GB2482680A - A method for determining the presence or absence of an element in a sample - Google Patents

Abstract

The method comprises receiving data generated from the sample and obtaining stored data relating to the element, the stored data comprising at least one mandatory reference data point and at least one optional reference data point. A relationship between the stored data and the received data is determined and the presence of the element in the sample is determined if the mandatory reference data point has a predetermined relationship with the received data. The determination is independent of a relationship between the optional reference data point and the received data generated from the sample. The method may produce a measure of the confidence that an element is present in the sample.                                                                   The method may also make use of parameters of an apparatus. The data may be spectral data indicating the elemental composition of the data, and the data points may be peaks in the spectral data.

Description

SPECTRAL ANALYSIS

The present invention relates to methods and apparatus suitable for analysing spectra. More specifically, but not exclusively, the present invention relates to methods and apparatus for analysing spectra indicating the elemental composition of a specimen.

Elemental analysis of a material is used in many industries where the purity of a material is important. For example, one such area where elemental analysis is used is in semiconductor wafer fabrication where it is important to detect any defects present on the surface of wafers. Elemental analysis provides information on the elemental composition of detected defects on the surface of a wafer which can be used to diagnose and address the cause of the defects.

One technique for elemental analysis of a specimen is Auger Electron Spectroscopy.

Auger Electron Spectroscopy uses the Auger effect whereby upon application of a photon or electron beam to a specimen, electrons are emitted from the material.

Each element may emit electrons having a variety of different energies and the electrons emitted by a particular element form a spectrum unique to that element.

The energies of electrons emitted from a specimen of interest can be processed to generate a spectrum of the energies and the spectrum can be analysed to determine elements present in the specimen.

A specimen to be analysed will usually contain more than one element and each element will emit electrons with particular energies. The energies of electrons that are emitted from all of the elements in a specimen will form a spectrum determined by the combination of elements in the specimen that is analysed. One proposed a method of analysing a specimen based upon known elemental spectra is to simulate spectra for different combinations of elements. A spectrum generated from a particular specimen may then be compared to simulated spectra to try to establish a match. The total possible number of spectra that may be generated from a specimen varies dependent upon the size of the set of possible candidate elements that may "S be present in the specimen together with the actual or assumed maximum number of S. * *. elements that are present in the specimen.

Table 1 shows the number of possible spectra that may be generated from a specimen, for a range of one to ten elements present in the specimen and based upon 80 possible candidate elements and 50 possible candidate elements. It can be seen from Table 1 that as the number of elements in the specimen increases, so too does the number of possible spectra that may be produced. In general, the number of elements in a specimen will not be known prior to analysis and so the total number of possible spectra will be determined by the sum of spectra produced for each number of elements in the specimen up to the assumed maximum number of possible elements for a known candidate set size, as shown in the bottom row of columns 2 and 3 for samples containing up to 10 elements. It can be seen that a very large number of spectra may be generated, and generation of a simulated spectra for each of the possible combinations, together with analysis of the simulated spectra with a spectrum generated from a specimen is therefore computationally expensive.

No. of elements in Number of spectra Number of spectra specimen (80 possible elements) (50 possible elements) 1 80 50 2 3160 1225 3 82160 19600 4 1.58458E+06 2.30300E+05 2.40400E+07 2.11876E+06 6 3.00500E+08 1.58907E+07 7 3.17672E+09 9.98844E+07 8 2.89875E+10 5.36879E+08 9 2.31900E+11 2.50543E+09 1.64649E+12 1.02723E+10 Total 1.91088E+12 1.34327E+10

TABLE I *S.. * * ****

Additionally, the numbers of spectra shown in Table 1 do not allow for the variations in spectra that may be caused by different relative quantities of elements and different chemical states of elements. For some combinations of elements, particular * ** spectral peaks may be either stronger or weaker than would be expected due to chemical and matrix effects of the combinations of elements. Different topographies of a surface will also have a strong influence on a spectrum generated from the surface due to differing electron collisions in the surface layer which affect the electrons emitted from the sample. The effects of the above factors are complex and difficult to predict.

Other methods for the identification of the elemental composition of a specimen based upon spectral data have been proposed. For example, it has been proposed that experimentally obtained data can be processed with reference to data relating to particular elements of interest. However such methods are disadvantageous in that they fail to take into account factors such as matrix effects and variations in topography on spectral data generated when combinations of elements are present in a specimen, which can lead to combination and overlap of spectral peaks that would be expected but difficult to predict, based upon data relating to a single element of interest. As such, accurate determination of the elemental composition of a sample using such techniques is difficult.

It is an object of some embodiments of the present invention to obviate or mitigate at least some of the problems set out above.

According to a first aspect of the invention there is provided a method for determining the presence or absence of an element in a sample. The method comprises receiving data generated from the sample and obtaining stored data relating to the element, the stored data comprising at least one mandatory reference data point and at least one optional reference data point. A relationship is determined between the stored data and the received data and the presence of the element in the sample is determined if the at least one mandatory reference data point has a predetermined * relationship with the received data, the determining being independent of any *.I*** * relationship between the at least one optional reference data point and the received data generated from the sample. ** **

* In this way, a priori knowledge of data points associated with an element (i.e. the stored data) can be used to determine the presence or absence of the element in a sample. Some data points associated with an element may not always be present in data generated from a sample containing the element under certain experimental conditions, whereas other data points may always be present in a sample. For example, some peaks may not be detected where there is a poor signal to noise ratio or where there is a large amount of background noise due to, for example, backscattered electrons originating from the sample material. Taking into account the relative likelihood of a data point associated with an element being present in a sample generated from the element can therefore allow some data points to be designated as mandatory and others to be designated as optional, thereby allowing a determination that an element is not present in a sample, where in fact it is present, to be avoided.

The method may further comprise determining a confidence of the presence of the element in the sample based upon a relationship between the at least one optional reference data point and the received data. The confidence may be based upon, for example, scores associated with optional reference data points and gap errors associated with optional data points.

That is, whilst some data points may not be required to be present in data generated from a sample, those points may still be useful in determining a confidence that the element is present in the sample. It will be appreciated that the confidence may also be based upon the mandatory data points.

The received data may comprise a plurality of received data points and determining a relationship between the at least one mandatory reference data point and the received data may comprise performing an identification operation between each of the at least one mandatory reference data points and the plurality of received data points. Each of the at least one mandatory reference data points and each of the received data points may have an associated energy value and the identification operation may be based upon the associated energy values.

*...** * * The identification operation may identify a particular received data point in the data if **.* and only if the energy value associated with the particular received data point is within a predetermined tolerance of an energy value associated with at least one mandatory reference data point. ** * * ** * S.

The element may be determined not to be present in the sample if the identification operation does not identify a received data point for each of the at least one mandatory reference data points.

The identification operation may identify a particular received data point in the data if and only if the particular received data point has an associated signal to background ratio greater than a predetermined value. For example, a data point may be considered present in the received data if it has an associated signal to background ratio greater than a predetermined minimum.

The at least one mandatory reference data point and the at least one optional reference data point associated with the element may be determined according to a ---training process. That is, data points associated with an element that are mandatory --or optional may be determined by processing known samples to generate data, and identifying those data points associated with the element that always appear in the generated data and those data points that do not always appear in the generated data.

Alternatively, the mandatory and optional data points for a particular data point may be determined from known data relating to the element.

The stored data relating to the element may comprise a plurality of data points and the method may further comprise obtaining at least one parameter of an apparatus used to generate the data from the sample and identifying the at least one mandatory reference data point and the at least one optional reference data point based upon the at least one parameter. That is, different parameters of an apparatus used to generate data from a sample may result in different data points being mandatory and * ** *** * optional. The mandatory and optional data points for an element may be determined according to a training process in which data is generated from a sample containing S.SS the element whilst different parameters are applied to the apparatus.

According to a second aspect of the invention there is provided a method for *:*. generating a score for a feature of data generated from a sample. The method comprises receiving the data generated from the sample and identifying the feature in the received data, the feature having a plurality of associated data points in the received data. At least two of the plurality of data points associated with the feature are processed to determine at least one property of the feature and the score for the feature is generated based upon the at least one property.

Prior art methods are concerned with a maximum point associated with a particular feature whereas the second aspect of the invention is concerned with considering properties of features relating to properties of points associated with the feature.

Where it is indicated that the feature has a plurality of associated data points, it will be appreciated that the feature may be defined by those data points. That is, the feature may be made up of its plurality of associated data points.

The received data may be defined by data points each defining a value on a first axis and a value on a second axis, the plurality of data points in the received data associated with a particular feature having values on the first axis such that the data points associated with the feature are adjacent to one another. For example, values on the first axis may be associated with channels on a charged particle energy analyser, and data points associated with the feature may be associated with values at a plurality of consecutive channels. The feature may have a shape defined by values of the plurality of associated data points.

The at least one property of each of the plurality of features may be selected from the group consisting of: an area of the feature, a width of the feature, a standard deviation of the feature, a signal to background ratio of the feature and a shape factor of the feature.

* * Two of the plurality of data points associated with each feature may be associated * ...S* * with local minima in the received data and the at least one property of each of the plurality of features may be based upon the local minima. Each of the plurality of data ** ** points may have an associated value, and the at least one property of each of the plurality of features may be determined based upon a difference between the values associated with the data points associated with local minima. For example, the local *:*. minima may be points of the plurality of points associated with the feature having a maximum and a minimum associated value (such as an energy value) and the at least one property may indicate a width of the feature.

The at least one property for a particular feature may be based upon all of the plurality of data points associated with the particular feature.

The method may further comprise generating background subtraction data from the received data, the background subtraction data comprising a plurality of background subtraction data points and determining a corresponding one of the background subtraction data points for each of the features. The at least one property of each of the plurality of features may be further based upon the corresponding one of the k.,nL#'nrn, in,4 * .1-dr n+b,n.4 + nnifl+c. C,-.r s, mnIc rwnn,t: ,-J +1, fn.,+ ,re +kn JLII 1J L1IJLI.,II'.JI I L4LQ JJII It.. I.11 I I1Jl, I.JJI L) ..JI LI iL IQLIJI III LI I received data may initially be determined and the at least one property may be determined by adjusting the initially determined property of the feature based upon --the corresponding one of the background subtraction data points. It has been found that such an adjustment provides a property for a feature which better reflects the feature where the feature lies close to other features.

Each of the plurality of features may have an associated value and determining a corresponding one of the background subtraction data points for a particular feature may be based upon the values associated with the features. For example the associated value may be a value associated with a maximum associated with the feature.

Each of the background subtraction data points may have a plurality of associated data points, each associated data point having an associated value, and determining a corresponding one of the background subtraction data points for a particular feature may comprise determining a background subtraction data point having at least one data point having an associated value greater than the value associated * S....

* S with the particular feature and at least one data point having an associated value smaller than the value associated with the particular feature. That is, a corresponding S. I background subtraction data point for a particular feature may be a background subtraction data point that encloses a maximum associated with the particular feature. In such a case, the associated values of the data points associated with the * . background subtraction data points may be energy values.

The method may further comprise obtaining stored data relating to an element, the stored data comprising a plurality of data points determining whether a relationship between the stored data and the feature is satisfied and, if the relationship is satisfied, generating a quantitative indication of the presence or absence of the element in the sample based upon the score associated with the feature. For example, the quantitative indication may be indicative of a confidence of whether the element is present in the sample.

According to a third aspect of the invention there is provided a method of determining a subset of a piurality of features in data generated from a sample, the method comprising generating a respective score for each of the plurality of features according to the second aspect of the invention and determining the subset of the plurality of features in the data based upon the generated scores.

The received data may be background subtraction data generated from data obtained from the sample. The subset of a plurality of features generated according to the third aspect of the invention may, for example, be used to determine a subset of a plurality of elements that are most likely to be present in a sample.

According to a fourth aspect of the invention there is provided a method of determining chemical shift data associated with an element based upon data generated from a sample. The method comprises generating a respective score for each of a plurality of features in the data generated from the sample according to the second aspect of the invention and selecting one of the plurality of features based upon the generated scores. Stored data relating to the element is obtained, the stored data comprising at least one data point associated with the element and the chemical shift data associated with the element is determined based upon the * ***** * selected one of the features and the at least one data point associated with the element. *..*

* Each of the plurality of features and data points may have an associated value and the chemical shift data may be determined based upon a difference between the *:*. value associated with the selected one of the plurality of features and the value associated with the at least one data point associated with the element.

The at least one data point associated with the element may be selected based upon its associated value, for example the at least one data point may be a data point having an associated value closest to the value associated with the selected one of the plurality of features.

The selected one of the p'urality of features may be selected such that it has a maximum associated score.

The chemical shift data may indicate a difference between expected values associated with data points associated with the element and values for corresponding data points associated with the element in the data generated from the sample.

According to a fifth aspect of the invention there is provided a method for determining a subset of elements from a plurality of elements, the subset of elements indicating elements of the plurality of elements most likely to be present in a sample. The method comprises obtaining stored data relating to each of the plurality of elements and receiving data generated from the sample, the received data comprising a plurality of data points, each data point having an associated value. The received data is processed a plurality of times, each processing being based upon a respective value for a parameter, to generate respective sets of intermediate data points, each intermediate data point having an associated value. The respective sets of intermediate data points are processed to determine relationships between intermediate data points in the respective sets of intermediate data points and a set of intermediate data points is determined based upon the determined relationships.

The subset of elements is determined based upon the determined set of intermediate * data points and the stored data relating to each of the plurality of elements.

* ** S.. * *

Determining relationships between intermediate data points, and using the *. .5 determined relationships to identify a set of intermediate data points in this way has been found to provide a set of intermediate data points that are most useful in determining a subset of elements from a plurality of elements. That is, data generated from a sample generally contains a large number of data points, and some of the data points are not useful in determining a subset of elements from a plurality of elements. The fifth aspect of the invention provides a way of identifying those data points that are useful in determining a subset of elements from a plurality of elements and therefore not only improves efficiency of the method, since unnecessary data points are not considered, but also provides improved determining of the subset of elements from the plurality of elements since false positive indications from the unnecessary data points are avoided.

The value associated with each of the data points may be an energy value.

The determined set of intermediate data points may comprise intermediate data points selected from at least one of the respective sets of intermediate data points.

The determined set of intermediate data points may be a set of intermediate data points having an energy value less than a predetermined threshold. Identification of -data points with a low energy value that are useful in determining a subset of elements from a plurality of elements is known to be particularly problematic, and the fifth aspect of the invention is particularly effective for identifying useful data points in the low energy region.

Processing the received data to generate each of the respective sets of intermediate data points may comprise performing a respective background subtraction process on the received data.

The received data may comprise a plurality of received data points and performing the background subtraction process may comprise processing pairs of the received data points, each pair of received data points having an associated difference, the difference being based upon the respective value for the parameter.

* ...* * Processing the respective sets of intermediate data points to determine relationships between intermediate data points in the respective sets of intermediate data points ** ** may comprise identifying corresponding data points in the sets of intermediate data points. S.. * . I...

Each of the intermediate data points may be a characteristic associated with a feature of the received data, the characteristic comprising a plurality of associated data points, and identifying corresponding data points in the sets of intermediate data points comprises determining whether a predetermined relationship between pluralities of data points associated with the characteristics is satisfied.

Each of the plurality of data points associated with a feature may have an associated value, each characteristic having an associated maximum data point having a relatively high associated va'ue, and an associated minimum data point having a relatively low associated value. Determining whether a predetermined relationship between pluralities of data points associated with characteristics is satisfied may comprise identifying a first characteristic and a second characteristic as corresponding characteristics if values for a maximum data point and a minimum data point for the first characteristic lie between values for a maximum data point and a minimum data point for the second characteristic. The first and second characteristics may be peaks defined by data points, for example peaks defined in data points of data generated by a background subtraction process. The maximum data point and minimum data point may, for example, each be points associated with points having an intensity value of zero in the data and may therefore define the extent (or width) of the characteristic..

Each of the intermediate data points may be a characteristic associated with a feature of the received data, the characteristic comprising a plurality of associated data points, each of the plurality of associated data points having an associated intensity and the relationships between data points in the respective sets of intermediate data points may be based upon a value of a data point associated with a feature having an associated intensity which is maximal. For example, the features may be peaks of the data and the relationships between data points may be based * upon whether a peak encloses a point associated with a feature that has a maximum *S*.* * * * intensity. I*I** * *

Determining the subset of elements may be based upon at least one data point ** ** having an energy value greater than the predetermined threshold. For example, the at least one data point having an energy value greater than the predetermined threshold may be determined according to the third aspect of the invention. S. S

S S S S.

According to a further aspect of the invention there is provided a method of generating a quantitative indication as to whether an element is present in a sample.

The method comprises obtaining stored data relating to the element, the stored data comprising a number of data points associated with the element and receiving data generated from the sample. The quantitative indication is generated based upon the number of data points associated with the element and a relationship between the received data and the stored data associated with the element.

By taking into account a number of data points associated with an element, a better inter-element comparison is achieved. For example, where a particular element has a relatively large number of associated data points, there are a relatively large number of data points to be identified in a sample in order that a strong indication of the presence of the element is provided. Conversely where an element has a relatively small number of associated data points, only a small number of data points need be identified in a sample in order that a strong indication of the presence of the element is provided, and a strong indication of an element with a small number of associated data points is therefore more likely. By taking into account the number of data points associated with the element, a more objective indication of the presence of an element is provided. That is, this aspect of the present invention allows for some degree of inter-element normalisation. The quantitative indication may be indicative of a confidence of whether the element is present in the sample.

The received data may comprise a plurality of received data points, and the relationship between the received data and the stored data may be determined by performing an identification operation between each of the data points associated with the element and the received data points.

Each of the received data points and each of the data points associated with the * : element may have an associated energy value, and the identification operation may * ..*** * be based upon the associated energy values. **** * * ****

* The identification operation may identify a particular received data point in the data if and only if the energy value associated with the particular received data point is * within a predetermined range of an energy value associated with one of the data points associated with the element. * 0*

The identification operation may identify a particular received data point with the one of the data points having an energy value within the predetermined range.

The quantitative indication may be based upon a number of the received data points identified by the identification operation and the number of data points associated with the element.

The method may further comprise generating a score associated with at least one of the received data points identified by the identification operation, and the quantitative indication may be further based upon the score. For example, a score may be generated for each of the received data points identified by the identification operation and the quantitative indication may be based upon a mean of the determined scores or a maximum of the determined scores. The determined scores may be used to generate a rank for the received data points and the quantitative indication may be based upon the rank. Furthermore, a score may be generated for each of the received data points, irrespective of whether the received data point is identified by the identification operation and the quantitative indication may be generated based upon a number of data points identified by the identification operation that have a rank greater than a predetermined threshold.

The quantitative indication may be further based upon a difference between an energy value associated with a received data point identified by the identification operation and an energy value associated with the one of the data points having an energy value within the predetermined range. For example, a difference value may be generated for each received data point identified by the identification operation, the difference value being based upon a difference in energy value associated with * the received data point and the one of the data points having an energy value within * the predetermined range and the quantitative indication may be based upon a mean of the difference values or a minimum of the difference values or a maximum of the **** * difference errors. The quantitative indication may be further based upon a tolerance S.* **S * used in the identification of received data points to identify received data points. 0p** * .

*:*. According to a further aspect of the invention there is provided a method for determining the presence or absence of an element in a sample. The method comprises obtaining stored data relating to the element, the stored data comprising a plurality of stored data points associated with the element, each data point associated with the element having an associated energy value and receiving data generated from the sample, the received data comprising a plurality of received data points, each of the plurality of received data points having an associated energy value and an associated score. The plurality of stored data points and a received data point having a maximum score are processed to determine an energy difference, the energy difference indicating a difference in energy value between a stored data point having an energy value closest to the received data point having a maximum score. At least one of the plurality of received data points is identified in the received data based upon the stored data and the determined energy difference and the presence or absence of the element is determined based upon the identified at least one of the plurality of received data points; Taking into account a data point having a maximum score to determine an energy difference in this way provides a more effective determination of the energy difference. For example, prior art methods generally use a data point selected based upon a single property, for example an intensity value associated with the data point, which does not consider other relevant properties of the data point.

Identifying at least one of the plurality of data points in the received data based upon the stored data and the determined energy difference may comprise performing an identification operation between each of the stored data points and the received data points, and the identification operation may be based upon the associated energy values and the determined energy difference.

The identification operation may identify a particular received data point in the received data if and only if the energy value associated with the particular received * S....

* data point is within the determined energy difference of an energy value associated with one of the stored data points. The identification operation may also allow for a S...

tolerance such that the identification operation identifies a particular received data point if the energy value is within the determined energy difference plus or minus the S...

* * allowed tolerance. S... S. S

SS * S.

The energy difference may be indicative of a chemical shift associated with the element in the sample.

The method may further comprise initially processing the received data generated from the sample to select the element for which the presence or absence is to be determined.

The score may be generated according to the second aspect of the invention.

According to a further aspect of the invention there is provided a method for determining the presence or absence of an element in a sample. The method comprises receiving data generated from the sample and receiving data indicating at least one parameter of an apparatus used to generate the data from the sample.

Stored data relating to the element is obtained and the presence or absence of the element is determined based upon the received data generated from the sample, the received data indicating at least one parameter of the apparatus and the obtained stored data.

Parameters of the apparatus used to generate data from a sample can influence the generated data. By taking into account parameters of the apparatus, the influence upon the generated data can be considered when determining the presence of absence of an element in the sample.

The stored data may comprise a plurality of data points associated with the element and determining the presence or absence of the element may comprise determining a subset of the plurality of data points associated with the element based upon the parameter.

**.* *1 * Determining the presence or absence of the element may comprise determining a * ** S..

relationship between the subset of the plurality of data points associated with the element and the received data and determining the presence of the element in the sample if the relationship satisfies a predetermined criterion, the determining being independent of any relationship between the plurality of data points associated with the element not in the subset. That is, the subset of the plurality of data points associated with the element may be mandatory data points whereas data points of the plurality of data points not in the subset may be optional data points.

The at least one parameter may be a parameter associated with deflection of charged particles. For example, the parameter may be associated with a voltage applied to a part of the apparatus associated with deflection of charged particles. For example the part of the apparatus may be an electrode that causes electrons emitted from a sample to be deflected onto a charged particle detector such as a hyperbolically shaped electrode.

According to a further aspect of the invention there is provided a method for determining one or more elements which are present in a sample. The method comprises receiving data generated from the sample and determining a subset of data points based upon the received data, each data point having an associated value. Possible elements in the sample are determined based upon the determined subset of data points. The received data is further processed based upon known data associated with the determined possible elements to determine one or more elements present in the sample.

Determining a subset of data points based upon the received data in this way reduces the number of data points to be considered and determining possible elements in the sample is therefore computationally less expensive. Furthermore, some of the data points in the received data may be caused by, for example, instrument noise, and these data points can cause false positive determinations of the presence of elements in a sample. Removing these data points from the data points used to determine the presence of elements in a sample can reduce the number of false positive results.

Determining a subset of data points based upon the received data may comprise determining data points most likely to be useful in determining the one or more *.. ..

* elements present in the sample. I** * * S.*

The received data may comprise a plurality of data points. Determining data points most likely to be useful in determining the one or more elements present in the sample may comprise processing the received data to determine at least one data point having a value in a first range and combining the determined at least one data point having a value in the first range with at least one further data point having a value not in the first range.

For example, a set of data points most likely to be useful in determining the one or more elements present in the sample may be selected from the received data generated from the sample, and the set may include at least one data point from a high energy range and at least one data point from a low energy range.

The value associated with each of the data points may be an energy value and the first range may be a low energy range of values.

Processing the received data to determine at least one data point having a value in the first range may comprise performing a background subtraction process on the received data to generate first background subtraction data and determining the at least one data point having a value in the first range from the first background subtraction data.

Performing a background subtraction process on the received data may comprise processing the received data a plurality of times, each processing being based upon a respective value for a parameter, to generate respective intermediate sets of data points. The respective intermediate sets of data points may be processed to determine correspondence between data points in the respective intermediate sets of data points and the first background subtraction data may be generated from the determined correspondence.

Processing the received data each of the plurality of times may comprise processing pairs of data points, each pair of data points having an associated difference, and the difference may have a predetermined relationship with the respective value for the I.....

* parameter.

I..... * I

The method may further comprise determining the at least one further data point **4I * having a value not in the first range. Determining the at least one further data point * may comprise identifying a plurality of data points having a value not in the first range and selecting a subset of the plurality of data points having a value not in the first range based upon at least one property of the identified plurality of data points.

Selecting the subset of the plurality of data points having a value not in the first range may be arranged to select data points based upon at least one characteristic of the data points.

Determining possible elements in the sample based upon the determined subset of data points may comprise comparing values associated with each data point of the subset of data points with stored data. The stored data may indicate a plurality of elements and at least one value associated with each of the plurality of elements.

Further processing the received data may comprise comparing data points associated with each of the determined possible elements with data points of the received data. The method may further comprise selecting a data point in the received data and identifying a data point associated with one of the possible elements based upon the selected data point.

The method may further comprise determining whether the identified data point associated with one of the possibte elements has previously been identified based upon a previous data point in the received data. If the identified data point associated with one of the possible elements has previously been identified, a relationship between the identified data point and selected data point may be determined and a relationship between the identified data point and the previous data point may be determined. The relationships may be compared to determine one of the identified data point and previous data point to be associated with the identified data point associated with one of the possible elements.

In each of the embodiments of the invention, the received data may be spectral data.

S... S5 * The data points and/or features may be peaks in the spectral data. The received data I,....

* may be data generated by a charged particle energy analyser. For example, the charged particle energy analyser may be selected from the group consisting of: an S...

* Auger mass spectrometer, an X-Ray photoelectron spectroscope and an Electron Probe Micro Analyser. * 4

Aspects of the invention may be combined in any convenient way. Features described in the context of one aspect of the invention can, where appropriate be applied to other aspects of the invention.

Aspects of the invention can be implemented in any convenient form. For example computer programs may be provided to carry out the methods described herein.

Such computer programs may be carried on appropriate computer readable media which term includes appropriate tangible storage devices (e.g. discs). Aspects of the invention can also be implemented by way of appropriately programmed computers.

Embodiments of the present invention will now be described, by way of example only, with reference to the accompanying drawings, in which: Figure 1 is a schematic illustration of a system for spectral analysis according to an embodiment of the present invention; Figure 1A is a schematic illustration showing a computer of the system of Figure 1 in further detail; Figure 2 is a schematic illustration showing part of the system of Figure 1 in further detail; Figure 3 is a flowchart showing processing carried out to determine the elemental composition of a specimen; Figure 4 is a flowchart showing iterative processing carried out to determine background subtraction peaks in spectral data generated by the system of Figure 1; Figures 5A to 5E are graphs showing spectra created after different numbers of * iterations of the processing of Figure 4; * ** S.. * S

Figure 6 is a flowchart showing processing carried out to identify peaks in spectral S...

data generated by the system of Figure 1; S..... * S

Figure 7 is a histogram showing a frequency distribution of spectral peaks for reference elements according to peak energy; * ..

Figure 8 is a flowchart showing processing carried out to identify significant peaks identified by the processing of Figure 4 in a low energy range of a spectrum; Figure 8A is a schematic illustration of a peak showing data points used in the determination of a peak shape factor; Figure 9 is a flowchart showing processing carried out to generate a score for peaks generated by the processing of Figure 6; Figure 10 is a schematic illustration of a spectral peak showing various features of the peak; Figure 11 is a flowchart showing processing carried out to identify possible elements present in a specimen based on the peaks determined by the processing of Figures 4 and 5; Figure 12 is a schematic illustration of a database structure for storing data associated with spectral peaks; Figure 13 is a schematic illustration of a peak type code; Figure 14 is a class diagram showing objects used to match spectral data to known data; Figure 15 is a flowchart showing processing carried out to match known data to S.....

* spectral data; and *.... * S

Figure 16 is a flowchart showing processing carried out to generate a likelihood of S. * the presence of an element in a specimen.

*IS.** * S Referring to Figure 1, an analyser 1 is arranged to detect charged particles ejected **** from a specimen 2. Charged particles are ejected from the specimen 2 following exposure of the surface of the specimen 2 to an electron beam 3 provided by an electron gun 3a. A computer 4 is arranged to receive and process data from the analyser 1. The computer 4 is arranged to communicate with a database 5. The database 5 stores reference data useful for determining whether experimental data is indicative of the presence of a particular element. The database 5 may also store data based upon the data received from the analyser 1. The analyser 1 comprises a multi-channel plate on its upper surface. A phosphor screen is located below the multi-channel plate, and a pattern on the phosphor screen created by arriving charged particles is detected by a photo diode array positioned below the phosphor screen. A suitable apparatus is an Auger Electron Spectroscope described in published International Patent Application Publication No. W09935668, (Prutton, El Gomati and Jacka), although any suitable charged particle energy analyser may be used. For example, the apparatus may be an X-Ray Photoelectron Spectroscope, or an Electron Probe Micro Analyser.

Figure IA shows the computer 4 in further detail. It can be seen that the computer comprises a CPU 4a which is configured to read and execute instructions stored in a volatile memory 4b which takes the form of a random access memory. The volatile memory 4b stores instructions for execution by the CPU 4a and data used by those instructions. For example, in use, data generated from the specimen 2 may be stored in the volatile memory 4b.

The Computer 4 further comprises non-volatile storage in the form of a hard disc drive 4c. The data generated from the specimen 2 may be stored on the hard disc drive 4c, together with other data used in the method such as theoretical data relating to elements. The computer 4 further comprises an I/O interface 4d to which are connected peripheral devices used in connection with the computer 4. More particularly, a display 4e is configured so as to display output from the computer 4. *.*..

* The display 4e may, for example, display a representation of the data generated from ****** * the specimen 2. Input devices are also connected to the I/O interface 4d. Such input devices include a keyboard 4f and a mouse 4g which allow user interaction with the **** computer 4. A network interface 4h allows the computer 4 to be connected to an * appropriate computer network so as to receive and transmit data from and to other *. computing devices. The CPU 4a, volatile memory 4b, hard disc drive 4c, I/O **** interface 4d, and network interface 4h, are connected together by a bus 4i.

Figure 2 schematically shows part of the arrangement of Figure 1 in further detail.

The analyser 1 comprises a position sensitive detector 6 which provides a plurality of channels 7, twelve channels being shown in Figure 2. Each of the channels 7 is arranged to detect charged particles landing on a particular area of the position sensitive detector 6. In some apparatus the number of channels is 2048 although the number of channels may vary between detectors.

The electron beam 3, provided by the electron gun 3a of Figure 1, causes charged particles to be ejected from the specimen 2 and some of the ejected particles are projected into one of the channels 7 provided by the position sensitive detector 6.

Charged particles ejected from the specimen 2 follow a trajectory before landing on the position sensitive detector 6. Four groups of trajectories El, E2, E3 and E4 are shown in Figure 2, with the trajectory followed by a particular charged particle being determined by its energy and charge. Assuming charged particles have equal charges (which is generally the case for Auger Electron Spectroscopy) then charged particles with low energies follow a trajectory in the group of trajectories El and land in channels of the position sensitive detector 6 that are relatively close to the specimen 2 while charged particles with high energies follow a trajectory in the group of trajectories E4 and land in a channel relatively far from the specimen 2. The relationship between the energy of an emitted charged particle and the particular channel in which the charged particle lands and is detected is quadratic and is described by equation (1) below: -=a(Ch)2 +b(Ch1)+c (1) where E, is the energy of charged particles landing at channel i, Ch1 is the ith channel, * V is the voltage applied to a negatively charged hyperbolically shaped electrode 9, * and a, b and c are calibration coefficients specific to the particular detector and **** * configuration which is used. S...

S.....

* S Changing the voltage applied to the hyperbolically shaped electrode 9 changes the . electric field between the hyperbolically shaped electrode 9 and a grounded * *:*. electrode 10 which affects the flight path of charged particles. Increasing the voltage applied to the hyperbolically shaped electrode 9 allows the range of charged particle energies that can be detected by the position sensitive detector 6 to be modified. For example, where the charged particles are negatively charged particles (e.g. electrons) increasing the negative voltage applied to the hyperbolically shaped electrode 9 reduces the trajectory that an emitted negatively charged particle of particular energy follows, thereby causing negatively charged particles having a relatively high energy to be projected onto the position sensitive detector even if such particles would be projected beyond the end of the position sensitive detector when a lower negative voltage is applied to the hyperbolically shaped electrode 9. As such, different voltages applied to the hyperbolically shaped electrode 9 cause different energy ranges of charged particles to be detected. It can be noted that the ratio between the highest and lowest energy detected by the position sensitive detector 6 remains constant as the voltage applied to the hyperbolically charged electrode is varied.

Methods for determining the elemental composition of a specimen 2 using the system of Figure 1 will now be described. At a very general level, and as is described in further detail below, spectral data from a specimen of interest is obtained, and features of the obtained data which are likely to be useful in the analysis of the specimen are identified. Having identified such features a degree of correspondence between the identified features and stored data associated with particular elements is determined so as to identify which elements are present in the specimen. The features of the obtained data used to determine elements present in the specimen may be peaks in the obtained data. In determining correspondence, various peak properties are taken into account, including a peak score which combines data indicating various peak characteristics. The methods described below first identify elements which might be present in the specimen of interest before further processing is carried out to identify which elements of the elements that might be * present in the specimen are considered most likely to be present in the specimen or ****** * used to rank the elements in order of significance. * * S. S.

Figure 3 shows this processing at a high level. The processing of Figure 3 can be *5S*S * considered to comprise four parts. A first part A comprises receiving spectral data. A second part B comprises processing the received data so as to generate various S...

processed versions of the data. Peaks which are likely to be useful in the analysis of the specimen can then be identified from the processed versions of the data, such peaks being identified based upon various peak properties which are included in the generation of a peak score. A third part C comprises identifying elements which might be present in the processed specimen, while a fourth part D comprises determining a likelihood that each element which might be present is in fact present.

The generated likelihoods can be processed with reference to a threshold so as to determine whether it is sufficiently likely that a particular element is present in the specimen for it to be considered that the particular element is present in the specimen.

Considering Figure 3 in further detail, at step SI raw spectral data S0 is received.

The raw spectral data S0 comprises a value indicative of (e.g. proportional to) a count of the number of charged particles detected in each of the N channels of the position sensitive detector 6 of Figure 2. From equation (1), the energy E of charged particles detected in each channel can be determined. As such, the raw spectral data So can be viewed as a count of charged particles having energy values E1 for i in the range I to N corresponding to the determined energy value for a respective channel Ch. The data S0 can be represented in graphical form, as described below with reference to Figures 5A to 5E.

At step S2 the data is pre-processed to remove dark current. Dark current is the residual current in a detector when there is no incident illumination, and contributes a small amount of noise to a measured signal. The dark current may be stored data that is known for the apparatus or may be determined at any time by generating data whilst no specimen is currently being analysed. Dark current may vary due to temperature changes and so will generally be measured regularly. Other pre- processing steps described in further detail below, or any other suitable pre- * processing, may also be carried out.

*I*.*S * S

S

S.....

* At step S3 a set of low energy background subtraction peaks B<100 is determined from the pre-processed data output from step S2 as described in further detail below *.S.

* with reference to Figure 8. The set B<100 is determined from three sets of background S.....

* subtraction peaks 2o 5O and each generated from the data output from step .* S2. Each of the sets 2o, 5O and is generated according to the processing described below with reference to Figure 4. The set of peaks B<100 is a set of peaks identified as significant peaks with energy less than 100eV.

At step S4 the data output from step S2 is further pre-processed to correct for gain.

Gain may be caused by unwanted substances on the detector, by damage to the detector or by features of the structure of the detector such as the arrangement of the channels in the multi-channel plate. A signal corresponding to the gain of a particular apparatus may be calculated by analysing a known control specimen with known features, such as carbon, in the apparatus. A function such as an exponential function is fitted to the data determined from analysing the known control specimen.

The function is fitted to regions of data which are known to not include any features associated with the analysed specimen such that the region only contains artefacts associated with the detector. The data to which the function is fitted may be determined by averaging a plurality of measured data determined under identical conditions, or by analysing the control specimen for a relatively long period of time.

The function may be fitted to the data in any convenient way, for example using the least squares method.

A gain correction factor for each channel ch, GCFCh, is determined by determining the value of the fitted equation at channel ch divided by the value of the data at channel ch after dark current subtraction. Measured data at a channel ch is then corrected for gain according to equation (Ia): CorrectedCh = GCFCh (MeasuredCh -DarkCurrentCh) (la) where: CorrectedCh is the gain corrected data at channel ch after dark current * subtraction; *.**.

* : MeasuredCh is the raw spectral data at channel ch; and * ***** * DarkCurrenth is the dark current at channel ch.

Note that the data (MeasuredCh -DarkCurrentCh) is determined at step S2. 0*** *

* The control specimen may be analysed by applying two different voltages to the * hyperbolically shaped electrode 9 of Figure 2. The differing voltages cause charged *:*. particles emitted from the control specimen to be projected into different channels of the position sensitive detector as explained above, according to equation (1) so that the full range of channels of the detector can be assessed. For example, Carbon may be used to determine the gain correction factor since Carbon has known features in only a relatively small energy range. Applying a low voltage to the hyperbolically shaped electrode causes features associated with Carbon to be detected at a high energy channel and an equation in the low energy range can be determined.

Applying a high energy voltage causes features associated with Carbon to be detected at a low energy channel and an equation in the high energy range can be determined. The equation is determined based upon regions containing no specimen features so that the region only contains artefacts associated with the detector. By applying different voltages to the hyperbolically shaped electrode, an equation can be fitted to all channels such that a Gain Correction Factor can be determined for each channel.

At step S5 a set of background subtraction peaks B50 is generated from the gain corrected data output from step S4 across the whole energy range of the acquired data according to the processing described below with reference to Figure 5 and at step S5A data associated with the peaks B50 is generated. Peaks having an associated energy of less than 100eV are discarded. The set of background subtraction peaks B50, in combination with the set of significant low energy peaks with energy less than 100eV B<100 identified at step S3 and after the filtering of step S8 forms a set of significant peaks at step S9 which is used to identify possible elements in the specimen as described below with reference to step SlO.

At step S6 the spectral data is further pre-processed to smooth the data. One suitable method of smoothing the data is by fitting a Savitzky-Golay polynomial to the gain-corrected data output from step S4. Details of the process of fitting a Savitzky-Golay polynomial can be found in A. Savitzky and Marcel J.E. Golay (1964) S.....

* "Smoothing and Differentiation of Data by Simplified Least Squares Procedures".

S.....

* S Analytical Chemistry, 36: 1627-1639, the contents of which are herein incorporated by reference. It will be appreciated that other data smoothing methods can be used, S...

such as moving averages, moving medians and Gaussian smoothing. The smoothing * S technique may be configurable and may be selected based on a tuning process using a set of training data. S... S. *

S SS

* At step S7 a set of peaks A is determined from the data output from step S6 as described with reference to the processing of Figure 6 and at step S7A the peaks A are scored as described in further detail below with reference to Figure 9. The scoring of the peaks A uses values determined from corresponding peaks in the set of peaks B50 generated at step S5 as described in further detail below. At step SI 1 the set of peaks A is used together with data stored in the database relating to the possible elements identified at step Sb to identify which elements of the possible elements identified at step Sb are most likely to be in the specimen.

At step S8 it is determined for each peak in the set B<100 generated at step S3 if there is a corresponding peak in the set A generated at step S7 within a configurable tolerance. A suitable tolerance has been found to be ±4eV. Any peak in the set A in the energy range less than 100eV that does not have a corresponding peak in the set B<100 is removed from the set A and is not considered further in the processing of step Si I. At step S9 a set of significant peaks SP is determined from the set of peaks B50 (>100eV) generated at step S5 and the set of peaks B<100 updated after the filtering of step S8. The generation of the set SP is described below with reference to Figure ii.

At step SiO the significant peaks SP are processed to determine a set E of elements that may be present in the specimen 2 of Figure 1. The processing of step Sb 0 is described in further detail below with reference to Figure ii. At step Sb I known data for each of the elements in the set E identified at step SbO is processed to identify matches between peaks included in the set A after the processing of step S8, and stored data, as described in further detail below with reference to Figure 15. At step S12 a likelihood of the presence of each of the elements in the set E in the specimen is determined as described in further detail below with reference to Figure 16.

S

* The generation of the set B<100 at step S3 uses three sets of background subtraction *5.

* peaks. Processing to generate background subtraction peaks is now described with s... reference to Figure 4. *5**

S

*S55** * Background subtraction peaks are identified using a background subtraction method *.* such as that described in H.E. Bauer (1995) "A fast and simple method for S...

*: .: background removal in Auger electron spectroscop', Fresenius' Journal of Analytical Chemistry, 353: 450-455, although any other suitable background subtraction method could be used. The background subtraction method of Bauer is an iterative process that takes as input a value L. The background subtraction data is subsequently processed to identify a set of peaks and the value is the maximum peak width of the identified peaks, although peaks with width less than may be generated.

The processing of Figure 4 generates a sequence of data sets S corresponding to the data after n iterations of background subtraction from the data set. Each iteration of background subtraction reduces at least one point in the data to zero. Other than two initial iterations which reduce the first and second points (being the two points having lowest associated energy) in the data to zero, the point in the data that is reduced to zero is a local minimum. A local minimum is defined as any value V in a consecutive sequence of values V, V1+1, .. V÷ for which V V1 and V+1 »= V. Local maxima are defined analogously. The value, S(E) indicates the determined intensity at energy E, for j in the range 1 to N, after n iterations and each data set S includes at least one value S(E) that is equal to zero that is not equal to zero in the data set generated by the n1th iteration if at least two points in the data set S1 are at zero and are separated by more than Li before the iteration. If two points in a data set S are not separated by more than Li then a further iteration is not performed.

Referring now to Figure 4, at step S20 the value Li is initialised. At step S21 a data set S is determined according to equation (2) below: S(E) = S'(E) -min{So'(E),(il. ..N)} , (j=1.. .N) (2) where: S0'(E1) is the value of the input data at the ith energy after appropriate pre- * : processing; and S.....

* S mm is a function which returns the minimum value included in a set of values provided as an argument. **.

S

S. S*S.

* S It can be seen from equation (2) that S is a set having at least one value S1(E1) equal to zero and all other values equal to the corresponding value of S0' after subtraction S...

of the value generated by the function mm.

At step S22 a value k is determined such that k is the smallest value for which Sl(Ek) is equal to zero. At step S23 a check is performed to determine if k is equal to 1. If it is determined that k is equal to 1 then the value S1(E1), corresponding to the value determined from the first channel of the position sensitive detector 6, is equal to zero and at step S24 the data set S2 corresponding to the data after a second iteration of background subtraction is generated such that all values in S2 are equal to the corresponding values in S1. If it is determined at step S23 that k is not equal to 1 then at step S25 a check is performed to determine if k is equal to 2. If it is determined that k is equal to 2 then at step S26 the value S2(E1) is set to zero and all other values of S2 are set to the corresponding values of S1. If it is determined that k is greater than 2 then at step S27 values of S2(EJ) for j in the range 1 to k-i are calculated according to equation (3) below: S2(Ej)_SI(Ei)_Sl(El).[EEJi_1...k_1 (3) where a has a value calculated according to equation (4) below: s1(E,) a=,nax s1(E1) ,i=2...k-1 (4) Ek -E1 If a has a value less then zero then a is set to zero. All other values of S2(E1) for] greater than or equal to k are set to the corresponding value of S1. The processing of steps S24, S26 and S27 generates a data set S2 in which S2(E,) is equal to zero 1:. . and all other values of S2(EJ) are greater than or equal to zero. The processing of equation (3) can also be seen to generate a data set in which S2(E1) is equal to zero.

The fraction in equation (3) will always be equal to 1 when j is equal to 1 as the top *** d and bottom of the fraction will be equal in this case. The value S2(E1) will therefore be equal to S1(E1)-S1(E1).1 which hasavalue zero. ) S..

The processing of each of steps S24, S26 and S27 passes to step S28. At step S28 * . the value k is set such that k is the maximum value in the range 1 to N which is such that Sl(Ek) is equal to zero. At step S29 a check is performed to determine if k is equal to N. If it is determined that k is equal to N then at step S30 the data set S3 corresponding to the third iteration of background subtraction is generated such that all values of S3 are equal to the corresponding values of S2. If it is determined that k is not equal to N then at step S3i a check is performed to determine if k is equal to N-i. If it is determined that k is equal to N-i then at step S32 the value S3(EN) is set to zero and all other values in the set S3 are set to the corresponding values of S2.

If it is determined that k is less than N-i then at step S33 values of S3(EJ) for j in the range k+i to N are calculated according to equation (5) below: S3(EI)S2(E])_S2(EN).[:J:kyk+1...N (5) where a has a value calculated according to equation (6) below. s(E.)

cL=max,i=k+1...N-1 (6) k EN -Ek If a is determined to have a value less than zero then a is again set to the value zero.

The processing of each of steps S30, S32 and S33, which is carried out at relatively high energy values, generally corresponds to the processing of steps S24, S26 and S27 which is carried out at relatively low energy values. Each of steps S30, S32 and S33 result in a data set S3 where S3(EN) is equal to zero, S3(E1) is equal to zero, and all other values of S3(E) are greater than or equal to zero. As will be described below, step S35 for subsequent iterations of background subtraction requires that two points in the data are equal to zero, and the processing to generate the data set * .* S3 ensures that this requirement is satisfied.

S

The processing of each of steps S30, S32 and S33 passes to step S34 where a counter n which counts through subsequent iterations of the background subtraction process is assigned the value 4. The processing of steps S35 to S40 generates a data set S comprising data values after n iterations of background subtraction.

At step S35 two values I and m are determined such that the following conditions are satisfied: (i) the energies E1 and Em are the two lowest energies for which the associated intensity in the data generated in the previous iteration is zero, that is the two lowest energies which are such that S1(E1) and Sn1(Em) are both equal to zero; (ii) the intensity associated with all energies between the energies represented by I and m is greater than 0, that is S1(E) is greater than zero for values of j between land m; (iii) the energies specified by I and m differ by at least the value specified at step S20, that is the value Em -E1 is greater than; and (iv) mis greater than I + 1.

At step S35a a check is carried out to determine whether values of I and m have been selected so as to satisfy the criteria of step S35. If this check is satisfied processing continues at step S36, otherwise processing ends on the basis that no further iterations can be carried out.

At step S36 a minimum value C is determined for all i according to equation (7) below: C=minl,i1+1...ni1 (7) * (EiEi)(EmEi) ) *e.S * * ** ** ** At step S37 the values of S(E) for j in the range j = l+1...m-1 are determined according to equation (8): I...

S S..

s(E1)=s,,1(E)_c.(E Ei)(Em _E),j +1*Jfll (8) where C is the value determined at step S36 according to equation (7).

Equations (7) and (8) are constructed in such a way as to ensure that all values S(E1) are positive or zero. This ensures that an over compensation of the background resulting in negative values is not possible. At step S38 the values of S(E) in the range j = l...l and m...N are determined according to equation (9) below: S(E)=S1(E),j=1...landj=m...N (9) That is, the intensities associated with energies in the range 1 to I and m to N after the nth iteration are set to be equal to the values associated with the corresponding energy after the (n-i)th iteration.

At step S39 it is determined if more iterations are required by comparing the value of the counter n to an upper limit. If it is determined at step S39 that more iterations are required then at step S40 the counter n is incremented and the processing of steps S35 to S39 is repeated. It has been found that a suitable number of iterations is 100, although other values of n could be used. The output of the processing of Figure 4 is the data set Sn.

It will be appreciated that the background subtraction process described above is illustrative only, and other background subtraction methods may be used.

Figures 5A to 5E each show a graph indicating signals corresponding to the data set S..... . . * . S calculated according to the processing of Figure 4 for varying values of n. Each graph shows a line 11 indicating the original signal corresponding to the data set 5o' and a line 12 indicating the cumulative background signal that is subtracted from S0' after n iterations to give S. A line 13 in each of Figures 5A to 5E indicates the signal corresponding to the data set S generated by subtracting the data represented by the line 12 from the data represented by the line 11. In Figures 5Ato 5D the line 12 is generated using a value of L equal to 100.

Figure 5A shows the data after the first 3 iterations of the background subtraction processing described above with reference to Figure 4 and the line 13 therefore corresponds to the data set S3. It can be seen from line 13 of Figure 5A that all data values have been reduced so that all data values are defined relative to a minimum data value which has been reduced to zero. This corresponds to the processing of step S21 of Figure 4. It can also be seen that the first and last data values have each been reduced to zero, corresponding to the processing of steps S23 to S27 and steps S29 to S33 of Figure 4. Figures 5B to 5D correspond to the data after a further 3, 9 and 19 iterations respectively (corresponding to the data sets S6, S12, and S22 respectively).

It can be seen from Figures 5A to 5D that as additional iterations are carried out, peaks present in the initial data can be more easily identified. It can also be seen that after 22 iterations, the resulting data being shown in Figure 5D, there are no energies having associated zero data values which are separated by at least (where = 100). As such, further iterations of the processing of steps S35 to S40 will not perform any further background subtraction (as detected at step S35a).

Figure 5E shows signals for equal to 20 with a value of n of 49. Here, it can be seen that a greater number of energy values now have associated data values of zero. This is because further iterations of the processing of steps S35 to S40 can be carried out, given that additional energies can be selected so as to satisfy the criterion relating to the value of t set out at step S35 (criterion (iii) above).

As set out in further detail below, identification of true positive low energy peaks is particularly difficult and the information from three different sets of background subtraction data is used to identify true positive peaks in the low energy region as is described below with reference to Figure 7.

* Background subtraction data generated by the processing of Figure 4 and illustrated ** with reference to Figures 5A to 5E above is processed to identify a set of peaks by identifying areas of the curve that are separated by two zero points. Other data which has not been processed so as to reduce points to zero, such as the data processed * S..

at step S7, is processed to identify peaks based on local minima and maxima.

*. Processing to identify peaks in data other than background subtraction data will now be described with reference to Figure 6.

As indicated above, during the processing of Figure 6, as described in further detail below, the data is processed to identify local maxima and local minima. The data is processed starting from the low energy range and moving towards the high energy range to identify consecutive maxima and minima in the data. For example to identify a maximum starting from a minimum with an intensity associated with a channel Ch1, the intensities associated with subsequent channels Ch1+1, Ch+2, ..., Ch are processed to identify a channel Chm for m in the range i+1 to n such that the value at channel Chm is less than the value at channel Chmi. Since channel Chm indicates that the intensity at consecutive channels is no longer increasing, this indicates that the intensity determined at channel Chmi, which was determined to be greater than or equal to the intensity at channel Chq for q in the range ito m-2, is a local maximum and is added to a set of maxima. Processing to identify minima proceeds analogously.

At step S45 a minimum value m1 is identified in the input spectral data in the manner set out above. At step S46 the minimum rn is added to a set MIN. At step S47 a maximum M1+1 in the data is identified in the manner set out above and at step S48 the maximum M+1 is added to a set MAX. At step S49 a check is carried out to determine if there is more data to be processed to identify maxima and minima. More data remains to be processed if the processing has not reached the data value determined at the channel of the position sensitive detector 6 corresponding to the highest energy value that the electron detector can detect. If it is determined that there is more data to be processed then the loop of steps S45 to S49 is repeated to identify the next maxima and minima in the data until there is no more data to be processed. As set out above, the first minimum m1 and first maximum M2 are the first detected minimum and maximum working from the lowest energy included in the spectrum. Subsequent maxima and minima are maxima and minima determined from S.....

* detections at channels that detect electrons with increasing energy. The maxima correspond to the summit of peaks in the signal representative of the data and so for any two maxima M and M1+2, M+2 indicates the summit of a peak that identifies electrons with a higher energy than M. S... * . S...

Whilst the processing of Figure 6 has been described above as beginning with channels in the low energy range and moving towards the high energy range, it will be appreciated that the processing may work in the reverse direction. That is, processing may begin with channels in the high energy range and move towards the low energy range. The processing of Figure 6 is described in further detail in Ludék Frank "Automated recognition of Auger electron spectra" Vacuum, vol. 6, numbers 7- 9, pages 437 to 440, 1986, which is herein incorporated by reference.

Identification of true positive low energy peaks can be problematic. Figure 7 is a histogram showing how many elements are expected, according to known elemental data, to produce a peak at a particular energy using Auger Mass Spectrometry.

Figure 7 shows energy values arranged in 25eV bins. As can be seen from Figure 7, more than 30 elements are expected to produce a peak in one of the low energy bins 14 and many other low energy bins indicate that a greater than average number of elements will produce a peak in a particular 25eV energy range. A single peak in the low energy region can therefore be a potential match to many elements, depending on the energy tolerance used to match peaks to elements.

Identification of true positive low energy peaks (that is peaks with energy below 100eV) is important because not only do many different elements have theoretical peaks in this region, but additionally true positive peaks in the low energy region can be hidden by peaks due to low energy secondary electrons emitted from the specimen that occur in this region.

Figure 8 shows processing to identify true positive low energy peaks, based on data generated by the processing of Figure 4. At step S54 the processing of Figure 4 is repeated three times, with values of L of 20, 50 and 100 to determine three sets of background subtraction data from the input data, each set of background subtraction data being subsequently processed to identify a respective set of background subtraction peaks 2O' 5O' in the respective background subtraction data. Each peak in the background subtraction data is defined by two points which each have a value of zero, together with a maximum, although a peak may have more than one local maximum. S...

At step S55 a peak b20 is selected from the set 2o such that the peak b20 has energy less than 100eV. At step S56 a check is performed to determine if the width of the peak b20 is greater than a first predetermined value. The width of the peak can be determined directly from the background subtraction data based upon the distance between the two zero points forming the left and right boundaries of the peak. If it is determined that the width is greater than the first predetermined value then processing passes to step S57 where a check is performed to determine if the standard deviation of the intensity of the peak is greater than a second predetermined value. The standard deviation of the intensity of the peak is determined by calculating the standard deviation of the values of each of the data points between the two zero points m11 and m+1 that define the width of the peak. If it is determined that the standard deviation is greater than the second predetermined value then at step S58 a check is performed to determine if a peak shape factor of the peak b20 is greater than a third predetermined value. Calculation of a peak shape factor is described in further detail below with reference to Figure 8A. If any of the checks of steps S56, S57 and S58 are not satisfied then processing passes to step S59 where it is determined if there are more peaks in the set A20 to be processed. If there are more peaks to be processed then processing returns to step S55 where a peak that has not previously been processed is selected.

The checks of steps S56 to S58 are intended to determine if characteristics of an identified peak correspond to characteristics of peaks that are true positive peaks.

The first, second and third predetermined values are determined according to a training process using a training set of data determined on the particular analyser 1 from known elements. Suitable values for one particular analyser have been found to be: width of peak b20 > 2.2 (i.e. the first predetermined value has a value of 2.2); standard deviation of intensity of peak > 0.005 (i.e. the second predetermined value has a value of 0.005); peak shape factor of 0.53 (i.e. the third predetermined value has a value of 0.53); * although, as indicated above, these values vary according to values determined for *::* the particular analyser, and additionally, other parameters may be used where a training process indicates that other parameters are useful in determining true positive peaks. 0*SS * * S...

** If each of the checks of steps S56, S57 and S58 are satisfied then processing passes to step S60. At step S60 a peak b50 in the set A50 that encloses the peak b20 is identified and at step S61 a peak blOO in the set A that encloses the peak b20 is identified. A peak, defined by two zero points m1 and m+1, is said to enclose a peak defined by two zero points m20*frl and m20+1, if the zero point m1 has energy less than or equal to m201 and the zero point m1+1 has energy greater than or equal to m20÷1. It should be noted that there may not be a peak b50 or a peak blOO that encloses the peak and in such a case the peak b50 or blOO is set to null.

At step S62 a set of peak values PV for each peak is determined based upon the currently processed b20 peak and its corresponding, if any, b50 and blOO peaks.

The values PV are shown in Table 2 below.

Term Definition dFirstMax Energy of the b50 maximum dSecond Max Energy of the bI 00 maximum dFirstDiff Abs(dFirstMax -energy of the b20 maximum) dSecondDiff Abs(dSecondMax -energy of the b20 maximum) dThirdDiff Abs(dFirstMax -dSecondMax) bFirstMatchExact A Boolean indicating whether at least one of the boundaries associated with the b50 peak matches a boundary associated with the b20 peak bSecondMatchExact A Boolean indicating whether at least one of the boundaries associated with the blOO peak matches a boundary associated with the b20 peak

TABLE 2

At step S63 a peak is added to the set of significant low energy peaks B<100 with the peak being one of the peaks b20, b50 or blOO selected based upon the values PV and the conditions shown in Table 3 below. The conditions shown in Table 3 were identified by analysing experimental data together with known data to determine rules I...

to identify true peaks in the experimental data. Whilst values are provided in Table 3, the values used for a particular analyser are configured according to a semi automated or automated training process using known data on the particular analyser to determine optimal values.

Peak Added to the List Condition b20Peak a) (dFirstDiff < 6.7 AND dSecondDiff < 4) OR (dFirstDiff < 4 AND dSecondDiff < 6.7) AND (dFirstMax!= 0.0 OR dSecondMatch!= 0.0)

OR

b) dFirstDiff < 4 AND dFirstMax!= 0.0

OR

C) dSecondDiff < 4 AND dSecondMax!= 0.0 b50Peak a) from 1st row AND dFirstDiff < 4 AND bFirstMatchExact == True

OR

b) from 1st row AND b50!= null

OR

d) dThirdDiff < 4 AND dFirstMax!= 0.0 AND dSecond Max!= 0.0 bi OOPeak a) from 1st row AND dSecondDiff < 5 AND bSecondMatchExact = True

OR

C) from i' row && blOO!= null

OR

d) from 2nd row

TABLE 3

The set B<100 may comprise peaks identified from any, or all, of the background subtraction peak sets A20, A50 and A. The set B<100 indicates the low energy peaks that are considered to be true positive peaks. All low energy peaks in the set A that do not correspond to a peak in the set B<100 are discarded (corresponding to step S8 of Figure 3).

The set of peaks B<100 is used together with the set of background subtraction peaks B50 generated at step S5 of Figure 3 to identify possible elements in the specimen.

The set of peaks B50 is generated from the data output from step S4 of Figure 3 after * ** S...., dark current removal and gain correction and is determined from data generated *. according to the background subtraction processing of Figure 4, with a value of A of 50, and subsequent identification of peaks from the resulting background subtraction data. Any peaks with energy less than 100eV in the set B50 are not considered as the set B<100 contains those peaks in the energy range less than 100eV determined to be true positive peaks iii that energy range.

The processing of step S5A of Figure 3 is arranged to generate data associated with each peak in the set B50. To do this the width, height, area, standard deviation and a peak shape factor of each peak in the set of peaks B50 is determined. The width is determined by the distance between the two minima of the peak (which are both zero points given the background subtraction process), the height is the value of the maximum intensity in the peak, and the standard deviation is determined as described above with reference to step S57 of Figure 8. The area is determined in any convenient way such as by using a trapezoidal integration formula as described below with reference to step S66 of Figure 9. The shape factor is determined according to equation (10) below: (y -y0),n >0 (10) 2n where y0 is the height of the point corresponding to the maximum M of the peak as shown in Figure 8A, and the y are the heights of the n data points, corresponding to values of the background subtraction data S, either side of the point yo five such data points being shown on either side of the maximum in Figure 8A. The value of the shape factor of a peak has been found to be a good indicator of whether a peak is a true positive peak indicative of the presence of an element in the specimen, and which can therefore reliably be used to determine the presence of an element in a specimen. * .

Before the sets of peaks B<100 and B50 are processed to determine possible elements present in the specimen, the set of peaks B<,00 is processed (at step S8) together with peaks in the set of peaks A in the energy region less than 100eV, generated at step S7 from the spectral data without background subtraction having first been performed on the spectral data. Any peaks in the set B<100 that do not match a peak * ** in the set A within a configurable tolerance, for example 4eV, are removed from the set A and not considered any further. Processing to identify and score the set of peaks A of steps 57 and S7A of Figure 3 is now described.

At step S7 of Figure 3 the set of peaks A is determined from the spectral data after dark current subtraction, gain correction and smoothing, that is, the data output from step S6 of Figure 3. The data output from step S6 of Figure 3 is processed according to the processing of Figure 6 to identify peaks as described above.

Each peak in the set A is processed to determine properties of the peak as will now be described with reference to Figures 9 and 10. At step S65 of Figure 9 a peak r corresponding to a maximum M1 in the set A (generated according to the processing of Figure 6) is selected such that the peak r has not previously been selected. At steps S66 to S70 values based upon properties of the peak r are determined. Figure shows an example peak, and the determination of properties of the peak r is described with reference to Figures 9 and 10. Figure 10 shows a peak r defined by a maximum 15 corresponding to a maximum M1 and two minima 16, 17 corresponding to minima m11, m1+1 either side of M1 in the spectral data S0 are shown.

Some of the properties for a peak r described below are determined using a corresponding background subtraction peak bs in either the set B<100 (for peaks with energy value less than 100eV) or the set B50. In the following, the term corresponding background subtraction peak, indicated by bs, is used to indicate a peak in either the set B<100 or the set B50 which encloses the maximum of the peak r between its two boundary points. Use of the background subtraction peaks in the determination of properties of a peak r allows the properties to be adjusted where they might otherwise not properly reflect true properties of the peak, for example where a peak lies on a shoulder of a further peak.

* ***** * S At step S66 a value Area representing the area of the peak r (shaded in Figure 10) is determined from the data. A,ea' is determined as the greater value of the area of the peak r and an area based upon the corresponding background subtraction peak bs. The area 18 of the peak r (shaded in Figure 10 and referred to * as Arear in the following) is first determined from the data. The combined areas 18 and 18a may be determined by a trapezoidal integration formula. In more detail, each peak is defined by a number of points, and the area under each peak can therefore be approximated to a plurality of trapezoids. The area of each trapezoid can be determined according to the standard calculation of Area = 1/2(hl + h2)*w where hi and h2 are the lengths of the parallel sides of the trapezoid, determined by two consecutive data points, and w is the distance between the data points along the x-axis. The total area indicated by the shaded area 18 together with area 18a can be determined by summing the trapezoidal areas. The area 18a is determined by first calculating the equation of the straight line between points 16 and 17 of Figure 10 using the standard equation y mx +c. For each data point along the line 22, the spectral intensity I is compared with the value y, with input x provided by the value of the data point. Where the value y is less than or equal to l, the value y is used in the determination of the area 18a, otherwise the value I is used in the determination of the area 18a. The total area 18a is determined in the same manner as described above for the combined area 18 and 18a, by summing the areas of trapezoids. Arear 18 is determined by subtracting area 18a from the combined areas 18 and 18a.

A value Areas is determined for each peak r, the value Area being based upon the area of the peak bs AreabS (that is, the area of the bs peak that encloses the maximum of the peak r). Given that background subtraction peaks are bounded by two zero points, the value AreabS may be determined by a trapezoidal integration formula. In more detail, each peak is defined by a number of points, and the area under each peak can therefore be approximated to a plurality of trapezoids. The area of each trapezoid can be determined according to the standard calculation of Area = + h2)*w set out above, and the total area of bs, AreabS, can be determined by summing the trapezoidal areas in the manner described above. * *

For each peak r, the value Area is determined according to equation (11) below: *S.* * S **** 55.55 * Areas = Areab * ( widthbs -Abs(Energy58 -Energy) r Width) * * bs *..* S. * S S * S ** where: Areab5 is the area of the peak bs, determined as described above; Energybsmax is the energy value of the point in bs which has maximum intensity; widthb$ is the distance between Energyb$max and the upper zero point of the peak bs if Energym,ax > Energybsmax, or the lower zero point of the peak bs if Energyax < Energybmax; Energyrmax is the energy value of the point in r which has maximum intensity; and Abs is a function which returns the absolute value of its parameter.

The area for the peak r, A rear, is determined by selecting the greater value of the area determined for the peak r, Arear and the value Area determined taking into account the corresponding peak bs according to equation (11) as described above.

At step S67 the right hand height as indicated by distance 19 is determined and at step S68 a width indicated by a distance 20 is determined. The width is determined at a height equidistant from the maximum 15 and a line 21, the line 21 being equidistant from the height of both minima 16,17 (sometimes referred to as the full width at half maximum height).

At step S69 a base width is determined for the peak r. The base width is selected as the greater of the width of the peak r, indicated by line 22 in Figure 10, and a value determined based upon the corresponding peak bs to the peak r. The value determined based upon the corresponding peak bs is determined according to a modified equation (11) in which the right hand side of equation (11) is multiplied by the width of the peak bs. All other peaks have their width adjusted in a similar * manner to that described above for the area. The width of the peak r and the width determined for the peak bs are compared so as to determine the maximum width, *,** and the width for the peak r is selected as the maximum. *...

* At step S70 the signal to background ratio (SB) is determined according to equation (12) below: ** I

1.: SB -Peaklntensty-Background (12)

Background

where (Peakintensity -Background) is the value in the data set Si,, for the final iteration of processing, generated in the determination of the set B50, at the energy corresponding to the energy of the maximum M, 15. That is, the value (Peaklntensity -Background) corresponds to the height of the line 13 in Figures 5A to 5E at the particular energy and after n iterations. The value Background is the value of the background signal at the energy corresponding to the energy of the maximum M and corresponds to the height of the line 12 in Figures 5A to 5E at the particular energy and after n iterations.

The values determined at steps S66 to S70 may be determined in any convenient order and by any suitable calculation.

At step S71 it is determined if there are peaks in the set A that have not been processed. If it is determined there are more peaks to be processed then at step S65 the next Peak r+2 corresponding to a maximum M+2 and two minima m+1, m+3, is selected and processed according to steps S66 to S70. If it is determined at step S71 that all peaks have been processed then at step S72 the values determined at steps S66 to S70 for each of the r1 peaks are processed to determine the maximum of each of the values across the i peaks.

At step S73 each of the values determined for each of the peaks r by the processing of steps S65 to S71 are normalised by dividing each value by the corresponding maximum value determined at step S72. For example, if the maximum area determined from the peaks in the set LF has a value Maxarea, each of the area values y corresponding to a respective peak is divided by the value Maxarea. All other parameters are processed analogously. At step S74 the peak score is determined for * each of the peaks by summing the normalised values generated at steps S66 to S70.

Each normalised value has a maximum value equal to 1 and the maximum peak ***.

score for each peak therefore is 5, given that five values are generated for each peak. * S

In an alternative embodiment the score for each peak may be determined by a weighted combination of features. The weights may be determined based upon a training set of data determined from known substances on the particular analyser.

The weights may be determined using, for example, factor analysis which is described in further detail below. Using weights in this way allows the peak score to be optimised, and where training sets of data determined from the particular analyser are used to determine the weights, the peak score may be optimised for the particular analyser. Some of the weights may, for example, take a value of zero such that particular features are not considered in the determination of the peak score, for example where it is determined that a particular feature is not useful in the determination of the peak score. Additionally, analysis of the data in this way allows other features to be used, where it is determined that such features are useful in determination of peak score for a training set of data. Determination of weights is discussed in further detail below.

As indicated above, at step S8 of Figure 3 peaks in the set A with energy less than 100eV are processed to determine whether there is a peak in the set B<100 with matching energy within a tolerance. For any peaks in the set A that do not have a matching peak in the set B<100, the peak in the set A is discarded and not considered in any of the subsequent processing.

A set of significant peaks SP is generated from a combination of B50 peaks and peaks in the set B<100. The set of significant peaks SP is used to identify elements that may be present in the specimen. Processing to identify the set of peaks SP will now be described with reference to Figure 11.

At step S80 the set of significant peaks SP is initialised by adding all of the peaks in the set B<100 to SP. At step S81 a counter variable n is set to 0. The counter variable n indicates the number of peaks in the set SP. At step S82 a peak bs in the set B50 is selected such that the peak bs is the highest scoring peak that has not previously been selected. Peak scores for peaks in the set B50 may be generated by summing normalised values for each of width, standard deviation, area, height and peak shape determined at step S5A of Figure 3. The values are normalised in a similar manner to the normalisation of steps S72 and S73 of Figure 9 by first determining the maximum of each of width, standard deviation, area, height and shape factor for the processed peaks in the set B50 and for each peak dividing each value by the corresponding maximum value.

At step S83 a check is performed to determine whether the normalised values for the selected peak bs are greater than defined minima. Suitable values for the defined minima are as follows, although other values may be used, for example values may be determined using a training process such as factor analysis as described below: Normalised width > 0.12; Normalised standard deviation > 0.022; Normalised area > 0.011; Normalised height> 0.03; and Normalised shape factor> 0.36, If the normalised values of the peak bs are greater than each of the above defined minima, then at step S84 the peak bs is added to the set SP, otherwise processing passes to step S82 where a further peak is selected.

At step S85 the counter variable n is incremented and at step S86 a check is performed to determine whether the value n is equal to 15. If the value n is equal to then it is determined that 15 peaks from the set B50 have been added to the set SP and processing continues at step S87. If the value n is not equal to 15 then at step S88 a further check is performed to determine whether there are more peaks in the set B50 to be processed. If there are more peaks to be processed then processing continues to step S82, otherwise processing continues to step S87 where a set of possible elements in the specimen is determined from the set of peaks SP. Whilst it has been described above that a maximum of 15 B50 peaks, as indicated by the check of steps S81 and S86, may be added to the set SP (together with the set of peaks BS<100), other values of n may be used. It should also be noted that less than peaks may pass the filter of step S83, and so less than 15 B50 peaks may be added to the set SP.

I

*.....

* It can be noted that the processing described with reference to Figure 11 *::::* corresponds to steps S9 and SlO of Figure 3. *

* The set SP is processed together with known element data to identify a set E of S...

elements that may be present in the specimen. That is, the peaks in the set SP are used as a basis for the identification of elements which may be included in the analysed sample. For each of the peaks in the set SP the energy value of the respective peak is compared with known data which includes the energy values of peaks associated with particular elements. If the energy value of a peak in the set SP matches the energy value of a peak associated with a particular element e within a tolerance, the element e is added to the set of elements that may be present in the specimen E. The identification of the set of elements E may be performed by a database query. It shQuld be noted that each peak in the set SP may identify more than one possible element and as such the number of elements identified may be greater than the number of peaks in the set SP.

The structure of the known data will first be described with reference to Figures 12 and 13, and the database search to identify the set of elements E will then be described with reference to the known data.

An Auger table 25 stores data on spectral peaks. The Auger table 25 has an ID field which acts as its primary key and an element ID field which identifies a record in an element table 26 which is associated with a particular record in the Auger table 25.

The Auger table 25 additionally has a transition type field, an intensity field, an energy field, and a type code. The energy field stores the energy of the known peak and is used to match known peak data to identified significant peaks by comparing the energy values within a tolerance of 10eV, although other tolerances may be suitable, The type code field stores a code that identifies properties of the known peak and is described in further detail below with reference to Figure 13. The transition type field and intensity field act as references to a Transition types table 27 and a peak intensity table 28 respectively.

Each of the records in the Auger table 25 is associated with one record in the element table 26. Each record in the element table 26 is associated with one or more * records of the Auger table 25. Relationships between records of the Auger table 25 * ***.S * S and the element table 26 are defined by values of the ElementlD field of the Auger table 25 which reference the ID field of the element table 26. *.**

S

* Each record of the element table 26 stores data associated with a particular element.

It should be noted that a particular element record will generally be associated with a plurality of records in the Auger table 25, being the peaks which are characteristic of that element.

The transition type table 27 stores data associated with the transition in the element that caused the particular peak, based upon the orbitals of the emitted electrons.

When an electron is removed from an inner-shell of an atom such as the K shell, an electron from a higher energy shell may fall into the gap left from the removed electron resulting in a release of energy. The release of energy may be transferred to another electron which is also ejected from the atom. Each transition is therefore determined by the energy shell of the initially removed electron, the energy shell of the electron moving to fill the gap and the energy shell of the electron which is also ejected from the atom. Each peak associated with an element results from a particular transition. The orbital associated with the initially removed electron may be one of a number of shells and each of the electron moving to fill the gap and the electron that is also ejected from the atom may additionally be from one of a number of shells. A number of different transitions are therefore generally possible for a single element. Transition type influences the peak shape and may be considered when identifying peaks.

The peak intensity table 28 has a level field indicating the expected relative height of the peak compared to the maximum peak height in the data, a value indicating a percentage range within which the peak height relative to the maximum peak height may lie, and a description field. The description field is a textual description of the intensity level which may be used to provide information to the user.

Referring now to Figure 13, a schematic illustration of a type code corresponding to the TypeCode field of the Auger table 25 of Figure 12 is shown. The type code provides data in a convenient and easily searched form, although any convenient way of storing the same data could be used. The type code uses hexadecimal digits to store data relating to a peak. Three digits 30 store a first condition associated with * ****.

the peak and three digits 31 store a second condition associated with the peak. The first and second conditions may indicate, for example, a range of voltages applied to the hyperbolic electrode 9 of Figure 2 for which a peak can be detected by the electron detector 1. For example, the range may be defined by a relatively low voltage represented by the three digits 30 and a relatively high voltage represented by the three digits 31. It will of course be appreciated that more than two conditions may be stored, for example a range of voltages and a range of beam energies for which a peak can be detected by the electron detector 1 may be stored.

The remainder of the type code stores data indicating properties of a peak that may be expected when the first and second conditions are satisfied. More specifically, a digit 32 indicates a detectability type, a digit 33 indicates a shape type and two digits 34 indicate a minimum signal to background ratio. The digits 32, 33, 34 indicate attributes associated with a known peak that may be required to be satisfied in order that a peak identified from a specimen is identified as indicating the presence of an element in the specimen. The attributes 32, 33, 34 will be described in further detail below with reference to processing to determine whether an element is present in a specimen.

The code for each peak may be determined by supervised classification using known data relating to elemental peaks and experimental data determined on a training set of elements. For example, spectra may be generated from specimens containing known compounds using the analyser 1, and the spectra may be manually labelled based on known information relating to the known compounds in the specimens and other information such as the voltage applied to the analyser. The manually marked spectra may then be automatically processed and classified into pre-defined types, for example indicating that the peak is detected at a particular voltage, andlor that the peak has a particular shape, such as that the peak is a shoulder peak appearing as a peak on a slope of a parent peak, or that the peak forms a plateau peak together with other peaks of the same element. The peak code can be used to group peaks with similar properties from different elements, for example so that peaks with similar properties which may be processed in a similar manner can be more efficiently processed.

S

a.....

* As indicated above, at step S87 of Figure 11 a database search is carried out to S *SS*S * 5 determine a set of elements E. The database search is based upon a comparison of the experimental data obtained from the specimen and known data stored for each S...

element that may be present in the specimen, and the set of peaks in the set SP is processed together with the known data to determine a set of possible elements in the specimen. The processing is based upon a comparison of the energy value of peaks in the set SP at the point having maximum intensity in the peak, with the Energy value of each Auger table 25 (indicating a peak associated with an element) whose TypeCode has a detectability type value that indicates that the peak associated with the Auger table 25 is expected to be detected for the conditions indicated by the TypeCode which correspond to the conditions under which the experimental data was obtained. In general terms, an element e is added to the set E if an energy value associated with a peak in the set SP matches an energy value associated with an Auger table 25 associated with the element e that has a detectability type indicating that a peak is expected to be detected, within a threshold.

The processing of Figure 11 described above identifies elements that may be present in the specimen, based upon the stored data described with reference to Figures 12 and 13. Identifying a subset of all the possible elements based upon the most significant peaks in a spectrum allows the search space to be considerably reduced.

Additionally, the TypeCode field of the Auger table 25 allows only peaks that are likely to be present in spectral data, generated under the same conditions as the experimental data, to be considered, which further reduces the search space. The spectral data is then further processed in greater detail considering all of the peaks present in the spectrum to determine a likelihood that each of the elements determined by the processing of Figure 11 is in fact present in the specimen (although in some cases the likelihood will be zero, indicating that the element is not present in the specimen). It can be noted that whilst the processing described above to determine the set of elements E has been with reference to the set of significant peaks SP generated from both sets of background subtraction peaks (i.e. B<100 and B50) , the subsequent processing is performed only with reference to the set of peaks A and the set of elements E. Figure 14 is a class diagram showing classes used for determining a likelihood that an element in the set E, identified by the processing described above with reference to Figure 11, is present in the specimen. At a high level, a Matcher class 40 stores * data relating to experimental peaks in the set A together with stored data for a plurality of elements. For each element in the set of elements E a corresponding instance of an Elementlnfo class 42 is retrieved. Each instance of the Elementlnfo class 42 has a set of associated PeakMatch classes 43 corresponding to theoretical S...

ess peaks associated with the Elementlnfo class 42. As described in further detail b&ow with reference to Figure 15, experimental peaks in the set A are compared to instances of the PeakMatch classes associated with the retrieved instances of the Elementlnfo classes corresponding to elements in the set of elements E. If an instance of a PeakMatch class has an energy variable within a threshold of an energy of an experimental peak (taking into account chemical shifting), then the IsMatched variable of the matching PeakMatch class 43 is set to true, and a Peak object 43A of the PeakMatch class 43, corresponding to the matching experimental peak, is filled. Each instance of the Elementlnfo class has a HashTable 44 of MatchStat classes 45. Each instance of the MatchStat class 45 stores data indicating statistics of properties of the matched theoretical and experimental peaks (corresponding to instances of the classes 43 and 43A).

In more detail, the Matcher class 40 includes a Store variable that stores elements which are to be considered, that is, the elements in the list of possible elements in the specimen E. A Tolerance variable indicates a level of tolerance allowed by the system within which the energy of a theoretical peak and the energy of a peak in the set A must lie in order that the two peaks are to be identified as a match. The tolerance may be input by a user or may be predefined in the system. An Elem Matches variable references an instance of the Hashtable class 41. The Hashtable represented by the instance of the Hashtable class 41 stores a plurality of instances of the Elementlnfo class 42. Each instance of the Elementlnfo class 42 stores data relating to a single element in the set of elements to be considered. The Matcher class 40 additionally has a MatchPeaks function that takes the set A as a parameter and performs matching of peaks in the set A with instances of the PeakMatch class 43 associated with elements stored in the Store variable (i.e. the elements in the set E).

Each instance of the Elementlnfo class includes an array list APMatches of peaks * that may be present in analysis of a specimen including the relevant element. Each instance of the Elementlnfo class further has a plurality of data fields storing known data associated with the particular element that may be used to determine whether **** the element is present in a specimen based upon the spectral data derived from the *..*S* * specimen during experimental analysis. **** * * ****

Each peak in the array list APMatches is an instance of the PeakMatch class 43.

Each instance of the PeakMatch class 43 includes a Boolean variable IsMatched.

The variable lsMatched indicates whether the respective peak has been matched with a peak of the experimentally obtained data. A variable Error indicates the difference in energy between the energy value IdealEnergy and the energy value of the matched peak in the experimentally obtained data. A variable IntensityLevel indicates the expected intensity of the peak relative to the maximum intensity for peaks of the element indicated by the instance of the Elementlnfo class 42 associated with the instance of the PeakMatch class 43. For example, a peak associated with an element that is expected to be highest will have a IntensityLevel value of 1, indicating that it is expected to be the highest peak. A Peak variable references a Peak object 43A storing information relating to the peak in the experimentally obtained data which has caused the match for the respective instance of the PeakMatch class 43.

A ScoreRank variable in the PeakMatch class 43 indicates the rank for the experimentally obtained peak responsible for the match, the rank being based upon a peak score generated by the processing of Figure 9. The rank is generated by sorting the list of peaks A in descending order using the compareTo operation of the Peak object 43A. The variable PeakType corresponds to the TypeCode of Figure 13 and the variable TransType corresponds to the Transition types object of Figure 12.

The Elementinfo class references a HashTable class 44, which is instantiated to generate a HashTable object storing instances of the MatchStat class 45. Each instance of the MatchStat class 45 stores data relating to a particular parameter associated with each PeakMatch object referenced by an Elementlnfo object. More specifically, for each parameter, the respective MatchStat object stores average, maximum and minimum values (although other values such as mode, median and standard deviation maybe stored) for that parameter based upon the PeakMatch *.S...

* objects associated with the respective Elementinfo object. A MatchStat object is instantiated for each of the variables shown in Table 4 below, although it will be appreciated that other combinations of features could be used: ** ** ****** _________________________________________________ * Normalised Peak Width Normalised Peak FWHM **** *:*. Normalised Peak Right Hand Height Normalised Peak Height Normalised Peak SB Ratio Score Rank Normalised Peak Area Error in Energy Position Gap Error Between Peaks

TABLE 4

It will be appreciated that some of the data to instantiate the classes of Figure 14 can be obtained from data stored in the database tables of Figure 12.

Processing carried out to match experimentally obtained peaks to PeakMatch objects associated with stored element peaks will now be described with reference to Figure 15. In general terms, the processing of Figure 15 determines for each experimental peak whether there is a corresponding theoretical peak associated with one of the elements in the set of possible elements E. Each experimental peak may be associated with more than one theoretical peak and as such, each experimental peak is compared with all theoretical peaks to determine if there is a match.

At step S95 the set A is input and at step S96 the highest scoring peak p in the set A that has not previously been selected is selected.

At step S97 a subset of elements in the set E, E, is determined where elements in the set E are elements having a peak which may be a match for the peak p. The set of elements E may be determined using a database query which compares the *.....

* energy value of the peak p with energy values of theoretical peaks associated with * elements in the set E. The comparison of energy value may take into account s..., chemical shifting. An element in the set E is included in the set E if and only if the *...

energy value of the peak p lies within a predetermined threshold of at least one peak * associated with the element. Where chemical shifting is allowed, the threshold may .... be based upon the maximum and minimum allowed chemical shift for an element *...

(corresponding to MaxShift and MinShift fields of table 26 of Figure 12).

At step S98 an element e in the set of elements E is selected. At step S99 a peak tp corresponding to an instance of a PeakMatch object 43 for the element e selected at step S98 is selected such that the energy value of tp is closest to the energy value of p. Identification of the closest peak tp to the peak p may be carried out by looping through all peaks associated with element e.

At step S100 a check is performed to determine if the peak tp has previously been matched with one of the peaks in the set A. The check of step S100 can be easily carried out based upon the variable IsMatched of the PeakMatch object 42 associated with the peak tp. If it is determined that the peak tp has not been previously matched then processing passes to step SlOl where the PeakMatch object is updated by adding data to the Peak variable which is associated with the peak p, and updating the IsMatched variable to a value of TRUE and the Error variable to indicate the difference between the energy value of the peak p and the energy value of the peak tp, and the ScoreRank variable to the rank of the peak p. If it is determined at step S100 that the peak tp has previously been matched, based upon the IsMatched variable, a peak p responsible for that match is identified based upon the Peak variable. Then, at step S102 the peak score S of the peaks p and pp, determined above with reference to Figures 9 and 10, are compared. The peak with the highest peak score is determined to be the matching peak and the PeakMatch object is updated, if necessary, by appropriate update of the Peak variable.

Processing from step SlOl passes to step S103 where it is determined whether the peak p is the first peak selected from the set A matched for a particular element e, and whether chemical shifting is to be considered for the element e. If it is the first peak to be considered for the element e (and therefore the highest scoring peak in S.....

* S the set A for the element e) and chemical shifting is to be considered then processing passes to step S104 where the chemical shift is determined. The chemical shift is determined based upon the Error value of the PeakMatch class 43, which indicates *.** the difference between the energy value of the peak p and the energy value of the S.....

* peak tp. Subsequent peaks tp for the element e will then take the calculated shift into account at step S99. That is, energy values for peaks associated with an element e *..* . : that has been determined as an element in which peaks have been chemically * shifted are determined as a theoretical energy value stored for the theoretical peak plus or minus the chemical shift value determined based upon the highest scoring peak for the element e.

Processing from each of steps S102, S103 and S104 passes to step S105. At step S105 a check is performed to determine if there are more elements in the set E with respect to which the currently selected peak has not been processed. If it is determined that there are more elements to be processed with respect to the peak p then processing passes to step S98 where a further element e is selected. If the peak p has been processed with all possible elements in the set E and all peaks tp associated with each of the elements in the set E (as determined by the checks of steps S104 and S105) then processing passes to step S106 where a check is performed to determine if there are more peaks to be processed. If it is determined that there are more peaks to be processed in the set A then processing passes to step S96 where a further peak p is selected and processed according to steps S97 to S105. The processing of Figure 15 to match experimentally obtained peaks to theoretical peaks may be carried out in any convenient way, for example by performing database queries to return the closest theoretical peak corresponding to each experimental peak.

As indicated above, where chemical shifting is considered, the comparison of peaks takes into account possible chemical shifting of the experimental data. Elements in different chemical states in a specimen, such as different valencies, can have slightly different energies for peaks identified by the electron detector to those stored in the known data described above. This phenomenon is known as chemical shifting, and information on chemical shifting can be stored in the peak information database as described above with reference to Figure 12. The Elements table 26 of Figure 12 has fields Max Shift, Mm Shift and Allow Shift. The value Allow Shift indicates if chemical S.....

* shifting is allowed for the particular element associated with the Element table 26, and the values Max Shift and Mm Shift indicate an amount of energy by which the energy of peaks associated with the element can be shifted in a positive and negative direction respectively. The value Max Shift and/or the value Mm Shift may S.....

* S be zero so that the shift may only be positive or negative respectively. Chemical shift values are then used when matching peaks to allow a further tolerance to that *..

defined in table 40 of Figure 14.

If an element is identified as having been subjected to a chemical shift the highest scoring peak added to the element is used to determine the actual chemical shift undergone by the element. All other identified peak matches added to the element then use this actual shift value when finding the nearest peak of the element. That is, at step S98, identification of a peak tp that is nearest to the peak p determines the peak tp having energy value that is closest to the energy value of p after adding the chemical shift value (which may be either positive or negative) determined at step S103. Chemical shifting data may be generated by any convenient method such as by adding a database relation between element and peak tables.

Processing to determine a likelihood that an element is present in a specimen will now be described with reference to Figure 16. At step SilO an element e in the set E is selected for processing. At steps Si Ii to SI 13 checks are performed to determine if criteria associated with the element e are satisfied by the experimentally obtained data and associated peaks generated by the processing previously described. The criteria that must be satisfied are a rating criteria check, a gap error criterion and a peak shape check. These criteria are described below.

At step Sill a rating criteria check is performed. Each PeakMatch object 43 of Figure 14 associated with an Elementinfo object 42 for a particular element has an associated detection type variable, stored as part of the type code of Figure 13. The detection type variable indicates how likely a particular peak is to be present and detected in experimentally obtained data if the associated element is present in a specimen. The detection type variable may indicate that a theoretical peak must be detected for an element to be considered present in a specimen. For each peak match tp that has a detection type value indicating the peak must be present, the check of step Sill is performed to determine if a corresponding peak has been * : detected in the experimentally obtained data. This check can be performed by checking the value of the IsMatched variable in the PeakMatch object associated **** with the peak tp. If any peak tp with a detection type value indicating a peak must be detected does not have an associated peak in the experimentally obtained data, then * the element is considered to not be present in the specimen and processing passes from step Sill to step S115 of Figure 16. If the check of step Sill is satisfied, S...

processing continues at step S112.

At step S112 a Gap Error criterion is checked. A gap error indicates the difference between the gap between any two peaks in the experimentally obtained data associated with a particular element and the gap between the corresponding peaks tp for that element. As such, where an element does not have two or more peaks tp no gap error can be determined, and processing passes directly to step Si 13. The gap error can be calculated according to equation (13) below: GapError = Abs(((pl) -(p2)) -((tpi) -(tP2))) (13) where p1, p2 are the energies associated with two peaks in the experimentally obtained data and determined to be associated with a particular element by the processing of Figure 15 and tPl, tP2 are the energies of the corresponding peak matches associated with the element. Differences between detected peaks and theoretical peaks can be a result of chemical shifting due to a valence change or due to experimental error, however the gap between the detected and theoretical peaks should remain reasonably consistent across peaks for the same element and transition type due to the same or similar shift being applied to each peak. The average gap error for an element across all pairs of peaks of the same transition type is therefore a good indication of whether or not there is a true match between peaks tp for a particular element and peaks in the experimentally obtained data. A maximum acceptable gap error can be set on a per transition type per element basis and the element will then be identified only if the processing generates an average gap error which is less than or equal to the maximum acceptable gap error. If the element does not have an average gap error less than or equal to the maximum acceptable gap error then the element is considered to not be present in the specimen and processing passes from step S112 to step S114, otherwise, * processing continues at step Si 13. *I*** * * *

* At step S113, a peak shape check is carried out. If an element has a high ranking peak with energy below 100eV then all other theoretical peaks with a high detection S...

* type (that is a peak that must be detected in order that the element can be *5S** * considered present in the specimen) should also have a signal to background ratio greater than or equal to a defined minimum signal to background ratio. The defined *:*. minimum signal to background ratio may be stored in the peak code of Figure 13.

This condition prevents a prominent high ranking peak from dominating the scoring results and leading to false positive element matches, where other peaks for the element do not indicate a match. If the peak shape check is not satisfied then processing passes from step S113 to step S115. Otherwise, processing continues at step S114. It will be appreciated that other properties may be considered. For example, where an element has plateau peaks, a check may be carried out to determine whether a group of peaks associated with the element lie within a certain energy range, and if the check is satisfied precise matching for each peak may then not be required.

If each of the checks set out above is satisfied then processing passes to step S114 where an element score is generated for element e. The element score is generated according to the values set out below in Table 5: Term 1. (No. Matched/Max for Elem) * (Max for Elem/Max No. Overall) 2. No. Matched/Max for Elem 3. 1-(Mean Error/Match Tolerance 4. 1-(Min Error/Match Tolerance) 5. Mean Score/5 6. Max Score/5 7. 1/(Mean Rank) 8. 1/(Min Rank) 9. (Num Hi Rank)/(Max No. Overall)

TABLE 5

where: No. Matched is the number of instances of the PeakMatch class 43 of Figure * 14 associated with the element e for which the IsMatched variable is equal to TRUE; Max for Elem is the number of instances of the PeakMatch class 43 ***S associated with the element e. Max for Elem considers only those instances of the PeakMatch class associated with the element e that are within the energy range that the electron detector is able to detect (determined by the voltage applied to the hyperbolically charged electrode) and only considers instances of the PeakMatch class associated with the element e which are reasonably expected to be detected, as determined by the peak code; Num Hi Rank indicates the number of instances of the PeakMatch class associated with the element e that have an associated experimentally obtained peak that has a peak score in the top n rankings, where n may be, for example 10; and Max No Overall is the maximum possible number of peak matches on a per element basis over all elements in the database identified as possible matches (i.e. in the set E) within the energy range of the spectrum data and with a detection type above a predefined threshold.

The term Max No Overall in term 1 compensates elements that have a high number of peaks to match within the energy range of the spectrum in comparison to those that have a low number of peaks within the energy range, and which are therefore easier to match.

Peaks that are within the energy range that the detector is able to detect can be determined using the Energy associated with each theoretical peak as shown in Figure 12. When considering whether a peak may be in the energy range of the spectrum, a matching tolerance is included so as not to exclude peaks near the edge of the energy range due to experimental error, chemical shift or other factors.

Each of the values in Table 5 are multiplied by an associated weight. The weights are chosen such that the sum of all of the weights is equal to 1, and therefore summing the weighted values of Table 5 gives a score in the range 0 to I indicating a likelihood of the presence or absence of the element e in the specimen 2 of Figure 1.

The weights may be determined using a training process such as factor analysis, as described below. Processing continues at step S115 where it is determined if there * are more elements to be processed. If it is determined that there are more elements ***.*.

* * to be processed then at step Si 10 a further element e is selected and the processing of steps Sill to Si 15 is repeated until all elements have been processed. The element scores for each element may be thresholded to generate an indication of the * elemental composition of the specimen 2. * .** * * ****

*:*. It has been indicated above that a training process may be used to determine weights. Training processes generally use training sets of data which are manually marked to indicate true positive peaks which are indicative of the presence of an element in the sample used to generate the training set of data. The training sets of data are generated using the same apparatus as the apparatus for which the weights are generated so that the generated weights are optimised for the apparatus. The peaks are scored based upon a plurality of different factors and the peaks are ranked based upon their associated scores. It is desirable that the true positive peaks are ranked higher than false positive peaks, and, given that the true positive peaks are known, weights may be generated such that the contribution of each factor to the score of each peak results in an optimal ranking, in which true positive peaks are ranked higher than false positive peaks. Weights may be determined using any suitable method such as, for example, factor analysis, although other methods of determining weights may be used such as Neural Networks and Decision Trees.

The training method described above may be used in the generation of peak scores at step S74 of Figure 9 and in the generation of element scores of step SI 14 of Figure 16.

It will be appreciated that the time for which data is acquired from a specimen may affect the quality of the data, and accordingly the reliability of any determination from the data. For example, longer acquisition times can result in features which are often not present after shorter acquisition times being present in data. Additionally, shorter acquisition times often also result in noisier data which can cause other features to be less easily detected. However, longer acquisition times require greater use of resources, such as use of an acquisition apparatus, which results in an overall reduction in the number of specimens that can be tested given a finite resource. As such, optimisation of acquisition time is desirable. One method of determining an optimal acquisition time is by analysing a specimen which contains a known element *..�** * and continuously calculating peak scores for peaks which are expected to be present in the data based on the presence of the element in the specimen. Alternatively, a *.e. score for the known element may be continuously calculated. An optimal acquisition time may be determined based upon the peak or element scores by determining an **...S * acquisition time at which the score is maximal. The optimal acquisition time may then be used as the acquisition time used to analyse specimens for which the elemental I..

*:*. composition is unknown.

Alternatively, continuous monitoring of element scores may also be used to determine when acquisition can be terminated for an unknown specimen. For example, where a score for a particular element exceeds a threshold value, it may be determined that the confidence that the particular element is present in a specimen is high enough, and data acquisition may be terminated.

The specimen may be a plurality of specimens or a specimen evaluated at different locations so as to determine when materials have reached an acceptable purity, for example when a particular element has been removed from a surface layer using sputtering. Continuously monitoring the spectra produced from the materials allows a determination of when the materials have reached an acceptable purity, for example when the amount of an element in the material has reduced to an acceptable level.

For example, where a specimen contains an impurity such as Carbon, the specimen may be subjected to a sputtering process so as to remove the impurity from the specimen. Continuous analysis of the specimen allows a determination of the level of Carbon in the specimen to be determined at a plurality of time points, and when the level has decreased sufficiently, the sputtering process may be terminated.

An estimate of noise in the spectrum may be generated from analysis of a specimen, and an optimal acquisition time may be determined based upon a time at which noise is minimal or the rate of decrease in noise approaches a minimum. Determination of noise provides a further method of optimising acquisition time, as well as a method for comparing different data sets, for example in a quality control process. Noise in acquired data may be determined by first performing dark current subtraction and gain correction on the spectral data. Where the specimen is a known specimen an area of the spectrum which is known to not contain peaks, based upon the composition of the known specimen, is selected. Alternatively, where the specimen * * 1 has an unknown composition, background subtraction may be used to identify areas which do not contain peaks by identifying regions of the curve in which variation from the background does not exceed a threshold. A polynomial is fitted to the area which does not contain peaks and the root mean square error (RMSE) of the polynomial is * determined using standard methods. The RMSE is normalised by dividing the error of the fit at each channel by the intensity value of the data at each channel. A plot of the I...

*. : normalised RMSE against the acquisition time or against the square root of (beam * current * acquisition time) is generated and from the generated plot an optimal acquisition time can be determined. Where the specimen is a known specimen, the determined optimal acquisition time may be used to estimate an optimal acquisition time for analysis of unknown specimens. The plot of the normalised RMSE could also be used to provide an indication of noise levels during acquisition, which could be used to indicate a possible fault in the acquisition instrument.

A further method of determining optimal acquisition time is using Receiver Operating Characteristic (ROC) curves. Sensitivity and specificity may be determined for different data acquisition times on known samples and the time at which sensitivity and specificity are optimal may be determined. The time at which sensitivity and specificity are optimal for a known sample may be used as a time for which data is acquired for a specimen to be analysed.

The above description indicates that known data associated with elements is used.

The known data may be generated by processing specimens of known elemental composition to generate data or, alternatively, the data may be determined from existing data on elemental spectral peaks. Generating known data by processing specimens will generate known data that is optimised for particular apparatus.

Although various embodiments of the invention have been described above, it will be appreciated that various modifications can be made to the described embodiments without departing from the spirit and scope of the invention. Indeed, the foregoing description should be considered in all respects illustrative and not limiting.

S

S..... * .

**.**. * . *S.. * S *... * * . S... * * S. * * .*

Claims (1)

1. CLAIMS1 A method for determining the presence or absence of an element in a sample comprising: receiving data generated from said sample; obtaining stored data relating to said element, said stored data comprising at least one mandatory reference data point and at least one optional reference data point; determining a relationship between said stored data and said received data; and determining the presence of said element in said sample if said at least one mandatory reference data point has a predetermined relationship with said received data, said determining being independent of any relationship between said at least one optional reference data point and said received data generated from said sample.

   2. A method according to claim 1 further comprising determining a confidence of the presence of said element in said sample based upon a relationship between said at least one optional reference data point and said received data.

   3. A method according to claim 1 or 2, wherein said received data comprises a plurality of received data points, and determining a relationship between said at least one mandatory reference data point and said received data comprises: performing an identification operation between each of said at least one mandatory reference data points and said plurality of received data points. * *

   4. A method according to claim 3, wherein each of said at least one mandatory reference data points and each of said received data points has an associated SI: energy value and said identification operation is based upon said associated energy * values. I... * S S...

   *:*. 5. A method according to claim 4, wherein said identification operation identifies a particular received data point in said data if and only if said energy value associated with said particular received data point is within a predetermined tolerance of an energy value associated with at least one mandatory reference data point.

   6. A method according to claim 5, wherein said element is determined not to be present in said sample if said identification operation does not identify a received data point for each of said at least one mandatory reference data points.

   7. A method according to any preceding claim, wherein said at least one mandatory reference data point and said at least one optional reference data point associated with said element are determined according to a training process.

   8. A method according to any preceding claim, wherein said stored data relating to said element comprises a plurality of data points and wherein the method further comprises: obtaining at least one parameter of an apparatus used to generate said data from said sample; and identifying said at least one mandatory reference data point and said at least one optional reference data point based upon said at least one parameter.

   9. A method according to any preceding claim, wherein said received data is spectral data.

   10. A method according to claim 9, wherein said data points are peaks in said spectral data.I* 11. A method according to claim 9 or 10, wherein said received data is data : generated by a charged particle energy analyser. ****I. 12. A method according to claim 11, wherein said charged particle energy *1.S..

   * analyser is selected from the group consisting of: an Auger mass spectrometer, an X- --,*-* *-*------------------RayphotoeIectron-spectroscope-and-an-Electron--Probe- Micro-AnalyserT----------------*1*I ** I

   I

   * 13. A computer program comprising computer readable instructions configured to cause a computer to carry out a method according to any one of claims I to 12.

   14. A computer readable medium carrying a computer program according to claim 13.

   15. A computer apparatus for determining the presence or absence of an element in a sample, the apparatus comprising: a memory storing processor readable instructions; and a processor arranged to read and execute instructions in said memory; wherein said processor readable instructions comprise instructions arranged to control the computer to carry out a method according to any one of claims 1 to 12.

   16. Apparatus for determining the presence or absence of an element in a sample, the apparatus comprising: means for receiving data generated from said sample; means for obtaining stored data relating to said element, said stored data comprising at least one mandatory reference data point and at least one optional reference data point; means for determining a relationship between said stored data and said received data; and means for determining the presence of said element in said sample if said at least one mandatory reference data point has a predetermined relationship with said received data, said determining being independent of any relationship between said at least one optional reference data point and said received data generated from said sample.

   17. A method for generating a score for a feature of data generated from a * sample, the method comprising: receiving said data generated from said sample; identifying said feature in said received data, said feature having a plurality of 1: associated data points in said received data; 0.***.* processing at least two of said plurality of data points associated with said -------feature--to-determine-atleast onepropertyDfsaidfet[;and -----generating said score for said feature based upon said at least one property.

   18. A method according to claim 17, wherein said received data is defined by a plurality of data points each defining a value on a first axis and a value on a second axis, said plurality of data points in said received data associated with a particular feature having values on the first axis such that the data points associated with the feature are adjacent to one another.

   19. A method according to claim 17 or 18, wherein said at least one property of each of said plurality of features is selected from the group consisting of: an area of the feature, a width of the feature, a standard deviation of the feature, a signal to background ratio of the feature and a shape factor of the feature.

   20. A method according to claim 17, 18 or 19, wherein two of said plurality of data points associated with each feature are associated with local minima in said received data and said at least one property of each of said plurality of features is based upon said local minima.

   21. A method according to claim 20, wherein each of said plurality of data points has an associated value, and wherein said at least one property of each of said plurality of features is determined based upon a difference between said values associated with said data points associated with local minima.

   22. A method according to any one of claims 17 to 21, wherein said at least one property for a particular feature is based upon all of the plurality of data points associated with the particular feature.

   * 23. A method according to any one of claims 17 to 22, further comprising: generating background subtraction data from said received data, said background subtraction data comprising a plurality of background subtraction data points; and ***** * determining a corresponding one of said background subtraction data points I.- , : wherein said at least one property of each of said plurality of features is further based upon said corresponding one of said background subtraction data points.

   24. A method according to claim 23, wherein each of said plurality of features has an associated value and said determining a corresponding one of said background subtraction data points for a particular feature is based upon said associated values.

   25. A method according to claim 24, wherein each of the background subtraction data points has a plurality of associated data points, each associated data point having an associated value, and determining a corresponding one of said background subtraction data points for a particular feature comprises determining a background subtraction data point having at least one data point having an associated value greater than said value associated with said particular feature and at least one data point having an associated value less than said value associated with said particular feature.

   26. A method of determining a subset of a plurality of features in data generated from a sample, the method comprising: generating a respective score for each of said plurality of features according to any one of claims 17 to 25; and determining said subset of said plurality of features in said data based upon said generated scores.

   27. A method of determining chemical shift data associated with an element based upon data generated from a sample, the method comprising: generating a respective score for each of a plurality of features in said data generated from said sample according to any one of claims 17 to 25; selecting one of said plurality of features based upon said generated scores; * : obtaining stored data relating to said element, said stored data comprising at least one data point associated with said element; and determining said chemical shift data associated with said element based upon said selected one of said features and said at least one data point associated with * said element. * #

   *. : 28. A method according to any one of claims 17 to 25, further comprising: obtaining stored data relating to an element, said stored data comprising a plurality of data points; determining whether a relationship between said stored data and said feature is satisfied; and if said relationship is satisfied, generating a quantitative indication of the presence or absence of said element in said sample based upon said score associated with said feature.

   29. A method according to any one of claims 17 to 28, wherein said received data is spectral data.

   30. A method according to claim 29, wherein said features are peaks in said spectral data.

   31. A method according to claim 29 or 30, wherein said received data is data generated by a charged particle energy analyser.

   32. A method according to claim 31, wherein said charged particle energy analyser is selected from the group consisting of: an Auger mass spectrometer, an X-Ray photoelectron spectroscope and an Electron Probe Micro Analyser.

   33. A computer program comprising computer readable instructions configured to cause a computer to carry out a method according to any one of claims 17 to 32.

   34. A computer readable medium carrying a computer program according to claim 33.

   35. Apparatus for generating a score for a feature of data generated from a sample, the apparatus comprising: a memory storing processor readable instructions; and a processor arranged to read and execute instructions in said memory; **.** * * wherein said processor readable instructions comprise instructions arranged to --control-the-computer-to-carryoutamethodacordir tcy öfläithi7T6 32 S... S. S

   36. Apparatus for generating a score for a feature of data generated from a sample, the apparatus comprising: means for receiving said data generated from said sample; means for identifying said feature in said received data, said feature having a plurality of associated data points in said received data; means for processing a plurality of said plurality of data points associated with said feature to determine at least one property of said feature; and means for generating said score for said feature based upon said at least one property.

   37. A method for determining a subset of elements from a plurality of elements, said subset of elements indicating elements of the plurality of elements most likely to be present in a sample, the method comprising: obtaining stored data relating to each of said plurality of elements; receiving data generated from said sample, said received data comprising a plurality of data points, each data point having an associated value; processing said received data a plurality of times, each processing being based upon a respective value for a parameter, to generate respective sets of intermediate data points, each intermediate data point having an associated value; processing said respective sets of intermediate data points to determine relationships between intermediate data points in said respective sets of intermediate data points; determining a set of intermediate data points based upon said determined relationships; and determining said subset of elements based upon said determined set of intermediate data points and said stored data relating to each of said plurality of elements.

   38. A method according to claim 37, wherein said value associated with each of said data points is an energy value. *.S. * S

   39. A method according to claim 37 or 38, wherein said determined set of intermediate data points comprises intermediate data points selected from at least --,*-,----------oneof-said-respectivesets-ofintermediatedatapiritsT --S... 0* S

   * 40. A method according to claim 37, 38 or 39, wherein said determined set of intermediate data points is a set of intermediate data points having an energy value less than a predetermined threshold.

   41. A method according to claim 37 to 40, wherein processing said received data to generate each of said respective sets of intermediate data points comprises performing a respective background subtraction process on said received data.

   42. A method according to claim 41, wherein said received data comprises a plurality of received data points and performing said background subtraction process comprises processing pairs of said received data points, each pair of received data points having an associated difference, the difference being based upon the respective value for the parameter.

   43. A method according to any one of claims 37 to 42, wherein processing said respective sets of intermediate data points to determine relationships between intermediate data points in said respective sets of intermediate data points comprises identifying corresponding data points in said sets of intermediate data points.

   44. A method according to claim 43, wherein each of said intermediate data points is a characteristic associated with a feature of said received data, said characteristic comprising a plurality of associated data points, and identifying corresponding data points in said sets of intermediate data points comprises determining whether a predetermined relationship between pluralities of data points associated with characteristics is satisfied.

   45. A method according to claim 44, wherein each of said plurality of data points associated with a characteristic has an associated value, each characteristic having an associated maximum data point having a highest associated value and an associated minimum data point having a lowest associated value, and wherein determining whether a predetermined relationship between pluralities of data points associated with characteristics is satisfied comprises: identifying a first characteristic and a second characteristic as corresponding - firsi characteristic lie between values for a maximum data point and a minimum data point * for the second characteristic.

   46. A method according to any one of claims 37 to 45, wherein each of said intermediate data points is a characteristic associated with a feature of said received data, said characteristic comprising a plurality of associated data points, each of said plurality of associated data points having an associated intensity, and wherein said relationships between data points in said respective sets of intermediate data points is based upon a value of a data point associated with a characteristic having an associated intensity which is maximal.

   47. A method according to any one of claims 37 to 46, wherein determining said subset of elements is based upon at least one data point having an energy value greater than said predetermined threshold.

   48. A method according to any one of claims 37 to 47, wherein said received data is spectral data.

   49. A method according to claim 48, wherein said plurality of intermediate data points are each a peak associated with a peak in said received spectral data.

   50. A method according to claim 48 or 49, wherein said received data is data generated by a charged particle energy analyser.

   51. A method according to claim 50, wherein said charged particle energy analyser is selected from the group consisting of: an Auger mass spectrometer, an X-Ray photoelectron spectroscope and an Electron Probe Micro Analyser.

   52. A computer program comprising computer readable instructions configured to cause a computer to carry out a method according to any one of claims 37 to 51. * S

   53. A computer readable medium carrying a computer program according to *.S.** * * claim 52.-------------------

   54. Apparatus for determining a subset of elements from a plurality of elements, * said subset of elements indicating elements most likely to be present in a sample, the apparatus comprising: a memory storing processor readable instructions; and a processor arranged to read and execute instructions in said memory; wherein said processor readable instructions comprise instructions arranged to control the computer to carry out a method according to any one of claims 37 to 51.

   55. Apparatus for determining a subset of elements from a plurality of elements, said subset of elements indicating elements of the plurality of elements most likely to be present in a sample, the apparatus comprising: means for obtaining stored data relating to each of said plurality of elements; means for receiving data generated from said sample, said received data comprising a plurality of data points, each data point having an associated value; means for processing said received data a plurality of times, each processing being based upon a respective value for a parameter, to generate respective sets of intermediate data points, each intermediate data point having an associated value; means for processing said respective sets of intermediate data points to determine relationships between intermediate data points in said respective sets of intermediate data points; means for determining a set of intermediate data points based upon said determined relationships; and means for determining said subset of elements based upon said determined set of intermediate data points and said stored data relating to each of said plurality of elements.

   56. A method of generating a quantitative indication as to whether an element is present in a sample, the method comprising: obtaining stored data relating to said element, said stored data comprising a number of data points associated with said element; receiving data generated from said sample; and generating said quantitative indication based upon said number of data points associated with said element and a relationship between said received data and said stored data associated with said element. * S

   IS.. . . . : 57. A method according to claim 56, wherein said received data comprises a * plurality of received data points, and said relationship between said received data and said stored data is determined by performing an identification operation between each of said data points associated with said element and said received data points.

   58. A method according to claim 57, wherein each of said received data points and each of said data points associated with said element has an associated energy value, wherein said identification operation is based upon said associated energy values.

   59. A method according to claim 58, wherein said identification operation identifies a particular received data point in said data if and only if said energy value associated with said particular received data point is within a predetermined range of an energy value associated with one of said data points associated with said element.

   60. A method according to claim 59, wherein said identification operation identifies a particular received data point with said one of said data points having an energy value within said predetermined range.

   61. A method according to claim 59 or 60, wherein said quantitative indication is based upon a number of said received data points identified by said identification operation and said number of data points associated with said element.

   62. A method according to claim 59, 60 or 61, further comprising generating a score associated with at least one of said received data points identified by said identification operation, wherein said quantitative indication is further based upon said score.

   63. A method according to any one of claims 59 to 62, wherein said quantitative indication is further based upon a difference between an energy value associated with a received data point identified by said identification operation and an energy S..' value associated with said one of said data points having an energy value with said S.,...* . predetermined range. 5..

   64. A computer program comprising computer readable instructions configured to * .. cause a computer to carry out a method according to any one of claims 56 to 63.

   65. A computer readable medium carrying a computer program according to claim 64.

   66. Apparatus for generating a quantitative indication of the presence or absence of an element in a sample, the apparatus comprising: a memory storing processor readable instructions; and a processor arranged to read and execute instructions in said memory; wherein said processor readable instructions comprise instructions arranged to control the computer to carry out a method according to any one of claims 56 to 63.

   67. Apparatus for generating a quantitative indication as to whether an element is present in a sample, the apparatus comprising: means for obtaining stored data relating to said element, said stored data comprising a number of data points associated with said element; means for receiving data generated from said sample; and means for generating said quantitative indication based upon said number of data points associated with said element and a relationship between said received data and said stored data associated with said element.

   68. A method for determining the presence or absence of an element in a sample comprising: obtaining stored data relating to said element, said stored data comprising a plurality of stored data points associated with said element, each data point associated with said element having an associated energy value; receiving data generated from said sample, said received data comprising a plurality of received data points, each of said plurality of received data points having an associated energy value and an associated score; processing said plurality of stored data points and a received data point S...having a maximum score to determine an energy difference, said energy difference indicating a difference in energy value between a stored data point having an energy -----------value-closest-to-said-receiveddata-pointhaviag niUiorér -- ::. identifying at least one of said plurality of received data points in the received data based upon said stored data and said determined energy difference; and determining the presence or absence of said element based upon said identified at least one of said plurality of received data points.

   69. A method according to claim 68, wherein identifying at least one of said plurality of data points in the received data based upon said stored data and said determined energy difference comprises performing an identification operation between each of said stored data points and said received data points, wherein said identification operation is based upon said associated energy values and said determined energy difference.

   70. A method according to claim 69, wherein said identification operation identifies a particular received data point in said received data if and only if said energy value associated with said particular received data point is within said determined energy difference of an energy value associated with one of said stored data points.

   71. A method according to claim68, 69 or 70, wherein said energy difference is indicative of a chemical shift associated with said element in said sample.

   72. A method according to any one of claims 68 to 71, further comprising initially processing said received data generated from said sample to select said element for which the presence or absence is to be determined.

   73. A method according to any one of claims 68 to 72, wherein said received data is spectral data.

   74. A method according to claim 73, wherein said received data points are peaks *....S * * in said spectral data. **�*.S. * S

   75. A method according to claim 73 or 74, wherein said received data is data generated by a charged particle energy analyser.SSS.S. * S

   -----76-A -method-according-toclaim75whriWsaid hãiéJ -filé ihrgy -- analyser is selected from the group consisting of: an Auger mass spectrometer, an X- * Ray photoelectron spectroscope and an Electron Probe Micro Analyser.

   77. A computer program comprising computer readable instructions configured to cause a computer to carry out a method according to any one of claims 68 to 76.

   78. A computer readable medium carrying a computer program according to claim 77.

   79. Apparatus for determining the presence or absence of an element in a sample, the apparatus comprising: a memory storing processor readable instructions; and a processor arranged to read and execute instructions in said memory; wherein said processor readable instructions comprise instructions arranged to control the computer to carry out a method according to any one of claims 68 to 76.

   80. Apparatus for determining the presence or absence of an element in a sample, the apparatus comprising: means for obtaining stored data relating to said element, said stored data comprising a plurality of stored data points associated with said element, each data point associated with said element having an associated energy value; means for receiving data generated from said sample, said received data comprising a plurality of received data points, each of said plurality of received data points having an associated energy value and an associated score; means for processing said plurality of stored data points and a received data point having a maximum score to determine an energy difference, said energy difference indicating a difference in energy value between a stored data point having an energy value closest to said received data point having a maximum score; means for identifying at least one of said plurality of received data points in the received data based upon said stored data and said determined energy difference; and * * means for determining the presence or absence of said element based upon *SS.. . . . . . * * said identified at least one of said plurality of received data points. * *

   *. : 81. A method for determining the presence or absence of an element in a sample * comprising: receiving data generated from said sample; receiving data indicating at least one parameter of an apparatus used to generate said data from said sample; obtaining stored data relating to said element; determining the presence or absence of the element based upon said received data generated from said sample, said rec&ved data indicating at least one parameter of said apparatus and said obtained stored data.

   82. A method according to claim 81, wherein said stored data comprises a plurality of data points associated with said element and wherein determining the presence or absence of the element comprises determining a subset of said plurality of data points associated with said element based upon said parameter.

   83. A method according to claim 82, wherein determining the presence or absence of the element comprises: determining a relationship between said subset of said plurality of data points associated with said element and said received data; and determining the presence of said element in said sample if said relationship satisfies a predetermined criterion, said determining being independent of any relationship between said plurality of data points associated with said element not in said subset.

   84. A method according to any one of claims 81 to 83, wherein said received data generated from said sample is spectral data.

   85. A method according to claim 84, wherein said received data generated from * . said sample comprises a plurality of data points and said plurality of data points are peaks in said spectral data. **.. * .

   86. A method according to claim 84 or 85, wherein said received data is data generated by a charged particle energy analyser.

   87. A method according to claim 86, wherein said at least one parameter is a * parameter associated with deflection of charged particles.

   88. A method according to claim 87, wherein said charged particle energy analyser is selected from the group consisting of: an Auger mass spectrometer, an X-Ray photoelectron spectroscope and an Electron Probe Micro Analyser.

   89. A computer program comprising computer readable instructions configured to cause a computer to carry out a method according to any one of claims 81 to 88.

   90. A computer readable medium carrying a computer program according to claim 89.

   91. Apparatus for determining the presence or absence of an element in a sample, the apparatus comprising: a memory storing processor readable instructions; and a processor arranged to read and execute instructions in said memory; wherein said processor readable instructions comprise instructions arranged to control the computer to carry out a method according to any one of claims 81 to 88.

   92. Apparatus for determining the presence or absence of an element in a sample, the apparatus comprising: means for receiving data generated from said sample; means for receiving data indicating at least one parameter of an apparatus used to generate said data from said sample; means for obtaining stored data relating to said element; means for determining the presence or absence of the element based upon said received data generated from said sample, said received data indicating at least *...S. . * . one parameter of said apparatus and said obtained stored data.

   * ** *** * * 93. A method for determining one or more elements which are present in a sample, comprising: receiving data generated from said sample; ---- e]ddTàach ----data point having an associated value; * determining possible elements in the sample based upon said determined subset of data points; and further processing the received data based upon known data associated with said determined possible elements to determine one or more elements present in said sample.

   94. A method according to claim 93 wherein determining a subset of data points based upon said received data comprises determining data points most likely to be useful in determining said one or more elements present in said sample.

   95. A method according to claim 94, wherein said received data comprises a plurality of data points and determining data points most likely to be useful in determining said one or more elements present in said sample comprises: processing said received data to determine at least one data point having a value in a first range; and combining said determined at least one data point having a value in said first range with at least one further data point having a value not in said first range.

   96. A method according to claim 95, wherein the value associated with each of said data points is an energy value.

   97. A method according to claim 96, wherein said first range is a low energy range of values.

   98. A method according to any one of claims 95 to 97, wherein processing said received data to determine at least one data point having a value in the first range comprises: performing a background subtraction process on said received data to **....* * generate first background subtraction data; and determining said at least one data point having a value in the first range fromsaid first background subtraction data. *S**

   99. A method according to claim 98, wherein performing a background -subtraction-process on-said-receiveddatacomprises: -- ::::hh; processing the received data a plurality of times, each processing being * based upon a respective value for a parameter, to generate respective intermediate sets of data points; processing said respective intermediate sets of data points to determine correspondence between data points in said respective intermediate sets of data points; and generating said first background subtraction data from said determined correspondence.

   Claims are truncated...