Chapter 3 - Experimental

3.1  Introduction

 

            A hot filament chemical vapour deposition chamber has been used to deposit diamond films onto silicon (100) substrates using a variety of input source gases, under standard deposition conditions (See Table 3.1).  The resulting films were examined using a variety of analytical techniques (See Chapter 2), (1) Auger Electron Spectroscopy (AES) to measure the dopant concentrations in the diamond films, (2) Laser Raman Spectroscopy (LRS) to analyze the quality of the films, namely their sp2-to-sp3 ratios and, (3) Scanning Electron Microscopy (SEM) to examine the film surface morphology, thickness and growth rates.

Pressure 20 Torr
Gases C/H,C/H/Cl, C/H/N and C/H/P systems
Total gas flow rate 100 or 200 sccm
Substrate temperature 800-1000°C (typically 900°C)
Filament temperature 2300-2400°C (filament current 6½-6¾A)
Filament/substrate distance 4 mm
Deposition time 6 hours

Table 3.1. Standard deposition conditions employed in a hot filament CVD reactor.

 

 

            Molecular beam mass spectrometry has been used to investigate the behaviour of different input source gas mixtures (Table 3.2) as diamond growth precursors in the hot filament CVD reactor, and to identify and understand the gas phase and gas-solid reaction mechanisms in the CVD process, and the changes in the reactions which occur in the presence of these different dopant gases.  Later in this chapter we describe the hot filament CVD apparatus (section 3.2), the molecular beam mass spectrometer design (section 3.3 and 3.4), its characterization, data collection and reduction (section 3.5) and a step-by-step explanation of the procedure used to calculate species concentrations from raw molecular beam mass spectrometric (MBMS) data (section 3.6).

 

Fig.3.1

Fig.3.2
Photograph of the Hot Filament Reactor.

3.2  Hot filament CVD reactor

 

            The deposition chamber is a standard HFCVD reactor consisting of a stainless steel six-way cross (Figures 3.1 and 3.2).  One of the flanges incorporates a glass window for viewing the substrate and filament and through which the pyrometer measures the filament temperature.  This window becomes coated with a yellowy-brown deposit after several deposition runs, which can be removed by cleaning with an IPA-soaked cloth.  Of the other five flanges, one is blank, one connects to the pump, one is attached to the moveable filament/substrate assembly, one is attached to the mass spectrometer, and the final one has feedthroughs for the gas inlet and pressure gauge.  This section is now split into 4 sub-sections to describe the CVD set-up in more detail.  They include the filament, the substrate, gas phase composition, and gas flow system.

 

(a)  The filament

 

            The CVD reactor employs a 0.25 mm thick tantalum filament which is prepared by winding a length of the wire into a coil with a diameter of ~4 mm and length ~1 cm.  Six turns of the wire is the standard used in all deposition experiments as well as molecular beam mass spectrometric studies. The tantalum wire filament is connected to a Variac power source and is heated to ~2300°C (achieved by passing a current of 6½-6¾ A through the filament) during film growth.  This provides the necessary energy to cause dissociation of the process gases.  A two-colour optical pyrometer (Land Infrared), placed in front of the glass window, is used to measure the filament temperature during diamond deposition.  Four cooling fans are positioned around the chamber to dissipate the heat produced by the filament during growth as well as during product distribution measurements.

 

(b)  The substrate

 

            Films were grown exclusively on silicon (100) substrates that had been pre-treated to enhance nucleation.  Pre-treatment involves manually abrading the smooth silicon surface with 1-3 mm diamond powder prior to deposition.  This is achieved by placing the substrate on a suitable surface (a ceramic tile), sandwiching a small amount of diamond powder (a cluster of diameter ca. 1-2 mm) between its surface and a grinding pad (usually another piece of silicon placed upside down) and rotating the grinding pad applying a gentle uniform pressure.  A well abraded substrate usually requires no more than 30 seconds of this procedure.  The substrate is then cleaned with IPA to remove all remaining diamond grit present on the surface.

 

            The substrate is then placed onto the substrate holder and positioned carefully under the filament to ensure even diamond growth.  Heat is supplied to the substrate by a coiled resistive wire which is protected by ceramic beads and embedded in the cement of the substrate holder.  The substrate heater temperature is controlled by an Iso-Tech D.C. power supply.  The maximum recommended voltage that should be applied to the heater without causing damage is 5 volts.  This produces a current flow of ~5A across the resistive wire, and a resultant substrate temperature of ~300-400°C.  The additional heating required to attain the optimum substrate temperature of 900°C is achieved from the heated filament (2300°C) 4 mm above the substrate surface.  A K-type thermocouple in direct contact with the substrate surface is employed to measure the temperature of the substrate during deposition.  The filament/substrate set-up is shown in Figure 3.3.

Fig.3.3

 

(c)  Gas phase composition

 

            The gas composition used in standard growth conditions is a 1% CH4/H2 mixture.  Other C/H containing precursor gases have been examined3.1,3.2 to study the effects on different hydrocarbons on the gas phase chemistry of the CVD process, and this will be discussed in detail in Chapter 4.  In Chapter 5, we replace methane gas with (1) a range of different halomethanes, namely CH4-nCln (n=1-4) in H2, taking care to always maintain a carbon-to-H2 ratio of 1:100 in the gas mixture, and (2) a gas mixture containing 1% CH4 in Cl2/H2, the amount of chlorine varying from 1% to 4% (2-8 atom %).  In Chapter 6 we examine the effects of nitrogen on the CVD process by substituting methane with a variety of C- and/or N-containing precursor gases, and in the final Results chapter (Chapter 7), we investigate the effects of adding phosphine on the growth behaviour of diamond films, using gas mixtures of 1% CH4 with varying amounts of PH3 (1000-5000 ppm).  A summary of the input source gases used is shown in Table 3.2.

 

Chapter #

 

Input source

gases used

Chemical

Formula

Performed

deposition

runs on

Performed

MBMS

analysis on

4

Methane

Ethane

Ethylene

Acetylene

n-Butane

1,3-Butadiene

CH4

C2H63.2

C2H43.2

C2H23.2

n-C4H103.2

C4H63.2

ü

ü

 

 

ü

5

Chloromethane

Dichloromethane

Trichloromethane

Tetrachloromethane

Tetrafluoromethane

Hydrogen chloride

Chlorine (molecule)

CH3Cl

CH2Cl2

CHCl3

CCl4

CF4

 

Cl2

ü

ü

ü

ü

ü

 

ü

ü

ü

ü

ü

ü

calibration only

ü

6

Ammonia

Methylamine

Hydrogen Cyanide

Nitrogen (molecule)

NH3

CH3NH2

HCN

N2

ü

ü

ü

ü

ü

ü

ü

ü

7

Phosphine

PH3

ü

ü

 

Table 3.2. Summary of the input gases used in the present work (except those taken from Reference 3.2). Details of the deposition results (including scanning electron micrographs (SEMs) of the films grown) and MBMS analyses will be discussed in full detail in their respective chapters in the Results section.

 


(d)  Gas flow system

 

            The gas flow system incorporated in the hot filament CVD set-up is shown in Figure 3.4.  The gases are stored in various vessels (See Table 3.3): gas cylinders, lecture bottles  and glass bulbs (gas or liquid sources).  The gases then flow through calibrated mass flow controllers (MFCs - Tylan General), after which they mix in a manifold and then join the main feed pipe leading to the top of the chamber.  A two-stage rotary vacuum pump draws gases through an outlet at the bottom of the chamber.  Both coarse (speedivalve) and fine (needle) valves are incorporated into the pump manifold enabling pressure adjustments in the reaction chamber.  There is also an air inlet to vent the chamber up to atmospheric pressure for filament/substrate replacements.  A direct measure of the process pressure, which is maintained at 20 Torr during the film growth, is made with a capacitance diaphragm gauge (Baratron).

fig.3.4

 

Gas Conversion Factors

 

Mass flow controllers (MFCs) are calibrated by referencing them to a standard (N2 in this case) while gas is flowing.  Because the MFC responds differently to gases other than the reference gas, a Gas Conversion Factor (GCF) is required.3.3  The gas conversion factors listed below are relative to N2 since all the mass flow controllers used (except for the ammonia MFC used only for NH3) are nitrogen MFC’s.  This conversion allows us to determine the actual flow rates of the different gases passing through the mass flow controllers.

 

 

 

 

Input gas

Gas

Source

Regulator

used

MFC type

GCF3.3

Set flow rates for 1 sccm of gas

H2

CH4

C2H63.2

C2H43.2

C2H23.2

n-C4H103.2

C4H63.2

   SGC

   SGC

   LB

   LB

   SGC

   LB

   LB

     B

     B

     B

     B

     B

     B

     B

a

a

a

a

a

a

a

1.01

0.72

0.50

0.60

0.58

0.26

0.32

0.99

1.39

2.00

1.67

1.72

3.85

3.13

CH3Cl (l)

CH2Cl2 (l)

CHCl3 (l)

CCl4 (l)

CF4

HCl

Cl2

   LB

   GB

   GB

   GB

   LB

   LB

   LB

     N

     N

     N

     N

     B

     SSP

     SSP

a

a

a

a

a

b

b

0.63

0.51

0.39

0.31

0.42

1.00

0.86

1.59

1.96

2.56

3.23

2.38

1.00

1.16

NH3 (l)

CH3NH2 (l)

HCN

N2

   LB

   LB

   GB

   SGC

     P

     N

     N

     B

c

c

a

a

  0.74*

0.51

0.76

1.00

1.35

1.96

1.32

1.00

PH3

   LB

     SS

b

0.76

1.32

 

Table 3.3., Detailed information on gas sources used, mass flow controller type and regulators required, together with characteristic gas conversion factors and flow rates.

___________________________________________________________________

Gas source:

(l) = liquid source - uses the vapour pressure of the volatile liquid.

SGC = standard gas cylinder, 2000 psi input pressure, 20 psi output pressure (BOC).

LB = lecture bottle,  maximum output pressure of 20 psi (Argo International Ltd).

GB = glass bulb, used for gases or liquid sources.  Pressures are low (typically < 1 atm ) therefore no regulators required.

 

Regulator used:

B = brass, SS = stainless steel, SSP = stainless steel with facility for purging with N2 or Argon, N = none, P = plastic regulator used with maximum output pressure of 2-6 psi (Matheson Gas Prod.).

 

MFC type:

a = nitrogen FC 260 viton MFC (Tylan General).

b = nitrogen FC 260 kalrez MFC (Tylan General).

c = ammonia FC 260 neoprene MFC (Tylan General).

 

GCF = Gas Conversion Factor relative to nitrogen gas (N2).

* NH3 mass flow controller used, \ GCF and gas flow rate = 1.

sccm = standard cubic centimetres per minute (cm3/min).

___________________________________________________________________

 

 

 

 


Safety Precautions and handling information

            Some of the gases used in the present work require special attention on safety precautions and handling procedures, and these are summarised below in Table 3.4.

 

Special Gases

Safety information and appropriate handling procedures.

HCl

Corrosive - blocks up MFC, and therefore requires a purged regulator and a special MFC for corrosive gases (N2 FC 260 kalrez).  The CVD chamber and MFC must be thoroughly pumped before venting the reactor to atmosphere for filament/substrate replacements.

Cl2

Corrosive - reacts with moisture to produce HCl which again can block up the MFC. This gas also requires a purged regulator and a special MFC for corrosive gases (N2 FC 260 kalrez).

NH3

Toxic and pungent - make sure laboratory is well ventilated. An ammonia MFC is used, and a special plastic regulator is incorporated.

CH3NH2

Blocks up MFC very easily. The MFC must be flushed frequently with N2 or argon before re-use.

HCN

Highly toxic - make sure laboratory is well ventilated and that there is an antidote kit ready if exposed to the gas.

PH3

Highly toxic - The CVD chamber and MFC must be thoroughly pumped.  Pyrolysis of PH3 results in the coating of large amounts of red phosphorus on the CVD chamber walls which requires regular cleaning.

 

Table 3.4. Safety precautions and general awareness of the hazardous gases that have been used in the present work.

3.3  Molecular beam mass spectrometry of diamond CVD

 

            Identifying and understanding both the chemical reactions and the physical transport mechanisms that contribute to the CVD process requires an in-situ diagnostic technique that permits quantitative determination of the concentrations of both free radical and stable species in the gas-phase, with minimal perturbation of the process environment.  Optical spectroscopy3.4 is a widely used technique, but is generally specific to a particular target species.  Gas chromatography3.5,3.6 and mass spectrometric3.7-3.10 studies have the advantages of generality and the fact that many stable species can be analyzed simultaneously, though recombination in the probe used to sample the process gas in these studies precludes detection of reactive gas species.  However, with careful design of the gas sampling system, mass spectrometry can be used to detect free radicals.  Molecular Beam Mass Spectrometry (MBMS) of the diamond growth environment in both hot filament and microwave plasma reactors was pioneered by Hsu and co-workers3.11-3.13, and has enabled quantitative measurements of the concentrations of H atoms and CH3 radicals, as well as stable species like CH4, C2H2, etc.

 

            This chapter gives a full description of the design and construction of a similar MBMS system with which we have obtained quantitative measurements of the gas phase composition in a hot filament CVD reactor under different process conditions.  A detailed account of the data collection and reduction procedures which enables the determination of the mole fractions of both the stable neutral and the free radical species prevalent in the hot-filament CVD process, are also included.


3.4  Molecular beam mass spectrometer design

 

            In designing a sensitive gas-phase analysis technique to optimize quantitative data collection from a reaction chamber such as the hot filament CVD reactor, careful considerations have to be made first about the problems specific to sampling from such a chamber during the CVD process (Table 3.5).

 

 

(1)  Very hot substrate/gas temperatures (900°C & 2000°C respectively),

(2)  Aggressive gases (H atoms, HCl, NH3, etc.),

(3)  Relatively high process pressures (20 Torr),

(4)  By-products (e.g. soot),

(5)  Low concentrations of reactant carbon species,

(6)  Regular venting of the CVD chamber,

(7)  Possibility of performing spatial distribution studies,

(8)  Detection of free radicals.

 

        Table 3.5. Problems specific to sampling from a hot filament CVD reactor.

 

Fig.3.5

The design of the two-stage MBMS has been optimized to sample from a filament assisted CVD process chamber operating at ~20 Torr.  Figure 3.5 shows a schematic diagram of the molecular beam mass spectrometer which has been coupled to the hot filament CVD reactor.

 

Fig.3.6

            Figure 3.6 shows the minor modifications that have been made to our standard process chamber so that the substrate and filament assembly can move in the x,z plane relative to the MBMS sampling orifice, thus enabling spatial distribution studies.3.14  A schematic diagram of the geometry of the MBMS sampling system, shown in Figure 3.7, indicates that in this study we do not sample gas via an orifice in the substrate, which would include the detection of hydrocarbon  species resulting from gas/solid heterogeneous reactions at the substrate surface, as was performed by Hsu.3.11  The sampling cone arrangement is positioned so as to sample the process gas mixture at the same radial distance from the filament as the substrate surface.  Therefore only the gas phase reaction mechanisms are studied.

fig.3.7

 

            Gas from the process chamber is extracted through a 100 mm diameter orifice in a stainless steel sampling cone.  A stainless steel sampling cone was chosen to withstand the high gas temperatures of around 900°C (a quartz sampling cone had been used but cracked at these temperatures), as well as aggressive gases such as H atoms, NH3, and HCl.  As Figure 3.8 shows, a water cooling jacket is incorporated into the apparatus body surrounding the sampling cone to prevent overheating.

Fig.3.8

 

            The sampled gas is then collimated by a 1 mm diameter skimmer and has an unobstructed path to the electron ionization chamber of a HAL/3F PIC 100 quadrupole mass spectrometer (Hiden Analytical, Warrington, England).  This equipment incorporates many features that are beneficial to this system, and will be discussed later in this section.

 

            Since the process pressure is typically 20 Torr and the mass spectrometer can only operate at pressures below 10-6 Torr, differential pumping is required.  Our two stage system utilizes turbomolecular pumps, the pumping speed being 240 l/s in the first stage and 70 l/s in the high vacuum MS chamber.  The pressure in each stage during gas sampling is typically 10-3 Torr and 5´10-7 Torr respectively.

            The signal detected by the mass spectrometer is proportional to the sum of the background gas and the gas introduced by the molecular beam.  In order to obtain the required species concentrations these two components need to be distinguished and the background effects eliminated.  This can be achieved by modulating the molecular beam using a piezoelectrically-driven vibrating reed chopper located between the sampling orifice and the collimating skimmer, some 5 mm from the MS skimmer.  An opto-reflective switch monitors when the chopper is in resonance and measures the resonant frequency (ca. 50 Hz).  Synchronised TTL signals are sent to the MS software which control pulse gating to the MS counter. The width of the signal counting window is variable and is set to 2 ms allowing signal to be collected only when the path of the molecular beam to the ionisation source is fully obstructed or unobstructed.  This enables the background signal component to be eliminated from the total signal, which was of great benefit especially when attempting quantitative analysis for radical species.

 

            A z-translator consisting of an edge-welded bellows assembly allows the distance between the skimmer and the extraction orifice to be varied.  A near linear increase in the detected signal observed as the skimmer-to-extractor distance is reduced from 12 cm to less than 2 cm (See Figure 3.9).  However, if the skimmer is moved yet closer the pumping speed near the orifice is reduced, becoming conductance limited, and causes the background pressure to increase so nullifying any benefit.  Figure 3.9 serves to illustrate the ‘quality’ of our molecular beam by comparing the observed signal vs. distance measurements with that expected for the limiting situation - namely a perfect molecular beam and completely free expansion from a point source.  The system also incorporates an x,y-adjuster in order to align the MS skimmer with the molecular beam, thereby ensuring maximum sensitivity.

Fig.3.9

 

There is a gate valve between the process chamber and the MBMS system so that the MS can be withdrawn and isolated from the process chamber when the latter is vented to atmosphere for filament/substrate replacements.

 

(a)  Choice of sampling cone orifice diameters.

 

Fig.3.10
Fig.3.10. SEM image of the stainless steel sampling cone

            Figure 3.10 shows scanning electron micrographs (SEMs) of the stainless steel sampling cone orifice (~100 mm in diameter) which was made by a spark erosion technique in our mechanical workshop.  The diameter of the extraction orifice as well as the collimating skimmer (1 mm) are selected to match the gas influx from the CVD chamber with the gas throughput of the turbomolecular pumps.  The sizes of the sampling cone orifice and the skimmer hole determine the pumping speeds required.  A large orifice will give large signals, but not only will it be difficult to pump away the sampled gas, but the process environment may be perturbed.  A small hole will produce good signal resolution and since less gas is being sampled through the orifice, it will be easier to pump.  However, detection of gas phase species will be compromised, particularly when the species we wish to analyze are in low concentrations (£1% in hydrogen).  Furthermore, since one of the by-products of the CVD process is soot (at the cooler regions of the chamber), the orifice could easily be blocked, thus further reducing signal intensity.  The sensitivity of our sampling system is mainly restricted by what was affordable for purchasing higher speed turbomolecular pumps, so a compromise had to be reached in terms of the orifice diameters employed in our system.  We do note that the available pumping speeds are close to the minimum that would be sensible, and that this system is not be very efficient for the detection of free radical species such as methyl radicals.

 

(b)  Two stage differential pumping

 

            The important condition to be met when sampling neutral (including free radical) species is the need for the quadrupole to operate in a background pressure of 10-6 to 10-7 Torr.  Thus one or two stages of differential pumping will be required depending on the gas pressure in the region to be sampled.  Since the process pressure during diamond CVD is typically 20 Torr, a two stage differential system is incorporated (See Figure 3.8).  Listed below are the parameters of importance:

 

a1 =      the diameter of the first sampling aperture between the high pressure gas region

            and the intermediate pumping stage (units - cm),

S1 =     the net intermediate stage pumping speed close to the aperture.  This will depend on the pumps used (units - litres/second),

a2 =      the diameter of the second sampling aperture which allows the gas sample

            (hopefully in the form of a molecular beam) to enter the quadrupole (units - cm),

S2 =     the pumping speed in the quadrupole space close to the second aperture (units -

            litres/second).

 

            The conductance C of an aperture (in the molecular flow regime) is given for H2 by the relation3.15:

C= 44.24F litres/second

where F = the area of the aperture (cm2),

            so that for a pressure p0 in the sampled region (20 Torr) the pressure p2 in the quadrupole is given by3.16:

       (3.1)

and the pressure in the intermediate stage by:

      (3.2)

In our MS set up:

a1 = 100 mm (i.e. 10-2 cm), a2 = 1 mm (i.e. 10-1 cm), S1 = 240 l/s and S2 = 70 l/s

 

            Substituting these values into equations (1) & (2) will give values for p2 and p1 of 10-7 Torr, and 10-4 Torr respectively.  The calculated pressures are in reasonable agreement with the measured values (5´10-7 Torr and 10-3 Torr respectively) given to (a) the turbo pumps’ expected inefficiency to pump lighter gas phase species such as hydrogen, (b) pumping restrictions due to the conductance of the pumping box and pipework, and (c) the non-absolute nature of pressures measured by ionisation gauges.

 

(c)  Hiden HAL/3F PIC 100 quadrupole mass spectrometer

 

            Figure 3.11 shows a photograph of the Hiden HAL/3F PIC 100 quadrupole mass spectrometer set up, which has been coupled to the hot filament CVD reactor.  This equipment incorporates many features that are beneficial to this system.  In the following sub-sections a description of its system components will be given, together with the data displays available, in particular the Multiple Ion Detection since this mode was used to obtain all quantitative measurements of the gas phase composition in the CVD process.  Also included is a description of the relevant probe parameters that were used allowing quantitative data to be obtained from the MBMS system.3.17


Figure 3.11. Photograph of our Hiden HAL/3F PIC 100 quadrupole mass spectrometer coupled to the hot filament reactor.

 

SYSTEM COMPONENTS

            Figure 3.12 shows a schematic of the system configuration.  The key components and functions are:

 

(1)  Main Control Unit

            This soft key driven unit is used to control the complete system.  The soft key functions are displayed on screen and are dynamically reassigned with the software menu.

 

(2)  RF Head

            This mounts on the probe and supplies the RF voltages to the quadrupole.


(3)  Probe

            This mounts inside the vacuum system, and contains the Source, Quadrupole mass filter and ion counting detector.

 

Figure 3.12. System block diagram.3.17

 

DATA DISPLAY

            It is possible to display data in many different forms.  The main menu which appears at switch on allows four such display modes.

 

(1)  BAR Mode           is used to obtain a histogram mass spectrum.

(2)  PROFile Mode     is used to acquire an ‘analogue’ representation of the mass spectrum showing peak shape.

(3)  MID Mode           Multiple Ion Detection mode is used for plotting the trend of various masses against time.

(4)  MAP Mode           is used to MAP the effect of a lens on the intensity of a particular ion.


THE MID MENU

 

            Throughout the data collection performed in the present work, the MID mode was used to obtain quantitative measurements of various stable species simultaneously.  This mode allows ions within the mass range of the instrument (0 to 100) to be monitored in any order as required, and intensity data may be presented as numerical values (counts per second) or as a scrolling graphical display.  The signal intensity of any given species was calculated by obtaining an average numerical value over 8 to 12 readings.  A step-by-step method for converting these measured signal intensities into species mole fractions is given in Section 3.6.

 

            The MID mode incorporates many functions (See Ref. 3.17, pp 30 to 34), but one worthy of mentioning is the SETUP menu which allows the per channel MID parameters to be set up.  By setting a particular current channel to monitor mass m ( in the MASS menu), we are able to detect the desired gas phase species.  Key functions apply to the currently selected channel, except for the CHAN key, which sets the current channel selection.  The current channel is the one with the channel number highlighted.  Parameters are displayed in two pages, one of channels 1 to 8, the other of channels 9 to 16, so up to 16 gas phase species can be monitored simultaneously.

 

STATUS AND TRIP MENUS

 

            Apart from the various available display modes, there are also two other important set-up modes in the main menu that we use:


(1)  STATUS menu

 

Fig.3.13

   F1,F2 - These are used to switch filaments on and off.  There are two filaments in the probe system, F1 and F2 (See Figure 3.13).  The F1 key selects and deselects filament 1 operation, and similarly the F2 key selects and deselects filament 2 operation.  When a filament is on the front panel green LED will light and after a few seconds the emission LED will light if the source and filament are in good condition.  An error message will be issued if there is a filament fault.

 

   DEGAS - Entering the time (n minutes) followed by the DEGAS key switches the analyser source into degas mode for n minutes, which can be between 1 and 100 minutes.  This removes any gradual build-up of impurities that may be present on the filament.  This facility is used typically once a month depending on how heavily the mass spectrometer was used.

 

(2)  TRIP menu

 

            This mode is used as an over pressure safety trip, in order to prevent damage to the detector caused by an unexpected surge of gas in the QMS chamber (e.g. air leaks).  A detection limit can be set for any given mass (e.g. H2, N2 and O2) over which the trip will be activated and the filament and the HT will be turned off.  In our set up, we have set values of 1000000, 5000 and 5000 counts per second for mass 2 (H2), mass 28 (N2) and mass 32 (O2) respectively to monitor possible air leaks into the MS chamber.


PROBE PARAMETERS

 

            Figure 3.13 shows a schematic of the probe.  There are a number of parameters which may be grouped in their function under the following headings:

 

(1)  Source                                                       Three parameters

(2)  Quadrupole Mass Analyser                        Two parameters

(3)  Ion detector                                               Two parameters

 

            The parameters may be adjusted in the TUNE menu, either by increment/decrement keys or by direct entry.  The former is probably the more convenient method since the parameters used are fairly consistent, requiring very little adjustment.  Typical values for the following parameters are listed in Table 3.6.

 

(1)  Source group

 

   EMISS parameter: Allows the electron emission current collected on the source cage to be set.  The value is displayed in microamperes (mA) on the screen when the TUNE menu is selected.

 

   ENERGY parameter: Sets the energy of the electrons used to produce ions as they leave the source.  The value is displayed in electron volts (eV) on the screen when the TUNE menu is selected.  This allows the alteration of the electron energy during quantitative analysis thus enabling the detection of reactive species using threshold ionization information.


   CAGE parameter: Sets the voltage on the source cage, which determines the energy of the ions as they leave the source. The value is displayed in volts (V) on the screen when the TUNE menu is selected.

 

(2)  Quadrupole mass filter

 

   RES’N parameter: Adjusts the ability of the mass filter to resolve between adjacent mass peaks at high mass values.  This parameter is displayed on an arbitrary ± 100% scale on the screen when the TUNE menu is selected.

 

   DELTAM parameter: Adjusts the ability of the mass filter to resolve between adjacent mass peaks at low mass values.  This parameter is displayed on an arbitrary ± 100% scale on the screen when the TUNE menu is selected.

 

(3)  Ion detector

 

   SEM parameter: Sets the voltage across the secondary electron multiplier detector and consequently determines the gain of the device.  The current value is displayed in volts (V) on the screen when the TUNE menu is selected.  Figure 3.14 shows a typical multiplier count rate/voltage curve.  As the SEM voltage is increased from a low value the count rate increases rapidly between the threshold voltage Vt, and a plateau voltage Vp.  The SEM value should be set very carefully.  If the voltage is set too high with a high beam current or high gas pressure then permanent damage to the multiplier may result.  It is recommended that the multiplier voltage should be restricted to the start of the plateau region (Vp), i.e. operating at about 100 to 200 volts above the plateau voltage.

   DISCRM parameter: Sets a voltage level on a comparator connected to the pulse amplifier.  If the pulse height is more negative than the discriminator level then the pulse is counted.  Its current value is displayed on an arbitrary scale of ± 100% on the screen when the TUNE menu is selected.  The discriminator level should be set below the most negative level of the background noise to prevent invalid counts, as shown in Figure 3.15.

 

Figure 3.14. Schematic showing the multiplier voltage curve.3.17


Figure 3.15. An illustration showing where the discrimination level should be set in order to eliminate any unwanted signal due to background noise.3.17

 

            It is important to note that the multiplier life time depends on the total charge collected by the multiplier and the background pressure. This has some implications as far as multiplier use is concerned:

 

(1)  It is wise to restrict the multiplier voltage to the start of the plateau region.  This minimises the operating gain (See Figure 3.16) and the collected charge for each count output.

 

(2)  It is not advisable to allow the probe to acquire high count rates for long periods of time unnecessarily.  The high voltage supplies can be disabled (and enabled) using the HT ON/OFF key when data is not actually being acquired.

 

(3)  The multiplier should be operated with a low background pressure (below 10-6 Torr) if possible.  The operating pressures used throughout the MBMS analysis are kept below 3´10-6 Torr.

 

Figure 3.16. A schematic showing a typical multiplier gain curve.  It is recommended that the multiplier voltage should be restricted to the start of the plateau region, Vp (See Figure 3.14).3.17

 

TUNING

 

            The precise values for the operating parameters depend specifically on the spectrometer and its application.  Typical settings used for each probe type are shown in Table 3.6.


 

DISCRM

0 to -20%

DELTAM

0 to -20%

RES’N

0 to -40%

SEM

2500 to 2650 V

ENERGY

vary depending on species being monitored.

CAGE

3.0 V

EMISS

120 to 140 mA

 

Table 3.6. Typical operating parameters of our MBMS system.

 

3.5  Characterization of the mass spectrometer

 

(a)  Energy scale calibration

 

            Since the actual electron energy in the ionization source of the MS is not necessarily the same as the potential applied to the cathode filament, the energy scale needs to be calibrated.  An accurate value for this electron energy is essential as precise threshold ionization energies need to be measured (See Section 3.5 (c) ).

 

The true MS cathode voltage can be accurately determined by measuring the ionization potential (IP) of Ar and correcting to its literature value of 15.76 eV.3.18  Measurements of the cathode voltage against argon signal were performed for each filament (See Figures 3.17 and 3.18), and the linear interpolation3.19 method was used to determine the threshold ionization potentials.  In this method, the linear section of the ionization efficiency curve (a curve produced when plotting the signal intensity of a given ion versus electron energy) is extrapolated back to zero signal intensity.  The intersection with the energy axis is the ionization potential of the gas being studied.  This linear interpolation method is not a perfect means of measurement, due to non-linearity of the graph at lower voltages.  This effect is caused by variation in the electron energy distribution interacting with the ionized species.  Although the average energy emitted by the ioniser may be insufficient to cause ionization of the species, a small spread in the energy distribution would induce some ionization.  We note that this small spread in the electron kinetic energy distribution (FWHM ~1 eV) may affect the accuracy of the linear extrapolation method.  Inspection of Figure 3.17 reveals that filament #1 produces an argon ionization potential of 16 eV compared to the literature value of 15.76 eV.  For filament #2 the observed argon IP was 18.4 eV, with an energy scale correction of 2.64 eV ± 0.3 eV.  For continuity filament #1 was used for all of the subsequent readings, and all cathode electron energies have had the calibration correction applied and are accurate to within 0.5 eV.

Fig.3.17

 

Fig.3.18

(b)  MS calibration

 

            When measuring the signal for species with a particular mass-to-charge (m/e) ratio it is important to eliminate or at least minimize interference from unwanted ions with the same (m/e) or those arising from fragmentation of other ionic species.  This can be achieved if it is possible to measure the signal of the species of interest at an electron ionization energy just below the ionization threshold of the interfering species.  For example, we can detect the signal for C2H4 (m/e = 28, IP = 10.51 eV) using an electron energy of 13.5 eV, so minimizing signal interference from CO (IP = 14.0 eV) and N2 (IP = 15.55 eV).  However, in cases where interference from fragmentation is unavoidable (e.g. detection of C2H4 in the presence of C2H6) corrections to the signals have to be made using observed fragmentation patterns.

 

            The relationship between the signal intensity, I, of a given species, i, measured at a given electron energy, Ei, and its mole fraction, Xi, is given by

Ii (Ei) = Si Xi                                                      (3.3)

where the sensitivity factor, Si, depends on the ionization cross section of i and the MS gas sampling efficiency, which may vary for different gas species and the local temperature, pressure and composition of the gas sample.  The concentrations of the stable species are determined by direct room temperature calibration of mixtures of known composition ensuring that, for each species monitored, we use the same user-selected electron ionization energy in both calibration and data collection cycles and that the total process pressure remains constant.  However, we find that an additional temperature dependent correction also needs to be made (see Section 3.5(d) and 3.5 (e) ).

 

(c)  Mass discrimination

 

            Mass discrimination effects in a molecular beam system on MBMS sampling efficiency are well known.3.11  These effects arise with a mixture of gases of varying atomic or molecular weights, for which it is observed that downstream the centre of the molecular beam has a disproportionately large concentration of higher mass moieties.  This effect is a possible consequence of pressure driven diffusion or Mach focusing.3.11  Large quantitative errors could be introduced by mass discrimination when measuring gases of varying compositions.

            This simple calibration procedure described above is justified only for small concentrations of input gas species (<1%) in hydrogen on the grounds that, for heavy species diluted in a large excess of hydrogen, the transport of the process gas through the sampling orifice and subsequent formation of a supersonic molecular beam are dominated by the mass transport properties of the hydrogen.  However, problems arise with this calibration procedure when the concentration of a particularly heavy species, such as Cl2, in the gas mixture is significantly greater than ~1%.  This is illustrated in Figure 3.19 which shows the attenuation of the CH4 signal (at room temperature) when different amounts of chlorine are added to a 1% CH4 in H2 mixture, an example chosen because of its relevance to the study of chlorine assisted CVD of diamond.  In these cases calibration was carried out using known amounts of the target gas of interest (e.g. CH4, C2H2, HCl,..) diluted in the appropriate H2/Cl2 mixture.

 

Figure 3.19. Variation of measured CH4+ signal as a function of gas composition.  The signal for 1% CH4 in H2 is attenuated as the chlorine concentration in the gas mixture increases.

 

Figure 3.20. Variation of measured Ne+ signal as a function of gas composition.  Similar signal attenuation is observed for 1% Ne in H2 as the chlorine concentration in the gas mixture increases.

 

This signal attenuation could, in principle, be due to CH4 reacting with the added Cl2 at room temperature or be a consequence of mass discrimination effects in the molecular beam mentioned above.  Evidence to support the latter explanation, i.e. that the light CH4 molecules are preferentially excluded from the centre of the beam by the heavy Cl2 molecules comes from the equivalent experiments in which we replace the 1% CH4 by 1% Ne in H2 (See Figure 3.20).  The Ne signal is observed to fall off in a similar way.

 

(d)  Temperature dependence of MS sampling efficiency

 

            Using pure H2 at 20 Torr, the variation of IH2 as a function of the local temperature, T, of the gas being sampled, as measured using a K-type thermocouple placed adjacent to the sampling orifice, shows a T -0.6 dependence (See Figure 3.21).  Similar experiments using pure samples of He (m/e = 4), Ne (m/e = 20) and Ar (m/e = 40) reveal that Si shows the same temperature dependence for all these pure gases.  In our results we assume a similar temperature dependence to the mass spectrometer sampling efficiency for all the hydrocarbon species of interest, and correct accordingly (by reference to the attenuation of the measured H2 signal).  This assumption is again justified on the grounds that, for small concentrations of heavy species in hydrogen, the mass transport properties are dominated by those of the hydrogen.

 

Figure 3.21. Variation of the detected signal as a function of the local temperature of the gas being sampled. For pure gases (H2 and Ne shown) the fit to the experimental data show a near T-0.6 dependence. For 2% Ar in H2 the Ar signal shows a T-1.6 dependence.


(e)  Thermal diffusion effects

 

            For a two component gas mixture the temperature dependence of the detected ion signals is different from that measured for a pure gas.  For example, Figure 3.21 shows that the Ar+ signal measured for a two component, 2% Ar in H2, gas mixture has a much greater temperature dependence (~-1.6) than that for a pure gas.  This indicates an additional thermal effect for a gas mixture whereby a temperature gradient induces preferential diffusion of the heavier component in the mixture away from the higher temperature filament/sampling orifice region.  This thermal diffusion (also known as the Soret effect3.20) has a major effect on the total concentration of any gas phase species measured in the region of the hot filament.  In the following chapters (4-7) the absolute species concentrations measured 4 mm from the filament are presented with no additional adjustment being made for the depletion of C/Cl/N/P-containing species due to thermal diffusion, except when comparing results from the CHEMKIN computer simulations (See Section 8).

 

(f)  Dissociation patterns

 

            As previously mentioned, in order to minimize interference arising from fragmentation of other ionic species, corrections have to be made to the signals using measured fragmentation patterns.  Fragment ions arise as a result of parent dissociation taking place in the ionization region of the mass spectrometer.  Table 3.7 below shows the fragments observed for various input gas species:


 


Gas introduced

Species observed in the mass spectrometer

CH4

CH4+, CH3+, CH2+, CH+

C2H2

C2H2+, C2H+, CH+

PH3

PH3+, PH2+, PH+, P+

 

Table 3.7. Species observed following electron bombardment of various input gases.  Full examples of these cracking patterns will be shown in the Results Section.

 

            These dissociation patterns are obtained by plotting a graph of the signal intensity of each species produced as a function of electron energy.  The ionisation potential (IP) of the parent ion and the appearance potentials (AP) of the species observed following electron bombardment can therefore be determined by linear interpolation methods described earlier (See section 3.3 (a) ).  The values obtained are then compared with literature values.3.21 Appropriate corrections are then made for signal contributions due to the fragmented species.

 

(g)  Ionization cross sections and potentials

 

            The signal intensity of a given species, measured at a particular electron energy, is dependent upon the efficiency with which it can be ionized.  This efficiency is expressed as a factor, the ionization cross section, which relates to the size and conformation of the species, as it is associated with the area presented to the incoming ionizing electron.

 

            Also related to the ionization cross section of a species is its ionization potential (IP).  For hydrogen atoms, the IP(H+) is the threshold energy required to produce H+ ions by direct electron impact,

 

IP(H+):     e- + H ® H+ + 2e-                                                    (3.4)

 

            This must be distinguished from the H+ ions produced via dissociative ionization; the appearance potential (AP) of H+,

 

AP(H+):     e- + H2 ® H + H+ + 2e-                                             (3.5)

 

whereby hydrogen molecules are split to produce both a hydrogen atom and cation.  Often the appearance potential of a given species is greater than its ionization potential.  For hydrogen, the values for IP(H+) and AP(H+) are 13.6 eV and 18.0 eV respectively.3.22  The difference between these values is typically the bond strength of the H2 molecule.  In general radical species have a lower ionization potential (IP) than neutral species, and hence a larger ionization cross section at any given electron energy.

 

(h)  Detection of radical species

 

            To date, models of the reaction mechanisms by which different hydrocarbon precursors are able to produce high quality CVD diamond have focused on methyl (CH3) radicals3.23,3.24 and/or acetylene (C2H2)3.25,3.26 as the active carbon species in the growth process.  Detection of low concentrations of free radicals (e.g. CH3 in a large excess of CH4 or CH3Cl) requires the use of the threshold ionization technique3.27 which distinguishes ions generated by direct electron impact of the radicals from those generated by dissociative ionization of the parent molecule (See section 3.5 (g) ).  Application of this technique for the detection of methyl radicals, for example, requires the electron energy of the MS ionization source to be maintained well above the ionization threshold of the CH3 radicals (IP = 9.84 eV), in order to maximize the detection sensitivity, yet sufficiently below the appearance potential (AP) of CH3+ from the dissociative ionization of CH4 (14.3 eV).  Signal interference from the parent molecule is thereby minimized.  In practice, all CH3+ signal measurements were made using an ioniser voltage centred at 13.5 eV, which resulted in a limited amount of fragmentation of the parent species but was readily corrected using measured fragmentation data.

 

(i)  Procedure for obtaining quantitative measurements of CH3 radicals

 

            In order to quantify the radical species we need to distinguish between the beam and background components of the MS signal since most of the radical species in the background component do not survive to be detected.  We achieve this by modulating the beam as described in section 3.4.  The dependence of the CH4+ signal with chopper delay (See Figure 3.22) shows that ~35% of the total signal comes from species in the molecular beam.


Figure 3.22. Illustrative diagram showing how the measured CH4+ signal varies as a function of the chopper delay.  This indicates that only ~35% of the total signal comes from species in the molecular beam.  The smooth curve through the points is included to guide the eye.

 

            Ideally we would eliminate the background components in all the measurements, thereby enabling the signal intensities of each species to be directly related to their ionization cross sections.  However, the restricted mechanical pumping speed of the MBMS system precludes such a procedure because of limitations in sensitivity of the system when modulating the beam.  Instead, in order to correct for the destruction of CH3 in the background we measured the proportion of the CH4 in the beam and background components (See Figure 3.22) and ratio the measured CH3 signal assuming that the CH3 radicals are similarly partitioned between beam and background but that none of the radical species in the background gas survive to the detector; our ‘corrected’ CH3 signal is thus an upper limit.

 

            Having corrected for loss of CH3 in the background we can now directly calibrate for CH3 by using measured ionization cross sections, Q, for methane3.28 and methyl radicals,3.29 at the respective electron energies used for detection, and the previously measured relationship between the CH4 signal intensity and its mole fraction, via

       (3.6)

 

In such a calibration procedure we assume that the mass discrimination factors for CH4 and CH3 in excess H2 on formation of the beam are equal.


3.7  References

3.1 C.A. Rego, P.W. May, C.R. Henderson, M.N.R. Ashfold, K.N. Rosser and N.M. Everitt, Diamond and Relat. Mater., 4, 770 (1995).

3.2  C.R. Henderson, BSc. 3rd Year Project, University of Bristol (1994).

3.3 Tylan General Instruction Manual (1991).

3.4  F.G. Celii and J.E. Butler, Ann. Rev. Phys. Chem., 42, 643 (1991).

3.5 C-H. Wu, M.A. Tamor, T.J. Potter and E.W. Kaiser, J. Appl. Phys., 68, 4825 (1990).

3.6 S.J. Harris and A.M. Weiner, J. Appl. Phys., 67, 6520 (1990).

3.7 S.J. Harris, A.M. Weiner and T.A. Perry, Appl. Phys. Lett., 53, 1605 (1989).

3.8 S.J. Harris, D.N. Belton, A.M. Weiner and S.J. Schmieg, J. Appl. Phys., 66 5353 (1989).

3.9 R. Beckmann, B. Sobisch and W. Kulisch, in Diamond Materials, The Electrochemical Society proceedings Series, Pennington, NJ, Vol. 93-17, p.1026 (1993).

3.10 I. Schmidt, C. Benndorf and P. Joeris, Diamond and Relat. Mater., 4, 725 (1995).

3.11 W.L. Hsu and D.M. Tung, Rev. Sci. Instrum., 63, 4138 (1992).

3.12 W.L. Hsu, J. Appl. Phys., 72, 3102 (1992).

3.13 W.L. Hsu, M.C. McMaster, M.E. Coltrin and D.S. Dandy, Jpn. J. Appl. Phys., 33, 2231 (1994).

3.14 C.A. Rego, P.W. May, C.R. Henderson, M.N.R. Ashfold, K.N. Rosser and    N.M.Everitt, in New Diamond Science and Technology, MYU, Tokyo, p.485 (1994).

3.15H S. Dushman Scientific Foundations of Vacuum Technique, edited by J.M

Lafferty (John Wiley & Sons, New York, 1962).

3.16Plasma Monitoring with Hiden Analytical Diagnostic Equipment, Hiden Analytical Warrington, England.

3.17Hiden Analytical Operating Manual, Hal II Ion Counting RGA Systems,          manual release 1.3, (1994).

3.18 W. Paul and M. Raether, Z. Physic, 140, 262 (1955).

3.19 R.W. Kiser, Introduction to Mass Spectrometry and Its Applications, Prentice- Hall, INC. Canada, 1965.

3.20 R.B. Bird, W.E. Stewart and E.N. Lightfoot, Transport Phenomena, John Wiley and Sons, New York, 1960, p.567.

3.21 F.H. Field and J.L. Franklin, Electron Impact Phenomena an the properties of

gaseous ions, Academic Press, New York, 1957

3.22 J. Berkowitz, Photoabsorption, Photoionization and Photoelectron Spectroscopy, Academic Press, London, 1977.

3.23M. Tsuda, M. Nakajima and S. Okinawa, J. Am. Chem. Soc., 108, 5780 (1986).

3.24 S.J. Harris, Appl. Phys. Lett., 56, 2298 (1990).

3.25 M. Frenklach and K.E. Spear, J. Mater. Res., 3, 133 (1988).

3.26 S. Skokov, B. Weiner and M. Frenklach, J. Phys. Chem., 98, 8 (1994).

3.27 G.C. Eltenton, J. Chem. Phys., 15, 455 (1947).

3.28 H. Chatham, D. Hils, R. Robertson and A. Gallagher, J. Chem. Phys., 81, 1770 (1984).

3.29 D.P. Wang, L.C. Lee and S.K. Srivastava, J. Chem. Phys., 152, 513 (1988).