!

If
.1

 

(n-

£33.15 .5343. Evan"... l: « ... a.

 

'E'tTC‘I ,
.46” +9 ”

 

 

 

LIBRARY
Michigan State Th. . t mm uh I
. . ISIS 066 I a e
U n Ive ISIty thesis entitled

 

REAL TIME MONITORING OF FRUIT SPIRIT DISTILLATION
WITH HIGH SPEED GAS CHROMATOGRAPHY

presented by

Deirdre Jane Lindemann

has been accepted towards fulfillment
of the requirements for the

Master of degree in Chemistry/Chemical
Science Engineering

 

 

%W%M

Major Professor’ 5 Sighat’ure

2.1. Ava 0511' 1.003

 

Date

MSU is an Affirmative Action/Equal Opportunity Institution

PLACE IN RETURN BOX to remove this checkout from your record.
T O AVOID FINB return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/CIRC/DateDue.p65-p.15

REAL TIME MONITORING OF FRUIT SPIRIT DISTILLATION WITH
HIGH SPEED GAS CHROMATOGRAPHY

By

Deirdre Jane Lindemann

A THESIS

Submitted to
Michigan State University
in partial fiilfillment of the requirements
for the degree of

MASTER OF SCIENCE
Chemistry and Chemical Engineering

2003

Abstract

REAL TIME MONITORING OF FRUIT SPIRIT DISTILLATION WITH
HIGH SPEED GAS CHROMATOGRAPHY

By

Deirdre Jane Lindemann

Fruit brandy distillation is typically performed using a batch still with a rectifying
column. The process of producing high quality spirits relies heavily on the distiller’s
sensory abilities to determine the appropriate moment to make cuts and apply
adjustments to the still. However, it would be preferred to have an instrumental method
of online analysis that could accomplish the same goal, while avoiding human error.
Recent advances in high-speed gas chromatography (HSGC) have made online analysis
of distillation a possibility. Two methods of high-speed gas chromatography were
examined for this purpose. The first method used the Chromatofast GC AcceleratorTM,
which reduces injection bandwidth and allows for good separation using short normal
bore columns and fast flow rates. The second method consisted of a headspace sampling
system, which relied on the selectivity of the column and proper temperature
programming for sufficient and fast separationi Distillations of “synthetic brandy-type”
solutions were performed in order to examine both methods for their ruggedness and
ability to analyze samples in a reasonable amount of time. The repeatability and
quantification were examined for each method. Finally, results on monitoring compound

behavior throughout a distillation are presented.

Table of Contents

List of Tables ................................................................................................. v.
List of Figures ................................................................................................ vi.
Introduction .............................................................................................. 1
1.1 Michigan Fruit Brandy Industry .................................................. 1
1.2 Brandy Production Process .......................................................... 3
1.3 Distillation ................................................................................... 6
1.4 Congener Compounds in Fruit Spirits ......................................... 8
1.5 Structure of a Batch Still .............................................................. 10
1.6 Quality Control ............................................................................ 13
1.7 Analysis of Fruit Spirits ............................................................... 13
1.8 Objective ...................................................................................... 14
1.9 Literature Cited ............................................................................ 15
Hi gh-Speed Gas Chromatography Background ....................................... 16
2.1 Development of High-Speed Gas Chromatography .................... 16
2.2 The Compromise Between Speed and Separation ....................... 18
2.3 Columns ....................................................................................... 19
2.4 Injector Considerations ................................................................ 21
2.5 Temperature Programming .......................................................... 22
2.6 Pressure Considerations ............................................................... 23
2.7 Carrier Gas Flow Rate ................................................................. 24
2.8 Detectors ...................................................................................... 24
2.9 Literature Cited ............................................................................ 25
Discussion of Fast GC Setups ........................................................................ 27
3.1 The Chromatofast GC Accelerator .............................................. 27
3.2 The Headspace Sampling Setup ................................................... 32
3.3 Literature Cited ............................................................................ 38
Materials and Methods ................................................................................... 39
4.1 Spirits Production Process ........................................................... 39
4.1.1 Creation of Synthetic Solutions ................................. 39
4.1.2 Distillation ................................................................. 39
4.2 Measurement of Condensation Time ........................................... 42
4.3 Analysis ........................................................................................ 43
4.4 Simple Distillation ....................................................................... 47
4.5 Literature Cited ............................................................................ 51

iii

Results: GC Accelerator ................................................................................ 52

5.1 Method Optimization ................................................................... 52
5.2 Standardization ............................................................................ 56
5.3 Distillation Behavior .................................................................... 58
5.4 Structural Integrity of GC Accelerator ........................................ 65
Results: Headspace Sampler .......................................................................... 68
6.1 Method Optimization ................................................................... 68
6.2 Standardization ............................................................................ 73
6.3 Distillation Behavior .................................................................... 75
6.4 Simple Distillation ....................................................................... 84
Summary and Conclusions ............................................................................ 92
Future Work ................................................................................................... 96

iv

Table 4.1.

Table 4.2.

Table 4.3.

Table 4.4.

Table 4.5.

Table 5.1.

Table 6.1.

List of Tables

Congener types and volume percentages used in Experiment 1 ........ 4O
Congener types and volume percentages used in Experiment 2. ...... 40
GC analysis parameters for the AcceleratorTM setup ......................... 46
GC analysis parameters for the headspace setup. .............................. 46

Congener types and volume percentages for simple distillation. ...... 50

A list of the retention times for chromatograms taken using
Fast GC with the GC AcceleratorTM and also conventional

GC. The boiling points for the congeners is also listed. ................... 55

Retention times for chromatograms taken using Fast GC
with the headspace sampler and also conventional GC.
Peaks were identified by comparing the retention times to

standard retention time for each of the compounds. .......................... 72

Figure 1.1.

Figure 1.2.

Figure 1.3.

Figure 1.4.

Figure 3.1.

Figure 3.2.

List of Figures

The fruit spirits production process". ................................................. 4
McCabe Thiele diagram for batch distillation with reflux

showing the effect of increased stages on improved

separationg. ......................................................................................... 7
Milligrams of volatile congeners in distilled spirits per 100

milliliters of alcohol. Fruit spirits have higher congener
concentrations than other distilled beverages“. ................................ 9
A 150 L Christian Carl Batch Stillé. .................................................. 12
Schematic of thr GC AcceleratorTM external components.

This air sampler makes it convenient for interfacing with a

still. There is a keypad and LCD for controlling

parametersl ......................................................................................... 28
Schematic of the GC Accelerator connected to a typical gas
chromatograph. The accelerator itself consists of three

restrictor columns R1, R2 and R3, two solenoid valves V1

and V2, a vacuum pump, cold trap, and an inlet for carrier

gas. Sample is pulled through R2 and R3 and into the cold

trap containing liquid nitrogen where cryfocusing of the

sample occurs. Carrier gas travels through V2 and sweeps

residual sample from R3 onto the trap as well as sweeps

through R2 to eliminate sample carryover. Finally, the

vi

Figure 3.3.

Figure 3.4.

Figure 3.5.

flow reverses in the cold trap, and the tube is flash heated.

From there, the sample is injected into the GC for

analysis” ........................................................................................... 30
Schematic of the online setup with the Arnold Holstein still

to the GC Accelerator and gas chromatograph. Sample

from the still is extracted through the Silco-Steel capillary

into the Accelerator where it is cryogenically frozen , then

flash heated and injected into the GC. ............................................... 31
A schematic of carrier gas flow through the entire system.

Helium from an air cylinder enters through the GC and

flows to the purge and trap adapter. The adapter splits the

flow and directs carrier gas to both the GC injector and the

injection valve. Helium flowing through the injector

travels through the column and the detector to ensure flow

through the GC. Carrier flowing through the injection

valve travels through the heated transfer line, where it

enters through the injector and eventually into the column

and detector. This ensures that sample is transferred to the

GC ...................................................................................................... 33
A close-up view of the sample loop inside the sampling

valve. In the load position, the vapor sample from the still

enters through the sample loop. In the inject position, the

vii

Figure 3.6.

Figure 4.1.

Figure 4.2.

Figure 4.3.

Figure 5.1.

carrier gas enters the sample loop and tranfers the sample to

the transfer line and eventually to the column4 .................................. 35
Schematic of the headspace sampling system connected to

the Shimadzu GC-17A. ...................................................................... 37
Schematic of a 10L Amold Holstein Still. ......................................... 41
Travel time for methanol condensation from the sampling

port above the condenser to the distillate collection port at

the bottom. It takes over three minutes before methanol

distillate begins to appear ................................................................... 44
A typical simple distillation apparatus. The round bottom

distillation flask used is 300mL and contains more than one

opening. An extra opening was needed to insert the sample

line of the headspace sampler in an area above the

distillation in the flask. The distillate collection flask is

somL‘ ................................................................................................. 49
A chromatogram of a brandy type solution taken with the

GC AcceleratorTM Using a 70°C run. Peaks were identified

by comparing the retention times from the chromatogram to

a standard set of retention times for each compound.

Isoamyl alcohol does not appear due to a much later elution

time .................................................................................................... 53

viii

Figure 5.2. Chromatogram of a sample from a brandy-type solution

using a conventional GC. The GC program from Table

was used. The sample was analyzed after distillation ....................... 54
Figure 5.3. Chromatogram of a sample from a brandy-type solution

using a conventional GC. Distillate cuts were taken and

analyzed after distillation. Peaks were identified by

comparing the retention times from the chromatogram to a

standard set of retention times for each compound. .......................... 5 7
Figure 5.4. Ethanol concentration measured after the distillation.

Samples of the distillate cuts were taken and analyzed

using conventional GC. Concentrations were found by

comparing peak areas to a standard curve for ethanol ....................... 60
Figure 5.5. Acetaldehyde concentration in the product as a function of

distillation time according to measurements made with the

GC Accelerator“. Concentrations are given as percent

volume of congener in a 40% ethanol/water mixture. ....................... 61
Figure 5.6. Ethyl acetate concentration in the product as a function of

distillation time according to measurements made with the

GC Accelerator”. Concentrations are given as percent

volume of congener in a 40% ethanol/water mixture. ....................... 62

ix

Figure 5.7.

Figure 5.8.

Figure 5.9.

Figure 6.1

Methanol concentration in the product as a function of

distillation time according to measurements made with the

GC AcceleratorTM. Concentrations are given as percent

volume of congener in a 40% ethanol/water mixture. ....................... 63
N-propanol concentration in the product as a function of

distillation time according to measurements made with the

GC Accelerator“. Concentrations are given as percent

volume of congener in a 40% ethanol/water mixture. ....................... 64
Chromatogram of a sample from a brandy-type solution

using the GC AcceleratorTM. After repeated measurements,

the chromatograms would develop poorly shaped peaks,

such as in this example. The shoulder on the ethanol peak

is most likely a result of the injection volume not being

“plug-flow” ........................................................................................ 66
Chromatogram for the distillation of a brandy-type mixture.

The sample was collected from the still using the headspace

sampler. The program used was the same as in Table 4.4.

Peaks were identified by comparing the retention times

from the analysis to standard retention times for each of the

compounds. ........................................................................................ 69

Figure 6.2

Figure 6.3

Figure 6.4.

Figure 6.5.

Chromatogram for a sample from a brandy-type solution

using a conventional GC. Distillate cuts were taken and

analyzed. Peaks were identified by comparing the retention

times from the chromatogram to standard retention times

for each of the compounds ................................................................. 70
A comparison of the chromatograms for the two methods.

The solid curve represents the headspace sampling analysis

and the dashed curve represents analysis with the GC

Accelerator“. The total elution time with the headspace

sampling system is 1.5 minutes whereas the total elution

time for the Accelerator“ is 2.5 minutes. Peaks were

identified by comparing the retention times from the

chromatogram to standard retention times for each of the

compounds. ........................................................................................ 71
The response curve for ethanol using the headspace

sampler. The peak areas do not increase with increasing
concentration. ................................................................................... 74
Ethanol concentration as a function of distillation time

taken afier distillation using a standard ethanol curve with

conventional GC. ............................................................................... 76

xi

Figure 6.6

Figure 6.7

Figure 6.8

Figure 6.9

Figure 6.10

The concentration behavior of acetaldehyde. The online
measurement is represented by the diamond plot and

corresponds to the left y-axis. The square plot represents

the behavior from post distillation measurements and

corresponds to the right y-axis. ........................................................ 77
The concentration behavior of ethyl acetate. The online
measurement is represented by the diamond plot and

corresponds to the left y-axis. The square plot represents

the behavior from post distillation measurements and

corresponds to the right y-axis. . ....................................................... 78
The concentration behavior of methanol. The online

measurement is represented by the diamond plot and

corresponds to the left y-axis. The square plot represents

the behavior from post distillation measurements and

corresponds to the right y-axis ........................................................... 80
The concentration behavior of n-propanol. The online

measurement is represented by the diamond plot and

corresponds to the left y-axis. The square plot represents

the behavior from post distillation measurements and

corresponds to the right y-axis. . ....................................................... 81
The concentration behavior of isoamyl alcohol. The online
measurement is represented by the diamond plot and

corresponds to the left y-axis. The square plot represents

xii

Figure 6.11

Figure 6.12

Figure 6.13

Figure 6.14

the behavior from post distillation measurements and

corresponds to the right y-axis ........................................................... 82
Ethanol concentration as a function of distillation time for

the simple distillations determined by conventional GC.

Ethanol concentration is given in percent volume of ethanol

per volume of distillate ...................................................................... 85
The concentration behavior of acetaldehyde during a

simple distillation. The online measurement of behavior is
represented by the diamond plot and corresponds to the left

y-axis. The square plot represents the behavior from post

distillation measurements and corresponds to the right y-

axis ..................................................................................................... 86
The concentration behavior of ethyl acetate during a simple
distillation. The online measurement is represented by the

diamond plot and corresponds to the left y—axis. The

square plot represents the behavior from post distillation
measurements and corresponds to the right y-axis. ........................... 87
The concentratrion behavior of methanol during a simple

distillation. The online measurement is represented by the

diamond plot and corresponds to the left y-axis. The

square plot represents the behavior from post distillation

measurements and corresponds to the right y-axis. ........................... 88

xiii

Figure 6.15

Figure 6.16.

The concentration behavior of n-propanol during a simple

distillation. The online measurement is represented by the

diamond plot and corresponds to the left y-axis. The

square plot represents the behavior from post distillation
measurements and corresponds to the right y-axis. ........................... 89
The concentration behavior of isoamyl alcohol during a

simple distillation. The online measurement is represented

by the diamond plot and corresponds to the left y-axis. The

square plot represents the behavior from post distillation

measurements and corresponds to the right y-axis ............................ 90

xiv

1. Introduction

1.1 Michigan Fruit Brandy Industry

In the past decade, there has been a growing interest in the fruit brandy industry in
the state of Michigan. Wineries in the state are interested in distillation for the
fortification of wines as well as to produce spirits. In addition, farmers are interested in
new ways to utilize surplus fruit. Major changes in state legislation made in 1996 have
made production and distribution of distilled spirits more economically feasible. This has
had a dramatic effect on the fruit brandy industry as the number of distilleries in
Michigan has increased from zero in 1996 to seven in 2000.

The United States faces many challenges in producing high quality fruit spirits.
First of all, fruit brandies contain hundreds to thousands of compounds, which greatly
affect the taste and aroma characteristics of the distilled product'. Second of all, some of
the compounds present in fruit brandy are regulated due to their toxic effects at high
levels of exposurel‘z‘3 . Therefore, much research needs to be done on several aspects of
fruit brandy production in order to ensure the quality of spirits in Michigan and the rest of
the country.

Brandy is the product obtained from the distillation of wine or fermented fruit
mash. This type of beverage consists mainly of ethanol and water, but it also contains an
array of various compounds that create a characteristic flavor and aroma for individual
products]. There are many different methods of production, often depending on the
location where it is made. For instance, there is the French style or alambic style of
brandy making, which involves the use of a batch pot still without rectification. This

means that there is little or no reflux for the enrichment of the vapor in the still. This type

requires multiple distillations to obtain high proof spirits and results in the “Armagnac”
or “Cognac” style brandies. It is often referred to as wine brandy and is stored in oak
barrels to obtain a characteristic color and essence from the woodl. The German style of
distillation involves the use of a batch still with rectification (the use of trays and partial
condensers). The still with rectification is designed to capture the essences of the fruit.
This type of distillation requires only a single pass through the still to obtain high proof
spirits and results in the “Eau-de-vie” or “Schnapps” type brandies. This type of brandy
is stored in glass and is served water clear}.

The German process of distillation is the only method examined in this thesis.
This process of fruit spirits distillation uses the entire fruit. As a result, there is a
relatively high production of congeners components (components other than ethanol and
water) as compared to other classes of distilled spirits such as whiskeys or vodkas4.
These compounds are further concentrated during the distillation process. These
congeners are key in creating the characteristic flavors and aromas of fruit spirits. Some
of these compounds can have adverse health effects if present in extremely high
concentrationsz. Others, if present in too high a concentration can produce an undesirable
taste or aroma]. Therefore, controlling the amounts of these congeners is of the utmost
importance in producing a quality product.

The concentrations of congeners can be controlled during distillation through
fractionation of the distillate through the use of trays and partial condensers]. It is
important to know when it is appropriate to perform these various actions during a
distillation in order to control the quality. This requires the sensory analysis of the

distiller, in other words his or her sense of taste and smell]. This can be difficult as the

senses can become overloaded after prolonged periods of exposure. Therefore an
instrumental technique may be useful in monitoring the process of distillation. Gas
chromatography has typically been used for analysis of fruit spirits after distillation,
however recent advances in high-speed gas chromatography have made it feasible for this
method to be used as an online technique”. Using this method for quality control could

make it easier for distillers to produce higher quality spirits.

1.2 Brandy Production Process

The process of distilling fruit spirits ( Figure 1.16) consists of four basic steps;
processing, fermentation, distillation, and storage. The basic starting material is some
kind of fruit including seed fruit (apples, pears), stone fruit (cherries, peaches), and berry
fruit among others. The value of these fruits is determined by high sugar content and
strong aroma. Although fruit that is less desirable may be used, it should not be moldy or
rotten as that will affect the quality of the distillate at the end].

The first step in the production of fruit brandy is the processing or the mashing of
the fruit. To begin, the fruit should be relatively clean and free from soil‘. Stems and
leaves should be removed whenever possible. Crushing of the fruit can be performed
many different ways. The fruit can be crushed under feet, through the use of a rolling
mill, or by hand with a mortar and pestle depending on the type of fruit and equipment
available. The goal is to process the fruit as completely as possible. The more broken

down the fruit is, the more efficiently and completely the fermentation will proceed].

gag" —>

Ripe Fruit

Mashed Fruit

 

“I

Il1IIII

 

W IIIIII

[II'IIII
"‘1“,th ‘.IIII

”IIIII MIIII. -

III IIIIIIII
I
IIIIIIIIIII IIIIIIIIIII

OWIIII

O I

(IIIIIII

 

O

 

 

 

Distillation

Consumption

Figure 1.1.

 

 

       

 
   

 

I I I‘ II II
VIIIIIIII ‘v‘yIlIII .MIIIIIII I IIIII
5,111 [WI “)1 .II‘IIIII .
I. ‘ II
" IIIII IIIIIIIIII III

IIIII‘II II . Il
mIIII

 
 

 
 

IIIII
II

II“
. II III III ,IIIIII‘IIII

Fermentation
1-2 Weeks

 

Storage
Months to
Years

The fruit spirits production process.6

After processing, the fruit is transferred to a fermenter where yeast will be added
in order to initiate fermentation. The primary and desired reaction in a fermentation is
the conversion of sugar to form ethanol. This is shown below.

CGHIZOG -) 2C2H50H + C02
In addition there are several side reactions that occur to form the congener compounds in
the mash. F errnentations must be kept at a temperature range between 15-20°C to
achieve an optimal reaction. The fermentation must be warm enough for the yeast to
react, but it has to be cool enough in order to avoid the growth of undesired
microorganisms. The pH must also be maintained in a range of 2.8 to 5.2, as this is the
range where yeast survives. This slightly acidic pH helps to avoid the growth of
unwanted microorganisms as wellI. It is also useful to stir the fermentation in order to
aid in heat transfer as well as to help vent CO; from the fermenter. Measurement of the
brix of the fermentation mash turns out to be an effective way of monitoring the progress
of the fermentation. (Brix is the percent of dissolved solids in the mash, which is
assumed to be mostly sugars.) The loss in sugar content corresponds to the production of
ethanol. Sugar content can be monitored by the use of a simple hand refractometer or
HPLC with a refractive index detector. Depending on the amount of stirring and the
conditions of the fermenter, fermentation generally lasts between ten days and two
weeksI.

When an acceptable time has passed for fermentation as determined by the
conversion of sugars, distillation can then be performed. The still is filled to about 65-
75% of volume with mash. Sometimes anti-foaming agents are added. This mash is

heated and the alcohol and other volatile compounds vaporize. The idea behind a

distillation is to obtain the greatest amount of ethanol with optimal concentration of
congeners to produce a product that is desirable in both taste and aroma. To accomplish
this, the distillation is typically broken up into three fractional cuts; heads, hearts, and
tails]. The heads portion contains relatively higher concentrations of lower boiling point
compounds and is generally discarded. The hearts fraction contains the optimal amount
of ethanol and congeners and is therefore saved as the potable product. The tails contains
relatively higher concentrations of higher boiling point compounds and is also discarded
due to an undesirable aroma and taste].

After collection of the distillate, the hearts fraction is stored at high proof for
several months to several years. The product is then diluted down to 40% and then is

bottled and made ready for consumption.

1.3 Distillation

Distillation may be carried out using one of two processes. First, there is
distillation without rectification where vapors produced by boiling are condensed and do
not return to the still. Second, there is distillation with reflux where part of the
condensate returns to the still and comes into contact with the rising vapors. These points
of contact are known as plates or stages and they help to make the vapor richer in the
higher boiling component7. They are also referred to as trays. It is known that the
greater number of trays or stages used, the more separation can be achieved”. This is
evident using a McCabe-Thiele diagram shown in figure 1.28. This figure shows the
mole fractions of ethanol in the vapor and liquid phase for a water ethanol distillation.

Operations were performed for a two stage and a seven stage separation. The seven stage

 

 

 

 

 

 

 

 

 

 

 

 

 

LO T r I r
.8- j
o
b de
YE 6~ c' x ,
d2
2 /
2 /
l' .,
.4 x0
F "w
R' '
.2 _ p=| otm
x
x wz
O L _l 1 I
O 2 4 .6 8 IO
XE

 

yE = Mole Fraction of Ethanol in Vapor
x5 = Mole Fraction of Ethanol in Water

 

 

 

Figure 1.2 McCabe Thiele diagram for batch distillation with
reflux showing the effect of increased stages on

improved separations.

separation yields a higher mole fraction in the distillate. This is typical of a distillation of
two components. Fruit brandy distillations are more complicated than this as there are

often hundreds of components in a fruit mash to be separated.

1.4 Congener Compounds in Fruit Spirits

Distillation of fruit spirits is more complicated than just a simple ethanol/water
distillation. Although the main components are ethanol and water, there are also
hundreds of congeners present in varying amounts. In fi'uit spirits production, the entire
fruit is used, including the skins, the pits, the pectin and seeds. This results in various
reactions in the fermentation and even distillation and storage that lead to the creation of
many congeners. In fact, fruit spirits contain a greater number of congeners compared to
other types of beverages such as whiskey, vodka, or gin. This can be seen in Figure 1.3,
which illustrates the typical components in various distilled beverages“. Depending on
the amount of these congeners, they can either have a positive or negative effect on the
taste and aroma of the resulting distillate. The distillation behavior of these congeners
depends on factors such as boiling point and equilibrium relationship to ethanol and
water. Conditions of the still and the environment are different each time a distillation is
run and each batch of mash, even if it is from the same fermentation, tends to have slight
differences. This leads to poor reproducibility of spirits from one distillation batch to the
next

There are various carbonyls, acids, and fuse] alcohols present in distilled spirits.
Most of these compounds are formed during the fermentation process. The main carbonyl

compounds present include esters such as ethyl formate and ethyl acetate and ketones,

Congener Amount (mg/100mL)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2000 J IMethanol
III Higher alcohols

1800 i I Carbonyl compounds
1600 I Ethyl acetate
1400 UOther esters
1200 F
wool
800 :1
600 I :l :g
400 I . . l
200 "

9" dbé‘se’er Weeease

009° g9“ 06” <29 49° 0’0“ .ng Qo‘ifi Qoe Q09 $ e“ 0 0°
’0 _<\0 ® 0% @9 09 I
0° s (be a <0 0°
‘60 6° 0° 0° 6
C90 Q‘ (99 60
v
‘0
60°

Figure 1.3. Milligrams of volatile congeners in distilled spirits per

100 milliliters of alcohol. Fruit spirits have higher
congener concentrations than other distilled

beverages“’6

such as acetone, aldehydes such as acetaldehyde and benzaldehydeg. Fusel alcohols,
which are alcohols with more than two carbons comprise the largest group of aroma
compounds in fruit spiritsg. The most important fusel alcohol present is isoamyl alcohol.
Other fusel alcohols include n-propanol, isobutyl alcohol, and amyl alcohol. The fatty
acids present are thought to be products of bacteria as well as a byproduct of the
formation of alcoholg. However they will not be discussed in this thesis.

Methanol is a major issue in the production of distilled spirits. Although
methanol is a desired compound for flavor and aroma, it is regulated for the health risks it
poses. Methanol can cause damage to many human organs, such as liver, kidneys, and
the nervous system. Methanol can also cause temporary or permanent blindness, or even
death3 . Therefore, the United States Environmental Protection Agency has set limits on
the amount of methanol that can be present in drinking water at 3.9 parts per million.

The acceptable exposure limit in air is set at 200 parts per million for an eight hour time

weighted average. Methanol in fruit spirits has been regulated in Europe fi'om a range of
400mg/100mL to IOOOmg/ IOOmL absolute alcohol depending on the variety of fruit spirit
being produced. According to the Bureau of Alcohol, Tobacco, and Firearms, the limit is

0.35%v/v for all fruit spirits, which translates to 700mg/100mL absolute alcoholz.

1.5 Structure of a Batch Still

The construction of the still is very important in controlling the various steps of a
distillation. A 150 L Christian Carl still is shown in Figure 1.36. There are varying
designs of these stills regarding the number and placement of trays, however, this figure

is a good indication of what is typical for brandy stills. Although design opinions vary, it

10

is agreed that the material used to make up the still must be copper. Copper is a very
good heat conductor, shows good resistance to fruit acids, and has catalytic properties
that contribute positively to the quality of the fruit spirits produced].

The basic design includes a pot, a column with trays, and condensers. The pot
contains the mash when the still is filled. In this figure, the pot is heated using a steam
jacket connected to a steam inlet. Some fruit mashes are rather viscous and pose
difficulties with uneven heating, foaming, and baking. Therefore, many stills may
contain a stirrer to ensure not only that the mash is evenly heated, but also to reduce the
occurrence of foaming and baking of mash to the surface of the pot. Trays can be opened
or closed. The particular still in the present work contains four trays, one over the pot,
and three in the column next to the still. A still may also contain a catalytic converter,
which is a vessel containing a high surface area copper packing used as both an extra tray
in distillation and means for removing cyanide containing compounds from the distillate6.
There is a function that allows for the distillate to enter through the bottom of the
catalytic converter, or allows for the catalytic converter to be bypassed.

Stills contain a condensing column that condenses the distillate vapor into liquid
when cooling water is passed through. There can also be partial condensers. The still
contains a partial condenser above the tray over the pot and one above the three trays in
the column adjacent to the pot. It is possible to choose to allow water to pass through
these condensers, or to leave them off. There is a flow regulator to control the flow rate
of cooling water through the condenser and partial condensers. By controlling the rate of
cooling water that passes through the condensers, one can control the rate of the

distillation.

ll

 

 

Steam Jacket - ' -Distillation Vapor Path
—Cooling Water Path

Partial Condensers

   
 
 

Total Condenser

 

Copper Packing

 

Ilfifllfl

Fermented Mash

 

 

 

 

 

 

 

 

  
    
 
     
 
 
 
   

 

S
Tray Control
Levers

Cooling Water
Control

Steam
Inlet

Distillate
Discharge

Removal
Port To Drain

 

 

Figure 1.4. A 150 L Christian Carl Batch Still“.

12

1.6 Quality Control

The control of quality can come from many aspects of the brandy production
process. It was mentioned before that the processing and fermentation steps of the
production process can affect quality. In the processing step, fruit of an adequate quality
must be used and proper maceration must be performed to ensure good quality. In the
fermentation, temperature, and pH must be considered to ensure that the fermentation
proceeds appropriately].

Finally, adjustments may be made to control the quality of the distillate. First,
appropriate fractionation is important. Separating the distillate into the heads, hearts, and
tails is essential. Knowing when to apply other devices, such as the catalytic converter or
partial condenser are important as well. This thesis will focus on the quality control

through manipulation of the distillation.

1.7 Analysis of Fruit Spirits

It is important to know when to make appropriate adjustments during the
distillation. The sensory analysis of the distiller has been used since the inception of the
practice of distillation. In other words, the distiller makes fractions and other adjustments
based on the taste and aroma of the distillate. This practice has been customary, but it
has its disadvantages. After prolonged periods of exposure, the distiller’s sense of taste
and smell begin to overload and become less reliable with each distillation. So it would
be advantageous to have some instrumental method to analyze the quality of distilled

spirits and be able to make adjustments during the distillationl.

l3

Gas chromatography (GC) has been used for years to analyze the concentration of
congeners present in the distillate]. It is a robust and reliable instrument with low limits
of detection capable of detecting trace concentrations”). The drawback to conventional
gas chromatography, however, is analysis time, which often ranges from 40 minutes to an
hour for distilled spirits. GC has only been useful for analysis afier the distillation.

Advances in high speed gas chromatography (HSGC) have now made it possible
to examine distillations using chromatographic methods as an online tool. HSGC can
utilize faster flow rates, shorter columns, and sometimes special injection techniques to
perform analyses that range from a few seconds to a few minutess. With this technique, it
is possible to collect a sample directly from a still and analyze in a fast enough manner

that adjustments could be made to control quality during a distillation.

1.8 Objective

The goal of this study will be to examine methods of high speed gas
chromatography and how they can be applied to online monitoring of a distillation. An
efficient and effective sampling method for extracting sample from a brandy still and
injecting it into a gas chromatograph will be demonstrated. It will also be important to
examine the speed and efficiency of the actual chromatographic separation. Finally, the
method will be applied the method to an actual distillation in order to monitor compounds

as they are volatilized.

l4

10.

Literature Cited
Tanner, H. Brunner, H.R., Fruit Distillation Today 3rd ed., (English Translation)
Heller Chemical Administration Society, Germany, 1982.
Bureau of Alcohol, Tobacco, and Firearms. “Alcohol and Tobacco Newsletter (1)
1998.” Available at <http://www.ttb.gov/pub/alctob_pub/cm98-1.pdf>.
Methanol, Material Safety and Data Sheet.
Piggot, J .R., Patterson, A. ed., Distilled Beverage F lavour:Recent Developments.
Ellis Horwood Ltd., 1989.
Sacks, R., Smith, H., Nowak, M., “High Speed Gas Chromatography.” Anal.
Chem. 69 (1998) 29A.
Claus, M. “An Investigation of the Relationship Between Tray Usage of the Still
and Congener Compound Concentration in Distilled Fruit Spirits”. MS. Thesis.
McCabe, Warren, Smith, Julian, Harriott, Peter. Unit Operations of Chemical
Engineering. McGraw, Hill, 2001.
Wankat, P.C., Equilibrium Staged Separations. Prentice Hall, New Jersey, 1998.
Suomalainen, H. and Lehtonen, M., “The Production of Aroma Compounds by
Yeast.”J. Inst. Brew., 53, (1) 1972.
Skoog, Douglas, Holler, James, Nieman, Timothy, Principles of Instrumental

Analysis. Harcourt, Brace and Co., 1998.

15

2. High Speed Gas Chromatography Background

2.1 Development of High Speed Gas Chromatography

The development of high speed gas chromatography began with the advent of
open tubular capillary columns by Golay in 19571’2. Open tubular columns have greater
permeability and require lower inlet pressures than packed columns. In addition, there is
a less rapid loss in column efficiency with increasing velocity3. Since it was more
practical to operate at higher flow rates, it was recognized that high speed gas
chromatography was possible. The establishment of the principles of high-speed
chromatography came in the 1960’s. Using open columns, in the 1960’s Desty and his
co-workers separated nine heptane isomers in 5 seconds using a gas tight syringe which
was struck sharply by a mallet to inject a narrow 10 ms vapor plug into a gas
chromatograph] ’2.

Despite such breakthroughs, open tubular columns gained slow acceptance due to
competition with packed columns, which had been used for years. Packed columns are
rugged and had high resolving power and high sample capacity. The first open tubular
columns were made of rigid glass, which made them difficult to install and maintain. In
addition, open tubular columns lacked good selectivity compared to packed columns
since there were not as many available stationary phases. The advent of fused-silica
columns along with other technological improvements made open tubular columns more
flexible and viable for everyday usez.

Unfortunately, once open tubular columns gained acceptance, the main goal in

their application was not speed as chromatographers were primarily concerned with

achieving high resolution. In fact, people often achieved much higher resolution than

16

needed by using extremely long capillary columns. The reason for this was that there
was a limited range of stationary phases, which resulted in low selectivity. So increasing
the column length was the simplest way to bypass this problem. If a column is long
enough, one can achieve adequate separation without having to be concerned with
stationary phases and temperature or pressure programming 2. Obviously, this leads to an
increase in analysis time.

In addition there were technological deficiencies in the instrumentation early in
the development of high-speed gas chromatography for which it took years to overcome.
First of all, commercial instruments were unable to handle the high speeds. Detectors
and other data collection systems were not fast enough to measure rapidly eluting
compounds. Also, oven and pressure capabilities were not advanced enough to produce
fast and precise programming rates. Additionally, the use of short columns needed for
fast separations led to a loss in peak capacity. This shortcoming restricted the complexity
of the sample that could be analyzed”

Fortunately, the needs of field portable GC devices and the need for
chromatographs with higher throughput led to greater developments in fast GC
technology. Advances in injection systems have reduced the bandwidth of the initial
injection to reduce band broadening. Faster detectors, such as flame ionization detectors
and time of flight mass spectrometers, can keep up with the faster elution of compounds.
In addition, gas chromatographs can be equipped with efficient ovens that are capable of
achieving faster temperature program rates. They are also available with more sensitive

pressure regulators. The growing need for faster separations combined with several

17

technological advancements have led to the recent growth in the use of high-speed gas

chromatography.

2.2 The Compromise between Speed and Separation

Unfortunately, everything comes at a cost and high-speed GC is no exception.
The measures taken for decreasing analysis time often decrease sample capacity and
selectivity, and also lead to broadening of peaks. Therefore, a compromise between
speed and separation must be found to meet the needs of the analysis at hand.

There are three major approaches that can be applied to achieve this compromise.
Minimizing the resolution to a value that is just sufficient is one option4. It was discussed
before that peaks are often over separated by using extremely long columns. Instead, one
could focus on the peaks or critical pairs that are important for the analysis and optimize
just enough to separate those compounds. This is probably the simplest method for
separation. Another option is to maximize the selectivity of the chromatographic
system4. Finding appropriate stationary phases, which are selective toward the
compounds of interest has been done. It has been shown that more selectivity can mean
less concern for column length and method development. Finally, another approach
involves speeding up analysis while maintaining the same resolution4. If the analysis
time is too long even with the minimum accepted resolution, methods that speed up
analysis without decreasing resolution can be applied. This is a more complicated route,
as it often requires major equipment changes such as the use of narrow bore columns or

new injectors.

18

There is really no universal method for high speed gas chromatography, so
finding a method that works is a challenge. The application determines the approach
needed to decrease analysis time. In choosing an approach, one needs to consider the
amount of speed truly needed to accomplish the goal, the required ruggedness of the
instrumentation needed, the complexity of the sample, sample phase, and method needed
for sample extraction. Sometimes several methods may need to be used in combination
in order to achieve the desired goal. Whatever the goal and method used, it is important
to look at every aspect of the GC instrumentation, from injection, to columns, to

detectors, to temperature and pressure programming, and carrier gas flow.

2.3 Columns

Two simple methods for reducing the separation time are to shorten the length of
the column and to decrease the stationary phase thickness3‘4’5. Shortening the column
obviously speeds up analysis by decreasing the distance the sample has to travel.
Decreasing stationary phase thickness reduces the amount of time that the sample
interacting with the stationary phase. These methods may speed up the analysis, however
they also decrease the efficiency. One must be careful not to lose desired resolution in
changing length or thickness.

Another common method for fast separation while maintaining column efficiency
is the use of narrow bore columns. This method would fall under the category of
maintaining constant resolution. A narrow bore column is one which has an inner
diameter of 0.1 mm, and below, compared to a normal bore which has an inner diameter

of 0.25mm and greater. Narrow bore columns make it possible to use shorter columns

19

and faster flow rates while maintaining resolution. The reason for the resolving power of

narrow bore columns can be seen in equations 2.1 and 2.2 belows.

HETP =

f .2- .
1.? Equation 2.1

HETP is the height equivalent to a theoretical plate; f(k) is a function depending on the
retention factor, r is the column radius, u is the average linear flow velocity of the mobile

phase, and D is the solute diffusion coefficient”. This equation can be rewritten to

include plate number (N).
N --= —-—P—I—;——— Equation 2.2
f (k) - r - u

This equation shows that when column length is shortened, reducing the column radius
and/or increasing linear velocity5 can maintain column efficiency. In fact, when the
length/ID ratio is unvaried, a short narrow bore column has the same efficiency as a
conventional column. Gas compression effects are lower and Van Deemter curves are
flatter with narrow inner diameter columns. Therefore, higher average velocities can be
used without sacrificing efficiency and the analysis time is shortened.5’6

The use of narrow bore columns is widespread, however it is not the universal
method for performing HSGC. Narrow bore columns require the use of very small
sample volumes often on the order of nanolitersl’3’5. Such small sample sizes are very
difficult to create and injection repeatability is also difficult to. In addition, narrow bore
columns require much higher pressures than some GC instruments can provide. Method
development is also difficult, especially in quality control6. However, it is becoming

more convenient to use this method given recent technological advances.

20

It has also been shown by other research groups that fast gas chromatography can
be performed without the use of narrow inner diameter columns. Good separation can be
performed using ordinary inner diameter columns with regular injection techniques. In a
great number of applications, the efficiency of a capillary column is much higher than
necessary. It is believed that normal bore columns can then be shortened to meet the
minimum separation requirements of medium complexity mixtures. Loss of efficiency
due to shortening columns can be compensated by carefiilly choosing selectivity in the
stationary phases. One can attempt to find one specific column that meets the separation
needs for the target compounds. However, selectivity can also be achieved using mixed
stationary phases or series coupled ensembles of two capillary columns7. These columns
can be designed for specific use with the compounds of interest. Increasing the
selectivity this way can allow for the use of a shorter column with a normal inner
diameter. Using normal inner diameter columns helps to maintain sample capacity which
is lost when using narrow inner diameter columns. Therefore, split ratios do not have to
be extremely high or the GC instrument can be run in splitless mode. It can also help to

avoid the need for more complicated instrumentation.

2.4 Injector Considerations

Most common injection systems are acceptable for high speed GC if applied
properly. One must remember, however, that it is often useful to minimize the injection
bandwidth whenever possible4. This will limit the amount of band broadening during
elution through the column and therefore, aid in separation. With a common splitless

injection, sometimes decreasing the amount of sample injected is enough to achieve

21

adequate separation. Sometimes more sample is needed to decrease bandwidth, so split
injection may be used. Many operating systems today are capable of achieving very high
split ratios that produce injection bandwidths on the range of milliseconds4.

Although a normal injector may be acceptable for some applications, it might be
advantageous to use a special injection technique. In addition to loss in resolution and
peak capacity from reducing the column length, the use of high velocities also increases
the occurrence of band broadening due to injection. Therefore, the sample’s bandwidth
not only needs to be reduced, but the peak capacity needs to be used more efficiently.
The use of a rapid and controlled injection system can help to alleviate this problem3.

This special injection technique involves cryofocusing or freezing a sample into a
very narrow plug. These plugs can have a bandwidth of smaller than lms. For
comparison, typical GC instruments produce injection bandwidths of 50-500ms'. This
very small plug can be injected into a shorter, regular inner diameter column, while using
faster flow rates and adequate separation can be achieved. Sacks and coworkers
developed such an instrument. This instrument contained a nitrogen cooled, electrically
heated cold trap inlet which could be used to decrease sample injection bandwidth. The
advantage to this method is that normal columns can be used so sample size restriction is
not an issue. In addition this added injection technique can have the technological
capability to supply the high pressures needed. Also this instrument can be added on to

existing chromatography equipment.

2.5 Temperature Programming

Temperature is a factor in controlling the speed and efficiency of a separation.

22

For samples of medium to high complexity with a wide range of boiling points, a
programmed analysis might be useful. In order to achieve high speed, it is helpful to
increase the rate of temperature programming compared to a regular GC. Modern GC
ovens can often produce programming rates of 50-100°C/min compared to the normally
used 2-10 °C /min. The one drawback to temperature programming is the cooling and
recovery time it takes to return the system back to initial conditions. This could be a
disadvantage when fast repeated measurements are required. However, there are recently
developed systems that not only have faster heating rates, but faster recovery times as
well. The EZ Flash system can have a heating rate up to 1200 °C/min and can cool from
300 °C to 50 0C in less than 30 seconds4.

It has also been shown that isothermal analysis can be done when analyzing
relatively simple mixturesz. Isothermal methods eliminate the need for recovery and
cooling time of the instrumentation. It also eliminates the need for more complicated

oven equipment such as the EZ Flash system.

2.6 Pressure Considerations

Pressure is also an important factor in hi gh-speed gas chromatography. For
example, the use of narrow bore columns requires high inlet pressures in order to
operate3’4’5. Pressure programming has even been studied as an alternative to temperature
programming in cases where the stationary phase or sample is thermally labile. With
pressure programming steeper gradients could be achieved and there is a more rapid

return to initial conditions). The newest gas chromatographs can be equipped with

23

electronic pressure control, which can achieve both extremely high pressures for narrow

bore capillary use and tunable pressure for programming purposes”.

2.7 Carrier Gas Flow Rate

Increasing the flow rate of the carrier gas is another way to speed up analysis time
in high-speed GC. This can be done until a loss of sufficient resolution is detected.
Another option is to change the type of carrier gas used2’4. One of the most widely used
carrier gases is helium, has one of the highest rates of diffusion. However, hydrogen has
an even faster diffusion rate. The drawback to hydrogen is its explosive nature. Despite
this problem, many modern chromatographs can be equipped with safety shut-off features

4

that protect against hydrogen leaks and decrease the chance for dangerous explosions .

This can make hydrogen gas a viable option.

2.8 Detectors

With faster elution of compounds from a high-speed GC, other instrumentation
needs to be fast enough to handle the speed of separation. Detectors, for example, must
have a fast enough response time to detect peaks with extremely narrow widths3’4'6. They
also need to have low limits of detection, especially in cases where a very small sample
size is used. Some examples of detectors with fast response times are flame ionization,

thermal conductivity, and time-of-flight mass spectrometry.

24

Literature Cited

Sacks, R., Smith, H., Nowak, M., “High Speed Gas Chromatography.” Anal.
Chem. 69 (1998) 29A.

Klee, M., Blumberg, L.M., “Theoretical and Practical Aspects of Fast Gas
Chromatography and Method Translation.” .1. C hromatogr. Sci. 40 (2002) 234-
247.

Ewels, B.A., Sacks R.D., “Electrically Heated Cold Trap Inlet System for High-
Speed Gas Chromatography”, Anal. Chem. 57 (1985), 2774-2779.

Korytar, P., Janssen, H., Matisova, E., Brinkman, U., “Practical Fast Gas
Chromatography: Methods, Instrumentation and Applications.” Trends Anal.
Chem. 21 (9-10) (2002) 558-

Bicchi, C., Brunelly, C, “Conventional Inner Diameter Short Capillary Columns:
An Approach to Speeding Up Gas Chromatographic Analysis of Medium
Complexity Samples.”J. C hromatogr. A 931 (2001) 129-140

David, F ., Gere, D.R., Scanlan, F., Sandra, P, “Instrumentation and Applications
of Fast High-Resolution Capillary Gas Chromatography.” J. Chromatogr. A 842
(1999) 309-319.

Coutant, C., Sacks, R., “Programmable Control of Column Selectivity for
Temperature Programmed GC.” Anal. Chem. 72 (2000) 5450-5458.

Mouradian, R.F., Levine, S.P., Sacks, R.D., “Evaluation of a Nitrogen-Cooled,
Electrically Heated Cold Trap Inlet for High-Speed Gas Chromatography.” J.

C hromatogr. Sci. 28 (1990) 643-648.

25

9. Peters, A., Sacks, R., “Inlet Pressure Programming for High-Speed Gas

Chromatography.”J. Chromatogr. Sci. 30 (1992) 643-648.

26

3. Experiment : Discussion of Fast GC Setups

3.1 The Chromatofast® GC AcceleratorTM

The first attempt at obtaining online distillation data involved the use of a special
cryofocusing injection apparatus called the Chromatofast® GC AcceleratorTM developed
by Richard Sacks and his co-workers at the University of Michigan. It is an inlet system
that attaches to a conventional gas chromatograph to give it high-speed GC capabilities.
It does this by reducing the effects of band broadening during injection in order to deliver
narrower peak widths. As mentioned in Chapter Two, decreasing band broadening
during injection can make it possible to achieve good separation even when using shorter,
normal inner diameter columns and faster flow rates. This is advantageous as it alleviates
the need for the use of narrow bore columns with strict sample size requirements. The
accelerator we used was purchased from Chromatofast®, Inc. for $21,100.

The Chromatofast® gas chromatograph setup is shown in Figure 3.1. This figure
shows the external components of the instrument. There is a sampling interface, which is
equipped with a vacuum pump for taking headspace samples. This particular model
consists of a Valco connector and a Silco-Steel capillary column for sample extraction.
The front panel of the AcceleratorTM contains a keypad and LCD display to monitor the
status of the instrument as well as to navigate the menu and enter data. There is also a
pressure regulator that can be adjusted between 0-60 psig. Finally, there is the heated
interface tube, which transfers the cryofocused sample from the AcceleratorTM to the GC

column. This tube is designed to extend through the wall of the GC oven through a hole

in the GC.

27

LCD

   
    

Air Sampler
(Silco Steel

Capillary)

Column
Connector

Keypad

 

Pressure Control (30-60psi)

Figure 3.1. Schematic of thr GC AcceleratorTM external components.
This air sampler makes it convenient for interfacing with
a still. There is a keypad and LCD for controlling

parameters'.

28

A schematic of the Accelerator“ connected to a conventional gas chromatograph
is shown in Figure 3.22. The accelerator itself consists of three restrictor columns R1, R2
and R3, two solenoid valves V1 and V2, a vacuum pump, cold trap, and an inlet for
carrier gas. The process of sampling and injection involves a sequence of steps,
sampling, flush, and fire. In the sampling stage, sample is pulled through R2 and R3 and
into the cold trap containing liquid nitrogen where cryfocusing of the sample occurs. In
the flush stage, canier gas travels through V2 and sweeps residual sample from R3 onto
the trap as well as sweeps through R2 to eliminate sample carryover. In the analysis step,
the flow reverses in the cold trap, and the tube is flash heated. From there, the sample is
injected into the GC for analysis3.

The basic on line analysis setup with a gas chromatograph and still is shown in
Figure 3.3. A hole was drilled in the vapor transfer pipe just above the condenser and
covered by a septum. The Silco steel air sampling line was connected through the septum
where it could collect distillate vapor prior to condensation. The Accelerator“ then
cryofocused the sample and injected it into the gas chromatograph for separation. Data

was then collected and analyzed using a computer connected to the GC.

29

 

 

 

 

 

 

 

 

 

 

 

 

 

Colurrm

 

 

C) V1 Trap

 

 

 

 

 

 

Figure 3.2. Schematic of the GC Accelerator connected to a typical
gas chromatograph”. The accelerator itself consists of three restrictor
columns R1, R2 and R3, two solenoid valves V1 and V2, a vacuum
pump, cold trap, and an inlet for carrier gas. Sample is pulled through
R2 and R3 and into the cold trap containing liquid nitrogen where
cryfocusing of the sample occurs. Carrier gas travels through V2 and
sweeps residual sample from R3 onto the trap as well as sweeps through
R2 to eliminate sample carryover. Finally, the flow reverses in the cold
trap, and the tube is flash heated. From there, the sample is injected

into the CC for analysis.

30

 

 

 

 

Shimadzu
GC- 1 7A

—

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

— GC Accelerator

Computer
0 T
g o 11 Ifi Chromatofast®

 

 

 

10L Arnold
Holstein Still

Figure 3.3. Schematic of the online setup with the Arnold Holstein still

connected to the GC AcceleratorTM and gas chromatograph. Sample

from the still is extracted through the Silco-Steel capillary into the

Accelerator where it is cryogenically frozen , then flash heated and

injected into the GC.

31

3.2 The Headspace Sampling Setup

The second setup was constructed on the basis that column selectivity would be
sufficient enough that adequate separation is possible without the use of a special
injection technique or narrow bore column. This headspace sampler setup is an efficient
method for extracting vapor samples while a distillation is occurring. The separation of
compounds relies on the selectivity of the chromatographic column and appropriate
temperature and pressure programming. Adequate separation was achieved even with the
use of shorter columns and faster flow rates.

A flow schematic of the headspace sampling instrument connected to a GC is
shown in Figure 3.4. This instrument consists of an injection valve assembly, heated
transfer line, and purge and trap adapter. Carrier gas supplies a constant flow to both the
GC and the headspace sampler assembly throughout the process. Carrier first flows
through a line in the GC, where it is passes through the purge and trap adapter. This
particular adapter came from Shimadzu, Inc., which made it compatible with our GC.
The adapter splits the flow and directs some to the injection valve and some to the GC
injector. From the injector, carrier gas flows to the column and eventually to the detector
to ensure constant circulation through GC instrument. From the injection valve, carrier
flows to the heated transfer line and again to the injector, which flows through the
column and the detector. This configuration ensures constant flow through the injection
valve and helps to transfer sample from the sample loop to the heated transfer line.

The injection valve serves as the heart of the apparatus as it allows sample to be
loaded into the sample loop and then injected onto the column. The injection valve used

was a six port, two position electrically actuated valve from VALCO Instruments

32

Carrier Gas (He)

. Heated
Sampling Tr sfer Line 0
\Klve/Loop 2“]

00%. J_\

 

 

 

 

 
   
 
   

 

 

 

 

Injector

Valve
Enclosure

Purge and /
Trap Adapter

Column

A

 

 

Detector (F ID) --“

 

 

 

 

Gas
Chromatograph

Figure 3.4. A schematic of carrier gas flow through the entire
system. Helium from an air cylinder enters through the GC and
flows to the purge and trap adapter. The adapter splits the flow
and directs carrier gas to both the GC injector and the injection
valve. Helium flowing through the injector travels through the
column and the detector to ensure flow through the GC. Carrier
flowing through the injection valve travels through the heated
transfer line, where it enters through the injector and eventually
into the column and detector. This ensures that sample is

transferred to the GC.

33

Incorporated. This particular valve has a two-inch standoff so that it can be encased in a
valve enclosure, which allows for heating and insulation. The injection valve is
connected to a resistively heated block. The temperature of the valve is maintained at
least at 120°C in order to ensure that the sample remains as a vapor inside the apparatus.
Temperature control is achieved using a Valco temperature controller and a thermocouple
that is connected just underneath the heating block.

The valve flow diagram shown in Figure 3.5 reveals how carrier gas and sample
flows through the system. This valve has six ports, 1 for the sample to be loaded, 2- for
venting waste, 3 and 6- for the sample loop, and 4 and 5 for the entrance of carrier gas.
In the load position, sample is loaded into the sample loop and directs to the vent.
Meanwhile carrier gas is flowing from the GC, through the valve directly to the heated
transfer line and onto the column. In the inject position, the ports are switched so that
carrier gas flows through the sample loop and pushes sample through to the transfer line.
From the transfer line, sample is transferred to the injector and onto the column to be
analyzed.

The heated transfer line comes from Tekmar Industries. It consists of copper
tubing with an outer diameter of 1/ 16” and an inner diameter of 0.02”. Its insulated line
is 60” long. It contains a K-type thermocouple, which is connected to an Omega CN9000
temperature controller. Again the temperature is maintained at least 120°C in order to
ensure that the sample remains a vapor. From the transfer line, sample enters into the
GC.

The computer is used for both instrument control and data acquisition. As a

trigger, the computer signals the electronic actuator on the valve to move into the load

34

Load

Sample In

,0 Vent/Waste
G 0

To 0 9
Column <— \0
_._/T\

l

Carrier/
Mobile Phase

 

 

Inject

Sample In

Vent/Waste

To
Column

 

 

 

Carrier/
Mobile Phase

Figure 3.5. A close-up view of the sample loop inside the sampling
valve. In the load position, the vapor sample from the still enters
through the sample loop. In the inject position, the carrier gas enters
the sample loop and tranfers the sample to the transfer line and

eventually to the column‘.

35

and inject positions. To achieve this, the electronic actuator was wired to the S8420
board of the GC computer. GC software was used to generate an external events table
that controlled the positions of the valve. The computer simultaneously collected data
from the computer by reading signal from the detector and developing chromatograms
during the analysis.

Figure 3.6 shows the entire online setup including the still, the headspace sampler
assembly, the GC, and the computer. The headspace assembly is connected to the still
through the sampling port just above the condenser. With this apparatus, the computer
triggers the opening of the injection valve to the load position. Simultaneously, the
computer begins data acquisition. While in the load position, pressure from the still
transfers vapor sample above the condenser into the sample loop. The sample travels
from the sampling port, to the sample loop in the injection valve, and through a heated
transfer line. The purge and trap adapter connects this transfer line to the gas

chromatograph where the sample travels through the column and detector to be analyzed.

36

Actuator Computer

Heated Transfer Line

Valve

 

 

 

 

 

 

 

 

 

 

Shimadzu :::
GC-17A "'

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Temperature Controllers

 

 

  

 

10 L Holstein Still

 

 

— - - — Vapor Path from Still - - — - - Temperature Control -
. Heated Transfer Line
Carrier Gas Path
----------- Temperature Control -
Valve Trigger Control Valve

 

 

 

Figure 3.6. Schematic of the headspace sampling system connected to the
Shimadzu GC-17A.

37

Literature Cited
Adapted from Chromatofast® GC Accelerator“ Air Sampler User Manual.

Adapted from Chromatofast® GC Accelerator“ Fast GC Operation Technical

Note.

Chromatofast® GC Accelerator“ Air Sampler User Manual.
Adapted from VALCO Flow Diagram Technical Note.

Shimadzu Purge and Trap Adapter Technical Note.

38

4. Materials and Methods

4.1 Spirits Production Process
4.1.1 Creation of Standard Solutions

Actual fruit fermentations were not used in performing the online experiments.
Fermentations generally take two weeks to ferment, which limits the number of
distillations that can be performed in a given period of time. Therefore standard solutions
were used to simulate distillation behavior. Creating a mixture that contains a similar
amount of congeners as that of a fermentation mash can be done. Solutions were made in
30 mL batches so experiments could be run in triplicate. These solutions consisted
mainly of ethanol and water. There are hundreds of congeners that exist in fruit spirits,
therefore, only the most prominent ones were investigated. The congeners used and their
volume percentages are listed in Tables 4.1 and 4.2. For the experiment with the
AcceleratorTM, the congeners chosen were acetaldehyde, ethyl acetate, ethyl formate,
acetone, methanol, n-propanol, and isoamyl alcohol. In the headspace sampling

experiment all the congeners were used except for ethyl formate and acetone.

4.1.2 Distillation

Distillation was performed using a 10 L Arnold Holstein still, as shown in Figure
4.1. This still is mainly constructed out of copper and consists of a pot, a steam jacket,
three bubble cap trays, a partial condenser, and a total condenser. The pot is where the
solution to be distilled resides. It is heated by the steam jacket, which uses electrically
heated steam. The bubble cap trays can be opened or closed so that one, two, or all three

trays may be activated. Just above the pot is a partial condenser, which can be turned on

39

 

 

 

 

 

Compound Volume (mL) Percent V/V Solution
Water 27486 91.62
Ethanol 2400 8.00
Acetaldehyde 1 5 0.05
Methanol 24 0.08
Ethyl Formate 6 0.02
Acetone 15 0.05
Ethyl Acetate 6 0.02
n-Propanol 24 0.08
Isoamyl Alcohol 24 0.08

Table 4.1. Congener types and volume percentages used in

Experiment 1.

 

 

 

 

 

Compound Volume (mL) Percent v/V Solution
Water 27507 91.69
Ethanol 2400 8.00
Acetaldehyde 1 5 0.05
Methanol 24 0.08
Ethyl Acetate 6 0.02
n-Propanol 24 0.08
Isoamyl Alcohol 24 0.08
Table 4.2. Congener types and volume percentages used in

Experiment 2.

40

 

Dis ' ation Vapor

— Cooling Water In
- - Cooling Water Out

 

 

 

 

Total Condenser

E Partial Condenser
[3 Bubble Cap Trays

[:3] Steam Jacket

 

 

 

 

From

 

Water

 

 

 

 

fififififififififlfififififlfififlflfifififi

 

 

 

 

 

 

 

 

 

    

 

Schematic of a 10L Arnold Holstein Still.
41

 

””””””’

Figure 4.1

or off. The final component of the still is the total condenser, which is constantly cooled
by running water in order to condense the vapor to form the resulting distillate. Cooling
water is supplied to both condensers by a hose connected to cold tap water. A flow
regulator located just behind the total condenser can control the flow of this water
manually. This can either increase or decrease the rate of distillate exiting from the still.

Distillation was performed first by adding the synthetic solution. Cooling water
should fill the condensing column and the partial condenser if it is used. Rectification is
relatively poor because of the small size of the still. Therefore, in order to obtain optimal
rectification, the partial condenser was used, and all three trays were closed. The electric
heater was turned on in order to heat the steam jacket. The distillation was then allowed
to run until distillate appears.

The distillate consists of mostly ethanol and water. Congeners exit the still at
different times during the distillation depending on their boiling points. Therefore,
depending on the point in the distillation, the distillate will contain different amounts of
congeners and ethanol. Normally, the distillate would be broken up into the heads,
hearts, and tails portions. However, to keep the analysis consistent, 75 mL cuts were

made.

4.2 Measurement of Condensation Time

It was important to understand how fast the vapor condenses when traveling from
the top of the condenser to the sampling port below in order to determine how fast a GC
analysis needed to be to obtain useful data. This measurement was made by injecting a

lmL sample of methanol into the sampling port above the condenser during an

42

ethanol/water distillation and recording the time that it takes for methanol to condense
and be collected at the sampling port at the bottom. Time was recorded at the moment of
injection and cuts were taken every 15 seconds for about five minutes. The cuts were
then analyzed with the gas chromatograph to determine at which cut methanol appears.

A plot of peak area verses time in Figure 4.2 gives an indication of the time that the
methanol starts to appear. It was found to appear at approximately 3 minutes. So in
developing an online method, the time it takes to extract the sample and analyze it must

be under 3 minutes.

4.3 Analysis

Online analyses were performed using a Shimadzu GC-17A gas chromatograph
with flame ionization detection. This particular GC was a base model that did not have
tunable pressure and could only perform splitless injections. In addition, the connection
between this instrument and the computer did not allow for the analysis software to
download GC parameters and to trigger the GC temperature program to run. Therefore,
any temperature program had to be entered manually to the instrument using the keypad
on the Shimadzu GC. The GC could only be started by manually pressing the start button
on the instrument panel as well. This GC was attached to the GC AcceleratorTM in the
first set of experiments and attached to the headspace sampler for the second set of
experiments.

For both the AcceleratorTM and headspace experiments, a 12.6 m Quadrex CW-
007 Carbowax column was used. This column had an inner diameter of 0.25mm and a

stationary phase thickness of lum. The temperatures of the injector and detector were

43

Methanol Peak Area

 

60000

50000 a

40000 4

30000 *

20000~

10000 ~

 

 

 

 

ll ILILIIIILJLIIIILLJLIIJILIIIIILIII
' T I l

0 50 100 150 200 250 300 350
Time (seconds)

Figure 4.2. Travel time for methanol condensation from the sampling
port above the condenser to the distillate collection port at the
bottom. It takes over three minutes before methanol distillate

begins to appear.

44

240°C and 255°C, respectively, and remained constant throughout the analysis. The
temperature and pressure conditions of the column varied depending on the method used.
The analysis parameters for each experiment are shown in Tables 4.3 and 4.4.

Acquisition with the AcceleratorTM began by turning on the sampling pump and
sampling for approximately fifteen seconds. Pressing the start button on the GC and
pressing start on the computer manually triggered the GC program. For the
AcceleratorTM method the GC was run isocratically at 70°C and at a pressure of 38 psi.
The elution time was approximately 2.5 minutes. Since this was an isocratic run there
was no need for recovery time, so the total analysis time was 2.5 minutes.

For acquisition with the headspace sampling system, analysis began by
simultaneously pressing start on the computer and pressing start on the GC instrument.
Pressing start on the computer triggered the injection valve to switch from the inject to
load position to allow vapor sample from the still to enter the sampling loop. At the same
time, pressing start began data acquisition. The sampling loop was programmed to stay
open for 9 seconds. Because the sample was vapor and because the pressure from the
still was used to transfer sample, it was difficult to determine the exact sample volume
that was being injected into the GC. Pressing start on the GC allowed for the temperature
program to begin. A pressure of 43.5 PS1 was used for this analysis. Separation was
only possible with a temperature program for this particular method of sampling. The
initial temperature was 40°C and this was held for 0.4 minutes. Next the temperature was
ramped 70°C/minute to 115°C. The GC run time was 1.5 minutes. There was a 1 minute
recovery time to cool the oven back down to initial conditions. The total analysis time

was 2.5 minutes.

45

 

Experiment 1

Accelerator Setup

 

 

 

 

 

 

 

 

 

 

 

 

Inlet Pressure 38 PSI
Injector Temperature 240 °C
Detector Temperature 255 °C
Sample Time 15 seconds
Temperature Program 70 °C Isocratic
GC Run Time 2.5 minutes
Recovery Time 0 minutes
Total Analysis Time 2.5 minutes
Table 4.3. GC analysis parameters for the AcceleratorTM

setup.

 

 

 

 

 

 

 

 

 

 

 

 

 

Experiment 2 Headspace Setup—Ix
Inlet Pressure 43.5 PSI
Sample Time 9 seconds
Injector Temperature 240 °C
Detector Temperature 255 °C
Temperature Program 40 °C hoeld 0.4 mrnutes.
Ramp 70 C/ minute to 115.
GC Run Time 1.5 minutes
Recovery Time 1.0 minutes
Total Analysis Time 2.5 minutes
Table 4.4. GC analysis parameters for the headspace

setup.

46

During the distillation, vapor samples were taken approximately every 150 L for
both setups. Compounds were identified by their relative retention times during the runs.
These retention times varied depending on the method used. The retention times for the
compounds along with boiling points in each experiment will be discussed further in the
chapters five and six as part of the results.

Concentration of ethanol was measured after distillation using a standard curve.
Samples of distillate cut were run through conventional GC to obtain ethanol
concentrations. With this information, ethanol was used as the internal standard. The
area percents of the congeners were compared to the area percent of ethanol. Calibration
curves were not used in this experiment due to the poor reproducibility in standardization

measurements.

4.4 Simple Distillation

Simple distillations were performed in order to test the reproducibility of the
headspace sampler. A distillation apparatus, as shown in Figure 4.31,was connected to
the headspace sampler. This apparatus was like most other simple distillation setups
except there was an extra port in the round bottom flask to be used to connect to the
headspace sampler to obtain online data.

A 1000 mL brandy type solution with similar percentages to both the
AcceleratorTM and headspace sampler experiments was used and the composition is
shown in Table 4.5. Approximately 150 mL of this solution was placed in a 300mL
round bottomed flask and attached to the distillation apparatus. This flask was immersed

in a heated silicon oil bath and the distillation was allowed to proceed. Ten milliliter cuts

47

were collected. Sample injections were made approximately every 2.5 minutes so that
there was at least one measurement for every cut. The distillation proceeded until the

temperature reached 100°C.

48

Thermometer

 

 

Inlet Adapter

& Three-way

A/Adapter Condenser
Vacuum Adapter

MN 0Pcn
" rm! l\ _ ll

/l

Water Outlet Water Inlet ".i'i.i-<—Clamp

~-

300 mL Flask 50 mL Flask

\
[Silicon Oil Bath 1 [Ice Water Bath |

 

 

Figure 4.3. A typical simple distillation apparatus. The round
bottom distillation flask used is 300mL and contains more than one
port. The extra port is needed to insert the sample line of the
headspace sampler in an area above the distillation in the flask. The

distillate collection flask is 50 mL'.

49

 

 

 

 

 

Compound Volume (mL) Percent V/v Solution
Water 916.9 91.69
Ethanol 80 8.00
Acetaldehyde 0.5 0.05
Methanol 0.8 0.08
Ethyl Acetate 0.2 0.02
n-Propanol 0.8 0.08
Isoamyl Alcohol 0.8 0.08
Table 4.5. Congener types and volume percentages for the simple
distillations.

50

Literature Cited

1) Simple Distillation apparatus. Adapted from

<http://www.ComCapeCanaveral/lab/4444/distillation.html>

51

5. Results : GC Accelerator

5.1 Method Optimization

The AcceleratorTM method successfully collected sample from the still and
generated a chromatogram. The next step was to optimize this method. An appropriate
sampling and GC program were developed to achieve speed while maintaining adequate
separation,. As shown in Chapter 4, the conditions that gave the best separation at the
time was an isocratic run at 70°C. An isocratic run allowed for repeated runs without
taking time to reset the system. The entire run time lasted about 2.5 minutes. The
chromatogram that resulted is shown in Figure 5.1. Figure 5.2 shows a chromatogram of
a similar brandy type mixture taken using normal GC injection and method. Comparing
the retention times given in the chromatograms to standard retention times of the
compounds for both the fast GC and conventional GC methods allowed for the
identification of the peaks. The AcceleratorTM method proved to be faster in comparison
with the conventional GC run. A list of the compounds present and the retention times
for both the AcceleratorTM and the normal GC run are shown in Table 5.1.

Isoamyl alcohol does not elute within the 2.5 minute run for the
AcceleratorTM method. The program chosen allowed for good separation and elution of
the other compounds. However, isoamyl alcohol eluted much later, beyond the
maximum required run time of three minutes. It was decided to run an experiment with
this particular program without including isoamyl alcohol. A second experiment was to
be run with temperature programming so that isoamyl alcohol could be studied as well.
However, the failing structural integrity of the Chromatofast® GC AcceleratorTM did not

allow for further experiments. This issue will be discussed later in the chapter.

52

 

 

 

 

 

 

 

 

 

 

750
ii
4
500
1r-
mV
3
250 __ 5
1 2
0 d- A ‘—
i i i i
0 0.5 1.0 1.5 2.0 2.5
Time (minutes)
1) Acetaldehyde 2) Ethyl Acetate 3) Methanol
4) Ethanol 5) N-propanol 6) Isoamyl Alcohol

 

 

 

Figure 5.1.

A chromatogram of a brandy type solution taken with the
GC AcceleratorTM Using a 70°C run. Peaks were identified
by comparing the retention times from the chromatogram
to a standard set of retention times for each compound.
Isoamyl alcohol does not appear due to a much later elution

time.

53

1000
900
800
700
600
mV
500
400
300
200

100

 

 

 

 

 

 

 

 

I I I

. _ 10 15
Time (min)

 

 

l) Acetaldehyde 2) Ethyl Acetate 3) Methanol
4) Ethanol 5) N-propanol 6) Isoamyl Alcohol

 

 

Figure 5.2

Chromatogram of a sample from a brandy-type solution
using a conventional GC. Distillate cuts were taken and
analyzed after distillation. Peaks were identified by
comparing the retention times from the chromatogram to a

standard set of retention times for each compound.

54

 

 

 

 

 

GC AcceleratorTM Conventional
Compound (min) GC(min.) Boiling Points (°C)

Acetaldehyde 0.801 :t 0.008 1.451 t 0.000 20.8
Ethyl Acetate 1.232 i 0.008 2.667 :t 0.002 77.0
Methanol 1.276 :1: 0.008 2.825 i 0.004 64.7
Ethanol 1.493 d: 0.006 3.563 i: 0.015 78.0
n-Propanol 2.308 t 0.009 5.996 :1: 0.019 97.0
Isoamyl Alcohol N/A 14.639 :t 0.051 132.0

Table 5.1. A list of the retention times for chromatograms taken using

Fast GC with the GC AcceleratorTM and also conventional

GC. The boiling points for the congeners is also listed.

55

5.2 Standardization

Standardization with the online setup turned out to be quite challenging. In an
attempt to develop a calibration curve for ethanol, standards were created at 40%, 50%
65%, and 85% and placed into 2mL GC autosampler vials. Prior to running the standards
through the fast GC, the vials were heated at a temperature of 75°C for at least five
minutes to approximate the conditions of the still as best as possible. Headspace above
the liquid in the vials was extracted for analysis and a calibration curve was plotted. This
plot is shown in Figure 5.3 with the standard deviation between the runs.

From this information it appears that the trends do not follow expected behavior.
Typically, peak area increases with an increasing concentration of ethanol; however, the
plot obtained from standardization showed a periodic increase and dip in peak area with
increasing concentration. This could have been due to the sample vials used to store the
ethanol standards. The vials may not have allowed for a consistent pressure to be
maintained. So the sample collected from the headspace in the vial could have
fluctuated. Inconsistent sampling from the GC AcceleratorTM could have also been the
problem. Looking at the error, the repeatability of the measurements was actually quite
reasonable, falling in the range of 0.48%-l .64%. Therefore the AcceleratorTM was most
likely not the problem. The problem most likely was due to pressure variations in the
sampling vials. This could also be due to the fact that it is difficult to closely imitate the
conditions that are present in the still when developing standard samples. Also, the
relative concentration of ethanol in the still during the distillation may actually be under

40%v/v rather than being equal to the measure of ethanol in the liquid distillate.

56

 

6.90E+06 1 ---—~ ~447. WWW- -- "LL“ WWW

6.80E+06 ~

6.70E+06 ~

6.60E+06 -

Peak Area

6.50E+06 -

6.40E+06 —

 

 

6.3OE+06 T r T T 1 T
30 40 50 60 70 80 90 1 00
Ethanol Concentration (%vlv)

 

Figure 5.3. A calibration curve for ethanol using the GC AcceleratorTM

method. Standard samples were made at 40%, 50%, 65%,
75%, and 85% v/v ethanol. These samples were heated to
75°C for at least 5 minutes. Rather than showing an increase
in peak area with increasing ethanol concentration, this curve

shows a periodic increase and decrease.

57

Studies on the relationship between the vapor and the liquid would need to be done to
resolve this issue.

The unusual behavior of the standards made it difficult to actually quantify the
amounts of ethanol and congeners that were present in the sampled vapor. So as an
alternative, it was decided that ethanol could be used as an internal standard. The amount
of ethanol could be measured after distillation using standard curves for conventional
GC. Ultimately all of the data would be normalized to make ethanol 40% in order to

make a better comparison between runs.

5.3 Distillation Behavior

The behavior of the compounds in the solution was studied during the distillation.
Chromatograms were taken using the online system. Peaks were identified by comparing
retention times from analysis to standard retention times for each of the congeners. Data
was collected for each congener and plots were generated to show trends during the
distillation. In addition, samples were collected from each cut that was taken closest to
the time of the online analysis.

The major component of the synthetic brandy solutions was ethanol. Ethanol
amounts were found after distillation by injecting samples from the distillate cuts into a
conventional GC. Ethanol concentrations were found using a standard ethanol curve
made for the conventional GC. Knowing this, the peak areas of the other compounds
were then compared to the peak area of ethanol in order to obtain relative amounts. The

ethanol response by conventional GC is shown in Figure 5.4. Ethanol starts near 80%v/v

58

and slowly decreases. Closer to the tail end of the distillation, the decrease in percent
volume of ethanol becomes sharper.

The major congener to emerge first is acetaldehyde. The behavior of this
compound during distillation over time is shown in Figure 5.5. Acetaldehyde starts off
with a relatively high concentration and decreases rather sharply as time goes on. This
makes sense given acetaldehyde’s low boiling point.

Ethyl acetate was studied next. The plot for its behavior is shown in Figure 5.6.
Ethyl acetate is another low boiling point compound. So like acetaldehyde, the relative
concentration starts out high and quickly drops.

Methanol emerged next. Methanol starts out relatively high in the early part of
the distillation and decreases towards the middle. The relative concentration then begins
to increase towards the end of the distillation. This behavior is shown in Figure 5.7.

The fusel alcohol, n-propanol was then examined. The behavior pattern is shown
in Figures 5.8. N-propanol starts out relatively low and increases as the distillation
occurs. This is probably due to the higher boiling point of n-propanol. There is a drop
towards the end of the distillation, which is probably a result of decreasing amount of n-

propanol in the pot.

59

Ethanol Concentration (%vlv)

 

 

 

 

75 —
70 —
65 —
60 —
55 1 1 i .
O 5 1O 15 20 25 30
Distillation Time (min.)
Figure 5.4. Ethanol concentration measured after the

distillation. Samples of the distillate cuts were
taken and analyzed using conventional GC.
Concentrations were found by comparing peak

areas to a standard curve for ethanol.

60

Acetaldehyde Concentration
(%vlv in 40% Ethanol/Water Mixture)

0.3 (W -

0.25 ..

0.2 r

 

0.15 a --

0.1 ~

0.05 a

 

 

 

Figure 5.5.

T i l T I

5 10 15 20 25

Distillation Time (min.)

Acetaldehyde concentration in the product as a
function of distillation time according to
measurements made with the GC AcceleratorTM.
Concentrations are given as percent volume of

congener in a 40% ethanol/water mixture.

61

 

Ethyl Acetate Concentration
(“/ovlv in 40% Ethanol/Water Mixture)

 

10*

 

 

Figure 5.6.

l I I +
5 10 15 20 25

0

DistillationTime (min.)

Ethyl acetate concentration in the product as a
function of distillation time according to
measurements made with the GC Accelerator“.
Concentrations are given as percent volume of

congener in a 40% ethanol/water mixture.

62

 

Methanol Concentration
(%vIv in 40% Ethanol/Water Mixture)

 

5.35 a

4.85 -

4.35 —

3.85 .

3.35 .

2.85 e

 

2.35

 

 

 

Figure 5.7

I I I I I

5 10 15 20 25

Distillation Time (min.)

Methanol concentration in the product as a function
of distillation time according to measurements made
with the GC Accelerator“. Concentrations are
given as percent volume of congener in a 40%

ethanol/water mixture.

63

N-Propanol Concentration
(%vlv in 40%EthanollWater Mixture)

 

4.6

4.1 e

3.6 a

3.1 r

2.6 ~

2.1 -

 

 

 

1.6

Figure 5.8

| I I I I

5 10 15 20 25

DistillationTime (min.)

N-propanol concentration in the product as a
function of distillation time according to
measurements made with the GC AcceleratorTM.
Concentrations are given as percent volume of

congener in a 40% ethanol/water mixture.

64

 

5.4 Structural Integrity of GC AcceleratorTM

The structural integrity is a very important consideration when looking at an
online method’s ability to monitor a distillation. Robustness is an area of weakness for
the GC Accelerator“. It affected the quality of the chromatographic peaks obtained and
affected the instrument’s ability to even function at all.

One main problem with the GC AcceleratorTM was that after repeated
measurements, the quality of the chromatograms became poor. Figure 5.9 illustrates
what would happen to the chromatograms after several measurements were taken. The
formation of poorly shaped peaks was noticed in the online readings. In addition, a very
distinct shoulder appeared on the ethanol peak during every measurement.

The formation of shoulders and other poorly shaped peaks was probably a result
of the sample not being “plug-like” upon injection into the GC. This could be due to
several things. The cryotrap itself may have no longer been efficient in freezing the
sample. This behavior could lead to wider injection bandwidths, which could lead to
poor efficienc. In addition, inefficient flushing of residual sample in the lines of the
AcceleratorTM could lead to the appearance of unexpected peaks. If sample from the
previous analysis is left behind in the lines, it could be injected along with the new
sample from the current run and distort the chromatogram. The solenoid valves may
have begun to fail to open and close properly also. This would cause an inefficient
transfer of sample and also lead to poor chromatograms.

Another drawback of the GC AcceleratorTM was that the liquid nitrogen transfer

line often became blocked by ice and would not allow liquid nitrogen to cool the

65

l 000

 

500 1
mV

 

 

 

 

 

.. Jill L

I I

0 1.0 1.5 2.0 2.5
Time (min.)

 

 

 

 

Figure 5.9 Chromatogram of a sample from a brandy-type solution
using the GC AcceleratorTM. After repeated measurements,
the chromatograms would develop poorly shaped peaks,
such as in this example. The shoulder on the ethanol peak is
most likely a result of the injection volume not being “plug-

flow”.

66

cryogenic trap. The liquid nitrogen was stored in a Dewar and transferred through Tygon
tubing into the cryogenic trap. If water was inside this tube, it would freeze upon contact
with the liquid nitrogen. The line then became blocked and no more liquid nitrogen
could be transferred to the trap to maintain a temperature of —80°C. Once blockage
occurred, the GC Accelerator“ was more or less rendered useless until the ice thawed
and essentially evaporated, which often took hours.

Finally, the GC Accelerator“ eventually failed to operate at all. The instrument
began to produce extremely small peaks or none at all. An attempt to repair the machine
was made, however this proved to be difficult. First of all, the instrument itself was very
compartmentalized, thus making it difficult to open up and reach the inner workings.
Second, the manual that accompanied the instrument was of little aid as far as
troubleshooting and instrument diagrams. Finally, Chromatofast® the company is no
longer in existence, so there was no hope for receiving technical assistance and repair
services.

Initially, the GC Accelerator“ achieved online readings. However, the failure of
the instrument to continue to operate eliminates it as a viable option for real-time

monitoring of a distillation.

67

6. Results: Headspace Sampler

6.1 Method Optimization

It was found that a special injection technique, such as the GC Accelerator® was
not necessarily required to achieve good separation while speeding up analysis. With a
short column and faster flow rate, along with the appropriate GC programming, fast,
efficient chromatographic readings could be obtained. The only thing that was needed
was a method for extracting vapor from the still. So a headspace sampler was
constructed to extract sample while using fast gas chromatography for analysis.

The GC program was shown previously in Table 4.4. A temperature program was
the best at achieving adequate separation. The elution time was approximately 1.5
minutes. Since a temperature program was used, time was needed to cool down the
system to initial conditions. This took approximately 1minute. The total run time of a
sample acquisition was 2.5 minutes, which is the same as that for the GC Accelerator®.
The chromatogram that resulted from this type of analysis is shown in Figure 6.1. A
comparison to a normal GC run of a similar mixture is shown in Figure 6.2. This method
definitely proves to be much faster than a normal GC run. The elution is also faster than
that of the GC Accelerator®. This can be seen in a comparison chromatogram in Figure
6.3. The top chromatogram shows the elution with the headspace sampler. The bottom
chromatogram represents the elution with the Accelerator®. Peaks were identified by
comparing the retention times from the chromatogram to standard retention times for
each of the compounds. A table with the retention times for all three methods is shown in

Table 6.1.

68

 

 

 

 

 

 

 

 

 

 

 

150
l 4
100‘
mV

50‘ 3
2
\ 5 6

.— l r. I / \

l A
0 0.5 1.0 1.5
Time (min.)

 

1) Acetaldehyde 2) Ethyl Acetate 3) Methanol
4) Ethanol 5) N-propanol 6) Isoamyl Alcohol

 

 

 

Figure 6.1 Chromatogram for the distillation of a brandy-type
mixture. The sample was collected from the still
using the headspace sampler. The program used
was the same as in Table 4.4. Peaks were identified
by comparing the retention times from the analysis
to standard retention times for each of the

compounds.

69

mV

1000
900

800
700
600
500
400
300
200

100

 

0 _igitz L l/ \/L

 

 

 

 

 

 

I I

Time (min.)

 

4) Ethanol

 

1) Acetaldehyde 2) Ethyl Acetate 3) Methanol

5) N-propanol 6) Isoamyl Alcohol

 

 

Figure 6.2

Chromatogram for a sample from a brandy-type
solution using a conventional GC. Distillate cuts
were taken and analyzed. Peaks were identified by
comparing the retention times from the
chromatogram to standard retention times for each

of the compounds.

70

 

250

200 a

150 ‘
mV

100 ,

 

 

 

 

 

 

 

 

 

 

Figure 6.3

 

Time (min)

A comparison of the chromatograms for the two methods.
The solid curve represents the headspace sampling analysis
and the dashed curve represents analysis with the GC
Accelerator”. The total elution time with the headspace
sampling system is 1.5 minutes whereas the total elution time
for the AcceleratorTM is 2.5 minutes. Peaks were identified by
comparing the retention times from the chromatogram to

standard retention times for each of the compounds.

71

 

 

 

 

Headspace Sampler GC AcceleratorTM Conventional
Compound (min.) (min) GC(min.)
Acetaldehyde 0.276 :t 0.007 0.801 d: 0.008 1.451 :t 0.000
Ethyl Acetate 0.582 :t 0.004 1.232 t 0.008 2.667 :1: 0.002
Methanol 0.620 i 0.004 1.276 :1: 0.008 2.825 :1: 0.004
Ethanol 0.719 :t 0.007 1.493 d: 0.006 3.563 at: 0.015
n-Propanol 0.992 :t 0.005 2.308 t 0.009 5.996 i 0.019
Isoamyl Alcohol 1.418 i 0.004 N/A 14.639 1 0.051
Table 6.1. Retention times for chromatograms taken using Fast GC

with the headspace sampler and also conventional GC.

Peaks were identified by comparing the retention times

from the chromatogram to standard retention times for

each of the compounds.

72

 

6.2 Standardization

Standardization with the headspace sampler proved to be just as difficult as with
the AcceleratorTM. This can be seen in the attempt to standardize ethanol. Ethanol
standards were made at 40%, 60%, 80% and 90% and placed in 2mL GC autosampler
vials. They were heated to a temperature of 75°C in order to approximate the conditions
of the still as best as possible. Figure 6.4 shows the calibration curve for ethanol and the
error between readings. The error range for triplicate runs at the same concentration was
44%-155%. Additionally, the expected trend of increasing peak area with increasing
concentration is not followed.

Again it is difficult to standardize headspace samples that have the same
conditions as the still. This strange behavior could also be due to the fact that the
sampling is not consistent. The headspace sampler relies on the pressure of the still or
sample vial to push sample through the sample loop. If the pressure fluctuates in the still
or the standard sample vial, the volume that enters the sample loop will be different every
time. Therefore, the volume injected into the GC will be different every time, thus
leading to irregular peak area readings for each run of a particular concentration.

Since repeatability was poor, quantification was again difficult. Comparing the
peak areas of the congeners to the peak areas of ethanol and then normalizing the values
to give a 40%v/v ethanol concentration was the method used for finding congener

concentration.

73

Peak Area

1

2.00E+06

1.50E+06

1

l

1.00E+06

l

5.00E+05

I

0.00E+00

 

-5.00E+05
20

Figure 6.4.

30

 

 

 

 

I I I I I r

40 50 60 70 80 90
Ethanol Concentration (%vlv)

 

_n__.__ .. __

100

The response curve for ethanol using the headspace

sampler. The peak areas do not increase with

increasing concentration.

74

6.3 Distillation Behavior

The behavior of all of the compounds was studied using the headspace sampler.
The concentrations for each of the compounds were calculated by comparing the peak
areas of the congeners to the peak area of ethanol. These concentrations were then
plotted to show percent volume of congener in a 40% ethanol/water mixture vs.
distillation time. A post-distillation analysis was also performed on the cuts of distillate
using a conventional GC. This was done in order to compare the behavior and amounts
measured by the fast GC to the reliable measurements of the conventional GC.

Ethanol measurements were taken after the distillation using conventional GC.
Samples from the distillate cuts were measured and compared to a standard ethanol
curve. The behavior of ethanol during the distillation is shown in Figure 6.5. As in the
previous experiments, ethanol starts out around 80% v/v of liquid and slowly decreases
until it gets closer to the tails portion where the decrease speeds up.

The behavior of acetaldehyde was studied both before and after distillation.
Figure 6.6 shows the comparison between online measurements of behavior and post-
distillation measurements. Other than the first point in the online measurements, the two
plots appear to have agreement in the trends. The amounts calculated according to the
online analysis are also markedly different from the amount measured according to the
regular GC.

The behavior of ethyl acetate is shown in Figure 6.7. The online and post-
distillation trends appear to be similar here. Like acetaldehyde however, the amounts
differ substantially. The online measurements give concentration values that are

significantly higher than the post-distillation measurements

75

Ethanol Concentration
(%vlv in ethanol/water mixture)

 

90

85*

80*

75*

70*

65*

60*

 

 

55

Figure 6.5.

I I I I I I

4 6 8 10 12 14
Distillation Time (min.)

Ethanol concentration as a function of distillation time
taken after distillation using a standard ethanol curve

with conventional GC.

76

 

 

 

Acetaldehyde Behavior

 

 

 

 

 

 

 

 

 

40 ~ 1' ——7 2
3 35 — _\ +On|ine Data T5 1-8
h c
i" +Post-Distillation T 1-5 .2
- 30 r ‘5
C 2 Data .4- 1 4 3
.—‘3 3 . ' g
2 iii 25 * g
E E a
3 0 o
c c 20 — c
o g g
o .. .5
o "J =
5 °\° 15 * 3'.-
e 3 3
o c . .5
; 1° ‘ 3
.5 0.
.\°
V 5 «
O r r I I I I " 0
o 2 4 6 8 10 12 14
Distillation Time (min.)
Fi ure 6.6. The concentration behavior of acetaldeh de. The online
g y

measurement is represented by the diamond plot and
corresponds to the left y-axis. The square plot represents
the behavior from post distillation measurements and

corresponds to the right y-axis.

77

(“Av/v in 40%EthanollWater Mixture)

 

 

 

Ethyl Acetate Behavior

 

 

 

 

 

 

 

 

--..._-1.4
. + Online Data
20 q ~ 1.2
’27 ~I- Post-Distillation Data 8
r: g " 1 E
2 2 E
:23 a ‘5“ 0 8 E
I: I! "" .
3 2 °
: o S
8 = a
a: g 10 - 1' 0.6 E
E w 5
= .s 9
° z 0.4 ‘g
a n.
at 5 «
~~ 0.2
0 r r r r r fl“ 0
0 2 4 6 8 1o 12 14
Distillation Time (min.)
Figure 6.7. The concentration behavior of ethyl acetate. The online

measurement is represented by the diamond plot and
corresponds to the left y-axis. The square plot represents
the behavior from post distillation measurements and

corresponds to the right y-axis.

78

(%vIv in 40% EthanolMlater Mixture)

Next methanol was examined. The plots for the behavior are shown in Figure 6.8.
The plot according to the post-distillation measurement, shows methanol decreasing
slowly towards the middle of the distillation and then increasing towards the end. The
plot according to online data shows just a straight increase as the distillation proceeds.
Actually it appears that the trends do agree except for the first point of online
measurement. The amounts again differ significantly between the two sets of
measurements.

Next, n-propanol was examined. The plot for both online and post-distillation
measurements is shown in Figure 6.9. The online analysis for n-propanol shows a much
different behavior compared to the normal GC analysis. According to the online
measurements, n-propanol starts out high and decreases steadily throughout the
distillation. This is different from the results from the conventional GC readings. It
starts out low and increases as the distillation proceeds and finally starts to decrease as
the amount of n-propanol decreases.

Finally, isoamyl alcohol was studied. The strange phenomena that happened with
n-propanol also occured here. According to the online plot, shown in Figure 6.10,
isoamyl alcohol starts out high and decreases throughout the distillation. The normal GC
plot, shown in Figure 6.10 shows it starting out low and increasing until the end where it
starts to decrease again. The values between online and post-distillation measurements
are very different as well.

On the whole the measurements taken with the online headspace apparatus

indicated different behavior compared to the analysis with the conventional GC.

79

Online Concentration
(%vlv in 40% Ethanol/Water Mixture)

Methanol Behavior

 

 

+ ' '
2 _ Online Analysus ., -. 0.49

+ Post-Distillation T
.. Analysis

 

 

 

“— 0.47

 

“ 0.45

 

is 0.43

 

~~ 0.41

 

 

 

 

O I I I T I I 0.39
DistillationTime (min.)

Figure 6.8 The concentration behavior of methanol. The online
measurement is represented by the diamond plot and
corresponds to the left y-axis. The square plot represents
the behavior from post distillation measurements and

corresponds to the right y-axis.

80

Post-Distillation Concentration
(%v/v in 40% Ethanol/Water Mixture)

N-Propanol Behavior

 

0.7 *

0.6 *

0.5 *

0.4 *

0.3 *

0.2 *

Online Concentration
(%vlv in 40% Ethanol/Water Mixture)

0.1 *

 

 

 

+ Online Analysis

*I- Post-Distillation
Analysis

 

 

 

 

I

.l.

 

 

Figure 6.9

I I I I

2 4 6 8

I I

10 12

Distillation Time (min.)

14

The concentration behavior of n-propanol. The online

measurement is represented by the diamond plot and

1.6

1.5

1.4

1.3

1.2

1.1

0.9

0.8

corresponds to the left y-axis. The square plot represents

the behavior from post distillation measurements and

corresponds to the right y-axis.

81

Post-Dlstillatlon Concentration
(%vlv in 40% Ethanol/Water Mixture)

Isoamyl Alcohol Behavior

 

 

 

 

 

 

 

 

 

 

1 g E +Online Analysis 1“ 2-4

’2‘ + Post-Distillation c g

é Analysis 5% S

.— 0 8 .4 E E
c 2 ' / " 1.9 a h
.9. as e a
g ‘5 o a
a 5 5 E
3 3 0.6 * u 3
c 5 c g
85 ”435
o l“ 1! m
C o =
e204* 25
o " o v

.5 .L o3 E

z «r-

3 0.98 g

g: 0.2 7 i .\°

0 . . . T .t . 0,4
0 2 4 6 8 1o 12 14
Distillation Time (min.)
Figure 6.10 The concentration behavior of isoamyl alcohol. The online

measurement is represented by the diamond plot and

corresponds to the left y-axis. The square plot represents

the behavior from post distillation measurements and

corresponds to the right y-axis.

82

One reason could be that this is the actual behavior of the vapor phase at the sampling
port above the condenser during the distillation. The distillation behavior itself is
inconsistent due to the size and construction of the Holstein still. Another reason could
be that the headspace sampler itself samples inconsistently since it relies on the pressure
inside the still to transfer the sample. As a result, the sample loop may not be collecting a
consistent amount of sample prior to injection, thus leading to sporadic readings from the
chromatograph.

Given the evidence so far, it would be more reasonable to conclude that it is the
inconsistency of the sampler that causes the behavior of the online readings. Looking at
the conventional GC measurements, there is little deviation in the error bars compared to
that of the online measurements. This would indicate a relatively constant distillation
behavior. Therefore, it is probably the headspace sampler that is causing the poor
repeatability. This will be studied further in the next section using simple distillation.

The fact that the amounts measured using the headspace sampler are different
from the amounts measured from the distillate in the conventional GC analysis also needs
to be examined. This fact could be explained using the McCabe-Thiele diagram. In the
sampling port of the still, there exists both liquid and vapor in equilibrium. Looking at
the McCabe-Thiele diagram it can be seen that the mole fraction of vapor does not equal
the mole fraction of liquid throughout most of the distillation process. So the amount of
distillate measured after the distillation is not going to correspond exactly to the vapor
sample measured at the top of the condenser. The vapor phase and liquid phase will be

related to each other according to their equilibrium properties, but it would take further

83

studies to determine this relationship in order to figure out actual amounts in the vapor

phase.

6.4 Simple Distillation

The headspace sampler was connected to a simple still to see if the behavior
measured from the online readings were due to inconsistencies in the still or from the
sampler itself. Three distillations were run with a solution similar to the ones made for
the previous experiments. GC acquisitions were performed approximately every 2.5
minutes and lOmL cuts were taken. Conventional GC measurements were also
performed for the sake of comparison.

Figure 6.11 shows the behavior of the ethanol concentration during the
distillations. Figures 6.12 through Figures 6.16 show the behavior of the congeners. Just
like the distillation with the IOL Holstein still, there is a difference between the
measurements made with the online headspace sampler setup and the post-distillation
conventional GC runs. The relative amount of congener and the behavior measured is
different between the online and conventional GC measurements. The most pronounced
differences in behavior between conventional GC measurements and online
measurements seem to appear in methanol in Figure 6.14, n-propanol in Figures 6.15 and,
and isoamyl alcohol in Figures 6.16.

The most interesting observation comes from looking at just the online
measurements and observing the error bars for the three runs. There seems to be a large

inconsistency in both the relative amounts and behaviors between the three runs,

84

Ethanol Concentration
("/ovlv In Ethanol/Water Mixture)

O)
O

 

(TI
0
1

A
O
1

()0
O
1

N
O
l

A
O
1

 

 

 

Figure 6.11.

5 10 15 20
Distillation Time (min.)
Ethanol concentration as a function of distillation time for the
simple distillations determined by conventional GC analysis.
Ethanol concentration is given in percent volume of ethanol

per volume of distillate.

85

Acetaldehyde Behavior

 

 

+ Online Analysis

2.5 - l 0.25
*I- Post-Distill Analysis

 

 

 

 

 

 

 

 

? A

'5 cg

E °3
:5 _ E:
725 2~l ~~o.2 g5.
I.“ a
a; 52
30 0'6
8&1'5q 770.1555
05 4 -5
,u. g...
.E.\° 3}!
7:" ~ —~ 2

i is

§ 0.5-" «0.05 a;

o 1 -o
0 5 10 15 20

Distillation Time (min.)

Figure 6.12 The concentration behavior of acetaldehyde during a simple
distillation. The online measurement is represented by the
diamond plot and corresponds to the left y-axis. The square
plot represents the behavior from post distillation

measurements and corresponds to the right y-axis.

86

Ethyl Acetate Behavior

 

1.6

1.4 ‘

Online Concentration
('/.vlv in 40% Ethanol/Water Mixture)

 

 

4>\ " 0.15

0.25

 

+ Online Analysis

+ Post-Distllation Analysis ~- 0.2

 

 

 

 

Post-Distillation Concentration
(%vlv In 40% EthanolIWater Mixture)

 

 

 

   

0.8 *
0.6 A *r 0.1
0.4 * ..
*~ 0.05
0.2 a
0 r r r “ O
0 5 10 15 20

Distillation Time (min.)

Figure 6.13 The concentration behavior of ethyl acetate during a simple

distillation. The online measurement is represented by the
diamond plot and corresponds to the left y-axis. The square
plot represents the behavior from post distillation

measurements and corresponds to the right y-axis.

87

Methanol Behavior

 

 

 

 

 

 

 

 

 

 

0.9 i 0.9
0.8 . +Oniine Analysis ._ 0.8 A
E . . . . e
i 0.7 . *I- Post-Distillation AnalySIS -. 0.7 g g
c E E
,3 E 0.6 * 8 '3
2 N c E
g a o.5~ 8 .
u g g c
1: n o “
8 5 0.4 ‘ g g
0 m .5. \-
§§03~ sé
c v ’ 9
o .5 a 5
g 0.2 ‘ a ;
.\- 1‘.
V 0.1 * *~ 0.1
O r r I O
0 5 10 15 20

Distillation Time (min.)

Figure 6.14 The concentration behavior of methanol during a simple
distillation. The online measurement is represented by the
diamond plot and corresponds to the left y-axis. The square
plot represents the behavior from post distillation

measurements and corresponds to the right y-axis.

88

Online Concentration
(%vlv in 40% Ethanol/Water Mixture)

N-propanol Behavior

 

 

 

 

 

 

 

 

 

 

 

 

 

1'4 .. 1.2
+ Online Analysis
1.2 * . _ . "
4* Post-Distlllation .. 1' -_ 1 E
1 _ Analysis 5’3
0 8 g E
«1— . c b
0.8 * 8 g
1' 5 E
O
0.6 “ . //\1> ._ 06 g E
E 111
.4 * 5 .
O ~~ 0.4 .2 is
g V
0 2 " u; E
i o
l \\ g 0.2 a. E
O '4 \i v
uh -L .-
-o.2 . , . o
0 5 10 15 20

Distillation Time (min.)

Figure 6.15 The concentration behavior of n-propanol during a simple
distillation. The online measurement is represented by the
diamond plot and corresponds to the left y-axis. The square
plot represents the behavior from post distillation

measurements and corresponds to the right y-axis.

89

Isoamyl Alcohol Behavior

 

1.5 a

0.5 *

Online Concentration
(°/ovlv in 40% EthanolMlater Mixture)

 

1.6

 

+ Online Analysis

 

 

-I- Post-Distillation Analysis

 

~~ 1.2

*- 0.8

** 0.6

 

*- 0.4

*~ 0.2

Post-Distillation Concentration
(%vlv in 40% Ethanol/Water Mixture)

 

 

 

 

 

Figure 6.16

f i -0.2
5 . 1O 15 20
Distillation Time (min.)

The concentration behavior of isoamyl alcohol during a
simple distillation. The online measurement is represented
by the diamond plot and corresponds to the left y-axis. The
square plot represents the behavior from post distillation

measurements and corresponds to the right y—axis.

90

especially for methanol, n-propanol, and isoamyl alcohol. This is not the case when
looking at the post-distillation measurements. The error bars for all compounds are
relatively smaller compared to the online measurements

The measurements with the conventional GC would indicate that the distillation
behavior for the simple distillation was fairly constant. Therefore, the inconsistency in
the online measurements is more likely a result of the variance in the sampling method.
So in order to improve consistency with the online measurements, it is recommended that

the focus be on improving the headspace sampling system.

91

7. Summary and Conclusions

The attempt to obtain online chromatographic analysis while monitoring a
distillation with the Chromatofast® GC AcceleratorTM was successful with respect to
actually achieving analyte separation from a still vapor sample. This method also
produced chromatograms in a time much faster than a conventional GC so that
measurements could be obtained every few minutes during the distillation. This makes
the technique a viable option for use in online monitoring.

Calibration proved to be difficult when measuring samples online. The ethanol
calibration curve did not give an increasing peak area with increasing ethanol
concentration. It was determined that the repeatability in the sampling of the
AcceleratorTM was reasonable so it most likely was not the cause of inconsistent data.
Rather it was more likely the pressure fluctuations in the standard sample vials causing
the problem. In addition, the calibration curve did not seem to correspond to the amounts
measured during distillation. This could be due to the inability to approximate still
conditions. It could also be due to the fact that there is actually less than 40%-80%
ethanol in the vapor phase at the top of the still. Quantification is something that still
needs to be studied further.

The behavior of the congeners was also measured using the online setup with the
GC AcceleratorTM. Trends were found for acetaldehyde, ethyl acetate, methanol, and n-
propanol. Isoamyl alcohol was not monitored in the online analysis due to its retention
time being much longer than 2.5 minutes.

Despite the success in obtaining online data, the poor structural integrity of the

Chromatofast® GC AcceleratorTM made the process of making repeated measurements

92

difficult. The use of liquid nitrogen allowed for the possibility of transfer line freezing.
In addition, the peak shapes quality worsened over time, most likely because of the
failure of the valves to open and shut properly. Finally the AcceleratorTM deteriorated to
the point where it would no longer collect data. Receiving technical assistance for
maintenance and repair on the machine was made almost impossible due to the failure of
Chromatofast® Inc. to stay in business. Therefore, using the GC AcceleratorTM as an
online tool is not practical.

This led to the development of the headspace sampling system. This sampling
system was able to extract vapor from the still and inject it directly into the GC. Fast
separation of all the major components was achieved.

The chromatograms from the headspace sampling system could be obtained in a
much shorter time than both the conventional GC and the GC AcceleratorTM method.
The headspace sampler also proved to be more rugged than the AcceleratorTM as
repeated measurements with the headspace sampler produced chromatograms of good
quality nearly every time. In addition, the cost of the headspace sampler was much more
affordable than the GC Accelerator“.

Despite the advantages, calibration was a problem for the headspace sampler as it
was for the previous method. Actually, it appeared to be worse than with the GC
AcceleratorTM. The ethanol calibration curve did not give increasing peak areas with
increasing concentration. In addition, the repeatability of the measurements was
extremely poor. These difficulties may be due to the inability to measure standards at

conditions that are close to the conditions occurring inside the still among other reasons.

93

As with, the AcceleratorTM, the behavior of acetaldehyde, ethyl acetate, methanol,
n-propanol, and isoamyl alcohol was studied with the headspace sampler online method.
In addition, the distillate was studied afterward with conventional GC for comparison.
The amounts and the behavior determined from the online measurements did not
correspond directly to those of the distillate measured with the conventional GC after
distillation. There were inconsistencies in behavior between the three runs for each
compound in the online analysis. However the post distillate analysis showed consistent
compound behavior between the three runs.

A study of simple distillations was done using the headspace sampling system to
compare to the results of the IOL distillation. Again, there was a difference in the
behavior and relative amounts of the compounds between the online analysis and the
conventional GC analysis. The online analysis also showed inconsistent behavior
between the three runs for each compound, while the distillate measurements with the GC
showed similar behavior between the three runs.

The reason for the differences in the behavior of the compounds between the
online measurements and the measurements taken of the distillate with conventional GC
is most likely inconsistent sampling by the headspace sampler. This probably explains
the poor repeatability in standardization and the inconsistency between the three runs in
distillation measurements as well. The difference in relative concentrations between
online vapor sampling and measurement of the distillate with conventional GC may be
explained as well. The concentrations of the compounds in the vapor are probably not
going to be the same as the concentrations in the liquid. So a relationship between the

vapor and the liquid needs to be studied in order to find actual concentrations in the

94

distillate. Through this, the online data will serve the desired purpose of allowing real-
time manipulation of the still to the change the composition of the distillate.

In conclusion, online measurement of a distillation using fast GC appears to be
feasible. Fast GC could be achieved with both of the methods studied in this thesis,
although the headspace sampling method proved to be more rugged and convenient than
the GC AcceleratorTM method. Behavior of the compounds chosen could be monitored,
however there was a problem with repeatability and finding absolute concentrations of
the compounds. With further study these problems could be overcome, therefore making
online monitoring with hi gh-speed gas chromatography a useful tool for monitoring

distillations.

95

8. Future Work

One of the major obstacles to overcome with the headspace sampling system is
the inconsistent sampling. As mentioned before, the headspace sampler relies strongly on
the pressure of the still or the headspace pressure of the sample vial. If the pressures are
inconsistent, then the amount of sample will most likely be inconsistent. A solution to
this problem might be to connect a pump between the still and the injection valve to
obtain a more consistent amount of sample. It might also be helpful to include some sort
of pressure regulator to maintain a constant pressure during sampling.

A reliable method for standardization should be found. Perhaps standards should
be placed in larger vials in order to better approximate the conditions of the still. In
addition, better quality vials should be used to prevent leaks and loss of analyte. Perhaps
vials should also be allowed to equilibrate longer when being heated to ensure that they
are heated consistently every time.

The other issue, which must be addressed, is the quantification of the compounds.
This is also related to the standardization. The inconsistency in sampling contributed
partly to this problem. However, there is also the fact that the concentrations of
compounds in the vapor probably do not correspond directly with the concentration of the
corresponding distillate. So it might be useful to look at the equilibrium behavior of the
compounds in question in order to determine the concentration relationship between the
vapor phase and the liquid phase.

It may be useful to look at alternatives to the headspace sampler as well. Solid

Phase Micro Extraction (SPME) might be a viable option. Using this method, it

might be easier to ensure that the volume injected into the GC is more consistent.

96

This could make it easier to develop calibration curves and obtain better quantitative
data. In addition, SPME concentrates samples, which could improve the detection of

trace compounds.

97

S

llllllllllllllllllll