semi合集-English.pdf - 第1745页
SEMI E42-0704 © SEMI 1995, 2004 14 2.2.4 Recipe Executor — The recipe executor (executor) is the com ponent of an agent that understands the contents of a recipe and is able to verify their syntactical correctness and to…
SEMI E42-0704 © SEMI 1995, 2004 13
2.2 Major Objects in RMS — Recipe Management is concerned with the major areas of functionality illustrated in
Figure 2.1: recipes
3
, recipe namespaces providing persistent recipe storage, recipe namespace managers that
provide management of the namespace and access to its recipes, and the recipe executor that executes recipes.
Figure 2.1
Major Objects of RMS
3 The recipe object in Figure 2.1 is not itself a standardized object. However, two recipe subtypes defined in Section 3 are standardized objects.
2.2.1 Recipes — Recipes provide a flexible,
manipulatable, and re-usable form for users to select
sequencing and settings to effect a particular result.
Equipment uses recipes in a variety of ways to control
the processing environment, sometimes using several
types of recipes together. Recipes may also be used for
maintenance activities, such as calibration or cleaning.
Recipes can be modified and copied from one
environment to another. Their inherent flexibility as a
form also becomes an endless source of problems
unless they can be managed. Misprocessing,
unintentionally running the wrong recipe for a given
product, is prohibitively expensive for factories.
RMS defines a recipe as an object with attributes —
information about the recipe — as well as content. It is
the attributes of the recipe that allow true management
of recipes, throughout their lifecycles, to occur.
2.2.2 Recipe Namespace — A recipe namespace in
RMS is analogous to a smart file directory. The
namespace provides long-term recipe storage
capability.
Basic operations are part of the namespace
specification. For example, recipes can be copied from
one namespace to another, copied within a namespace,
renamed, created, updated, and deleted. They can be
downloaded to, and uploaded from, an application
called a recipe executor. Several recipes can be linked
together to form a set.
The namespace also can serve as a recipe pool that can
be shared by a group of equipment of a common type.
The distributed recipe namespace is a namespace able
to utilize and manage recipe storage provided by such
an equipment group.
2.2.3 Recipe Namespace Manager — The recipe
namespace itself is passive, a container for recipes. The
recipe namespace manager (manager) provides the
dynamic element that manages the namespace. The
manager represents the interface for the namespace to
the external world and the internal decision authority
within the namespace.
SEMI E42-0704 © SEMI 1995, 2004 14
2.2.4 Recipe Executor — The recipe executor (executor) is the component of an agent that understands the
contents of a recipe and is able to verify their syntactical correctness and to validate that it can be executed under
the current configuration. It executes a recipe by reading its contents and applying them appropriately in order to
achieve a desired result.
The recipe executor also may provide additional limited storage capability, but it is restricted from modifying
existing recipes, except under special conditions, to prevent unexpected and/or unwanted change.
2.3 Agent — As used in RMS, an agent is a system, or subsystem, in a factory, that has a physical aspect. It consists
of one or more applications that provide and/or use service resources, such as a recipe namespace manager and/or a
recipe executor, as illustrated in Figure 2.2.
Figure 2.2
An Example of an Agent and Component Applications
Agent is a generalization that covers traditional equipment, supervisory or aggregate systems such as stations, cells,
and clusters, and intelligent subsystems within equipment, such as process modules within a cluster. Other types of
agents may also be implemented, such as those dedicated to recipe namespace management.
An agent that provides a recipe executor is called an executing agent. Equipment is an executing agent with
associated hardware to which the recipe applies and which it uses to do work.
2.4 Implementations — RMS can be implemented on different platforms within a factory. A host controller may
only provide recipe namespace management, while a diskless cluster module may provide only recipe execution.
Traditional equipment capable of operating in stand-alone mode is required to provide a recipe namespace that is
available on powerup as well as a recipe executor component. Additional examples of implementations of RMS are
provided in Related Information 1.
RMS defines the services provided through an external interface. Communications between or within applications
that are internal to an agent are not covered by RMS, so long as the attributes and operations comply with RMS
requirements.
3 Recipes
This section describes the concepts concerning recipes.
3.1 Motivations — Definitions of recipe structure, attributes, and operations address the following issues:
• maintenance of recipe attributes for management and traceability,
• provision of a standard method for transferring attributes separately, as well as with the recipe body,
• reduction or elimination of the need for dedicated recipe editors,
• support for applications requiring multiple recipe types, such as formats or language syntaxes,
• run-to-run control,
SEMI E42-0704 © SEMI 1995, 2004 15
• support for sorting by factory approval and
certification levels,
• the ability to tune a generic recipe for a specific
installation of agent as well as the same agent over
time, and
• the ability to share recipes across executing agents
supporting the same recipe syntax or language(s).
3.2 Basic Concepts — There are two types of recipes
addressed by RMS, the managed recipe that is stored
in a recipe namespace and the execution recipe that is
stored in recipe execution storage. This section
describes the basic concepts concerning both types of
recipes.
3.2.1 Types of Recipes — There are three types of
recipe objects shown in Figure 3.1. Most of Recipe
Management is concerned with the management of
recipes stored in a recipe namespace. These recipes are
called managed recipes. Recipes are also stored by the
recipe executor in the recipe execution storage, and
these are called execution recipes.
Figure 3.1
Recipe Types
The recipe supertype is an abstract type that shows the
common elements and attributes of the managed recipe
and the execution recipe and their differences. The
recipe supertype is not itself intended to be
implemented. That is, recipe object types formally
defined in RMS are either managed recipes or
execution recipes. As indicated by the light lines, the
supertype is not a standardized object.
3.2.2 Recipe Structure — Both the managed and the
execution recipe have a body
4
and a set of attributes.
The body (contents) of the recipe contains the data used
by the recipe executor for its execution process. The
body is also not a standardized object and may not be
accessed through Object Services.
Figure 3.2 illustrates all of the attributes defined by
RMS that are common to both the managed recipe and
the execution recipe. Not all of these attributes are
required for a minimum implementation of RMS. The
rules for setting the values of certain attributes,
however, are different for the managed recipe and the
execution recipe.
Figure 3.2
Recipe Supertype
The identification attributes, those attributes used for
purposes of identification, are discussed in Section 3.3.
The other attributes are discussed in later sections
specific to each recipe type.
The managed recipe also may maintain a set of
attributes that are specific to a single executing agent.
The agent-specific attributes are technically attributes
of the association between the recipe and the executing
agent. In RMS, however, these attributes are treated as
attributes of a component of the recipe called the agent-
specific dataset, in order to allow them to be managed
within the namespace along with the other attributes of
the recipe. Agent-specific attributes allow the recipe to
be personalized for particular executing agents (i.e.,
particular recipe executors).
4 The recipe’s body corresponds to a “process program” in SEMI E5
and SEMI E30.