3. {‘13,
3.... ‘5: .

. -
. 3 ..

Annual. ..

35“.... 0%”

3. .. .s c

r «‘51 .t .

.1“: Run» . ‘

.

2... Hi.

. km
3; , J.

. J. 23. .
)1 .s... .
human. “(an
. r1“: 1...:
. 1.53.".31... I

.l D Q 1.
.... .....: .muu(.
lulu“?

’

up

 

gggfiggnfig .9; $5”? 5%?

loll-II

45am~§¥¥§vr

gagafigfi . ...

 

" '0‘??? I“.

3004
g / 3 (I? .':J'

This is to certify that the
thesis entitled

REAL TIME FRUIT SPIRIT DISTILLATION ANALYSIS
WITH HIGH SPEED GAS CHROMATOGRAPHY

USING LIQUID PHASE SAMPLING

presented by

Robert Edward Boes Jr.

has been accepted towards fulfillment
of the requirements for the

MS. degree in Chemistl

 

 

‘ ,./"\\ ‘
x. _ _- ; K I A. i‘ M 2 (w .t ’ '
‘ 7 ~ I‘ . l . C— 4_ ”‘5 a k “\_ -‘f 1 ‘ 9—. ~ (H ,
]

 

Major Professor’s Signjmure I

5““ AA /\ \/ 2 ( v ff,"

 

Date

MS U is an Affirmative Action/Equal Opportunity Institution

LIBRARY
Michigan State
University

 

PLACE IN RETURN Box to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECAILED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 cJCTfic/omeoue.indd-p.1s

REAL TIME FRUIT SPIRIT DISTILLATION ANALYSIS
WITH HIGH SPEED GAS CHROMATOGRAPHY
USING LIQUID PHASE SAMPLING

By

Robert Edward Boes Jr.

A THESIS

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

MASTER OF SCIENCE
Department of Chemistry

2005

Abstract

REAL TIME FRUIT SPIRIT DISTILLATION ANALYSIS
WITH HIGH SPEED GAS CHROMATOGRAPHY
USING LIQUID PHASE SAMPLING

By

Robert Edward Boes Jr.

Fruit brandy distillation is a growing industry. Several different spirit distillation
techniques have been performed to improve both production and quality. Recent
advances in high speed gas chromatography have made online analysis of distillation
possible. Over the last couple years a few methods have surfaced that have provided a
good starting point from which to continue with research. Two of these setups were
thoroughly studied resulting in the development of a new high speed gas chromatography
method. The new “liquid-phase sampler” method expanded upon the old setups, still
relying heavily on prOper column selectivity and temperature programming for maximum
compound separation. The new process introduced a peristaltic pump allowing for
constant sample flow, a specially designed liquid-phase sampler for continued sampling,
and the use of a GC auto injector for accurate sample injections. Synthetic brandy
solutions were used to analyze the new setup’s ruggedness and reproducibility. Several
distillations were performed examining changes in temperature, pressure, flow rate, and
concentration as a function of distillation volume and time. The behavior of several
compounds was analyzed in real time with the new method. The liquid-phase sampler
setup proved to be cheaper, less complex, and a very efficient way of monitoring fruit

brandy distillations with high speed gas chromatography.

Table of Contents

List of Tables ....................................................................................................... v.
List of Figures ...................................................................................................... vi.
1. Introduction .................................................................................................... 1
1.1 Fruit Brandy Distillation ............................................................... 1
1.2 Structure of a Batch Still ............................................................... 2
1.3 Composition of Fruit Spirits ......................................................... 4
1.4 Quality Control and Analysis ........................................................ 7
1.5 Objective ....................................................................................... 8
1.6 Literature Cited ............................................................................. 9
2. High Speed Gas Chromatography ................................................................. 10
2.1 Background and Development ...................................................... 10
2.2 Speed and Separation .................................................................... 11
2.3 GC Columns .................................................................................. 13
2.4 Injector Considerations ................................................................. 14
2.5 Temperature Programming ........................................................... 15
2.6 Pressure Considerations ................................................................ 15
2.7 Carrier Gas F low Rate .................................................................. 16
2.8 Detectors ....................................................................................... 1 6
2.9 Literature Cited ............................................................................. 17
3. Discussion of High Speed GC Setup ............................................................. 18
3.1 Previous Setups and Results ......................................................... 18
3.2 Liquid-Phase Sampler Setup ......................................................... 20
3.3 Literature Cited ............................................................................. 26
4. Materials and Methods ................................................................................... 27
4.1 Spirits Production Process ........................................................... 27
4.1.1 Creation of Synthetic Solutions ............................... 27
4.1.2 Distillation ................................................................ 27
4.2 Pump Calibration ......................................................................... 30
4.3 Analysis ........................................................................................ 31
4.3.1 Gas Chromatograph Setup ....................................... 31
4.3.2 Top Tray - Online Analysis ..................................... 36
4.3.3 Post Distillation Analysis ......................................... 37
4.4 Literature Cited ............................................................................ 38

iii

Modeling the Liquid-Phase Sampler .............................................................. 39

5.0 Ideal Reactors ............................................................................... 39
5.1 Residence Time Distribution of Ideal Reactors ........................... 41
5.2 Residence Time Distribution of Liquid-Phase Sampler .............. 43
5.3 Literature Cited ............................................................................ 50
Results: Liquid-Phase Sampler ...................................................................... 51
6.1 Still Reproducibility ..................................................................... 5 1
6.2 Method Optimization ................................................................... 57
6.3 Standardization ............................................................................ 57
6.4 Distillation Behavior .................................................................... 69
6.5 Literature Cited ............................................................................ 78
Summary and Conclusions ........................................................................... 79
Future Work .................................................................................................. 84

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 the liquid-

phase sampler experiment .................................................................. 37

Pump calibration results for the two main settings used

during the experiment ........................................................................ 41

Gas chromatograph temperature analysis parameters for the

liquid-phase sampler setup ................................................................. 42

Gas chromatograph pressure analysis parameters for the

liquid-phase sampler setup ................................................................. 43

Gas chromatograph auto injector analysis parameters for

the liquid-phase sampler setup ........................................................... 44

Mean residence time results for the liquid-phase sampler.

Two different flow-rates were used during the experiment ............... 56

Retention times for chromatograms taken using the liquid-
phase sampler. Peaks were identified by comparing the
retention times to standard retention time for each of the

compounds. ........................................................................................ 68

Figure 1.1.

Figure 1.2.

Figure 3.1.

Figure 3.2.

Figure 3.3.

Figure 3.4.

Figure 4.1.

Figure 5.1.

List of Figures

A 150 L Christian Carl Batch sun? .....................................

Milligrams of volatile congeners in distilled spirits per 100
milliliters of alcohol. Fruit spirits have higher congener

concentrations than other distilled beverages”. ....................

Schematic of the previously used headspace sampling setup

connected to the Shimadzu GC-17A1. . .................................

A round plate made of plexiglas was created to replace the
glass plate in the top tray of the still. A small hole was
drilled in the center of the plate to allow for the transfer line
to run through it. A similar plexiglas plate was also made

for the bottom to allow for the sample to return to the still. .

Schematic of the liquid-phase sampler. Liquid flows into
the vial fi'om the bottom inlet filling up the vial. The liquid
then exits through the top outlet to keep a constant flow
running through the vial. ........................................................

Schematic of the liquid-phase sampler setup connected to
the Simadzu GC-17A auto injector. Liquid sample travels
from the top tray through a peristaltic pump and on to the

GC. The unused sample returns to the still. ..........................
Schematic of a 10L Arnold Holstein Still]. ...........................

Two diagrams of ideal reactors; the plug flow reactor and
the continuous stirred tank reactor. The liquid-phase

sampler diagram is also shown in comparison. ....................

vi

............ 3

............ 5

............ 19

............ 21

............ 23

............ 24

............ 29

............ 4O

Figure 5.2.

Figure 5.3.

Figure 5.4.

Figure 5.5.

Figure 6.1.

Figure 6.2.

Residence time distribution of a plug flow reactor, resulting
from a step tracer input. The tracer output signal (F) is zero
from the point of injection to the mean residence time (t).
At this point the reactor is totally clear out and nothing but

the tracer remains in the reactor (F=1)2. . .......................................... 42

Residence time distribution of an ideal continuous stirred
tank reactor resulting fi'om a step tracer input. The tracer
output signal increases in an exponential form until nothing

but the tracer remains in the ractorz. .................................................. 44

The relationship between concentration of ethanol and the
total volume exiting the liquid-phase sampler. Sample cuts
of 0.25 ml were taken and analyzed by the GC. After about
5 ml of sample has been collected the concentration of

ethanol is approximately zero. ........................................................... 45

The residence time distribution of the liquid-phase sampler
plotted against the plug flow and continuous stirred tank
reactor models at a mean residence time of 20.12 seconds.
The liquid-phase sampler analysis is represented by the
round plot. The CSTR analysis is represented by the
square plot. The PFR analysis is represented by the
diamond plot. The liquid—phase sampler appears to behave

somewhere in between the two ideal models. ................................... 48

Pressure of the bottom pot on the still as a function of
distillation volume. Pressure increases in a linear

relationship with volume ................................................................... 52

Temperature as a function of distillation volume on the
small still. The temperature levels off during the middle of

the distillation with sharp increases both initially and

vii

Figure 6.3.

Figure 6.4.

Figure 6.5.

Figure 6.6.

Figure 6.7.

Figure 6.8.

Figure 6.9.

towards the end portion of the run. This is characteristic of

a third order polynomial relationship. ............................................... 53

Time of distillate cuts as a function of distillation volume.
75 milliliter cuts were taken and the time recorded. The

data appears to increase in an exponential form. ............................... 55

Flow rate of the distillate as a function of distillation
volume. Flow rate has the characteristics of a decreasing in

a linear relationship with respect to distillation volume. .................. 56

Chromatogram for the distillation of a brandy-type
mixture. The sample was collected fi'om the top tray of the
still using the liquid phase sampler. The programs used
were the same as in Tables 4.4 and 4.5. Peaks were
identified by comparing the retention times from the
analysis to standard retention times for each of the

compounds. ........................................................................................ 58

The standard calibration curves for ethanol. The static
measurement is represented by the diamond plot and the

flow measurement is represented by the square plot. ........................ 61

The standard calibration curves for acetaldehyde. The
static measurement is represented by the diamond plot and

the flow measurement is represented by the square plot. .................. 63

The standard calibration curves for ethyl acetate. The static
measurement is represented by the diamond plot and the

flow measurement is represented by the square plot. ........................ 64

The standard calibration curves for methanol. The static

measurement is represented by the diamond plot and the

viii

Figure 6.10.

Figure 6.11.

Figure 6.12.

Figure 6.13.

Figure 6.14.

flow measurement is represented by the square plot. ........................ 65
The standard calibration curves for n-propanol. The static
measurement is represented by the diamond plot and the

flow measurement is represented by the square plot. ........................ 67

The standard calibration curves for isoamyl alcohol. The
static measurement is represented by the diamond plot and

the flow measurement is represented by the square plot. .................. 68

Ethanol concentration as a fimction of distillation time.
Data was taken from the top tray with the liquid-phase
sampler and also after distillation. Standard ethanol
calibration curves were used for each analysis. The online
measurement is represented by the square plot and the post
distillation measurement is represented by the diamond

plot. ............. 70

Acetaldehyde concentration as a function of distillation
time. Data was taken from the top tray with the liquid-
phase sampler and also after distillation. Standard
acetaldehyde calibration curves were used for each
analysis. The online measurement is represented by the
square plot and the post distillation measurement is

represented by the diamond plot ........................................................ 72

Ethyl acetate concentration as a function of distillation
time. Data was taken from the top tray with the liquid-
phase sampler and also afier distillation. Standard ethyl
acetate calibration curves were used for each analysis. The
online measurement is represented by the square plot and
the post distillation measurement is represented by the

diamond

ix

Figure 6.15.

Figure 6.16.

Figure 6.17.

plot. ................................................................................................... 73
Methanol concentration as a function of distillation time.

Data was taken from the top tray with the liquid-phase

sampler and also after distillation. Standard methanol

calibration curves were used for each analysis. The online
measurement is represented by the square plot and the post
distillation measurement is represented by the diamond

plot. .................................................................................................... 74

N-propanol concentration as a function of distillation time.
Data was taken from the top tray with the liquid-phase
sampler and also after distillation. Standard n-propanol
calibration curves were used for each analysis. The online
measurement is represented by the square plot and the post
distillation measurement is represented by the diamond

plot. .................................................................................................... 75

Isoamyl alcohol concentration as a function of distillation
time. Data was taken from the top tray with the liquid-
phase sampler and also after distillation. Standard isoamyl
alcohol calibration curves were used for each analysis. The
online measurement is represented by the square plot and
the post distillation measurement is represented by the

diamond plot. ..................................................................................... 76

1. Introduction

1.1 Fruit Brandy Distillation

Many challenges are encountered 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‘u. Therefore, much research needs to be done on several aspects of fruit
brandy production in order to ensure the quality of spirits for consumption4.

Brandy is the product obtained from the distillation of wine or fermented fruit
mash]. Distilling fruit spirits involves using one of two processes. First, there is
distillation without rectification where vapors produced by boiling are condensed and do
not return to the stills. This means that there is little or no reflux for the enrichment of the
vapor in the still. This French or alambic style of brandy making requires multiple
distillations to obtain high proof spirits]. Second, there is distillation with rectification
(the use of trays and partial condensers) where part of the condensate returns to the still
and comes into contact with the rising vapors. These points of contact are known as
stages or trays and they help to make the vapor richer in the higher boiling component5 .

It is known that the greater number of trays or stages used, the more separation can be
achieved”. This German style of distillation is designed to capture the essences of the
fruit and requires only a single pass through the still to obtain high proof spin'ts'. The

German batch distillation process will be examined in this thesis.

1.2 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 1.37. There are varying
designs of these stills in terms of the number and placement of the trays, however, this
figure is a good indication of what is typical for brandy stills. Although design opinions
vary, it 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 contains four trays, one over the pot, and three in the
column next to the still4. 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 a means for removing cyanide containing compounds from the distillate7. 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

 

 

Steam Jacket
Partial Condensers
Total Condenser
Copper Packing

 

- - -Distillation Vapor Path
—Cooling Water Path

 

A

 

 

Fermented Mash

 

 

 

 

 

 

 

lnlet

 

 

 

Figure 1.1

.........‘

 

I
i
i
i

 

 

 

Tray Control
Levers

Cooling Water

Control

    

    
 

To Drain

A 150 L Christian Carl Batch Still7.

“35,9.-._._._._._._._..1.

-‘lll-‘U----V‘"'C’F

~‘\.~\.-W-sw--i-t-\-~-...-;-._---.__

  
  
 

  

 
  

  

displayed 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 distillation4.

1.3 Composition of 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 (components other than ethanol and water) present in varying
amounts. In fruit 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”. These congeners 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 next4.

Congener Amount (mgl100mL)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I Methanol j
2000 « .
El Higher alcohols
1800 .
, I Carbonyl compounds
1600 - ‘ I Ethyl acetate
1400 . A ~ . DOther esters _j
1200 . I
1000 ~ 1
800 . .
600 I .. j j
400 I l
200 ' H I
It i

 

0° 0% Q Q69” 0 3% Q 0040 r0960 09600 >9? “‘9 A
c)“ \e 0" vi 0° 5
(0° 4‘ o 00 {9 08
Q 0060 Q80 9° \0
69 .5?
(P
9

Figure 1.2 Milligrams of volatile congeners in distilled spirits per 100
milliliters of alcohol. Fruit spirits have higher congener

concentrations than other distilled beverages”.

 

There are various carbonyls, acids, and fusel 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,
such as acetone, aldehydes such as acetaldehyde and benzaldehyde. 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 alcohol4.

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.
Methanol in fruit spirits has been regulated in Europe from a range of alcohol depending
on the variety of fruit spirit being produced. According to the Bureau of Alcohol,
Tobacco, and F irearrns, the limit is 0.35%v/v for all fruit spirits, which translates to
700mg/100mL absolute alcoholz.

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].

1.4 Quality Control and Analysis

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. The concentrations of congeners can be controlled during distillation
through fractionation of the distillate through the use of trays, the catalytic converter, and
partial condenser‘. 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 (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 after 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 minutesl 1. With this technique,

it is possible to collect a sample directly from a still and analyze it in a fast enough

manner that adjustments could be made to control quality during a distillation4.

1.5 Objective

The goal of this study was 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 was demonstrated. It was also important to examine the speed and
efficiency of the actual chromatographic separation. Finally, the method was applied to

an actual distillation in order to monitor compounds as they were volatilized“.

1.6

10.

11.

Literature Cited

Tanner, H. Brunner, H.R., Fruit Distillation Today 3'“ 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/a1ctob _pub/cm98-1.pdt>.

Methanol, Material Safety and Data Sheet.

Lindemann, D. “Real Time Monitoring of Fruit Spirit Distillation with High
Speed Gas Chromatography”. 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.

Claus, M. “An Investigation of the Relationship Between Tray Usage of the Still
and Congener Compound Concentration in Distilled Fruit Spirits”. MS. Thesis.

Piggot, J.R., Patterson, A. ed., Distilled Beverage Flavour: Recent Developments.
Ellis Horwood Ltd., 1989.

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.

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

2. High Speed Gas Chromatography

2.1 Background and Development

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“.

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
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

10

stationary phases and temperature or pressure programming 2. Obviously, this leads to an
increase in analysis time4.

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

chromatography“.

2.2 Speed and Separation
Unfortunately, everything comes at a cost and high speed GC is no exception.

The measures taken for decreasing analysis time ofien decrease sample capacity and

11

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 hand4.

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 option5 . 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
systems. 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 resolutions. 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 injectors4.

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

12

to look at every aspect of the GC instrumentation, from injection, to columns, to

detectors, to temperature and pressure programming, and carrier gas flow4.

2.3 GC Columns

Two simple methods for reducing the separation time are to shorten the length of
the column and to decrease the stationary phase thickness3'5’6. 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 carefirl not to lose desired resolution in
changing length or thickness4.

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
and faster flow rates while maintaining resolution4.

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’3'6. 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

13

development is also difficult, especially in quality control7. However, it is becoming
more convenient to use this method given recent technological advances“.

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 mixtures4. Loss of efficiency
due to shortening columns can be compensated by carefiilly choosing selectivity in the
stationary phaseé. 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 columnsg. 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 oflen useful to minimize the injection

bandwidth whenever possibles. This will limit the amount of band broadening during

14

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
adequate separation4. 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 millisecondss.

2.5 Temperature Programming

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

well4.

2.6 Pressure Considerations

Pressure is also an important factor in high speed gas chromatography. For
example, the use of narrow bore columns requires high inlet pressures in order to
operate3’5’6. Pressure programming has even been studied as an alternative to temperature

programming in cases where the stationary phase or sample is thermally labile. With

15

pressure programming steeper gradients could be achieved and there is a more rapid
return to initial conditionsg. The newest gas chromatographs can be equipped with
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 usedz’s. 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

5

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

This can make hydrogen gas a viable option4.

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 widths3'5'7. 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“.

l6

2.9

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.” J. Chromatogr. 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.

Lindemann, D. “Real Time Monitoring of Fruit Spirit Distillation with High
Speed Gas Chromatography”. MS. Thesis.

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. Chromatogr. 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.

Peters, A., Sacks, R., “Inlet Pressure Programming for High-Speed Gas
Chromatography.” J. Chromatogr. Sci. 30 (1992) 643-648.

17

3. Experiment: Discussion of the Liquid-Phase Sampler Setup

3.1 Previous Setups & Results

A schematic of the previous headspace sampling setup connected to the still and
GC is shown in Figure 3.1. This instrument consisted of an injection valve assembly,
heated transfer line, and purge and trap adapter. Carrier gas supplied a constant flow to
both the headspace sampler assembly and the GC. The headspace assembly was
connected to the still through the sampling port just above the condenser. The sample
traveled from the sampling port, to the sample loop in the injection valve, and through a
heated transfer line. The purge and trap adapter connected this transfer line to the gas
chromatograph where the sample traveled through the column and detector to be
analyzed].

This setup required that work be done in the vapor/liquid phase. This created
several problems. First, an inconsistent volume of sample was delivered to the CG. The
pressure of the still was relied upon to extract a sample volume. With this setup it was
very important that a fixed volume be injected into the chromatograph every time for
precise analysis of compound retention times and peak areas. Even if a pump had been
present, drawing a consistent vapor/liquid sample would still have had its challenges.
Second, in order to keep the sample in the vapor phase a heated transfer line was needed.
This line had to be kept at a high enough temperature to assure the gaseous phase was
maintained. Pressure would also have to be controlled. While this was definitely able to
be done it was not the most effective means of going about the experiment. Third, all the
required items to make the process work properly made the entire setup very complex

and bulky. This made the experiment very hard to move around from one place to

18

Actuator Heated Transfer Line Computer

Valve

 

 

 

 

 

 

 

 

 

 

 

 

 

III
..-\.
III
l.:'-
“-1
I..\
III -'*-

 

 

 

 

 

 

 

 

 

 

 

 

 

 

   
    

 

 

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 3.1 Schematic of the previously used headspace sampling

setup connected to the Shimadzu GC-17Al.

l9

another. These three main factors contributed in large part to the unsuccessfiilness of this
setup. It was clear that a new, more efficient method would work much better then any

of the current renditions.

3.2 Liquid-Phase Sampler Setup

The new liquid-phase sampler setup involved replacing the current glass plate on
the top tray of the column with a newly cut plexiglas plate. The plate measured 66mm in
diameter with a depth of 5mm. This plate was needed so that a hole could easily be
drilled through the material allowing for the fluid transfer line to pass into the top tray.
See Figure 3.2. A small enough hole was drilled in the center of the plate (4mm in
diameter) to create a good seal with the transfer line. Another identical plate was created
to allow for the return sample to be placed back into the bottom of the still.

The transfer line ran from the top tray of the still, through a peristaltic pump, into
the liquid-phase sampler, and then back to the still. The line was made of platinum cured
silcon tubing. The tubing was purchased from VWR Industries. It had an inner diameter
of 0.79mm and an outer diameter of 3.99mm. The total length of the tubing was
122.3mm. This silcon tubing provided both flexibility and a wide temperature range on
which to transfer fluids.

A variable flow peristaltic pump was used to remove fluid from the top tray of the
column and deposit it into the liquid-phase sampler to be analyzed by the GC. The
peristaltic pump used for the experiment was purchased from VWR for $169.77. Its

dimensions were 16.8cm L x 12cm H x 11.5cm D and weighed approximately 0.57kg

20

 

 

10 L Holstein Still Liquid Transfer Line

Figure 3.2

Plexiglas Plate

A round plate made of plexiglas was created to replace
the glass plate in the top tray of the still. A small hole
was drilled in the center of the plate to allow for the
transfer line to run through it. A similar plexiglas plate
was also made for the bottom to allow for sample to

return to the still.

21

It consisted of 3 plastic rollers to reduce flow pulsation, prevent siphoning, and eliminate
the need for a check valve. Depending on the tubing size it was able to produce flow-
rates from 0.4 to 85.0 ml/min. It could support a temperature range of -62 to 260C. The
pump provided a constant flow of sample through the liquid-phase sampler.

The liquid-phase sampler is the key to the entire setup and is the main reason the
experiment is successful. The liquid phase sampler’s design was based off of a regular
GC auto injector vial purchased from National Scientific Company. The dimensions of
the clear glass vial were 12mm x 32mm. It held approximately 2 milliliters in volume.
The vial was modified by a glass shop to create two small holes in its side. This is seen
in Figure 3.3. The first hole (the inlet) was made as close to the bottom of the vial as
possible so that minimal sample residence time would result. The second hole (the
outlet) was created about halfway up the vial. This was so that enough liquid would be
present in the vial for the syringe to take up during analysis. Both an inlet and outlet
were needed to make sure that a constant flow of liquid would continuously run through
the vial.

In order to deliver a consistent sample to the GC the AOC-20i Shimadzu Auto
Injector was used. It consisted of a 10 micro-liter syringe, a sample holder vial rack, and
a computer program to supply the parameters. The vial rack had to be modified to hold
the liquid-phase sampler properly. A small piece was cut out of the side of the holder
rack which allowed the liquid phase sampler to sit upright and provide sturdiness during

the experiment. The auto injector provided the best method of sample injection into the

GC.

22

Auto Injector

Syringe

Liquid

GC vial

 

\Xj/

 

 

 

Flow Outlet

Liquid

 

 

 

 

 

 

Flow Inlet

Figure 3.3

 

 

 

 

i

C

 

<9

Schematic of the liquid-phase sampler. Liquid flows
into the vial from the bottom inlet filling up the vial.
The fluid then exits through the top outlet to keep a

constant flow running through the vial.

23

Peristaltic Pump Computer

\IIL

l
I |—_ - [:3
Shimadzu
J : GC-l 7A
__l

Auto Injector

 

 

 

 

 

 

 

E_.

10 L Holstein Still

 

 

 

 

— -- — Liquid Transfer Path from Still to GC

— — — Return Liquid Transfer Path from GC to Still

 

 

 

Figure 3.4 Schematic of the liquid-phase sampler setup connected
to the Shimadzu GC-17A auto injector. Liquid sample
travels from top tray through a peristaltic pump and on

to the GC. The unused sample returns to the still.

The entire liquid-phase sampler setup can be seen in Figure 3:4. The liquid
sample travels from the top tray of the column, to the peristaltic pump, and then through
the liquid-phase sampler where it can be taken up by the GC auto-inj ector syringe. The
unused sample travels from the liquid-phase sampler back to the still where it is
redistilled. This whole process is a continuous cycle run throughout the experiment. GC
analysis is conducted in the same way is if there were a normal static sample being

examined.

25

3.3 Literature Cited

1. Lindemann, D. “Real Time Monitoring of Fruit Spirit Distillation with High
Speed Gas Chromatography”. 2003 MS. Thesis.

26

4. Materials and Methods

4.1 Spirits Production Process
4.1.1 Creation of Synthetic Solutions

Actual fermentations often take weeks to process, limiting the total number of
distillations that can be performed. In order to run multiple batches within a given time
period, standard solutions of mixtures were made. These solutions contained similar
amounts of congeners present as in most fermentations. Solutions were made in 8 liter
batches, enough to make a single distillation. As soon as the solution was made it was
used for the distillation. This was to avoid any evaporation or possible compound
formation. The standard solutions consisted mainly of ethanol and water. With hundreds
of compounds present in fruit spirits it would be difficult to be completely accurate in
analysis. Therefore only ethanol and the five main congeners were examined. The
congeners used and their volume percentages are listed in Table 4.1. For the liquid-phase
sampler experiment the congeners chosen were acetaldehyde, ethyl acetate, methanol, n-

propanol, and isoamyl alcoholl.

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

27

 

 

 

 

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 4.1 Congener types and volume percentages used in the

liquid-phase sampler experiment'.

28

 

 

 

‘ Distillation Vapor Path

— Cooling Water In

- - Cooling Water Out

 

 

 

 

 

r. S
wrava.
nwr
aumr
kndw.
COD

aCOC
che
a l
mfiflb
wrath
taou
SPTB

 

 

 

 

 

 

 

:fidfififififififififififijfififififififi

 

 

 

 

 

 

 

 

 

 

 

 

Schematic of a 10L Arnold Holstein Stiu'.

Figure 4.1

29

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 milliter cuts were

made throughout the distillation].

4.2 Pump Calibration

The peristaltic pump was not self calibrating and the flow rates depended upon
the size of tubing used in the experiment. Two main settings were used during the
experiment, High 6 and Low 6. The pump was calibrated to determine the flow rates at

both these settings. For the Low 6 setting a 50mL graduated cylinder was used. The

30

cylinder was filled to the 50ml mark and the time was recorded. Six separate trials were
done and the times averaged. The Low 6 setting was determined to have a flow rate of
3.73 mein. In the same way the High 6 setting was determined, only a lOOOmL
graduated cylinder was used instead of the 100mL. The flow rate at this setting was

found to be 9.77m1/min. Table 4.2 displays the overall results.

4.3 Analysis
4. 3.1 Gas Chromatograph Setup

Online analyses were performed using a Shimadzu GC-17A gas chromatograph
with flame ionization detectionl. The GC had both a tunable pressure and split injection
analysis. All temperature, pressure, and GC parameters were entered through the
computer’s analysis software then downloaded to the GC. Everything was directly
controlled through the computer.

For the liquid-phase sampler experiment, 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 1pm. The temperatures of the injector and detector were 240°C and 255°C,
respectively, and remained constant throughout the analysis]. The conditions of the
column were set up for the experiment. The GC temperature and pressure analysis
parameters are shown in Tables 4.3 and 4.4.

The GC AOC-20i Shimadzu Auto Injector was used to inject a sample into the
GC instead of manually injecting with a syringe. The auto-injector was controlled

through the GC and computer software program. The auto injector parameters are shown

in Table 4.5.

31

 

Low 6 Setting High 6 Setting

 

 

Graduated Cylinder 50 ml 1000 ml
Average Time 804 sec 6144 see
Average Flow Rate 3.73 ml/min 9.77ml/min

 

Table 4.2 Pump calibration results for the two main settings used

during the experiment.

32

 

 

Injector Temperature 240°C

 

 

 

 

 

 

 

 

 

 

 

 

Detector Temperature 255°C

Max Temperature 399°C

Equilibrium Time 0.3 minutes

Column Tern erature Pro ram 400C held 0'4 minutes

p g Ramp 70°C /min to 115°C

Temperature Run Time 1.5 minutes

Recovery (Cooling) Time 1.72 minutes

Detector Analysis Time 2.0 minutes

Total GC Run Time 3.22 minutes

Table 4.3 Gas chromatography temperature analysis parameters

for the liquid-phase sampler setup.

33

 

 

 

 

 

 

 

 

 

 

 

 

 

Injector Mode Split

Gas Helium @ 600 kPa
Column Pressure Program 157 kPa Isobaric
GC Run Time 1.5 minutes
Set P Pressure 100 kPa

Set P Flow 3.0 ml/min
Septum Purge 5.6 ml/min
Oven Temperature 40 C
Velocity 107 cm/sec
Column Flow 5.8 ml/min
Spilt Ratio 34:1

Split Flow 198 ml/min

 

 

 

Table 4.4 Gas chromatography pressure analysis parameters for

the liquid-phase sampler setup.

34

 

Dwell Time

0.3 seconds

 

 

 

 

 

 

 

 

 

 

 

Number Rinses w/Sample 2
Number Rinses w/Solvent 2
Viscosity Composition Time 1 second
Number Syringe Pumps 3
Injection Speed Fast
Plunger Suction Speed Low
Injection Type Traditional
Syringe Injection Speed High
Sample Volume Used 0.5 pl
Needle Height From Vial Bottom 3.5 mm

 

 

Table 4.5 Gas chromatography Auto Injector analysis parameters

for the liquid-phase sampler setup.

 

4. 3.2 Top Tray — Online Analysis

Acquisition with the liquid-phase sampler setup began by turning on the all three
gas cylinders (helium, hydrogen, air) for the GC. The GC and auto injector were turned
on. The temperature and pressure parameters were downloaded to the GC. Right before
liquid started forming in the top tray of the still the pump was turned on. The pump was
turned to its highest speed setting, in order to fill the line (prime it) with sample from the
top tray. As soon as liquid started entering the liquid-phase sampler the pump was placed
on the High 6 setting and the analysis program was started on the computer. The auto
injector proceeded to operate and inject a sample from the liquid-phase sampler into the
GC.

Separation was best with a specific temperature program for this method of
sampling. The temperature program began as soon as injection took place. The initial
temperature of the oven was 40°C and this was held for 0.4 minutes. Next the
temperature was ramped 70°C/minute to 115°C. The temperature program time was 1.5
minutes. In order to return the oven back to its initial conditions a cooling period was
needed. The fastest way to achieve this was by opening the column door to the GC until
the oven temperature had reached 40°C. The experiment was set up so that as soon as the
oven reached 40°C a new sample was injected. This was adjusted by increasing the
detector analysis time to 2.0 minutes, and by adjusting several of the auto injector
parameters. The total GC run time was 3.22 minutes. Or in other words, every 3 minutes

13 seconds a new sample could be injected into the GC.

36

4.3.3 Post Distillation Analysis

75 milliliter cuts of distillate were taken from the still. As soon as distillate
started appearing from the bottom of the condenser a stop watch was started to keep track
of the times of the cuts. At each out several items were recorded. The time of each cut
was recorded. The pressure of the still pot was recorded. The temperature above the
total condenser was recorded. All these items were recorded throughout the experiment.
Following distillation, samples of each 75 milliliter cut were placed into vials and then
analyzed with the GC. The same temperature/pressure programs were used as that for the
top tray - online analysis. Compounds were identified by their relative retention times
during the runs. The retention times for the compound along with boiling points will be
discussed in the results]. The concentrations of the six main congeners were measured

using individual calibration curves.

37

4.4 Literature Cited

1. Lindemann, D. “Real Time Monitoring of Fruit Spirit Distillation with High
Speed Gas Chromatography”. 2003 MS. Thesis.

38

5. Modeling the Liquid-Phase Sampler

5.1 Ideal Reactors

The liquid—phase sampler was an important part of this experiment. All the
data sampling and analysis was dependant upon the continuous flow of liquid through
this vial. Understanding the fluid velocity distribution inside the vessel during the
experiment would be very beneficial. This would allow us to relate the liquid that is
being pumped into the vial with the sample that is actually taken up by the syringe
and then injected into the GC. The behavior of the liquid—phase sampler can be
predicted by modeling it after an ideal reactor. Two types of commonly used
industrial reactors are the PF R and the CSTR.

A PFR, or Plug-F low Reactor, is a tubular reactor. It typically consists of a
long cylindrical pipe and is normally operated at steady state. The reactants are fed
into one end and the products are withdrawn at the other end. The reactants are
continually consumed as they flow through the reactor. Thus in modeling the tubular
reactor, we assume that the concentration varies continuously in the axial direction
through the reactor].

A CSTR, or Continuous-Stirred Tank Reactor, is normally run at steady state
by feeding the reactants into the reactor and continuously withdrawing the products.
The CSTR is generally modeled as having no spatial variations in concentration,
temperature, or reaction rate throughout the vessel. Since the temperature and
concentration are identical everywhere within the vessel, they are also the same at the
point of discharge from the vessel]. The two different types of ideal reactors along

with the liquid-phase sampler model used in the experiment can be seen in Figure 4.1.

39

Plug Flow Reactor

Input >6 Output

Mixed Flow CSTR

4.,

‘\

W—a Output

 

 

Input

 

 

Liquid-Phase Sampler

\/\/\/

Output <—--— §
g

b

 

 

 

Input

Figure 5.1 Two diagrams of ideal reactors; the plug flow reactor
and the continuous stirred tank reactor. The liquid-

phase sampler diagram is also shown in comparison.

40

5.2 Residence Time Distribution of Ideal Reactors

The residence-time distribution of a reactor is a characteristic of the mixing
that occurs in the chemical reactor. In an ideal plug-flow reactor, all the atoms of
material leaving the reactor have been inside it for exactly the same amount of time.
An atom of material entering a PFR will travel from the inlet to the outlet in a period
of time equal to the mean residence time. For a steady-state, constant density system;
the mean residence time (t) can be described as the total volume of the reactor (V)
divided by the total flow-rate (0)1. The residence-time distribution of a PF R due to a
step tracer input can be seen in Figure 4.2. A step tracer is one in which the entire
contents of the inlet flow line are changed permanently. The tracer output signal (F)
for the PFR is zero from the point of injection to the mean residence time (t). At this
point the reactor is totally cleared out and nothing but the tracer remains in the reactor
(F = 1). Any input tracer, or stimuli, injected at the inlet will exit in exactly the same
manner after a lag time defined as t = Wu. The PFR is one class of reactors in which
all the atoms have the same residence time'.

In other reactor types, the various atoms spend different times inside the
reactor; that is, there is a distribution of residence times of the material within the
reactor. For an ideal CSTR the feed introduced at any given time becomes
completely mixed with the material already in the reactor. Some of the atoms
entering the CSTR leave it ahnost immediately, because material is being
continuously withdrawn from the reactor. Other atoms remain in the reactor almost

forever, because all the material is never removed from the reactor at one time. Many

41

l—d

b------------------------------—

 

F=1 {f<t<oo}

Tracer Output Signal (F)

F=0{0<t<i}

 

 

 

Figure. 5.2 Residence time distribution of a plug flow reactor,
resulting from a step tracer input. The tracer output
signal (F) is zero from the point of injection to the mean
residence time (t). At this point the reactor is totally
cleared out and nothing but the tracer remains in the

reactor (F = 1)2.

42

of the atoms, however, leave the reactor after spending a period of time somewhere in
the vicinity of the mean residence timel. The residence-time distribution of a CSTR
due to a step tracer can be seen in Figure 4.3. Any step input tracer injected at the
inlet will start to mix immediately with the contents of the reactor. The tracer then
proceeds to exit the reactor in an exponential form, until nothing but the tracer

remains in the reactor.

5.3 Residence Time Distribution of Liquid—Phase Sampler

An experiment was performed to determine the residence-time distribution of
the flow vial. In order to collect data the pump flow rate had to be slowed down. The
pump setting was placed on Low 6, which was approximately 3.73 ml/min. A sample
of 100% pure ethanol was run through the liquid—phase sampler to start with. At time
equals zero distilled water (step injection tracer) began to be pumped through the
liquid—phase sampler. The fluid coming from the output of the liquid—phase sampler
was collected in regular 2ml GC vials. Sample cuts were taken every 0.25 ml.
Twenty samples were collected from the flow vial. These twenty 0.25 ml samples
were then analyzed with the GC to determine the concentration of ethanol as a
function of time and volume. The ethanol peak area data was used to produce the
graph seen in Figure 4.4.

Ethanol concentration starts out at approximately 100%. As time proceeds
ethanol concentration starts to reduce and the concentration of distilled water
increases. As can be seen, afier about 5mL of cumulative volume has been collected

from the liquid—phase sampler the concentration of ethanol is practically zero.

43

--------------------------------------------------------------

H

F=1—exp(-t/f)

Tracer Output Signal (F)

 

 

 

Figure 5.3 Residence time distribution of an ideal continuous
stirred tank reactor, resulting from a step tracer input.
The tracer output signal increases in an exponential
form until nothing but the tracer remains in the

reactorz.

44

Ethanol Concentration %VN

Figure 5.4

'4‘; —" firm —— ——r— — -'-j ———- ~ ’Y *- * -"' i

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5
Cumulative Volume, mL

The relationship between concentration of ethanol and
total volume exiting the liquid—phase sampler. Sample
cuts of 0.25 ml were taken and analyzed by GC. After
about 5 ml of sample has been collected the

concentration of ethanol is approximately zero.

45

Collecting 5 ml of sample took approximately 1.34 minutes. As can be seen from the
graph there appears to be very few ethanol particles after four or five minutes. This
indicates that the liquid—phase sampler has minimal dead zone areas in which sample
could remain in the vial over extended periods of time. The mean residence time or
holding time of the liquid—phase sampler was determined to be 20.12 seconds at a flow
rate of 3.73 ml/min. These results are summed up in Table 4.3. This is approximately
how long it takes a single molecule on average to travel through the liquid—phase sampler
at this flow-rate.

This data was compared with the two ideal reactor models discussed earlier, the
CSTR and PF R. Instead of looking at the decrease in ethanol concentration the data was
reversed to show the increase in water concentration as a function of time. This data was
then matched up with the PFR and CSTR residence-time distribution plots at a mean
residence time of 20.12 seconds. The three plots can be seen in Figure 4.5.

The ideal PF R has a water concentration of zero up until the mean residence time.
At this point it jumps to 100%, signifying the reactor has been cleared out. All the
particles have been discharged. The CSTR increases in an exponential form over the
entire range of variables, maximizing at 100%. At the mean residence time the
concentration of the water in the CSTR is about 62%V/V. According to this model
roughly 62% of the ethanol molecules spend 20.12 seconds or less in the reactor, and
38% of the molecules spend longer then 20.12 seconds in the reactor.

The liquid—phase sampler residence-time distribution behaves somewhere in
between the two ideal models. According to the plot, at the mean residence time there

are about 50% of ethanol molecules and 50% water molecules in the vial. The liquid—

46

 

Experiment 1

Experiment 2

 

 

 

Pump Setting Low 6 High 6
Pump Flow-Rate (1)), ml/min 3.73 9.77
Volume of Reactor (V), ml 1.25 1.25
Mean Resrdence Time (t) = V/ 0, 20.12 7.68
(seconds)
Table 5.1 Mean residence time results for the liquid-phase

sampler. Two different flow-rates were used during the

experiment.

47

 

Water Concentration %VN

0 l0

Figure 5.5

 

 

[— ; .._..4 .__ L..# J

+ Liquid- Phase Sarrpler f i

 

|
I
ll *CSTR
i —t—PFR j

20 30 40 50 6O 70 80 90

Time, sec

The residence time distribution of liquid-phase sampler
plotted against the plug flow and continuous stirred
tank reactor models at a mean residence time of 20.12
seconds. The liquid—phase sampler analysis is
represented by the round plot. The CSTR analysis is
represented by the square plot. The PFR analysis is
represented by the diamond plot. The liquid—phase
sampler appears to behave somewhere in between the

two ideal models.

48

phase sampler does appear to represent the PFR more accurately in general. If the liquid—-
phase sampler did completely adhere to one of the ideal models we could describe it in
terms of its reactor design equation. The design equation relates moles, volume,
temperature, flow-rate, time, and the rate of reaction within the reactor. A design
equation for the liquid—phase sampler is not needed for this experiment. While we can
not model the liquid—phase sampler exactly after the PFR or CSTR, the residence-time
distribution comparison shows that the liquid-phase sampler does share a similar

relationship to an ideal reactor.

49

5.4 Literature Cited

1. F ogler, S. H., Elements of Chemical Reaction Engineering 3rd ed., Prentice Hall
Inc., New Jersey, 1999.

2. Levenspiel, 0., Chemical Reaction Engineering 2'"! ed., John Wiley & Sons, Inc.,
1972.

3. Lindemann, D. “Real Time Monitoring of Fruit Spirit Distillation with High

Speed Gas Chromatography”. MS. Thesis.

50

6. Results: Liquid-Phase Sampler

6.1 Still Reproducibility

The reliability of the still was important for data analysis. If the conditions of the
still could not be reproduced over time, this would have an affect on the liquid-phase
sampler experimental data. For the experiment a total of five trials were run. Each trial
was conducted in exactly the same way and was carried out according to the materials
and methods section above. During each of these trials a number of different distillation
variables were recorded and examined.

The first variable to be analyzed was the pressure of the still. The pressure
conditions were not controlled; rather the pressure was left to increase as the distillation
proceeded. A pressure gauge, measuring the pressure in the bottom pot, was used to
record the pressure as a function of distillate volume. The corresponding plot can be seen
in Figure 6.1. The pressure of the still increases in a linear relationship with the amount
of cumulative volume collected. The R2 value and standard deviations indicate a decent
reproducibility.

Temperature was the next variable to be examined. This still did not have
thermometers in the individual trays, therefore, the temperature in the top tray was
unknown for this experiment. However, there was a temperature gauge directly above
the total condenser. This temperature was recorded as a function of distillate volume.
The graph is shown in Figure 6.2. The temperature initially increases sharply from about
53°C to 63°C. The temperature then levels off during the middle of the run, holding at

roughly 63-65°C. After approximately 700 ml of volume has been collected the

51

Pressure, Bar

0.40 T- a -- v -- - i a v ,_ _V___ _____L __._-- I»- E- _.

0.35 1.
l

0.30 1
l
0.25 l

0.20 J

  
   

, y = 0.0003x + 0.1108
0.15 l R2 = 0.9985

l

0.10 .> ‘
0.05

0.00 --= -—
0 100 200 300 400 500 600 700 800 900 1000

Cumulative Distillation Volume, ml

 

-r __.. an __T___._. _ _..‘ _ 1-2%-- If...“ f__.

Figure 6.1 Pressure of the bottom pot on the still as a function of
distillation volume. Pressure increases in a linear

relationship with volume.

52

Temperature, Degrees C

 

 

l

75 l y= 115-07x3 -0.0002x2 + 0.0974x+48.586 l
|= R2 = 0.9885 ‘

70 ~ ‘
l l
l

55 1 l
1 .

60 1 j
l l

55 1 ~

50 . —— —.— ———~ —- __. . ——-—. —— -. ——- . -——- —_~ -- ..
0 100 200 300 400 500 600 700 800 900 1000

Cumulative Distillation Volume, mi
Figure 6.2 Temperature as a function of distillation volume on the

small still. The temperature levels off during the middle
of the distillation with sharp increases both initially and
towards the end portion of the run. This is

characteristic of a third order polynomial relationship.

53

temperature again starts to rise at a rapid rate, exceeding 75°C. This is characteristic of a
third order polynomial function and the data fits this type of relationship.

Another variable analyzed was time, or time of the cuts. 75 milliliter cuts of
distillate volume were taken and the time of each cut was recorded. Figure 6.3 displays
the time of each cut as a function of cumulative volume. As can be seen the data tends to
rise in an exponential form. An exponential curve was fitted with the data. The R2 value
was also shown indicating how well it matches up with the exponential curve. The curve
doesn’t fit the data that great, but it does show a general trend in the data. The standard
deviations increase as the amount of total volume collected increases.

The flow rate of the still was also of interest. Even though the flow rate was set to
a specific setting, the distillate volume did not exit the still at the same rate during the
experiment. A number of factors such as pressure, temperature, and solution composition
affect the flow-rate. The flow rate was calculated as a function of distillation volume and
the plot is displayed in Figure 6.4. Not only does the flow rate decrease over time, but it
has the characteristics of a decreasing in a linear relationship with respect to distillation
volume, as indicated by the R2 value. It starts out initially at a flow rate of around
38ml/min. By the time the experiment had ended the flow rate had reduced to roughly
7ml/min, approximately five magnitudes less then when it started. The standard

deviations of the measurements show no direct pattern.

54

Time of Cuts, See

Figure 6.3

 

y = 151 .41e°-°°34" f
R2 = 0.9522 l

100 200 300 400 500 600 700 800 900 1000
Cumulative Distillation Volume, mi

Time of distillate cuts as a function of distillation
volume. 75 milliliter cuts were taken and the time
recorded. The date appears to increase in an

exponential form.

55

Flowrate, milmin

 
  

y = -0.0374x + 42.94
R2 = 0.9527

  

0 ‘i T_‘ i—l 7" TTTC _'_ ifl'“ T" _T_ _' 7 ‘TI_-—_ __ v" ‘ — " "_—_ _ T7 _ _ ' IV" _ '

0 100 200 300 400 500 600 700 800 900 1000
Cumulative Distillation Volume, ml

Figure 6.4 Flow rate of the distillate as a function of distillation
volume. Flow rate has the characteristics of a
decreasing in a linear relationship with respect to

distillation volume.

56

 

6.2 Method Optimization

Fast and efficient chromatographic readings could be obtained with the liquid-
phase sampler setup. This was achieved with a shorter column and faster flow rate, along
with the appropriate GC programming]. The GC temperature and pressure programs
were shown previously in Tables 4.4 and 4.5. 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 1.72 minutes. The total run time of a sample
acquisition was 3.22 minutes. The chromatograrn that resulted from this type of analysis
is shown in Figure 6.2. As can be seen ethyl acetate and methanol come off extremely
close to one another. Using isothermal conditions these two peaks would not be
separated. 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 the flow vial experiment is shown in Table 6.1.

6.3 Standardization

All six main compounds needed to be standardized. A calibration curve for each
compound was made. While producing the calibration curve for ethanol it was
discovered that actually two calibration curves were needed for each compound. One set
of curves was needed for post-distillation (static) analysis. This would be the normal way
of analyzing the data. The other set of calibration curves was needed for analysis with

the liquid-phase sampler. The movement of liquid through the vial did affect the

57

mV

1000

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

4
750 -
5 6
500 —
1 2
250 - 3
a. ii A
I l T
0 0.5 1.0 1.5 2.0
Time, min
1) Acetaldehyde 2) Ethyl Acetate 3) Methanol
4) Ethanol 5) N-propanol 6) Isoamyl Alcohol
Figure 6.5 Chromatogram for the distillation of a brandy-type

mixture. The sample was collected from the top
tray of the still using the liquid-phase sampler. The
programs used were the same as in Tables 4.4 and
4.5. Peaks were identified by comparing the
retention times from the analysis to standard

retention times for each of the compounds.

58

 

Compound Retention Time (min) Boiling Points (°C)

 

 

Acetaldehyde 0.355 :t 0.003 20.8
Ethyl Acetate 0.727 :t 0.005 77.0
Methanol 0.774 t 0.003 64.7
Ethanol 0.928 t 0.004 78.0
N-propanol 1.150 t 0.014 97.0
Isoamyl Alcohol 1.609 d: 0.009 132.0

 

Table 6.1 Retention times for the chromatograph taken using the
liquid-phase sampler. Peaks were identified by
comparing the retention times from the chromatograph

to standard retention times for each of the compounds.

59

 

calibration data. Therefore a total of twelve calibration curves were needed to analyze
both top tray data and post-distillation data.

Ethanol is by far the main component of any alcoholic beverage and, therefore,
one of the most important to examine. To develop the calibration curve for ethanol
sixteen standards were made over the range of interest. Standards were created at 30%,
35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, and 100%.
To produce the static calibration curve these standards were placed in 2 milliliter GC auto
sampler vials and run with the GC using the parameters and temperature/pressure
programs give in Tables 4.3, 4.4, and 4.5. To obtain the flow calibration curve these
same standards were pumped through the liquid-phase sampler at 9.77ml/min (high 6
setting). This flow rate was chosen because it is the same flow rate at which the liquid-
phase sampler experiment was conducted. Data was taken with the GC using the same
software programs as before. Both the static and flow calibration plots can be seen in
Figure 6.5. In both plots the peak area increases in a linear relationship with respect to
the ethanol concentrations. The equations describing these linear trendlines are also
supplied. The R2 values (20.99) and the standard deviations for each set of data, as seen
on the graph, signify a good calibration. As can be seen the flow calibration data
produced greater peak areas then the static calibration data at the same ethanol
concentrations. This turned out to be the common theme in several of the compounds.

Standardization with acetaldehyde proved to be quite challenging. Due to
acetaldehyde’s low boiling point (approximately 208°C), its viscosity, and continuous

evaporation in the hood it was very hard to produce volumes of accurate amounts.

60

Peak Area

9.0E+05 -. JI-rw~~-

8.5E+06
8.0E+06 i
7.5E+06 i
7.0E+05 i
6.5E+06 J
6.0E+06 -«
5.5E+06 j
5.01=.+05 l1

4.5E+06

4.0E+06 i—r ——

25

Figure 6.6

  
   
 

y(flow) = 47286x + 4E+06
R2 = 0.9938

y(static) = 44993x + 4E+06
R2 = 0.9953

oStatic Data l ‘

1,

[-———-——~—— --~ - - —v—-—-— ~ -—T——— -'—-—~ »-—.~— “m — 1~~——- - "v- ‘ - ‘~———— ‘ ~---

35 45 55 65 75 85 95
Ethanol Concentration %VN

The standard calibration curves for ethanol. The static
measurement is represented by the diamond plot and

the flow measurement is represented by the square plot.

61

- Flow Data .

Syringing small samples of liquid acetaldehyde was just about impossible. In an attempt
to solve this problem larger standard samples were created. The standards were also
made up according to mass instead of volume. Using an analytical balance five standards
were made over the range of interest. Volumetric samples, by mass, were made at 0.1%,
0.3%, 0.5%, 0.7%, and 1.0%. These were run in the same way as ethanol, with the two
different methods (static and flow). The calibration curves for both can be seen in Figure
6.6. The R2 values are again really good, but the errors in measurement are worse. As
the concentrations of acetaldehyde increased so did the standard deviations. Also it can
be seen that the flow data is not necessarily above the static data as it was for ethanol.
This is probably just an error in attempting to make good acetaldehyde standards based
on the problems given above.

Ethyl acetate and methanol were analyzed together. These two compounds have
GC retention times very close to one another and thus their peak areas overlap
considerably. It was decided that it would be best to standardize them together to model
the synthetic solutions after actual brandy-type solutions. Standards were made at 0.1%,
0.3%, 0.5%, 0.6%, 0.7% for ethyl acetate and 0.3%, 0.4%, 0.5%, 0.7%, 1.0% for
methanol. These standards were run with the GC using the static and flow methods. The
calibration plots can be seen in Figures 6.7 and 6.8. As is expected the R2 values in
general are not as good as the other compounds. Because of the peak overlap the
software program must make certain approximations as to where the dividing line should
be, and thus peak area approximations. These approximations obviously create error; and

this can be seen in the calibration curves for both ethyl acetate and methanol.

62

Peak Area

4.0E+05 l
l
3.0E+05 j

l
2.0E+05 .

l
1.0E+05

0.0E+00 i

0.0

Figure 6.7

y(flow) = 481795x + 30253

    
 
 

R2 = 0.9938

F ____._ ._ -

0 Static Data
I Flow Data

 

___._ ._T_____ _—'I ___ ___

0.4 0.6 0.8
Acetaldehyde Concentration %VN

 

— ' I.

1.0

y(static) = 534590x + 13154
R2 = 0.9947

The standard calibration curves for acetaldehyde. The

static measurement is represented by the diamond plot

and the flow measurement is represented by the square

plot.

63

Peak Area

6.0E+05 i
l

i
l
4.0E+05 j
j

3.0E+05 a

 

2.0E+05 1
l
1.0E+05 j

l

0.0E+00 r _—

0

Figure 6.8

y(static) = 824857x - 51033 I
R2 = 0.9598

  
   
 

y(flow) = 636217x - 6991.7
R2 = 0.9844

 

j 0 Static Data j

l - Flow Data j

L7 -__. . . 7, l
0.1 0.2 0.3 0.4 0.5 0.6 0.7

Ethyl Acetate Concentration %VN

The standard calibration curves for ethyl acetate. The
static measurement is represented by the diamond plot
and the flow measurement is represented by the square

plot.

64

l

0.8

Peak Area

4.0E+05

l
l
3.01305 1.
l
l

1.0E+05

0.0E+00 4 ~._

0.0

Figure 6.9

   
   

y(flow) = 426754x + 98218
R2 = 0.958

y(static) = 497046x + 47555 i

 

R2 = 0.9974 ,

l

l

1, LL -___ _, L l

, oStatic Data j j

j I Flow Data l 1

_WM , .-..-_ ”In- . _.___._ _.._ ‘..______..._ L. _. l
0.2 0.4 0.6 0.8 1.0 1.2

Methanol Concentration %VN

The standard calibration curves for methanol. The static
measurement is represented by the diamond plot and

the flow measurement is represented by the square plot.

65

Five samples were made for N-Propanol. These standards were made at 0.1%,
0.3%, 0.5%, 0.7%, and 1.0%. The static and flow methods were used with the GC to
analyze the data. The results are displayed in Figure 6.9. The R2 values and standard
deviations are reasonable. The calibration curve for the flow method is again above that
for the static. This is similar to the calibration data for ethanol and is expected.

Isoamyl Alcohol was that last compound to be analyzed. Five samples were again
prepared over the range of interest. Standards were made at 0.1%, 0.3%, 0.5%, 0.7%,
and 1.0%. These were analyzed with the GC and the appropriate static and flow
calibration curves were produced. The plots can be seen in Figure 6.10. R2 values and
standard deviations are again good. The flow data resides above the static as in the

previous compounds.

66

Peak Area

1.4E+05 - a v a L, v v a the i r f r r f __ —————~———~-—~

    
 

l
1.2E+06 l
l y(flow) = 1E+06x + 100747 N
j R2 = 0.999 1
1.0E+06 j l
l l
8.0E+05 j j
l l
6.0E+05 i y(static) = 978417x+ 92690 “
l R2 = 0.9949
4.0E+05 l ‘
i - l ‘
‘ Static Data
2.05+05 i l ° i l
1 ;_ _ ___-___ m ,
0.0900 Lua —-—e—— -——-—-—-~ +--—— _,______ —.——-———— -— ——..—t
0.0 0.2 0.4 0.5 0.8 1.0 1.2

N-Propanol Concentration %VN

Figure 6.10 The standard calibration curves for n-propanol. The
static measurement is represented by the diamond plot
and the flow measurement is represented by the square

plot.

67

Peak Area

1.6E-I-06 1— Jr--— --— —————— -—— ——- -A~—— fl..— -._-.# ___ -W _..___.-____-

1.4E+06 l

  
   
 

l
1.2E+06 ,
l y(flow) = 1E+06x + 20555 t
l R2=0.999 I
1.0E+06 -
8.0E+05 «‘
j y(static) = 1E+06x + 26751
60805 j R2=0.9952
4.0E+05 j
__ I- __ --___ L I
20E 05 ‘ FioStaticData .
. + '1 .
l o ' I Flow Data I
l a a- »— i
0.0E+00 - --——- —-—- . ___ . a- L- - _._ __ I
0.0 0.2 0.4 0.5 0.8 1.0 1.2

isoamyl Alcohol Concentration %VN

Figure 6.11 The standard calibration curves for isoamyl alcohol. The
static measurement is represented by the diamond plot
and the flow measurement is represented by the square

plot.

68

 

6.4 Distillation Behavior

The behavior of all of the compounds during the distillation was studied using the
liquid-phase sampler setup. Chromatography spectra were taken using this setup. The
concentrations for each of the compounds as they materialized in the top tray were
calculated by comparing the peak areas of the congeners to the peak areas of their
respective calibration curves. These concentrations were then plotted to show percent
volume of congener vs. distillation time. A post-distillation analysis was also performed
on the cuts of distillate using the same GC programming as that for the top tray data.
This was done in order to compare the behavior and amounts between the congeners
present in the top tray and those present after distillation. The trends between the two
sets of data were shown.

The major component of the synthetic brandy solutions was ethanol. Ethanol
measurements were taken during the distillation with the liquid-phase sampler and also
on the 75 milliliter cuts alter the distillation. The concentrations of ethanol were found
using the standard ethanol calibration curves. The behavior of ethanol for both methods
during the distillation and how they relate to each other is shown in Figure 6.11. The
post-distillation data shows ethanol starting out near 85%v/v ethanol. The trend slightly
increases initially and maintains a concentration of about 90%v/v for the first 20 minutes
of the distillation. The concentration then starts to drop off towards the end of the run,
eventually reaching about 40%v/v afier 45 minutes. The top tray data shows ethanol
starting out at about the same concentration as that of the post—distillation, 85%v/v. But

instead of increasing initially, it proceeds to decline over time and continues to decrease

69

Ethanol Concentration %VN

120 -. —~-—-
110—1
100 _

 

Figure 6.12

Ethanol Behavior

 

F—o— Post-Distillation Data I j
5 + TOP Tray - Online Data ; ‘

 

     

 

 

1 0 1 5 20 25 30 35 40 45 50
Distillation Time, min

Ethanol concentration as a function of distillation time.
Data was taken from the top tray with the liquid-phase
sampler and also after distillation. Standard ethanol
calibration curves were used for each analysis. The
online measurement is represented by the square plot
and the post distillation measurement is represented by

the diamond plot.

70

 

throughout the distillation. The data drops off rather sharply compared to the post-
distillation data and is practically all but gone from the top tray after 30 minutes.

The first major congener to emerge from the GC column is acetaldehyde. Figure
6.12 shows the behavior comparison between the top tray measurements and the post-
distillation measurements. Both plots show a sharp decrease in concentration over time.
Only one measurement could be obtained from the top tray due to the speed at which
acetaldehyde passed through the system.

Ethyl acetate was examined next. Its behavior can be seen in Figure 6.13. The top
tray and post-distillation measurements again show similar trends. Both show a sharp
decrease in concentration over time. After about 10 minutes ethyl acetate has blown
through the top tray. After 15 minutes it can be seen to have left the still completely.

The behavior of methanol is shown in Figure 6.14. The post-distillation plot
shows methanol initially decreasing. Then towards the middle of the distillation it begins
to show an increasing trend. The top tray data shows a pretty sharp decrease in methanol
concentration for the first 5-10 minutes. After that though, it decreases in a more gradual
fashion until the end of the distillation.

Figure 6.15 shows the comparison of plots for n-propanol. Overall the plots for
both the top tray and post-distillation tend to agree with each other. They both show a
gradual decrease in concentration over time. However in the post-distillation data the
concentration increases briefly at first before falling.

Isoamyl alcohol came of the GC column last. Its behavior is seen in Figure 6.16.
These two plots are very similar in comparison. Both show a slight increase in

concentration and then a gradual decline over time. What is interesting about this set of

71

Acetaldehyde Concentration %VN

3.0 1 ——~ —— ”w —— _..-__.E LE ,v

 

Acetaldehyde Behavior

 

 

‘w"—’ l

l _._ Post-Distillation Data-raj
j + Top Tray - Online Data j j

 

 

0.5 +—
0 5
Figure 6.13

 

 

1 0 1 5 20 25 30 35 40 45 50
Distillation Tme, min

Acetaldehyde concentration as a function of distillation
time. Data was taken from the top tray with the liquid-
phase sampler and also after distillation. Standard
acetaldehyde calibration curves were used for each
analysis. The online measurement is represented by the
square plot and the post distillation measurement is

represented by the diamond plot.

72

Ethyl Acetate Behavior

 

 

 

1.6‘1- ‘ww— WE‘LL A.
l l
1.4 —" I _ g
j 1' +Post—Distillation Data j
1.2 j +Top Tray-Online Data 1 l

 

Ethyl Acetate Concentration %VN

 

‘0.2 +7 fl‘ " ‘1 1‘" r “ . ' -'*—~———‘r -- --
5 10 15 20 25 30 35 40 45 50

Distillation Time, min

 

 

 

._1___

Figure 6.14 Ethyl acetate concentration as a function of distillation
time. Data was taken from the top tray with the liquid-
phase sampler and also after distillation. Standard
ethyl acetate calibration curves were used for each
analysis. The online measurement is represented by the
square plot and the post distillation measurement is

represented by the diamond plot.

73

Methanol Concentration %VN

Methanol Behavior

12 E_-L____._______ _# +. # #m* ,_.,A

 

1.0

 

 

0.8 1 k ._ //

0.6 7'

 

 

 

 
 
 
 

0.4

 

l

i eo—Post-Distiilation Data i

 

 

 

. I
0.2 l +Top Tray-Online Data l l
l l
0.0 «l
l
l l
02 s--— «a —s 9*. _ . «H- , —. — ...—— —— ._.. ___
O 5 10 15 20 25 30 35 4O 45 50

Distillation Time, min

Figure 6.15 Methanol concentration as a function of distillation
time. Data was taken from the top tray with the liquid-
phase sampler and also after distillation. Standard
methanol calibration curves were used for each
analysis. The online measurement is represented by the
square plot and the post distillation measurement is

represented by the diamond plot.

74

N-Propanoi Concentration %VN

N-Propanol Behavior
2.0 —~ _._ — -._ _‘____ ___.._ ___

 

l

1.8 1' ,_ ., . *

l ; +Post—Distillaiion Data
1.6 i '

,_.______ *h __—‘..._—_+—.. .! _

1r i —-— Top Tray - Online Data !

/ ._ .. . _- __L, LL.

 

 

 

 

 

 

 

 

 

0 5 10 15 20 25 30 35 40 45 50
Distillation Time, min

Figure 6.16 N-propanol concentration as a function of distillation
time. Data was taken from the top tray with the liquid-
phase sampler and also after distillation. Standard n-
propanol calibration curves were used for each analysis.
The online measurement is represented by the square
plot and the post distillation measurement is

represented by the diamond plot.

75

Isoamyl Alcohol Concentration %VN

2.7 ~- - ~*

2.4

2.1 -
I

1.8 *

 

1.2 4

 

0.9

 

0.6 *1.
0.3 i

0.0

Figure 6.17

Isoamyl Alcohol Behavior

 

 

If S;— Post-Distillation Data 1
’ _.__ Top Tray - Online Data .

 

 

 

Distillation Time, min

Isoamyl alcohol concentration as a function of
distillation time. Data was taken from the top tray with
the liquid-phase sampler and also after distillation.
Standard isoamyl alcohol calibration curves were used
for each analysis. The online measurement is
represented by the square plot and the post distillation

measurement is represented by the diamond plot.

76

plots is that at the start the concentration from the top tray is higher then the post-

distillation initial concentration.

77

6.6 Literature Cited

1. Lindemann, D. “Real Time Monitoring of Fruit Spirit Distillation with High
Speed Gas Chromatography”. 2003 MS. Thesis.

78

7. Summary and Conclusions

Previous problems with the headspace sampler setup lead to the development of
the new liquid-phase sampler setup. This setup integrated many of the same ideas and
techniques as before but allowed for a consistent sample transfer from the still to the GC.
This created a successful high speed gas chromatography setup which was capable of
analyzing distillations in real time.

Understanding the flow process happening within the liquid-phase sampler was
important. The relationship between the sample being extracted from the still and that
being injected into the GC can be seen through liquid-phase sampler’s residence-time
distribution. Its behavior with respect to a plug flow reactor and a continuous stirred tank
reactor was observed. The residence-time distribution of the liquid-phase sampler resides
somewhere in between the two ideal models. It does not follow the clear cut PFR
distribution nor does it in any way adhere to the exponential form of the CSTR. The
liquid-phase sampler appears to match up more accurately with the PF R model overall.

In the beginning stages before the mean residence time and towards the end it is basically
equivalent with the PFR. Around the mean residence time it deviates minutely from the
PFR model, not following the clear cut ideal pattern. This is a result of the mixing
happening within the vessel, reacting like a CSTR. The passage of liquid through the
liquid-phase sampler appears to be uniform with a minimal dead zone where molecules
might accumulate over time. Modeling the liquid-phase sampler after the PFR would be
a decent approximation. While not absolute ideal, the data provides a good indication

that the liquid-phase sampler is a good receptacle for GC analysis sampling.

79

The variable conditions of the IOL Arnold Holstein still were recorded to confirm
its reproducibility. The pressure, temperature, time, and flow rate can be duplicated in a
precise manner over several trials. A linear relationship is shown to exist for pressure
and flow rate as a function of distillation volume. The cut times verses distillation
volume expressed an exponential relationship. Temperature as a function of distillation
volume exhibited a third order polynomial relationship. While this may suggest the still
conditions are reliable, it does not necessarily imply a high quality control of distillation
product. For this, compound composition and concentration were analyzed in detail.

Calibration of the six main compounds was successful. To accurately represent
the data it was determined that actually two separate calibrations were needed, one for
flow sample and one for static sample. The standard solutions resulted in a good set of
calibration curves for both the static and flow data. A repetitive trend was seen in the
calibration. For ethanol, n-propanol, and isoamyl alcohol the flow set of data was clearly
above that of the static data for a given concentration. For the other three compounds
(acetaldehyde, ethyl acetate, and methanol) a similar occurrence was expected between
the flow and static, but errors in program analysis of the peak area and preparing accurate
standard solutions prohibited in showing this type of correlation.

The reason for this type of relationship is unclear. It could be the result of the
movement of liquid during analysis, creating slight temperature, pressure, and volume
differences. It is plausible that a larger sample is being injected into the GC during the
flow analysis then during a normal static injection. The volume of liquid in the liquid-
phase sampler during flow analysis was not equivalent to the volume of liquid present in

the GC vial during static analysis. This may have created some differences in injection

80

sample size. Also, the auto injector syringe may take up slightly more sample due to the
flow of liquid through the liquid-phase sampler, thus forcing more solution into the
syringe needle during injection. The larger sample size would create larger peak areas.
Analyzing various standards at several different flow rates would provide a good
indication about what is occurring here.

The distillation behavior of ethanol, acetaldehyde, ethyl acetate, methanol, n-
propanol, and isoamyl alcohol in the top tray was studied with the liquid-phase sampling
method. In addition, the distillate was studied afierward under the same GC conditions
for comparison. The amounts and the behavior determined from the top tray did not
correspond directly to those of the distillate measured with the GC after distillation.
Some general trends were observed. All of the top tray concentrations were shown to be
less then those taken from the post-distillation. This was true from start to finish. In fact
only once, in the case for isoamyl alcohol, did the concentration in the top tray exceed the
post-distillation data. The top tray concentrations all decrease over time, except for
isoamyl alcohol where there is a slight increase initially. This is not the case for the post-
distillation concentrations of certain compounds.

For acetaldehyde and ethyl acetate the decreasing plots are expected in both the
top tray and after distillation. These two compounds reside in the still for a short period
of time, due to their low boiling points, and come off rather rapidly. For the other
compounds the fall and rise of concentrations observed post-distillation does not seem to
directly correlate with those in the top tray. The top tray concentrations should give an
indication of what to expect in the post-distillation data. The two sets of data should not

be exactly similar, but should show some type of connection. Looking at the plots, there

81

may be a small indication due to the change in slope for the top tray data. It is hard to
make any conclusions based on this though. In general, no specific relationship could be
seen from these results. For the most part, all the top tray data simply decreased over
time. Several reasons for this type of behavior are possible. First, the data observed
could be the actual relationship that the top tray produces as the compounds pass through
it. While this may be the most obvious explanation it does not seem likely. The top tray
concentrations would be expected to show some type of relationship with the post-
distillation data. Second, the calibration data may be faulty. Again this may be the case,
and the reasons for this were described above. But overall the calibration results obtained
do indicate a good standardization. Third, and probably the most likely, the Arnold
Holstein still could be poorly constructed. The proportion of the column to the size of the
pot indicates an engineering defect. Smaller stills in general do not produce the same
quality of spirits as larger ones are capable of. The rate at which the distillation
proceeded confirms this proposition. Attempts were made to slow down the speed of the
still, changing pressure and distillate flow rate, but this seemed to have little effect. As a
result, during distillation, all compounds were basically blown right through the column,
with little or no rectification. This in turn would show a top tray concentration reduction
for all compounds throughout the distillation. Data analysis on a larger, higher quality
constructed still would be needed to confirm this hypothesis.

In conclusion, the liquid-phase sampler proved to be a very feasible setup for high
speed gas chromatography. This method achieved rapid sampling and adequate
separation of all important compounds. The method produced chromatograms in a time

much faster than a conventional GC so that measurements could be obtained every few

82

minutes during the distillation. Behavior of the compounds could be analyzed as they
passed through the top tray of the column. More studies involving different distillation
apparatuses and setup improvements are required to further the pursuit of monitoring fruit

spirits in real time using high speed gas chromatography.

83

8. Future Work

The liquid-phase sampler setup has been proven to work effectively. The next
step would be to try out the setup on a larger, higher quality still. A properly engineered
still would provide improvement in the results. Obtaining data from such a still would
perhaps better show the unknown relationship between the top tray and the post-
distillation data. Each of the individual compounds could then be studied more in depth.

It would be beneficial to observe the other trays on the still to see how they
compare with the top tray as well as with the post-distillation data. Observing the still
pot while the distillation is occurring might also be profitable. Using these liquid-phase
sampler results, manipulation of the still to achieve higher product quality should be
examined. Actual fruit fermentations and distillations could be done and analyzed with
the liquid-phase sampler setup. Synthetic solutions are great to do research with, but
acquiring data from the actual fruit spirit distillations would be something of interest.

Higher speed and resolution are always being strived for. As high speed gas
chromatography becomes more common and available better techniques will surface,
resulting in better performance. This perhaps will allow for the development of an even
faster sampling rate then the 3.22 minutes now achieved while maintaining compound
resolution. It might be adventitious to find a new technique that can separate more then
just the six main compounds. This would probably require changing several GC parts,
parameters, and programs. With hundreds of compounds present in any given fi'uit spirit,
a way to examine these smaller concentrations as they move up through the trays would

be of great use.

84

 

 

uijijjjjtjjjjjujijjju