Citation
Design and performance criteria in the development of a new instrumental method for ultramicro elemental analysis

Material Information

Title:
Design and performance criteria in the development of a new instrumental method for ultramicro elemental analysis
Added title page title:
Ultramicro elemental analysis
Creator:
Miller, C. David, 1931-
Place of Publication:
[Gainesville]
Publisher:
[s.n.]
Publication Date:
Language:
English
Physical Description:
vi, 107 l. : illus. ; 28 cm.

Subjects

Subjects / Keywords:
Calcium ( jstor )
Carbon ( jstor )
Carbon dioxide ( jstor )
Combustion ( jstor )
Copper oxides ( jstor )
Hydrides ( jstor )
Hydrogen ( jstor )
Nitrogen ( jstor )
Oxygen ( jstor )
Silica gel ( jstor )
Instrumental analysis ( lcsh )
Microchemistry ( lcsh )
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis - University of Florida.
Bibliography:
Bibliography: l. 104-106.
General Note:
Manuscript copy.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
11062489 ( OCLC )
ocm11062489
0030112408 ( ALEPH )

Downloads

This item has the following downloads:

ELNDBIG8G_WFWZ8Y.xml

AA00012906_00001.pdf

designperforma00mill_0000.txt

designperforma00mill_0049.txt

designperforma00mill_0089.txt

AA00012906_00001_0032.txt

designperforma00mill_0028.txt

AA00012906_00001_0071.txt

designperforma00mill_0002.txt

designperforma00mill_0084.txt

designperforma00mill_0105.txt

AA00012906_00001_0073.txt

AA00012906_00001_0081.txt

AA00012906_00001_0023.txt

AA00012906_00001_0009.txt

AA00012906_00001_0051.txt

designperforma00mill_0055.txt

designperforma00mill_0007.txt

AA00012906_00001_0038.txt

AA00012906_00001_0115.txt

AA00012906_00001_0007.txt

designperforma00mill_0035.txt

AA00012906_00001_0082.txt

designperforma00mill_0034.txt

designperforma00mill_0015.txt

AA00012906_00001_0006.txt

AA00012906_00001_0095.txt

designperforma00mill_0041.txt

AA00012906_00001_0020.txt

AA00012906_00001_0061.txt

AA00012906_00001_0050.txt

designperforma00mill_0078.txt

designperforma00mill_0047.txt

AA00012906_00001_0054.txt

AA00012906_00001_0046.txt

designperforma00mill_0103.txt

designperforma00mill_0081.txt

AA00012906_00001_0089.txt

AA00012906_00001_0004.txt

AA00012906_00001_0041.txt

AA00012906_00001_0086.txt

designperforma00mill_0024.txt

AA00012906_00001_0075.txt

AA00012906_00001_0068.txt

designperforma00mill_0086.txt

AA00012906_00001_0088.txt

AA00012906_00001_0025.txt

AA00012906_00001_0003.txt

AA00012906_00001_0015.txt

AA00012906_00001_0048.txt

AA00012906_00001_0055.txt

AA00012906_00001_0047.txt

designperforma00mill_0071.txt

designperforma00mill_0093.txt

AA00012906_00001_0076.txt

designperforma00mill_0088.txt

designperforma00mill_0020.txt

AA00012906_00001_0045.txt

designperforma00mill_0066.txt

designperforma00mill_0033.txt

AA00012906_00001_0105.txt

designperforma00mill_0094.txt

AA00012906_00001_0092.txt

designperforma00mill_0110.txt

AA00012906_00001_pdf.txt

AA00012906_00001_0053.txt

AA00012906_00001_0085.txt

AA00012906_00001_0030.txt

AA00012906_00001_0066.txt

designperforma00mill_0003.txt

designperforma00mill_0009.txt

designperforma00mill_0018.txt

designperforma00mill_0080.txt

designperforma00mill_0101.txt

designperforma00mill_0090.txt

AA00012906_00001_0059.txt

AA00012906_00001_0058.txt

designperforma00mill_0011.txt

designperforma00mill_0058.txt

AA00012906_00001_0069.txt

AA00012906_00001_0057.txt

designperforma00mill_0106.txt

designperforma00mill_0059.txt

AA00012906_00001_0091.txt

AA00012906_00001_0104.txt

AA00012906_00001_0017.txt

designperforma00mill_0091.txt

designperforma00mill_0075.txt

designperforma00mill_0045.txt

AA00012906_00001_0090.txt

designperforma00mill_0092.txt

AA00012906_00001_0097.txt

AA00012906_00001_0043.txt

designperforma00mill_0050.txt

designperforma00mill_0096.txt

designperforma00mill_0079.txt

designperforma00mill_0029.txt

designperforma00mill_0098.txt

designperforma00mill_0014.txt

ELNDBIG8G_WFWZ8Y_xml.txt

designperforma00mill_0063.txt

AA00012906_00001_0013.txt

AA00012906_00001_0026.txt

designperforma00mill_0025.txt

AA00012906_00001_0008.txt

designperforma00mill_0052.txt

AA00012906_00001_0022.txt

designperforma00mill_0104.txt

AA00012906_00001_0019.txt

designperforma00mill_0107.txt

designperforma00mill_0031.txt

designperforma00mill_0004.txt

designperforma00mill_0037.txt

designperforma00mill_0095.txt

AA00012906_00001_0011.txt

designperforma00mill_0109.txt

designperforma00mill_0097.txt

AA00012906_00001_0060.txt

designperforma00mill_0108.txt

designperforma00mill_0100.txt

designperforma00mill_0046.txt

AA00012906_00001_0010.txt

AA00012906_00001_0042.txt

AA00012906_00001_0049.txt

AA00012906_00001_0110.txt

designperforma00mill_0023.txt

designperforma00mill_0112.txt

designperforma00mill_0006.txt

AA00012906_00001_0005.txt

designperforma00mill_0039.txt

AA00012906_00001_0070.txt

designperforma00mill_0076.txt

AA00012906_00001_0077.txt

AA00012906_00001_0114.txt

designperforma00mill_0043.txt

designperforma00mill_0038.txt

AA00012906_00001_0018.txt

designperforma00mill_0036.txt

designperforma00mill_0061.txt

designperforma00mill_0005.txt

AA00012906_00001_0084.txt

designperforma00mill_0008.txt

AA00012906_00001_0102.txt

designperforma00mill_0069.txt

AA00012906_00001_0037.txt

AA00012906_00001_0062.txt

AA00012906_00001_0100.txt

designperforma00mill_0102.txt

designperforma00mill_0113.txt

AA00012906_00001_0065.txt

AA00012906_00001_0072.txt

designperforma00mill_0065.txt

designperforma00mill_0053.txt

designperforma00mill_0064.txt

AA00012906_00001_0083.txt

AA00012906_00001_0103.txt

designperforma00mill_0074.txt

AA00012906_00001_0111.txt

designperforma00mill_0114.txt

designperforma00mill_0060.txt

AA00012906_00001_0016.txt

AA00012906_00001_0101.txt

designperforma00mill_0027.txt

designperforma00mill_0057.txt

designperforma00mill_0021.txt

AA00012906_00001_0079.txt

designperforma00mill_0083.txt

designperforma00mill_0051.txt

AA00012906_00001_0107.txt

designperforma00mill_0068.txt

designperforma00mill_0062.txt

designperforma00mill_0030.txt

AA00012906_00001_0040.txt

designperforma00mill_0099.txt

AA00012906_00001_0039.txt

AA00012906_00001_0094.txt

AA00012906_00001_0067.txt

AA00012906_00001_0029.txt

AA00012906_00001_0080.txt

AA00012906_00001_0028.txt

AA00012906_00001_0074.txt

designperforma00mill_0082.txt

AA00012906_00001_0002.txt

designperforma00mill_0026.txt

designperforma00mill_0072.txt

AA00012906_00001_0109.txt

AA00012906_00001_0036.txt

AA00012906_00001_0098.txt

AA00012906_00001_0027.txt

designperforma00mill_0048.txt

designperforma00mill_0111.txt

designperforma00mill_0016.txt

designperforma00mill_0017.txt

designperforma00mill_0044.txt

AA00012906_00001_0064.txt

AA00012906_00001_0063.txt

designperforma00mill_0012.txt

AA00012906_00001_0093.txt

AA00012906_00001_0056.txt

designperforma00mill_0073.txt

AA00012906_00001_0024.txt

designperforma00mill_0085.txt

AA00012906_00001_0012.txt

AA00012906_00001_0014.txt

AA00012906_00001_0001.txt

AA00012906_00001_0087.txt

AA00012906_00001_0113.txt

AA00012906_00001_0112.txt

designperforma00mill_0013.txt

AA00012906_00001_0034.txt

AA00012906_00001_0099.txt

designperforma00mill_0087.txt

AA00012906_00001_0078.txt

designperforma00mill_0070.txt

AA00012906_00001_0106.txt

designperforma00mill_0001.txt

AA00012906_00001_0044.txt

designperforma00mill_0056.txt

designperforma00mill_0010.txt

AA00012906_00001_0052.txt

AA00012906_00001_0033.txt

designperforma00mill_0042.txt

designperforma00mill_0032.txt

designperforma00mill_0054.txt

AA00012906_00001_0031.txt

designperforma00mill_0022.txt

AA00012906_00001_0035.txt

AA00012906_00001_0096.txt

designperforma00mill_0040.txt

designperforma00mill_0077.txt

AA00012906_00001_0108.txt

designperforma00mill_0019.txt

AA00012906_00001_0021.txt

designperforma00mill_0067.txt


Full Text











DESIGN AND PERFORMANCE CRITERIA IN
THE DEVELOPMENT OF A
NEW INSTRUMENTAL METHOD

FOR ULTRAMICRO ELEMENTAL ANALYSIS








By
C. DAVID MILLER


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











UNIVERSITY OF FLORIDA
April, 1964












ACKNOWLEDGMENTS


The author extends his sincere appreciation to his

supervisory committee, to the faculty of the Chemistry

Department of the University of Florida, and to his

colleagues for their interest and their friendly cooperation

during the course of this investigation. He takes

particular pleasure in expressing gratitude to his research

director, Dr. J. D. Winefordner, for his constant sincerity

and willingness in providing the counsel, guidance, and

encouragement necessary for the realization of this project.

The helpful suggestions and the cooperation of the

research and development group of the American Instrument

Company are deeply appreciated. Mr. William C. Hampton is

particularly deserving of special thanks in this regard.

Additional recognition is due the American Instrument

Company for the financial support of this research, and the

assistance is highly valued.

Finally, the author gratefully acknowledges his

indebtedness to his wife, Rita Sue, for her continued

inspiration and understanding.











TABLE OF CONTENTS


ACKNOWLEDGMENTS . .
LIST OF TABLES . .
LIST OF FIGURES . .
Section


I. INTRODUCTION .
II. EXPERIMENTAL CONSIDERATIONS .
Initial Functional Objectives and
Preliminary Concepts .
Selection of Components and Operational
Procedures .. .
Critical Parameters in the Finalized
System and Extension to Nitrogen
Determination .
III. CONSTRUCTIONAL AND OPERATIONAL DETAILS .
Experimental System .
Operational Aspects .
IV. RESULTS AND DISCUSSION .
V. EXTENSIONS TO ADDITIONAL ANALYSES .
VI. SUMMARY ...................


APPENDIX . .
LITERATURE CITED .
BIOGRAPHICAL SKETCH .


Page
ii
iv
V


1
8

8

11


36
55
55
69
77
93
98
101
104
107


II-












LIST OF TABLES


Table Page

1. Test Results for Accuracy of Carbon Analysis 34

2. Peak Height to Half-Peak-Width Ratios. 40

3. Heater Power Requirements. ... 66
4. Variation of Bridge Current With Voltage 69

5. Precision for a Typical Carbon and Hydrogen
Determination ... 81

6. Precision for a Typical Nitrogen Determination 82

7. Analytical Results for Carbon and Hydrogen 84
8. Analytical Results for Nitrogen. .. 90

9. Results for Oxygen Analyses. ... 95












LIST OF FIGURES
Figure Page

1. Preliminary System and Readout for Carbon
and Hydrogen Analysis .. 10

2. Preliminary Design of Sample Introduction -
System. ... 14

3. Circuit for Conductance Measurement of Water
and Thermal Conductivity Measurement of
Carbon Dioxide 17
4. Details of the Retention and Conversion
System. .. .. 26

5. Variations in Component Arrangements and
Readouts for Carbon and Hydrogen Analysis 30

6. Comparison of Sensitivity to Carbon Dioxide
Using Packed and Unpacked Columns Under
Identical Flow Conditions ......... 33

7. Finalized Flow System and Readout for Carbon
and Hydrogen Analysis .. 37

8. Comparison of Combustion Efficiencies Using
Oxygen and Nitrogen Carrier Gases .. 38
9. Response to Carbon Dioxide Versus Equili-
bration Time. .. 46

10. Dependence of Response on Flow Rate .... 47

11. Reduction of Atmospheric Contamination During
Nitrogen Analyses 53

12. Instrumental Arrangement for Carbon and
Hydrogen Analysis ........... 56

13. Details of Conversion and Detection Systems 57
14. Details of Orifice Mounting .. 58










Figure Page

15. Details of Cell Construction 60

16. Tube Packings for Combustion, Absorption, and
Conversion .. 62

17. Details of Sample Introduction Port 63

18. Schematic Diagram of Power Supply and Bridge
Circuit. .. 67

19. Stability of Power Supply With Change in Input
Voltage. . ... 68

20. Encapsulation of Powdered Samples. ... 71

21. Readouts for Carbon Dioxide, Hydrogen, and
Nitrogen . .. 78

22. Response Curves for Carbon Dioxide, Hydrogen,
and Nitrogen ... 79

23. Prototype Model of American Instrument Co.
Carbon and Hydrogen Analyzer .. 105













SECTION I


INTRODUCTION


Elemental analysis may be defined as the determination

of the total percentage of each element present in a given

compound without regard to molecular arrangement. Analyses

of carbon, hydrogen, nitrogen, etc., find primary application

in the verification of empirical formulae of pure organic

compounds, or in the assay of mixtures whose constituents

exhibit substantial compositional differences. Information

essential to structural identification is provided by

inclusion of complimentary data from functional group

analysis.

The method for carbon and hydrogen first studied by

Pregl in 1910 may be regarded as the classical procedure of

microchemical analysis. The corresponding Liebig macro-

procedure now finds little or no application due to excessive

time requirements for complete combustion. Numerous modifi-

cations of the original techniques have been proposed with

the object of simplifying the determination and minimizing

the inherent errors. In general, however, the basic

principles of Pregl's method have become well accepted, and

the method is more widely used than any other.











The current determination is based upon complete

conversion in a stream of oxygen of the carbon and hydrogen

in the sample to carbon dioxide and water. This is

effected by passage of the pyrolysis products through a

combustion tube packed with copper oxide and lead chromate

maintained at a temperature of 6800C. Carbon dioxide is

absorbed by sodium hydroxide on asbestos, and water is

retained by magnesium perchlorate. The corresponding

weight increases in the absorption tubes are related to the

percentage composition of the sample.

Pregl has also been credited with adaptation of the

classical macro-Dumas nitrogen determination to the micro-

technique which has been accepted as reliable for most

forms of nitrogen. The procedure involves combustion of

the sample in an atmosphere of carbon dioxide, in which

carbon and hydrogen are converted to carbon dioxide and

water over a copper oxide catalyst, and oxides of nitrogen

are reduced to elemental nitrogen over copper. The nitrogen

is collected and measured in a nitrometer containing a

confining solution of 50 per cent aqueous potassium

hydroxide which absorbs the carbon dioxide.

Instrumentation automating the basic techniques for

carbon, hydrogen, and nitrogen, as well as the direct

determination of oxygen has recently been made available by

Coleman Instruments, Inc. (5,6,7). The "empty-tube" method










for carbon and hydrogen which uses a quartz combustion tube

containing baffles and a small quantity of silver has been

investigated and recommended by Belcher, Ingram, Spooner,

Kainz, and Horwatitsch (3,4,17,22). Improved combustion

efficiency has been observed upon substitution of various

oxides for the conventional copper oxide packing. K8rbl

(25,26), for example, has obtained excellent results at

reduced operating temperatures with the decomposition

products of silver permanganate.

Very few methods are available which are capable

of circumventing combustion requirements for elemental

analysis although a neutron scattering technique has been

used by Finston and Yellin (11) to determine carbon-hydrogen

ratios. A number of efforts directed toward elimination of

absorption tube weighing rely upon conductometric, mano-

metric, or volumetric measurements. Malissa and Pell (28),

for example, have determined carbon, hydrogen, and sulfur

conductometrically while Frazer (13) has investigated a gas

volumetric method for the simultaneous determination of

carbon, hydrogen, and nitrogen. Hozumi and Kirsten (16)

have recently reported a similar approach to the combined

analysis based upon dry combustion of the sample in a sealed

tube and the subsequent weighing of the quantity of mercury

required to replace each of the combustion gases.










A non-dispersive infrared gas analyzer has been used

by Kuck and co-workers (27) to determine the carbon dioxide

released during a conventional combustion process after

adjusting the gas mixtures to a constant volume and a

standard reference pressure. Haber and Gardiner (14)

employed a commercial version of the electrolytic moisture

cell of Keidel (23) for the analysis of carbon and hydrogen.

The method involved quantitative conversion of carbon

dioxide to water by passing it over hot lithium hydroxide

deposited on a carborundum substrate. The subsequent

coulometric determinations of water provided absolute

measurements for both elements.

Procedures based upon measurement of the differences

in thermal conductivity between the resident gas and the

combustion products offer, in general, greater speed and

less experimental manipulation. This type of analysis was

first applied to carbon and hydrogen by Duswalt and Brandt

(9) and Sundberg and Maresh (40) who used conventional
combustion techniques coupled with a gas chromatographic

determination of carbon dioxide and acetylene, an appropri-

ate conversion product of water. The gases were trapped

in a U-tube immersed in liquid nitrogen and, when the

combustion had been completed, were introduced into a gas

chromatograph where they were separated and detected.

Oxygen and helium carrier gases were used respectively.










Vogel and Quattrone (45) devised a somewhat more complicated

procedure involving a bomb combustion of the sample in an

oxygen atmosphere and a direct chromatographic analysis of

carbon dioxide and water. A system of even greater

complexity was employed by Nightingale and Walker (31) for

the simultaneous determination of carbon, hydrogen, and

nitrogen. An induction furnace was used to burn the sample

in a helium atmosphere and the resultant nitrogen, carbon

dioxide, and acetylene (the conversion product of water)

were separated by temperature-programmed gas chromatography.

In addition to the large sample requirements the methods

described were inaccurate and time-consuming, and were

limited by basic disadvantages which will be discussed in

a later section.

In an improvement of his original work, Maresh (29)
incorporated a simultaneous nitrogen determination, and the

method formed the basis of a commercial analyzer produced

by the Fisher Scientific Company (12). The instrument

again relied upon a separation of nitrogen, carbon dioxide,

and acetylene which were initially collected at liquid

nitrogen temperatures. Inherent problems included high

operational costs and inefficiencies associated with the

inert combustion atmosphere.

Simon and co-workers (8,33) determined carbon,
hydrogen, and nitrogen by expanding the mixture of combustion









6

products into a series of filters which selectively absorbed

water and carbon dioxide. Static thermal conductivity

measurements between various portions of the system pro-

vided, under equilibrium conditions, data relating

concentration differences between composites and individual

components. The procedure of Walisch (46) depended upon

sample combustion over copper oxide in a helium carrier

gas containing 3 per cent oxygen, followed by the reduction

of oxygen and the oxides of nitrogen over copper. The

signal corresponding to nitrogen was subtracted from that

of the carbon dioxide-nitrogen composite, and water was

determined independently. The principle of analysis was

adopted by the Technicon Instruments Corporation in the

development of a commercial carbon, hydrogen, and nitrogen

analyzer (41). An instrument designed and manufactured

by the F and M Scientific Corporation (10) used a series

arrangement of two katharometers to measure the water,

nitrogen, and carbon dioxide produced by burning the sample

in a helium atmosphere.

Of the methods described, those based on a gas

chromatographic detection of combustion products provide

greater operational simplicity, are less susceptible to

human error, and are capable of substantially reducing

analysis time although the quoted times range between 10

and 105 minutes per sample. Deviations, however, from









7
traditional techniques, generally resulting from the use of

inert carrier gases to increase sensitivity, impose limi-

tations on combustion efficiency and restrict the range of

compounds capable of analysis. The singular and distinct

advantage of these systems arising from the reduction in

required methodological skills is often accompanied by

prohibitively high cost of equipment. The desirability of

a system combining maximum operational efficiency and

economy with increased speed and ease of analysis was

recognized, and it was toward this objective that these

research efforts were directed.













SECTION II


EXPERIMENTAL CONSIDERATIONS

Initial Functional Objectives and Preliminary Concepts


The procedures reported by Duswalt and Brandt (9)

and Sundberg and Maresh (29,40) first indicated the

feasibility of replacement of conventional gravimetric or

volumetric measurements of combustion products by a gas

chromatographic system. This innovation was accompanied by

little or no net increase in accuracy or speed of analysis,

and the additional requirements of a gas chromatograph and

a liquid nitrogen system limited the technique from an

economical standpoint.

It was proposed, therefore, that a system based on
a dynamic sensing device be designed which would provide a

more rapid, efficient, and economical means for the micro-

determination of carbon and hydrogen. To achieve maximum

speed of analysis, initial experimental procedures were

based upon the use of small samples and a method for flash

combustion providing rapid pyrolysis. In addition, it was

desired to avoid retention of combustion products. Evalu-

ation of this approach was facilitated by construction of











the system illustrated in Figure 1. A pipe T coupled to a

short tube filled with calcium carbide (30 mesh) was

inserted between the injection port and a 1 foot silica gel

(60-80 mesh) column of a commercial chromatograph. A

pyrolyzer probe consisting of a platinum filament connected

to a high current transformer was coated with a small

amount of benzoic acid and was fitted into the T-connector

which, in turn, was wrapped with heating tape to prevent

condensation of water. When equilibrium had been estab-

lished, a current of 55 amperes was passed through the

platinum loop for a period of 1 second, affording a measured

ignition temperature of 1000C. Using pure oxygen as the

carrier gas, combustion and conversion products were

separated and were swept through a 100,000 ohm thermistor

detector. Signals corresponding to the resultant carbon

dioxide and acetylene were similar to those of the illus-

trative chromatogram in Figure 1. Although the thermal

conductivities of carbon dioxide and acetylene are both

lower than that of oxygen, a negative peak was observed for

low concentrations of acetylene. With increasing acetylene

concentrations, W-shaped peaks occurred, exhibiting

alternate negative and positive deflections. These were

due to the maximum occurring in the thermal conductivity-

composition curve of the binary mixture 02-C2H2 at low

acetylene concentrations.

























Silica Gel Column


--Carbon Dioxide


Combustion


Pig. 1.-Preliminary System and Readout for Carbon and
Hydrogen Analysis











In spite of inherent procedural difficulties, the

system indicated the potential of a direct combustion-

detection approach to carbon and hydrogen analysis, and

subsequent techniques were based on improvement of this

method.


Selection of Components and Operational Procedures


Even though efforts were directed, initially, toward

the development of an instrumental technique for which an

eventual commercial apparatus would be readily adaptable,

alternate methods were studied to establish a system of

optimum efficiency and accuracy. Basic theoretical and

experimental considerations governing the choice of

components, conditions, and their interrelationships were

extended, where possible, to enable additional elemental

analyses with appropriate modifications.

Combustion and sample introduction systems

The preliminary experimental method for oxidative

degradation proved unsatisfactory for a number of reasons.

Sublimation or pre-vaporization of samples with high vapor

pressures was unavoidable. Efficient weighing or sample-

handling techniques could not be established. A lack of

combustion spontaneity resulted in peak asymmetry, pre-

venting the use of peak height measurement. A definite but

insufficient improvement was made by fabricating a small,











tight, conical coil of #22 gauge platinum wire and pro-

viding for its insertion into a quartz combustion tube,

thus enabling the weighing of solid samples. Passage of a

high current through the cone provided rapid but incomplete

combustion of the sample. In the case of benzoic acid, for

example, a substantial amount of the sublimate appeared on

the walls of the quartz tube. A further concept involved

the use of a weighable platinum tube inserted between two

heavy electrodes through which the carrier gas was passed.

Difficulties in these approaches included the unavoidable

disturbance of equilibrium conditions prior to the intro-

duction of each sample and severe initial losses through

volatilization or spattering due to the finite time re-

quired for the system to reach a threshold combustion

temperature.

It became evident that a stationary packed column

heated to the highest practical temperature and coupled

with an appropriate method for sample introduction would

eliminate these difficulties and would provide maximum

combustion efficiency. A Vycor tube (18 in. x 7 mm I.D.)

with a side arm (one inch from the upper end) for oxygen

flow was packed with wire-form copper oxide, and the upper

end was fitted with a serum cap. The tube was vertically

connected to a chromatographic system and was wrapped with

60 turns of #22 gauge Nichrome wire to which approximately











70 volts were applied. Careful introduction of small

varying amounts of ethanol by means of a 10 il precision

syringe produced sharp, symmetrical carbon dioxide peaks

displaying good linearity. Provision for introduction of

solids without interruption of the carrier gas flow was

made by devising a sliding valve which positioned the

sample between two parallel plates and subsequently trans-

ported it into the hot zone by means of a gravity feed as

illustrated in Figure 2. Details of an improved valve,

based on a rotary concept, which proved to be extremely

satisfactory during the remainder of the investigation,

are presented in a later section.

Further improvements in the combustion system pro-

vided ignition temperatures of approximately 10500C and

allowed furnace operation on line voltage. The tube was

packed to the half-way mark with copper oxide but was heated

at least 6 inches above this level to facilitate decompo-

sition of flashback pyrolysis products. The spacing of the

Nichrome coils was adjusted for provision of highest

temperature at the copper oxide surface, the site at which

sample decomposition occurred.

Detection systems

Determination of combustion product concentrations

by relative thermal conductivity measurements provided a

satisfactory method for initial experimental work. The





















Sample


02-


Fig. 2.-Preliminary Design of Sample Introduction System











general requirements of stability and speed and linearity

of response were satisfied, and adequate sensitivity to

carbon dioxide was obtained. However, because of the use

of oxygen as a carrier gas and the similarities in thermal

conductivities for oxygen and water, it was not practical

to measure water directly by this technique. Other detec-

tion systems were constructed and tested in an initial

effort to eliminate conversion requirements. The hydrogen

flame detector, used conventionally, shows little or no

response to water, carbon dioxide, or other permanent gases.

Using this apparent disadvantage, provision was made for

the premixing of hydrogen, containing an ethanol contaminant,

with the oxygen carrier gas and for the introduction of

this mixture into an appropriate burner-collector electrode

assembly. It was assumed that the diluent effect of the

normally non-responsive carbon dioxide and water would

result in significant negative response in comparison to

the elevated background signal provided by the constant

ethanol bleed. The procedure was discarded, however, due

to its relative complexity and critical operational

requirements.

A second approach based on conductance measurements

utilized a hygrometer element consisting of thin palladium

wires wound concentrically about a plastic core which was

coated with a small amount of lithium chloride in a











polyvinyl alcohol binder. The element was positioned in a

tube and the electrodes were connected to an a.c. bridge

circuit, the unbalance of which was representative of the

quantity of water passing through the detector. Improved

results were obtained with a more favorable cell geometry

approaching point source detection. Six parallel strands

of thin wire were evenly spaced along the inside wall of a

short length of 1/16 inch I. D. heavy-walled pyrex tubing.

A light coating of polyvinyl alcohol was applied and

alternate strands of wire were joined and connected to the

simple d.c. circuit shown in Figure 3. The response was

extremely rapid and was specific for water. Carbon

dioxide was determined by a conventional thermal conduc-

tivity detector. Linearity and reproducibility, however,

did not meet the requirements of the determination. A

variety of other detection systems were considered but from

a standpoint of simplicity and reliability, thermal

conductivity methods proved most satisfactory.

In choosing an appropriate katharometer, many factors

were taken into consideration. Although the filament

detector generally exhibits greater stability, conventional

glass-clad thermistors were selected in preference to hot

wires because of their decreased susceptibility to oxidation,

and their higher ambient temperature response in an oxygen

atmosphere. The cell was designed in such a manner that










































Fig. 3.-Circuit for Conductance Measurement of Water and
Thermal Conductivity Measurement of Carbon Dioxide










both the reference and the sample sensing elements were

located in the direct path of the carrier gas. Unlike the

diffusion cell, this provided a rapid quantitative response

which was somewhat sensitive to flow fluctuations.

Thermistors with a room temperature resistance of 100,000

ohms were used for initial studies, but later work indicated

improved stability with less expensive 2,000 ohm thermistors.

The desirable linearity of output signal with
respect to concentration is not completely predictable from

theoretical considerations. A strict mathematical treat-

ment taking into account all subsidiary effects is

complicated and impractical due to a lack of accurate data.

In general, the temperature-response behavior of a sensing

thermistor over a narrow temperature range may be repre-

sented by

RT= 1R 1E + (2 T1)

where the coefficient o is a function of temperature and

is given approximately by


T

where the constant 3 is defined by


RT2 2 1
UT =
1 .










Thus, within a small temperature interval, thermistor

resistance is a linear function of temperature. The

response, dE/dy2 to a particular component has been

shown by Smith and Bowden (34) to correspond to

dE dE dT dh
dUz aT dT3 dya U
where E is the out-of-balance bridge voltage, y2 is the

mole fraction of the non-carrier component in the detector,

T is the sensing thermistor temperature, and h is the

thermistor surface heat transfer coefficient. The dE/dT

and dT/dh derivatives are calculated from instrument

characteristics while dh/dy2 depends essentially on the

thermal conductivity of the binary gas mixture in the de-

tector. The possible effect of non-ideality of the binary

mixtures, oxygen-carbon dioxide and oxygen-hydrogen with

respect to thermal conductivity was considered. The

analogous nitrogen-carbon dioxide mixture, for example,

expresses a negative deviation from ideality as illustrated

by Keulemans (24). Departures from linearity, however, due

to these and other factors were either compensatory or

negligible in the experimental studies. This was evident

from the excellent linearity over a concentration range at

least as wide as that governed by combustion restrictions.

SResults are presented in a later section.











A two transistor zener-stabilized power supply with

a variable d.c. output voltage of 14.8-26.3 volts was con-

structed. The excellent regulation obtained in spite of

large fluctuations in line voltage is shown in Figure 19.

A Wheatstone bridge circuit, designed to accept a wide

resistance range of thermistors and to provide for suitable

sensitivity and attenuation variation, was constructed and

used in all studies.

Conversion systems

A three-fold advantage was realized in the replace-

ment of a direct thermal detection of water by the measure-

ment of an appropriate gaseous conversion product. When

retention techniques are required, anomalous desorption of

water from active column sites causes non-reproducible

tailing, and the eventual change in the substrate-solid

support ratio results in variations in retention and response.

The sensitivity of a katharometer to water in an oxygen

carrier gas is too low to provide a signal-to-noise ratio

commensurate with the precision and accuracy requirements

of the analysis. Thermostatic control of the column-

detector system at a temperature in excess of 1000C is

avoided because ambient temperatures may be used in con-

version product measurement.

The initial experimental system, previously discussed,

utilized a conversion of water to acetylene by reaction with











calcium carbide. Sensitivity was more than adequate with

respect to carbon dioxide but was inadequate with respect

to acetylene. In addition, anomalous negative response was

observed for acetylene due to longitudinal diffusion within

the column, with subsequent depression of its maximum

concentration in the carrier gas. An immediate improvement

developed from a consideration of the reaction sequence

CmHn + (m + 4)02--m C02 + H20

SH20 + CaC2 --- Ca(OH)2 + n C2H2

in contrast to which the reaction of water with an active

metal hydride


n H20 + CaH2 --- Ca(OH)2 + n H2

would provide a distinct increase in sensitivity as a result

of the more favorable stoichiometry and the higher thermal

conductivity of hydrogen. In addition, non-uniformities

resulting from peak spreading could be eliminated due to

the elution of hydrogen prior to carbon dioxide.

Detrimental effects of two reactions on quantitative
measurements of carbon dioxide were considered

2 CO2 + CaH2 -- Ca(COOH)2

C02 + Ca(0H)2--.CaCO + H20

It was proven experimentally, however, that under the

temperature and flow conditions of the analysis no carbon











dioxide retention occurred as a result of its passage

through calcium hydride or calcium hydroxide. This was

indicated by a series of careful injections of carbon

dioxide at points immediately preceding and immediately

following a tube packed with calcium hydride. Similar

injections were made after total depletion of the calcium

hydride with water. Peak heights, corresponding to carbon

dioxide concentrations, were completely unaffected by

either of the packing.

Calcium hydride was selected over a series of other

metal hydrides for several reasons. Its reaction with

water to release hydrogen was rapid and quantitative. From

a practical standpoint, it was found to be more easily

handled than the sodium, lithium, or lithium-aluminum

analogs; i.e., a container of calcium hydride could be

opened and exposed briefly to the atmosphere innumerable

times without observable decomposition. Its hydroxide

displayed the least tendency to retain water, as evidenced

by the pronounced reduction in tailing over that experienced

with the other hydroxides. This corresponded to prediction

because calcium hydride is neither hygroscopic nor does it

form a hydrate. Its small but definite affinity for water

under the conditions of the preliminary systems, however,

was sufficient to produce critical design problems.

An initial conversion system consisted of a short











pyrex tube equal in diameter to the combustion tube and

packed with approximately 1/4 inch of 40 mesh calcium

hydride. The peak height response was observed to decrease

markedly after the combustion of three or four samples.

Asymmetric peaks were particularly evident in the case of

large samples or those with a substantial hydrogen content

and were attributed, in part, to the observable condensation

of water prior to its passage through the calcium hydride,

resulting in extremely slow evolution of hydrogen. A

corrective procedure was attempted in which the conversion

tube was placed in an oven controlled at 110C. The system

was first tested for evidences of thermal degradation of

the calcium hydride by plotting the effluent signal versus

temperature as indicated by a thermometer attached to the

tube. No hydrogen evolution was noted during elevation of

the oven temperature to 1300C. Although some improvement

due to elimination of condensation problems was observed,

no noticeable gain in reactivity or reduction in tailing

accompanied the temperature increase. Resultant erratic

values for hydrogen precluded the use of this system.

A unique application of an apparatus patterned after

the electrolytic moisture detector of Keidel (23) was

evaluated in an effort to obtain a more spontaneous release

of hydrogen. A hygrometer element, identical to the one

previously described under "Detection systems" was coated











by immersion in syrupy phosphoric acid. Subsequent de-

hydration by applying a voltage to the electrodes until no

further hydrogen evolution was observed left a thin film of

phosphorus pentoxide. The element was encased in a glass

sleeve and was positioned directly below the combustion

tube. Carbon dioxide generated during a combustion process

was unretained and was detected in the normal manner. The

water, however, was irreversibly absorbed by the phosphorus

pentoxide. Application of a relatively high voltage to the

electrodes resulted in the rapid electrolytic release of

oxygen and hydrogen into the oxygen carrier stream. The

hydrogen was detected conventionally by thermal conductivity.

The conversion, however, was found to be incomplete and

losses were incurred through inefficient collection of water

on the phosphorus pentoxide.

A number of additional chemical conversions were

investigated, including the reaction of water with aluminum

carbide to produce methane. This proved ineffective due

to the extremely slow reaction rate. Rejection of other

systems was based also on reactivity or detection deficien-

cies.

A simple and accurate method for measurement of water

was eventually devised which eliminated virtually all of the

difficulties previously encountered. A standard unsealed

Kimax melting point capillary tube was fitted at one end











with a short plug of asbestos and was packed with a 1 inch

layer of silica gel (60-80 mesh). Tight connection to the

combustion tube was made through a serum vial cap, and a

1/8 inch I. D. pyrex tube packed with a 1/4 inch layer of

calcium hydride was positioned directly below the capillary

tube by means of a silicone rubber insert. The calcium

hydride was pushed tightly against the seal to eliminate

all possibility of condensation prior to conversion. A

small moving furnace designed to operate at a constant

temperature of approximately 2000C encircled the capillary

tube as indicated in Figure 4. The velocity of the carrier

gas through the capillary was nearly twenty times greater

than its velocity through the combustion tube or through

previously designed conversion tubes. This was indicated

by the reciprocal ratio of the respective cross-sectional

areas. Similarly, a five-fold increase in oxygen velocity

was calculated for the calcium hydride tube. The extremely

high transfer rate of.carbon dioxide through this portion

of the system eliminated any possibility of its retention

by the silica gel or its reaction with calcium hydride or

calcium hydroxide. Water, on the other hand, was retained

with excellent efficiency by the silica gel and in no

instances, regardless of sample size or hydrogen content,

was an overload of the desiccant noted. The slight conden-

sation on the walls of the capillary tube, which was















SSilver Wool

---- Combustion Tube

--Rubber Serum Cap




S--- Moving Furnace




Capillary Tube






j;i---Silica Gel Packing


-I-- Silicone Rubber Sleeve
Calcium Hydride


Glass Wool Retainer


Fig. 4.-Details of the Retention and Conversion System











sometimes observed immediately after a sample combustion,

was of no consequence since all water droplets were swept

into the silica gel before the appearance of the carbon

dioxide peak. Positioning the moving furnace at the upper

portion of the capillary tube as well as the use of silver

wool as a heat transfer agent in the combustion tube aided

significantly in the prevention of condensation. In an

actual analysis the sample was introduced with the moving

furnace remaining in the upper position. Immediately after

observance of the carbon dioxide peak maximum (30-45

seconds), the furnace was dropped to its lower position

where it was seated against the silicone rubber insert for

3-5 seconds, vaporizing the entrapped water almost

instantaneously. Attenuation and recorder lead polarity

were adjusted and the resultant signal for hydrogen appeared

within 30-45 seconds. Incorporation of the calcium hydride

in the capillary tube directly below the silica gel proved

unsatisfactory due to prohibitively large flow restrictions

arising from the expansion accompanying the conversion to

calcium hydroxide. The use of a larger tube prevented this

difficulty, facilitated rapid hydride replacement, and

insured uniform release of hydrogen.

Readout systems

It was presumed at the start of the investigation

that a readout system providing signal integration over the











time coordinate would be required for an analysis. As work

progressed, it was observed that combustion rates were suf-

ficiently constant, for virtually all substances encountered,

that peak height measurements could be used advantageously.

This provided a simpler and more rapid readout, offered a

wider chart span on the same recorder, eliminated the

additional cost of integrating equipment, and facilitated

adaptation to a digital readout system, if desired.

A Minneapolis Honeywell 1 millivolt Brown recorder

with a response time of 1 second, full scale, was used for

all reported results. The specified accuracy, of + 0.25 per

cent was recognized to be a limiting factor although perfor-

mance data for the system indicated this to be a somewhat

conservative figure. Because the maximum bridge out-of-

balance voltage was in excess of 10 millivolts, a Sargent

Model TR recorder was preferentially selected for use with

the prototype instrument discussed in the Appendix, by

virtue of the + 0.1 per cent accuracy specified for this

range. The Sargent recorder offers the additional advantage

of an improved pen system and a displacement adjustment in

spite of a slightly reduced chart span (9% in. vs. 10 in.).

Response magnitudes for carbon dioxide and hydrogen are

approximately proportional to the differences between their

respective thermal conductivities and that of oxygen.










Setting 02,02) 1 we find An(H2,02) 15 at 25C.

For most compounds a two- to four-fold attenuation was

required to adjust the amplitude of the hydrogen signal to

approximately that of carbon dioxide. A double-pole double-

throw switching mechanism was provided for reversal of

recorder-lead polarity, enabling the use of the full chart

span for both carbon dioxide and hydrogen peaks.

Column systems and arrangement of components

The preliminary experimental procedure relied upon
the use of a silica gel column for the separation of carbon

dioxide and acetylene. In later work, a silica gel packing

was also used to separate hydrogen from carbon dioxide.

Two basic arrangements of components, and the resultant

readouts, with no electrical reversal of polarity, are

shown in Figure 5. The second system was devised for a

more completely automatic approach and provided a re-

inversion of the hydrogen signal thus facilitating the use

of automatic attenuation. The procedure involved passage

of the combustion and conversion products through the silica

gel on which carbon dioxide was initially retained while

hydrogen was swept through the first side of the detector

producing a negative signal. During the flow of hydrogen

through the delay column, the carbon dioxide was eluted

from the silica gel producing a positive signal on the re-

corder, and then flowed into the ascarite filter where it









--Sample Port




--Combustion Tube




Calcium Hydride

Silica Gel Column

Detector

02


-Sample Port


SCombustion Tube





Calcium Hydride

Silica Gel Column


-- Detector

Ascarite

02- Delay Column
02--


Fig. 5.-Variations in Component Arrangements and Readouts
for Carbon and rlo;rogen Analysis










was permanently retained. The delay column which consisted

of approximately 50 feet of empty 1/8 inch plastic tubing

was so adjusted in length that after the first side of the

detector was cleared of carbon dioxide, the hydrogen passed

through the second side of the detector, thereby producing

a positive signal on the recorder. Diffusion of hydrogen

was almost negligible during its long transport through

the column, but analytical results suffered from such

factors as non-uniform release of hydrogen, increased column

impedance due to carbon dioxide absorption by the ascarite,

and non-linearity of response to carbon dioxide. Carbon

dioxide peaks were non-Gaussian, exhibiting pronounced

asymmetry with increasing sample size. No specific improve-

ment in elution symmetry was noted upon replacement of the

silica gel with a 6 foot column packed with 30 per cent

hexamethylphosphoramide on chromosorb. A number of problems

associated with the combined usage of oxygen as a carrier

gas and a silica gel column for retention of carbon dioxide

became apparent. Although sensitivity to hydrogen was more

than adequate, it was necessary to utilize the maximum

bridge output with a 1 millivolt recorder to obtain signifi-

cant peak amplitudes for carbon dioxide. Unfavorable

signal-to-noise ratios were observed under these conditions,

and a large attenuation ratio was required for measurement

of the two components. Column overload at higher concen-

tration levels caused deviations from linearity requiring










corrective working curves, and excessive peak widths for

carbon dioxide resulted in undesirably long analysis times.

Of even greater consequence were indications of changes

in column characteristics during extended operation of the

system, resulting in variations in both retention volume

and response.

To test the capabilities of a more ideal system

requiring no chromatographic separation, the calcium

hydride was replaced by a desiccant, and a 1/8 inch x 8

foot length of empty plastic tubing was substituted for the

silica gel column. This allowed observation of performance

characteristics with regard to measurement of carbon

dioxide, initially circumventing any problems associated

with the determination of water. Polyethylene samples were

used for the test runs due to weighing convenience. Figure

6 indicates the improvement in sensitivity provided by this

system as compared to that using a silica gel column. Flow

rates were identical (40 cc/min.) and, in each instance,

the time required for the initial appearance of the carbon

dioxide peak was virtually the same. The seven-to-eight-

fold increase in sensitivity was accompanied by excellent

linearity throughout an even wider range of concentrations

than was normally suggested (further data are presented in

a later section), and stability, readout ease, and speed

of analysis were vastly improved. The accuracy afforded by

















1.00-

0.90 -

0.80

0.70


0.60 Unpacked Column

* 0.50

S0.40

o / Packed Silica Gel Column
o 0.30

0.20

0.10

0
0 0.1 0 0. 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Weight of Polyethylene, mg



Fig. 6.-Comparison of Sensitivity to Carbon Dioxide Using
Packed and Unpacked Columns Under Identical Flow
Conditions










the combined combustion and detection techniques was next

tested, and results are presented in Table 1.


TABLE 1

TEST RESULTS FOR ACCURACY OF CARBON ANALYSIS


Compound %C, Found %C, Theoretical

B-Naphthol 83.5 85.3
Naphthalene 93.6 93.7
Benzophenone 85.8 85.7
Oxalic acid 19.0 19.0
Phthalic acid 57.8 57.8
9-Fluorenone 86.6 86.6
Triphenylphosphine 82.4 82.4



Sample handling and weighing

A Cahn Gram Electrobalance (Model G, Cahn Instrument

Company, Paramount, California) was selected for all sample

weighing. The general performance of the balance proved

to be outstanding due to its insensitivity to environmental
conditions such as vibration, temperature variation, air

currents, etc. The specified limits of precision and
accuracy for the weight range of interest are 0.1 micrograms
and 0.5 micrograms, respectively. Exact proportionality,
however, rather than absolute weighing accuracy is essential
to this work. Because very critical weighing were required
the balance was seroed with each determination. The










calibration was found to remain extremely constant, but

was checked, nevertheless, after every five or six

determinations. For ultimate precision, the 0-1 milligram

range of the balance was used for all weighing, even

though, in some instances, the total weight exceeded 1

milligram. When it became necessary to encapsulate a

sample, a balancing capsule weighing slightly less than

the sample capsule was placed on the righthand pan, and

the net tare weight was obtained through electrical sub-

traction by the zero circuit. If, in this procedure, the

total weights of the net tare and the sample were in excess

of 1 milligram, the 1 milligram calibrating weight was

added to the righthand pan and the additional weight was

obtained in the usual manner.

For non-crystalline or viscous liquid samples,

containers were fabricated from 6 mm squares of aluminum

foil formed about a 1.5 mm rod and crimped on one end.

After sample introduction the opposite end was crimped and

the weighing was completed. A preferred method of sample

enclosure required the use of seamless copper capsules of

equivalent dimensions. These facilitated the weighing of

volatile liquid samples and continuously provided a fresh

layer of copper oxide rather than aluminum oxide at the

combustion site. They were not used during the course of

the investigation due to difficulties in fabrication.










However, it was later found that, where an extrusion process

failed, an electrodeposition of copper on a slightly tapered

iron mandrel proved quite satisfactory. The high co-

efficient of thermal expansion of copper allowed removal of

the capsule by immersion in boiling water.

Critical Parameters in the Finalized System and
Extension to Nitrogen Determination

Evaluation of all factors previously discussed led

to the construction of an apparatus using the flow system

illustrated in Figure 7. This arrangement of components

was selected by virtue of expediency and its provision for

maximum experimental versatility. A similar, but somewhat

more advantageous arrangement was used in the design of the

prototype instrument discussed in the Appendix.

Choice of oxygen as a carrier gas

The necessity for an oxygen atmosphere to provide

maximum combustion efficiency for carbon and hydrogen

analyses was recognized by Liebig and Pregl in their

initial studies. In the present system where an extremely

rapid and uniform release of combustion products is

essential, an inert carrier gas such as helium or nitrogen

can not be used for general application. The inadequacy of

a non-oxidizing carrier gas is shown in Figure 8. Nitrogen

was selected for this test rather than helium due to the

















--- Sample Port H2
CO 2

---Combustion Tube





-- Moving Heater
-- Capillary Tube
Silica Gel
Calcium Hydride
S-- Filter

-- Delay Column
Detector

02
H2 Release
and Polarity
Reversal



Fig. 7.-Finalized Flow System and Readout for Carbon and
Hydrogen Analysis








38


--- I-- :





S -O---
I 4I-
: Ia' ie 'G r G: s

*.. .- d i -7 ; t .....
--- I --
0 : o


a--i--i 1 .-. I 'd ":d 1
-- 0 q. -
















at 40.0 cc/sin, at 40.0 cc/min.

Jig. 8.-Comparison of Combustion Efficiencies Using Oxygen
and Nitrogen Carrier Gases
Fig. 8.---m i of C ie Ui Oxygen










and Nitrogen Carrier Gases










near identity of its thermal conductivity with that of

oxygen. The carbon dioxide response factors for benzoic

acid were similar; R02 1.13 mv/mg and %2 1.10 mv/mg.

However, upon introduction of a sample of activated carbon

(approximately 60% C), RO2 = 0.98 mv/mg whereas virtually

no response was observed for an even larger sample of

carbon using a nitrogen carrier gas. A second carbon

sample was encapsulated and was surrounded with approxi-

mately 5 milligrams of silver permanganate. The carbon

dioxide peak height response in the nitrogen carrier gas

was, in this case, 0.28 mv/mg with some evidence of peak

spreading.

Further information regarding combustion spontaneity

in oxygen was derived from the following data. Peak height

to half-peak-width ratios were calculated for a series of

samples ranging in ease of combustion. Sample weights were

chosen so that peak height magnitudes remained essentially

the same. The maximum chart speed was used to allow more

accurate half-peak-width measurements. The results, as
shown in Table 2, indicated no inconsistencies or peak
spreading. Oxygen provided the additional advantage of

rapid regeneration of copper oxide which may have been

reduced during the combustion process.

A differential thermal analysis of copper oxide

versus carborundum was conducted in a helium flow. The












TABLE 2
PEAK HEIGHT TO HALF-PEAK-WIDTH RATIOS


Peak Half-Peak
Height Width
Sample Wt,pg mm mm PH/HPW

Benzoic acid 661 196 4.2 46.7
Biphenyl 497 201 4.3 46.8
Potassium acid
phthalate 906 184 3.9 46.2
Caffeine 902 195 4.2 45.9
Teflon 1730 176 3.9 45.1
Activated carbon
(~-60 per cent 0) 722 187 4.0 46.8











resultant endotherm occurring initially at 925*C suggested

a partial degradation of the copper oxide at this temperature.

This was substantiated in later work by evidences of oxygen

release under similar conditions. A second differential

thermal analysis of copper oxide, substituting oxygen for

helium, showed no thermal band below a temperature of

10550C. In excess of this temperature an endotherm corres-
ponding to a slow fusion process was noted. Another

important advantage, therefore, was realized in the use of

oxygen rather than an inert carrier gas because its

repression of copper oxide decomposition allowed the use

of a higher furnace temperature with a resultant increase

in combustion efficiency.

Combustion tube packing and retention systems

The oxidizing portion of the combustion tube for a

conventional system for microanalysis of carbon and hydrogen

is packed with a mixture of 2 parts of copper oxide and 1

part of lead chromate, according to the procedure suggested

by Steyermark (38). The furnace temperature for this

system, however, is relatively low (6800C) as opposed to

that of the method under investigation (>10000C). Experi-

mental results with lead chromate-copper oxide mixtures

indicated that it was unnecessary, in the present system,

to add an additional oxidant to the copper oxide. Later

work, however, proved the necessity for addition of a











section of platinized asbestos to the combustion tube in a

similar technique for nitrogen determinations. This

functioned to catalyze the combustion of any unburned

pyrolysis products. Although the use of an oxygen atmos-

phere substantially reduced this possibility, it was

considered advantageous to include a layer of platinized

asbestos in the carbon and hydrogen combustion tube.

Chlorine, bromine, iodine, and oxides of sulfur were

initially retained on a 1%i inch section of tightly rolled

silver gauze positioned in the lower portion of the

combustion tube. This, however, provided a relatively small

surface area for reduction and required indentations in the

walls of the tube to prevent shifting. On the other hand,

silver wool was found to retain its position, to provide a

much larger surface area, and to be much easier to clean.

It functioned also, as an excellent heat transfer medium,

assisting in the prevention of condensation of water in the

lower portion of the tube. Compounds with a high fluorine

content were analyzed successfully with no special pre-

cautions. However, for the continual analysis of fluoro-

compounds, a layer of magnesium oxide or aluminum oxide

positioned between the platinized asbestos and the copper

oxide is advocated.

Combustion of nitrogen-containing compounds in an

oxygen atmosphere would be expected to produce, essentially,











a mixture of nitrogen dioxide and its dimer, dinitrogen
tetroxide, which would include the oxidation product of any
nitric oxide initially formed. Certain organic, such as
diazo-compounds, would be presumed to decompose with the

release of elemental nitrogen, and the combustion temperature
would be far too low for any oxidation to the monoxide.
Because the thermal conductivities of nitrogen and oxygen

are nearly identical, virtually no response to elemental
nitrogen would be superimposed upon the carbon dioxide
signal. A number of reactants have been proposed for the

retention of nitrogen dioxide. Traditionally, lead peroxide

maintained at a temperature of 175-1900C is used (37), and
absorbs nitrogen dioxide according to the reaction

2 NO2 + 2 Pb02--Pb(N02)2 + PbO + 1 02

More recently, specially prepared manganese dioxide has

been used for this purpose. Onoe (32) found that activated
manganese dioxide prepared from manganese sulfate and
ammonium peroxydisulfate or from potassium permanganate and
methanol gave the best results, and concluded that the
reaction was simply

MnO2 + 2 NO2 -Mn(N03)2

Vecera and Snobl (43) maintained that the hydrated amorphous
gel [MnO(OH)23 binds nitrogen dioxide with its surface
hydroxyl groups according to the equation










Mn(OH)4 + N204-.Hn(NO3)2 + 2 H20 ,

the amount of absorbed nitrogen dioxide being in proportion

to the surface. They observed excellent absorbent
properties for this gel and described a procedure for its

preparation. Experimentally, a short filter packed with a

microanalytical grade of manganese dioxide followed by a

layer of magnesium perchlorate was used. Its efficiency

was tested by injecting 5 cc volumes of nitrogen dioxide

and sulfur dioxide immediately before and after the filter.

Complete retention was found even after numerous repeated

injections.

It was also necessary to provide a desiccant prior

to the combustion tube to remove traces of moisture present

in the oxygen carrier gas. This was particularly essential

because the silica gel capillary tube served to accumulate
water over extended periods of operation. Trusell and Diehl

(42) indicated the superior desiccant properties of anhydrous

magnesium perchlorate and, in this application, it was found

preferable to the 5A molecular sieve originally used. For

oxygen containing an appreciable amount of hydrogen or

hydrocarbons, initial purification using a pre-burner and

a carbon dioxide-water absorption system would be required.

Thermal equilibrium and flow rate
Two factors contributed significantly to variations











in detector response. The influence of block temperature,

a function of the time allowed for the detector to reach

thermal equilibrium, is shown in Figure 9. Because

conduction of heat away from the thermistors increases the

temperature of the cell block until an equilibrium

temperature is reached, a decrease in response to carbon

dioxide with time is anticipated due to the convergence of

the thermal conductivities of oxygen and carbon dioxide at

higher temperatures. Little or no response change is

expected or observed for hydrogen since the ratio of the

thermal conductivities for hydrogen and oxygen remain

essentially constant over a wide temperature range. It

is evident, however, that the system should be allowed to

"warm up" for at least one hour before an analysis is run,

and that frequent standardizations are required for the

first two and a half-hour period.

The necessity for a critical control of the oxygen

flow rate is indicated in Figure 10. The diminishing

response to carbon dioxide corresponds to that normally

encountered when a carrier gas flow rate is increased to a

point where the residence time of the component in the cell

has a significant effect on the resultant signal. It is

apparent, in this instance, that if measurements of carbon

dioxide were based on peak areas rather than peak heights,

any errors due to flow variations would be severely




















1.65



1.60



1.55


S 1. 50-


1.45

N 1.45

0 1 1 1 1 I I1 1
0 0 20 40 60 80 100 120 140 160 180 200

Time Elapsed, Min., After Switching on Thermistor
Detector



Fig. 9.-Response to Carbon Dioxide Versus Equilibration Time






























o 002 Response

* H2 Response


Oxygen Flow Rate, cc/min.


Fig. 10.-Dependence of Response on Flow Rate


1.40


0
"g 1.355

o

1.30
0






1.25
P4


S1.20



1.15
e











magnified since peak widths and peak heights decrease

simultaneously with increased flow. A similar decrease in

response to hydrogen was observed but only at much higher

flow rates. The response maximum for methane in oxygen

was found to be intermediate between that for carbon

dioxide and that for hydrogen.

Slight variations in flow impedance occur within the

system for numerous reasons. Changes in line voltage alter

the furnace temperature resulting in expansion or contraction

of the copper oxide. Increases in the amount and the degree

of compression of calcium hydroxide in the converter tube

as well as expansion due to formation of calcium hydroxide

produce additional flow restrictions. Rapid increase in

temperature of the silica gel in the capillary tube with

subsequent expansion introduces transient flow changes. In

order to minimize or to eliminate these effects a throttling

orifice was constructed and installed in the system. The

flow of gas through an orifice depends only to a certain

extent on the pressure differential across the orifice.

Where the upstream pressure is held constant and the pressure

at the discharge end is decreased, the exit gas flow will

increase to a critical value. Further decreases in the

downstream pressure will cause no increase in flow until the
critical flow condition has been exceeded. The critical

condition is obtained for an oxygen flow through an orifice











if the upstream pressure is approximately two or more times
that of the downstream pressure (2). The equation for
critical flow is

q -CApK VU/ ,

where q is the mass rate of flow of the gas; 0 is the dis-

charge coefficient; A is the cross-sectional area of the

orifice; p is the upstream static pressure of the gas; K
is a dimensional coefficient involving the gas constant;
M is the molecular weight of the gas; T is the absolute

temperature of the gas upstream; and U is a constant that
is characteristic of the gas and is given by

U = k[2/(k + 1)] ,

where k is the ratio of specific heats (C p/C) and

a (k+l) / (k-l) .

An orifice of approximately 0.0025 inch diameter was
developed in a thin steel disk using the method suggested
by Marks (30). The ratio of the absolute upstream pressure
to the downstream pressure was measured and was found to be

roughly equal to 2 for this orifice. The equation for
critical flow indicates that a 5 change in temperature
of the gas upstream would produce a corresponding change
in flow of + 0.8 per cent. Therefore, in the final proto-
type model, the orifice and the secondary pressure regulator
were positioned in the constant temperature section of the


instrument.











Nitrogen determination

A simple modification of the finalized system for

carbon and hydrogen analysis provided a rapid method for

the determination of nitrogen. In initial studies the

combustion tube was repacked with a 4- inch layer of copper

oxide (20 mesh) followed by a 4 inch layer of freshly

hydrogen-reduced copper (20 mesh) and a 1 inch layer of

silver wool. The conversion system was removed and, when

helium was used as a carrier gas, the manganese dioxide

cartridge was replaced by a filter packed with magnesium

perchlorate and ascarite. Only the desiccant was required

for a carbon dioxide carrier gas, and the combustion tube

was connected directly to the filter.

Sample analysis was identical to that in the carbon

and hydrogen determination with the exception that a

catalyst was generally added. Combustion products were

swept through the copper section of the tube where oxides

of nitrogen were reduced to elemental nitrogen. Halogens

and oxides of sulfur were retained by the silver, and water

and carbon dioxide were absorbed where required. The

nitrogen peak emerged in 30-45 seconds after introduction

of the sample. The reaction
n [ o 0
C0mn (+ (M + 02 -C1) m2 + H20 + 2 N2

is less favorable, stoichiometrically, for nitrogen than











for carbon dioxide formation. For those compounds containing

high percentages of nitrogen, a carbon dioxide carrier gas

afforded sufficient sensitivity for the analysis, but it

was found advantageous to use helium for general application.

Good combustion characteristics, with respect to nitrogen

release were observed for most compounds even though an

inert carrier gas was used. It is reasonable to assume

from thermodynamic considerations that, during the com-

bustion process, free nitrogen would be formed even more

readily than carbon dioxide or water due to its higher

stability.

The small amount of air which is introduced with the

sample is of no consequence in the system for carbon and

hydrogen because oxygen and air have virtually identical

thermal conductivities. In the case of the nitrogen

analysis, however, where helium or carbon dioxide is used

as a carrier gas, a significant error may be introduced.

The slight Venturi effect within the valve draws the en-

trapped air into the system and the subsequent peak is

partially superimposed upon the nitrogen peak which precedes

it. A modification of the procedure for sample introduction
reduced this contribution to a minimum. The valve (see

Figure 17) was first rotated to the extreme clockwise

position to clear the inner sample compartment of air.

After a few seconds the encapsulated sample was dropped into










the upper compartment which was subsequently filled with a

loosely fitting aluminum plug. The valve was then quickly

rotated to the extreme counterclockwise position and back

to the extreme clockwise position. Because the helium in

the inner compartment was under slight pressure, diffusion

of almost all outside air was prevented. The magnitude of

the resultant air peak was generally at least twenty times

less than that encountered during normal operation, and the

peak was often completely eliminated as in the case indi-

cated in Figure 11. Due to the somewhat slow diffusion

process the peak was considerably broad, and followed that

corresponding to sample nitrogen with only slight overlap.

Therefore, the contribution of atmospheric nitrogen was

negligible providing the valve was operated in the manner

described.

High "pseudo-nitrogen peaks" were initially obtained

for several non-nitrogenous compounds such as sucrose,

dextrose, and citric acid. It was later discovered that

injections of methane produced the same effect which was

attributed to incomplete combustion and subsequent detection

of certain low molecular weight pyrolysis products. This

difficulty was completely eliminated by the addition of a

layer of platinized asbestos to the copper oxide portion

of the combustion tube packing.


















i--i-- -; i---- :--
--: ....--- -_- --: '---o --'', -- --4-- -i----'-



'.-- Kl Nitrogen Peak Resultant from Introduction
of Air During Normal Operation of Valve

.. L _. ) ..... -
-- .. .. ... o .. .!. .. .. .... ... .. .


- E Elimination of Air Peak By
-i--i- -- Modified Operation of Valve



---K --


Fig. ll.-Reduction of Atmospheric Contamination During
Nitrogen Analyses


a
4,
St)
- --

0
:
- ^
; W j_











Addition of catalysts

In only very few instances was it found necessary

or advantageous to add a combustion catalyst to the sample

being analyzed for carbon and hydrogen. Under less favorable

conditions, where larger samples and lower temperatures are

employed, those compounds containing the alkali or alkaline

earth metals in the absence of sulfur or phosphorus would

leave an ash of the carbonate resulting in low carbon

values. Steyermark (36) suggests the addition of potassium

dichromate or vanadium pentoxide in the analysis of these

compounds. Efficiencies of a large number of catalysts

have been evaluated (15,18,19,20,21,44) but it is generally

agreed that tricobalt tetroxide, cobaltic oxide plus silver,

and silver permanganate are among the most satisfactory.

Silver permanganate was prepared by the reaction of silver

nitrate and potassium permanganate, and was used as a

general catalyst where required.













SECTION III


CONSTRUCTIONAL AND OPERATIONAL DETAILS

Experimental System


The apparatus used for this study was assembled so

that a variety of modifications could be made simply and

rapidly. Major components were mounted on a network of

1/2 inch diameter rods which insured freedom from vibration.

Ambient temperature was not controlled, but remained

essentially constant at 25*0. It was found necessary to

provide thermal shielding for only three components; the

pre-filter, the sample introduction port, and the detector.

The instrumental arrangement is shown in Figures 12

and 13. U.S.P. grade oxygen (Linde Company, New York,

N.Y.) was used throughout the investigation and all con-

tacted surfaces were thoroughly cleared of oil, grease, or

other contaminants. A standard two stage oxygen pressure

regulator (Linde Company) equipped with a 1/4 inch straight

thread hex-nipple was used for upstream pressure control.

A female 1/4 inch two-piece swivel hose connector was

modified to house the throttling orifice as shown in Figure

14. The inner section of the hose end portion was machined





1 II I


PNIi
PP I P hI


Fig. 12.-Instrumental Arrangement for Carbon and Hydrogen Analysis







I-

ii i ~i

Ii


i 'NI


Fig. 13.-Details of Conversion and Detection Systems


- I


5


57












































Fig. 14.-Details of Orifice Mounting










to a flat surface to accommodate two 7/16 inch neoprene

o-rings between which the orifice disk was positioned. The

hole was made by denting the steel disk (0.010 in. x

7/16-in. D.) with a sharp punch and by carefully filing

the back until a breakthrough was obtained. The opening,

as observed under a microscope, appeared to be quite round

and was estimated to have a diameter of 0.0025 inch by

comparison to a wire of that size. Connection to the de-

tector was made with heavy-walled tubing and all further
connections up to the combustion tube were made as short

and rigid as possible.

The detector block was machined from 304 stainless

steel and is shown in Figure 15. Individual empty cell

volumes were approximately 0.4 cc. Type G126, 2,000 ohm,

hermetically sealed Fenwal thermistors (Allied Electronics

Corporation, Chicago, Ill.) were seated in position. These

required no sealing washers. The leads were insulated with

teflon sleeves, and the cell was tested for leaks at an

oxygen pressure of 50 p.s.i.g. The detector assembly was

mounted in an aluminum minibox (2 3/4 x 2 1/8 x 1 5/8 in.),

and the spaces between the walls were tightly packed with

glass wool. The outlet from the reference cell was connected
directly to a plain end flow metering tube (tube size 1A-

15-1; Ace Glass Company, Vineland, N. J.) with a heavy wall
rubber coupling. The rotameter was connected, in turn, to








60









r----~i7~-~: I-:-------




Glass
Wool---


L.------------------I
END VIEW






Aluminum -1/8 in. Stainless
Minibox Steel Tubing








S--


L-----------------------
SIDE VIEW

Material: 304 Stainless
Steel
Scale: Full


Fig. 15.-Details of Cell Construction










a pyrex tube (18 in. x 7 mm O.D.) packed with 20-60 mesh

magnesium perchlorate (Anhydrone; Coleman Instruments, Inc.

Maywood, Ill.) and 8-20 mesh ascarite (Arthur H. Thomas

Company, Philadelphia, Pa.) as indicated in Figure 16. The

tube was wrapped with a sheet of aluminum foil to reflect

heat radiated from the furnace in order to prevent thermal

expansion and contraction of the packing. A short heavy

rubber connector was used to join the filter tube and the

sample introduction port. Details of the valve assembly

and construction are presented in Figure 17. The body and

the central rotating plate were machined from 304 stainless

steel, and the upper and lower rotary port seats were

machined from Kel-F. All moving surfaces were hand-lapped

to prevent leakage. The 1/4 inch outlet allowed rapid

connection and replacement of the combustion tube by means

of a short piece of rubber tubing. The connector was pro-

tected from the heat of the furnace by wrapping it with

aluminum foil. Because it was essential to maintain the

injection port at room temperature to prevent pre-vapori-

zation of volatile samples, a large aluminum heat reflector

(6 1/2 x 10 in.) was positioned directly below the valve.

The "ycor combustion tube (18 in. x 7 mm O.D.,

743172, Corning Glass Works, Corning, N.Y.) was packed with
copper oxide (Cuprox; 29-130, Coleman Instruments, Inc.),

silver metal wool (E. H. Sargent and Co., Chicago, Ill.) and

















0







Cd a


0 0
P 4I
ro










o
Fig. 16.-be Packings for
Conversion
0




0




Oxygen Combustion
Filter Tube for
Carbon and
Hydrogen
Determi-
nation

Fig. 16.-Tube Packings for
Conversion


0 I'. ,


0 E-4I 0)

cd
U6- U H
I 3,
soi
g- Cf H
CD
N Iw
so
*rf -^ 1
HO




0 -
Pd 0
4 00
(0.0 = C r

(U1 Or1 !
to

W
Z P4


o



o :);




Combustion
Tube for
Nitrogen
Determi-
nation


Nitrogen Di-
oxide Filter

4)

oI I





Carbon Dioxide-
Water Filter
for Nitrogen
Determination


Combustion, Absorption, and
































TOP VIEW


S! > tKel-F
--------------






--- in. Stainless Steel Tubing


SIDE VIEW

Material: 304 Stainless
Steel
Scale: Full

Fig. 17.-Details of Sample Introduction Port










5 per cent platinized asbestos (P-152, Fisher Scientific
Co., Pittsburgh, Pa.) as shown in Figure 16. For nitrogen

analysis, an additional packing of metallic copper (Cuprin;

29-120, Coleman Instruments, Inc.) was used, as indicated.

An open end Kimax Capillary Tube (1.5 x 100 mm) was
packed with 1 inch of 60-80 mesh silica gel (Coast Engineer-

ing Company, California) and was connected to the combustion
tube by insertion through a 6 mm rubber sleeve type serum

stopper. The conversion tube (1 3/4 in. x 4 mm 0.D.) was
packed with about a 5/16 inch layer of 40 mesh calcium
hydride (Metal Hydrides, Inc., Congress Street, Beverly,
Mass.) as shown in Figure 16. The hydride layer was re-
tained by a tight plug of glass wool and was positioned
firmly against a 1/8 inch silicone rubber sleeve (0.125 in.

O.D., 0.062 in. I.D.; Reiss Manufacturing Corporation,
Little Falls, N. J.) which was resistant to the heat and
allowed rapid removal and replacement of the tube.

The filter for retaining oxides of nitrogen, con-

sisting of a pyrex tube (3 in. x 7 mm 0.D.) packed with
14-30 mesh activated manganese dioxide (33-110, Coleman

Instruments, Inc.) and magnesium perchlorate, is illustrated
in Figure 16. This was connected to a 6 foot length of
plastic tubing (3/16 in. O.D., 1/8 in. I.D.) which was
coiled around a cylindrical form. Plastic was chosen rather

than metal to reduce the possibility of surface adsorption











of hydrogen. A short length of rubber tubing joined the

column to the inlet of the reference detector, and the out-

let was connected to a flow meter. The flow meter was of

the soap-film type and was constructed from a 5 cc Mohr

measuring pipet by cutting off the ends and attaching a

glass T fitted with a medicine dropper bulb. This allowed

more accurate and more rapid flow measurements than were

possible with conventional higher volume soap-film meters.

A stop watch with a 10 second sweep which could be read to

the nearest 0.02 second was used to check all flow rates.

An 11 foot length of #22 gauge nichrome wire was

used for the furnace winding. This provided 110 turns

about the 7 mm Vycor combustion tube and allowed 3 1/2

inch double-twisted leads at either end for reduction of

lead resistance. Small bolts were used to connect the

nichrome heater to the line cord. The coil was compressed

at the point corresponding to the surface of the copper

oxide and was expanded at either end. This provided

maximum heat at the sample contact point. Using an optical

pyrometer, the temperature of the copper oxide at line

voltage (- 117 volts a.c.) operation was observed to lie

between 1000 and 10500C. For the capillary tube heater,

approximately 11 inches of #24 gauge nichrome wire were

wound 34 times about a 5/32 inch mandrel leaving 1 inch

double-twisted leads at either end. The spiral was encased











in a pyrex tube which was held in position with a standard

laboratory thermometer clamp. Vertical adjustment was

provided by sliding the clamp along a 1/2 inch diameter rod.

A Powerstat (The Superior Electric Company, Bristol, Conn.)

was used to supply the required 6 volts. Operational data

for both heaters are presented in Table 3.


Nichrome Wire Gaug

Wire Length, inches

Cold Resistance,

Applied a.c. Volta

Current, amperes
operating temper

Power Requirement


TABLE 3

HEATER POWER REQUIREMENTS

Combustion
Tube Heater

ge No. 22

s 132

ohms 3.0

Ige, volts 117


;at
nature)

s, watts


7.5
880


Capillary
Tube Heater

24

11

11.4

6


2

12


A schematic diagram of the regulated power supply

and bridge circuit is shown in Figure 18. The solid-state

power supply, using two transistors and a zener diode,

provided excellent regulation with minimum a.c. ripple.

Performance characteristics under extreme input voltage

variations are indicated in Figure 19.












q C /* -IN2069 SILICONE RECTIFIERS
24V 2.71 2N1372

---- -4.7K
IOO1"F 4.7K 1 3K
SIOK


2N1382 ---- IOK
-30K-10 TURN
-IK70 500n

70
__-- 10K 500K
12 V Z


-RECORDER
2K
THERMISTO




Fig. 18.-Schematic Diagram of Power Supply and Bridge Circuit








68










27- 26.3 Volts

25

23

4 21-

19-


8 17

o 1 14.8 Volts
15-

13


30 40 50 60 70 80 90 100 110 120 130
A. C. Input, Volts



Fig. 19.-Stability of Power Supply With Change in Input
Voltage








69

Wire-wound bridge resistors (1% precision, 1/4 watt)

and potentiometers were used to reduce noise. The power

supply and the bridge circuit were housed in separate

chassis to provide greater flexibility and to expedite

circuit changes during the experimental work. The bridge

current-voltage relationship is shown in Table 4.


TABLE 4

VARIATION OF BRIDGE CURRENT WITH VOLTAGE

Bridge Voltage, Bridge Current,
volts ma

14.8 7
16.0 8
18.0 9
20.0 10
22.0 11
24.0 12
26.3 13


All analytical results were obtained using a bridge voltage

of 14.8 volts. A considerably greater signal-to-noise ratio

was obtained at this voltage which more than compensated for

the slight gain in sensitivity at increased voltages.

Operational Aspects

Standards and sampling procedure

Benzoic acid (N.B.S. Standard Sample #39g) was chosen

as a calibrating standard for all reported carbon and hydro-

gen determinations. To allow easier and more rapid handling,










the benzoic acid was compressed by the method used for

preparation of KBr pellets. Small chips, exhibiting excel-

lent coherence, were weighed and were introduced directly

into the system. Certain types of samples, however,

required encapsulation. Among these were liquids, non-

crystalline solids, and hygroscopic, air-oxidizable, or

explosive compounds. In addition, a few samples, such as

benzophenone, which exhibited a tendency to burn pre-

maturely before contacting the copper oxide, or to adhere

to the walls of the upper portion of the combustion tube,

required special handling.

The capsules used during these studies weighed less

than 2 milligrams and were prepared by wrapping 6 mm squares

of aluminum foil around a 1.5 mm melting point tube. In the

case of liquids the sample was injected by means of a 1

microliter syringe (7001-N, The Hamilton Company, Inc.,

Whittier, California) into a small amount of combustion

catalyst which was pre-tared with the capsule. This served

to absorb the liquid and to minimize volatilization or

leakage during the weighing procedure. Highly volatile,

non-viscous samples could not be analyzed during the course

of this investigation, but it was felt that with the design

of a seamless capsule that these would present no difficulty

in future work. Powdered solids were handled by the method

illustrated in Figure 20. The capsule was formed in the







































IU
% n. a OM


Fig. 20.-Encapsulation of Powdered Samples
k-I











normal manner on a 3/4 inch length of capillary tube using

a pre-weighed square of aluminum. The other end of the tube

was inserted into the powder and an appropriate amount of

the sample was introduced into the capsule by means of a

metal plunger. In this manner the compound was exposed to

the air for only a few seconds before being permanently

sealed and no losses were incurred through spillage or

adherence to the outside walls.

The percentage composition of benzoic acid (68.85 %

C, 4.95 % H) approaches the mean composition expected for

most non-halogen-containing organic compounds. This factor,

as well as its availability in extremely pure form and its

lack of affinity for moisture, contributed to its selection

as a calibration standard. A weight of 0.65 milligrams of

benzoic acid was considered to be optimum because it allowed

good weighing precision and did not require a high sensi-

tivity setting. For maximum accuracy, the sample weight

was selected so that the resultant carbon dioxide signal

corresponded approximately to that of the standard without

an attenuation change. This was justified by the substantial-

ly lower percentage accuracy requirements for hydrogen.

Peak amplitudes were adjusted, in each case, to lie between

50 and 100 per cent of full scale for greater readant
accuracy.











Instrumental procedure

With oxygen flowing through the system, power was

switched on to the furnace, to the detector, to the capil-

lary heater, and to the recorder. After a few minutes the

oxygen flow rate was adjusted to 40.0 cc/min. by regulation

of the upstream pressure. Periodic checks on flow values

were made using the soap-film flow meter. The instrument

was allowed to equilibrate for at least two hours before

running a sample. It was considered advisable to turn off

only the furnace heater during breaks in extended periods

of operation.

With the input shorted attenuatorr switch at the

"S" position), the recorder was adjusted so that the pen

rested exactly on the zero position of the chart. Recorder

amplifier sensitivity was adjusted for maximum gain and

was then reduced to the first point where no oscillation

occurred at any position on the scale. With the attenuator

in the position of maximum sensitivity (X1), the 30,000

ohm helipot was turned to zero the recorder. A benzoic

acid sample weighing close to 0.65 milligrams was introduced

into the sample port and then into the system by clockwise

rotation of the valve. After the appearance of the carbon

dioxide peak, the bridge sensitivity was set by the following

procedure. The zero control was first adjusted to provide

a signal corresponding exactly to the initial peak height.











The 500,000 ohm potentiometer was then adjusted to shift

the deflection to 75 per cent of full scale. The silica

gel was cleared of water by lowering the capillary tube

heater, and the converter tube was filled with fresh

calcium hydride.

With the attenuator at the X1 position and the re-

corder at the zero position, a second sample of benzoic

acid weighing approximately 0.65 milligram was introduced

into the combustion tube. As soon as the carbon dioxide

peak appeared the heater was lowered to the silica gel

portion of the capillary tube and was allowed to rest on the

silicone rubber insert of the converter tube for 5 seconds.

The heater was returned to its original position, the

recorder polarity switch was set at the "H2" position, the

attenuator was adjusted to X4, and the recorder was re-

zeroed, if necessary. Immediately after the hydrogen peak

emerged, the polarity switch was returned to the "002"

position, the attenuator was reset to X1, and the system

was ready to accept a sample.

For those samples with known approximate compo-

sitions, quantities were selected which corresponded roughly

to carbon contents of 0.45 milligrams (the approximate

weight of carbon in 0.65 mg of benzoic acid). In the case

of a sample of totally unknown content, it was sometimes

advantageous to make a quick preliminary determination so










that a more appropriate sample weight and attenuation

range could be selected. Weights for both the benzoic acid

standards and the samples were recorded to the nearest 0.1

microgram.

The weighed sample was introduced into the system,

and the complete procedure used for standardization was

repeated. In most instances the correct attenuation value

for hydrogen could be readily estimated. However, in

those cases where the amplitude of the hydrogen signal did

not follow prediction, it was possible to increase or to

decrease the attenuation during the emergence of the peak,

providing the recorder had been previously zeroed at the

highest sensitivity position.

Because peak heights for carbon and hydrogen were
linear with concentration over the prescribed range, per-

centage compositions could be calculated using the

relationships

%C ( 2 P C stamp/ sam stamp.
XHOstatd.
(P.H. 2 stand./ stand. stan


H 2 ( samp/ t'samp. Attnsam
X saH.. WtHstand.
(P' 'H2 stand./ stand. Attn stand. stand


An engineering ruler was used to read peak heights to within

the nearest 0.005 inch. For samples with carbon content










below 55 per cent, greater analytical accuracy was obtained

by comparison to a standard also of low carbon content.

Sucrose (N.B.S. Standard Sample #17) was often used for

this purpose.

Nitrogen analyses were performed using the modified

system and sample introduction technique discussed on pages

50-52. Helium (Ohio Chemical and Surgical Equipment
Company, Madison, Wisconsin), which was found to be more

satisfactory as a carrier gas than carbon dioxide, was

employed for the major portion of the determinations, and

the capillary tube-conversion system of the carbon and

hydrogen apparatus was replaced with a magnesium perchlorate-

ascarite filter. Reagent grade ammonium oxalate (A-679,

Fisher Scientific Company) was used as a standard for the

analysis of most compounds with nitrogen content ranging

between 10 and 30 per cent. A convenient weight for cali-

bration purposes was 0.5 milligrams. At a helium flow rate

of 40.0 cc/min. this required an attenuation of X2 with no

change in the bridge sensitivity adjustment used for the

carbon and hydrogen analysis. It was again found advisable

to weigh the samples in quantities corresponding to the

standards in amount of nitrogen released. Under these

conditions the nitrogen composition was obtained using the

equation
P.H. / Wt
%N samp./ Wtsamp. %N
S "H. stand./ Wstand. stand.













SECTION IV


RESULTS AND DISCUSSION


Before any standardizations or analyses were con-

ducted, the system was thoroughly tested for leaks. This

was accomplished most conveniently by closing off the exit

of the sample side of the detector cell and noting the

corresponding flow decrease on the rotameter. If the flow

did not decrease to zero after several minutes, portions

of the system were isolated until the point of leakage was

determined. Precision and accuracy were found to be

extremely dependent upon the maintenance of a totally leak-

tight system and leakage could generally be recognized by

failure of a component peak to return to the base line.

Other malfunctions during an analysis were readily indicated

by an anomalous response pattern such as the occurrence of

double peaks, negative peaks, peak broadening, etc. It was

necessary, of course, to minimize the recorder deadband

before any determinations were made.

Typical readouts for carbon and hydrogen, and for

nitrogen are shown in Figure 21. Linearity over a normal

concentration range is indicated by the curves of Figure

22. A benzoic acid standard was used for the carbon and

77

























0
U U









o i 4 0:
SI 4I



















0.6013 mg Benzoic 0.4371 mg Ammonium
o 0














Acid Oxalate Monobydrate
Fi.601 m-Readouts for Carbon Dioxide, Hydrogen, and Nitrogen

Fig. 21.-Eeadouts for Carbon Dioxide, Hydrogen, and Nitrogen





































-N2, X2

-H2 X4

C02' X1


0 0.1 0.2 0.3 0.4 0.5 0.6

Weight of Standard, mg


Fig. 22.-Response
Nitrogen


Curves for Carbon Dioxide, Hydrogen, and


0.7 0.8











hydrogen values, and ammonium oxalate monohydrate was used

for the nitrogen values. Two attenuation settings were re-

quired to provide maximum readout accuracy for hydrogen

over the entire weight range. Prior to each of the

determinations the flow rate was measured and was set, if

necessary, to exactly 40.0 cc/min. The responses and the

response ratios did not necessarily remain constant during

extended periods of operation, but expressed variations

with general changes in conditions such as replacement of

combustion tube packing, partial deactivation of calcium

hydride by atmospheric moisture, etc. When the system had

been allowed to stabilize for a sufficient length of time

and when there were no observable fluctuations in operating

conditions, such as flow rate or line voltage, the required

frequency of standardization was reduced to a minimum. For

extremely critical analyses, however, it was found advisable

to run a standard immediately after the sample and to use a

fresh layer of calcium hydride for each set of determinations.

To test the precision of this procedure a series of six

samples of benzoic were analyzed, each by comparison with

a separate benzoic acid standard. These were purposely not

run in sequence in order to determine the effect of response

variations. The results are tabulated in Table 5. Similar

data were obtained for the nitrogen analysis of alanine

using ammonium oxalate monohydrate as a standard, and these

are presented in Table 6.






TABLE 5
PRECISION FOR A TYPICAL CARBON AND HYDROGEN DETERMINATION


DeBenzoin Benzoic Acid Benzoin Composition, Found
Determi-
nation Wt., P.H., inches Wt., P.H., inches Deviation Deviation
No. pg x 10 pg x 10 % 0 from % H from
002 H2 002 H2 mean, % mean, %


594.3
627.9
571.1
672.0
637.3
598.9


78.95 53.75*
83.05 54.10*
74.35 92.30
89.50 58.80*
83.85 55.55*
76.80 94.45


612.3
661.7
632.1
623.0
619.2
654.4


70.70 95.75
75.95 97.20
71.60 90.05
72.20 97.10
70.75 91.15
73.10 91.30
Mean


*
At an attenuation of X4. (All other hydrogen values
of X2.)

Theoretical Values,
Benzoic Acid
Benzoin
Mean Experimental Values, Benzoin
Average Deviation
Standard Deviation
% Standard Deviation


79.21
79.34
79.13
79.12
79.28
79.04
79719


+0.02
+0.15
-0.06
-0.07
+0.09
-0.15
Mean


5.72
5.81
5.62
5.56
5.86
5.60
5-.67


+0.03
+0.12
-0.07
-0.13
+0.17
-0.09


were obtained at an attenuation


68.85
79.23
79.19
0.09
0.11
0.14


4.95
5.70
5.69
0.10
0.12
2.1









PRECISION FOR A


TABLE 6

TYPICAL NITROGEN DETERMINATION


Determi- Ammonium oxalate
nation Alanine monohydrate Alanine, Deviation
No. Wt., P.H., t., P.., % N from
pg inches x 10 pg inches x 10 found mean, %


911.6 62.25
874.6 58.85
852.3 57.25
900.8 62.50
854.2 58.05
891.7 61.40


861.4
791.2
823.9
841.0
772.1
805.0


74.35
67.90
70.75
75.55
65.05
69.30


Theoretical Values
Ammonium Oxalate Monohydrate
Alanine
Mean Experimental Value, Alanine
Average Deviation
Standard Deviation
% Standard Deviation


15.59
15.45
15.96
15.64
15.90
15.76
Mean T.7


19.71
15.72
15.72
0.16
0.18
1.1


-0.13
-0.27
+0.24
-0.08
+0.18
+0.04










To determine the accuracy of the method, a range of

compounds exhibiting wide structural and compositional

variations were analyzed. The method of repeated standardi-

zations was again used for all reported determinations,

even in those cases where excellent response consistency

was obtained. Analytical reagents were selected for

analysis wherever possible. Those which were known to be

deliquescent were first thoroughly dried at a reduced

pressure. No combustion catalyst was used for carbon and

hydrogen analyses except in the case of liquids or organo-

metallics containing alkali or alkaline earth metals.

The accuracy for carbon and hydrogen is indicated

by the data presented in Table 7. A single determination

was conducted for each compound and, with the exception of

potassium acid phthalate, no preliminary values were dis-

carded. Of the 20 compounds reported, the average absolute

deviation from accepted percentages was + 0.15 per cent for

carbon and + 0.12 per cent for hydrogen. No deviations

exceeded + 0.46 per cent. The acceptable absolute deviation

from the theoretical percentages when using the traditional

gravimetric method is + 0.3 per cent (39).

Carbon analysis of phosphorus-containing compounds

by conventional techniques requires extremely strong

ignition to avoid retention of carbon dioxide by the oxides

of phosphorus. Even with the observance of this precaution,













Compound
Acetanilide


Benzo-
phenone



4-Bromo-
aceto-
phenone

o-Bromo-
benzoic
Acid

Caffeine


TABLE 7
ANALYTICAL RESULTS FOR CARBON AND HYDROGEN


% C piff.
Structure Origin Theor. Anal. %
CH CONHCHS Merck 71.09 70.89 -0.20
3 5 and Co.
(U.S.P.)

(CCH5)2C0 Fisher 85.69 85.75 +0.06
Scien-
tific
Co.,
No.B-270
BrC6H4COCH5 Eastman 48.28 48.24 -0.04
3 Organic
Chemicals
No.1886
BrC6H4COOH Matheson 41.82 41.59 -0.23
Coleman
and Bell
No.7383
CH3NCON(CH )COC:0N:CHNCH Eastman 49.48 49.36 -0.12
3It--I 3Organic
Chemicals
No.355


3.54



2.51



5.19


3.84



2.42



5.31


Diff.
%
-0.19


+0.06


+0.30



-0.09



+0.12


I


~


I


% H
Theor. Anal.
6.71 6.52


5.53 5.59


I


I







TABLE 7 Continued


% C Diff. % H Diff.
Compound Structure Origin Theor. Anal. % Theor. Anal. %
Diallyldi- (CH2:CHCH2)2PBr(C6H5)2 Synthe- 62.25 62.24 -0.01 5.80 5.78 -0.02
phenyl sized,
phosphonium Univ. of
Bromide Florida
p-Dichloro- C6H4C12 The Will 49.03 48.81 -0.22 2.74 2.86 +0.12
benzene Corp.
2,2"-Di- (C1C6H4)2C6H4 Synthe- 72.26 72.13 -0.13 4.04 3.96 -0.08
chloro-m- sized,
ter- Univ. of
phenyl*+ Maryland
9-Fluorenone C6H4CC 6H4 Eastman 86.65 86.69 +0.03 4.47 4.56 -0.09
Organic
Chemicals
No.3008
Hexamethyl- [(CH )2N)PPO Fisher 40.22 40.49 +0.27 10.12 10.27 +0.15
phos- Scien-
phoramide*+ tific Co.
No. H-
343
Naph- 1loH8 N.B.S. 93.71 93.49 -0.22 6.29 6.24 -0.05
thalene Standard
No. 38b







TABLE 7 Cont


Compound
Potassium
Acid
Phthalate*
Quinalizarin


Structure Origin
KOOCC6H4COOH N.B.S.
Standard
No. 84e

(HO)206H2(CO)206H2(OH)2 Fisher
Scien-
tific
Co. No.
Q-4


;inued


% C
Theor. Anal.
47.05 46.90


61.77 61.61


Diff.

-0.15


-0.16


% H
Theor. Anal.
2.47 2.40


2.96 5.13


Salicylic HOC6H4COOH Fisher 60.87 60.69 -0.18 4.38 4.25 -0.13
Acid Scien-
tific Co.
No. A-277

Sucrose C1222011 N.B.S. 42.11 42.14 +0.03 6.47 6.38 -0.09
Standard
No. 17

Sulfanilamide H2NC6H4SO2NH2 Nutri- 41.85 41.82 -0.03 4.68 4.60 -0.08
tional
Biochemi-
cals
Corp.


Diff.

-0.07


+0.17


I








TABLE 7 Continued


% C Diff. % H Diff.
Compound Structure Origin Theor. Anal. % heor. Anal. %

Tetra-n-butyl- [CH (CH2) 34NI Eastman 52.03 52.49 +0.46 9.76 9.50 -0.26
ammonium Organic
Iodide Chemicals
No. 4702

4,4,4-Tri- SCH:CHCH:CCOCH2COCF Eastman 43.25 43.49 +0.14 2.25 2.49 +0.14
fluoro-1- Organic
(2-thienyl)- Chemicals
1,3-butane- No. 7260
dione

Triphenyl- (CgH6) P Metal and 82.43 82.41 -0.02 5.76 5.83 +0.08
phosphine Thermite
Corp.

Vanillin CH O(OH)C6H3CHO Eastman 63.15 62.94 -0.21 5.30 5.46 +0.16
Organic
Chemicals
No. 273

Silver permanganate added

+Liquid sample










the carbon contents reported for phosphorus compounds by

commercial analytical laboratories have generally been 1-3

per cent lower than the theoretical values. Excellent

results, however, were obtained in this study for similar

or identical compounds with no additional catalytic or

temperature requirements. This was attributed to the high

combustion efficiency of the system and to its acceptance

of small samples.

Of all the compounds analyzed, only one gave con-

sistently low results. Potassium acid phthalate (47.05% C,

2.47% H) repeatedly showed a hydrogen content of 2.35 +

0.05 per cent and a carbon content of 44.0 + 0.1 per cent.
This corresponded very closely to the values calculated

for the reaction


2 + 15 02---K2C03 + 15 C02 + 5 H20 .
SCOOH

Results were not noticeably improved by the addition of

cobaltic oxide, but encapsulation of the sample with

approximately 5 milligrams of silver permanganate gave

values within 0.15 per cent of the theoretical percentages.

The silver permanganate prepared in this laboratory showed

no background response for carbon, hydrogen, or nitrogen

and was recommended as a general catalyst for all compounds

which were suspected to exhibit anomalous combustion










characteristics. Substances with high halogen or sulfur

content were analyzed with no difficulty even though the

silver gauze used initially was of relatively low surface

area. It was noted in earlier work that the chlorine re-

leased during an analysis of a chloro-compound partially

oxidized the silver bromide formed during previous analyses

of bromo-compounds, and expelled a small amount of bromine

into the silica gel. Although this condition showed no

adverse effect on the accuracy of the determinations, it

was corrected by the use of silver wool in place of the

silver gauze.

The results of a series of 10 nitrogen analyses are

presented in Table 8. The average deviation was within

0.22 per cent of the theoretical percentage and no

individual deviation exceeded 0.43 per cent. The allowable

error in the conventional micro-Dumas method for nitrogen

is + 0.2 per cent ( 0.3% for high percentages of nitrogen)

(35). A small amount of silver permanganate was added to
the sample for almost all reported determinations. In

addition to its catalytic effect, it served to reoxidize

the copper formed by the reduction of copper oxide during

the combustion process. At line voltage furnace operation,

some thermal decomposition of the copper oxide was evidenced

by excessive oxidation of the copper layer. It was, there-

fore, recommended that the furnace be operated at a somewhat

lower temperature (<10000C) during nitrogen analyses.







TABLE 8
ANALYTICAL RESULTS FOR NITROGEN


Compound
Acetanilide

Ammonium
Nitrate
Asparagin

p-Bromo-
aniline*


Structure

CH3CONHC6H5

NH4NO3

NH2COCH2CH(NH2)COOH*H20

BrC6H4NH2


Origin
Merck and Co.
(U.S.P.)
Fisher Scientific
Co., No. A-676
Fisher Scientific
Co., No. A-912
Eastman Organic
Chemicals
No. 473


% N
Theor. Anal.
10.36 10.51

35.00 35.28

18.66 18.49

8.14 8.30


Caffeine* CH NCON(CH3)COC:CN:CHNCH Eastman Organic 28.85 29.12 +0.27
31 3 N1 o. 355
Creatine NH20(:NH)N(CH3)CH20OOH Nutritional 32.05 31.69 -0.36
Biochemicals
Corp.
Cystine* [SCH2CH(NH2)COOH]2 Merck and Co. 11.66 11.50 -0.16
No. 72612


Diff., %
+0.15

+0.28

-0.17

+0.16


I


I


I


Z






TABLE 8 Continued


Compound

Hexamethyl-
phosphor-
amide*+
Quinaldic
Acid*

Sulfanilamide*


Structure

[(CH3)2N]3PO


O6H4N:CCOOHCH:CH

H2NC6HS402NH2


Origin

Fisher Scientific
Co., No. H-343

Eastman Organic
Chemicals
No. 2055
Nutritional
Biochemicals
Corp.


%N
Theor. Anal.

23.45 23.02


8.09 7.92


16.27 16.31


Silver permanganate added
+Liquid sample


Diff., %

-0.43


-0.17


+0.04


I











The initial results for nitrogen determinations

using a carbon dioxide carrier gas were satisfactory for

high nitrogen percentages but the signal-to-noise ratio

was too low to provide sufficiently accurate values for

compounds of low nitrogen content. All reported results

were, therefore, obtained using a helium carrier gas.

Experimental studies have indicated that the

accuracy of this method of analysis is comparable to that

of the conventional micro-Dumas technique for nitrogen,

and exceeds that of the conventional Pregl technique for

carbon and hydrogen. Its main advantages lie in its speed,

simplicity, reduction in operational requirements, and its

acceptance of small samples. With two functional systems

and a single recorder, a complete carbon, hydrogen, and

nitrogen analysis may be performed in less than ten minutes

even though two weighing are required. New commercially

available instruments which allow CHN determinations to be

performed on a single sample require considerably longer

analysis time and suffer from inaccuracies inherent in

combining the procedures.

It is anticipated that, with additional improvements,

the system described will require less frequent standardi-

zation and will offer even greater stability, precision and

accuracy.












SECTION V


EXTENSIONS TO ADDITIONAL ANALYSES


A simple modification of the apparatus originally

designed for the determination of carbon and hydrogen

provided a satisfactory analysis for nitrogen. Adaptation

of the general approach to the analysis of oxygen, sulfur,

and halogens was, therefore, investigated.

Oxygen

Recognition of the difficulties involved in the

direct analysis of oxygen in organic compounds has generally

led to the calculation of oxygen content by difference.

Good results, however, have been reported for certain

materials through the use of a Pregl method coupled with

a volumetric or gravimetric determination of the combustion

products. The analysis is based upon the ignition of the

compound in the presence of carbon at 11200C using an

oxygen-free carrier gas such as nitrogen or helium. At

this temperature the following pertinent reactions occur

Organic 0 + C O

CO2 + C -2 CO

H20 + C -CO + H2











The quantity of carbon monoxide produced is stoichiometrically

related to the amount of oxygen present in the sample. Oxi-

dation of carbon monoxide to carbon dioxide by passage

through a solution containing iodine pentoxide and by subse-

quent titration of the iodine formed allows a determination

of the total amount of carbon monoxide released during the

combustion process. An alternate method requires oxidation

of the carbon monoxide by passage over hot copper oxide,

and a gravimetric determination of the resultant carbon

dioxide collected in an ascarite absorption tube.

A combustion tube was packed with a 6 inch layer of

20 mesh spectrographically pure carbon chips followed by

a 1 inch layer of silver wool. The Nichrome heating coil

was compressed to provide a maximum temperature in the area

of the upper portion of the carbon layer. A second

combustion tube, packed with copper oxide, was positioned

below the first tube, and a magnesium perchlorate filter

was added to retain the water formed during the oxidation

of hydrogen. Helium, de-oxygenated by passage through a

copper-filled furnace tube at 4500 (S-36518, E. H. Sargent

and Co.), was used as a carrier gas. A method of sample

introduction identical to that used for the nitrogen analysis

was chosen to reduce atmospheric contamination to a minimum.

Erratic results were obtained for an initial series of

compounds, and it was found that non-oxygen-containing




Full Text
26
Silver Wool
Combustion Tube
Rubber Serum Cap
Moving Furnace
Capillary Tube
Silica Gel Packing
Silicone Rubber Sleeve
Calcium Hydride
Glass Wool Retainer
Pig. 4.-Details of tbe Retention and Conversion System


65
of hydrogen. A short length of rubber tubing joined the
column to the inlet of the reference detector, and the out
let was connected to a flow meter. The flow meter was of
the soap-film type and was constructed from a 5 cc Mohr
measuring pipet by cutting off the ends and attaching a
glass T fitted with a medicine dropper bulb. This allowed
more accurate and more rapid flow measurements than were
possible with conventional higher volume soap-film meters.
A stop watch with a 10 second sweep which could be read to
the nearest 0.02 second was used to check all flow rates.
An 11 foot length of #22 gauge nichrome wire was
used for the furnace winding. This provided 110 turns
about the 7 mm Vycor combustion tube and allowed 3 1/2
inch double-twisted leads at either end for reduction of
lead resistance. Small bolts were used to connect the
nichrome heater to the line cord. The coil was compressed
at the point corresponding to the surface of the copper
oxide and was expanded at either end. This provided
maximum heat at the sample contact point. Using an optical
pyrometer, the temperature of the copper oxide at line
voltage (~ 117 volts a.c.) operation was observed to lie
between 1000 and 1050C. For the capillary tube heater,
approximately 11 inches of #24 gauge nichrome wire were
wound 34 times about a 3/52 inch mandrel leaving 1 inch
double-twisted leads at either end. The spiral was encased


Fig. 18.-Schematic Diagram of Power Supply and Bridge Circuit
cr>
-o


75
that a more appropriate sample weight and attenuation
range could he selected. Weights for both the benzoic acid
standards and the samples were recorded to the nearest 0.1
microgram.
The weighed sample was introduced into the system,
and the complete procedure used for standardization was
repeated. In most instances the correct attenuation value
for hydrogen could be readily estimated. However, in
those cases where the amplitude of the hydrogen signal did
not follow prediction, it was possible to increase or to
decrease the attenuation during the emergence of the peak,
providing the recorder had been previously zeroed at the
highest sensitivity position.
Because peak heights for carbon and hydrogen were
linear with concentration over the prescribed range, per
centage compositions could be calculated using the
relationships
%C
Py/Wt*samp.
./ "samp,
d./ W15 stand. X
Wt,
'samp.
An engineering ruler was used to read peak heights to within
the nearest 0.005 inch. For samples with carbon content


17
Ther-
mistor
Circuit

Fig. 3.-Circuit for Conductance Measurement of Water and
Thermal Conductivity Measurement of Carbon Dioxide


5
for carbon and hydrogen which uses a quartz combustion tube
containing baffles and a small quantity of silver has been
investigated and recommended by Belcher, Ingram, Spooner,
Kainz, and Horwatitsch (3,4,17,22), Improved combustion
efficiency has been observed upon substitution of various
oxides for the conventional copper oxide packing, KBrbl
(25,26), for example, has obtained excellent results at
reduced operating temperatures with the decomposition
products of silver permanganate.
Very few methods are available which are capable
of circumventing combustion requirements for elemental
analysis although a neutron scattering technique has been
used by Finston and Yellin (11) to determine carbon-hydrogen
ratios. A number of efforts directed toward elimination of
absorption tube weighing rely upon conductometric, mano-
metric, or volumetric measurements. Malissa and Pell (28),
for example, have determined carbon, hydrogen, and sulfur
conductometrically while Frazer (13) has investigated a gas
volumetric method for the simultaneous determination of
carbon, hydrogen, and nitrogen. Hozumi and Kirsten (16)
have recently reported a similar approach to the combined
analysis based upon dry combustion of the sample in a sealed
tube and the subsequent weighing of the quantity of mercury
required to replace each of the combustion gases.


45
in detector response. The influence of block temperature,
a function of the time allowed for the detector to reach
thermal equilibrium, is shown in Figure 9. Because
conduction of heat away from the thermistors increases the
temperature of the cell block until an equilibrium
temperature is reached, a decrease in response to carbon
dioxide with time is anticipated due to the convergence of
the thermal conductivities of oxygen and carbon dioxide at
higher temperatures. Little or no response change is
expected or observed for hydrogen since the ratio of the
thermal conductivities for hydrogen and oxygen remain
essentially constant over a wide temperature range. It
is evident, however, that the system should be allowed to
"warm up" for at least one hour before an analysis is run,
and that frequent standardizations are required for the
first two and a half-hour period.
The necessity for a critical control of the oxygen
flow rate is indicated in Figure 10. The diminishing
response to carbon dioxide corresponds to that normally
encountered when a carrier gas flow rate is increased to a
point where the residence time of the component in the cell
has a significant effect on the resultant signal. It is
apparent, in this instance, that if measurements of carbon
dioxide were based on peak areas rather than peak heights,
any errors due to flow variations would be severely


52
corrective working curves, and excessive peak widths for
carbon dioxide resulted in undesirably long analysis times.
Of even greater consequence were indications of changes
in column characteristics during extended operation of the
system, resulting in variations in both retention volume
and response.
To test the capabilities of a more ideal system
requiring no chromatographic separation, the calcium
hydride was replaced by a desiccant, and a 1/8 inch x 8
foot length of empty plastic tubing was substituted for the
silica gel column. This allowed observation of performance
characteristics with regard to measurement of carbon
dioxide, initially circumventing any problems associated
with the determination of water. Polyethylene samples were
used for the test runs due to weighing convenience. Figure
6 indicates the improvement in sensitivity provided by this
system as compared to that using a silica gel column. Flow
rates were identical (40 cc/min.) and, in each instance,
the time required for the initial appearance of the carbon
dioxide peak was virtually the same. The seven-to-eight-
fold increase in sensitivity was accompanied by excellent
linearity throughout an even wider range of concentrations
than was normally suggested (further data are presented in
a later section), and stability, readout ease, and speed
of analysis were vastly improved. The accuracy afforded by


UNIVERSITY OF FLORIDA
3 1262 085lili^


SECTION I
INTEODUCTION
Elementad analysis may be defined as the determination
of the total percentage of each element present in a given
compound without regard to molecular arrangement. Analyses
of carbon, hydrogen, nitrogen, etc., find primary application
in the verification of empirical formulae of pure organic
compounds, or in the assay of mixtures whose constituents
exhibit substantial compositional differences. Information
essential to structural identification is provided by
inclusion of complimentary data from functional group
analysis.
The method for carbon and hydrogen first studied by
Pregl in 1910 may be regarded as the classical procedure of
microchemical analysis. The corresponding Liebig macro-
procedure now finds little or no application due to excessive
time requirements for complete combustion. Numerous modifi
cations of the original techniques have been proposed with
the object of simplifying the determination and minimizing
the inherent errors. In general, however, the basic
principles of Pregl's method have become well accepted, and
the method is more widely used than auay other.
1


76
below 35 per cent, greater analytical accuracy was obtained
by comparison to a standard also of low carbon content.
Sucrose (N.B.S. Standard Sample #17) was often used for
this purpose.
Nitrogen analyses were performed using the modified
system and sample introduction technique discussed on pages
50-52. Helium (Ohio Chemical and Surgical Equipment
Company, Madison, Wisconsin), which was found to be more
satisfactory as a carrier gas than carbon dioxide, was
employed for the major portion of the determinations, and
the capillary tube-conversion system of the carbon and
hydrogen apparatus was replaced with a magnesium perchlorate-
ascarite filter. Heagent grade ammonium oxalate (A-679,
Fisher Scientific Company) was used as a standard for the
analysis of most compounds with nitrogen content ranging
between 10 and 30 per cent. A convenient weight for cali
bration purposes was 0.5 milligrams. At a helium flow rate
of 40.0 cc/min. this required an attenuation of X2 with no
change in the bridge sensitivity adjustment used for the
carbon and hydrogen analysis. It was again found advisable
to weigh the samples in quantities corresponding to the
standards in amount of nitrogen released. Under these
conditions the nitrogen composition was obtained using the
equation


Figure Page
15. Details of Cell Construction 60
16. Tube Packings for Combustion, Absorption, and
Conversion 62
17. Details of Sample Introduction Port 63
18. Schematic Diagram of Power Supply and Bridge
Circuit > 67
19. Stability of Power Supply With Change in Input
Voltage 68
20. Encapsulation of Powdered Samples 71
21. Readouts for Carbon Dioxide, Hydrogen, and
Hitrogen 78
22. Response Curves for Carbon Dioxide, Hydrogen,
and nitrogen 79
23. Prototype Model of American Instrument Co.
Carbon and Hydrogen Analyzer 103
vi


89
characteristics Substances with high halogen or sulfur
content were analyzed with no difficulty even though the
silver gauze used initially was of relatively low surface
area. It was noted in earlier work that the chlorine re
leased during an analysis of a chloro-compound partially
oxidized the silver bromide formed during previous analyses
of bromo-compounds, and expelled a small amount of bromine
into the silica gel. Although this condition showed no
adverse effect on the accuracy of the determinations, it
was corrected by the use of silver wool in place of the
silver gauze.
The results of a series of 10 nitrogen analyses axe
presented in Table 8. The average deviation was within
+0.22 per cent of the theoretical percentage and no
individual deviation exceeded + 0.43 per cent. The allowable
error in the conventional micro-Dumas method for nitrogen
is + 0.2 per cent (+ 0.3% for high percentages of nitrogen)
(35). A small amount of silver permanganate was added to
the sample for almost all reported determinations. In
addition to its catalytic effect, it served to reoxidize
the copper formed by the reduction of copper oxide during
the combustion process. At line voltage furnace operation,
some thermal decomposition of the copper oxide was evidenced
by excessive oxidation of the copper layer. It was, there
fore, recommended that the furnace be operated at a somewhat
lower temperature (<1000C) during nitrogen analyses.


14
Sample
I
Fig. 2.-Preliminary Design of Sample Introduction System


SECTION III
CONSTRUCTIONAL AND OPERATIONAL DETAILS
Experimental System
The apparatus used for this study was assembled so
that a variety of modifications could be made simply and
rapidly. Major components were mounted on a network of
1/2 inch diameter rods which insured freedom from vibration.
Ambient temperature was not controlled, but remained
essentially constant at 25C. It was found necessary to
provide thermal shielding for only three components; the
pre-filter, the sample introduction port, and the detector.
The instrumental arrangement is shown in Figures 12
and 13. U.S.P. grade oxygen (Linde Company, New York,
N.Y.) was used throughout the investigation and all con
tacted surfaces were thoroughly cleared of oil, grease, or
other contaminants. A standard two stage oxygen pressure
regulator (Linde Company) equipped with a 1/d inch straight
thread hex-nipple was used for upstream pressure control.
A female 1/d inch two-piece swivel hose connector was
modified to house the throttling orifice as shown in Figure
Id. The inner section of the hose end portion was machined
55


44
Mn(OH)4 + U204Ma(N03)2 + 2 HgO ,
the amount of absorbed nitrogen dioxide being in proportion
to the surface. They observed excellent absorbent
properties for this gel and described a procedure for its
preparation. Experimentally, a short filter packed with a
microanalytical grade of manganese dioxide followed by a
layer of magnesium perchlorate was used. Its efficiency
was tested by injecting 5 cc vollimes of nitrogen dioxide
and sulfur dioxide immediately before and after the filter.
Complete retention was found even after numerous repeated
inj ections.
It was also necessary to provide a desiccant prior
to the combustion tube to remove traces of moisture present
in the oxygen carrier gas. This was particularly essential
because the silica gel capillary tube served to accumulate
water over extended periods of operation. Trusell and Diehl
(42) indicated the superior desiccant properties of anhydrous
magnesium perchlorate and, in this application, it was found
preferable to the 5A molecular sieve originally used. For
oxygen containing an appreciable amount of hydrogen or
hydrocarbons, initial purification using a pre-burner and
a carbon dioxide-water absorption system would be required.
Thermal equilibrium and flow rate
Two factors contributed significantly to variations


48
magnified since peak widths and peak heights decrease
simultaneously with increased flow. A similar decrease in
response to hydrogen was observed but only at much higher
flow rates. The response maximum for methane in oxygen
was found to be intermediate between that for carbon
dioxide and that for hydrogen.
Slight variations in flow impedance occur within the
system for numerous reasons. Changes in line voltage alter
the furnace temperature resulting in expansion or contraction
of the copper oxide. Increases in the amount and the degree
of compression of calcium hydroxide in the converter tube
as well as expansion due to formation of calcium hydroxide
produce additional flow restrictions. Eapid increase in
temperature of the silica gel in the capillary tube with
subsequent expansion introduces transient flow changes. In
order to minimize or to eliminate these effects a throttling
orifice was constructed and installed in the system. The
flow of gas through an orifice depends only to a certain
extent on the pressure differential across the orifice.
Where the upstream pressure is held constant and the pressure
at the discharge end is decreased, the exit gas flow will
increase to a critical value. Further decreases in the
downstream pressure will cause no increase in flow until the
critical flow condition has been exceeded. The critical
condition is obtained for an oxygen flow through an orifice


40
TABLE 2
PEAK HEIGHT TO HALF-PEAK-WIDTH RATIOS
Sample
Wt,Mg
Peak
Height
mm
Half-Peak
Width
mm
PH/HPW
Benzoic acid
661
196
4.2
46.7
Biphenyl
497
201
4.3
46.8
Potassium acid
phthalate
906
184
3.9
46.2
Caffeine
902
193
4.2
45.9
Teflon
1730
176
3.9
45.1
Activated carbon
(~ 60 per cent C)
722
187
4.0
46.8


83
To determine the accuracy of the method, a range of
compounds exhibiting wide structural and compositional
variations were analyzed. The method of repeated standardi
zations was again used for all reported determinations,
even in those cases where excellent response consistency
was obtained. Analytical reagents were selected for
analysis wherever possible. Those which were known to be
deliquescent were first thoroughly dried at a reduced
pressure. No combustion catalyst was used for carbon and
hydrogen analyses except in the case of liquids or organo-
metallics containing alkali or alkaline earth metals.
The accuracy for carbon and hydrogen is indicated
by the data presented in Table 7. A single determination
was conducted for each compound and, with the exception of
potassium acid phthalate, no preliminary values were dis
carded. Of the 20 compounds reported, the average absolute
deviation from accepted percentages was + 0.15 per cent for
carbon and + 0.12 per cent for hydrogen. No deviations
exceeded + 0.46 per cent. The acceptable absolute deviation
from the theoretical percentages when using the traditional
gravimetric method is + 0.3 per cent (39).
Carbon analysis of phosphorus-containing compounds
by conventional techniques requires extremely strong
ignition to avoid retention of carbon dioxide by the oxides
of phosphorus. Even with the observance of this precaution,


LIST OF TABLES
Table Page
1. Test Hesults for Accuracy of Carbon Analysis 34
2. Peak Height to Half-Peak-Uidtb Ratios 40
3 Heater Power Requirements 66
4. Variation of Bridge Current With Voltage ... 69
5. Precision for a Typical Carbon and Hydrogen
Determination 81
6. Precision for a Typical Nitrogen Determination 82
7. Analytical Results for Carbon and Hydrogen . 84
8. Analytical Results for Nitrogen 90
9. Results for Oxygen Analyses 95
iv


35
calibration was found, to remain extremely constant, but
was checked, nevertheless, after every five or six
determinations. For ultimate precision, the 0-1 milligram
range of the balance was used for all weighings, even
though, in some instances, the total weight exceeded 1
milligram. When it became necessary to encapsulate a
sample, a balancing capsule weighing slightly less than
the sample capsule was placed on the righthand pan, and
the net tare weight was obtained through electrical sub
traction by the zero circuit. If, in this procedure, the
total weights of the net tare and the sample were in excess
of 1 milligram, the 1 milligram calibrating weight was
added to the righthand pan and the additional weight was
obtained in the usual manner.
For non-crystalline or viscous liquid samples,
containers were fabricated from 6 mm squares of aluminum
foil formed about a 1.5 nim rod and crimped on one end.
After sample introduction the opposite end was crimped and
the weighing was completed. A preferred method of sample
enclosure required the use of seamless copper capsules of
equivalent dimensions. These facilitated the weighing of
volatile liquid samples and continuously provided a fresh
layer of copper oxide rather than aluminum oxide at the
combustion site. They were not used during the course of
the investigation due to difficulties in fabrication.


80
hydrogen values, and ammonium oxalate monohydrate was used
for the nitrogen values. Two attenuation settings were re
quired to provide maximum readout accuracy for hydrogen
over the entire weight range. Prior to each of the
determinations the flow rate was measured and was set, if
necessary, to exactly 40.0 cc/min. The responses and the
response ratios did not necessarily remain constant during
extended periods of operation, hut expressed variations
with general changes in conditions such as replacement of
combustion tube packing, partial deactivation of calcium
hydride by atmospheric moisture, etc. When the system had
been allowed to stabilize for a sufficient length of time
and when there were no observable fluctuations in operating
conditions, such as flow rate or line voltage, the required
frequency of standardization was reduced to a minimum. Por
extremely critical analyses, however, it was found advisable
to run a standard immediately after the sample and to use a
fresh layer of calcium hydride for each set of determinations.
To test the precision of this procedure a series of six
samples of benzoic were analyzed, each by comparison with
a separate benzoic acid standard. These were purposely not
run in sequence in order to determine the effect of response
variations. The results are tabulated in Table 5. Similar
data were obtained for the nitrogen analysis of alanine
using ammonium oxalate monohydrate as a standard, and these
are presented in Table 6.


69
Wire-wound, bridge resistors (1% precision, 1/4 watt)
and potentiometers were used to reduce noise. The power
supply and the bridge circuit were housed in separate
chassis to provide greater flexibility and to expedite
circuit changes during the experimental work. The bridge
current-voltage relationship is shown in Table 4.
TABLE 4
VARIATION OF BRIDGE CURRENT WITH VOLTAGE
Bridge Voltage,
volts
14.8
16.0
18.0
20.0
22.0
24.0
26.3
Bridge Current,
ma
7
8
9
10
11
12
13
All analytical results were obtained using a bridge voltage
of 14.8 volts. A considerably greater signal-to-noise ratio
was obtained at this voltage which more than compensated for
the slight gain in sensitivity at increased voltages.
Operational Aspects
Standards and sampling procedure
Benzoic acid (N.B.S. Standard Sample #39g) was chosen
as a calibrating standard for all reported carbon and hydro
gen determinations. To allow easier and more rapid handling,


Magnesium Perchlorate Ascarite Magnesium Perchlorate
62
|
I
I
Oxygen
Filter
*0
d
Pi
Pi
o
o
d
0
> rH
rH O
H O
CO >
'd
0
N CO
H O
d p
H CQ
P cd 3
H 03
Pi <
Combustin
Tube for
Carbon and
Hydrogen
Determi
nation
d
H
&
u
d>
Pi
Pi
o
o
d
> rH
H O
H O
CO >
*3
a>
03
H O
d p
H CO
P 0)
cd ,a
rH 03
Pi<
Combustion
Tube for
Nitrogen
Determi
nation
H &§
OSH
1
'd
o d
rH -d
H CO C5
CQ
CO rH
cd O
rH O
O >
cd
o w
Converter Tube
nn
0
p
a cd
3 d
H o
03 rH
03 .d
d o
bO d
cd o)
SPh
0
CO
0 0
d *d
cd *H
bOX
S3
so
Nitrogen Di
oxide Filter
0)
Carbon Dioxide-
Water Filter
for Nitrogen
Determination
Fig. 16.-Tube Packings for Combustion, Absorption, and
Conversion


TABLE 6
PRECISION FOR A TYPICAL NITROGEN DETERMINATION
Determi
nation
Alanine
Ammonium oxalate
monohydrate
Alanine,
Deviation
No.
ME
P.H. ,
inches x 10
vt.,
MS
P.H.,
inches x
10
% N
found
from
mean, %
1
911.6
62.25
861.4
74.35
15.59
-0.13
2
874.6
58.85
791.2
67.90
15.45
-0.27
3
832.3
57-25
823.9
70.75
15.96
+0.24
4
900.8
62.50
841.0
73-55
15.64
-0.08
5
854.2
58.05
772.1
65.05
15.90
+0.18
6
891.7
61.40
805.0
69.30
Mean
15.76
15775
+0.04
% N
Theoretical Values
Ammonium Oxalate Monohydrate
19.71
Alanine
15.72
Mean Experimental Value, Alanine
15.72
Average Deviation
0.16
Standard Deviation
0.18
% Standard Deviation
1.1


19
Thus, within a small temperature interval, thermistor
resistance is a linear function of temperature. The
response, cLE/dy^ to a particular component has been
shown by Smith and Bowden (34) to correspond to
dE dE dT dh
3y 3f 3E dyg *
where E is the out-of-balance bridge voltage, y2 is the
mole fraction of the non-carrier component in the detector,
T is the sensing thermistor temperature, and h is the
thermistor surface heat transfer coefficient. The dE/dT
and dT/dh derivatives are calculated from instrument
characteristics while dh/dyg depends essentially on the
thermal conductivity of the binary gas mixture in the de
tector. The possible effect of non-ideality of the binary
mixtures, oxygen-carbon dioxide and oxygen-hydrogen with
respect to thermal conductivity was considered. The
analogous nitrogen-carbon dioxide mixture, for example,
expresses a negative deviation from ideality as illustrated
by Keulemans (24). Departures from linearity, however, due
to these and other factors were either compensatory or
negligible in the experimental studies. This was evident
from the excellent linearity over a concentration range at
least as wide as that governed by combustion restrictions.
' Results are presented in a later section.


TABLE 03? CONTENTS
Page
ACKNOWLEDGMENTS ii
LIST OF TABLES iv
LIST OF FIGURES v
Section
I. INTRODUCTION 1
II. EXPERIMENTAL CONSIDERATIONS 8
Initial Functional Objectives and
Preliminary Concepts 8
Selection of Components and Operational
Procedures 11
Critical Parameters in tbe Finalized
System and Extension to Nitrogen
Determination 36
III. CONSTRUCTIONAL AND OPERATIONAL DETAILS. ... 55
Experimental System 55
Operational Aspects 69
IV. RESULTS AND DISCUSSION 77
V. EXTENSIONS TO ADDITIONAL ANALYSES 93
VI. SUMMARY 98
APPENDIX. 101
LITERATURE CITED 104
BIOGRAPHICAL SKETCH 107
iii


9
the system illustrated in Figure 1. A pipe T coupled to a
short tube filled with calcium carbide (30 mesh) was
inserted between the injection port and a 1 foot silica gel
(60-80 mesh) column of a commercial chromatograph. A
pyrolyzer probe consisting of a platinum filament connected
to a high current transformer was coated with a small
amount of benzoic acid and was fitted into the T-connector
which, in turn, was wrapped with heating tape to prevent
condensation of water. When equilibrium had been estab
lished, a current of 55 amperes was passed through the
platinum loop for a period of 1 second, affording a measured
ignition temperature of 1000C. Using pure oxygen as the
carrier gas, combustion and conversion products were
separated and were swept through a 100,000 ohm thermistor
detector. Signals corresponding to the resultant carbon
dioxide and acetylene were similar to those of the illus
trative chromatogram in Figure 1. Although the thermal
conductivities of carbon dioxide and acetylene are both
lower than that of oxygen, a negative peak was observed for
low concentrations of acetylene. With increasing acetylene
concentrations, W-shaped peaks occurred, exhibiting
alternate negative and positive deflections. These were
due to the maximum occurring in the thermal conductivity-
composition curve of the binary mixture at low
acetylene concentrations.


42
section of platinized asbestos to tbe combustion tube in a
similar technique for nitrogen determinations. This
functioned to catalyze the combustion of any unburned
pyrolysis products. Although the use of an oxygen atmos
phere substantially reduced this possibility, it was
considered advantageous to include a layer of platinized
asbestos in the carbon and hydrogen combustion tube.
Chlorine, bromine, iodine, and oxides of sulfur were
initially retained on a lH inch section of tightly rolled
silver gauze positioned in the lower portion of the
combustion tube. This, however, provided a relatively small
surface area for reduction and required indentations in the
walls of the tube to prevent shifting. On the other hand,
silver wool was found to retain its position, to provide a
much larger surface area, and to be much easier to clean.
It functioned also, as an excellent heat transfer medium,
assisting in the prevention of condensation of water in the
lower portion of the tube. Compounds with a high fluorine
content were analyzed successfully with no special pre
cautions. However, for the continual analysis of fluoro-
compounds, a layer of magnesium oxide or aluminum oxide
positioned between the platinized asbestos and the copper
oxide is advocated.
Combustion of nitrogen-containing compounds in an
oxygen atmosphere would be expected to produce, essentially,


if the upstream pressure is approximately two or more times
that of the downstream pressure (2). The equation for
critical flow is
q CApKVMT ,
where q is the mass rate of flow of the gas¡ C is the dis
charge coefficient; A is the cross-sectional area of the
orifice; p is the upstream static pressure of the gas; K
is a dimensional coefficient involving the gas constant;
M is the molecular weight of the gas; T is the absolute
temperature of the gas upstream; and U is a constant that
is characteristic of the gas and is given by
tr k[2/(k + l)]s ,
where k is the ratio of specific heats (Cp/Cv) and
s > (k+1) / (k-1) .
An orifice of approximately 0.0025 inch diameter was
developed in a thin steel disk using the method suggested
by Marks (30). The ratio of the absolute upstream pressure
to the downstream pressure was measured and was found to be
roughly equal to 2 for this orifice. The equation for
critical flow indicates that a + 5 change in temperature
of the gas upstream would produce a corresponding change
in flow of + 0.8 per cent. Therefore, in the final proto
type model, the orifice and the secondary pressure regulator
were positioned in the constant temperature section of the
instrument


41
resultant endotherm occurring initially at 925C suggested
a partial degradation of the copper oxide at this temperature.
This was substantiated in later work by evidences of oxygen
release under similar conditions. A second differential
thermal analysis of copper oxide, substituting oxygen for
helium, showed no thermal band below a temperature of
1055C. In excess of this temperature an endotherm corres
ponding to a slow fusion process was noted. Another
important advantage, therefore, was realized in the use of
oxygen rather than an inert carrier gas because its
repression of copper oxide decomposition allowed the use
of a higher furnace temperature with a resultant increase
in combustion efficiency.
Combustion tube packing and retention systems
The oxidizing portion of the combustion tube for a
conventional system for microanalysis of carbon and hydrogen
is packed with a mixture of 2 parts of copper oxide and 1
part of lead chromate, according to the procedure suggested
by Steyermark (38). The furnace temperature for this
system, however, is relatively low (680C) as opposed to
that of the method under investigation (> 1000C). Experi
mental results with lead chromate-copper oxide mixtures
indicated that it was unnecessary, in the present system,
to add an additional oxidant to the copper oxide. Later
work, however, proved the necessity for addition of a


70
the benzoic acid, was compressed by tbe method used for
preparation of KBr pellets. Small chips, exhibiting excel
lent coherence, were weighed and were introduced directly
into the system. Certain types of samples, however,
required encapsulation. Among these were liquids, non-
crystalline solids, and hygroscopic, air-oxidizable, or
explosive compounds. In addition, a few samples, such as
benzophenone, which exhibited a tendency to burn pre
maturely before contacting the copper oxide, or to adhere
to the walls of the upper portion of the combustion tube,
required special handling.
The capsules used during these studies weighed less
than 2 milligrams and were prepared by wrapping 6 mm squares
of aluminum foil around a 1.5 mm melting point tube. In the
case of liquids the sample was injected by means of a 1
microliter syringe (7001-N, The Hamilton Company, Inc.,
Whittier, California) into a small amount of combustion
catalyst which was pre-tared with the capsule. This served
to absorb the liquid and to minimize volatilization or
leakage during the weighing procedure. Highly volatile,
non-viscous samples could not be analyzed during the course
of this investigation, but it was felt that with the design
of a seamless capsule that these would present no difficulty
in future work. Powdered solids were handled by the method
illustrated in Figure 20. The capsule was formed in the


39
near identity of its thermal conductivity with that of
oxygen. The carhon dioxide response factors for feenzoic
acid were similar; Rn = 1.13 mv/mg and R = 1.10 mv/mg.
u2
However, upon introduction of a sample of activated carhon
(approximately 60% C), R,-, = 0.98 mv/mg whereas virtually
u2
no response was observed for an even larger sample of
carbon using a nitrogen carrier gas. A second carbon
sample was encapsulated and was surrounded with approxi
mately 5 milligrams of silver permanganate. The carbon
dioxide peak height response in the nitrogen carrier gas
was, in this case, 0.28 mv/mg with some evidence of peak
spreading.
Further information regarding combustion spontaneity
in oxygen was derived from the following data. Peak height
to half-peak-width ratios were calculated for a series of
samples ranging in ease of combustion. Sample weights were
chosen so that peak height magnitudes remained essentially
the same. The maximum chart speed was used to allow more
accurate half-peak-width measurements. The results, as
shown in Table 2, indicated no inconsistencies or peak
spreading. Oxygen provided the additional advantage of
rapid regeneration of copper oxide which may have been
reduced during the combustion process.
A differential thermal analysis of copper oxide
versus carborundum was conducted in a helium flow. The


99
under low-load conditions but are as yet not capable of
determining small weight differences in heavy absorption
tubes to sin acceptable degree of accuracy.
The detection method which has been described
complements the new weighing innovations and eliminates
the critical volumetric measurements and programmed weighing
which are considered to represent the major source of error
in traditional microanalyses. The system circumvents the
meticulous and exacting techniques previously required of
a skilled operator and reduces the analysis time to less
than ninety seconds after introduction of the sample. Care
ful selection of design and operational parameters allows
compositional data to be obtained for an extremely wide
range of organic compounds with accuracy which exceeds
that of conventional methods.
In principle, a sample (solid, liquid, or gas) is
introduced into a tube packed with copper oxide at a fixed
temperature of approximately 1000C through which pure
oxygen flows at a constant rate. Combustion occurs rapidly
and uniformly under the conditions of the analysis. All
combustion products are retained with the exception of
carbon dioxide, the concentration of which is accurately
measured as it passes through a thermistor detector. Water,
initially absorbed by a desiccant, is expelled by a movahle
heater into a tube containing a layer of calcium hydride.


TABLE 7 Continued
% C
Diff.
% H
Diff.
Compound
Structure
Origin
Theor.
Anal.
%
Theor.
Anal.
%
Potassium
Acid
Phthalate*
koocc6h4cooh
N.B.S.
Standard
No. 84e
47.05
46.90
-0.15
2.47
2.40
-0.07
Quinalizarin
(ho)2c6h2(co)2c6h2(oh)
2 Fisher
Scien
title
Co. No.
Q-4
61.77
61.61
-0.16
2.96
5.13
+0.17
Salicylic
Acid
HOCgH^COOH
Fisher
Scien
tific Co.
No. A-277
60.87
60.69
-0.18
4.38
4.25
-O.I3
Sucrose
C12H2211
N.B.S.
Standard
No. 17
42.11
42.14
+0.03
6.47
6.38
-0.09
Sulfanilamide
H2NC6H4S02NH2
Nutri
tional
Biochemi
cals
Corp.
41.85
41.82
-0.03
4.68
4.60
-0.08


DESIGN AND PERFORMANCE CRITERIA IN
THE DEVELOPMENT OF A
NEW INSTRUMENTAL METHOD
FOR ULTRAMICRO ELEMENTAL ANALYSIS
By
C. DAVID MILLER
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
April, 1964

ACKNOWLEDGMENTS
The author extends his sincere appreciation to his
supervisory committee, to the faculty of the Chemistry
Department of the University of Florida, and to his
colleagues for their interest and their friendly cooperation
during the course of this investigation. He takes
particular pleasure in expressing gratitude to his research
director, Dr. J. D. Winefordner, for his constant sincerity
and willingness in providing the counsel, guidance, and
encouragement necessary for the realization of this project.
The helpful suggestions and the cooperation of the
research and development group of the American Instrument
Company are deeply appreciated. Mr. William C. Hampton is
particularly deserving of special thanks in this regard.
Additional recognition is due the American Instrument
Company for the financial support of this research, and the
assistance is highly valued.
Finally, the author gratefully acknowledges his
indebtedness to his wife, Bita Sue, for her continued
inspiration and understanding.
ii

TABLE 03? CONTENTS
Page
ACKNOWLEDGMENTS ii
LIST OF TABLES iv
LIST OF FIGURES v
Section
I. INTRODUCTION 1
II. EXPERIMENTAL CONSIDERATIONS 8
Initial Functional Objectives and
Preliminary Concepts 8
Selection of Components and Operational
Procedures II
Critical Parameters in tbe Finalized
System and Extension to Nitrogen
Determination 36
III. CONSTRUCTIONAL AND OPERATIONAL DETAILS. ... 55
Experimental System 55
Operational Aspects 69
IV. RESULTS AND DISCUSSION 77
V. EXTENSIONS TO ADDITIONAL ANALYSES 93
VI. SUMMARY 98
APPENDIX. . 101
LITERATURE CITED 104
BIOGRAPHICAL SKETCH 107
iii

LIST OF TABLES
Table Page
1. Test Hesults for Accuracy of Carbon Analysis . 34
2. Peak Height to Half-Peak-Uidtb Ratios 40
3« Heater Power Requirements 66
4. Variation of Bridge Current With Voltage ... 69
5. Precision for a Typical Carbon and Hydrogen
Determination 81
6. Precision for a Typical Nitrogen Determination 82
7. Analytical Results for Carbon and Hydrogen . . 84
8. Analytical Results for Nitrogen 90
9. Results for Oxygen Analyses 95
iv

LIST OP FIGURES
Figure Page
1. Preliminary System and Readout for Carbon
and Hydrogen Analysis 10
2. Preliminary Design of Sample Introduction -
System 14
3° Circuit for Conductance Measurement of Water
and Thermal Conductivity Measurement of
Carbon Dioxide 17
4. Details of the Retention and Conversion
System 26
5. Variations in Component Arrangements and
Readouts for Carbon and Hydrogen Analysis . . 50
6. Comparison of Sensitivity to Carbon Dioxide
Using Packed and Unpacked Columns Under
Identical Flow Conditions 53
7. Finalized Flow System and Readout for Carbon
and Hydrogen Analysis 57
8. Comparison of Combustion Efficiencies Using
Oxygen and Nitrogen Carrier Gases 58
9. Response to Carbon Dioxide Versus Equili¬
bration Time 46
10. Dependence of Response on Flow Rate 47
11. Reduction of Atmospheric Contamination During
Nitrogen Analyses 53
12. Instrumental Arrangement for Carbon and
Hydrogen Analysis 56
15. Details of Conversion and Detection Systems . 57
14. Details of Orifice Mounting 58
v

Figure Page
15. Details of Cell Construction 60
16. Tube Packings for Combustion, Absorption, and
Conversion 62
17. Details of Sample Introduction Port 63
18. Schematic Diagram of Power Supply and Bridge
Circuit > 67
19. Stability of Power Supply With Change in Input
Voltage 68
20. Encapsulation of Powdered Samples 71
21. Readouts for Carbon Dioxide, Hydrogen, and
Nitrogen 78
22. Response Curves for Carbon Dioxide, Hydrogen,
and Nitrogen 79
23. Prototype Model of American Instrument Co.
Carbon and Hydrogen Analyzer 103
vi

SECTION I
INTBODUCTION
Elementad analysis may be defined as the determination
of the total percentage of each element present in a given
compound without regard to molecular arrangement. Analyses
of carbon, hydrogen, nitrogen, etc., find primary application
in the verification of empirical formulae of pure organic
compounds, or in the assay of mixtures whose constituents
exhibit substantial compositional differences. Information
essential to structural identification is provided by
inclusion of complimentary data from functional group
analysis.
The method for carbon and hydrogen first studied by
Pregl in 1910 may be regarded as the classical procedure of
microchemical analysis. The corresponding Liebig macro-
procedure now finds little or no application due to excessive
time requirements for complete combustion. Numerous modifi¬
cations of the original techniques have been proposed with
the object of simplifying the determination and minimizing
the inherent errors. In general, however, the basic
principles of Pregl's method have become well accepted, and
the method is more widely used than any other.
1

2
The current determination is based upon complete
conversion in a stream of oxygen of the carbon and hydrogen
in the sample to carbon dioxide and water. This is
effected by passage of the pyrolysis products through a
combustion tube packed with copper oxide and lead chromate
maintained at a temperature of 680°C. Carbon dioxide is
absorbed by sodium hydroxide on asbestos, and water is
retained by magnesium perchlorate. The corresponding
weight increases in the absorption tubes are related to the
percentage composition of the sample.
Pregl has also been credited with adaptation of the
classical macro-Dumas nitrogen determination to the micro¬
technique which has been accepted as reliable for most
forms of nitrogen. The procedure involves combustion of
the sample in an atmosphere of carbon dioxide, in which
carbon and hydrogen are converted to carbon dioxide and
water over a copper oxide catalyst, and oxides of nitrogen
are reduced to elemental nitrogen over copper. The nitrogen
is collected and measured in a nitrometer containing a
confining solution of 50 per cent aqueous potassium
hydroxide which absorbs the carbon dioxide.
Instrumentation automating the basic techniques for
carbon, hydrogen, and nitrogen, as well as the direct
determination of oxygen has recently been made available by
Coleman Instruments, Inc. (5,6,7). The "empty-tube" method

3
for carbon and hydrogen which uses a quartz combustion tube
containing baffles and a small quantity of silver has been
investigated and recommended by Belcher, Ingram, Spooner,
Kainz, and Horwatitsch (3,4,17,22), Improved combustion
efficiency has been observed upon substitution of various
oxides for the conventional copper oxide packing. KBrbl
(25,26), for example, has obtained excellent results at
reduced operating temperatures with the decomposition
products of silver permanganate.
Very few methods are available which are capable
of circumventing combustion requirements for elemental
analysis although a neutron scattering technique has been
used by Finston and Yellin (11) to determine carbon-hydrogen
ratios. A number of efforts directed toward elimination of
absorption tube weighing rely upon conductometric, mano-
metric, or volumetric measurements. Malissa and Pell (28),
for example, have determined carbon, hydrogen, and sulfur
conductometrically while Frazer (13) has investigated a gas
volumetric method for the simultaneous determination of
carbon, hydrogen, and nitrogen. Hozumi and Kirsten (16)
have recently reported a similar approach to the combined
analysis based upon dry combustion of the sample in a sealed
tube and the subsequent weighing of the quantity of mercury
required to replace each of the combustion gases.

A noii-dispersive infrared gas analyzer lias been used
by Kuck and co-workers (27) to determine the carbon dioxide
released during a conventional combustion process after
adjusting the gas mixtures to a constant volume and a
standard reference pressure. Haber and Gardiner (14)
employed a commercial version of the electrolytic moisture
cell of Keidel (23) for the analysis of carbon and hydrogen.
The method involved quantitative conversion of carbon
dioxide to water by passing it over hot lithium hydroxide
deposited on a carborundum substrate. The subsequent
coulometric determinations of water provided absolute
measurements for both elements.
Procedures based upon measurement of the differences
in thermal conductivity between the resident gas and the
combustion products offer, in general, greater speed and
less experimental manipulation. This type of analysis was
first applied to carbon and hydrogen by Duswalt and Brandt
(9) and Sundberg and Maresh (40) who used conventional
combustion techniques coupled with a gas chromatographic
determination of carbon dioxide and acetylene, an appropri¬
ate conversion product of water. The gases were trapped
in a U-tube immersed in liquid nitrogen and, when the
combustion had been completed, were introduced into a gas
chromatograph where they were separated and detected.
Oxygen and helium carrier gases were used respectively.

5
Vogel and Quattrone (45) devised a somewhat more complicated
procedure involving a bomh combustion of the sample in an
oxygen atmosphere and a direct chromatographic analysis of
carbon dioxide and water. A system of even greater
complexity was employed by Nightingale and Walker (51) for
the simultaneous determination of carbon, hydrogen, and
nitrogen. An induction furnace was used to burn the sample
in a helium atmosphere and the resultant nitrogen, carbon
dioxide, and acetylene (the conversion product of water)
were separated by temperature-programmed gas chromatography.
In addition to the large sample requirements the methods
described were inaccurate and time-consuming, and were
limited by basic disadvantages which will be discussed in
a later section.
In an improvement of his original work, Maresh (29)
incorporated a simultaneous nitrogen determination, and the
method formed the basis of a commercial analyzer produced
by the Fisher Scientific Company (12). The instrument
again relied upon a separation of nitrogen, carbon dioxide,
and acetylene which were initially collected at liquid
nitrogen temperatures. Inherent problems included high
operational costs and inefficiencies associated with the
inert combustion atmosphere.
Simon and co-workers (8,35) determined carbon,
hydrogen, and nitrogen by expanding the mixture of combustion

6
products into a series of filters which selectively absorbed
water and carbon dioxide* Static thermal conductivity
measurements between various portions of the system pro¬
vided, under equilibrium conditions, data relating
concentration differences between composites and individual
components* The procedure of Valisch (46) depended upon
sample combustion over copper oxide in a helium carrier
gas containing 3 per cent oxygen, followed by the reduction
of oxygen and the oxides of nitrogen over copper. The
signal corresponding to nitrogen was subtracted from that
of the carbon dioxide-nitrogen composite, and water was
determined independently. The principle of analysis was
adopted by the Technicon Instruments Corporation in the
development of a commercial carbon, hydrogen, and nitrogen
analyzer (41)* An instrument designed and manufactured
by the P and M Scientific Corporation (10) used a series
arrangement of two katharometers to measure the water,
nitrogen, and carbon dioxide produced by burning the sample
in a helium atmosphere.
Of the methods described, those based on a gas
chromatographic detection of combustion products provide
greater operational simplicity, are less susceptible to
human error, and are capable of substantially reducing
analysis time although the quoted times range between 10
and 105 minutes per sample. Deviations, however, from

7
traditional techniques, generally resulting from the use of
inert carrier gases to increase sensitivity, impose limi¬
tations on combustion efficiency and restrict the range of
compounds capable of analysis. The singular and distinct
advantage of these systems arising from the reduction in
required methodological skills is often accompanied by
prohibitively high cost of equipment. The desirability of
a system combining maximum operational efficiency and
economy with increased speed and ease of analysis was
recognized, and it was toward this objective that these
research efforts were directed.

SECTION II
EXPERIMENTAL CONSIDERATIONS
Initial Functional Objectives and Preliminary Concepts
The procedures reported by Duswalt and Brandt (9)
and Sundberg and Maresh (29,40) first indicated the
feasibility of replacement of conventional gravimetric or
volumetric measurements of combustion products by a gas
chromatographic system. This innovation was accompanied by
little or no net increase in accuracy or speed of analysis,
and the additional requirements of a gas chromatograph and
a liquid nitrogen system limited the technique from an
economical standpoint.
It was proposed, therefore, that a system based on
a dynamic sensing device be designed which would provide a
more rapid, efficient, and economical means for the micro-
determination of carbon and hydrogen. To achieve maximum
speed of analysis, initial experimental procedures were
based upon the use of small samples and a method for flash
combustion providing rapid pyrolysis. In addition, it was
desired to avoid retention of combustion products. Evalu¬
ation of this approach was facilitated by construction of
8

9
the system illustrated in Figure 1. A pipe T coupled to a
short tube filled with calcium carbide (30 mesh) was
inserted between the injection port and a 1 foot silica gel
(60-80 mesh) column of a commercial chromatograph. A
pyrolyzer probe consisting of a platinum filament connected
to a high current transformer was coated with a small
amount of benzoic acid and was fitted into the T-connector
which, in turn, was wrapped with heating tape to prevent
condensation of water. When equilibrium had been estab¬
lished, a current of 55 amperes was passed through the
platinum loop for a period of 1 second, affording a measured
ignition temperature of 1000°C. Using pure oxygen as the
carrier gas, combustion and conversion products were
separated and were swept through a 100,000 ohm thermistor
detector. Signals corresponding to the resultant carbon
dioxide and acetylene were similar to those of the illus¬
trative chromatogram in Figure 1. Although the thermal
conductivities of carbon dioxide and acetylene are both
lower than that of oxygen, a negative peak was observed for
low concentrations of acetylene. With increasing acetylene
concentrations, W-shaped peaks occurred, exhibiting
alternate negative and positive deflections. These were
due to the maximum occurring in the thermal conductivity-
composition curve of the binary mixture at low
acetylene concentrations.

10
Fig. 1.-Preliminary System and Readout for Carbon and
Hydrogen Analysis

11
In spite of inherent procedural difficulties, the
system indicated the potential of a direct combustion-
detection approach to carbon and hydrogen analysis, and
subsequent techniques were based on improvement of this
method.
Selection of Components and Operational Procedures
Even though efforts were directed, initially, toward
the development of an instrumental technique for which an
eventual commercial apparatus would be readily adaptable,
alternate methods were studied to establish a system of
optimum efficiency and accuracy» Basic theoretical and
experimental considerations governing the choice of
components, conditions, and their interrelationships were
extended, where possible, to enable additional elemental
analyses with appropriate modifications.
Combustion and sample introduction systems
The preliminary experimental method for oxidative
degradation proved unsatisfactory for a number of reasons.
Sublimation or pre-vaporization of samples with high vapor
pressures was unavoidable. Efficient weighing or sample¬
handling techniques could not be established. A lack of
combustion spontaneity resulted in peak asymmetry, pre¬
venting the use of peak height measurement. A definite but
insufficient improvement was made by fabricating a small,

12
tight, conical coil of #22 gauge platinum wire and pro¬
viding for its insertion into a quartz combustión tube,
thus enabling the weighing of solid samples. Passage of a
high current through the cone provided rapid but incomplete
combustion of the sample. In the case of benzoic acid, for
example, a substantial amount of the sublimate appeared on
the walls of the quartz tube. A further concept involved
the use of a weighable platinum tube inserted between two
heavy electrodes through which the carrier gas was passed.
Difficulties in these approaches included the unavoidable
disturbance of equilibrium conditions prior to the intro¬
duction of each sample and severe initial losses through
volatilization or spattering due to the finite time re¬
quired for the system to reach a threshold combustion
temperature.
It became evident that a stationary packed column
heated to the highest practical temperature and coupled
with an appropriate method for sample introduction would
eliminate these difficulties and would provide maximum
combustion efficiency. A Vycor tube (18 in. x 7 mm I.D.)
with a side arm (one inch from the upper end) for oxygen
flow was packed with wire-form copper oxide, and the upper
end was fitted with a servan cap. The tube was vertically
connected to a chromatographic system and was wrapped with
60 turns of #22 gauge Nichrome wire to which approximately

13
70 volts were applied. Careful introduction of small
varying amounts of ethanol by means of a 10 ¡al precision
syringe produced sharp, symmetrical carbon dioxide peaks
displaying good linearity. Provision for introduction of
solids without interruption of the carrier gas flow was
made by devising a sliding valve which positioned the
sample between two parallel plates and subsequently trans¬
ported it into the hot zone by means of a gravity feed as
illustrated in Figure 2. Details of an improved valve,
based on a rotary concept, which proved to be extremely
satisfactory during the remainder of the investigation,
are presented in a later section.
Further improvements in the combustion system pro¬
vided ignition temperatures of approximately 1050°C and
allowed furnace operation on line voltage. The tube was
packed to the half-way mark with copper oxide but was heated
at least 6 inches above this level to facilitate decompo¬
sition of flashback pyrolysis products. The spacing of the
Nichrome coils was adjusted for provision of highest
temperature at the copper oxide surface, the site at which
sample decomposition occurred.
Detection systems
Determination of combustion product concentrations
by relative thermal conductivity measurements provided a
satisfactory method for initial experimental work. The

14
Sample
i
Fig. 2.-Preliminary Design of Sample Introduction System

15
general requirements of stability and speed and linearity
of response were satisfied, and adequate sensitivity to
carbon dioxide was obtained. However, because of tbe use
of oxygen as a carrier gas and tbe similarities in thermal
conductivities for oxygen and water, it was not practical
to measure water directly by this technique. Other detec¬
tion systems were constructed and tested in an initial
effort to eliminate conversion requirements. The hydrogen
flame detector, used conventionally, shows little or no
response to water, carbon dioxide, or other permanent gases.
Using this apparent disadvantage, provision was made for
the premixing of hydrogen, containing an ethanol contaminant,
with the oxygen carrier gas and for the introduction of
this mixture into an appropriate burner-collector electrode
assembly. It was assumed that the diluent effect of the
normally non-responsive carbon dioxide and water would
result in significant negative response in comparison to
the elevated background signal provided by the constant
ethanol bleed. The procedure was discarded, however, due
to its relative complexity and critical operational
requirements.
A second approach based on conductance measurements
utilized a hygrometer element consisting of thin palladium
wires wound concentrically about a plastic core which was
coated with a small amount of lithium chloride in a

16
polyvinyl alcohol hinder. The element was positioned in a
tube and the electrodes were connected to an a.c. bridge
circuit, the unbalance of which was representative of the
quantity of water passing through the detector. Improved
results were obtained with a more favorable cell geometry
approaching point source detection. Six parallel strands
of thin wire were evenly spaced along the inside wall of a
short length of 1/16 inch I. D. heavy-walled pyrex tubing.
A light coating of polyvinyl alcohol was applied and
alternate strands of wire were joined and connected to the
simple d.c. circuit shown in Figure 3« The response was
extremely rapid and was specific for water. Carbon
dioxide was determined by a conventional thermal conduc¬
tivity detector. Linearity and reproducibility, however,
did not meet the requirements of the determination. A
variety of other detection systems were considered but from
a standpoint of simplicity and reliability, thermal
conductivity methods proved most satisfactory.
In choosing an appropriate katharometer, many factors
were taken into consideration. Although the filament
detector generally exhibits greater stability, conventional
glass-clad thermistors were selected in preference to hot
wires because of their decreased susceptibility to oxidation,
and their higher ambient temperature response in an oxygen
atmosphere. The cell was designed in such a manner that

17
Ther-
mistor
Circuit
•
Fig. 3.-Circuit for Conductance Measurement of Water and
Thermal Conductivity Measurement of Carbon Dioxide

18
toth the reference and the sample sensing elements were
located in the direct path of the carrier gas. Unlike the
diffusion cell, this provided a rapid quantitative response
which was somewhat sensitive to flow fluctuations.
Thermistors with a room temperature resistance of 100,000
ohms were used for initial studies, but later work indicated
improved stability with less expensive 2,000 ohm thermistors.
The desirable linearity of output signal with
respect to concentration is not completely predictable from
theoretical considerations. A strict mathematical treat¬
ment taking into account all subsidiary effects is
complicated and impractical due to a lack of accurate data.
In general, the temperature-response behavior of a sensing
thermistor over a narrow temperature range may be repre¬
sented by
where the coefficient «is a function of temperature and
is given approximately by
T‘
where the constant /ñ is defined by
1

19
Thus, within a small temperature interval, thermistor
resistance is a linear function of temperature. The
response, dE/dy^ , to a particular component has been
shown by Smith and Bowden (34) to correspond to
dE dE dT dh
dyg 3f 3E dy-g *
where E is the out-of-halance bridge voltage, y2 is the
mole fraction of the non-carrier component in the detector,
T is the sensing thermistor temperature, and h is the
thermistor surface heat transfer coefficient. The dE/dT
and dT/dh derivatives are calculated from instrument
characteristics while dh/dyg depends essentially on the
thermal conductivity of the binary gas mixture in the de¬
tector. The possible effect of non-ideality of the binary
mixtures, oxygen-carbon dioxide and oxygen-hydrogen with
respect to thermal conductivity was considered. The
analogous nitrogen-carbon dioxide mixture, for example,
expresses a negative deviation from ideality as illustrated
by Keulemans (24). Departures from linearity, however, due
to these and other factors were either compensatory or
negligible in the experimental studies. This was evident
from the excellent linearity over a concentration range at
least as wide as that governed by combustion restrictions.
Results are presented in a later section.

20
A two transistor zener-stabilized power supply with
a variable d.c. output voltage oí 14-,8-26.3 volts was con¬
structed. The excellent regulation obtained in spite of
large fluctuations in line voltage is shown in Figure 19.
A Wheatstone bridge circuit, designed to accept a wide
resistance range of thermistors and to provide for suitable
sensitivity and attenuation variation, was constructed and
used in all studies.
Conversion systems
A three-fold advantage was realized in the replace¬
ment of a direct thermal detection of water by the measure¬
ment of an appropriate gaseous conversion product. When
retention techniques are required, anomalous desorption of
water from active column sites causes non-reproducible
tailing, and the eventual change in the substrate-solid
support ratio results in variations in retention and response.
The sensitivity of a katharometer to water in an oxygen
carrier gas is too low to provide a signal-to-noise ratio
commensurate with the precision and accuracy requirements
of the analysis. Thermostatic control of the column-
detector system at a temperature in excess of 100°C is
avoided because ambient temperatures may be used in con¬
version product measurement.
The initial experimental system, previously discussed,
utilized a conversion of water to acetylene by reaction with

21
calcium carbide. Sensitivity was more than adequate with
respect to carbon dioxide but was inadequate with respect
to acetylene. In addition, anomalous negative response was
observed for acetylene due to longitudinal diffusion within
the column, with subsequent depression of its maximum
concentration in the carrier gas. An immediate improvement
developed from a consideration of the reaction sequence
CmHn + (m + £)02^m C02 + § HgO
| H20 + | CaC2 —- £ Ca(0H)2 + £ C2H2
in contrast to which the reaction of water with an active
metal hydride
| H20 + 2 CaH2 —Ca(0H)2 + §
would provide a distinct increase in sensitivity as a result
of the more favorable stoichiometry and the higher thermal
conductivity of hydrogen. In addition, non-uniformities
resulting from peak spreading could be eliminated due to
the elution of hydrogen prior to carbon dioxide.
Detrimental effects of two reactions on quantitative
measurements of carbon dioxide were considered
2 C02 + CaH2 —— Ca(C00H)2
GO 2 + Ca(0H)2 —►■CaCO^ + H20 .
It was proven experimentally, however, that under the
temperature and flow conditions of the analysis no carbon

22
dioxide retention occurred as a result of its passage
through calcium hydride or calcium hydroxide. This was
indicated by a series of careful injections of carbon
dioxide at points immediately preceding and immediately
following a tube packed with calcium hydride. Similar
injections were made after total depletion of the calcium
hydride with water. Peak heights, corresponding to carbon
dioxide concentrations, were completely unaffected by
either of the packings.
Calcium hydride was selected over a series of other
metal hydrides for several reasons. Its reaction with
water to release hydrogen was rapid and quantitative. Prom
a practical standpoint, it was found to be more easily
handled than the sodium, lithium, or lithium-aluminum
analogs; i.e., a container of calcium hydride could be
opened and exposed briefly to the atmosphere innumerable
times without observable decomposition. Its hydroxide
displayed the least tendency to retain water, as evidenced
by the pronounced reduction in tailing over that experienced
with the other hydroxides. This corresponded to prediction
because calcium hydride is neither hygroscopic nor does it
form a hydrate. Its small but definite affinity for water
under the conditions of the preliminary systems, however,
was sufficient to produce critical design problems.
An initial conversion system consisted of a short

23
pyrex tube equal in diameter to tbe combustion tube and
packed with, approximately 1/4 inch, of 40 mesh calcium
hydride. The peak height response was observed to decrease
markedly after the combustion of three or four samples.
Asymmetric peaks were particularly evident in the case of
large samples or those with a substantial hydrogen content
and were attributed, in part, to the observable condensation
of water prior to its passage through the calcium hydride,
resulting in extremely slow evolution of hydrogen. A
corrective procedure was attempted in which the conversion
tube was placed in an oven controlled at 110°C. The system
was first tested for evidences of thermal degradation of
the calcium hydride by plotting the effluent signal versus
temperature as indicated by a thermometer attached to the
tube. No hydrogen evolution was noted during elevation of
the oven temperature to 130°C. Although some improvement
due to elimination of condensation problems was observed,
no noticeable gain in reactivity or reduction in tailing
accompanied the temperature increase. Resultant erratic
values for hydrogen precluded the use of this system.
A unique application of an apparatus patterned after
the electrolytic moisture detector of Keidel (23) was
evaluated in an effort to obtain a more spontaneous release
of hydrogen. A hygrometer element, identical to the one
previously described under "Detection systems" was coated

24
by immersion in syrupy phosphoric acid. Subsequent de¬
hydration by applying a voltage to the electrodes until no
further hydrogen evolution was observed left a thin film of
phosphorus pentoxide. The element was encased in a glass
sleeve and was positioned directly below the combustion
tube. Carbon dioxide generated during a combustion process
was unretained and was detected in the normal manner. The
water, however, was irreversibly absorbed by the phosphorus
pentoxide. Application of a relatively high voltage to the
electrodes resulted in the rapid electrolytic release of
oxygen and hydrogen into the oxygen carrier stream. The
hydrogen was detected conventionally by thermal conductivity.
The conversion, however, was found to be incomplete and
losses were incurred through inefficient collection of water
on the phosphorus pentoxide.
A number of additional chemical conversions were
investigated, including the reaction of water with aluminum
carbide to produce methane. This proved ineffective due
to the extremely slow reaction rate. Rejection of other
systems was based also on reactivity or detection deficien¬
cies.
A simple and accurate method for measurement of water
was eventually devised which eliminated virtually all of the
difficulties previously encountered. A standard unsealed
Kimax melting point capillary tube was fitted at one end

25
with a short plug of asbestos and was packed with a 1 inch
layer of silica gel (60-80 mesh). Tight connection to the
combustion tube was made through a serum vial cap, and a
1/8 inch I. D. pyrex tube packed with a 1/4 inch layer of
calcium hydride was positioned directly below the capillary
tube by means of a silicone rubber insert. The calcium
hydride was pushed tightly against the seal to eliminate
all possibility of condensation prior to conversion. A
small moving furnace designed to operate at a constant
temperature of approximately 200°C encircled the capillary
tube as indicated in Figure 4. The velocity of the carrier
gas through the capillary was nearly twenty times greater
than its velocity through the combustion tube or through
previously designed conversion tubes. This was indicated
by the reciprocal ratio of the respective cross-sectional
areas. Similarly, a five-fold increase in oxygen velocity
was calculated for the calcium hydride tube. The extremely
high transfer rate of.carbon dioxide through this portion
of the system eliminated any possibility of its retention
by the silica gel or its reaction with calcium hydride or
calcium hydroxide. Water, on the other hand, was retained
with excellent efficiency by the silica gel and in no
instances, regardless of sample size or hydrogen content,
was an overload of the desiccant noted. The slight conden¬
sation on the walls of the capillary tube, which was

26
Silver Wool
Combustion Tube
Rubber Serum Cap
Moving Furnace
Capillary Tube
Silica Gel Packing
Silicone Rubber Sleeve
Calcium Hydride
Glass Wool Retainer
Fig. 4.-Details of tbe Retention and Conversion System

27
sometimes observed immediately after a sample combustion,
was of no consequence since all water droplets were swept
into the silica gel before the appearance of the carbon
dioxide peak. Positioning the moving furnace at the upper
portion of the capillary tube as well as the use of silver
wool as a heat transfer agent in the combustion tube aided
significantly in the prevention of condensation. In an
actual analysis the sample was introduced with the moving
furnace remaining in the upper position. Immediately after
observance of the carbon dioxide peak maximum (30-45
seconds), the furnace was dropped to its lower position
where it was seated against the silicone rubber insert for
3-5 seconds, vaporizing the entrapped water almost
instantaneously. Attenuation and recorder lead polarity
were adjusted and the resultant signal for hydrogen appeared
within 30-45 seconds. Incorporation of the calcium hydride
in the capillary tube directly below the silica gel proved
unsatisfactory due to prohibitively large flow restrictions
arising from the expansion accompanying the conversion to
calcium hydroxide. The use of a larger tube prevented this
difficulty, facilitated rapid hydride replacement, and
insured uniform release of hydrogen.
Readout systems
It was presumed at the start of the investigation
that a readout system providing signal integration over the

28
time coordinate would be required for an analysis. As work
progressed, it was observed that combustion rates were suf¬
ficiently constant, for virtually all substances encountered,
that peak height measurements could be used advantageously.
This provided a simpler and more rapid readout, offered a
wider chart span on the same recorder, eliminated the
additional cost of integrating equipment, and facilitated
adaptation to a digital readout system, if desired.
A Minneapolis Honeywell 1 millivolt Brown recorder
with a response time of 1 second, full scale, was used for
all reported results. The specified accuracy, of + 0.25 per
cent was recognized to be a limiting factor although perfor¬
mance data for the system indicated this to be a somewhat
conservative figure. Because the maximum bridge out-of-
balance voltage was in excess of 10 millivolts, a Sargent
Model TR recorder was preferentially selected for use with
the prototype instrument discussed in the Appendix, by
virtue of the + 0.1 per cent accuracy specified for this
range. The Sargent recorder offers the additional advantage
of an improved pen system and a displacement adjustment in
spite of a slightly reduced chart span (9'A in. vs. 10 in.).
Response magnitudes for carbon dioxide and hydrogen are
approximately proportional to the differences between their
respective thermal conductivities and that of oxygen.

29
Setting A^Cq2>02) - 1 we find AX^^j a _ 15 at 250(3.
For most compounds a two- to four-fold attenuation was
required to adjust the amplitude of the hydrogen signal to
approximately that of carbon dioxide. A double-pole double-
throw switching mechanism was provided for reversal of
recorder-lead polarity, enabling the use of the full chart
span for both carbon dioxide and hydrogen peaks.
Column systems and arrangement of components
The preliminary experimental procedure relied upon
the use of a silica gel column for the separation of carbon
dioxide and acetylene. In later work, a silica gel packing
was also used to separate hydrogen from carbon dioxide.
Two basic arrangements of components, and the resultant
readouts, with no electrical reversal of polarity, are
shown in Figure 5* The second system was devised for a
more completely automatic approach and provided a re-
inversion of the hydrogen signal thus facilitating the use
of automatic attenuation. The procedure involved passage
of the combustion and conversion products through the silica
gel on which carbon dioxide was initially retained while
hydrogen was swept through the first side of the detector
producing a negative signal. During the flow of hydrogen
through the delay column, the carbon dioxide was eluted
from the silica gel producing a positive signal on the re¬
corder, and then flowed into the ascarite filter where it

°2~
â–  Sample Port
â–  Combustion Tube
â–  Calcium Hydride
Silica Gel Column
Detector
CO-,
30
°2 —
Sample Port
Combustion Tube
Calcium Hydride
Silica Gel Column
co0
Detector
• Ascarite
Delay Column IT
Ho
u
Fig. 5--Variations in Component Arrangements and Readouts
for Carbon and Hydrogen Analysis

31
was permanently retained. The delay column which consisted
of approximately 50 feet of empty 1/8 inch plastic tubing
was so adjusted in length that after the first side of the
detector was cleared of carbon dioxide, the hydrogen passed
through the second side of the detector, thereby producing
a positive signal on the recorder. Diffusion of hydrogen
was almost negligible during its long transport through
the column, but analytical results suffered from such
factors as non-uniform release of hydrogen, increased column
impedance due to carbon dioxide absorption by the ascarite,
and non-linearity of response to carbon dioxide. Carbon
dioxide peaks were non-Gaussian, exhibiting pronounced
asymmetry w-tth increasing sample size. No specific improve¬
ment in elution symmetry was noted upon replacement of the
silica gel with a 6 foot column packed with 50 per cent
hexamethylphosphoramide on chromosorb. A number of problems
associated with the combined usage of oxygen as a carrier
gas and a silica gel column for retention of carbon dioxide
became apparent. Although sensitivity to hydrogen was more
than adequate, it was necessary to utilize the maximum
bridge output with a 1 millivolt recorder to obtain signifi¬
cant peak amplitudes for carbon dioxide. Unfavorable
signal-to-noise ratios were observed under these conditions,
and a large attenuation ratio was required for measurement
of the two components. Column overload at higher concen¬
tration levels caused deviations from linearity requiring

52
corrective working curves, and excessive peak widths for
carbon dioxide resulted in undesirably long analysis times.
Of even greater consequence were indications of changes
in column characteristics during extended operation of the
system, resulting in variations in both retention volume
and response.
To test the capabilities of a more ideal system
requiring no chromatographic separation, the calcium
hydride was replaced by a desiccant, and a 1/8 inch x 8
foot length of empty plastic tubing was substituted for the
silica gel column. This allowed observation of performance
characteristics with regard to measurement of carbon
dioxide, initially circumventing any problems associated
with the determination of water. Polyethylene samples were
used for the test runs due to weighing convenience. Figure
6 indicates the improvement in sensitivity provided by this
system as compared to that using a silica gel column. Flow
rates were identical (40 cc/min.) and, in each instance,
the time required for the initial appearance of the carbon
dioxide peak was virtually the same. The seven-to-eight-
fold increase in sensitivity was accompanied by excellent
linearity throughout an even wider range of concentrations
than was normally suggested (further data are presented in
a later section), and stability, readout ease, and speed
of analysis were vastly improved. The accuracy afforded by

COp Peak; Height, it
33
Weight of Polyethylene, mg
Fig. 6.-Comparison of Sensitivity to Carhon Dioxide Using
Packed and Unpacked Columns Under Identical Flow
Conditions

34
the combined combustion and detection techniques was next
tested, and results are presented in Table 1.
TABLE 1
TEST RESULTS FOR ACCURACY OF CARBON ANALYSIS
Compound
%C, Found
%C, Theoretical
/S-Naphthol
83.5
83.3
Naphthalene
93.6
93.7
Benzophenone
85.8
85.7
Oxalic acid
19.0
19.0
Phthalic acid
57.8
57.8
9-Fluorenone
86.6
86.6
Triphenylphosphine
82.4
82.4
Sample handling and weighing;
A Cahn Gram Electrobalance (Model G, Cahn Instrument
Company, Paramount, California) was selected for all sample
weighings. The generad performance of the balance proved
to be outstanding due to its insensitivity to environmental
conditions such as vibration, temperature variation, air
currents, etc. The specified limits of precision and
accuracy for the weight range of interest are 0.1 micrograms
and 0.5 micrograms, respectively. Exact proportionality,
however, rather than absolute weighing accuracy is essential
to this work. Because very critical weighings were required
the balance was zeroed with each determination. The

35
calibration was found to remain extremely constant, but
was checked, nevertheless, after every five or six
determinations. For ultimate precision, the 0-1 milligram
range of the balance was used for all weighings, even
though, in some instances, the total weight exceeded 1
milligram. When it became necessary to encapsulate a
sample, a balancing capsule weighing slightly less than
the sample capsule was placed on the righthand pan, and
the net tare weight was obtained through electrical sub¬
traction by the zero circuit. If, in this procedure, the
total weights of the net tare and the sample were in excess
of 1 milligram, the 1 milligram calibrating weight was
added to the righthand pan and the additional weight was
obtained in the usual manner.
For non-crystalline or viscous liquid samples,
containers were fabricated from 6 mm squares of aluminum
foil formed about a 1.5 nim rod and crimped on one end.
After sample introduction the opposite end was crimped and
the weighing was completed. A preferred method of sample
enclosure required the use of seamless copper capsules of
equivalent dimensions. These facilitated the weighing of
volatile liquid samples and continuously provided a fresh
layer of copper oxide rather than aluminum oxide at the
combustion site. They were not used during the course of
the investigation due to difficulties in fabrication.

36
However, it was later found, that, where an extrusion process
failed, an electrodeposition of copper on a slightly tapered
iron mandrel proved quite satisfactory. The high co¬
efficient of thermal expansion of copper allowed removal of
the capsule hy immersion in boiling water.
Critical Paryneters in the Finalized System and
Extension to Nitrogen Determination
Evaluation of all factors previously discussed led
to the construction of an apparatus using the flow system
illustrated in Figure 7. This arrangement of components
was selected by virtue of expediency and its provision for
maximum experimental versatility. A similar, but somewhat
more advantageous arrangement was used in the design of the
prototype instrument discussed in the Appendix.
Choice of oxygen as a carrier gas
The necessity for an oxygen atmosphere to provide
maximum combustion efficiency for carbon and hydrogen
analyses was recognized by Liebig and Pregl in their
initial studies. In the present system where an extremely
rapid and uniform release of combustion products is
essential, an inert carrier gas such as helium or nitrogen
can not be used for general application. The inadequacy of
a non-oxidizing carrier gas is shown in Figure 8. Nitrogen
was selected for this test rather than helium due to the

Sample Port
Combustión Tube
Moving Heater
Capillary Tube
Silica Gel
Calcium Hydride
Filter
Delay Column
Detector
Reversal
Fig. 7.-Finalized Flow System and Readout for Carbon and
Hydrogen Analysis

Fig. 8.-Comparison of Combustion Efficiencies Using Oxyge:
and Nitrogen Carrier Gases
V>J
00

39
near identity of its thermal conductivity with that of
oxygen. The carhon dioxide response factors for benzoic
acid were similar; Rn = 1.13 mv/mg and R„ = 1.10 mv/mg.
u2 w2
However, upon introduction of a sample of activated carbon
(approximately 60% C), R,-, = 0.98 mv/mg whereas virtually
u2
no response was observed for an even larger sample of
carbon using a nitrogen carrier gas. A second carbon
sample was encapsulated and was surrounded with approxi¬
mately 5 milligrams of silver permanganate. The carbon
dioxide peak height response in the nitrogen carrier gas
was, in this case, 0.28 mv/mg with some evidence of peak
spreading.
Further information regarding combustion spontaneity
in oxygen was derived from the following data. Peak height
to half-peak-width ratios were calculated for a series of
samples ranging in ease of combustion. Sample weights were
chosen so that peak height magnitudes remained essentially
the same. The maximum chart speed was used to allow more
accurate half-peak-width measurements. The results, as
shown in Table 2, indicated no inconsistencies or peak
spreading. Oxygen provided the additional advantage of
rapid regeneration of copper oxide which may have been
reduced during the combustion process.
A differential thermal analysis of copper oxide
versus carborundum was conducted in a helium flow. The

40
TABLE 2
PEAK HEIGHT TO HALF-PEAK-WIDTH RATIOS
Sample
Wt,Mg
Peak
Height
mm
Half-Peak
Width
mm
PH/HPW
Benzoic acid
661
196
4.2
46.7
Biphenyl
497
201
4.3
46.8
Potassium acid
phthalate
906
184
3.9
46.2
Caffeine
902
193
4.2
45.9
Teflon
1730
176
3.9
45.1
Activated carbon
(~ 60 per cent C)
722
187
4.0
46.8

41
resultant endotherm occurring initially at 925°C suggested
a partial degradation of the copper oxide at this temperature.
This was substantiated in later work by evidences of oxygen
release under similar conditions. A second differential
thermal analysis of copper oxide, substituting oxygen for
helium, showed no thermal band below a temperature of
1055°C. In excess of this temperature an endotherm corres¬
ponding to a slow fusion process was noted. Another
important advantage, therefore, was realized in the use of
oxygen rather than an inert carrier gas because its
repression of copper oxide decomposition allowed the use
of a higher furnace temperature with a resultant increase
in combustion efficiency.
Combustion tube packing and retention systems
The oxidizing portion of the combustion tube for a
conventional system for microanalysis of carbon and hydrogen
is packed with a mixture of 2 parts of copper oxide and 1
part of lead chromate, according to the procedure suggested
by Steyermark (38). The furnace temperature for this
system, however, is relatively low (680°C) as opposed to
that of the method under investigation (> 1000°C). Experi¬
mental results with lead chromate-copper oxide mixtures
indicated that it was unnecessary, in the present system,
to add an additional oxidant to the copper oxide. Later
work, however, proved the necessity for addition of a

42
section of platinized asbestos to tbe combustion tube in a
similar technique for nitrogen determinations. This
functioned to catalyze the combustion of any unburned
pyrolysis products. Although the use of an oxygen atmos¬
phere substantially reduced this possibility, it was
considered advantageous to include a layer of platinized
asbestos in the carbon and hydrogen combustion tube.
Chlorine, bromine, iodine, and oxides of sulfur were
initially retained on a lH inch section of tightly rolled
silver gauze positioned in the lower portion of the
combustion tube. This, however, provided a relatively small
surface area for reduction and required indentations in the
walls of the tube to prevent shifting. On the other hand,
silver wool was found to retain its position, to provide a
much larger surface area, and to be much easier to clean.
It functioned also, as an excellent heat transfer medium,
assisting in the prevention of condensation of water in the
lower portion of the tube. Compounds with a high fluorine
content were analyzed successfully with no special pre¬
cautions. However, for the continual analysis of fluoro-
compounds, a layer of magnesium oxide or aluminum oxide
positioned between the platinized asbestos and the copper
oxide is advocated.
Combustion of nitrogen-containing compounds in an
oxygen atmosphere would be expected to produce, essentially,

a mixture of nitrogen dioxide and its dimer, dinitrogen
tetroxide, which would include the oxidation product of any
nitric oxide initially formed. Certain organics, such as
diazo-compounds, would he presumed to decompose with the
release of elemental nitrogen, and the combustion temperature
would he far too low for any oxidation to the monoxide.
Because the thermal conductivities of nitrogen and oxygen
are nearly identical, virtually no response to elemental
nitrogen would he superimposed upon the carbon dioxide
signal. A number of reactants have been proposed for the
retention of nitrogen dioxide. Traditionally, lead peroxide
maintained at a temperature of 175-190°C is used (57), and
absorbs nitrogen dioxide according to the reaction
2 N02 + 2 Pb02-*-Pb(N02)2 + PbO + \ 02 .
More recently, specially prepared manganese dioxide has
been used for this purpose. Onoe (32) found that activated
manganese dioxide prepared from manganese sulfate and
ammonium peroxydisulfate or from potassium permanganate and
methanol gave the best results, and concluded that the
reaction was simply
Mn02 + 2 N02 —Mn(N03)2 .
Vecera and Snobl (43) maintained that the hydrated amorphous
gel [MnO(OH)2] binds nitrogen dioxide with its surface
hydroxyl groups according to the equation

44
Mn(OH)4 + U204—Ma(N03)2 + 2 HgO ,
the amount of absorbed nitrogen dioxide being in proportion
to the surface. They observed excellent absorbent
properties for this gel and described a procedure for its
preparation. Experimentally, a short filter packed with a
microanalytical grade of manganese dioxide followed by a
layer of magnesium perchlorate was used. Its efficiency
was tested by injecting 5 cc vollimes of nitrogen dioxide
and sulfur dioxide immediately before and after the filter.
Complete retention was found even after numerous repeated
inj ections.
It was also necessary to provide a desiccant prior
to the combustion tube to remove traces of moisture present
in the oxygen carrier gas. This was particularly essential
because the silica gel capillary tube served to accumulate
water over extended periods of operation. Trusell and Diehl
(42) indicated the superior desiccant properties of anhydrous
magnesium perchlorate and, in this application, it was found
preferable to the 5A molecular sieve originally used. For
oxygen containing an appreciable amount of hydrogen or
hydrocarbons, initial purification using a pre-burner and
a carbon dioxide-water absorption system would be required.
Thermal equilibrium and flow rate
Two factors contributed significantly to variations

45
in detector response. The influence of block temperature,
a function of the time allowed for the detector to reach
thermal equilibrium, is shown in Figure 9. Because
conduction of heat away from the thermistors increases the
temperature of the cell block until an equilibrium
temperature is reached, a decrease in response to carbon
dioxide with time is anticipated due to the convergence of
the thermal conductivities of oxygen and carbon dioxide at
higher temperatures. Little or no response change is
expected or observed for hydrogen since the ratio of the
thermal conductivities for hydrogen and oxygen remain
essentially constant over a wide temperature range. It
is evident, however, that the system should be allowed to
"warm up" for at least one hour before an analysis is run,
and that frequent standardizations are required for the
first two and a half-hour period.
The necessity for a critical control of the oxygen
flow rate is indicated in Figure 10. The diminishing
response to carbon dioxide corresponds to that normally
encountered when a carrier gas flow rate is increased to a
point where the residence time of the component in the cell
has a significant effect on the resultant signal. It is
apparent, in this instance, that if measurements of carbon
dioxide were based on peak areas rather than peak heights,
any errors due to flow variations would be severely

COp Peak Height Response, mv/mg Polyethylene
46
Time Elapsed, Min., After Switching on Thermistor
Detector
Fig. 9.-Response to Carbon Dioxide Versus Equilibration Time

Peak Height Response, mv/mg Benzoic Acid
47
Pig. 10.-Dependence of Response on Plow Rate

48
magnified since peak widths and peak heights decrease
simultaneously with increased flow. A similar decrease in
response to hydrogen was observed but only at much higher
flow rates. The response maximum for methane in oxygen
was found to be intermediate between that for carbon
dioxide and that for hydrogen.
Slight variations in flow impedance occur within the
system for numerous reasons. Changes in line voltage alter
the furnace temperature resulting in expansion or contraction
of the copper oxide. Increases in the amount and the degree
of compression of calcium hydroxide in the converter tube
as well as expansion due to formation of calcium hydroxide
produce additional flow restrictions. Eapid increase in
temperature of the silica gel in the capillary tube with
subsequent expansion introduces transient flow changes. In
order to minimize or to eliminate these effects a throttling
orifice was constructed and installed in the system. The
flow of gas through an orifice depends only to a certain
extent on the pressure differential across the orifice.
Where the upstream pressure is held constant and the pressure
at the discharge end is decreased, the exit gas flow will
increase to a critical value. Further decreases in the
downstream pressure will cause no increase in flow until the
critical flow condition has been exceeded. The critical
condition is obtained for an oxygen flow through an orifice

if the upstream pressure is approximately two or more times
that of the downstream pressure (2). The equation for
critical flow is
q - CApKVMÜTÍ ,
where q is the mass rate of flow of the gas¡ C is the dis¬
charge coefficient; A is the cross-sectional area of the
orifice; p is the upstream static pressure of the gas; K
is a dimensional coefficient involving the gas constant;
M is the molecular weight of the gas; T is the absolute
temperature of the gas upstream; and U is a constant that
is characteristic of the gas and is given by
U - k[2/(k + l)]s ,
where k is the ratio of specific heats (Cp/Cv) and
s > (k+1) / (k-1) .
An orifice of approximately 0.0025 inch diameter was
developed in a thin steel disk using the method suggested
by Marks (30). The ratio of the absolute upstream pressure
to the downstream pressure was measured and was found to be
roughly equal to 2 for this orifice. The equation for
critical flow indicates that a + 5° change in temperature
of the gas upstream would produce a corresponding change
in flow of + 0.8 per cent. Therefore, in the final proto¬
type model, the orifice and the secondary pressure regulator
were positioned in the constant temperature section of the
instrument

50
Nitrogen determination
A simple modification of the finalized system for
carbon and hydrogen analysis provided a rapid method for
the determination of nitrogen. In initial studies the
combustion tube was repacked with a 4 inch layer of copper
oxide (20 mesh) followed by a 4 inch layer of freshly
hydrogen-reduced copper (20 mesh) and a 1 inch layer of
silver wool. The conversion system was removed and, when
helium was used as a carrier gas, the manganese dioxide
cartridge was replaced by a filter packed with magnesium
perchlorate and ascarite. Only the desiccant was required
for a carbon dioxide carrier gas, and the combustion tube
was connected directly to the filter.
Sample analysis was identical to that in the carbon
and hydrogen determination with the exception that a
catalyst was generally added. Combustion products were
swept through the copper section of the tube where oxides
of nitrogen were reduced to elemental nitrogen. Halogens
and oxides of sulfur were retained by the silver, and water
and carbon dioxide were absorbed where required. The
nitrogen peak emerged in 30-45 seconds after introduction
of the sample. The reaction
Wo + (m + i)02 m C02 + I H2° + ! N2
is less favorable, stoichiometrically, for nitrogen than

51
for carbon dioxide formation. For those compounds containing
high percentages of nitrogen, a carbon dioxide carrier gas
afforded sufficient sensitivity for the analysis, but it
was found advantageous to use helium for general application.
Good combustion characteristics, with respect to nitrogen
release were observed for most compounds even though an
inert carrier gas was used. It is reasonable to assume
from thermodynamic considerations that, during the com¬
bustion process, free nitrogen would be formed even more
readily than carbon dioxide or water due to its higher
stability.
The small amount of air which is introduced with the
sample is of no consequence in the system for carbon and
hydrogen because oxygen and air have virtually identical
thermal conductivities. In the case of the nitrogen
analysis, however, where helium or carbon dioxide is used
as a carrier gas, a significant error may be introduced.
The slight Venturi effect within the valve draws the en¬
trapped air into the system and the subsequent peak is
partially superimposed upon the nitrogen peak which precedes
it. A modification of the procedure for sample introduction
reduced this contribution to a minimum. The valve (see
Figure 1?) was first rotated to the extreme clockwise
position to clear the inner sample compartment of air.
After a few seconds the encapsulated sample was dropped into

52
tlie upper compartment which was subsequently filled with a
loosely fitting aluminum plug» The valve was then quickly
rotated to the extreme counterclockwise position and hack
to the extreme clockwise position. Because the helium in
the inner compartment was under slight pressure, diffusion
of almost all outside air was prevented. The magnitude of
the resultant air peak was generally at least twenty times
less than that encountered during normal operation, and the
peak was often completely eliminated as in the case indi¬
cated in Figure 11. Due to the somewhat slow diffusion
process the peak was considerably broad, and followed that
corresponding to sample nitrogen with only slight overlap.
Therefore, the contribution of atmospheric nitrogen was
negligible providing the valve was operated in the manner
described.
High "pseudo-nitrogen peaks" were initially obtained
for several non-nitrogenous compounds such as sucrose,
dextrose, and citric acid. It was later discovered that
injections of methane produced the same effect which was
attributed to incomplete combustion and subsequent detection
of certain low molecular weight pyrolysis products. This
difficulty was completely eliminated by the addition of a
layer of platinized asbestos to the copper oxide portion
of the combustion tube packing.

53
~~mT
~rrn:
w
4?
O
Efcfi
• •:
Nitrogen Peak Resultant from Introduction
of Air During Normal Operation of Valve
■tí
-p
td
-p
o
«
t>
rH
aJ
—
'd —<
© -
-P :—
.p in Elimination of Air Peak By
° -- Modified Operation of Valve
ctí
>
1
4
mr.
Attenuation, XI (2 times that normally used during a
nitrogen determination)
Fig. 11.-Reduction of Atmospheric Contamination During
Nitrogen Analyses

Addition of catalysts
In only very few instances was it found necessary
or advantageous to add a combustion catalyst to the sample
being analyzed for carbon and hydrogen. Under less favorable
conditions, where larger samples and lower temperatures are
employed, those compounds containing the alkali or alkaline
earth metals in the absence of sulfur or phosphorus would
leave an ash of the carbonate resulting in low carbon
values. Steyermark (36) suggests the addition of potassium
dichromate or vanadium pentoxide in the analysis of these
compounds. Efficiencies of a large number of catalysts
have been evaluated (15,18,19,20,21,44) but it is generally
agreed that tricobalt tetroxide, cobaltic oxide plus silver,
and silver permanganate are among the most satisfactory.
Silver permanganate was prepared by the reaction of silver
nitrate and potassium permanganate, and was used as a
general catalyst where required.

SECTION III
CONSTRUCTIONAL AND OPERATIONAL DETAILS
Experimental System
The apparatus used for this study was assembled so
that a variety of modifications could be made simply and
rapidly. Major components were mounted on a network of
1/2 inch diameter rods which insured freedom from vibration.
Ambient temperature was not controlled, but remained
essentially constant at 25°C. It was found necessary to
provide thermal shielding for only three components; the
pre-filter, the sample introduction port, and the detector.
The instrumental arrangement is shown in Figures 12
and 13. U.S.P. grade oxygen (Linde Company, New York,
N.Y.) was used throughout the investigation and all con¬
tacted surfaces were thoroughly cleared of oil, grease, or
other contaminants. A standard two stage oxygen pressure
regulator (Linde Company) equipped with a 1/d inch straight
thread hex-nipple was used for upstream pressure control.
A female 1/d inch two-piece swivel hose connector was
modified to house the throttling orifice as shown in Figure
Id. The inner section of the hose end portion was machined
55

LLL
Fig. 12.-Instrumental Arrangement for Carbon and Hydrogen Analysis
cr

57
Pig. 13.-Details of Conversion and Detection Systems

58
Pig. 14.-Details of Orifice Mounting

59
to a flat surface to accommodate two 7/16 inch neoprene
o-rings between which the orifice disk was positioned. The
hole was made by denting the steel disk (0.010 in. x
7/16 in. D.) with a sharp punch and by carefully filing
the back until a breakthrough was obtained. The opening,
as observed under a microscope, appeared to be quite round
and was estimated to have a diameter of 0.0025 inch by
comparison to a wire of that size. Connection to the de¬
tector was made with heavy-walled tubing and all further
connections up to the combustion tube were made as short
and rigid as possible.
The detector block was machined from 504 stainless
steel and is shown in Figure 15. Individual empty cell
volumes were approximately 0.4 cc. Type G126, 2,000 ohm,
hermetically sealed Fenwal thermistors (Allied Electronics
Corporation, Chicago, Ill.) were seated in position. These
required no sealing washers. The leads were insulated with
teflon sleeves, and the cell was tested for leaks at an
oxygen pressure of 50 p.s.i.g. The detector assembly was
mounted in an aluminum minibox (2 5/4 x 2 1/8 x 1 5/8 in.),
and the spaces between the walls were tightly packed with
glass wool. The outlet from the reference cell was connected
directly to a plain end flow metering tube (tube size 1A-
15-1; Ace Glass Company, Vineland, F. J.) with a heavy wall
rubber coupling. The rotameter was connected, in turn, to

60
SIDE VIEW
Material: 304 Stainless
Steel
Scale: Full
Pig. 15.-Details of Cell Construction

61
a pyrex tube (18 in. x 7 mm O.D.) packed with. 20-60 mesh
magnesium perchlorate (Anhydrone; Coleman Instruments, Inc.
Maywood, Ill.) and 8-20 mesh ascarite (Arthur H. Thomas
Company, Philadelphia, Pa.) as indicated in Figure 16. The
tube was wrapped with a sheet of aluminum foil to reflect
heat radiated from the furnace in order to prevent thermal
expansion and contraction of the packing. A short heavy
rubber connector was used to join the filter tube and the
sample introduction port. Details of the valve assembly
and construction are presented in Figure 17. The body and
the central rotating plate were machined from 304 stainless
steel, and the upper and lower rotary port seats were
machined from Zel-F. All moving surfaces were hand-lapped
to prevent leakage. The 1/4 inch outlet allowed rapid
connection and replacement of the combustion tube by means
of a short piece of rubber tubing. The connector was pro¬
tected from the heat of the furnace by wrapping it with
aluminum foil. Because it was essential to maintain the
injection port at room temperature to prevent pre-vapori¬
zation of volatile samples, a large aluminum heat reflector
(6 1/2 x 10 in.) was positioned directly below the valve.
The ' ycor combustion tube (18 in. x 7 mm O.D.,
743172, Corning Glass Works, Corning, II.Y.) was packed with
copper oxide (Cuprox; 29-130, Coleman Instruments, Inc.),
silver metal wool (E. H. Sargent and Co., Chicago, Ill.) and

Magnesium Perchlorate Ascarite Magnesium Perchlorate
62
I
I
l
Oxygen
Filter
«d
U
Pi
Pi
o
o
u
0
> rH
rH O
•H O
CQ >
'd
0
N CO
•H O
d p
•tH CQ
P cd ¿3
H 03
Combustión
Tube for
Carbon and
Hydrogen
Determi¬
nation
•d
•H
&
u
d>
Pi
Pi
o
o
u
> rH
H O
•H O
CO >
'd
a>
« co
•H O
d p
•H CO
P 0)
cd ,a
rH 03
Combustion
Tube for
Nitrogen
Determi¬
nation
â– d e
&■§
oen
a a)
p od
•H *H
O U
rH *d
•H <13
CQ C5
CQ
“ CQ rH
Cd O
rH O
O >
cd
o w
Converter Tube
nn
03
P
a d
3 p
•H O
03 rH
03 .a
d Ü
faO d
cd 03
SPh
0
CO
0 0
ri *d
cd *H
bOK
S3
so
Nitrogen Di¬
oxide Filter
a>
Carbon Dioxide-
Water Filter
for Nitrogen
Determination
Fig. 16.-Tube Packings for Combustion, Absorption, and
Conversion

63
SIDE VIEW
Material: 304 Stainless
Steel
Scale: Pull
Pig. 17.-Details of Sample Introduction Port

64
5 per cent platinized asbestos (P-152, Fisher Scientific
Co., Pittsburgh, Pa.) as shown in Figure 16. For nitrogen
analysis, an additional packing of metallic copper (Cuprin;
29-120, Coleman Instruments, Inc.) was used, as indicated.
An open end Kimax Capillary Tube (1,5 i 100 mm) was
packed with 1 inch of 60-80 mesh silica gel (Coast Engineer¬
ing Company, California) and was connected to the combustion
tube by insertion through a 6 mm rubber sleeve type serum
stopper. The conversion tube (1 in. x 4 mm O.D.) was
packed with about a 5/16 inch layer of 40 mesh calcium
hydride (Metal Hydrides, Inc., Congress Street, Beverly,
Mass.) as shown in Figure 16. The hydride layer was re¬
tained by a tight plug of glass wool and was positioned
firmly against a 1/8 inch silicone rubber sleeve (0.125 in.
O.D., 0.062 in. I.D.; Reiss Manufacturing Corporation,
Little Falls, N. J.) which was resistant to the heat and
allowed rapid removal and replacement of the tube.
The filter for retaining oxides of nitrogen, con¬
sisting of a pyrex tube (3 in. x 7 mm O.D.) packed with
14-30 mesh activated manganese dioxide (33-110, Coleman
Instruments, Inc.) and magnesium perchlorate, is illustrated
in Figure 16. This was connected to a 6 foot length of
plastic tubing (3/16 in. O.D., 1/8 in. I.D.) which was
coiled around a cylindrical form. Plastic was chosen rather
than metal to reduce the possibility of surface adsorption

65
of hydrogen. A short length of rubber tubing joined the
column to the inlet of the reference detector, and the out¬
let was connected to a flow meter. The flow meter was of
the soap-film type and was constructed from a 5 cc Mohr
measuring pipet by cutting off the ends and attaching a
glass T fitted with a medicine dropper bulb. This allowed
more accurate and more rapid flow measurements than were
possible with conventional higher volume soap-film meters.
A stop watch with a 10 second sweep which could be read to
the nearest 0.02 second was used to check all flow rates.
An 11 foot length of #22 gauge nichrome wire was
used for the furnace winding. This provided 110 turns
about the 7 mm Vycor combustion tube and allowed 3 1/2
inch double-twisted leads at either end for reduction of
lead resistance. Small bolts were used to connect the
nichrome heater to the line cord. The coil was compressed
at the point corresponding to the surface of the copper
oxide and was expanded at either end. This provided
maximum heat at the sample contact point. Using an optical
pyrometer, the temperature of the copper oxide at line
voltage (~ 117 volts a.c.) operation was observed to lie
between 1000 and 1050°C. For the capillary tube heater,
approximately 11 inches of #24 gauge nichrome wire were
wound 34 times about a 3/52 inch mandrel leaving 1 inch
double-twisted leads at either end. The spiral was encased

66
in a pyrex tube which, was held in position with a standard
laboratory thermometer clamp. Vertical adjustment was
provided by sliding the clamp along a 1/2 inch diameter rod.
A Powerstat (The Superior Electric Company, Bristol, Conn.)
was used to supply the required 6 volts. Operational data
for both heaters are presented in Table 3-
TABLE 3
HEATER POWER REQUIREMENTS
Combustion
Tube Heater
Nichrome Wire Gauge No. 22
Wire Length, inches 132
Cold Resistance, ohms 3.0
Applied a.c. Voltage, volts 117
Current, amperes (at
operating temperature) 7.5
Power Requirements, watts 880
Capillary
Tube Heater
24
11
11.4
6
2
12
A schematic diagram of the regulated power supply
and bridge circuit is shown in Figure 18. The solid-state
power supply, using two transistors and a zener diode,
provided excellent regulation with minimum a.c. ripple.
Performance characteristics under extreme input voltage
variations are indicated in Figure 19.

Fig. 18.-Schematic Diagram of Power Supply and Bridge Circuit
cr>
-o

Output, Volts
68
Pig. 19.-Stability of Power Supply With Change in Input
Voltage

69
Wire-wound, bridge resistors (1% precision, 1/4 watt)
and potentiometers were used to reduce noise. The power
supply and the bridge circuit were housed in separate
chassis to provide greater flexibility and to expedite
circuit changes during the experimental work. The bridge
current-voltage relationship is shown in Table 4.
TABLE 4
VARIATION OF BRIDGE CURRENT WITH VOLTAGE
Bridge Voltage,
volts
14.8
16.0
18.0
20.0
22.0
24.0
26.3
Bridge Current,
ma
7
8
9
10
11
12
13
All analytical results were obtained using a bridge voltage
of 14.8 volts. A considerably greater signal-to-noise ratio
was obtained at this voltage which more than compensated for
the slight gain in sensitivity at increased voltages.
Operational Aspects
Standards and sampling procedure
Benzoic acid (N.B.S. Standard Sample #39g) was chosen
as a calibrating standard for all reported carbon and hydro¬
gen determinations. To allow easier and more rapid handling,

70
the benzoic acid was compressed by tbe method used for
preparation of KBr pellets. Small chips, exhibiting excel¬
lent coherence, were weighed and were introduced directly
into the system. Certain types of samples, however,
required encapsulation. Among these were liquids, non-
crystalline solids, and hygroscopic, air-oxidizable, or
explosive compounds. In addition, a few samples, such as
benzophenone, which exhibited a tendency to burn pre¬
maturely before contacting the copper oxide, or to adhere
to the walls of the upper portion of the combustion tube,
required special handling.
The capsules used during these studies weighed less
than 2 milligrams and were prepared by wrapping 6 mm squares
of aluminum foil around a 1.5 mm melting point tube. In the
case of liquids the sample was injected by means of a 1
microliter syringe (7001-N, The Hamilton Company, Inc.,
Whittier, California) into a small amount of combustion
catalyst which was pre-tared with the capsule. This served
to absorb the liquid and to minimize volatilization or
leakage during the weighing procedure. Highly volatile,
non-viscous samples could not be analyzed during the course
of this investigation, but it was felt that with the design
of a seamless capsule that these would present no difficulty
in future work. Powdered solids were handled by the method
illustrated in Figure 20. The capsule was formed in the

Fig. 20.-Encapsulation of Powdered Samples

72
normal manner on a 3/4 inch, length of capillary tube using
a pre-weighed square of aluminum. The other end of the tuhe
was inserted into the powder and an appropriate amount of
the sample was introduced into the capsule by means of a
metal plunger. In this manner the compound was exposed to
the air for only a few seconds before being permanently
sealed and no losses were incurred through spillage or
adherence to the outside walls.
The percentage composition of benzoic acid (68.85 %
C, 4.95 % H) approaches the mean composition expected for
most non-halogen-containing organic compounds. This factor,
as well as its availability in extremely pure form and its
lack of affinity for moisture, contributed to its selection
as a calibration standard. A weight of 0.65 milligrams of
benzoic acid was considered to be optimum because it allowed
good weighing precision and did not require a high sensi¬
tivity setting. For maximum accuracy, the sample weight
was selected so that the resultant carbon dioxide signal
corresponded approximately to that of the standard without
an attenuation change. This was Justified by the substantial¬
ly lower percentage accuracy requirements for hydrogen.
Peak amplitudes were adjusted, in each case, to lie between
50 and 100 per cent of full scale for greater readant
accuracy

73
Instrumental procedure
With oxygen flowing through the system, power was
switched on to the furnace, to the detector, to the capil¬
lary heater, and to the recorder. After a few minutes the
oxygen flow rate was adjusted to 40.0 cc/min. hy regulation
of the upstream pressure. Periodic checks on flow values
were made using the soap-film flow meter. The instrument
was allowed to equilibrate for at least two hours before
running a sample. It was considered advisable to turn off
only the furnace heater during breaks in extended periods
of operation.
With the input shorted (attenuator switch at the
"S" position), the recorder was adjusted so that the pen
rested exactly on the zero position of the chart. Recorder
amplifier sensitivity was adjusted for maximum gain and
was then reduced to the first point where no oscillation
occurred at any position on the scale. With the attenuator
in the position of maximum sensitivity (XI), the 50,000
ohm helipot was turned to zero the recorder. A benzoic
acid sample weighing close to 0.65 milligrams was introduced
into the sample port and then into the system by clockwise
rotation of the valve. After the appearance of the carbon
dioxide peak, the bridge sensitivity was set by the following
procedure. The zero control was first adjusted to provide
a signal corresponding exactly to the initial peak height.

74
The 500,000 ohm potentiometer was then adjusted to shift
the deflection to 75 per cent of full scale. The silica
gel was cleared of water by lowering the capillary tube
heater, and the converter tube was filled with fresh
calcium hydride.
With the attenuator at the XI position and the re¬
corder at the zero position, a second sample of benzoic
acid weighing approximately 0.65 milligram was introduced
into the combustion tube. As soon as the carbon dioxide
peak appeared the heater was lowered to the silica gel
portion of the capillary tube and was allowed to rest on the
silicone rubber insert of the converter tube for 5 seconds.
The heater was returned to its original position, the
recorder polarity switch was set at the "Hg" position, the
attenuator was adjusted to X4, and the recorder was re¬
zeroed, if necessary. Immediately after the hydrogen peak
emerged, the polarity switch was returned to the "COg"
position, the attenuator was reset to XI, and the system
was ready to accept a sample.
For those samples with known approximate compo¬
sitions, quantities were selected which corresponded roughly
to carbon contents of 0.45 milligrams (the approximate
weight of carbon in 0.65 mg of benzoic acid). In the case
of a sample of totally unknown content, it was sometimes
advantageous to make a quick preliminary determination so

75
that a more appropriate sample weight and attenuation
range could he selected. Weights for both the benzoic acid
standards and the samples were recorded to the nearest 0.1
microgram.
The weighed sample was introduced into the system,
and the complete procedure used for standardization was
repeated. In most instances the correct attenuation value
for hydrogen could be readily estimated. However, in
those cases where the amplitude of the hydrogen signal did
not follow prediction, it was possible to increase or to
decrease the attenuation during the emergence of the peak,
providing the recorder had been previously zeroed at the
highest sensitivity position.
Because peak heights for carbon and hydrogen were
linear with concentration over the prescribed range, per¬
centage compositions could be calculated using the
relationships
%C
%&
Py/^'samp.
./ "samp,
d./ W15 stand. X
Wt,
'samp.
An engineering ruler was used to read peak heights to within
the nearest 0.005 inch. For samples with carbon content

76
below 35 per cent, greater analytical accuracy was obtained
by comparison to a standard also of low carbon content.
Sucrose (N.B.S. Standard Sample #17) was often used for
this purpose.
Nitrogen analyses were performed using the modified
system and sample introduction technique discussed on pages
50-52. Helium (Ohio Chemical and Surgical Equipment
Company, Madison, Wisconsin), which was found to be more
satisfactory as a carrier gas than carbon dioxide, was
employed for the major portion of the determinations, and
the capillary tube-conversion system of the carbon and
hydrogen apparatus was replaced with a magnesium perchlorate-
ascarite filter. Reagent grade ammonium oxalate (A-679,
Fisher Scientific Company) was used as a standard for the
analysis of most compounds with nitrogen content ranging
between 10 and 30 per cent. A convenient weight for cali¬
bration purposes was 0.5 milligrams. At a helium flow rate
of 40.0 cc/min. this required an attenuation of X2 with no
change in the bridge sensitivity adjustment used for the
carbon and hydrogen analysis. It was again found advisable
to weigh the samples in quantities corresponding to the
standards in amount of nitrogen released. Under these
conditions the nitrogen composition was obtained using the
equation

SECTION IV
RESULTS AND DISCUSSION
Before any standardizations or analyses were con¬
ducted, tie system was thoroughly tested for leaks. This
was accomplished most conveniently hy closing off the exit
of the sample side of the detector cell and noting the
corresponding flow decrease on the rotameter. If the flow
did not decrease to zero after several minutes, portions
of the system were isolated until the point of leakage was
determined. Precision and accuracy were found to he
extremely dependent upon the maintenance of a totally leak-
tight system and leakage could generally he recognized hy
failure of a component peak to return to the hase line.
Other malfunctions during an analysis were readily indicated
hy an anomalous response pattern such as the occurrence of
double peaks, negative peaks, peak broadening, etc. It was
necessary, of course, to minimize the recorder deadband
before any determinations were made.
Typical readouts for carbon and hydrogen, and for
nitrogen are shown in Pigure 21. Linearity over a normal
concentration range is indicated by the curves of Figure
22. A benzoic acid standard was used for the carbon and
77

Fig. 21.-Readouts for Carbon Dioxide, Hydrogen, and Nitrogen
-o
oo

10
9
8
7
6
5
4
3
2
1
0
22
79
Weight of Standard, mg
-Response Curves for Carbon Dioxide, Hydrogen, and
Nitrogen

80
hydrogen values, and ammonium oxalate monohydrate was used
for the nitrogen values. Two attenuation settings were re¬
quired to provide maximum readout accuracy for hydrogen
over the entire weight range. Prior to each of the
determinations the flow rate was measured and was set, if
necessary, to exactly 40.0 cc/min. The responses and the
response ratios did not necessarily remain constant during
extended periods of operation, hut expressed variations
with general changes in conditions such as replacement of
combustion tube packing, partial deactivation of calcium
hydride by atmospheric moisture, etc. When the system had
been allowed to stabilize for a sufficient length of time
and when there were no observable fluctuations in operating
conditions, such as flow rate or line voltage, the required
frequency of standardization was reduced to a minimum. Por
extremely critical analyses, however, it was found advisable
to run a standard immediately after the sample and to use a
fresh layer of calcium hydride for each set of determinations.
To test the precision of this procedure a series of six
samples of benzoic were analyzed, each by comparison with
a separate benzoic acid standard. These were purposely not
run in sequence in order to determine the effect of response
variations. The results are tabulated in Table 5. Similar
data were obtained for the nitrogen analysis of alanine
using ammonium oxalate monohydrate as a standard, and these
are presented in Table 6.

TABLE 5
PRECISION FOR A TYPICAL CARBON AND HYDROGEN DETERMINATION
Determi¬
nation
No.
Benzoin
Benzoic Acid
Benzoin Composition, Found
Wt. ,
Rg
P.H. ,
X
co2
inches Wt.,
10 Rg
H2
P.H. ,
X
co2
inches
10
h2
% C
Deviation
from
mean, %
% H
Deviation
from
mean, %
1
59^.3
78.95
53.75* 612.3
70.70
95.75
79.21
+0.02
5.72
+0.03
2
627.9
83.05
54.10* 661.7
75.95
97.20
79.34
+0.15
5.81
+0.12
5
571.1
74.35
92.30 632.1
71.60
90.05
79.13
-0.06
5.62
-0.07
4
672.0
89.50
58.80* 623.0
72.20
97.10
79.12
-0.07
5.56
-O.I3
5
637.3
83.85
55.55* 619.2
70.75
91.15
79.28
+0.09
5.86
+0.17
6
598.9
76.80
94.45 654.4
73.10
91.30
79.04
-O.15
5.60
-0.09
*
Mean
79719
Mean
5.69
*
At an attenuation of X4. (All other hydrogen values were obtained at an attenuation
Theoretical Values,
% C
% H
Benzoic Acid
68.85
4.95
Benzoin
79.23
5.70
Mean Experimental Values, Benzoin
79.19
5.69
Average Deviation
0.09
0.10
Standard Deviation
0.11
0.12
% Standard Deviation
0.14
2.1
00

TABLE 6
PRECISION FOR A TYPICAL NITROGEN DETERMINATION
Determi¬
nation
Alanine
Ammonium oxalate
monohydrate
Alanine,
Deviation
No.
M6
P.H. ,
inches x 10
vt.,
MS
P.H.,
inches x 10
% N
found
from
mean, %
1
911.6
62.25
861.4
74.35
15.59
-0.13
2
874.6
58.85
791.2
67.90
15.45
-0.27
3
832.3
57.25
823.9
70.75
15.96
+0.24
4
900.8
62.50
841.0
73-55
15.64
-0.08
5
854.2
58.05
772.1
65.05
15.90
+0.18
6
891.7
61.40
805.0
69.30
Mean
15.76
15775“
+0.04
% N
Theoretical Values
Ammonium Oxalate Monohydrate
19.71
Alanine
15.72
Mean Experimental Value, Alanine
15.72
Average Deviation
0.16
Standard Deviation
0.18
% Standard Deviation
1.1

83
To determine the accuracy of the method, a range of
compounds exhibiting wide structural and compositional
variations were analyzed. The method of repeated standardi¬
zations was again used for all reported determinations,
even in those cases where excellent response consistency
was obtained. Analytical reagents were selected for
analysis wherever possible. Those which were known to be
deliquescent were first thoroughly dried at a reduced
pressure. No combustion catalyst was used for carbon and
hydrogen analyses except in the case of liquids or organo-
metallics containing alkali or alkaline earth metals.
The accuracy for carbon and hydrogen is indicated
by the data presented in Table 7. A single determination
was conducted for each compound and, with the exception of
potassium acid phthalate, no preliminary values were dis¬
carded. Of the 20 compounds reported, the average absolute
deviation from accepted percentages was + 0.15 per cent for
carbon and + 0.12 per cent for hydrogen. No deviations
exceeded + 0.46 per cent. The acceptable absolute deviation
from the theoretical percentages when using the traditional
gravimetric method is + 0.3 per cent (39).
Carbon analysis of phosphorus-containing compounds
by conventional techniques requires extremely strong
ignition to avoid retention of carbon dioxide by the oxides
of phosphorus. Even with the observance of this precaution,

TABLE 7
ANALYTICAL RESULTS FOR CARBON AND HYDROGEN
% C
Diff.
% H
Diff.
Compound
Structure
Origin
Theor.
Anal •
%
Theor,
Anal,
%
Acetanilide
CHjCONHCgH^
Merck
and Co.
(U.S.P.)
71.09
70.89
-0.20
6.71
6.52
-0.19
Benzo-
phenone
(c6h5)2co
Fisher
Scien¬
tific
Co.,
No.B-270
85.69
85.75
+0.06
5*53
5.59
+0.06
4-Bromo-
aceto-
phenone
BrC^H^COCH^
Eastman
Organic
Chemicals
No.1886
48.28
48.24
-0.04
3.54
3.84
+0.30
o-Bromo-
benzoic
Acid
BrCgH^COOH
Matheson
Coleman
and Bell
No.7383
41.82
41.59
-0.23
2.51
2.42
-0.09
Caffeine
CH5NCON(CH5)COC:CN:
: CHNCH-Eastman
'’Organic
Chemicals
No.355
49.48
4-9.36
-0.12
5.19
5.31
+0.12
oo

TABLE 7 - Continued
Compound.
Structure
Origin
% C
Theor. Anal.
Diff.
%
% H
Theor. Anal.
Diff.
%
Diallyldi-
phenyl
phosphonium
Bromide
(CH2:CHCH2)2PBr(C6H5)2 Synthe-
^ sized,
Univ. of
Florida
62.25
62.24
-0.01
5.80
5.78
-0.02
p-Dichloro-
benzene
C6H4C12
The Will
Corp.
49.03
48.81
-0.22
2.74
2.86
+0.12
2 ^"-Di¬
ehl or o-m-
ter-
phenyl*+
(cic6h4)2c6h4
Synthe¬
sized,
Univ. of
Maryland
72.26
72.13
-0.13
4.04
3.96
-0.08
9-Fluorenone
c6h4coc6h4
Eastman
Organic
Chemicals
No.3008
86.65
86.69
+0.03
4.47
4.56
-O.09
Hexamethyl-
phos-
phoramide*+
[(ch5)2n]5po
Fisher
Scien¬
tific Co.
No. H-
343
40.22
40.49
+0.27
10.12
10.27
+0.15
Naph¬
thalene
C10H8
N.B.S.
Standard
No. 38b
93.71
95.49
-0.22
6.29
6.24
-0.05
co
Vl

TABLE 7 - Continued
% C
Diff.
% H
Diff.
Compound
Structure
Origin
Theor.
Anal •
%
Theor.
Anal.
%
Potassium
Acid
Phthalate*
koocc6h4cooh
N.B.S.
Standard
No. 84e
47.05
46.90
-0.15
2.47
2.40
-0.07
Quinalizarin
(ho)2c6h2(co)2c6h2(oh)
2 Fisher
¿Scien¬
title
Co. No.
Q-4
61.77
61.61
-0.16
2.96
5.13
+0.17
Salicylic
Acid
HOCgH^COOH
Fisher
Scien¬
tific Co.
No. A-277
60.87
60.69
-0.18
4.38
4.25
-O.I3
Sucrose
C12H22°11
N.B.S.
Standard
No. 17
42.11
42.14
+0.03
6.47
6.38
-0.09
Sulfanilamide
H2NC6H4S02NH2
Nutri¬
tional
Biochemi¬
cals
Corp.
41.85
41.82
-0.03
4.68
4.60
-0.08

TABLE 7 - Continued
Compound
Structure
Origin
% C
Theor. Anal.
Diff.
%
% H
Theor. Anal.
Diff.
%
Tetra-n-butyl-
ammonium
Iodide
[ch3(ch2)5]4ni
Eastman
Organic
Chemicals
No. 4702
52.03
52.49
+0.46
9.76
9.50
-0.26
4,4,4-Tri-
fluoro-1-
(2-thienyl)-
1,3-hutane-
dione
SCH:CHCH:CC0CHoC0CF-,
• 1 2 3
Eastman
Organic
Chemicals
No. 7260
43.25
43.49
+0.14
2.25
2.49
+0.14
Triphenyl-
phosphine
«°6h5)3p
Metal and
Thermite
Corp.
82.43
82.41
-0.02
5.76
5.83
+0.08
Vanillin
CH,0(0H)CfiH,CH0
Eastman
63.15
62.94
-0.21
5.30
5.46
+0.16
Organic
Chemicals
No. 2?3
Silver permanganate added
+Liquid sample
00
-o

88
the carbon contents reported for phosphorus compounds by
commercial analytical laboratories have generally been 1-3
per cent lower than the theoretical values. Excellent
results, however, were obtained in this study for similar
or identical compounds with no additional catalytic or
temperature requirements. This was attributed to the high
combustion efficiency of the system and to its acceptance
of small samples.
Of all the compounds analyzed, oily one gave con¬
sistently low results. Potassium acid phthalate (47.05% C,
2.47% H) repeatedly showed a hydrogen content of 2.35 +
0.05 per cent and a carbon content of 44.0 + 0.1 per cent.
This corresponded very closely to the values calculated
for the reaction
OC00K
+ 15 05—E-CO- + 15 C0~ + 5 H.0 .
COOH 2 2 3 2 2
Results were not noticeably improved by the addition of
cobaltic oxide, but encapsulation of the sample with
approximately 5 milligrams of silver permanganate gave
values within 0.15 per cent of the theoretical percentages.
The silver permanganate prepared in this laboratory showed
no background response for carbon, hydrogen, or nitrogen
and was recommended as a general catalyst for all compounds
which were suspected to exhibit anomalous combustion

89
characteristics» Substances with high halogen or sulfur
content were analyzed with no difficulty even though the
silver gauze used initially was of relatively low surface
airea. It was noted in earlier work that the chlorine re¬
leased during an analysis of a chioro-compound partially
oxidized the silver bromide formed during previous analyses
of bromo-compounds, and expelled a small amount of bromine
into the silica gel. Although this condition showed no
adverse effect on the accuracy of the determinations, it
was corrected by the use of silver wool in place of the
silver gauze.
The results of a series of 10 nitrogen analyses axe
presented in Table 8. The average deviation was within
+0.22 per cent of the theoretical percentage and no
individual deviation exceeded + 0.43 per cent. The allowable
error in the conventional micro-Dumas method for nitrogen
is + 0.2 per cent (+ 0.3% for high percentages of nitrogen)
(35). A small amount of silver permanganate was added to
the sample for almost all reported determinations. In
addition to its catalytic effect, it served to reoxidize
the copper formed by the reduction of copper oxide during
the combustion process. At line voltage furnace operation,
some thermal decomposition of the copper oxide was evidenced
by excessive oxidation of the copper layer. It was, there¬
fore, recommended that the furnace be operated at a somewhat
lower temperature (< 1000oC) during nitrogen analyses.

TABLE 8
ANALYTICAL RESULTS FOR NITROGEN
% N
Compound
Structure
Origin
Theor.
Anal,
Diff., %
Acetanilide
ch5conhc6h5
Merck and Co.
(U.S.P.)
10.36
10.51
+0.15
Ammonium
Nitrate
nh4no3
Fisher Scientific
Co., No. A-676
35.00
35.28
+0.28
Asparagin
NH2C0CH2CH(NH2)C00H-H20
Fisher Scientific
Co., No. A-912
18.66
18.49
-0.17
p-Bromo-
aniline*
BrC6H4NH2
Eastman Organic
Chemicals
No. 473
8.14
8.30
+0.16
Caffeine*
CHzNCON(CH., )COC:CN:CHNCH*
t>\ 0 i 0
Eastman Organic
No. 355
28.85
29.12
+0.27
Creatine
NH2C(:NH)N(CH5)CH2COOH
Nutritional
Biochemicals
Corpo
32.05
31.69
-0.36
Cystine*
[SCH2CH(NH2)C00H]2
Merck and Co.
No. 72612
11.66
11.50
-0.16
vD
O

TABLE 8 - Continued
Compound
Structure
Origin
% N
Theor. Anal.
Diff., %
Hexamethyl-
phosphor-
amide*+
[(ch5)2n]5po
Fisher Scientific
Co., No. H-343
23.45
23.02
-O.43
Quinaldic
Acid*
C^-H.-N: CCOOHCH: CH
LO |
Eastman Organic
Chemicals
No. 2055
8.09
7.92
-O.17
Sulf anilamide *
H2NC6H4S02NH2
Nutritional
Biochemicals
Corp.
16.27
16.31
+0.04
Silver permanganate added
+Liquid sample

92
The initial results for nitrogen determinations
using a carbon dioxide carrier gas were satisfactory for
high nitrogen percentages but the signal-to-noise ratio
was too low to provide sufficiently accurate values for
compounds of low nitrogen content. All reported results
were, therefore, obtained using a helium carrier gas.
Experimental studies have indicated that the
accuracy of this method of analysis is comparable to that
of the conventional micro-Dumas technique for nitrogen,
and exceeds that of the conventional Pregl technique for
carbon and hydrogen. Its main advantages lie in its speed,
simplicity, reduction in operational requirements, and its
acceptance of small samples. With two functional systems
and a single recorder, a complete carbon, hydrogen, and
nitrogen analysis may be performed in less than ten minutes
even though two weighings are required. New commercially
available instruments which allow CHN determinations to be
performed on a single sample require considerably longer
analysis time and suffer from inaccuracies inherent in
combining the procedures.
It is anticipated that, with additional improvements,
the system described will require less frequent standardi¬
zation and will offer even greater stability, precision and
accuracy

SECTION V
EXTENSIONS TO ADDITIONAL ANALYSES
A simple modification of the apparatus originally
designed for the determination of carbon and hydrogen
provided a satisfactory analysis for nitrogen. Adaptation
of the general approach to the analysis of oxygen, sulfur,
and halogens was, therefore, investigated.
Oxygen
Recognition of the difficulties involved in the
direct analysis of oxygen in organic compounds has generally
led to the calculation of oxygen content by difference.
Good results, however, have been reported for certain
materials through the use of a Pregl method coupled with
a volumetric or gravimetric determination of the combustion
products. The analysis is based upon the ignition of the
compound in the presence of carbon at 1120°C using an
oxygen-free carrier gas such as nitrogen or helium. At
this temperature the following pertinent reactions occur
Organic 0 + C -*-C0
C02 + C — 2 CO
HgO + C CO + H2 .
93

94
The quantity of carton monoxide produced is stoichiometrically
related to the amount of oxygen present in the sample. Oxi¬
dation of carbon monoxide to carbon dioxide by passage
through a solution containing iodine pentoxide and by subse¬
quent titration of the iodine formed allows a determination
of the total amount of carbon monoxide released during the
combustion process. An alternate method requires oxidation
of the carbon monoxide by passage over hot copper oxide,
and a gravimetric determination of the resultant carbon
dioxide collected in an ascarite absorption tube.
A combustion tube was packed with a 6 inch layer of
20 mesh spectrographically pure carbon chips followed by
a 1 inch layer of silver wool. The Nichrome heating coil
was compressed to provide a maximum temperature in the area
of the upper portion of the carbon layer. A second
combustion tube, packed with copper oxide, was positioned
below the first tube, and a magnesium perchlorate filter
was added to retain the water formed during the oxidation
of hydrogen. Helium, de-oxygenated by passage through a
copper-filled furnace tube at 450°C (S-56518, E. H. Sargent
and Co.), was used as a carrier gas. A method of sample
introduction identical to that used for the nitrogen analysis
was chosen to reduce atmospheric contamination to a minimum.
Erratic results were obtained for an initial series of
compounds, and it was found that non-oxygen-containing

95
materials produced carbon dioxide peaks as well. This was
due, in part, to tbeir passage in the vapor phase through
the oxidizing portion of the combustion tube. A substantial
improvement was made by elimination of the copper oxide and
by direct determination of carbon monoxide. Besults for a
series of 5 compounds, using benzoic acid as a standard,
are shown in Table 9. Discrepancies in these values could
be attributed to positive or negative contributions to the
thermal conductivity of the composite consisting of carbon
monoxide and small amounts of unconverted hydrogen or low
molecular weight pyrolysis products such as methane or
ethane. It was evident that further work would necessarily
require a column separation of carbon monoxide from these
and other contaminants.
TABLE 9
EESULTS FOB OXYGEN ANALYSES
Compound
Benzoin
Benzophenone
/¡-Naphthol
Sucrose
%Q
Theor. Anal.'
15.08 14.3
8.78 9.0
11.10 11.1
51.42 50.2
31.55 31.6
Vanillin

96
Sulfur and halogens
Numerous experimental difficulties were associated
with the determination of sulfur, chlorine, bromine, and
iodine by the use of a chromatographic detection system.
Nevertheless, when an oxygen carrier gas was swept through
a combustion tube packed with quartz chips, rapid responses
were obtained for chloro-, bromo-, and sulfur-containing-
organics, and the resultant signals were identified with
chlorine, bromine, and sulfur dioxide, respectively.
Column separations were required, however, for each determi¬
nation and, in the case of sulfur compounds, the necessity
for conversion of the mixture of sulfur dioxide and sulfur
trioxide to a single component became evident. Hapid cor¬
rosion of thermistor leads, tubing and connectors led to
the discontinuance of these efforts pending more satisfactory
detection conditions.
Carbon in slow burning compounds
Certain analyses, such as the carbon content in steel,
require large samples and strong ignition over an extended
time interval. A simple adaptation of the original system
was shown to accommodate such determinations. The silica
gel packing of the capillary tube was replaced by a 2 inch
layer of 60-80 mesh 5A molecular sieve, and a larger moving
heater was constructed to jacket the packing. Time require¬
ments for complete combustion were indicated by running an

97
initial sample and by noting the time, after successive
lowerings of the heater, at which no additional carbon
dioxide was expelled from the capillary tube- Complete
retention of carbon dioxide was observed, and its release
into the system was rapid and reproducible. A standard
was selected which was similar in structure and in carbon
content to the sample.
Moisture and water of composition
Due to the high sensitivity obtained by conversion
of water to hydrogen, the system proved to be extremely
efficient for the determination of trace amounts of moisture
and for the degree of hydration of inorganic compounds. It
was necessary only to release any extraneous moisture which
had collected in the silica gel immediately prior to the
introduction of each sample.

SECTION VI
SUMMARY
The microanalytical techniques for carbon, hydrogen,
and nitrogen in use today are essentially those which were
investigated forty or fifty years ago with few modifications
or improvements. The development of ultramicro procedures
has been hindered by the lack of sufficiently sensitive and
accurate weighing equipment capable of weighing both the
sample and absorption tubes to the same degree of precision.
The argument for the use of larger samples has generally not
been based upon this consideration but rather upon the "non-
homogeneous nature of the compounds to be analyzed and the
subsequent higher weight requirements for representative
sampling." It is evident, however, that micromethods should
not be applied to samples which are known to be hetero¬
geneous unless acceptable mechanical means are available to
provide extremely uniform mixtures. With the advent of thin
layer or gas chromatography and microsublimation or micro-
distillation techniques there has been an increasing demand
for methods requiring sub-milligram quantities of sample.
New balances now allow rapid and precise sample weighing
98

99
under low-load conditions but are as yet not capable of
determining small weight differences in heavy absorption
tubes to sin acceptable degree of accuracy.
The detection method which has been described
complements the new weighing innovations and eliminates
the critical volumetric measurements and programmed weighing
which are considered to represent the major source of error
in traditional microanalyses. The system circumvents the
meticulous and exacting techniques previously required of
a skilled operator and reduces the analysis time to less
than ninety seconds after introduction of the sample. Care¬
ful selection of design and operational parameters allows
compositional data to be obtained for an extremely wide
range of organic compounds with accuracy which exceeds
that of conventional methods.
In principle, a sample (solid, liquid, or gas) is
introduced into a tube packed with copper oxide at a fixed
temperature of approximately 1000°C through which pure
oxygen flows at a constant rate. Combustion occurs rapidly
and uniformly under the conditions of the analysis. All
combustion products are retained with the exception of
carbon dioxide, the concentration of which is accurately
measured as it passes through a thermistor detector. Water,
initially absorbed by a desiccant, is expelled by a movahle
heater into a tube containing a layer of calcium hydride.

100
Hydrogen is released quantitatively and is measured by tbe
same detector. Peak heights are linear functions of
concentrations, and percentage compositions are calculated
by comparison to a single standard (usually benzoic acid).
For nitrogenous compounds the procedure is repeated with
the exception that the nitrogen content of the sample is
completely converted to elemental nitrogen which is
detected in a helium carrier gas.
It is hoped that the method, by virtue of its speed,
accuracy, and ease of operation, will prove to be a valuable
asset to the synthetic organic, pharmaceutical, or medical
laboratory as well as to structural research and industrial
process control.

APPENDIX

102
The basic principles and operational and structural
details of the described instrumental procedure have been
applied in the production of a commercial carbon and
hydrogen analyzer (1) by the American Instrument Company
of Silver Spring, Maryland, the prototype model of which
is shown in Figure 23.
The instrument is so constructed as to allow rapid
replacement of component parts, and provides precision flow
control and temperature maintenance. The pre-conditioning
column is positioned prior to the pressure regulator, and
the reference detector cell is situated between the
throttling orifice and the sample introduction port to
insure maximum stability. Power requirements are sub¬
stantially reduced by enclosing the combustion tube in a
ceramic jacket which, in turn, is further insulated to
prevent radiational losses. The 250°C heater is moved to
either extremity of the capillary tube by a rotary dial
mechanism. The complete unit is packaged in a cabinet 24
inches high, 12 inches wide, and 10 inches long. Output
signals are measured on a 5 millivolt recorder (Model TR,
E. H. Sargent and Co.), and an electrobalance (Model G,
Cahn Instrument Co.) is recommended for sample weighing.
The production of a similar instrument for the determination
of nitrogen is planned for the near future.

Fig. 23.-Prototype Model of American Instrument Co. Carbon
and Hydrogen Analyzer

LITERATURE CITED
1. American Instrument Co., Inc., Preliminary Data Sheet
2399.
2. Anon., "Fluid Meters, Their Theory and Application,"
5th ed., p. 81f, ASME, New Tork.
3. Belcher, R., Ingram, G., Anal. Chim. Acta 4, 118 (1950)
4. Belcher, R., Spooner, C. E., J. Chem. Soc. 194-3. 313.
5. Coleman Instruments, Inc., Bulletin B-273.
6. Ibid.. Bulletin B-258B.
7. Ibid.. Paper No. 114 Presented at the Pittsburgh
Confernece on Analytical Chemistry and Applied
Spectroscopy, March, 1964.
8. Clerc, J. T., Dohner, R., Sauter, W.. Simon. W.. Helv.
Chim. Acta 46, 2369 (1963).
9. Duswalt, A. A., Brandt. W. W.. Anal. Chem. 32, 272
(1960).
10. F & M Scientific Corp., Bulletin 1800.
11. Finston, H. L., Tellin, E., Anal. Chem. 35, 336 (1963).
12. Fisher Scientific Co., Bulletin FS-257.
15.Frazer, J. W., Mikrochim. Acta 1962. 993.
14. Haber, H. S., Gardiner. K. W.. Microchem. J. 6. 83
(1962). " “
15. Horacek, J., Kbrbl, J.. Pechanec, V.. Mikrochim. Acta
I960, 294.
16. Hozumi, K., Kirsten, W. J., Anal. Chem. 35. 1522 (1963)
17. Ingram, G., Analyst 73. 548 (1948).
18. Kainz, G., Horwatitsch, H., Z. anal. Chem. 176. 175
(I960). “
104

105
19. Ibid., 177, 521 (I960).
20. Ibid., p. 344.
21. Ibid., 178, 406 (1961).
22. Ibid.. 184, 363 (1961).
23. Keidel, I*. A., Anal. Chem. ¿1, 2043 (1959).
24. Keulemans, A. I. M., "Gas Chromatography," 2nd ed.,
p. 90, Eeinhold, New York, 1959.
25. Kürbl, J., Collection Czech. Chem. Commun. 20. 953
(1955). —
26. Ibid., p. 1026.
27. Kuck, J. A., Berry, J. W., Andreatch, A. J., Lentz,
P. A., Anal. Chem. ¿4, 403 (1962).
28. Malissa, H., Pell, E., Michrochem. J., Symp. Ser. 2,
371 (1962), through Chem. Abstr. 58, 6191e (1963).“
29. Maresh, C., Sundberg, 0. E., Hofstader, R. A.,
Gerhardt, G. E., Michrochem. J., Symp. Ser. 2, 387
(1962), through C. A. 58, 6l9lf (1963).
30. Marks, R., Rev. Sci. Instr. 28, 381 (1957).
31. Nightingale, C. P., Walker, J. M., Anal. Chem. 34.
1435 (1962). —
32. Onoe, T., Ann. Rep. Takamine Lab. 10, 128 (1958),
through C. A. 56, 8003f (1962).
33» Simon, W., Sommer, P. Z., Lysey, G. H., Michrochem. J.
1962. 239. “
34. Smith, B. D., Bowden, W. W., Anal. Chem. ¿6, 82 (1964).
35. Steyermark, A., "Quantitative Organic Microanalysis,"
2nd ed., p. 171, Academic Press, New York, 1961.
36. Ibid., p. 222.
37. Ibid.. pp. 233, 238.
38. Ibid., p. 234.
39. Ibid.. p. 255.

106
40. Sundberg, 0. E., Maresh, C., Anal. Chem. 32, 274 (I960).
41. Technicon Instruments Corp., Bulletin No. CHN-1.
42. Trusell, F., Diehl, H., Anal. Chem. 35, 674 (1963).
43. Vecera. n.. Snobl. D.. Collection Czech. Chem. Commun.
2£, 2013 (I960).
44. Vecera, M., Snobl, D., Synek, L., Mikrochim. Acta
1961, 370, through C. A. ¿6, 9420d (1962J.
45. Vogel, A. M., Quattrone J. J., Anal. Chem. 32. 1754
(I960).
46. Ualisch, V., Chem. Ber. 2!, 2314 (1961).

BIOGEAPHICAL SKETCH
C. David Miller was born April 7, 1931» at Baltimore,
Maryland. He was graduated from Leonia High. School, Leonia,
New Jersey in 194-8, and in 1952 he received the degree of
Bachelor of Arts from Columbia University.
He was employed as an analytical chemist by the
E. I. Du Pont de Nemours Company of Gibbstown, New Jersey
during 1952 and 1953» Following two years of service in
the Chemical Corps of the United States Army he returned
to Columbia University to conduct research in physical-
analytical chemistry. In 1957 he entered the Graduate
School of the University of Maryland and he received the
degree of Master of Science in organic chemistry in 1959.
He then accepted a position with the American Instrument
Company of Silver Spring, Maryland where he was engaged in
analytical instrumentation development until his enrollment
in the graduate school of the University of Florida in
September, 1961. During the pursuit of his work toward the
degree of Doctor of Philosophy he was the recipient of a
research grant sponsored by the American Instrument Company.
C. David Miller is married to the former Hita Sue
Showalter. He is a member of the American Chemical Society
and the Electron Microscopy Society of America.
107

This dissertation was prepared under the direction
of the chairman of the candidate's supervisory committee
and has been approved by all members of that committee. It
was submitted to the Dean of the College of Arts and
Sciences and to the Graduate Council, and was approved as
partial fulfillment of the requirements for the degree of
Doctor of Philosophy.
April 18, 1964
Dean, College o
s/And Sciences
Supervisory Committee:
Dean, Graduate School

3 1262 08554 8716



Fig. 10.-Dependence of Response on Flow Rate
Peak Height Response, mv/mg Benzoic Acid
H*
I-*
H
H*
H*





H*
ro
ro
VJ
VI
o
VI
o
VJ1
o
-f*


10
9
8
7
6
5
4
3
2
1
0
22
79
Weight of Standard, mg
-Response Curves for Carbon Dioxide, Hydrogen, and
Nitrogen


by immersion in syrupy phosphoric acid. Subsequent de
hydration by applying a voltage to the electrodes until no
further hydrogen evolution was observed left a thin film of
phosphorus pentoxide. The element was encased in a glass
sleeve and was positioned directly below the combustion
tube. Carbon dioxide generated during a combustion process
was unretained and was detected in the normal manner. The
water, however, was irreversibly absorbed by the phosphorus
pentoxide. Application of a relatively high voltage to the
electrodes resulted in the rapid electrolytic release of
oxygen and hydrogen into the oxygen carrier stream. The
hydrogen was detected conventionally by thermal conductivity.
The conversion, however, was found to be incomplete and
losses were incurred through inefficient collection of water
on the phosphorus pentoxide.
A number of additional chemical conversions were
investigated, including the reaction of water with aluminum
carbide to produce methane. This proved ineffective due
to the extremely slow reaction rate. Rejection of other
systems was based also on reactivity or detection deficien
cies.
A simple and accurate method for measurement of water
was eventually devised which eliminated virtually all of the
difficulties previously encountered. A standard unsealed
Kimax melting point capillary tube was fitted at one end


60
SIDE VIEW
Material: 304 Stainless
Steel
Seale: Full
Fig. 15.-Details of Cell Construction


53
#
o
Nitrogen Peak Resultant from Introduction
of Air During Normal Operation of Valve
p
cd
-p
o

>
rH
cd
'd
-p
cd
-p
o


>
rH
cd
>
Elimination of Air Peak By
Modified. Operation of Valve
I
3"
Attenuation, XI (2 times that normally used during a
nitrogen determination)
Fig. 11.-Reduction of Atmospheric Contamination During
Nitrogen Analyses


52
tlie upper compartment which was subsequently filled with a
loosely fitting aluminum plug The valve was then quickly
rotated to the extreme counterclockwise position and hack
to the extreme clockwise position. Because the helium in
the inner compartment was under slight pressure, diffusion
of almost all outside air was prevented. The magnitude of
the resultant air peak was generally at least twenty times
less than that encountered during normal operation, and the
peak was often completely eliminated as in the case indi
cated in Figure 11. Due to the somewhat slow diffusion
process the peak was considerably broad, and followed that
corresponding to sample nitrogen with only slight overlap.
Therefore, the contribution of atmospheric nitrogen was
negligible providing the valve was operated in the manner
described.
High "pseudo-nitrogen peaks" were initially obtained
for several non-nitrogenous compounds such as sucrose,
dextrose, and citric acid. It was later discovered that
injections of methane produced the same effect which was
attributed to incomplete combustion and subsequent detection
of certain low molecular weight pyrolysis products. This
difficulty was completely eliminated by the addition of a
layer of platinized asbestos to the copper oxide portion
of the combustion tube packing.


63
/ \
/ \
SIDE VIEW
Material: 304 Stainless
Steel
Scale: Pull
Pig. 17.-Details of Sample Introduction Port


51
for carbon dioxide formation. For those compounds containing
high percentages of nitrogen, a carbon dioxide carrier gas
afforded sufficient sensitivity for the analysis, but it
was found advantageous to use helium for general application.
Good combustion characteristics, with respect to nitrogen
release were observed for most compounds even though an
inert carrier gas was used. It is reasonable to assume
from thermodynamic considerations that, during the com
bustion process, free nitrogen would be formed even more
readily than carbon dioxide or water due to its higher
stability.
The small amount of air which is introduced with the
sample is of no consequence in the system for carbon and
hydrogen because oxygen and air have virtually identical
thermal conductivities. In the case of the nitrogen
analysis, however, where helium or carbon dioxide is used
as a carrier gas, a significant error may be introduced.
The slight Venturi effect within the valve draws the en
trapped air into the system and the subsequent peak is
partially superimposed upon the nitrogen peak which precedes
it. A modification of the procedure for sample introduction
reduced this contribution to a minimum. The valve (see
Figure 1?) was first rotated to the extreme clockwise
position to clear the inner sample compartment of air.
After a few seconds the encapsulated sample was dropped into


TABLE 7 Continued
Compound.
Structure
Origin
% C
Theor. Anal.
Diff.
%
% H
Theor. Anal.
Diff.
%
Diallyldi-
phenyl
phosphonium
Bromide
(CH2:CHCH2)2PBr(C6H5)2 Synthe-
^ sized,
Univ. of
Florida
62.25
62.24
-0.01
5.80
5.78
-0.02
p-Dichloro-
benzene
C6H4C12
The Will
Corp.
49.03
48.81
-0.22
2.74
2.86
+0.12
2 ^"-Di
ehl oro-m-
ter-
phenyl*+
(cic6h4)2c6h4
Synthe
sized,
Univ. of
Maryland
72.26
72.13
-0.13
4.04
3.96
-0.08
9-Fluorenone
c6h4coc6h4
Eastman
Organic
Chemicals
No.3008
86.65
86.69
+0.03
4.47
4.56
-O.09
Hexamethyl-
phos-
phoramide*+
[(ch5)2n]5po
Fisher
Scien
tific Co.
No. H-
343
40.22
40.49
+0.27
10.12
10.27
+0.15
Naph
thalene
C10H8
N.B.S.
Standard
No. 38b
93.71
95.49
-0.22
6.29
6.24
-0.05
oo


SECTION VI
SUMMARY
The microanalytical techniques for carbon, hydrogen,
and nitrogen in use today are essentially those which were
investigated forty or fifty years ago with few modifications
or improvements. The development of ultramicro procedures
has been hindered hy the lack of sufficiently sensitive and
accurate weighing equipment capable of weighing both the
sample and absorption tubes to the same degree of precision.
The argument for the use of larger samples has generally not
been based upon this consideration but rather upon the "non-
homogeneous nature of the compounds to be analyzed and the
subsequent higher weight requirements for representative
sampling." It is evident, however, that micromethods should
not be applied to samples which are known to be hetero
geneous unless acceptable mechanical means are available to
provide extremely uniform mixtures. With the advent of thin
layer or gas chromatography and microsublimation or micro-
distillation techniques there has been an increasing demand
for methods requiring sub-milligram quantities of sample.
New balances now allow rapid and precise sample weighing
98


11
In spite of inherent procedural difficulties, the
system indicated the potential of a direct combustion-
detection approach to carbon and hydrogen analysis, and
subsequent techniques were based on improvement of this
method.
Selection of Components and Operational Procedures
Even though efforts were directed, initially, toward
the development of an instrumental technique for which an
eventual commercial apparatus would be readily adaptable,
alternate methods were studied to establish a system of
optimum efficiency and accuracy Basic theoretical and
experimental considerations governing the choice of
components, conditions, and their interrelationships were
extended, where possible, to enable additional elemental
analyses with appropriate modifications.
Combustion and sample introduction systems
The preliminary experimental method for oxidative
degradation proved unsatisfactory for a number of reasons.
Sublimation or pre-vaporization of samples with high vapor
pressures was unavoidable. Efficient weighing or sample
handling techniques could not be established. A lack of
combustion spontaneity resulted in peak asymmetry, pre
venting the use of peak height measurement. A definite but
insufficient improvement was made by fabricating a small,


TABLE 7
ANALYTICAL RESULTS FOR CARBON AND HYDROGEN
% C
Diff.
% H
Diff.
Compound
Structure
Origin
Theor.
Anal
%
Theor,
Anal,
%
Acetanilide
CHjCONHCgH^
Merck
and Co.
(U.S.P.)
71.09
70.89
-0.20
6.71
6.52
-0.19
Benzo-
phenone
(c6h5)2co
Fisher
Scien
tific
Co.,
No.B-270
85.69
85.75
+0.06
5*53
5.59
+0.06
4-Bromo-
aceto-
phenone
BrC^H^COCH^
Eastman
Organic
Chemicals
No.1886
48.28
48.24
-0.04
3.54
3.84
+0.30
o-Bromo-
benzoic
Acid
BrCgH^COOH
Matheson
Coleman
and Bell
No.7383
41.82
41.59
-0.23
2.51
2.42
-0.09
Caffeine
CH5NCON(CH5)COC:CN:
: CHNCH-Eastman
'Organic
Chemicals
No.355
49.48
4-9.36
-0.12
5.19
5.31
+0.12
oo


TABLE 8 Continued
Compound
Structure
Origin
% N
Theor. Anal.
Diff., %
Hexamethyl-
phosphor-
amide*+
[(ch5)2n]5po
Fisher Scientific
Co., No. H-343
23.45
23.02
-O.43
Quinaldic
Acid*
C^-H.-N: CCOOHCH: CH
LP ^ 1
Eastman Organic
Chemicals
No. 2055
8.09
7.92
-O.17
Sulf anilamide *
H2NC6H4S02NH2
Nutritional
Biochemicals
Corp.
16.27
16.31
+0.04
Silver permanganate added
+Liquid sample


Sample Port
Combustion Tube
Moving Heater
Capillary Tube
Silica Gel
Calcium Hydride
Filter
Delay Column
Detector
Reversal
Fig. 7.-Finalized Flow System and Readout for Carbon and
Hydrogen Analysis


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID ELNDBIG8G_WFWZ8Y INGEST_TIME 2013-01-23T16:13:18Z PACKAGE AA00012906_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES


7
traditional techniques, generally resulting from the use of
inert carrier gases to increase sensitivity, impose limi
tations on combustion efficiency and restrict the range of
compounds capable of analysis. The singular and distinct
advantage of these systems arising from the reduction in
required methodological skills is often accompanied by
prohibitively high cost of equipment. The desirability of
a system combining maximum operational efficiency and
economy with increased speed and ease of analysis was
recognized, and it was toward this objective that these
research efforts were directed.


20
A two transistor zener-stabilized power supply with
a variable d.c. output voltage of 14-,8-26.3 volts was con
structed. The excellent regulation obtained in spite of
large fluctuations in line voltage is shown in Figure 19.
A Wheatstone bridge circuit, designed to accept a wide
resistance range of thermistors and to provide for suitable
sensitivity and attenuation variation, was constructed and
used in all studies.
Conversion systems
A three-fold advantage was realized in the replace
ment of a direct thermal detection of water by the measure
ment of an appropriate gaseous conversion product. When
retention techniques are required, anomalous desorption of
water from active column, sites causes non-reproducible
tailing, and the eventual change in the substrate-solid
support ratio results in variations in retention and response.
The sensitivity of a katharometer to water in an oxygen
carrier gas is too low to provide a signal-to-noise ratio
commensurate with the precision and accuracy requirements
of the analysis. Thermostatic control of the column-
detector system at a temperature in excess of 100C is
avoided because ambient temperatures may be used in con
version product measurement.
The initial experimental system, previously discussed,
utilized a conversion of water to acetylene by reaction with


2
The current determination is based upon complete
conversion in a stream of oxygen of the carbon and hydrogen
in the sample to carbon dioxide and water. This is
effected by passage of the pyrolysis products through a
combustion tube packed with copper oxide and lead chromate
maintained at a temperature of 680C. Carbon dioxide is
absorbed by sodium hydroxide on asbestos, and water is
retained by magnesium perchlorate. The corresponding
weight increases in the absorption tubes are related to the
percentage composition of the sample.
Pregl has also been credited with adaptation of the
classical macro-Dumas nitrogen determination to the micro
technique which has been accepted as reliable for most
forms of nitrogen. The procedure involves combustion of
the sample in an atmosphere of carbon dioxide, in which
carbon and hydrogen are converted to carbon dioxide and
water over a copper oxide catalyst, and oxides of nitrogen
are reduced to elemental nitrogen over copper. The nitrogen
is collected and measured in a nitrometer containing a
confining solution of 50 per cent aqueous potassium
hydroxide which absorbs the carbon dioxide.
Instrumentation automating the basic techniques for
carbon, hydrogen, and nitrogen, as well as the direct
determination of oxygen has recently been made available by
Coleman Instruments, Inc. (5,6,7). The "empty-tube" method


16
polyvinyl alcohol binder. The element was positioned in a
tube and the electrodes were connected to an a.c. bridge
circuit, the unbalance of which was representative of the
quantity of water passing through the detector. Improved
results were obtained with a more favorable cell geometry
approaching point source detection. Six parallel strands
of thin wire were evenly spaced along the inside wall of a
short length of 1/16 inch I. D. heavy-walled pyrex tubing.
A light coating of polyvinyl alcohol was applied and
alternate strands of wire were joined and connected to the
simple d.c. circuit shown in Figure 3 The response was
extremely rapid and was specific for water. Carbon
dioxide was determined by a conventional thermal conduc
tivity detector. Linearity and reproducibility, however,
did not meet the requirements of the determination. A
variety of other detection systems were considered but from
a standpoint of simplicity and reliability, thermal
conductivity methods proved most satisfactory.
In choosing an appropriate katharometer, many factors
were taken into consideration. Although the filament
detector generally exhibits greater stability, conventional
glass-clad thermistors were selected in preference to hot
wires because of their decreased susceptibility to oxidation,
and their higher ambient temperature response in an oxygen
atmosphere. The cell was designed in such a manner that


13
70 volts were applied. Careful introduction of small
varying amounts of ethanol by means of a 10 ul precision
syringe produced sharp, symmetrical carbon dioxide peaks
displaying good linearity. Provision for introduction of
solids without interruption of the carrier gas flow was
made by devising a sliding valve which positioned the
sample between two parallel plates and subsequently trans
ported it into the hot zone by means of a gravity feed as
illustrated in Figure 2. Details of an improved valve,
based on a rotary concept, which proved to be extremely
satisfactory during the remainder of the investigation,
are presented in a later section.
Further improvements in the combustion system pro
vided ignition temperatures of approximately 1050C and
allowed furnace operation on line voltage. The tube was
packed to the half-way mark with copper oxide but was heated
at least 6 inches above this level to facilitate decompo
sition of flashback pyrolysis products. The spacing of the
Nichrome coils was adjusted for provision of highest
temperature at the copper oxide surface, the site at which
sample decomposition occurred.
Detection systems
Determination of combustion product concentrations
by relative thermal conductivity measurements provided a
satisfactory method for initial experimental work. The


5
Vogel and Quattrone (45) devised a somewhat more complicated
procedure involving a bomb combustion of the sample in an
oxygen atmosphere and a direct chromatographic analysis of
carbon dioxide and water. A system of even greater
complexity was employed by Nightingale and Walker (51) for
the simultaneous determination of carbon, hydrogen, and
nitrogen. An induction furnace was used to burn the sample
in a helium atmosphere and the resultant nitrogen, carbon
dioxide, and acetylene (the conversion product of water)
were separated by temperature-programmed gas chromatography.
In addition to the large sample requirements the methods
described were inaccurate and time-consuming, and were
limited by basic disadvantages which will be discussed in
a later section.
In an improvement of his original work, Maresh (29)
incorporated a simultaneous nitrogen determination, and the
method formed the basis of a commercial analyzer produced
by the Fisher Scientific Company (12). The instrument
again relied upon a separation of nitrogen, carbon dioxide,
and acetylene which were initially collected at liquid
nitrogen temperatures. Inherent problems included high
operational costs and inefficiencies associated with the
inert combustion atmosphere.
Simon and co-workers (8,35) determined carbon,
hydrogen, and nitrogen by expanding the mixture of combustion


Addition of catalysts
In only very few instances was it found necessary
or advantageous to add a combustion catalyst to the sample
being analyzed for carbon and hydrogen. Under less favorable
conditions, where larger samples and lower temperatures are
employed, those compounds containing the alkali or alkaline
earth metals in the absence of sulfur or phosphorus would
leave an ash of the carbonate resulting in low carbon
values. Steyermark (36) suggests the addition of potassium
dichromate or vanadium pentoxide in the analysis of these
compounds. Efficiencies of a large number of catalysts
have been evaluated (15,18,19,20,21,44) but it is generally
agreed that tricobalt tetroxide, cobaltic oxide plus silver,
and silver permanganate are among the most satisfactory.
Silver permanganate was prepared by the reaction of silver
nitrate and potassium permanganate, and was used as a
general catalyst where required.


A noii-dispersive infrared gas analyzer lias been used
by Kuck and co-workers (27) to determine the carbon dioxide
released during a conventional combustion process after
adjusting the gas mixtures to a constant volume and a
standard reference pressure. Haber and Gardiner (14)
employed a commercial version of the electrolytic moisture
cell of Keidel (23) for the analysis of carbon and hydrogen.
The method involved quantitative conversion of carbon
dioxide to water by passing it over hot lithium hydroxide
deposited on a carborundum substrate. The subsequent
coulometric determinations of water provided absolute
measurements for both elements.
Procedures based upon measurement of the differences
in thermal conductivity between the resident gas and the
combustion products offer, in general, greater speed and
less experimental manipulation. This type of analysis was
first applied to carbon and hydrogen by Duswalt and Brandt
(9) and Sundberg and Maresh (40) who used conventional
combustion techniques coupled with a gas chromatographic
determination of carbon dioxide and acetylene, an appropri
ate conversion product of water. The gases were trapped
in a U-tube immersed in liquid nitrogen and, when the
combustion had been completed, were introduced into a gas
chromatograph where they were separated and detected.
Oxygen and helium carrier gases were used respectively.


36
However, it was later found, that, where an extrusion process
failed, an electrodeposition of copper on a slightly tapered
iron mandrel proved quite satisfactory. The high co
efficient of thermal expansion of copper allowed removal of
the capsule hy immersion in boiling water.
Critical Paryneters in the Finalized System and
Extension to Nitrogen Determination
Evaluation of all factors previously discussed led
to the construction of an apparatus using the flow system
illustrated in Figure 7. This arrangement of components
was selected by virtue of expediency and its provision for
maximum experimental versatility. A similar, but somewhat
more advantageous arrangement was used in the design of the
prototype instrument discussed in the Appendix.
Choice of oxygen as a carrier gas
The necessity for an oxygen atmosphere to provide
maximum combustion efficiency for carbon and hydrogen
analyses was recognized by Liebig and Pregl in their
initial studies. In the present system where an extremely
rapid and uniform release of combustion products is
essential, an inert carrier gas such as helium or nitrogen
can not be used for general application. The inadequacy of
a non-oxidizing carrier gas is shown in Figure 8. Nitrogen
was selected for this test rather than helium due to the


a mixture of nitrogen dioxide and its dimer, dinitrogen
tetroxide, which would include the oxidation product of any
nitric oxide initially formed. Certain organics, such a3
diazo-compounds, would he presumed to decompose with the
release of elemental nitrogen, and the combustin temperature
would he far too low for any oxidation to the monoxide.
Because the thermal conductivities of nitrogen and oxygen
are nearly identical, virtually no response to elemental
nitrogen would he superimposed upon the carbon dioxide
signal. A number of reactants have been proposed for the
retention of nitrogen dioxide. Traditionally, lead peroxide
maintained at a temperature of 175-190C is used (57), and
absorbs nitrogen dioxide according to the reaction
2 N02 + 2 Pb02-*-Pb(N02)2 + PbO + \ 02 .
More recently, specially prepared manganese dioxide has
been used for this purpose. Onoe (32) found that activated
manganese dioxide prepared from manganese sulfate and
ammonium peroxydisulfate or from potassium permanganate and
methanol gave the best results, and concluded that the
reaction was simply
Mn02 + 2 N02 Mn(N03)2 .
Vecera and Snobl (43) maintained that the hydrated amorphous
gel [MnO(OH)2] binds nitrogen dioxide with its surface
hydroxyl groups according to the equation


LITERATURE CITED
1. American Instrument Co., Inc., Preliminary Data Sheet
2399.
2. Anon., "Fluid Meters, Their Theory and Application,"
5th ed., p. 81f, ASME, New York.
3. Belcher, R., Ingram, G., Anal. Chim. Acta 4, 118 (1950)
4. Belcher, R., Spooner, C. E., J. Chem. Soc. 194-3. 313.
5. Coleman Instruments, Inc., Bulletin B-273.
6. Ibid.. Bulletin B-258B.
7. Ibid., Paper No. 114 Presented at the Pittsburgh
Confernece on Analytical Chemistry and Applied
Spectroscopy, March, 1964.
8. Clerc, J. T., Dohner, R., Sauter, W.. Simon. W.. Helv.
Chim. Acta 46, 2369 (1963).
9. Duswalt, A. A., Brandt. W. W.. Anal. Chem. 32, 272
(1960).
10. F & M Scientific Corp., Bulletin 1800.
11. Finston, H. L., Yellin, E., Anal. Chem. 336 (1963).
12. Fisher Scientific Co., Bulletin FS-257.
15.Frazer, J. W., Mikrochim. Acta 1962. 993.
14. Haber, H. S., Gardiner. K. W., Microchem. J. 6. 83
(1962).
15. Horacek, J., Kbrbl, J.. Pechanec, V.. Mikrochim. Acta
I960. 294.
16. Hozumi, K., Kirsten, W. J., Anal. Chem. 35. 1522 (1963)
17. Ingram, G., Analyst 73. 548 (1948).
18. Kainz, G., Horwatitsch, H., Z. anal. Chem. 176. 175
(I960).
104


LIST OF FIGURES
Figure Page
1. Preliminary System and Readout for Carton
and Hydrogen Analysis 10
2. Preliminary Design of Sample Introduction -
System 14
.3. Circuit for Conductance Measurement of Water
and Thermal Conductivity Measurement of
Carbon Dioxide 17
4. Details of the Retention and Conversion
System 26
5. Variations in Component Arrangements and
Readouts for Carbon and Hydrogen Analysis . 30
6. Comparison of Sensitivity to Carbon Dioxide
Using Packed and Unpacked Columns Under
Identical Flow Conditions 33
7. Finalized Flow System and Readout for Carbon
and Hydrogen Analysis 37
8. Comparison of Combustion Efficiencies Using
Oxygen and Nitrogen Carrier Gases 38
9. Response to Carbon Dioxide Versus Equili
bration Time 4-6
10. Dependence of Response on Flow Rate 47
11. Reduction of Atmospheric Contamination During
Nitrogen Analyses 53
12. Instrumental Arrangement for Carbon and
Hydrogen Analysis 56
13. Details of Conversion and Detection Systems 57
14. Details of Orifice Mounting 58
v


64
5 per cent platinized asbestos (P-152, Fisher Scientific
Co., Pittsburgh, Pa.) as shown in Figure 16. For nitrogen
analysis, an additional packing of metallic copper (Cuprin;
29-120, Coleman Instruments, Inc.) was used, as indicated.
An open end Kimax Capillary Tube (1,5 i 100 mm) was
packed with 1 inch of 60-80 mesh silica gel (Coast Engineer
ing Company, California) and was connected to the combustion
tube by insertion through a 6 mm rubber sleeve type serum
stopper. The conversion tube (1 in. x 4 mm O.D.) was
packed with about a 5/16 inch layer of 40 mesh calcium
hydride (Metal Hydrides, Inc., Congress Street, Beverly,
Mass.) as shown in Figure 16. The hydride layer was re
tained by a tight plug of glass wool and was positioned
firmly against a 1/8 inch silicone rubber sleeve (0.125 in.
O.D., 0.062 in. I.D.; Heiss Manufacturing Corporation,
Little Falls, N. J.) which was resistant to the heat and
allowed rapid removal and replacement of the tube.
The filter for retaining oxides of nitrogen, con
sisting of a pyrex tube (3 in. x 7 mm O.D.) packed with
14-30 mesh activated manganese dioxide (33-110, Coleman
Instruments, Inc.) and magnesium perchlorate, is illustrated
in Figure 16. This was connected to a 6 foot length of
plastic tubing (3/16 in. O.D., 1/8 in. I.D.) which was
coiled around a cylindrical form. Plastic was chosen rather
than metal to reduce the possibility of surface adsorption


Fig. 20.-Encapsulation of Powdered Samples


This dissertation was prepared under the direction
of the chairman of the candidate's supervisory committee
and has been approved by all members of that committee. It
was submitted to the Dean of the College of Arts and
Sciences and to the Graduate Council, and was approved as
partial fulfillment of the requirements for the degree of
Doctor of Philosophy.
April 18, 1964
Dean, College o
s/And Sciences
Supervisory Committee:
Dean, Graduate School


12
tight, conical coil of #22 gauge platinum wire and pro
viding for its insertion into a quartz combustin tube,
thus enabling the weighing of solid samples. Passage of a
high current through the cone provided rapid but incomplete
combustion of the sample. In the case of benzoic acid, for
example, a substantial amount of the sublimate appeared on
the walls of the quartz tube. A further concept involved
the use of a weighable platinum tube inserted between two
heavy electrodes through which the carrier gas was passed.
Difficulties in these approaches included the unavoidable
disturbance of equilibrium conditions prior to the intro
duction of each sample and severe initial losses through
volatilization or spattering due to the finite time re
quired for the system to reach a threshold combustion
temperature.
It became evident that a stationary packed column
heated to the highest practical temperature and coupled
with an appropriate method for sample introduction would
eliminate these difficulties and would provide maximum
combustion efficiency. A Vycor tube (18 in. x 7 mm I.D.)
with a side arm (one inch from the upper end) for oxygen
flow was packed with wire-form copper oxide, and the upper
end was fitted with a serum cap. The tube was vertically
connected to a chromatographic system and was wrapped with
60 turns of #22 gauge Nichrome wire to which approximately


Pig. 21.-Readouts for Carbon Dioxide, Hydrogen, and Nitrogen
-o
oo


96
Sulfur and halogens
Numerous experimental difficulties were associated
with, the determination of sulfur, chlorine, bromine, and
iodine by the use of a chromatographic detection system.
Nevertheless, when an oxygen carrier gas was swept through
a combustion tube packed with quartz chips, rapid responses
were obtained for chloro-, bromo-, and sulfur-containing-
organics, and the resultant signals were identified with
chlorine, bromine, and sulfur dioxide, respectively.
Column separations were required, however, for each determi
nation and, in the case of sulfur compounds, the necessity
for conversion of the mixture of sulfur dioxide and sulfur
trioxide to a single component became evident. Sapid cor
rosion of thermistor leads, tubing and connectors led to
the discontinuance of these efforts pending more satisfactory
detection conditions.
Carbon in slow burning compounds
Certain analyses, such as the carbon content in steel,
require large samples and strong ignition over an extended
time interval. A simple adaptation of the original system
was shown to accommodate such determinations. The silica
gel packing of the capillary tube was replaced by a 2 inch
layer of 60-80 mesh 5A molecular sieve, and a larger moving
heater was constructed to jacket the packing. Time require
ments for complete combustion were indicated by running an


31
was permanently retained. The delay column which consisted
of approximately 50 feet of empty 1/8 inch plastic tubing
was so adjusted in length that after the first side of the
detector was cleared of carbon dioxide, the hydrogen passed
through the second side of the detector, thereby producing
a positive signal on the recorder. Diffusion of hydrogen
was almost negligible during its long transport through
the column, but analytical results suffered from such
factors as non-uniform release of hydrogen, increased column
impedance due to carbon dioxide absorption by the ascarite,
and non-linearity of response to carbon dioxide. Carbon
dioxide peaks were non-Gaussian, exhibiting pronounced
asymmetry w-tth increasing sample size. No specific improve
ment in elution symmetry was noted upon replacement of the
silica gel with a 6 foot column packed with 50 per cent
hexamethylphosphoramide on chromosorb. A number of problems
associated with the combined usage of oxygen as a carrier
gas and a silica gel column for retention of carbon dioxide
became apparent. Although sensitivity to hydrogen was more
than adequate, it was necessary to utilize the maximum
bridge output with a 1 millivolt recorder to obtain signifi
cant peak amplitudes for carbon dioxide. Unfavorable
signal-to-noise ratios were observed under these conditions,
and a large attenuation ratio was required for measurement
of the two components. Column overload at higher concen
tration levels caused deviations from linearity requiring


LLL
Fig. 12.-Instrumental Arrangement for Carbon and Hydrogen Analysis


94
The quantity of carton monoxide produced is stoichiometrically
related to the amount of oxygen present in the sample. Oxi
dation of carbon monoxide to carbon dioxide by passage
through a solution containing iodine pentoxide and by subse
quent titration of the iodine formed allows a determination
of the total amount of carbon monoxide released during the
combustion process. An alternate method requires oxidation
of the carbon monoxide by passage over hot copper oxide,
and a gravimetric determination of the resultant carbon
dioxide collected in an ascarite absorption tube.
A combustion tube was packed with a 6 inch layer of
20 mesh spectrographically pure carbon chips followed by
a 1 inch layer of silver wool. The Nichrome heating coil
was compressed to provide a maximum temperature in the area
of the upper portion of the carbon layer. A second
combustion tube, packed with copper oxide, was positioned
below the first tube, and a magnesium perchlorate filter
was added to retain the water formed during the oxidation
of hydrogen. Helium, de-oxygenated by passage through a
copper-filled furnace tube at 450C (S-56518, E. H. Sargent
and Co.), was used as a carrier gas. A method of sample
introduction identical to that used for the nitrogen analysis
was chosen to reduce atmospheric contamination to a minimum.
Erratic results were obtained for an initial series of
compounds, and it was found that non-oxygen-containing


59
to a flat surface to accommodate two 7/16 inch neoprene
o-rings between which the orifice disk was positioned. The
hole was made by denting the steel disk (0.010 in. x
7/16 in. D.) with a sharp punch and by carefully filing
the back until a breakthrough was obtained. The opening,
as observed under a microscope, appeared to be quite round
and was estimated to have a diameter of 0.0025 inch by
comparison to a wire of that size. Connection to the de
tector was made with heavy-walled tubing and all further
connections up to the combustion tube were made as short
and rigid as possible.
The detector block was machined from 504 stainless
steel and is shown in Figure 15. Individual empty cell
volumes were approximately 0.4 cc. Type G126, 2,000 ohm,
hermetically sealed Fenwal thermistors (Allied Electronics
Corporation, Chicago, Ill.) were seated in position. These
required no sealing washers. The leads were insulated with
teflon sleeves, and the cell was tested for leaks at an
oxygen pressure of 50 p.s.i.g. The detector assembly was
mounted in an aluminum minibox (2 5/4 x 2 1/8 x 1 5/8 in.),
and the spaces between the walls were tightly packed with
glass wool. The outlet from the reference cell was connected
directly to a plain end flow metering tube (tube size 1A-
15-1; Ace Glass Company, Vineland, F. J.) with a heavy wall
rubber coupling. The rotameter was connected, in turn, to


66
in a pyrex tube which, was held in position with a standard
laboratory thermometer clamp. Vertical adjustment was
provided by sliding the clamp along a 1/2 inch diameter rod.
A Powerstat (The Superior Electric Company, Bristol, Conn.)
was used to supply the required 6 volts. Operational data
for both heaters are presented in Table 3-
TABLE 3
HEATER POWER REQUIREMENTS
Combustion Capillary
Tube Heater Tube Heater
Nichrome Wire Gauge No.
22
24
Wire Length, inches
132 11
Cold Resistance, ohms 3.0
Applied a.c. Voltage, volts 117
Current, amperes (at
operating temperature) 7.5
Power Requirements, watts 880
11.4
6
2
12
A schematic diagram of the regulated power supply
and bridge circuit is shown in Figure 18. The solid-state
power supply, using two transistors and a zener diode,
provided excellent regulation with minimum a.c. ripple.
Performance characteristics under extreme input voltage
variations are indicated in Figure 19.


SECTION IV
RESULTS AND DISCUSSION
Before any standardizations or analyses were con
ducted, the system was thoroughly tested for leaks. This
was accomplished most conveniently hy closing off the exit
of the sample side of the detector cell and noting the
corresponding flow decrease on the rotameter. If the flow
did not decrease to zero after several minutes, portions
of the system were isolated until the point of leakage was
determined. Precision and accuracy were found to he
extremely dependent upon the maintenance of a totally leak-
tight system and leakage could generally he recognized by
failure of a component peak to return to the base line.
Other malfunctions during an analysis were readily indicated
by an anomalous response pattern such as the occurrence of
double peaks, negative peaks, peak broadening, etc. It was
necessary, of course, to minimize the recorder deadband
before any determinations were made.
Typical readouts for carbon and hydrogen, and for
nitrogen are shown in Figure 21. Linearity over a normal
concentration range is indicated by the curves of Figure
22. A benzoic acid standard was used for the carbon and
77


97
initial sample and by noting the time, after successive
lowerings of the heater, at which no additional carbon
dioxide was expelled from the capillary tube- Complete
retention of carbon dioxide was observed, and its release
into the system was rapid and reproducible. A standard
was selected which was similar in structure and in carbon
content to the sample.
Moisture and water of composition
Due to the high sensitivity obtained by conversion
of water to hydrogen, the system proved to be extremely
efficient for the determination of trace amounts of moisture
and for the degree of hydration of inorganic compounds. It
was necessary only to release any extraneous moisture which
had collected in the silica gel immediately prior to the
introduction of each sample.


105
19. Ibid.. 177, 521 (I960).
20. Ibid., p. 344.
21. Ibid., 178, 406 (1961).
22. Ibid.. 184, 363 (1961).
23. Keidel, I>. A., Anal. Chem. 1, 2043 (1959).
24. Keulemans, A. I. M., "Gas Chromatography," 2nd ed.,
p. 90, Eeinhold, New York, 1959.
25. Krbl, J., Collection Czech. Chem. Commun. 20. 953
(1955).
26. Ibid., p. 1026.
27. Kuck, J. A., Berry, J. W., Andreatch, A. J., Lentz,
P. A., Anal. Chem. 4, 403 (1962).
28. Malissa, H., Pell, E., Michrochem. J., Symp. Ser. 2,
371 (1962), through Chem. Abstr. 58, 6191e (1963)
29. Maresh, C., Sundberg, 0. E., Hofstader, R. A.,
Gerhardt, G. E., Michrochem. J., Symp. Ser. 2, 387
(1962), through C. A. 58, 6l9lf (1963).
30. Marks, R., Rev. Sci. Instr. 28, 381 (1957).
31. Nightingale, C. P., Walker, J. M., Anal. Chem. 34,
1435 (1962).
32. Onoe, T., Ann. Rep. Takamine Lab. 10, 128 (1958),
through C. A. 56, 8003f (1962).
33 Simon, W., Sommer, P. Z., Lysey, G. H., Michrochem. J.
1962. 239.
34. Smith, B. D., Bowden, W. W., Anal. Chem. 6, 82 (1964).
35. Steyermark, A., "Quantitative Organic Microanalysis,"
2nd ed., p. 171, Academic Press, New York, 1961.
36. Ibid., p. 222.
37. Ibid.. pp. 233, 238.
38. Ibid., p. 234.
39. Ibid.. p. 255.


10
Fig. 1.-Preliminary System and Readout for Carbon and
Hydrogen Analysis


95
materials produced carbon dioxide peaks as well. This was
due, in part, to tbeir passage in the vapor phase through
the oxidizing portion of the combustion tube. A substantial
improvement was made by elimination of the copper oxide and
by direct determination of carbon monoxide. Besults for a
series of 5 compounds, using benzoic acid as a standard,
are shown in Table 9. Discrepancies in these values could
be attributed to positive or negative contributions to the
thermal conductivity of the composite consisting of carbon
monoxide and small amounts of unconverted hydrogen or low
molecular weight pyrolysis products such as methane or
ethane. It was evident that further work would necessarily
require a column separation of carbon monoxide from these
and other contaminants.
TABLE 9
EESULTS FOB OXYGEN ANALYSES
Compound
Benzoin
Benzophenone
/¡-Naphthol
Sucrose
%Q
Theor. Anal.'
15.08 14.3
8.78 9.0
11.10 11.1
51.42 50.2
31.55 31.6
Vanillin


73
Instrumental procedure
With oxygen flowing through the system, power was
switched on to the furnace, to the detector, to the capil
lary heater, and to the recorder. After a few minutes the
oxygen flow rate was adjusted to 40.0 cc/min. hy regulation
of the upstream pressure. Periodic checks on flow values
were made using the soap-film flow meter. The instrument
was allowed to equilibrate for at least two hours before
running a sample. It was considered advisable to turn off
only the furnace heater during breaks in extended periods
of operation.
With the input shorted (attenuator switch at the
"S" position), the recorder was adjusted so that the pen
rested exactly on the zero position of the chart. Recorder
amplifier sensitivity was adjusted for maximum gain and
was then reduced to the first point where no oscillation
occurred at any position on the scale. With the attenuator
in the position of maximum sensitivity (XI), the 50,000
ohm helipot was turned to zero the recorder. A benzoic
acid sample weighing close to 0.65 milligrams was introduced
into the sample port and then into the system by clockwise
rotation of the valve. After the appearance of the carbon
dioxide peak, the bridge sensitivity was set by the following
procedure. The zero control was first adjusted to provide
a signal corresponding exactly to the initial peak height.


APPENDIX


34
the combined combustion and detection techniques was next
tested, and results are presented in Table 1.
TABLE 1
TEST RESULTS FOR ACCURACY OF CARBON ANALYSIS
Compound
%C, Found
%C, Theoretical
/S-Naphthol
83.5
83.3
Naphthalene
93.6
93.7
Benzophenone
85.8
85.7
Oxalic acid
19.0
19.0
Phthalic acid
57.8
57.8
9-Fluorenone
86.6
86.6
Triphenylphosphine
82.4
82.4
Sample handling and weighing
A Cahn Gram Electrobalance (Model G, Cahn Instrument
Company, Paramount, California) was selected for all sample
weighings. The generad performance of the balance proved
to be outstanding due to its insensitivity to environmental
conditions such as vibration, temperature variation, air
currents, etc. The specified limits of precision and
accuracy for the weight range of interest are 0.1 micrograms
and 0.5 micrograms, respectively. Exact proportionality,
however, rather than absolute weighing accuracy is essential
to this work. Because very critical weighings were required
the balance was zeroed with each determination. The


COp Peak; Height, it
33
Weight of Polyethylene, mg
Fig. 6.-Comparison of Sensitivity to Carhon Dioxide Using
Packed and Unpacked Columns Under Identical Flow
Conditions


88
the carbon contents reported for phosphorus compounds by
commercial analytical laboratories have generally been 1-3
per cent lower than the theoretical values. Excellent
results, however, were obtained in this study for similar
or identical compounds with no additional catalytic or
temperature requirements. This was attributed to the high
combustion efficiency of the system and to its acceptance
of small samples.
Of all the compounds analyzed, oily one gave con
sistently low results. Potassium acid phthalate (47.05% C,
2.47% H) repeatedly showed a hydrogen content of 2.35 +
0.05 per cent and a carbon content of 44.0 + 0.1 per cent.
This corresponded very closely to the values calculated
for the reaction
OC00K
+ 15 05K~C0, + 15 C0~ + 5 H.0 .
COOH 2 2 3 2 2
Results were not noticeably improved by the addition of
cobaltic oxide, but encapsulation of the sample with
approximately 5 milligrams of silver permanganate gave
values within 0.15 per cent of the theoretical percentages.
The silver permanganate prepared in this laboratory showed
no background response for carbon, hydrogen, or nitrogen
and was recommended as a general catalyst for all compounds
which were suspected to exhibit anomalous combustion


61
a pyrex tube (18 in. x 7 mm O.D.) packed with. 20-60 mesh
magnesium perchlorate (Anhydrone; Coleman Instruments, Inc.
Maywood, Ill.) and 8-20 mesh ascarite (Arthur H. Thomas
Company, Philadelphia, Pa.) as indicated in Figure 16. The
tube was wrapped with a sheet of aluminum foil to reflect
heat radiated from the furnace in order to prevent thermal
expansion and contraction of the packing. A short heavy
rubber connector was used to Join the filter tube and the
sample introduction port. Details of the valve assembly
and construction are presented in Figure 17. The body and
the central rotating plate were machined from 304 stainless
steel, and the upper and lower rotary port seats were
machined from Zel-F. All moving surfaces were hand-lapped
to prevent leakage. The 1/4 inch outlet allowed rapid
connection and replacement of the combustion tube by means
of a short piece of rubber tubing. The connector was pro
tected from the heat of the furnace by wrapping it with
aluminum foil. Because it was essential to maintain the
injection port at room temperature to prevent pre-vapori
zation of volatile samples, a large aluminum heat reflector
(6 1/2 x 10 in.) was positioned directly below the valve.
The ycor combustion tube (18 in. x 7 mm O.D.,
743172, Corning Glass Works, Corning, M.Y.) was packed with
copper oxide (Cuprox; 29-130, Coleman Instruments, Inc.),
silver metal wool (E. H. Sargent and Co., Chicago, Ill.) and


23
pyrex tube equal in diameter to tbe combustion tube and
packed with, approximately 1/4 inch, of 40 mesh calcium
hydride. The peak height response was observed to decrease
markedly after the combustion of three or four samples.
Asymmetric peaks were particularly evident in the case of
large samples or those with a substantial hydrogen content
and were attributed, in part, to the observable condensation
of water prior to its passage through the calcium hydride,
resulting in extremely slow evolution of hydrogen. A
corrective procedure was attempted in which the conversion
tube was placed in an oven controlled at 110C. The system
was first tested for evidences of thermal degradation of
the calcium hydride by plotting the effluent signal versus
temperature as indicated by a thermometer attached to the
tube. No hydrogen evolution was noted during elevation of
the oven temperature to 130C. Although some improvement
due to elimination of condensation problems was observed,
no noticeable gain in reactivity or reduction in tailing
accompanied the temperature increase. Resultant erratic
values for hydrogen precluded the use of this system.
A unique application of an apparatus patterned after
the electrolytic moisture detector of Keidel (23) was
evaluated in an effort to obtain a more spontaneous release
of hydrogen. A hygrometer element, identical to the one
previously described under "Detection systems" was coated


74
The 500,000 ohm potentiometer was then adjusted to shift
the deflection to 75 per cent of full scale. The silica
gel was cleared of water by lowering the capillary tube
heater, and the converter tube was filled with fresh
calcium hydride.
With the attenuator at the XI position and the re
corder at the zero position, a second sample of benzoic
acid weighing approximately 0.65 milligram was introduced
into the combustion tube. As soon as the carbon dioxide
peak appeared the heater was lowered to the silica gel
portion of the capillary tube and was allowed to rest on the
silicone rubber insert of the converter tube for 5 seconds.
The heater was returned to its original position, the
recorder polarity switch was set at the "Hg" position, the
attenuator was adjusted to X4, and the recorder was re
zeroed, if necessary. Immediately after the hydrogen peak
emerged, the polarity switch was returned to the "COg"
position, the attenuator was reset to XI, and the system
was ready to accept a sample.
For those samples with known approximate compo
sitions, quantities were selected which corresponded roughly
to carbon contents of 0.45 milligrams (the approximate
weight of carbon in 0.65 mg of benzoic acid). In the case
of a sample of totally unknown content, it was sometimes
advantageous to make a quick preliminary determination so


22
dioxide retention occurred as a result of its passage
through calcium hydride or calcium hydroxide. This was
indicated by a series of careful injections of carbon
dioxide at points immediately preceding and immediately
following a tube packed with calcium hydride. Similar
injections were made after total depletion of the calcium
hydride with water. Peak heights, corresponding to carbon
dioxide concentrations, were completely unaffected by
either of the packings.
Calcium hydride was selected over a series of other
metal hydrides for several reasons. Its reaction with
water to release hydrogen was rapid and quantitative. Prom
a practical standpoint, it was found to be more easily
handled than the sodium, lithium, or lithium-aluminum
analogs; i.e., a container of calcium hydride could be
opened and exposed briefly to the atmosphere innumerable
times without observable decomposition. Its hydroxide
displayed the least tendency to retain water, as evidenced
by the pronounced reduction in tailing over that experienced
with the other hydroxides. This corresponded to prediction
because calcium hydride is neither hygroscopic nor does it
form a hydrate. Its small but definite affinity for water
under the conditions of the preliminary systems, however,
was sufficient to produce critical design problems.
An initial conversion system consisted of a short


Output, Volts
68
Pig. 19.-Stability of Power Supply With Change in Input
Voltage


ACKNOWLEDGMENTS
The author extends his sincere appreciation to his
supervisory committee, to the faculty of the Chemistry
Department of the University of Florida, and to his
colleagues for their interest and their friendly cooperation
during the course of this investigation. He takes
particular pleasure in expressing gratitude to his research
director, Dr. J. D. Winefordner, for his constant sincerity
and willingness in providing the counsel, guidance, and
encouragement necessary for the realization of this project.
The helpful suggestions and the cooperation of the
research and development group of the American Instrument
Company are deeply appreciated. Mr. William C. Hampton is
particularly deserving of special thanks in this regard.
Additional recognition is due the American Instrument
Company for the financial support of this research, and the
assistance is highly valued.
Finally, the author gratefully acknowledges his
indebtedness to his wife, Hita Sue, for her continued
inspiration and understanding.
ii


103
Fig. 23.-Prototype Model of American Instrument Co. Carbon
and Hydrogen Analyzer


2-
Sample Port
Combustion Tube
Calcium Hydride
Silica Gel Column
Detector
CO-,
30
2
Sample Port
Combustion Tube
Calcium Hydride
Silica Gel Column
CO,
Detector
Ascarite
Delay Column IT
Ho
u
Fig. 5.-Variations in Component Arrangements and Readouts
for Carbon and Hydrogen Analysis


92
The initial results for nitrogen determinations
using a carbon dioxide carrier gas were satisfactory for
high nitrogen percentages but the signal-to-noise ratio
was too low to provide sufficiently accurate values for
compounds of low nitrogen content. All reported results
were, therefore, obtained using a helium carrier gas.
Experimental studies have indicated that the
accuracy of this method of analysis is comparable to that
of the conventional micro-Dumas technique for nitrogen,
and exceeds that of the conventional Pregl technique for
carbon and hydrogen. Its main advantages lie in its speed,
simplicity, reduction in operational requirements, and its
acceptance of small samples. With two functional systems
and a single recorder, a complete carbon, hydrogen, and
nitrogen analysis may be performed in less than ten minutes
even though two weighings are required. New commercially
available instruments which allow CHN determinations to be
performed on a single sample require considerably longer
analysis time and suffer from inaccuracies inherent in
combining the procedures.
It is anticipated that, with additional improvements,
the system described will require less frequent standardi
zation and will offer even greater stability, precision and
accuracy


21
calcium carbide. Sensitivity was more than adequate with
respect to carbon dioxide but was inadequate with respect
to acetylene. In addition, anomalous negative response was
observed for acetylene due to longitudinal diffusion within
the column, with subsequent depression of its maximum
concentration in the carrier gas. An immediate improvement
developed from a consideration of the reaction sequence
CmHn + (m + £)02^m C02 + § HgO
| H20 + | CaC2 $ Ca(0H)2 + £ C2H2
in contrast to which the reaction of water with an active
metal hydride
| H20 + S CaH2 Ca(0H)2 + §
would provide a distinct increase in sensitivity as a result
of the more favorable stoichiometry and the higher thermal
conductivity of hydrogen. In addition, non-uniformities
resulting from peak spreading could be eliminated due to
the elution of hydrogen prior to carbon dioxide.
Detrimental effects of two reactions on quantitative
measurements of carbon dioxide were considered
2 C02 + CaH2 Ca(C00H)2
C02 + Ca(0H)2 CaCO^ + H20 .
It was proven experimentally, however, that under the
temperature and flow conditions of the analysis no carbon


Orifice
Pig. 14.-Details of Orifice Mounting


57
Pig. 13.-Details of Conversion and Detection Systems


TABLE 7 Continued
Compound
Structure
Origin
% C
Theor. Anal.
Diff.
%
% H
Theor. Anal.
Diff.
%
Tetra-n-butyl-
ammonium
Iodide
[ch3(ch2)5]4ni
Eastman
Organic
Chemicals
No. 4702
52.03
52.49
+0.46
9.76
9.50
-0.26
4,4,4-Tri-
fluoro-1-
(2-thienyl)-
1,3-hutane-
dione
SCH:CHCH:CC0CHoC0CF-,
1 2 3
Eastman
Organic
Chemicals
No. 7260
43.25
43.49
+0.14
2.25
2.49
+0.14
Triphenyl-
phosphine
(c6h5)3p
Metal and
Thermite
Corp.
82.43
82.41
-0.02
5.76
5.83
+0.08
Vanillin
CH,0(0H)CfiH,CH0
Eastman
63.15
62.94
-0.21
5.30
5.46
+0.16
Organic
Chemicals
No. 2?3
Silver permanganate added
+Liquid sample
00
-o


50
Nitrogen determination
A simple modification of the finalized system for
carbon and hydrogen analysis provided a rapid method for
the determination of nitrogen. In initial studies the
combustion tube was repacked with a 4 inch layer of copper
oxide (20 mesh) followed by a 4 inch layer of freshly
hydrogen-reduced copper (20 mesh) and a 1 inch layer of
silver wool. The conversion system was removed and, when
helium was used as a carrier gas, the manganese dioxide
cartridge was replaced by a filter packed with magnesium
perchlorate and ascarite. Only the desiccant was required
for a carbon dioxide carrier gas, and the combustion tube
was connected directly to the filter.
Sample analysis was identical to that in the carbon
and hydrogen determination with the exception that a
catalyst was generally added. Combustion products were
swept through the copper section of the tube where oxides
of nitrogen were reduced to elemental nitrogen. Halogens
and oxides of sulfur were retained by the silver, and water
and carbon dioxide were absorbed where required. The
nitrogen peak emerged in 30-45 seconds after introduction
of the sample. The reaction
CmHnNo + + jf)2 m C02 + | H20 + §
is less favorable, stoichiometrically, for nitrogen than


18
both the reference and the sample sensing elements were
located in the direct path of the carrier gas. Unlike the
diffusion cell, this provided a rapid quantitative response
which was somewhat sensitive to flow fluctuations.
Thermistors with a room temperature resistance of 100,000
ohms were used for initial studies, but later work indicated
improved stability with less expensive 2,000 ohm thermistors.
The desirable linearity of output signal with
respect to concentration is not completely predictable from
theoretical considerations. A strict mathematical treat
ment taking into account all subsidiary effects is
complicated and impractical due to a lack of accurate data.
In general, the temperature-response behavior of a sensing
thermistor over a narrow temperature range may be repre
sented by
) ] ,
where the coefficient is a function of temperature and
is given approximately by
CK
where the constant p is defined by


6
products into a series of filters which selectively absorbed
water and carbon dioxide Static thermal conductivity
measurements between various portions of the system pro
vided, under equilibrium conditions, data relating
concentration differences between composites and individual
components The procedure of Valisch (46) depended upon
sample combustion over copper oxide in a helium carrier
gas containing 3 per cent oxygen, followed by the reduction
of oxygen and the oxides of nitrogen over copper. The
signal corresponding to nitrogen was subtracted from that
of the carbon dioxide-nitrogen composite, and water was
determined independently. The principle of analysis was
adopted by the Technicon Instruments Corporation in the
development of a commercial carbon, hydrogen, and nitrogen
analyzer (41) An instrument designed and manufactured
by the P and M Scientific Corporation (10) used a series
arrangement of two katharometers to measure the water,
nitrogen, and carbon dioxide produced by burning the sample
in a helium atmosphere.
Of the methods described, those based on a gas
chromatographic detection of combustion products provide
greater operational simplicity, are less susceptible to
human error, and are capable of substantially reducing
analysis time although the quoted times range between 10
and 105 minutes per sample. Deviations, however, from


SECTION II
EXPERIMENTAL CONSIDERATIONS
Initial Functional Objectives and Preliminary Concepts
The procedures reported by Duswalt and Brandt (9)
and Sundberg and Maresh (29,40) first indicated the
feasibility of replacement of conventional gravimetric or
volumetric measurements of combustion products by a gas
chromatographic system. This innovation was accompanied by
little or no net increase in accuracy or speed of analysis,
and the additional requirements of a gas chromatograph and
a liquid nitrogen system limited the technique from an
economical standpoint.
It was proposed, therefore, that a system based on
a dynamic sensing device be designed which would provide a
more rapid, efficient, and economical means for the micro-
determination of carbon and hydrogen. To achieve maximum
speed of analysis, initial experimental procedures were
based upon the use of small samples and a method for flash
combustion providing rapid pyrolysis. In addition, it was
desired to avoid retention of combustion products. Evalu
ation of this approach was facilitated by construction of
8


DESIGN AND PERFORMANCE CRITERIA IN
THE DEVELOPMENT OF A
NEW INSTRUMENTAL METHOD
FOR ULTRAMICRO ELEMENTAL ANALYSIS
By
C. DAVID MILLER
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
April, 1964


TABLE 5
PRECISION FOR A TYPICAL CARBON AND HYDROGEN DETERMINATION
Determi
nation
No.
Benzoin
Benzoic Acid
Benzoin Composition, Found
Wt. ,
MS
P.H. ,
X
co2
inches Wt.,
10 Mg
H2
P.H. ,
X
co2
inches
10
h2
% C
Deviation
from
mean, %
% H
Deviation
from
mean, %
1
59^.3
78.95
53.75* 612.3
70.70
95.75
79.21
+0.02
5.72
+0.03
2
627.9
83.05
54.10* 661.7
75.95
97.20
79.34
+0.15
5.81
+0.12
3
571.1
74.35
92.30 632.1
71.60
90.05
79.13
-0.06
5.62
-0.07
4
672.0
89.50
58.80* 623.0
72.20
97.10
79.12
-0.07
5.56
-0.13
5
637.3
83.85
55.55* 619.2
70.75
91.15
79.28
+0.09
5.86
+0.17
6
598.9
76.80
94.45 654.4
73.10
91.30
79.04
-O.15
5.60
-0.09
*
Mean
79719
Mean
5.69
*
At an attenuation of X4. (All other hydrogen values were obtained at an attenuation
Theoretical Values,
% C
% H
Benzoic Acid
68.85
4.95
Benzoin
79.23
5.70
Mean Experimental Values, Benzoin
79.19
5.69
Average Deviation
0.09
0.10
Standard Deviation
0.11
0.12
% Standard Deviation
0.14
2.1
00


BIOGRAPHICAL SKETCH
C. David Miller was born April 7, 1931 at Baltimore,
Maryland. He was graduated from Leonia High. School, Leonia,
New Jersey in 194-8, and in 1952 he received the degree of
Bachelor of Arts from Columbia University.
He was employed as an analytical chemist by the
E. I. Du Pont de Nemours Company of Gibbstown, New Jersey
during 1952 and 1953 Following two years of service in
the Chemical Corps of the United States Army he returned
to Columbia University to conduct research in physical-
analytical chemistry. In 1957 he entered the Graduate
School of the University of Maryland and he received the
degree of Master of Science in organic chemistry in 1959.
He then accepted a position with the American Instrument
Company of Silver Spring, Maryland where he was engaged in
analytical instrumentation development until his enrollment
in the graduate school of the University of Florida in
September, 1961. During the pursuit of his work toward the
degree of Doctor of Philosophy he was the recipient of a
research grant sponsored by the American Instrument Company.
C. David Miller is married to the former Hita Sue
Showalter. He is a member of the American Chemical Society
and the Electron Microscopy Society of America.
107


106
40. Sundberg, 0. E., Maresh, C., Anal. Chem. 32, 274 (I960).
41. Technicon Instruments Corp., Bulletin No. CHN-1.
42. Trusell, F., Diehl, H., Anal. Chem. 35, 6?4 (1963).
43. Vecera. M.. Snobl. D.. Collection Czech. Chem. Commun.
2, 2013 (I960).
44. Vecera, M., Snobl, D., Synek, L., Mikrochim. Acta
1961, 370, through C. A. 6, 9420d (1962J.
45. Vogel, A. M., Quattrone J. J., Anal. Chem. 32. 1754
(I960).
46. Walisch, V., Chem. Ber. §4, 2314 (1961).


28
time coordinate would be required for an analysis. As work
progressed, it was observed that combustion rates were suf
ficiently constant, for virtually all substances encountered,
that peak height measurements could be used advantageously.
This provided a simpler and more rapid readout, offered a
wider chart span on the same recorder, eliminated the
additional cost of integrating equipment, and facilitated
adaptation to a digital readout system, if desired.
A Minneapolis Honeywell 1 millivolt Brown recorder
with a response time of 1 second, full scale, was used for
all reported results. The specified accuracy, of + 0.25 per
cent was recognized to be a limiting factor although perfor
mance data for the system indicated this to be a somewhat
conservative figure. Because the maximum bridge out-of-
balance voltage was in excess of 10 millivolts, a Sargent
Model TH recorder was preferentially selected for use with
the prototype instrument discussed in the Appendix, by
virtue of the + 0.1 per cent accuracy specified for this
range. The Sargent recorder offers the additional advantage
of an improved pen system and a displacement adjustment in
spite of a slightly reduced chart span (9'A in. vs. 10 in.).
Besponse magnitudes for carbon dioxide and hydrogen are
approximately proportional to the differences between their
respective thermal conductivities and that of oxygen.


SECTION V
EXTENSIONS TO ADDITIONAL ANALYSES
A simple modification of the apparatus originally
designed for the determination of carbon and hydrogen
provided a satisfactory analysis for nitrogen. Adaptation
of the general approach to the analysis of oxygen, sulfur,
and halogens was, therefore, investigated.
Oxygen
Recognition of the difficulties involved in the
direct analysis of oxygen in organic compounds has generally
led to the calculation of oxygen content by difference.
Good results, however, have been reported for certain
materials through the use of a Pregl method coupled with
a volumetric or gravimetric determination of the combustion
products. The analysis is based upon the ignition of the
compound in the presence of carbon at 1120C using an
oxygen-free carrier gas such as nitrogen or helium. At
this temperature the following pertinent reactions occur
Organic 0 + C -*-C0
C02 + C 2 CO
HgO + C CO + H2 .
93


72
normal manner on a 3/4 inch, length of capillary tube using
a pre-weighed square of aluminum. The other end of the tuhe
was inserted into the powder and an appropriate amount of
the sample was introduced into the capsule by means of a
metal plunger. In this manner the compound was exposed to
the air for only a few seconds before being permanently
sealed and no losses were incurred through spillage or
adherence to the outside walls.
The percentage composition of benzoic acid (68.85 %
C, 4.95 % H) approaches the mean composition expected for
most non-halogen-containing organic compounds. This factor,
as well as its availability in extremely pure form and its
lack of affinity for moisture, contributed to its selection
as a calibration standard. A weight of 0.65 milligrams of
benzoic acid was considered to be optimum because it allowed
good weighing precision and did not require a high sensi
tivity setting. For maximum accuracy, the sample weight
was selected so that the resultant carbon dioxide signal
corresponded approximately to that of the standard without
an attenuation change. This was Justified by the substantial
ly lower percentage accuracy requirements for hydrogen.
Peak amplitudes were adjusted, in each case, to lie between
50 and 100 per cent of full scale for greater readant
accuracy


100
Hydrogen is released quantitatively and is measured by tbe
same detector. Peak heights are linear functions of
concentrations, and percentage compositions are calculated
by comparison to a single standard (usually benzoic acid).
For nitrogenous compounds the procedure is repeated with
the exception that the nitrogen content of the sample is
completely converted to elemental nitrogen which is
detected in a helium carrier gas.
It is hoped that the method, by virtue of its speed,
accuracy, and ease of operation, will prove to be a valuable
asset to the synthetic organic, pharmaceutical, or medical
laboratory as well as to structural research and industrial
process control.


Fig. 8.-Comparison of Combustion Efficiencies Using Oxygen
and Nitrogen Carrier Gases
V>J
Co


COp Peak Height Response, mv/mg Polyethylene
46
Time Elapsed, Min., After Switching on Thermistor
Detector
Fig. 9.-Response to Carbon Dioxide Versus Equilibration Time


27
sometimes observed immediately after a sample combustion,
was of no consequence since all water droplets were swept
into the silica gel before the appearance of the carbon
dioxide peak. Positioning the moving furnace at the upper
portion of the capillary tube as well as the use of silver
wool as a heat transfer agent in the combustion tube aided
significantly in the prevention of condensation. In an
actual analysis the sample was introduced with the moving
furnace remaining in the upper position. Immediately after
observance of the carbon dioxide peak maximum (30-45
seconds), the furnace was dropped to its lower position
where it was seated against the silicone rubber insert for
3-5 seconds, vaporizing the entrapped water almost
instantaneously. Attenuation and recorder lead polarity
were adjusted and the resultant signal for hydrogen appeared
within 30-45 seconds. Incorporation of the calcium hydride
in the capillary tube directly below the silica gel proved
unsatisfactory due to prohibitively large flow restrictions
arising from the expansion accompanying the conversion to
calcium hydroxide. The use of a larger tube prevented this
difficulty, facilitated rapid hydride replacement, and
insured uniform release of hydrogen.
Readout systems
It was presumed at the start of the investigation
that a readout system providing signal integration over the


25
with a short plug of asbestos and was packed with a 1 inch
layer of silica gel (60-80 mesh). Tight connection to the
combustion tube was made through a serum vial cap, and a
1/8 inch I. D. pyrex tube packed with a 1/4 inch layer of
calcium hydride was positioned directly below the capillary
tube by means of a silicone rubber insert. The calcium
hydride was pushed tightly against the seal to eliminate
all possibility of condensation prior to conversion. A
small moving furnace designed to operate at a constant
temperature of approximately 200C encircled the capillary
tube as indicated in Figure 4. The velocity of the carrier
gas through the capillary was nearly twenty times greater
than its velocity through the combustion tube or through
previously designed conversion tubes. This was indicated
by the reciprocal ratio of the respective cross-sectional
areas. Similarly, a five-fold increase in oxygen velocity
was calculated for the calcium hydride tube. The extremely
high transfer rate of.carbon dioxide through this portion
of the system eliminated any possibility of its retention
by the silica gel or its reaction with calcium hydride or
calcium hydroxide. Water, on the other hand, was retained
with excellent efficiency by the silica gel and in no
instances, regardless of sample size or hydrogen content,
was an overload of the desiccant noted. The slight conden
sation on the walls of the capillary tube, which was


29
Setting A^Cq2>02) 1 we find ~ 15 at 25C.
For most compounds a two- to four-fold attenuation was
required to adjust the amplitude of the hydrogen signal to
approximately that of carbon dioxide. A double-pole double-
throw switching mechanism was provided for reversal of
recorder-lead polarity, enabling the use of the full chart
span for both carbon dioxide and hydrogen peaks.
Column systems and arrangement of components
The preliminary experimental procedure relied upon
the use of a silica gel column for the separation of carbon
dioxide and acetylene. In later work, a silica gel packing
was also used to separate hydrogen from carbon dioxide.
Two basic arrangements of components, and the resultant
readouts, with no electrical reversal of polarity, are
shown in Figure 5. The second system was devised for a
more completely automatic approach and provided a re-
inversion of the hydrogen signal thus facilitating the use
of automatic attenuation. The procedure involved passage
of the combustion and conversion products through the silica
gel on which carbon dioxide was initially retained while
hydrogen was swept through the first side of the detector
producing a negative signal. During the flow of hydrogen
through the delay column, the carbon dioxide was eluted
from the silica gel producing a positive signal on the re
corder, and then flowed into the ascarite filter where it


TABLE 8
ANALYTICAL RESULTS FOR NITROGEN
% N
Compound
Structure
Origin
Theor
Anal,
Diff., %
Acetanilide
ch5conhc6h5
Merck and Co.
(U.S.P.)
10.56
10.51
+0.15
Ammonium
Nitrate
nh4no3
Fisher Scientific
Co., No. A-676
55.00
55.28
+0.28
Asparagin
NH2C0CH2CH(NH2)C00H-H20
Fisher Scientific
Co., No. A-912
18.66
18.49
-0.17
p-Bromo-
aniline*
BrC6H4NH2
Eastman Organic
Chemicals
No. 475
8.14
8.50
+0.16
Caffeine*
CHzNCON(CH,)COC:CN:CHNCH*
Eastman Organic
No. 555
28.85
29.12
+0.27
Creatine
NH2C(:NH)N(CH5)CH2COOH
Nutritional
Biochemicals
Corpo
52.05
51.69
-0.56
Cystine*
[SCH2CH(NH2)C00H]2
Merck and Co.
No. 72612
11.66
11.50
-0.16
vD
O


102
The basic principles and operational and structural
details of the described instrumental procedure have been
applied in the production of a commercial carbon and
hydrogen analyzer (1) by the American Instrument Company
of Silver Spring, Maryland, the prototype model of which
is shown in Figure 23.
The instrument is so constructed as to allow rapid
replacement of component parts, and provides precision flow
control and temperature maintenance. The pre-conditioning
column is positioned prior to the pressure regulator, and
the reference detector cell is situated between the
throttling orifice and the sample introduction port to
insure maximum stability. Power requirements are sub
stantially reduced by enclosing the combustion tube in a
ceramic jacket which, in turn, is further insulated to
prevent radiational losses. The 250C heater is moved to
either extremity of the capillary tube by a rotary dial
mechanism. The complete unit is packaged in a cabinet 24
inches high, 12 inches wide, and 10 inches long. Output
signals are measured on a 5 millivolt recorder (Model TR,
E. H. Sargent and Co.), and an electrobalance (Model G,
Cahn Instrument Co.) is recommended for sample weighing.
The production of a similar instrument for the determination
of nitrogen is planned for the near future.


15
general requirements of stability and speed and linearity
of response were satisfied, and adequate sensitivity to
carbon dioxide was obtained. However, because of tbe use
of oxygen as a carrier gas and tbe similarities in thermal
conductivities for oxygen and water, it was not practical
to measure water directly by this technique. Other detec
tion systems were constructed and tested in an initial
effort to eliminate conversion requirements. The hydrogen
flame detector, used conventionally, shows little or no
response to water, carbon dioxide, or other permanent gases.
Using this apparent disadvantage, provision was made for
the premixing of hydrogen, containing an ethanol contaminant,
with the oxygen carrier gas and for the introduction of
this mixture into an appropriate burner-collector electrode
assembly. It was assumed that the diluent effect of the
normally non-responsive carbon dioxide and water would
result in significant negative response in comparison to
the elevated background signal provided by the constant
ethanol bleed. The procedure was discarded, however, due
to its relative complexity and critical operational
requirements.
A second approach based on conductance measurements
utilized a hygrometer element consisting of thin palladium
wires wound concentrically about a plastic core which was
coated with a small amount of lithium chloride in a